Equipment attitude data analysis method and system based on adaptive weighting

Abstract

The invention discloses an equipment attitude data analysis method and system based on adaptive weighting, and belongs to the technical field of unmanned aerial vehicle attitude analysis. The system comprises a flight sensing module, a scene analysis module, a parameter setting module and a dynamic adjustment module. The flight sensing module is used for collecting historical records of the unmanned aerial vehicle and environment data and operation data in the flight process. The scene analysis module builds a virtual scene through the environment data and the operation data, and maps and simulates and divides a temperature difference area in real time; the parameter setting module sets influence objects according to the temperature difference area, and analyzes historical records to set weight constraint coefficients for the influence objects; and the dynamic adjustment module is used for calculating the stability index in real time and dynamically adjusting the weight of each influence object according to the stability index and the weight constraint coefficient. The weight of the sensor is dynamically adjusted through a self-adaptive weighting mechanism, and the accuracy and flight stability of attitude data of the unmanned aerial vehicle are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle attitude analysis, in particular to a device attitude data analysis method and system based on adaptive weighting. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicle technology, unmanned aerial vehicles are increasingly widely used in environmental monitoring, logistics transportation, agricultural plant protection and other fields. Stable flight and high-precision control of unmanned aerial vehicles put stringent requirements on attitude estimation. The accuracy of attitude data directly affects the flight safety and task execution efficiency of unmanned aerial vehicles.

[0003] At present, in the aspect of unmanned aerial vehicle attitude data fusion, a fixed weight data fusion strategy is usually adopted, which can easily lead to an increase in estimation error and affect the control effect of the unmanned aerial vehicle. There are some drawbacks, for example: 1. In high-temperature or low-temperature environments, there is a significant temperature difference between the windward surface and the leeward surface of the unmanned aerial vehicle. In a high-temperature scenario, the windward surface sensor has a higher data accuracy than the leeward surface due to the lower temperature, while in a low-temperature scenario, the windward surface has a lower accuracy than the leeward surface due to the impact of cold air. 2. Even at the same temperature, different positions of the same type of sensor will have performance differences due to temperature sensitivity, and the existing technology cannot quantify individual differences based on historical data. 3. In the existing technology, in the scenarios of motor harmonic distortion and loss of control, the fixed weight fusion algorithm is still mechanically executed, and the sensor distortion is not suppressed by dynamic weight reduction. Therefore, at present, a more intelligent and efficient unmanned aerial vehicle attitude data analysis technology solution is needed to solve the above problems. SUMMARY

[0004] The purpose of the present application is to provide a device attitude data analysis method and system based on adaptive weighting to solve the problems raised in the background.

[0005] To solve the above technical problems, the present application provides a device attitude data analysis method based on adaptive weighting, comprising:

[0006] S100, collecting historical records of the unmanned aerial vehicle, as well as environmental data and running data during flight.

[0007] The historical records include environmental data and running data at different historical times.

[0008] The environmental data includes the wind speed and temperature around the unmanned aerial vehicle. The running data includes a three-dimensional model of the unmanned aerial vehicle, GPS parameters, attitude parameters, temperature parameters and motor current.

[0009] The GPS parameters include spatial position, flight direction and flight speed; the attitude parameters are collected in real time by the inertial sensor and specifically include acceleration, angular velocity and angle. The temperature parameters include temperatures of different positions inside the unmanned aerial vehicle. A three-dimensional temperature field is generated by combining the temperature parameters with the three-dimensional model.

[0010] The operation data are used for describing environmental information, flight state and unmanned aerial vehicle performance, and are the basis for the unmanned aerial vehicle to perceive the environment and realize autonomous flight.

[0011] A multi-source data foundation is constructed to provide environmental, operation and historical data support for virtual scene building. The three-dimensional temperature field is dynamically mapped to the heat distribution of the unmanned aerial vehicle, and the physical entity is converted into a digital model that can be quantitatively analyzed.

[0012] S200, a virtual scene is built by using environmental data and operation data to map and simulate the temperature difference area in real time. Specifically, the following steps are included:

[0013] S201, a virtual scene is built, and a three-dimensional model of the unmanned aerial vehicle is loaded in the virtual scene. The three-dimensional model is adjusted in real time in the three-dimensional space in terms of pitch, roll and yaw angle states according to the attitude parameters.

[0014] S202, environmental data and operation data at the same time in the historical record are obtained, and the relationship between wind speed, temperature, flight direction and flight speed and the three-dimensional temperature field is analyzed, and a training set is constructed.

[0015] S203, the training set is input into the physical model for training. The trained physical model simulates and analyzes the three-dimensional temperature field TF fut at future time t in the virtual scene according to the current three-dimensional temperature field TF now , environmental data and operation data.

[0016] S204, the temperature difference at the same position in the three-dimensional temperature field TF now and the three-dimensional temperature field TF fut is compared, and the area with a temperature difference greater than a threshold value is divided as a temperature difference area.

[0017] The dynamic prediction of temperature change and the positioning of risk area are realized, the physical environment is mapped into a calculable virtual space, and a space basis is provided for sensor weight adjustment.

[0018] S300, an influence object is set according to the temperature difference area, and a weight constraint coefficient is set for each influence object according to the historical record. Specifically, the following steps are included:

[0019] S301, the temperature difference area is mapped in the three-dimensional model in the virtual scene, and the inertial sensor in the temperature difference area is taken as the influence object. All inertial sensors collecting the same attitude parameters are combined into a virtual sensor.

[0020] The virtual sensor is virtually generated by an algorithm, specifically a same type of posture parameter set collected by all inertia sensors, the number of virtual sensors is the same as the number of posture parameter types, and different virtual sensors correspond to different posture parameter types.

[0021] The inertia sensor actually exists in reality and can simultaneously collect multiple types of posture parameters. Each inertia sensor can exist in multiple virtual sensors, specifically according to the number of collected posture parameter types.

[0022] S302, set the three-dimensional temperature field TF fut of each influence object as a predicted temperature, and set a reference temperature t ref Combine the posture parameters and temperature parameters in the historical record to calculate the fluctuation index of each influence object. Specifically, it includes:

[0023] S3021, establish a three-dimensional temperature field according to the temperature parameters in the historical record, and analyze the temperature change of the influence object AO VS located in the virtual sensor VS.

[0024] S3022, obtain the predicted temperature t VS of the influence object AO pre , mark the historical time period with a temperature equal to the predicted temperature t rpe and a duration greater than the threshold, and uniformly set N time points in each historical time period.

[0025] S3023, set a reference object for each historical time period, and analyze the posture parameter difference between the influence object AO VS and the reference object at each time point, and calculate the standard deviation of all time point differences as the fluctuation coefficient of the corresponding historical time period.

[0026] When analyzing each time point in the historical time period, the temperature and posture parameters of each inertia sensor in the virtual sensor VS are substituted into the formula to calculate the abnormal coefficient YX of each inertia sensor:

[0027]

[0028] In the formula, a is a constant, ZTC m is the posture parameter of the inertia sensor at the mth time point, t i is the temperature of the inertia sensor at the ith time point; and the inertia sensor with the smallest abnormal coefficient is selected as the reference object.

[0029] Log term: capture sensor output mutation through accumulation of continuous difference of posture parameters.

[0030] Temperature deviation term: calculate the average relative deviation of historical temperature and reference temperature, reflecting the stability of the sensor affected by temperature.

[0031] The sensor with the best temperature adaptability is selected as the reference object to provide a reliable benchmark for the fluctuation index calculation.

[0032] When the reference object is the impact object, the fluctuation coefficient of the corresponding historical time period is zero.

[0033] By calculating the abnormal coefficient, the inertial sensor with normal fluctuation amplitude of the collected parameters and temperature close to the reference temperature in the corresponding historical time period is selected, and the reference object is used as the comparison reference of other impact objects to analyze the abnormal fluctuation degree of the data collection of each impact object.

[0034] S3024, the average value of the fluctuation coefficients of all historical time periods is taken as the fluctuation index of the impact object AO VS , and the fluctuation index of each impact object in the virtual sensor VS is calculated respectively.

[0035] S3025, by analogy, the fluctuation index of each impact object in each virtual sensor is calculated.

[0036] S303, get the temperature t now in the environmental data, set the basic coefficient WC0 and the adjustment coefficient WC ad . According to the fluctuation index, set the weight constraint coefficient for each inertial sensor in the virtual sensor. Specifically:

[0037] Calculate the average value of the fluctuation indexes of all impact objects in the virtual sensor FLU ave , and substitute it into the formula to calculate the weight constraint coefficient QZ of each impact object:

[0038]

[0039] In the formula, k, h, c, s are constants respectively, and FLU is the fluctuation index of the impact object. The weight constraint coefficient of the inertial sensor of other non-impact objects in the virtual sensor is set as WC0.

[0040] The constants c and s are the center offset of the modified Sigmoid function, and the constants k and h are used to control the parameters of the slope of the Sigmoid function. The larger the value, the steeper the function curve.

[0041] Quantify the reliability of the sensor in the temperature abnormal area, generate adaptive weight constraints through historical fluctuation analysis, and ensure the data reliability in high temperature environment.

[0042] S400, calculate the stability index in real time, and dynamically adjust the weight of each impact object according to the stability index and the weight constraint coefficient. Specifically, it includes:

[0043] S401, analyze the interval length t between the future time t and the current time dur , calculate the difference between the weight constraint coefficient of each inertial sensor in the virtual sensor and the base coefficient WC0, and then divide it by the interval length t respectively dur to obtain the adjustment speed.

[0044] When the weight constraint coefficient is greater than the base coefficient WC0, the adjustment speed is positive, and the interval length t dur is gradually increased to adjust the weight.

[0045] When the weight constraint coefficient is less than the base coefficient WC0, the adjustment speed is negative, and the interval length t dur is gradually reduced to adjust the weight.

[0046] When the weight constraint coefficient is equal to the base coefficient WC0, the adjustment speed is zero, and the interval length t dur does not change the weight.

[0047] S402, adjust the weight coefficient of each inertial sensor in the virtual sensor according to the adjustment speed, and calculate the comprehensive attitude parameter of the corresponding virtual sensor by combining the latest weight coefficient and attitude parameter using the weighted algorithm.

[0048] S403, calculate the stability index of the unmanned aerial vehicle in real time according to the running data, and increase the adjustment speed of each inertial sensor in the virtual sensor when the stability index is less than the threshold value, until the stability index is not less than the threshold value.

[0049] The calculation formula of the stability index STI is:

[0050]

[0051] In the formula, γ is the attenuation index, is the heading angle deviation, σgp is the standard deviation of the flight speed vector, I1 is the fundamental wave current amplitude of the motor, and I u is the u-th harmonic current amplitude.

[0052] γ is an empirical coefficient for adjusting the influence weight of THD on the stability index. When γ is large, a slight increase in THD will cause the stability index to drop sharply. When γ is small, higher THD is allowed. The stability standard is dynamically adjusted according to different tasks.

[0053] THD represents the proportion of harmonic components in the total current of the motor driving current, reflecting the motor torque ripple and mechanical vibration intensity. When THD increases, the motor vibration intensifies, and the unmanned aerial vehicle body has a resonance risk, resulting in poor power stability.

[0054] THD<15% indicates normal, and THD>25% indicates stability risk.

[0055] σgp is used to describe the jitter degree of the flight trajectory of the unmanned aerial vehicle, and the unit is m / s. The smaller the value of σgp is, the smaller the speed fluctuation is, which means that the flight is stable. The larger the value of σgp is, the greater the trajectory jitter is, and the control performance of the unmanned aerial vehicle is reduced.

[0056] The difference between the GPS heading angle and the command heading angle is represented, and the sensor fusion consistency is reflected. The GPS heading angle is the heading angle calculated by the speed vector, and the command heading angle is the heading angle analyzed by the command signal. The angle deviation is converted into a stable index influence weight.

[0057] The stable index couples the current harmonic and the GPS heading deviation to evaluate the stability. Specifically, the normalized score of the stability of the power, the trajectory and the heading is integrated, and the multi-source heterogeneous data is converted into a single decision index.

[0058] In the interval duration, the weight coefficient is always between the weight constraint coefficient and the basic coefficient WC0.

[0059] The weight adjustment is coupled with the flight stability in real time, the unstable state is quickly responded through multi-dimensional indexes, and the data fusion accuracy is optimized.

[0060] The physical environment, historical fluctuation and real-time state are converted into a calculable parameter system through mathematical modeling, so as to provide a quantitative decision basis for the attitude control of the unmanned aerial vehicle in a complex environment.

[0061] The application also provides an equipment attitude data analysis system based on adaptive weighting, which comprises a flight perception module, a scene analysis module, a parameter setting module and a dynamic adjustment module.

[0062] The flight perception module is used to collect the historical records of the unmanned aerial vehicle, and the environmental data and operation data in the flight process.

[0063] The historical records of the unmanned aerial vehicle and real-time flight data are collected. A three-dimensional temperature field is generated by combining the temperature parameters with the three-dimensional model, and the environmental information, flight state and unmanned aerial vehicle performance data are integrated.

[0064] A multi-source heterogeneous data base is constructed to provide complete input for virtual scene building and temperature difference analysis, and to support the data requirements of the subsequent adaptive weighting algorithm.

[0065] The scene analysis module builds a virtual scene through environmental data and operation data, and maps and simulates the temperature difference area in real time.

[0066] The three-dimensional model of the unmanned aerial vehicle is loaded in the virtual scene, the physical model is trained in combination with the historical data, and the future three-dimensional temperature field is predicted. By comparing the current and predicted temperature fields, the area with a temperature difference value exceeding a threshold is divided into a "temperature difference area".

[0067] The dynamic simulation of temperature change and risk area identification map the physical environment to a virtual model for quantifiable analysis, providing space for sensor weight adjustment.

[0068] The parameter setting module is used to set the influence object according to the temperature difference area, and analyze the historical record to set the weight constraint coefficient for each influence object.

[0069] The inertial sensor in the temperature difference area is marked as an "influence object", and is grouped into a "virtual sensor" according to the attitude parameter type. The fluctuation index of each influence object is calculated based on historical temperature and attitude data, combined with the current environment temperature and reference temperature, and the weight constraint coefficient is dynamically generated using the Sigmoid function.

[0070] By quantifying the fluctuation of sensor data, the weight constraint coefficient is adaptively allocated, ensuring that the sensor weight in the temperature abnormal area can be dynamically adjusted with the change of the environment, and the data reliability is improved.

[0071] The dynamic adjustment module is used to calculate the stability index in real time, and dynamically adjust the weight of each influence object according to the stability index and the weight constraint coefficient.

[0072] The adjustment speed is calculated according to the difference between the weight constraint coefficient and the basic coefficient, and the weight of each sensor in the virtual sensor is updated at regular intervals. Combined with the real-time calculation of the stability index, the weight adjustment is dynamically accelerated or decelerated.

[0073] The weight adjustment is linked in real time with the flight stability, and the multi-dimensional index is used to quickly respond to the unstable state, optimize the sensor data fusion effect, and ensure the control accuracy and flight safety of the unmanned aerial vehicle.

[0074] The flight perception module provides a multi-source data basis; the scene analysis module realizes virtual positioning of the temperature risk area; the parameter setting module generates adaptive weight constraints through historical fluctuation analysis; and the dynamic adjustment module couples stability feedback to optimize weight distribution in real time.

[0075] The attitude data reliability and control stability of the unmanned aerial vehicle in complex environments are improved, and it is especially suitable for high-risk scenarios such as temperature mutation.

[0076] Compared with the prior art, the beneficial effects achieved by the present application are:

[0077] Adaptive environmental perception advantage: by constructing a three-dimensional temperature field and predicting the temperature difference area, the system can actively identify temperature sensitive areas (such as the position of the inertial sensor), which significantly improves the adaptability to complex thermal environments compared with traditional static sensor weight distribution.

[0078] Dynamic weight optimization advantage: Based on volatility index and real-time temperature data, the weight constraint coefficient of each sensor is dynamically calculated using the Sigmoid function, realizing the smooth adjustment of the weight with the change of the environment. This mechanism avoids the error accumulation problem caused by fixed weight in the prior art, and improves the accuracy of the attitude data when the temperature difference changes sharply.

[0079] Multi-source data fusion advantage: The same type of sensor data is aggregated and processed through virtual sensors, and the comprehensive attitude parameters are output through a weighting algorithm. This design solves the data coordination problem of heterogeneous sensors (such as temperature, acceleration, and angular velocity), and greatly improves the anti-interference ability and data robustness compared with the single sensor dependent scheme.

[0080] Stability closed-loop control advantage: The system state is controlled in real time through the stability index (coupling current harmonic, heading deviation, and speed fluctuation), and the weight adjustment process is dynamically accelerated or decelerated. This closed-loop mechanism enables the system to actively intervene in sensor weight distribution when flight stability decreases, significantly enhancing flight safety margin compared with traditional open-loop control.

[0081] Systematic decision advantage: The environmental perception (temperature difference zone), parameter setting (weight constraint), and dynamic adjustment (stability index feedback) are integrated into a modular architecture, forming a complete closed loop from data acquisition to control decision. Compared with fragmented solutions, this integrated design improves system response speed and decision coordination.

[0082] The prior art cannot perceive the temperature difference effect of the windward / backwind surface, ignores the individual temperature sensitivity difference of the sensor, and separates weight distribution and stability control, forming a vicious cycle of "environmental misjudgment→weight mismatch→control instability". The present application realizes three-dimensional self-adaptation of environment-individual-state through dynamic positioning of temperature difference zone, personalized weight constraint driven by volatility index, and real-time linkage acceleration of stability index, solving the attitude control problem in high / low temperature complex scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0083] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, together with the embodiments of the application, to explain the application, and do not constitute a limitation on the application. In the drawings:

[0084] Figure 1 is a flowchart of the device attitude data analysis method based on adaptive weighting of the present application;

[0085] Figure 2 is a structural schematic diagram of the device attitude data analysis system based on adaptive weighting of the present application. DETAILED DESCRIPTION

[0086] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0087] With reference to Figure 1 The present application provides an adaptive weighting-based device posture data analysis method, comprising:

[0088] S100, collecting historical records of the unmanned aerial vehicle, and environmental data and operation data in the flight process.

[0089] The historical records include environmental data and operation data at different historical times.

[0090] The environmental data includes wind speed and temperature around the unmanned aerial vehicle. The operation data includes a three-dimensional model of the unmanned aerial vehicle, GPS parameters, attitude parameters, temperature parameters, and motor current.

[0091] The GPS parameters include spatial position, flight direction, and flight speed; the attitude parameters are collected by an inertial sensor in real time, and specifically include acceleration, angular velocity, and angle. The temperature parameters include the temperature at different positions inside the unmanned aerial vehicle. A three-dimensional temperature field is generated by combining the temperature parameters with the three-dimensional model.

[0092] The operation data is used to describe environmental information, flight state, and unmanned aerial vehicle performance, and is the basis for the unmanned aerial vehicle to perceive the environment and realize autonomous flight.

[0093] A multi-source data base is constructed to provide environmental, operation, and historical data support for virtual scene building. A three-dimensional temperature field is used to dynamically map the thermal distribution of the unmanned aerial vehicle, and physical entities are converted into quantifiable digital models.

[0094] S200, a virtual scene is built by using the environmental data and operation data to map and simulate the division of temperature difference zones in real time. Specifically, it includes:

[0095] S201, a virtual scene is built, a three-dimensional model of the unmanned aerial vehicle is loaded in the virtual scene, and the three-dimensional model is adjusted in real time according to the attitude parameters to adjust the angle state of pitch, roll, and yaw in the three-dimensional space.

[0096] S202, environmental data and operation data at the same time in the historical records are obtained, the relationship between wind speed, temperature, flight direction, flight speed, and the three-dimensional temperature field is analyzed, and a training set is constructed.

[0097] S203, the training set is input into a physical model for training, and the trained physical model is used to simulate the temperature difference zones according to the current three-dimensional temperature field TF nowand environmental data and operation data, simulate a three-dimensional temperature field TF at future time t in a virtual scene fut .

[0098] S204, compare the three-dimensional temperature field TF now and the three-dimensional temperature field TF fut at the same position, divide the area where the temperature difference is greater than the threshold value as the temperature difference area.

[0099] Realize dynamic prediction of temperature change and positioning of risk area, map the physical environment to a calculable virtual space, and provide a space basis for sensor weight adjustment.

[0100] S300, set the influence object according to the temperature difference area, and set the weight constraint coefficient for each influence object according to the historical record. Specifically, it includes:

[0101] S301, map the temperature difference area in the three-dimensional model in the virtual scene, and take the inertial sensor in the temperature difference area as the influence object. All inertial sensors collecting the same posture parameter are combined into a virtual sensor.

[0102] The virtual sensor is generated by algorithm, specifically a set of the same type of posture parameters collected by all inertial sensors. The number of virtual sensors is the same as the number of posture parameter types, and different virtual sensors correspond to different posture parameter types.

[0103] The inertial sensor actually exists in reality and can collect multiple types of posture parameters at the same time. Each inertial sensor can exist in multiple virtual sensors, specifically according to the number of collected posture parameter types.

[0104] S302, set the temperature at the position of each influence object in the three-dimensional temperature field TF fut as the predicted temperature, set the reference temperature t ref , and calculate the fluctuation index of each influence object combined with the posture parameter and temperature parameter in the historical record. Specifically, it includes:

[0105] S3021, establish a three-dimensional temperature field according to the temperature parameter in the historical record, and analyze the temperature change at the position of the influence object AO VS in the virtual sensor VS.

[0106] S3022, obtain the predicted temperature t VS of the influence object AO pre , mark the historical time period where the temperature is equal to the predicted temperature t pre and the duration is greater than the threshold value, and set N time points uniformly in each historical time period.

[0107] S3023, set a reference object for each historical time period, analyze the influence object AO at each time point VS The difference between the posture parameters of the reference object and the posture parameters of the influence object AO at each time point is calculated, and the standard deviation of the difference at all time points is calculated as the fluctuation coefficient of the corresponding historical time period.

[0108] When analyzing each time point in the historical time period, the temperature and posture parameters of each inertial sensor in the virtual sensor VS are substituted into the formula to calculate the abnormality coefficient YX of each inertial sensor:

[0109]

[0110] In the formula, a is a constant, ZTC m is the posture parameter of the inertial sensor at the mth time point, t i is the temperature of the inertial sensor at the ith time point; the inertial sensor with the smallest abnormality coefficient is selected as the reference object.

[0111] Log term: Through the accumulation of continuous difference of posture parameters, the sudden change of sensor output (such as acceleration jump caused by sudden temperature rise) is captured.

[0112] Temperature deviation term: Calculate the relative deviation mean of historical temperature and reference temperature, reflect the stability of sensor affected by temperature.

[0113] Screen out the sensor with the best temperature adaptability as the reference object to provide a reliable benchmark for the fluctuation index calculation.

[0114] When the reference object is the influence object, the fluctuation coefficient of the corresponding historical time period is zero.

[0115] Through the calculation of the abnormality coefficient, the inertial sensor with normal fluctuation amplitude of collected parameters and temperature close to the reference temperature in the corresponding historical time period is screened out, and the reference object is taken as the comparison reference of other influence objects, to analyze the abnormal fluctuation degree of data collection of each influence object.

[0116] S3024, the average value of the fluctuation coefficients of all historical time periods is taken as the fluctuation index of the influence object AO VS , and the fluctuation index of each influence object in the virtual sensor VS is calculated respectively.

[0117] S3025, in this way, the fluctuation index of each influence object in each virtual sensor is calculated.

[0118] S303, get the temperature t now in the environmental data, set the basic coefficient WC0 and the adjustment coefficient WC ad . According to the fluctuation index, set the weight constraint coefficient for each inertial sensor in the virtual sensor. Specifically:

[0119] The average value of the fluctuation index FLU of all influence objects in the virtual sensor is calculated ave The weight constraint coefficient QZ of each influence object is calculated by substituting the formula:

[0120]

[0121] In the formula, k, h, c, and s are constants, and FLU is the fluctuation index of the influence object. The weight constraint coefficient of the inertial sensor of other non-influence objects in the virtual sensor is set as WC0.

[0122] The constants c and s are the center offsets of the modified Sigmoid function, and the constants k and h are parameters for controlling the slope of the Sigmoid function. The greater the values, the steeper the function curve.

[0123] The reliability of the quantification sensor in the temperature abnormal area is quantified, and the adaptive weight constraint is generated by historical fluctuation analysis to ensure the data reliability in high-temperature environment.

[0124] S400, the stability index is calculated in real time, and the weight of each influence object is dynamically adjusted according to the stability index and the weight constraint coefficient. Specifically, it includes:

[0125] S401, analyze the interval length t dur between the future time t and the current time, calculate the difference between the weight constraint coefficient of each inertial sensor in the virtual sensor and the basic coefficient WC0, and then divide it by the interval length t dur to obtain the adjustment speed.

[0126] When the weight constraint coefficient is greater than the basic coefficient WC0, the adjustment speed is positive, and the interval length t dur is adjusted by gradually increasing the weight.

[0127] When the weight constraint coefficient is less than the basic coefficient WC0, the adjustment speed is negative, and the interval length t dur is adjusted by gradually reducing the weight.

[0128] When the weight constraint coefficient is equal to the basic coefficient WC0, the adjustment speed is zero, and the interval length t dur does not change the weight.

[0129] S402, the weight coefficient of each inertial sensor in the virtual sensor is adjusted at a fixed time according to the adjustment speed, and the latest weight coefficient and attitude parameter are combined by using the weighted algorithm to calculate the comprehensive attitude parameter of the corresponding virtual sensor.

[0130] S403, the stability index of the unmanned aerial vehicle is calculated in real time according to the running data, and the adjustment speed of each inertial sensor in the virtual sensor is increased until the stability index is not less than the threshold value.

[0131] The calculation formula of the stability index STI is:

[0132]

[0133] In the formula, γ is the attenuation index, is the heading angle deviation, σgp is the standard deviation of the flight speed vector, I1 is the motor fundamental current amplitude (corresponding to the blade frequency), I u is the u-th harmonic current amplitude (u = 2, 3, 4, 5).

[0134] γ is an empirical coefficient for adjusting the influence weight of THD on the stability index. If γ has a large value, a slight increase in THD will cause the stability index to drop sharply. If γ has a small value, higher THD is allowed, and the stability standard is dynamically adjusted according to different tasks.

[0135] THD represents the proportion of harmonic components in the motor driving current to the total current, reflecting the motor torque ripple and mechanical vibration intensity. If THD increases, the motor vibration will intensify, and there is a risk of resonance in the unmanned aerial vehicle body, resulting in poor power stability.

[0136] THD < 15% indicates normal, and THD > 25% indicates a stability risk.

[0137] σgp is used to describe the degree of jitter of the flight trajectory of the unmanned aerial vehicle, with a unit of m / s. If σgp has a small value, the speed fluctuation is small, meaning that the flight is stable. If σgp has a large value, the trajectory jitter is large, and the control performance of the unmanned aerial vehicle is reduced.

[0138] represents the difference between the GPS heading angle and the command heading angle (unit: radian), reflecting the consistency of sensor fusion. The GPS heading angle is the heading angle calculated by the speed vector, and the command heading angle is the heading angle analyzed from the command signal. By the angle deviation is converted into a stability index influence weight.

[0139] The stability index couples the current harmonics and the GPS heading deviation to evaluate the stability. Specifically, the normalized score of the stability of power, trajectory, and heading is integrated, and the multi-source heterogeneous data is converted into a single decision index.

[0140] In S404, the weight coefficient is always between the weight constraint coefficient and the basic coefficient WC0 within the interval length.

[0141] The weight adjustment is coupled with the flight stability in real time, and the multi-dimensional index is used to quickly respond to unstable states, and the data fusion accuracy is optimized.

[0142] Through mathematical modeling, the physical environment (three-dimensional temperature field), historical fluctuations (sensor reliability), and real-time state (motor current / trajectory) are converted into a computable parameter system, providing quantitative decision-making basis for the attitude control of the UAV in complex environments.

[0143] Referring to Figure 2 The application also provides a device attitude data analysis system based on adaptive weighting, comprising a flight perception module, a scene analysis module, a parameter setting module, and a dynamic adjustment module.

[0144] The flight perception module is used to collect historical records of the UAV, as well as environmental data and operating data during flight.

[0145] The historical records of the UAV (including environmental data such as wind speed and temperature, and operating data such as GPS parameters, attitude parameters, temperature parameters, and motor current) and real-time flight data are collected. A three-dimensional temperature field is generated by combining temperature parameters with a three-dimensional model, and environmental information, flight state, and UAV performance data are integrated.

[0146] A multi-source heterogeneous data base is constructed to provide complete input for virtual scene building and temperature difference analysis, supporting the data needs of subsequent adaptive weighting algorithms.

[0147] The scene analysis module builds a virtual scene through environmental data and operating data, and real-time mapping and simulation of temperature difference zones.

[0148] The UAV three-dimensional model is loaded in the virtual scene, and the physical model is trained based on historical data to predict the future three-dimensional temperature field. By comparing the current and predicted temperature fields, the area with a temperature difference exceeding the threshold is classified as a "temperature difference zone".

[0149] Dynamic simulation of temperature changes and risk area identification are achieved, and the physical environment is mapped to a virtual model for quantitative analysis, providing spatial basis for sensor weight adjustment.

[0150] The parameter setting module is used to set the impact object according to the temperature difference zone, and to analyze the historical records to set the weight constraint coefficient for each impact object.

[0151] The inertial sensors in the temperature difference zone are marked as "impact objects", and grouped into "virtual sensors" according to the attitude parameter type. Based on historical temperature and attitude data, the fluctuation index of each impact object is calculated (by comparing the standard deviation of the reference object), and the weight constraint coefficient is dynamically generated using the Sigmoid function based on the current environmental temperature and reference temperature.

[0152] By quantifying the sensor data fluctuation, the weight constraint coefficient is adaptively allocated to ensure that the sensor weight in the temperature abnormal area can be dynamically adjusted with environmental changes, improving data reliability.

[0153] The dynamic adjustment module is configured to calculate the stability index in real time, and dynamically adjust the weight of each influence object according to the stability index and the weight constraint coefficient.

[0154] The adjustment speed is calculated according to the difference between the weight constraint coefficient and the basic coefficient, and the weight of each sensor in the virtual sensor is updated at a fixed time. In combination with the stability index calculated in real time, the weight adjustment is accelerated or decelerated dynamically.

[0155] The weight adjustment is linked with the flight stability in real time, the multi-dimensional index (power, trajectory, heading) is used to quickly respond to the unstable state, the sensor data fusion effect is optimized, and the control accuracy and flight safety of the unmanned aerial vehicle are ensured.

[0156] The flight perception module provides a multi-source data basis; the scene analysis module realizes virtual positioning of the temperature risk area; the parameter setting module generates an adaptive weight constraint through historical fluctuation analysis; and the dynamic adjustment module is coupled with stability feedback to optimize the weight distribution in real time.

[0157] The attitude data reliability and control stability of the unmanned aerial vehicle in a complex environment are improved, and the method is especially suitable for high-risk scenes such as temperature mutation.

[0158] In embodiment 1, it is assumed that the THD of the unmanned aerial vehicle is 25%, σgp is 0.6, 8°, and γ is 5, and the stability index is calculated by substituting the formula:

[0159]

[0160] The stability index of the unmanned aerial vehicle is 0.21.

[0161] It should be noted that, in this document, the relationship terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or device.

[0162] Finally, it should be noted that: the above only describes the preferred embodiments of the present application, and is not used to limit the present application, although the present application has been described in detail with reference to the foregoing embodiments, and those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A device attitude data analysis method based on adaptive weighting, characterized in that: The method includes: S100 collects historical data of the drone, as well as environmental and operational data during flight; S200: It builds virtual scenes using environmental and operational data, and maps and simulates temperature difference zones in real time. S300: Set the influencing objects according to the temperature difference zone, and analyze the historical records to set weight constraint coefficients for each influencing object; S400: Calculates the stability index in real time and dynamically adjusts the weights of each affected object based on the stability index and weight constraint coefficients.

2. The device attitude data analysis method based on adaptive weighting according to claim 1, characterized in that: In S100, historical records include environmental data and operational data from different historical periods; Environmental data includes wind speed and temperature around the drone; operational data includes the drone's 3D model, GPS parameters, attitude parameters, temperature parameters, and motor current. GPS parameters include spatial position, flight direction, and flight speed; attitude parameters are collected in real time by inertial sensors, specifically including acceleration, angular velocity, and angle; temperature parameters include the temperature at different locations inside the UAV; and a three-dimensional temperature field is generated by combining the temperature parameters with a three-dimensional model.

3. The device attitude data analysis method based on adaptive weighting according to claim 2, characterized in that: S200 includes: S201. Build a virtual scene, load the 3D model of the drone into the virtual scene, and adjust the pitch, roll and yaw angles of the 3D model in 3D space in real time according to the attitude parameters. S202. Obtain environmental and operational data from the same time period in historical records, analyze the relationship between wind speed, temperature, flight direction and flight speed and the three-dimensional temperature field, and construct a training set; S203. Input the training set into the physical model for training. After training, the physical model is based on the current three-dimensional temperature field TF. now In addition to environmental and operational data, the three-dimensional temperature field TF at a future time t is simulated and analyzed in a virtual scene. fut ; S204, Comparison of three-dimensional temperature field TF now and three-dimensional temperature field TF fut The temperature difference at the same location is used to divide the area where the temperature difference is greater than the threshold and this area is designated as the temperature difference zone.

4. The device attitude data analysis method based on adaptive weighting according to claim 3, characterized in that: The S300 includes: S301. Map the temperature difference zone in the 3D model within the virtual scene, and treat the inertial sensors within the temperature difference zone as the affected objects; combine all inertial sensors that collect the same attitude parameters into a virtual sensor. S302, TF three-dimensional temperature field fut The temperature at the location of each affected object is used as the predicted temperature, and a reference temperature t is set. ref By combining the attitude and temperature parameters from historical records, the fluctuation index of each affected object is calculated; S303. Obtain the temperature t from the environmental data. now Set the base coefficient WC0 and the adjustment coefficient WC. ad Weight constraint coefficients are set for each inertial sensor within the virtual sensor based on the fluctuation index.

5. The device attitude data analysis method based on adaptive weighting according to claim 4, characterized in that: S302 includes: S3021. Establish a three-dimensional temperature field based on temperature parameters from historical records, and analyze the influencing object AO in the virtual sensor VS. VS Temperature changes at the location; S3022, Obtain the affected object AO VS Predicted temperature t pre The temperature is marked as equal to the predicted temperature t. pre Furthermore, for historical periods whose duration exceeds the threshold, N time points are evenly set within each historical period. S3023. Set a reference object for each historical time period and analyze the influencing object AO at each time point. VS The difference in attitude parameters between the object and the reference object is used to calculate the standard deviation of the difference at all time points as the fluctuation coefficient for the corresponding historical time period. S3024, The average fluctuation coefficient of all historical time periods is used as the object of influence AO VS The fluctuation index is calculated for each of the other affected objects in the virtual sensor VS. S3025. Similarly, calculate the fluctuation index of each affected object in each virtual sensor.

6. The device attitude data analysis method based on adaptive weighting according to claim 5, characterized in that: In S3023, when analyzing the temperature and attitude parameters of each inertial sensor in the virtual sensor VS at each time point within the historical period, the anomaly coefficient YX of each inertial sensor is calculated by substituting them into the formula: In the formula, α is a constant, ZTC m Let t be the attitude parameters of the inertial sensor at the m-th time point. i Let be the temperature of the inertial sensor at the i-th time point; select the inertial sensor with the smallest anomaly coefficient as the reference.

7. The device attitude data analysis method based on adaptive weighting according to claim 4, characterized in that: In S303, the average fluctuation index (FLU) of all affected objects in the virtual sensor is calculated. ave Substitute the values ​​into the formula to calculate the weight constraint coefficient QZ for each affected object: In the formula, k, h, c, and s are constants, and FLU is the fluctuation index of the affected object; the weight constraint coefficient of the inertial sensors of other non-affected objects in the virtual sensor is set to WC0.

8. The device attitude data analysis method based on adaptive weighting according to claim 4, characterized in that: The S400 includes: S401. Analyze the time interval t between future time t and current time. dur Calculate the difference between the weight constraint coefficient of each inertial sensor in the virtual sensor and the basic coefficient WC0, and then divide each difference by the interval time t. dur Get the speed adjustment; S402. Adjust the weight coefficients of each inertial sensor in the virtual sensor according to the adjustment speed, and use a weighted algorithm to calculate and output the comprehensive attitude parameters of the corresponding virtual sensor by combining the latest weight coefficients and attitude parameters. S403. Calculate the stability index of the UAV in real time based on the operation data. If the stability index is less than the threshold, increase the adjustment speed of each inertial sensor in the virtual sensor until the stability index is not less than the threshold, then decrease the adjustment speed. S404. During the interval, the weight coefficient always remains between the weight constraint coefficient and the basic coefficient WC0.

9. The device attitude data analysis method based on adaptive weighting according to claim 8, characterized in that: In S403, the formula for calculating the Stability Index (STI) is as follows: In the formula, γ is the decay exponent. For the heading angle deviation, σgp is the standard deviation of the flight velocity vector, I1 is the amplitude of the fundamental motor current, and I u The value is the amplitude of the u-th harmonic current.

10. A device attitude data analysis system based on adaptive weighting, characterized in that: The system includes a flight perception module, a scene analysis module, a parameter setting module, and a dynamic adjustment module; The flight perception module is used to collect the drone's historical records, as well as environmental and operational data during flight. The scene analysis module builds virtual scenes using environmental and operational data, and maps and simulates temperature difference zones in real time. The parameter setting module is used to set the influencing objects according to the temperature difference zone and to set weight constraint coefficients for each influencing object by analyzing historical records. The dynamic adjustment module is used to calculate the stability index in real time and dynamically adjust the weights of each affected object based on the stability index and the weight constraint coefficient.

Images