Landing control method and platform of modular square cabin type fire extinguishing and rescue unmanned aerial vehicle

Abstract

The application discloses a take-off and landing control method and platform of a modular square cabin type fire extinguishing and rescue unmanned plane, relates to the technical field of unmanned plane flight control, and comprises the following steps: triggering a state synchronization instruction after receiving a flight task, and obtaining an unmanned plane state self-checking result; establishing a time sequence environment data set by performing environment situation awareness; performing dynamic take-off and landing score discrimination by configuring a sliding time window for extraction; and performing segmented take-off and landing control optimization based on the extraction result and the state self-checking result, and using the segmented take-off and landing control optimization for take-off and landing control management. The application solves the technical problems that the traditional take-off and landing control method is difficult to effectively realize dynamic perception and fusion control on non-electric variables such as wind speed field, and that the take-off and landing risk assessment of the modular square cabin type fire extinguishing and rescue unmanned plane is not accurate, segmented control is not fine, and task continuity is poor in a complex environment, and achieves the technical effects of accurate take-off and landing risk assessment, fine segmented control and continuous task execution of the modular square cabin type fire extinguishing and rescue unmanned plane in a complex environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle flight control, in particular to a take-off and landing control method and platform of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle. BACKGROUND

[0002] In the fire extinguishing and rescue scene, the take-off and landing control of the unmanned aerial vehicle is crucial for the accurate processing of non-electric variables (wind speed field, heat convection intensity, etc.). The existing technology mostly adopts the traditional fixed parameter control strategy, which has obvious shortcomings in complex environments such as semi-open fire scenes. Since the non-electric variables (such as dynamic wind speed and sudden heat convection) in the fire scene have strong time-varying characteristics, the traditional method cannot real-time fuse the unmanned aerial vehicle state and non-electric variable dynamic data, and it is difficult to establish an accurate take-off and landing risk assessment model. At the same time, it lacks segmented fine control driven by non-electric variables, resulting in poor adaptability of the unmanned aerial vehicle to non-electric variable interference during take-off and landing, and the inability to realize load transfer and task reconstruction based on non-electric variable characteristics (such as obstacle geometric position) after recovery. These problems result in low take-off and landing safety and poor task efficiency of the unmanned aerial vehicle in complex environments, which is difficult to meet the needs of emergency rescue scenes dominated by non-electric variables. SUMMARY

[0003] The present application provides a take-off and landing control method and platform of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle, which solves the technical problems that the traditional take-off and landing control method cannot effectively realize dynamic perception and fusion control of non-electric variables such as wind speed field, resulting in inaccurate take-off and landing risk assessment, non-fine segmented control and poor task continuity of the modular shelter type fire extinguishing and rescue unmanned aerial vehicle in complex environments.

[0004] In a first aspect, the present application provides a take-off and landing control method of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle, the method comprising: triggering a state synchronization instruction after receiving a flight task, and obtaining a state self-checking result of the fire extinguishing and rescue unmanned aerial vehicle according to the state synchronization instruction; performing environment situation awareness according to the state synchronization instruction, and establishing a time sequence environment data set, the time sequence environment data set including wind speed field, heat convection intensity, visibility, and obstacle characteristics; after configuring a sliding time window to extract the time sequence environment data set, performing dynamic take-off and landing scoring discrimination according to the extraction result and the state self-checking result; if the dynamic take-off and landing scoring discrimination result is a pass result, performing segmented take-off and landing control optimization based on the extraction result and the state self-checking result, and establishing a segmented take-off and landing control optimization result; and performing take-off and landing control management using the segmented take-off and landing control optimization result.

[0005] In a second aspect of the present application, a take-off and landing control platform of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle is provided, and the platform comprises: a flight task acquisition module, configured to trigger a state synchronization instruction after receiving a flight task, and acquire a state self-checking result of the fire extinguishing and rescue unmanned aerial vehicle according to the state synchronization instruction; an environment data set construction module, configured to perform environment situation awareness according to the state synchronization instruction, and construct a time sequence environment data set, wherein the time sequence environment data set comprises a wind speed field, a thermal convection intensity, a visibility, and an obstacle feature; a take-off and landing score discrimination module, configured to extract the time sequence environment data set by configuring a sliding time window, and perform dynamic take-off and landing score discrimination according to an extraction result and the state self-checking result; an optimization result construction module, configured to, if the dynamic take-off and landing score discrimination result is a pass result, perform segmented take-off and landing control optimization based on the extraction result and the state self-checking result, and construct a segmented take-off and landing control optimization result; and a take-off and landing control execution module, configured to perform take-off and landing control management by using the segmented take-off and landing control optimization result.

[0006] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0007] In the present application, after receiving a flight task, a state synchronization instruction is triggered, a state self-checking result of an unmanned aerial vehicle is acquired, and environment situation awareness is performed, a time sequence environment data set containing a wind speed field, a thermal convection intensity, etc. is constructed, and dynamic take-off and landing score discrimination is performed after extracting features by using a sliding time window. If the discrimination passes, segmented take-off and landing control optimization is performed based on an extraction result and a self-checking result, control optimization results of stages such as hovering preparation and dynamic descent are constructed, and then take-off and landing control management is performed. At the same time, by using a critical escape trigger, a multi-dimensional score system, and segmented fine control, combined with a task closed-loop design of a modular shelter, precise control of unmanned aerial vehicle take-off and landing in a complex environment and task continuity are achieved, take-off and landing safety and efficiency in a fire extinguishing and rescue scene are improved, and the technical effects of precise take-off and landing risk assessment, fine segmented control, and continuous task execution of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle in a complex environment are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 is a flowchart of a take-off and landing control method of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle provided by the embodiments of the present application.

[0010] Figure 2A structure schematic diagram of a take-off and landing control platform of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle provided by the embodiment of the application.

[0011] Legend: flight task acquisition module 1, environment data set construction module 2, take-off and landing score discrimination module 3, optimization result construction module 4, take-off and landing control execution module 5. DETAILED DESCRIPTION

[0012] The application provides a take-off and landing control method and platform of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle, and aims to solve the technical problems that the traditional take-off and landing control method cannot effectively realize dynamic perception and fusion control on non-electric variables such as wind speed field, and thus leads to inaccurate take-off and landing risk assessment, non-fine segmented control and poor task continuity of the modular shelter type fire extinguishing and rescue unmanned aerial vehicle in a complex environment.

[0013] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0014] It should be noted that the terms "first", "second" and the like in the specification and the above drawings of the application are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment one, as shown in the figure, a take-off and landing control method of a modular shelter type fire extinguishing and rescue unmanned aerial vehicle, wherein the method comprises: Figure 1

[0016] Step A100: after receiving a flight task, triggering a state synchronization instruction, and obtaining a state self-checking result of the fire extinguishing and rescue unmanned aerial vehicle according to the state synchronization instruction.

[0017] In the embodiment of the application, the state synchronization instruction is a core control instruction for synchronously obtaining the state of the fire extinguishing and rescue unmanned aerial vehicle and external environment data.

[0018] ​Specifically, after receiving the flight task, a state synchronization instruction is triggered, which first activates the multi-sensor detection network built in the UAV. The power system sensor collects real-time motor speed data, the battery management system feeds back voltage parameters, the inertial measurement unit synchronously outputs attitude information such as pitch angle and yaw angle, and the landing gear position sensor feeds back the retraction state, such as showing that the landing stage is locked. The data of these core detection points are transmitted to the on-board central processor through the CAN bus.

[0019] Then, the central processor performs a three-level verification process on the collected data: first, threshold comparison is performed, such as triggering a preliminary warning when the motor temperature exceeds 85°C; second, time domain trend analysis is carried out, such as judging the power decay rate through 5 groups of battery voltage data (22.2V, 22.1V, 22.0V, 21.9V, 21.8V); finally, the historical health model is called, and the current vibration frequency spectrum (such as propeller vibration frequency) is compared with the baseline data of the new machine to generate a health score for each system, such as a power system score of 92 and a flight control system score of 95.

[0020] The construction of the historical health model begins with the baseline data collection of the new machine when it is delivered from the factory. Through a multi-sensor array (such as a vibration acceleration sensor and a temperature sensor), the running parameters of each system under standard working conditions are obtained, such as the propeller unloaded vibration frequency of 120Hz and the motor rated temperature of 45°C, forming an initial health benchmark library. During the service process of the UAV, the full-state running data after each task is automatically collected according to the flight period, and after removing the noise through Kalman filtering, the key characteristic parameters are extracted, such as the vibration spectrum peak value of the power system and the battery voltage decay slope. The time series analysis method is used to construct the health trend model of each system. When the UAV undergoes maintenance or component replacement, the system triggers the model update mechanism, merges the running data of the new component with the historical data, adjusts the model parameters through the weighted moving average algorithm, and forms a dynamic and iterative health evaluation benchmark. For example, after replacing the propeller of a certain UAV, its vibration spectrum peak value shifts from 120Hz to 118Hz, and the model automatically includes this parameter in the historical database and recalculates the normal fluctuation threshold (118Hz±5Hz) to ensure the accuracy of the health score. This model continuously accumulates running data throughout the life cycle to construct a multi-dimensional health evaluation system that includes dynamic thresholds and decay trends of each system, providing a reliable historical reference for real-time state verification.

[0021] After the completion of the check, the state abnormality is discriminated according to the ISO26262 functional safety standard, if the score of a subsystem is lower than 70 points, such as the battery health score is 65 points, an abnormal early warning containing a fault code (such as BMS-007) is generated, and is pushed to the ground control terminal in a priority classification manner, and a high priority fault needs to be responded within 10 seconds. Finally, all detection and discrimination results are integrated to form a state self-checking result containing multiple core indicators, such as the power system is normal, the flight control system is normal, and the battery capacity is remaining 68%, and is stored in the on-board memory in JSON format and is uploaded to the unmanned aerial vehicle management system synchronously.

[0022] Through real-time collection by multiple sensors, three-level data verification and standardized abnormality discrimination process, the accuracy and comprehensiveness of the state self-checking result are ensured, and reliable equipment state data support is provided for subsequent environmental situation awareness and dynamic take-off and landing decision-making.

[0023] Step A200: environmental situation awareness according to the state synchronization instruction, establishing a time-series environmental data set, the time-series environmental data set including wind speed field, thermal convection intensity, visibility, obstacle characteristics.

[0024] Optionally, after receiving the flight task and triggering the state synchronization instruction, the multi-sensor cooperative collection mechanism is started synchronously. During the collection process of the time-series environmental data set, the parameter range of each environmental characteristic is quantified, and the data collection standard is shown in Table 1. The three-dimensional ultrasonic anemometer captures the spatial wind speed field data in real time at a sampling frequency of 10 Hz, for example, at a height of 10 meters from the ground, the X-axis wind speed is 3.2 m / s, the Y-axis wind speed is -1.5 m / s, and the Z-axis wind speed is 0.8 m / s, forming a three-dimensional vector wind speed sequence; the infrared thermal imager detects thermal radiation signals through the 8-14 μm wave band, and the thermal convection intensity is obtained by conversion according to Planck's law, for example, the thermal convection intensity at a height of 20 meters above a certain fire point is 4.7 kW / m², and 25 frames of temperature field distribution data are generated per second.

[0025] The binocular vision sensor combines with the laser radar to construct an environmental perception network, the vision system obtains a 640x480 pixel grayscale image through the RGB-D camera, and the laser radar transmits 16 laser beams at a frequency of 5 Hz to obtain the three-dimensional coordinate information of the obstacles. For example, a building debris with a height of 5 meters is detected 15 meters in front, and its coordinate range is (100, 200, 0)-(120, 210, 5), and an obstacle contour feature vector is generated through a feature point extraction algorithm. The visibility detection uses the principle of infrared scattering to measure the attenuation coefficient of aerosol to 780 nm wavelength light, and converts to obtain the current visibility distance, and updates the data once per second.

[0026] All sensor data is calibrated by a synchronous clock and stored in a ring buffer in timestamp order. The ring buffer continuously records sensor data such as wind speed field, thermal convection intensity, visibility, and obstacle characteristics in timestamp order through a circular storage mechanism, forming a time-series environmental data set containing wind speed field three-dimensional vector (timestamp+X / Y / Z axis wind speed), thermal convection intensity (timestamp+power density), visibility (timestamp+distance value), and obstacle characteristics (timestamp+three-dimensional coordinates+profile vector). For example, at 14:30:25, the data set records wind speed field (3.2,-1.5,0.8) m / s, thermal convection 4.7 kW / m², visibility 8.5 m, and obstacle coordinates (100,200,0)-(120,210,5). Subsequent real-time data is added every second, forming a continuous time series of environmental situation.

[0027] Through multi-sensor collaborative collection, clock synchronization calibration, and structured data storage, a time-series environmental data set containing wind speed field, thermal convection intensity, visibility, and obstacle characteristics is constructed, providing real-time, multi-dimensional environmental data support for subsequent feature extraction based on a sliding time window and dynamic take-off and landing risk assessment.

[0028] Table 1: Time-series environmental data set parameter table

[0029] ;

[0030] Step A300: After configuring the sliding time window to extract the time-series environmental data set, dynamic take-off and landing scoring is determined based on the extraction results and the state self-checking results.

[0031] In the embodiments of the present application, the sliding time window is a data processing mechanism for dynamically extracting features from the time-series environmental data set (containing wind speed field, thermal convection intensity, etc.).

[0032] In one embodiment of the present application, after completing the construction of the time-series environmental data set, a sliding time window configuration module is started to perform feature extraction on the data set. First, set the time window basic parameters, such as setting the window length to 5 seconds (corresponding to 50 groups of 10 Hz sampled wind speed data) and the sliding step to 1 second, ensuring that an analysis window covering the latest 5 seconds of environmental data is generated every 1 second. Taking three-dimensional wind speed field data as an example, in the 14:30:25-14:30:30 time window, the X-axis wind speed sequence [3.2,3.1,3.3,3.0,2.9] m / s is extracted, and its mean value 3.1 m / s, standard deviation 0.15 m / s, and fluctuation frequency (main frequency obtained by FFT transformation) are calculated, forming the time-frequency feature vector of the wind speed field.

[0033] For the heat convection intensity data, a variable window strategy is adopted: when a sudden change in heat convection intensity is detected (such as a change rate exceeding 1 kW / m² / s), the window is automatically reduced to 2 seconds (corresponding to 50 frames of temperature field data) to capture the transient change characteristics. For example, the heat convection intensity of a certain fire point suddenly rises from 4.7 kW / m² to 5.9 kW / m² within 2 seconds, and the rising slope of 0.6 kW / m² / s and peak value of 5.9 kW / m² are extracted as characteristic parameters through sliding window extraction.

[0034] The extraction of visibility is combined with time series continuity analysis, such as the visibility continuously decreasing from 8.5 meters to 5.2 meters within 10 consecutive sliding windows (10 seconds), and the system extracts the decay rate of 0.33 meters / second as a trend feature.

[0035] When performing sliding window processing on obstacle coordinate data, first take 5 seconds as the window length and 1 second as the sliding step to intercept the time series coordinate sequence, such as the three-dimensional coordinates of the obstacle during 14:30:25-14:30:30. Taking a certain building debris as an example, its first frame coordinate in the window is (100, 200, 5) (14:30:25), and the last frame coordinate is (100.3, 200.1, 5.1) (14:30:30). The coordinate offset is calculated by the Euclidean distance formula: meters, since the offset of 0.33 meters is less than the threshold of 0.5 meters (which is determined by a person skilled in the art according to actual conditions), it is determined as a static obstacle. The stability score is converted according to the proportion of the offset to the threshold, and the formula is: stability = 1-(offset / threshold) x 100%, i.e. 1-(0.33 / 0.5) x 100% = 34%, and the final stability score of the obstacle in the current window is 100%-34% = 66%, which is used for subsequent obstacle impact feature scoring and path planning stability evaluation.

[0036] After all the extracted features are standardized, they are combined in timestamp order to form an extraction result dataset containing wind speed time-frequency features, heat convection rate, visibility trend, obstacle stability, etc. For example, the extraction result record at 14:30:30 is: [wind speed average 3.1 m / s, heat convection slope 0.6 kW / m² / s, visibility decay 0.33 m / s, obstacle stability 66%].

[0037] Then, according to the extraction results, environmental change fitting and risk trend fitting are performed to identify critical escape conditions, and if triggered, the scoring is skipped and the segmented takeoff and landing control optimization is entered directly. The specific steps are described in detail in A310-A330.

[0038] Further, the extraction result is divided into three levels of time characteristics and a time spectrum vector is established, a first score result of environmental characteristics such as wind speed and thermal convection is calculated, and a dynamic take-off and landing score is distinguished in combination with a second score result of the state self-checking, and specific steps are described in detail in A340-A370.

[0039] By configuring a multi-scale sliding time window combined with a fixed step and a variable window, time-frequency feature extraction and three-level time characteristic division are performed on time-series environmental data, the dynamic change law of environmental parameters such as wind speed field and thermal convection is captured, and accurate environmental characteristic data support is provided for subsequent dynamic take-off and landing score discrimination and segmented control optimization based on the extraction result.

[0040] Step A400: If the dynamic take-off and landing score discrimination result is a pass result, control optimization of segmented take-off and landing is performed based on the extraction result and the state self-checking result, and a segmented take-off and landing control optimization result is established.

[0041] Specifically, if the dynamic take-off and landing score discrimination result is a pass result, based on the extraction result and the state self-checking result, control optimization of the hovering preparation phase, control optimization after time-series influence prediction of the dynamic descent segment path fitting, and landing control optimization after state update of the landing contact segment are performed in sequence, and the three results are output as the landing control optimization result, and specific steps are described in detail in A410-A440.

[0042] Step A500: Using the segmented take-off and landing control optimization result for take-off and landing control management.

[0043] Specifically, the dynamic descent control optimization result is used for dynamic descent control of the fire-fighting rescue unmanned aerial vehicle, a descent state self-checking result is established, and stable abnormality discrimination is performed on the result, if it is abnormal, a temporary hovering instruction is generated, and after automatic attitude compensation, the dynamic descent control is continued, and specific steps are described in detail in A510-A530.

[0044] Further, the method provided in the embodiment of the application comprises:

[0045] A310: Environmental change fitting is performed according to the extraction result, and an environmental change prediction result is established.

[0046] A320: Risk trend fitting is performed using the environmental change prediction result, and trigger identification of critical escape is performed on the risk trend fitting result.

[0047] A330: If critical escape is triggered, the dynamic take-off and landing score discrimination is cancelled, and control optimization of segmented take-off and landing based on the extraction result and the state self-checking result is directly performed.

[0048] In the embodiments of the present application, critical escape is an emergency triggering mechanism for sudden high-risk environment in unmanned aerial vehicle take-off and landing control.

[0049] Specifically, after completing the feature extraction of the time series environment data set by the sliding time window, first, based on the extraction results, such as the wind speed field sequence [3.2, 3.5, 3.8, 4.1, 4.3] m / s, the thermal convection intensity sequence [4.5, 4.7, 4.9, 5.1, 5.3] kW / m² and other data in the last 5 seconds, environment change fitting is performed:

[0050] Step a: a cubic polynomial fitting algorithm is used to model the wind speed data, and the historical data sequence of the wind speed field is extracted from the time series environment data set by the sliding time window, for example, the wind speed sampling value [3.2, 3.5, 3.8, 4.1, 4.3] m / s in the last 5 seconds is obtained, and the corresponding time stamp t=0 to t=4 seconds. A cubic polynomial fitting algorithm y=at³+bt²+ct+d is used to curve fit the sequence, and the least square method is used to solve the coefficients, so that the mean square error of the fitting curve and the historical data is minimized. The calculation result is y=0.2t²+0.1t+3.2, because the coefficient a of the cubic term is 0, the actual degradation is a quadratic polynomial, the fitting goodness R² of the equation to the historical data is greater than 0.95, indicating that the model can effectively represent the wind speed change trend. When predicting the wind speed in the future 1 second (t=5) and 2 seconds (t=6), the equation is substituted to get y(5)=0.2×5²+0.1×5+3.2=4.6 m / s, y(6)=0.2×6²+0.1×6+3.2=5.1 m / s, that is, the wind speed will rise to 4.6 m / s and 5.1 m / s respectively in the next 2 seconds.

[0051] Step b: for the thermal convection intensity data, the historical sequence [4.5, 4.7, 4.9, 5.1, 5.3] kW / m² with obvious recent growth trend is selected, based on the physical characteristics of the exponential increase of thermal convection at the fire scene with time, an exponential growth model y=y0×k^t is selected for modeling. Where y0 is the initial value, and k is the growth coefficient. Through logarithmic linearization processing of the historical data, lny=lny0+t×lnk, the linear regression is used to obtain y0=4.5, k=1.05, and the model expression is y=4.5×1.05^t. When predicting the thermal convection intensity after 2 seconds, t=2 is substituted to get y=4.5×1.05²≈5.8 kW / m², that is, the thermal convection intensity will grow from the current 5.3 kW / m² to 5.8 kW / m² after 2 seconds, reflecting the rapid enhancement trend of fire thermal radiation.

[0052] Step c: For the visibility data, the system extracts the historical sequence through a sliding time window, such as the visibility values in the last 10 seconds [8.5, 8.2, 7.9, 7.5, 7.2, 6.8, 6.5, 6.1, 5.8, 5.5] m / s. Considering the exponential decay characteristic of smoke concentration at the fire scene over time, an exponential function model y = y0 x e^(-kt) is used for fitting. Through least squares calculation, the initial value y0 = 8.5 and the decay coefficient k = 0.05 are obtained, and the fitting equation is y = 8.5 x e^(-0.05t). When predicting the visibility in the next 2 seconds, t = 11 and t = 12 are substituted into the equation, y(11) = 8.5 x e^(-0.05 x 11) ≈ 5.2 m, and y(12) = 8.5 x e^(-0.05 x 12) ≈ 5.0 m, reflecting the continuous downward trend of visibility.

[0053] Step d: For the obstacle feature, if a dynamic obstacle such as a moving building debris is detected, its historical coordinate sequence is extracted, for example, the coordinates of a certain obstacle in the last 5 seconds (100, 200, 5), (101, 201, 5.1), (102, 203, 5.2). A linear regression model is used for trajectory fitting, obtaining the displacement equation x = 100 + 0.4t, y = 200 + 0.6t, z = 5 + 0.04t (t is time, unit: seconds), and the model accuracy is verified by the root mean square error (RMSE < 0.2 meters). The obstacle coordinates after 2 seconds are (100.8, 201.2, 5.08), and combined with the obstacle contour feature data (such as length x width x height = 20 m x 10 m x 5 m), the risk prediction result of the dynamic obstacle is generated.

[0054] Then, the risk trend fitting is performed using the environmental change prediction results obtained by the above steps. The wind speed safety threshold can be set to 5 m / s, the thermal convection safety threshold can be set to 6 kW / m², etc. (the specific threshold values are determined by those skilled in the art according to the actual situation), and the risk trend function is calculated. When the wind speed prediction value will exceed the threshold in 2 seconds, such as the aforementioned prediction of 5.1 m / s, and the risk trend fitting curve shows that the risk index will increase from the current 30% to 75% (exceeding the critical escape threshold of 50%) in the next 2 seconds, the critical escape mechanism is triggered. For example, if the thermal convection prediction value increases from 5.3 kW / m² to 5.8 kW / m² in 2 seconds, although it does not reach the threshold, the risk growth rate exceeds 0.5 kW / m² / s, which will also trigger the escape identification.

[0055] If the critical escape is triggered, the system immediately cancels the dynamic take-off and landing score judgment, directly calls the obstacle feature in the extraction result, such as the obstacle coordinates (100, 200, 5)-(120, 210, 8) in front of 10 meters and the state self-checking result (such as battery power 68%, power system normal), enters the subsection take-off and landing control optimization. At this time, the shortest safe path is preferentially selected, the optimal height of the hovering preparation stage is calculated as 20 meters, the descending speed of the dynamic descending stage is set as 1.5 m / s, and the attitude compensation parameters are adjusted in real time, such as the pitch angle adjustment-3°, to ensure that the take-off and landing operation is completed before the risk (such as strong wind) comes.

[0056] Through the fitting of the extraction result and the risk trend modeling, combined with the preset critical escape threshold, the rapid identification and response to the sudden high-risk environment are realized, the time delay of the traditional scoring judgment process is avoided, the subsection take-off and landing control optimization is triggered in advance, and the safety and timeliness of the unmanned aerial vehicle in the complex environment are ensured.

[0057] Further, the step A300 in the method provided in the embodiments of the present application comprises:

[0058] A340: dividing the extraction result into three levels of time features.

[0059] A350: establishing a time spectrum vector according to the three levels of time features, using the time spectrum vector to perform scoring calculation on the wind speed influence feature, the thermal convection disturbance feature, the visible disturbance feature, and the obstacle influence feature, and establishing a first scoring result.

[0060] A360: performing take-off and landing scoring analysis according to the state self-checking result, and establishing a second scoring result.

[0061] A370: using the first scoring result and the second scoring result to perform dynamic take-off and landing scoring judgment.

[0062] In the embodiments of the present application, the three levels of time features refer to the extraction result of the time series environment data set by the sliding time window, which is divided into three levels of feature categories according to the time scale.

[0063] Optionally, after completing the feature extraction of the time series environment data set by the sliding time window, the system first divides the environmental feature parameters in the extraction result into a high-frequency segment (1-5 seconds), a medium-frequency segment (5-30 seconds), and a low-frequency segment (more than 30 seconds) according to the three levels of time feature division rules. Taking the wind speed field data as an example, the high-frequency segment captures the instantaneous wind speed mutation, such as the wind speed rising from 3.2 m / s to 4.5 m / s within 3 seconds; the medium-frequency segment analyzes the wind speed trend, such as the average wind speed 3.8 m / s and the change rate 0.2 m / s² within 10 seconds; and the low-frequency segment identifies the environmental background wind speed, such as the average wind speed 4.0 m / s within 30 seconds, to form a three-level time feature matrix.

[0064] Then, a multi-dimensional time spectrum vector is established according to the three-level time characteristics, for example, the high-frequency wind speed mutation amplitude, the medium-frequency thermal convection change rate, the low-frequency visibility attenuation trend and other parameters are mapped as vector dimensions, such as [0.8, 0.6, 0.4], the numerical value represents the feature influence weight. The vector is used to calculate the quantitative score of the wind speed influence feature (such as instantaneous strong wind deduction 15 points), the thermal convection disturbance feature (such as thermal convection mutation deduction 20 points), the visible disturbance feature (such as visibility sudden drop deduction 10 points), and the obstacle influence feature (such as dynamic obstacle deduction 25 points). The scores are added according to the weight to obtain the first score result, and the full score is 100 points.

[0065] At the same time, the state self-checking result is analyzed for take-off and landing scoring, for example, the power system health score is 92 points, the battery endurance score is 85 points, and the flight control system score is 90 points. The second score result is calculated according to the preset weight (such as power system 40%, battery 30%, flight control 30%), 92x0.4+85x0.3+90x0.3=89.3 points. Finally, the first score result and the second score result are weighted and fused according to the proportion of 6:4, such as 75x0.6+89.3x0.4=81.1 points. If the total score exceeds the preset threshold (such as 70 points), it is determined that the dynamic take-off and landing scoring discrimination is passed.

[0066] By dividing the environmental characteristics according to the three-level time scale and constructing the time spectrum vector, combined with the quantitative score of the unmanned aerial vehicle state self-checking, the multi-dimensional evaluation of the take-off and landing risk in the complex environment is realized. Compared with the single time scale scoring method in the prior art, the accuracy and environmental adaptability of the dynamic take-off and landing scoring discrimination are significantly improved, and a scientific decision basis is provided for the segmented take-off and landing control optimization.

[0067] Further, the step A400 in the method provided by the embodiment of the application comprises:

[0068] A410: according to the extraction result and the state self-checking result, the hovering control optimization is performed, and a hovering preparation stage control optimization result is established.

[0069] A420: after path fitting of the dynamic descent segment, time sequence influence prediction of the extraction result is performed, and a time sequence influence prediction result is established.

[0070] A430: according to the path fitting result, the time sequence influence prediction result and the state self-checking result, the control optimization of the dynamic descent is performed, and a dynamic descent segment control optimization result is established.

[0071] A440: after the state update of the landing contact section fire rescue unmanned aerial vehicle, start the landing control optimization after state update, establish the landing section control optimization result, and output the hovering preparation stage control optimization result, dynamic descent section control optimization result, and landing section control optimization result as the landing control optimization result.

[0072] Specifically, when performing segmented take-off and landing control optimization based on the extraction result and the state self-checking result, first enter the control optimization of the hovering preparation stage. Based on the wind speed field and obstacle features extracted by the sliding time window, combined with the state self-checking result, set the hovering target height to 20 meters and the horizontal position deviation threshold to ≤0.5 meters. The PID controller takes the real-time collected height data (accuracy ±0.1 meters) and horizontal coordinates as inputs, calculates the height error (the difference between the current height and 20 meters) and the horizontal position error (the Euclidean distance between the current coordinates and the target coordinates), and generates a motor power adjustment signal by using the proportional term to quickly respond to error changes (such as a height deviation of 1 meter, a proportional coefficient Kp=0.5, and an output of 0.5 control amount), the integral term to eliminate static error (such as a continuous deviation of 0.5 meters, an integral coefficient Ki=0.1, and a cumulative output of 0.1xtime control amount), and the differential term to predict error trends (such as a height change rate of 0.3 m / s, a differential coefficient Kd=0.2, and an output of 0.2x0.3 control amount). The three are superimposed to generate a motor power adjustment signal, such as a 5%-10% increase or decrease in quadcopter motor power, to adjust the unmanned aerial vehicle attitude in real time. When the height is stable at 20 meters ±0.3 meters, the horizontal deviation is ≤0.5 meters, and the duration is more than 3 seconds, the hovering preparation stage control optimization result is established to ensure the stable attitude before landing.

[0073] Subsequently, the path fitting is performed on the dynamic descent section. The B-spline curve algorithm is used to smooth the landing path (such as the three-dimensional trajectory from 20 meters height to 5 meters height), and the time series influence prediction of the extraction result is performed, for example, according to the past 5 seconds of wind speed sequence [2.8, 3.1, 3.3, 3.0, 3.2] m / s to predict the future 10 seconds wind speed change trend, and the time series influence prediction result is established. Then, combined with the path fitting result (such as a descent rate of 1.5 m / s), the time series prediction result (the wind speed may rise to 3.5 m / s), and the state self-checking result (the power system output is stable), the control parameters of the dynamic descent are optimized by the particle swarm optimization algorithm to determine the optimal descent angle (-15°) and attitude compensation amount (yaw angle adjustment ±2°), and the dynamic descent section control optimization result is formed.

[0074] In the landing contact section, the state of the unmanned aerial vehicle is updated in real time, such as the height is reduced to 5 meters, the speed is updated to 0.5 m / s, and the landing gear state is ready for landing, and then the state updated landing control optimization is started. Based on the landing surface flatness data scanned by the laser radar in real time (such as flatness error ≤3cm) and the sensor data of state self-checking (such as height sensor error ±0.1m), the landing control optimization result of the landing section is established by using the fuzzy control algorithm to optimize the buffer intensity and the landing angle (ideal landing angle ≤3°) during landing. The control optimization results of the three stages are integrated into the complete landing control optimization result output.

[0075] By scientifically dividing the landing process into three stages of hovering, dynamic descending and landing contact, and combining the environment extraction results and state self-checking data for fine control optimization, the whole process precision control from macro path planning to micro attitude adjustment is realized, the safety and stability of the unmanned aerial vehicle landing in complex environment are improved, and technical support is provided for the reliable take-off and landing of the modularized shelter type fire fighting and rescue unmanned aerial vehicle in the fire field and other harsh environments.

[0076] Further, the step A410 in the method provided by the embodiment of the application comprises:

[0077] A411: Obtain the current position coordinates of the fire fighting and rescue unmanned aerial vehicle and the landing platform position coordinates.

[0078] A412: Perform landing path planning according to the current position coordinates, the landing platform position coordinates and the obstacle features, and establish a landing path planning result.

[0079] A413: Input the landing path planning result, the extraction result and the state self-checking result into the adaptive segmented planning channel, perform segmented planning, and establish the hovering section, the dynamic descending section and the landing contact section.

[0080] Specifically, before performing the hovering control optimization, the current position coordinates and the landing platform position coordinates, such as longitude and latitude coordinates and altitude, are obtained by the Beidou positioning module and the visual positioning system carried by the unmanned aerial vehicle at a sampling frequency of 10Hz, and the positioning accuracy can reach centimeter level. At the same time, combined with the obstacle features obtained by the laser radar scanning, such as the building debris with the coordinate range of (116°, 39°, 5m)-(117°, 40°, 10m) existing in front of 10m, the improved A* algorithm is used for landing path planning and the landing path planning result is obtained:

[0081] First, the current position coordinates of the unmanned aerial vehicle (116°, 39°) and the altitude of 5 meters are taken as the starting point, and the landing platform coordinates (116°, 39°, 0 meters) are taken as the end point. A grid map with a resolution of 1 meter x 1 meter x 1 meter is constructed in three-dimensional space. The algorithm obtains obstacle features such as building debris in the coordinate range (116°, 39°, 5 meters)-(117°, 40°, 10 meters) 10 meters in front of the laser radar, and sets a 5-meter range around the obstacle as a no-passing area to ensure a safety distance. The evaluation function uses f(n)=g(n)+h(n), where g(n) is the three-dimensional Euclidean distance cost from the starting point to the current node, such as a cost of 1 per 1 meter of movement, and h(n) is the heuristic estimate cost of the current node to the end point, which uses three-dimensional Manhattan distance. The algorithm expands the nodes by priority queue according to the f(n) value, and selects the node with the smallest cost to expand each time until the end node is reached. Finally, key path points such as (116, 39, 40 meters) are selected from the expanded path to generate a smooth three-dimensional landing curve, ensuring that the path maintains a safety distance of not less than 5 meters from the obstacle throughout the entire path.

[0082] Subsequently, the landing path planning results (such as the sequence of key path points), the environmental features extracted by the sliding time window (such as wind speed 3 m / s, thermal convection intensity 4 kW / m²), and the state self-check results (such as battery power 70%, flight control system normal) are input into the adaptive segmented planning channel. Based on the fuzzy logic algorithm, the path is automatically divided into: hovering segment, dynamic descent segment, landing contact segment according to time scale and task stage. The specific segmented take-off and landing control optimization parameters are shown in Table 2.

[0083] Through multi-source positioning data fusion, real-time modeling of obstacle features, and adaptive segmentation algorithm, precise control from global path planning to local stage subdivision is realized, ensuring that the landing path of the unmanned aerial vehicle in complex environment has safety and trajectory smoothness, providing reliable initial conditions and segmented control basis for subsequent hovering control optimization.

[0084] Table 2: Segmented take-off and landing control optimization parameter table

[0085] ;

[0086] Further, the method provided in the embodiments of the present application comprises the following steps:

[0087] A510: using the dynamic descent segment control optimization result to control the dynamic descent of the fire-fighting and rescue unmanned aerial vehicle, and performing the descent state self-check of the fire-fighting and rescue unmanned aerial vehicle to establish the descent state self-check result.

[0088] A520: performing stable abnormality discrimination of the descent state self-check result, and if the discrimination result is an abnormal result, generating a temporary hovering instruction.

[0089] A530: continue to perform dynamic descent control after automatic attitude compensation according to the temporary hovering instruction.

[0090] Specifically, first, according to the dynamic descent segment control optimization result, such as the optimal descent rate 1.5 m / s and the descent angle -15°, the unmanned aerial vehicle dynamic descent control is implemented. At the same time, the built-in sensor is used to collect the descent state data in real time at a frequency of 100 Hz, including height (accuracy ± 0.1 meters), horizontal speed (accuracy ± 0.2 m / s), pitch angle (accuracy ± 1°) and other parameters, and the descent state self-checking result is established. For example, when the unmanned aerial vehicle descends from 20 meters to 10 meters according to the planned path, the system records the real-time height 10.2 meters, the descent rate 1.48 m / s, and the pitch angle -14.5°, and the deviation from the optimization result is within 5%.

[0091] Then, the system performs stable abnormality discrimination on the descent state self-checking result, and sets the height fluctuation exceeding ± 0.5 meters, the horizontal speed deviation exceeding ± 0.3 m / s or the attitude angle deviation exceeding ± 3° as the abnormal threshold. If the height is detected to suddenly jump from 10.2 meters to 10.8 meters (deviation 0.6 meters) at a certain moment, the abnormality discrimination result is abnormal, and a temporary hovering instruction is immediately generated to control the unmanned aerial vehicle to maintain stable hovering at the current height.

[0092] After receiving the temporary hovering instruction, the system starts the automatic attitude compensation mechanism, adjusts the four-rotor motor output power (such as the left and right motor power difference ± 5%), corrects the yaw angle deviation (such as from the current +4° to ±1°), and after the attitude is stable, continues to perform dynamic descent control to restore to the planned descent rate and path.

[0093] By combining dynamic descent control with real-time state self-checking, using a preset abnormal threshold to trigger a temporary hovering and attitude compensation mechanism, the rapid response to sudden unstable states during descent is achieved, and compared with the traditional fixed parameter descent control, the safety and control robustness of the unmanned aerial vehicle during the descent process in complex environments are significantly improved, ensuring the smooth execution of the dynamic descent phase.

[0094] Further, the step A450 in the method provided by the embodiment of the application comprises:

[0095] A451: according to the obstacle feature, a take-off phase is planned, and a starting buffer segment, an initial segment, a middle segment and a height cut-in segment are established.

[0096] A452: control optimization of the starting buffer segment, the initial segment, the middle segment and the height cut-in segment is performed respectively, and take-off control optimization results are generated.

[0097] In one embodiment, when optimizing the take-off stage control based on the extraction result and the state self-check result, first, obstacle feature data is obtained by laser radar and visual sensor to detect that there is an obstacle with a height of 8 meters 10 meters in front of the take-off point, for example, the coordinate range is (100, 200, 0)-(120, 210, 8). According to the obstacle feature, the take-off process is divided into four stages: a starting buffer segment (0-5 meters in height), an initial segment (5-20 meters), a middle segment (20-50 meters), and a height cutting-in segment (50-100 meters of target height), and a 10% height overlap zone is set between each stage to ensure smooth transition.

[0098] For starting buffer segment control optimization, based on the extracted wind speed field data (such as ground wind speed 2m / s) and the state self-checked motor output power, the PID controller optimizes the ascending rate to 1m / s, while keeping the horizontal position deviation ≤1 meter, to avoid collision with obstacles due to air flow disturbance at the initial stage of take-off. When optimizing the initial segment control, combining the thermal convection intensity (such as 3kW / m² at a height of 5 meters) and the current battery capacity, the fuzzy control algorithm is used to adjust the motor tilt angle, the ascending rate is increased to 2m / s, and the obstacle avoidance path is planned, such as shifting 5 meters to the right, to ensure a safe distance ≥3 meters from the obstacle.

[0099] The middle segment control optimization is based on the visibility data (such as 10 meters at a height of 20 meters) and the flight control system state (pitch angle deviation ≤2°), and the particle swarm optimization algorithm is used to determine the optimal ascending angle 15°, the ascending rate is maintained at 2m / s, and the obstacle avoidance sensor array is started, with a scanning range of 360°, to update the obstacle coordinates in real time. The height cutting-in segment generates a smooth height cutting-in trajectory according to the target height (100 meters) and the remaining battery capacity through the model predictive control (MPC) algorithm, and finally reaches the target height at a rate of 0.5m / s, with a horizontal position error ≤0.5 meters.

[0100] By scientifically dividing the take-off process into four stages and combining obstacle features and environmental data for segmented control optimization, the full-process safety control from the ground to the target height is realized, which significantly improves the take-off success rate and flight stability of the unmanned aerial vehicle in complex obstacle environments compared to the traditional one-time take-off method, providing technical support for the rapid deployment of modularized shelter type fire fighting and rescue unmanned aerial vehicles in high-risk scenes such as fire scenes.

[0101] Further, step A540 in the method provided by the embodiments of the present application comprises:

[0102] A541: When any fire fighting and rescue unmanned aerial vehicle returns to the modularized shelter, the fire fighting and rescue unmanned aerial vehicle is judged for task continuation execution.

[0103] A542: If the task continues to execute the discrimination passes, the load compartment carries out the fire-fighting rescue unmanned aerial vehicle load transfer, and the flight task is reconstructed. The fire-fighting rescue unmanned aerial vehicle is controlled to take off according to the reconstructed flight task.

[0104] In the embodiments of the application, the modularized shelter is a mobile modularized facility used in conjunction with the fire-fighting rescue unmanned aerial vehicle, and has functions of take-off and landing control, state monitoring, load transfer and task management.

[0105] Optionally, after the fire-fighting rescue unmanned aerial vehicle completes the task and returns to the modularized shelter, the system first starts the task continuation execution discrimination mechanism. The state self-checking result of the unmanned aerial vehicle is read, such as the remaining battery power of 35%, the power system loss rate of 8%, and the sensor accuracy deviation of ≤5%, and the task priority (such as emergency rescue required when the fire field temperature is continuously higher than 500°C) and the environmental data (such as the current wind speed of 4m / s and the visibility of 6 meters) are comprehensively evaluated. If the power threshold is ≥30% and the key system state is normal, such as the flight control system error ≤2°, and the task does not reach the termination condition, such as the fire source is not completely extinguished, the discrimination result is passed.

[0106] If the task continues to execute the discrimination passes, the system controls the load compartment to perform the load transfer operation. For example, the spent fire extinguishing bomb load (weight 10kg) is unloaded, and a new infrared sensor module (weight 5kg) is loaded. Subsequently, according to the latest situation of the fire field, such as the new fire source coordinates (100, 200, 5) and the rescue demand, the flight task parameters are reconstructed, including the target height adjustment to 30 meters, the cruise speed setting to 15m / s, the task time length extension to 40 minutes, and the generation of new take-off and landing control instructions.

[0107] Finally, based on the reconstructed flight task, the take-off control optimization result is called, such as the initial buffer segment rising rate of 1m / s and the height cut-in segment angle of 12°, the unmanned aerial vehicle is controlled to take off from the shelter, and the rescue task is continued.

[0108] Through the real-time discrimination of the unmanned aerial vehicle state, the automatic load transfer and the task dynamic reconstruction mechanism, the continuous operation capability of the modularized shelter type fire-fighting rescue unmanned aerial vehicle in the complex fire field environment is realized. Compared with the traditional single task mode, the rescue efficiency and the task continuity are significantly improved, and efficient equipment support is provided for multiple rounds of fire-fighting rescue operations.

[0109] Further, the step A100 in the method provided in the embodiments of the application comprises:

[0110] A110: state abnormality discrimination is performed on the state self-checking result, and a state abnormality discrimination result is generated.

[0111] A120: a state abnormality warning of the fire-fighting rescue unmanned aerial vehicle is given according to the state abnormality discrimination result.

[0112] In one embodiment, after receiving the flight task and triggering the state synchronization instruction, the state data of key components such as the power system, the battery, and the flight control system are first collected in real time by the built-in sensor network of the unmanned aerial vehicle (sampling frequency 100 Hz) to form a state self-checking result. For example, the power system temperature sensor collects a value of 75°C (normal threshold ≤ 80°C), the battery remaining capacity is 30% (low capacity threshold ≤ 20%), and the flight control system pitch angle deviation is 1.5° (allowable deviation ≤ 3°), and these data constitute the original data set of the state self-checking.

[0113] Then, the state self-checking result is executed to perform abnormality discrimination. A method combining threshold comparison and trend analysis is adopted: for static parameters such as temperature and capacity, a hard threshold is set, such as battery capacity < 20% and power system temperature ≥ 80°C; for dynamic parameters such as attitude deviation, the change rate is analyzed, such as pitch angle deviation change rate > 0.5° / s. If a parameter exceeds the threshold or the change rate is abnormal, such as the battery capacity suddenly drops to 18%, a state abnormality discrimination result is generated, and the abnormal type is marked as battery capacity emergency and the abnormal timestamp is recorded.

[0114] According to the state abnormality discrimination result, a multi-modal early warning mechanism is used to report the abnormality warning. For emergency abnormalities, such as power system overheating, an audible and visual alarm (1000 Hz high-frequency buzzing of the buzzer and flashing of the red LED) is triggered and a wireless alarm is sent to the ground station; for general abnormalities, such as low capacity, a yellow warning icon is displayed on the unmanned aerial vehicle control panel and recorded to the log. For example, when the discrimination result is the pitch angle abnormality of the flight control system, the audible and visual alarm and wireless transmission are started at the same time, prompting the operator that the flight control attitude deviation is out of limit and suggesting to return.

[0115] By constructing a multi-level state parameter monitoring system and combining threshold discrimination and trend analysis algorithm, real-time abnormality detection and early warning of the key states of the unmanned aerial vehicle are realized, which significantly improves the accuracy of abnormality identification and the timeliness of early warning compared with the traditional single threshold alarm method, and provides a reliable state guarantee mechanism for the safe operation of the modularized shelter type fire extinguishing and rescue unmanned aerial vehicle in complex task scenarios.

[0116] In summary, the take-off and landing control method of the modularized shelter type fire extinguishing and rescue unmanned aerial vehicle provided by the embodiments of the application has the following technical effects:

[0117] This application triggers state synchronization instructions during the take-off and landing process of the modular box-type fire-fighting and rescue UAV to obtain state self-inspection results and perform environmental situation awareness, configures a sliding time window to extract the time series environmental data set, and then performs dynamic take-off and landing scoring judgment. If the judgment passes, the control optimization of segmented take-off and landing is performed based on the extraction results and self-inspection results, and the control optimization results of each stage are established and used for take-off and landing control management, thereby achieving precise control of the UAV's take-off and landing process, and significantly improving the safety and reliability of the modular box-type fire-fighting and rescue UAV's take-off and landing in complex environments, achieving the technical effects of precise take-off and landing risk assessment, fine segmented control and continuous mission execution of the modular box-type fire-fighting and rescue UAV in complex environments.

[0118] Example 2, as Figure 2 As shown, based on the same inventive concept as the aforementioned embodiment 1, this embodiment of the present application provides a modularized cabin-type firefighting and rescue drone take-off and landing control platform, the platform comprising:

[0119] The flight mission acquisition module 1 is used to trigger a state synchronization instruction after receiving a flight mission, and obtain a state self-test result of the fire-fighting and rescue UAV according to the state synchronization instruction.

[0120] The environmental data set construction module 2 is used to perform environmental situation awareness according to the state synchronization instruction and establish a time series environmental data set, which includes wind speed field, thermal convection intensity, visibility, and obstacle characteristics.

[0121] The take-off and landing score determination module 3 is used to configure a sliding time window to extract the time series environment data set, and then perform dynamic take-off and landing score determination based on the extraction result and the state self-check result.

[0122] The optimization result construction module 4 is used to perform segmented take-off and landing control optimization based on the extraction result and the state self-check result if the dynamic take-off and landing scoring judgment result is a passing result, and establish a segmented take-off and landing control optimization result.

[0123] The take-off and landing control execution module 5 performs take-off and landing control management by using the segmented take-off and landing control optimization result.

[0124] Furthermore, the take-off and landing score determination module 3 is used to perform the following steps:

[0125] According to the extraction result, environmental change fitting is performed, and an environmental change prediction result is established; the risk trend fitting result is subjected to critical escape triggering identification; if critical escape is triggered, the dynamic take-off and landing score discrimination is cancelled, and the segmented take-off and landing control optimization based on the extraction result and the state self-checking result is directly performed.

[0126] Further, the take-off and landing score discrimination module 3 is configured to perform the following steps:

[0127] The extraction result is divided into three levels of time characteristics; a time spectrum vector is established according to the three levels of time characteristics; the time spectrum vector is used to perform score calculation of wind speed influence characteristics, thermal convection disturbance characteristics, visible disturbance characteristics, and obstacle influence characteristics, and a first score result is established; take-off and landing score analysis is performed according to the state self-checking result, and a second score result is established; dynamic take-off and landing score discrimination is performed using the first score result and the second score result.

[0128] Further, the optimization result construction module 4 is configured to perform the following steps:

[0129] According to the extraction result and the state self-checking result, hover control optimization is performed, and hover preparation stage control optimization result is established; after path fitting of the dynamic descent segment, time sequence influence prediction of the extraction result is performed, and time sequence influence prediction result is established; according to the path fitting result, the time sequence influence prediction result, and the state self-checking result, control optimization of the dynamic descent is performed, and dynamic descent segment control optimization result is established; after the landing contact segment is subjected to fire extinguishing and rescue unmanned aerial vehicle state updating, state updated landing control optimization is started, and landing segment control optimization result is established; the hover preparation stage control optimization result, the dynamic descent segment control optimization result, and the landing segment control optimization result are output as landing control optimization result.

[0130] Further, the optimization result construction module 4 is configured to perform the following steps:

[0131] The current position coordinates of the fire extinguishing and rescue unmanned aerial vehicle and the landing platform position coordinates are obtained; according to the current position coordinates, the landing platform position coordinates, and the obstacle characteristics, a landing path planning result is established; the landing path planning result, the extraction result, and the state self-checking result are input into an adaptive segmented planning channel, and segmented planning is performed, and a hover segment, a dynamic descent segment, and a landing contact segment are established.

[0132] Further, the take-off and landing control execution module 5 is configured to perform the following steps:

[0133] The dynamic descent section control optimization result is used to control the dynamic descent of the fire-fighting and rescue unmanned aerial vehicle, and a descent state self-checking is performed to establish a descent state self-checking result; a stable abnormality discrimination of the descent state self-checking result is performed, and if the discrimination result is an abnormal result, a temporary hovering instruction is generated; after automatic attitude compensation according to the temporary hovering instruction, the dynamic descent control is continuously performed.

[0134] Further, the optimization result construction module 4 is used to perform the following steps:

[0135] According to the obstacle features, a take-off stage planning is performed to establish a starting buffer section, an initial section, a middle section and a height cut-in section; and control optimization of the starting buffer section, the initial section, the middle section and the height cut-in section is respectively performed to generate take-off control optimization results.

[0136] Further, the take-off and landing control execution module 5 is used to perform the following steps:

[0137] When any fire-fighting and rescue unmanned aerial vehicle returns to the modular shelter, a task continuation execution discrimination of the fire-fighting and rescue unmanned aerial vehicle is performed; if the task continuation execution discrimination passes, after load transfer of the fire-fighting and rescue unmanned aerial vehicle through the load cabin, a flight task is reconstructed, and the fire-fighting and rescue unmanned aerial vehicle is controlled to take off according to the reconstructed flight task.

[0138] Further, the flight task acquisition module 1 is used to perform the following steps:

[0139] The state abnormality discrimination result is used to report a state abnormality warning of the fire-fighting and rescue unmanned aerial vehicle.

[0140] The take-off and landing control platform of the modular shelter type fire-fighting and rescue unmanned aerial vehicle provided in the embodiment can perform the take-off and landing control method of the modular shelter type fire-fighting and rescue unmanned aerial vehicle provided in any embodiment of the application, has the corresponding function modules and beneficial effects of the execution method.

[0141] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy mutual differentiation, and do not limit the protection scope of the present application.

[0142] The foregoing DETAILED DESCRIPTION, including the above section titled "Detailed Description," is not to be taken as limiting the scope of the application. Various modifications, combinations, and equivalents can be apparent to those skilled in the art and can be made once the nature of the application is understood. Any modification, combination, or equivalent, which falls within the principles and the scope of the present application, is intended to be included in the present application. In some instances, the actions or steps can be performed in different order from those described herein, and still achieve desirable results. Additionally, the process depicted in the figures can not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

Claims

1. A modularized cabin-type firefighting and rescue UAV take-off and landing control method, characterized in that: The method comprises: After receiving the flight mission, trigger the status synchronization instruction and obtain the status self-test result of the fire-fighting and rescue UAV according to the status synchronization instruction; Performing environmental situation awareness according to the state synchronization instruction and establishing a time series environmental data set, wherein the time series environmental data set includes wind speed field, thermal convection intensity, visibility, and obstacle characteristics; After configuring a sliding time window to extract the time series environment data set, dynamic take-off and landing scoring is performed based on the extraction results and the status self-check results; If the dynamic take-off and landing scoring result is a passing result, performing a segmented take-off and landing control optimization based on the extraction result and the state self-check result, and establishing a segmented take-off and landing control optimization result; Performing take-off and landing control management using the segmented take-off and landing control optimization result; The step of performing dynamic take-off and landing scoring based on the extraction result and the state self-check result includes: Performing environmental change fitting based on the extraction results to establish environmental change prediction results; Using the environmental change prediction results to perform risk trend fitting, and performing critical escape trigger identification on the risk trend fitting results; If a critical escape is triggered, the dynamic takeoff and landing scoring judgment is canceled, and the control optimization of the staged takeoff and landing based on the extraction result and the state self-check result is directly performed; The step of performing dynamic take-off and landing scoring based on the extraction result and the state self-check result further includes: Dividing the extraction results into three levels of time features; Establishing a time spectrum vector based on the three-level time characteristics, and using the time spectrum vector to perform scoring calculations on wind speed impact characteristics, thermal convection disturbance characteristics, visual disturbance characteristics, and obstacle impact characteristics to establish a first scoring result; Performing a takeoff and landing scoring analysis based on the status self-check result to establish a second scoring result; The first scoring result and the second scoring result are used to perform dynamic take-off and landing scoring judgment.

2. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 1, wherein: The step of performing control optimization for staged take-off and landing based on the extraction result and the state self-check result, and establishing a staged take-off and landing control optimization result, includes: Performing hover control optimization based on the extraction results and the state self-check results, and establishing a hover preparatory stage control optimization result; After performing path fitting on the dynamic descending segment, executing a time series impact prediction of the extraction result to establish a time series impact prediction result; Performing dynamic descent control optimization based on the path fitting results, the timing impact prediction results, and the state self-check results, and establishing a dynamic descent section control optimization result; After the state of the fire-fighting and rescue drone is updated in the landing contact section, the landing control optimization after the state update is started, and the landing section control optimization result is established. The hovering preparation stage control optimization result, the dynamic descent stage control optimization result, and the landing stage control optimization result are output as the landing control optimization result.

3. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 2, wherein: Before performing hover control optimization according to the extraction result and the state self-check result, the method includes: Get the current location coordinates and landing platform coordinates of the firefighting and rescue drone; Perform landing path planning based on the current position coordinates, the landing platform position coordinates, and obstacle characteristics, and establish a landing path planning result; The landing path planning result, the extraction result, and the state self-check result are input into an adaptive segment planning channel, segment planning is performed, and a hovering segment, a dynamic descent segment, and a landing contact segment are established.

4. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 2, wherein: The method of performing take-off and landing control management by utilizing the segmented take-off and landing control optimization result includes: Using the dynamic descent segment control optimization result to perform dynamic descent control of the fire-fighting and rescue UAV, and performing a descent state self-check of the fire-fighting and rescue UAV to establish a descent state self-check result; executing a stability abnormality determination of the descent state self-test result, and generating a temporary hovering instruction if the determination result is an abnormal result; After automatic attitude compensation is performed according to the temporary hovering instruction, dynamic descent control is continued.

5. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 1, wherein: The control optimization for staged takeoff and landing based on the extraction result and the state self-check result further includes: Plan the takeoff phase based on the obstacle characteristics, and establish the initial buffer segment, initial segment, middle segment, and altitude entry segment; The control optimization of the initial buffer section, initial section, middle section and altitude cut-in section is performed respectively to generate the takeoff control optimization result.

6. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 1, wherein: After performing take-off and landing control management using the segmented take-off and landing control optimization result, the method includes: When any firefighting and rescue drone returns to the modular cabin, the firefighting and rescue drone is judged to continue its mission; If the mission is judged to continue, the payload of the fire-fighting and rescue UAV is transferred through the payload compartment, the flight mission is reconstructed, and the fire-fighting and rescue UAV is controlled to take off according to the reconstructed flight mission.

7. The method for controlling the take-off and landing of a modularized box-type firefighting and rescue UAV according to claim 1, wherein: The step of obtaining the status self-test result of the fire-fighting and rescue drone according to the status synchronization instruction includes: Performing state abnormality determination on the state self-test result to generate a state abnormality determination result; According to the abnormal state judgment result, an abnormal state warning of the fire-fighting and rescue drone is reported.

8. A modularized cabin-type firefighting and rescue UAV take-off and landing control platform, characterized by: A method for controlling the take-off and landing of a modularized modular firefighting and rescue UAV according to any one of claims 1 to 7, wherein the platform comprises: A flight mission acquisition module is used to trigger a status synchronization instruction after receiving a flight mission, and obtain a status self-test result of the fire-fighting and rescue UAV according to the status synchronization instruction; An environmental data set construction module is used to perform environmental situation awareness according to the state synchronization instruction and establish a time series environmental data set, wherein the time series environmental data set includes wind speed field, thermal convection intensity, visibility, and obstacle characteristics; A take-off and landing score determination module is configured to extract the time series environment data set through a sliding time window and then perform dynamic take-off and landing score determination based on the extraction result and the state self-check result; an optimization result building module, configured to, if the dynamic take-off and landing scoring judgment result is a passing result, perform control optimization for segmented take-off and landing based on the extraction result and the state self-check result, and establish a segmented take-off and landing control optimization result; The take-off and landing control execution module uses the segmented take-off and landing control optimization result to perform take-off and landing control management.

Images