Load weight measuring method based on rotor pulling force of unmanned aerial vehicle
By cleaning and filtering lift data based on UAV rotors, combined with aerodynamic models and adaptive filters, the problem of real-time high-precision measurement of UAV payload weight during flight was solved, achieving stable and reliable estimation in dynamic environments and supporting flight control and mission planning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HANGZHOU STAR SHUTTLE TECHNOLOGY CO LTD
- Filing Date
- 2026-02-11
- Publication Date
- 2026-05-12
AI Technical Summary
Existing methods for measuring the payload weight of unmanned aerial vehicles (UAVs) are difficult to achieve real-time, online, and high-precision measurement during flight. They are also limited by additional hardware burdens and sensitivity to environmental changes, resulting in large measurement errors and insufficient robustness, making it difficult to meet the requirements of high-precision autonomous flight control.
By acquiring the raw lift values of the UAV rotor in stable flight, data cleaning and validity verification are performed, and lift values within a reasonable range are selected. The load weight is calculated by combining the rotor aerodynamic model and real-time attitude information, and an adaptive filter is introduced to process dynamic flight conditions, thereby realizing multi-sensor data fusion and sensor health management.
It significantly improves the directness and timeliness of payload weight measurement, provides stable and reliable weight estimation in complex flight environments, supports flight control and mission planning, and enhances the system's survivability and output capabilities in dynamic environments.
Smart Images
Figure CN122020528A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a method for measuring load weight based on the rotor thrust of a UAV. Background Technology
[0002] Currently, when drones perform tasks such as transportation, surveying, and agricultural spraying, the real-time and accurate perception of their payload weight is crucial for flight safety, energy management, and mission planning. Traditional payload measurement methods mainly rely on static ground weighing before takeoff or indirect estimations such as empirical models of battery current and thrust, making it difficult to achieve real-time, online, and accurate measurement during flight.
[0003] Some existing technologies attempt to obtain weight information by installing sensors on the landing gear or analyzing power system parameters in flight. However, these methods are often limited by additional hardware burdens, complex calibration processes, or are sensitive to changes in the UAV's flight attitude, maneuvering state, and environmental conditions (such as atmospheric density). This results in large measurement errors and insufficient robustness under dynamic and complex actual flight conditions, making it difficult to meet the requirements of high-precision autonomous flight control for real-time and reliable load status feedback. Summary of the Invention
[0004] This invention provides a load weight measurement method based on the rotor thrust of a UAV to solve existing technical problems, thereby addressing the difficulty of existing technologies in meeting the requirements of high-precision autonomous flight control for real-time and reliable load status feedback.
[0005] To solve the above-mentioned technical problems, according to one aspect of the present invention, more specifically, a method for measuring load weight based on the rotor thrust of a UAV, comprising the following steps: S1. When the UAV is in a stable flight state, acquire multiple raw lift measurements generated by each rotor within a set sampling period to form a set of raw lift values for each rotor. S2. Perform data cleaning and validity verification on each set of raw lift values, remove outliers, and calculate the preliminary lift value for each rotor. S3. Calculate the global lift reference value based on the initial lift values of all rotors, and set a reasonable screening range based on this reference value; S4. Select the preliminary lift values that fall within the reasonable selection range as the effective rotor lift values; S5. Based on the effective rotor lift value, the real-time attitude information of the UAV, and the aerodynamic model of the rotor, calculate the total lift of the UAV and the current load weight. Among them, load weight Calculated using the following core formula: ; In the above formula, Indicates the number of effective rotors; Indicates the first One effective rotor lift value; Indicates the first The thrust coefficient compensation factor for each rotor; Represents gravitational acceleration; and These represent the real-time pitch and roll angles of the drone, respectively. This indicates the drone's baseline weight.
[0006] Furthermore, data cleaning and validity verification are performed on each set of raw lift values, including: Determine whether the variance of all values in the set is less than a preset minimum variance threshold. (1) If so, the corresponding sensor data is determined to be invalid and the set is discarded; (2) If not, then filter out the sets whose values are sorted by size first. and after Data points, among which This is a preset ratio; The arithmetic mean of the remaining data points after filtering is calculated and used as the initial lift value of the rotor.
[0007] Furthermore, set a reasonable filtering range, specifically: Calculate the arithmetic mean of all initial lift values. and standard deviation ; Set the lower limit of the filter range to The upper limit is ,in The coefficients are preset based on the confidence level.
[0008] Furthermore, the tensile coefficient compensation factor The method for determining it is as follows: (1) Obtain the atmospheric density at the current flight altitude Rotor speed and the characteristic parameters of the rotor blades; (2) Obtain the results online or by looking up a table using the following model: ; in, This is the rotor thrust coefficient; The rotor radius; This represents the theoretical pulling force value under the current rotational speed and standard atmospheric conditions.
[0009] Furthermore, the statement that the drone is in a stable flight state means: The rates of change of the drone's pitch, roll, and yaw angles are all below their respective preset thresholds, and the fluctuation range of the acceleration of each axis in the body coordinate system is within the preset range.
[0010] Furthermore, the load weight measurement method also includes an adaptive dynamic estimation step for the load weight: (1) Execute the load weight measurement method based on UAV rotor thrust in multiple consecutive measurement cycles to obtain the load weight time series. ; (2) An adaptive filter based on a motion model is used to process the time series in order to suppress high-frequency noise and track the gradual change trend of the load; The state update process of the adaptive filter includes a process noise covariance matrix that is adjusted according to the flight state (acceleration, deceleration, hovering).
[0011] Furthermore, the load weight measurement method also includes sensor health management and data fusion steps: (1) Record the reliability score of historical data for each lift measurement channel; (2) In the current calculation, the effective rotor lift value is... Assign a weight This weight is positively correlated with the reliability score of its channel.
[0012] Furthermore, based on the effective rotor lift value and weight allocation The total lift is calculated using a weighted summation method: ; In the above formula, This indicates the magnitude of the total lift.
[0013] This invention provides a method for measuring the load weight based on the rotor thrust of a UAV. Compared with existing technologies, this method achieves the following advantages: 1. This invention obtains load information by directly calculating the rotor resultant force during flight, significantly improving the directness and timeliness of measurement. Its core algorithm can effectively fuse multi-sensor data and dynamically compensate for the influence of factors such as flight attitude and aerodynamic changes during calculation. Therefore, it can provide stable and reliable weight estimation when the UAV is hovering, cruising, or even performing certain maneuvers, providing key data support for flight control, endurance assessment, and mission planning.
[0014] 2. This invention introduces a multi-fault-tolerant mechanism including data cleaning, validity screening, and sensor health management. The system can automatically identify and eliminate abnormal data caused by momentary sensor malfunctions, airflow disturbances, or rotor anomalies (such as blade icing or minor damage), ensuring the reliability of core computational data. Even if some measurement channel data is temporarily unavailable, the system can still continue operating based on the remaining valid data through weighted fusion, greatly enhancing the survivability and continuous output capability of the entire measurement system in complex and unpredictable flight environments.
[0015] 3. This invention cleverly addresses the challenges posed by dynamic flight conditions by integrating an adaptive dynamic estimation step. When the UAV accelerates, decelerates, or turns, the airframe acceleration interferes with the instantaneous calculations based on static equilibrium. This method uses a model-based filter (such as a Kalman filter) to optimize the continuous weight estimation sequence. By adaptively adjusting process noise, it effectively smooths measurement fluctuations caused by maneuvering overload, distinguishes between actual load changes and apparent changes caused by motion, and outputs a smooth, low-hysteresis, high-quality estimate that tracks the gradual trend of actual load changes, meeting the stringent requirements of closed-loop flight control for the stability of state feedback signals. Attached Figure Description
[0016] Figure 1 This is a flowchart of the present invention. Detailed Implementation
[0017] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0018] Example 1 like Figure 1 As shown, this embodiment takes a quadcopter drone in a stable hovering state as an example to explain in detail the implementation process of the method described in this application.
[0019] Step S1, Data Acquisition The drone's flight controller initiates a stable hovering state. At this point, each of the four onboard rotor motors (e.g., brushless motors) is connected to a sensor capable of real-time measurement of rotor thrust (e.g., high-sensitivity strain gauges or miniature piezoelectric sensors integrated into the motor base). The controller continuously acquires data for one second at a frequency of 100Hz through the data channel of each sensor, obtaining a set of 100 raw thrust values for each rotor, denoted as . , , , Meanwhile, the inertial measurement unit reports the current attitude as: pitch angle. = 1.5°, roll angle = -0.8°.
[0020] Step S2: Data Cleaning and Preliminary Calculation Each raw dataset is processed. Taking rotor 1 as an example, among its 100 raw values, there are a few spikes caused by instantaneous airflow disturbances. Therefore: 1. Calculate the set variance. If it is much greater than the preset minimum variance value, the sensor data is deemed valid.
[0021] 2. Filter out the 5% of data points with the largest and smallest effective values (i.e., filter out the 5 largest and 5 smallest values).
[0022] 3. Calculate the arithmetic mean of the remaining 90 data points to obtain the initial lift value of rotor 1. Similarly, we obtain , , .visible Significantly high.
[0023] Step S3: Validity Screening Calculate the arithmetic mean of all initial lift values. Standard deviation Take the coefficient. The filtering range is then: lower limit upper limit .
[0024] , , All values falling within this range are considered valid rotor lift values.
[0025] If the value exceeds the upper limit, it is considered an outlier and removed (this may correspond to a temporary sensor malfunction or minor foreign object entanglement on the blade). At this point, the effective number of rotors... .
[0026] Step S4: Core Load Calculation Calculations are performed based on the core formula described above. Given: drone baseline weight .
[0027] gravitational acceleration .
[0028] Attitude compensation factor: The value is close to 1, indicating that the hovering posture is nearly horizontal.
[0029] Tensile coefficient compensation factor Since it is hovering near sea level and the rotational speeds of all normal rotors are stable, based on the model or pre-calibration table, take... (Compensation for actual air density and motor efficiency).
[0030] Will , , and corresponding Substitute into the formula: ; ; The calculated current load weight is approximately 0.30 kg. The system outputs this result, which can be used to determine if there is overload or for fine-tuning of flight control parameters.
[0031] Step S5: Sensor Health Update The data channel corresponding to rotor 3 was determined to have an output anomaly in this test. The system lowered the reliability score of this channel, and if its data returns to normal in the next test cycle, a lower weight will be assigned to its data. If the abnormality persists, mark it as needing to be checked.
[0032] Example 2 like Figure 1 As shown, this embodiment demonstrates how the method of this application maintains the stability and accuracy of load measurement through dynamic estimation when the UAV performs climb and turn maneuvers.
[0033] Step S1: Scene and data initialization.
[0034] With a known payload of 0.5 kg, the drone began performing a "climb + right roll turn" maneuver. The controller continuously executed the weight measurement process at 50 ms intervals.
[0035] Mule S2, Single-cycle Data Processing and Challenges During a specific period, the drone's attitude angle changes significantly: =15° (nose tilted up) =10° (right roll). The inertial measurement unit shows some fluctuations in angular velocity and acceleration.
[0036] 1. After data collection and cleaning, four preliminary lift values were obtained. However, one of the values (rotor 2) was slightly affected by the complex airflow outside the turn, and was retained after screening, but its reliability score was temporarily lowered.
[0037] 2. Directly apply the core formula mentioned above to calculate the instantaneous load weight. Due to the large attitude angle, The attitude compensation effect is significant. Simultaneously, based on the current rotational speed and estimated air density (slightly reduced due to climb), the attitude compensation of each rotor was dynamically calculated using a model. .
[0038] 3. However, due to the vertical acceleration (climbing acceleration) of the organism, the static formula of the simplified sheep will introduce instantaneous errors. The period directly calculated... There was a brief fluctuation, showing a weight of 0.53 kg.
[0039] Step S3: Adaptive Dynamic Estimation The system calls an adaptive filter (such as an extended Kalman filter) to process the continuous weight estimation sequence.
[0040] State model: The filter will load weight As a slowly changing state quantity (assuming the load does not change abruptly).
[0041] Process noise adaptation: Due to the current maneuvering flight state (unstable hovering), the system automatically increases the process noise covariance matrix. The corresponding elements in the [database]. This is equivalent to telling the filter: "The uncertainty of the current model has increased, please trust the latest observations more."
[0042] Filter update: The filter is based on the increase The system updates the state using an observation model that includes acceleration information. It can partially distinguish between the "apparent weight changes" caused by body acceleration and the actual load changes. After filtering and smoothing, it outputs the final load weight estimate for this cycle.
[0043] Step S4: Weighted Data Fusion When calculating the total lift, the system uses the real-time reliability score of each channel. Assign weights. For example, the updated scores for each channel in this cycle are as follows: , , , Then the weight .
[0044] Before substituting into the core formula, the "total lift" used for calculation... These are the effective lift values. The weighted sum, rather than a simple averaging, is used. This reduces the impact of significant data fluctuations on the final result, further improving the robustness of the overall estimation under complex flight conditions.
[0045] Step S5: Continuous Output and Application The system continuously outputs the payload weight value at a high frequency (e.g., 20Hz), after adaptive dynamic estimation and weighted fusion processing. This data stream is very smooth and can stably reflect the actual 0.5kg payload without being disturbed by violent maneuvers. The flight controller can use this stable weight information to optimize the maneuver control law in real time, ensuring flight safety and performance.
[0046] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.
Claims
1. A method for measuring load weight based on the rotor thrust of a UAV, characterized in that, Includes the following steps: S1. When the UAV is in a stable flight state, acquire multiple raw lift measurements generated by each rotor within a set sampling period to form a set of raw lift values for each rotor. S2. Perform data cleaning and validity verification on each set of raw lift values, remove outliers, and calculate the preliminary lift value for each rotor. S3. Calculate the global lift reference value based on the initial lift values of all rotors, and set a reasonable screening range based on this reference value; S4. Select the preliminary lift values that fall within the reasonable selection range as the effective rotor lift values; S5. Based on the effective rotor lift value, the real-time attitude information of the UAV, and the aerodynamic model of the rotor, calculate the total lift of the UAV and the current load weight. Among them, load weight Calculated using the following core formula: ； In the above formula, Indicates the number of effective rotors; Indicates the first One effective rotor lift value; Indicates the first The thrust coefficient compensation factor for each rotor; Represents gravitational acceleration; and These represent the real-time pitch and roll angles of the drone, respectively. This indicates the drone's baseline weight.
2. The load weight measurement method based on UAV rotor thrust according to claim 1, characterized in that: Data cleaning and validity verification were performed on each set of raw lift values, including: Determine whether the variance of all values in the set is less than a preset minimum variance threshold. (1) If so, the corresponding sensor data is determined to be invalid and the set is discarded; (2) If not, then filter out the sets whose values are sorted by size first. and after Data points, among which This is a preset ratio; The arithmetic mean of the remaining data points after filtering is calculated and used as the initial lift value of the rotor.
3. The load weight measurement method based on UAV rotor thrust according to claim 1, characterized in that: Set a reasonable filtering range, specifically as follows: Calculate the arithmetic mean of all initial lift values. and standard deviation ; Set the lower limit of the filter range to The upper limit is ,in The coefficients are preset based on the confidence level.
4. The load weight measurement method based on UAV rotor thrust according to claim 1, characterized in that: The tensile coefficient compensation factor The method for determining it is as follows: (1) Obtain the atmospheric density at the current flight altitude Rotor speed and the characteristic parameters of the rotor blades; (2) Obtain the results online or by looking up a table using the following model: ； in, This is the rotor thrust coefficient; The rotor radius; This represents the theoretical pulling force value under the current rotational speed and standard atmospheric conditions.
5. The load weight measurement method based on UAV rotor thrust according to claim 1, characterized in that: The drone being in a stable flight state means: The rates of change of the drone's pitch, roll, and yaw angles are all below their respective preset thresholds, and the fluctuation range of the acceleration of each axis in the body coordinate system is within the preset range.
6. The load weight measurement method based on UAV rotor thrust according to claim 1, characterized in that: The load weight measurement method further includes an adaptive dynamic estimation step for the load weight: (1) Execute the load weight measurement method based on UAV rotor thrust in multiple consecutive measurement cycles to obtain the load weight time series. ; (2) An adaptive filter based on a motion model is used to process the time series in order to suppress high-frequency noise and track the gradual change trend of the load; The state update process of the adaptive filter includes a process noise covariance matrix that is adjusted according to the flight state.
7. The load weight measurement method based on UAV rotor thrust according to claim 6, characterized in that: The load weight measurement method also includes sensor health management and data fusion steps: (1) Record the reliability score of historical data for each lift measurement channel; (2) In the current calculation, the effective rotor lift value is... Assign a weight This weight is positively correlated with the reliability score of its channel.
8. The load weight measurement method based on UAV rotor thrust according to claim 7, characterized in that: Based on the lift value of each effective rotor and weight allocation The total lift is calculated using a weighted summation method: ； In the above formula, This indicates the magnitude of the total lift.