A method and system for intelligent analysis of posture data

Abstract

The present invention relates to the technical field of posture data analysis, and specifically to a posture data intelligent analysis method and system, comprising the following steps: obtaining acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of a posture sensor, calculating environmental disturbance parameters, and obtaining environmental disturbance values. In the present invention, the posture change rate is calculated in real time to match the trend, the instantaneous fluctuation is smoothed by a sliding window method, and the weight is adjusted adaptively to optimize data consistency and prediction capabilities. The instantaneous angular velocity and acceleration offset in a high-speed motion state are dynamically compensated to reduce errors, and the rotation matrix parameters and inertia error correction amount are adjusted at the same time to improve the stability of the posture solution, and the data processing flow is optimized as a whole, so that the posture data can maintain high stability and accuracy in different motion modes, and the long-term trend matching and inertia error compensation capabilities are improved, which is particularly suitable for complex environments and high-speed dynamic scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of posture data analysis, and in particular to a posture data intelligent analysis method and system. Background Art

[0002] The field of attitude data analysis technology includes the technology of collecting, processing and analyzing the position, direction and motion state of objects in space. The core content includes the acquisition of attitude data, data preprocessing, attitude solution, feature extraction and analysis. The acquisition of attitude data usually relies on inertial measurement units, visual sensors or optical tracking systems, which can collect data such as acceleration, angular velocity, magnetic field strength, etc. Data preprocessing involves steps such as noise filtering, sensor data calibration and coordinate conversion to improve the accuracy and consistency of data. Attitude solution is mainly based on attitude estimation algorithms, which convert the raw data collected by sensors into representations such as attitude angles or quaternions. Feature extraction involves extracting key motion parameters from attitude data, such as angular velocity change trends, attitude stability, etc., to provide a basis for subsequent analysis. Attitude data analysis includes pattern recognition, trajectory analysis and behavior classification. Through mathematical modeling and data analysis methods, the potential laws and information in attitude data are mined to support applications such as motion analysis, intelligent monitoring and human-computer interaction.

[0003] Among them, the intelligent analysis method of posture data refers to the intelligent analysis technology based on posture data, which processes the collected posture information in multiple dimensions to obtain specific posture features and motion patterns. It covers key technical matters such as data synchronization, feature extraction, data modeling and pattern recognition. Data synchronization is used to time align the data collected by different sensors to ensure the consistency of analysis. Feature extraction obtains posture change trends, key feature points and posture stability parameters by analyzing the time series of posture data. Data modeling relies on statistical analysis and mathematical modeling methods to construct data association relationships and feature expression models, providing a basis for subsequent pattern recognition. Pattern recognition uses classification algorithms and pattern matching technologies to classify and analyze posture data and identify specific motion states and behavior patterns. Mainly relying on data processing and mathematical modeling methods, posture data is intelligently analyzed to support posture recognition and motion analysis needs in complex scenarios.

[0004] In the process of attitude data processing, the existing technology is limited by static or single time window calculation methods, which makes it difficult to adapt to dynamically changing attitude data, resulting in short-term fluctuations and environmental interference having a significant impact on data accuracy. The preprocessing of attitude data mainly relies on fixed filtering methods, which cannot be adaptively adjusted according to the actual motion state, resulting in error accumulation of data during intense motion or large environmental fluctuations. In the process of attitude solution, data modeling mostly uses linear methods, which makes it difficult to effectively model complex motion patterns, resulting in attitude deviations under high-speed motion conditions, affecting real-time performance and data accuracy. Inertial error compensation relies on fixed parameters and does not consider changes in motion states, which easily causes error accumulation and affects the overall stability of attitude data. The lack of a trend matching adjustment mechanism makes it impossible to dynamically optimize attitude data in combination with historical trends, resulting in short-term fluctuations affecting the overall stability of the data. Overall, the existing technology is difficult to provide stable, continuous, and low-error attitude data in complex environments and high-speed motion conditions, affecting application performance in the fields of high-precision motion monitoring, intelligent interactive devices, and automated navigation. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent analysis method for posture data.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a posture data intelligent analysis method, comprising the following steps: S1: Obtain the acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, calculate the environmental disturbance parameters, and obtain the environmental disturbance value; S2: calling the environmental disturbance value, calculating the angular velocity change rate and the attitude angle change rate, smoothing the attitude data stream, and obtaining the attitude data balance value; S3: calling the attitude data balance value, calculating the short-term, medium-term and long-term attitude change trends, obtaining the corresponding attitude trends, calculating the trend matching degree, screening the trend fitting parameters, judging the deviation between the short-term trend and the long-term trend, and if the deviation exceeds the threshold, adjusting the calculation parameters of the inertial sensor characteristic items, obtaining the trend optimization weight parameters, and obtaining the trend matching adjustment value; S4: calling the trend matching adjustment value, calculating the instantaneous angular velocity mutation amount and acceleration offset direction, adjusting the attitude integration step, and obtaining the inertia deviation correction amount; S5: calling the inertia deviation correction amount, adjusting the rotation matrix parameters, correcting the attitude angle calculation error, obtaining the optimized attitude data stream, and obtaining the attitude data analysis result.

[0007] As a further solution of the present invention, the environmental disturbance parameters include motion state influencing factors, magnetic field interference coefficients, and temperature and humidity variation coefficients; the attitude data balance values ​​include angular velocity smoothing parameters, attitude angle weight adjustment factors, and abnormal frame error correction amounts; the trend optimization weight parameters include trend matching adjustment coefficients, short-term offset correction factors, and long-term attitude fitting weights; the inertia deviation correction amounts include rotation compensation adjustment values, motion inertia correction factors, and acceleration error balance amounts; and the optimized attitude data stream includes attitude angle dynamic balance values, inertia compensation calculation amounts, and attitude matrix calibration parameters.

[0008] As a further solution of the present invention, the step of obtaining the environmental disturbance parameters is: S101: Acquire acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, calculate the fluctuation rate of the acceleration data, obtain the standard deviation of the acceleration data within the time window, and obtain the acceleration fluctuation rate; S102: Calculating a magnetic field offset value based on the acquired magnetic field strength data, by calculating the root mean square value of the three-axis components of the magnetic field strength and comparing it with the magnetic field reference value to obtain a magnetic field offset amplitude; S103: calling the acceleration fluctuation rate, the magnetic field offset amplitude and the temperature and humidity change rate, and comparing them with the set threshold value to determine whether they exceed the set threshold value. If any parameter exceeds the threshold value, the corresponding environmental disturbance parameter is recorded, using the formula: ; The environmental disturbance parameters are obtained by calculation, where: represents the environmental disturbance parameter, represents the acceleration volatility, represents the magnitude of the magnetic field deviation, Represents the rate of change of temperature and humidity.

[0009] As a further solution of the present invention, the step of acquiring the posture data balance value is: S201: calling the environmental disturbance parameter, collecting the angular velocity data sequence, calculating the angular velocity change rate, obtaining the first-order difference of the angular velocity data within a fixed time window, averaging the first-order difference of the angular velocity data, identifying the average change rate of the angular velocity, and obtaining the angular velocity change rate; S202: Calculating the attitude angle change rate based on the angular velocity change rate, identifying the attitude angle change amount per unit time by performing time differentiation processing on the attitude angle data, and calculating the average value thereof to obtain the attitude angle change rate; S203: calling the attitude angle change rate to determine whether it exceeds the set attitude stability threshold. If it exceeds, the attitude data stream is smoothed using a sliding window, and the weight of the abnormal frame data is adjusted using the formula: ; Calculate and obtain the balance value of posture data; in, Represents the balance value of posture data, Represents the first The weight of the frame, Represents the attitude angle data of the corresponding frame, Represents the window frame number.

[0010] As a further solution of the present invention, the step of obtaining the trend optimization weight parameter is: S301: calling the posture data balance value, calculating the short-term, medium-term and long-term posture change trends, respectively calculating the short-term, medium-term and long-term posture angle change averages within a fixed time window, calculating the posture change rate of the time period through a sliding window, and storing the calculation results as posture trends; S302: calling the posture trend, calculating the trend matching degree of the current posture data stream, using trend similarity calculation, calculating the Euclidean distance between the short-term, medium-term, and long-term trend data and the current data stream, and normalizing them to obtain a trend matching value, thereby obtaining the trend matching degree; S303: calling the trend matching degree to determine the deviation between the short-term trend and the long-term trend. If the deviation exceeds the set threshold, adjusting the calculation parameters of the inertial sensor characteristic items, using the formula: ; Calculate and obtain trend optimization weight parameters; in, represents the trend optimization weight parameter, Representative The short-term attitude trend value of the frame, Representative The long-term attitude trend value of the frame, Represents the total number of data frames used in the calculation, Adjustment factor for trend matching.

[0011] As a further solution of the present invention, the step of obtaining the inertia deviation correction amount is: S401: calling the trend optimization weight parameter, calculating the instantaneous angular velocity mutation amount and the acceleration offset direction, collecting the angular velocity data stream and calculating the angular velocity change value between adjacent frames, calculating the instantaneous angular velocity mutation amount by time difference normalization, obtaining the change of the three-axis components of acceleration at the same time, and calculating the offset angle of the acceleration vector direction relative to the previous frame, generating the angular velocity mutation amount and the acceleration offset direction; S402: calling the angular velocity mutation amount and the acceleration offset direction to determine whether the current motion state meets the high-speed motion condition, comparing the angular velocity threshold and the acceleration change threshold, if either one exceeds the set range, then the current state is determined to be a high-speed motion state, otherwise it is marked as a low-speed or stable motion state, and a high-speed motion determination result is generated; S403: calling the high-speed motion determination result, if the high-speed motion condition is met, adjusting the posture integration step length, calculating the rotational inertia compensation value, using the formula: ; Calculate and obtain the inertia deviation correction value; in, represents the inertia deviation correction, Representative The instantaneous angular velocity mutation of the frame, Represents the acceleration offset direction of the corresponding frame, Represents the total number of data frames used in the calculation, is the inertia compensation adjustment coefficient.

[0012] As a further solution of the present invention, the step of acquiring the optimized posture data stream is: S501: calling the inertia deviation correction amount, adjusting the rotation matrix parameters of the attitude calculation model, acquiring the three-axis angular velocity and linear acceleration data based on the inertial sensor data stream, and calculating the compensation term of the rotation matrix, and correcting the quaternion parameters in the rotation matrix according to the inertia deviation correction amount to obtain the rotation matrix optimization parameters; S502: calling the rotation matrix optimization parameters, calculating the smoothness of the posture data curve, smoothing the posture data using a sliding window filtering method, evaluating the smoothness of the posture curve by calculating the mean square error of the posture angle change in the window, and obtaining the smoothness of the posture data; S503: calling the smoothness of the posture data, correcting the posture angle calculation error, calculating the mean error between the smoothed posture data and the original data, and dynamically adjusting the posture angle, using the formula: ; Calculate and obtain the optimized posture data stream; in, Represents the posture data stream, is the original attitude angle, is the attitude angle after smoothing, Represents the total number of calculation frames, is the attitude correction factor.

[0013] A posture data intelligent analysis system, comprising: The environmental disturbance calculation module obtains acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data, calculates the attitude disturbance intensity value and environmental fluctuation amplitude value, and weights them to obtain the environmental disturbance value; The attitude data balance module calls the environmental disturbance value, calculates the angular velocity change rate and the attitude angle change rate, filters and processes the attitude data stream, and obtains the attitude data balance value; The trend matching adjustment module calls the attitude data balance value, calculates the short-term attitude trend, the medium-term attitude trend, and the long-term attitude trend, calculates the trend deviation, and determines whether it exceeds the trend deviation threshold. If it exceeds, the calculation parameters of the inertial sensor feature items are adjusted, the trend optimization weight parameters are calculated, the trend fitting degree is corrected, and the trend matching adjustment value is obtained; The inertia deviation correction module calls the trend matching adjustment value, calculates the instantaneous angular velocity mutation amount and the acceleration offset direction, calculates the difference between the two, and obtains the inertia deviation correction amount; The attitude output optimization module calls the inertia deviation correction amount, adjusts the rotation matrix parameters, corrects the attitude angle calculation error, smoothes the optimized attitude data stream, and obtains the attitude data analysis result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the attitude change rate and trend matching are calculated in real time, the sliding window method is used to smooth the instantaneous fluctuations, and the weights are adjusted adaptively to optimize data consistency and prediction ability. The instantaneous angular velocity and acceleration offset in the high-speed motion state are dynamically compensated to reduce errors. At the same time, the rotation matrix parameters and the inertia error correction amount are adjusted to improve the stability of the attitude solution. The data processing flow is optimized as a whole, so that the attitude data can maintain high stability and accuracy in different motion modes, and the long-term trend matching and inertia error compensation capabilities are improved. It is particularly suitable for complex environments and high-speed dynamic scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a main step flow chart of the present invention; Figure 2 A flow chart of the steps for obtaining environmental disturbance parameters of the present invention; Figure 3 A flowchart of the steps for obtaining the balance value of the posture data of the present invention; Figure 4 A flow chart of the steps of obtaining the trend optimization weight parameters of the present invention; Figure 5 A flow chart of the steps for obtaining the inertia deviation correction amount of the present invention; Figure 6 The figure is a flow chart of the steps of obtaining the optimized posture data stream of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 , a posture data intelligent analysis method, comprising the following steps: S1: Obtain the acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, calculate the acceleration fluctuation rate, magnetic field offset amplitude, and temperature and humidity change rate, and determine whether they exceed the threshold. If so, record the environmental disturbance parameters; S2: Call the environmental disturbance parameters, calculate the angular velocity change rate and attitude angle change rate, and determine whether the attitude stability parameter exceeds the threshold. If it exceeds the threshold, use the sliding window to smooth the attitude data stream, adjust the weight of the abnormal frame data, and obtain the attitude data balance value; S3: Call the attitude data balance value, calculate the short-term, medium-term and long-term attitude change trends, obtain the attitude trend of the corresponding time period, calculate the trend matching degree of the current attitude data stream, filter the trend fitting parameters, and judge the deviation between the short-term trend and the long-term trend. If the deviation exceeds the threshold, adjust the calculation parameters of the inertial sensor feature items to obtain the trend optimization weight parameters; S4: Call the trend optimization weight parameter, calculate the instantaneous angular velocity mutation amount and acceleration offset direction, and determine whether it is in a high-speed motion state. If it is in a high-speed motion state, adjust the attitude integration step, calculate the rotational inertia compensation value, and obtain the inertia deviation correction value; S5: Call the inertia deviation correction value, adjust the rotation matrix parameters of the attitude calculation model, calculate the smoothness of the attitude data curve, correct the attitude angle calculation error, and obtain the optimized attitude data stream.

[0019] Environmental disturbance parameters include motion state influencing factors, magnetic field interference coefficients, and temperature and humidity variation coefficients. Attitude data balance values ​​include angular velocity smoothing parameters, attitude angle weight adjustment factors, and abnormal frame error corrections. Trend optimization weight parameters include trend matching adjustment coefficients, short-term offset correction factors, and long-term attitude fitting weights. Inertia deviation corrections include rotation compensation adjustment values, motion inertia correction factors, and acceleration error balances. The optimized attitude data stream includes attitude angle dynamic balance values, inertia compensation calculation amounts, and attitude matrix calibration parameters.

[0020] See also Figure 2 , the steps to obtain the environmental disturbance parameters are: S101: Acquire acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, and calculate the fluctuation rate of the acceleration data. The calculation method is to obtain the standard deviation of the acceleration data within a certain time window to obtain the acceleration fluctuation rate; The acceleration, angular velocity, magnetic field strength, temperature, humidity and vibration data of the attitude sensor are obtained, and the fluctuation rate of the acceleration data is calculated. The calculation method is to obtain the standard deviation of the acceleration data within a certain time window. A series of data points output by the acceleration sensor are collected in a laboratory environment. For example, data is recorded every 0.1 second during a 30-second test period. The standard deviation is calculated from these data. This standard deviation reflects the degree of fluctuation of the sensor output data. This calculation process provides key information for monitoring the quality and stability of the sensor and obtains the acceleration fluctuation rate.

[0021] S102: Calculate the magnetic field offset value based on the acquired magnetic field strength data, and obtain the magnetic field offset amplitude by calculating the root mean square value of the three-axis components of the magnetic field strength and comparing it with the magnetic field reference value; Calculate the magnetic field offset amplitude by calculating the root mean square value of the three-axis components of the magnetic field strength and comparing it with the magnetic field reference value. For example, a standard geomagnetic field strength is 50μT. By comparing the actual measured value with this reference value, the calculated root mean square value can help identify whether the device's magnetic field sensor is affected by external interference or the device's own magnetic material. This comparative analysis can ensure the accuracy and reliability of device functions in magnetic field-sensitive applications such as smartphones or navigation systems, and obtain the magnetic field offset amplitude.

[0022] S103: Call the acceleration fluctuation rate, magnetic field offset amplitude and temperature and humidity change rate, and compare them with the set threshold to determine whether they exceed the set threshold. If any parameter exceeds the threshold, the corresponding environmental disturbance parameter is recorded, using the formula: ; The environmental disturbance parameters are obtained by calculation, where: represents the environmental disturbance parameter, represents the acceleration volatility, represents the magnitude of the magnetic field deviation, Represents the rate of change of temperature and humidity; Parameter acquisition process and calculation: Calculate the acceleration volatility ; Assume that the acceleration data (unit: ) is as follows, with a sampling interval of 0.1s and a total of 10 data points: ; Acceleration rate Calculate the standard deviation: ; in, is the average value: ; Calculate the standard deviation: ; ; ; Calculate the magnetic field offset amplitude : Assume that the three-axis component data of the magnetic field (unit: μT): ; Reference magnetic field value μT.

[0023] Calculation of RMS magnetic field strength: ; ; ; Magnetic field offset amplitude: ; Calculate the rate of change of temperature and humidity : Assume that the temperature data changes over time (unit: °C) and the sampling interval is 1 minute: ; Humidity data (unit: %RH): ; Calculate the rate of temperature change: ; Calculate the rate of change of humidity: ; Take the average of the two as the comprehensive temperature and humidity change rate: ; Calculate environmental disturbance parameters : Substitute the calculated parameters into the formula: ; ; ; ; Result analysis: Calculated environmental disturbance parameters , which indicates the overall disturbance of the sensor's vibration, magnetic field change, temperature and humidity fluctuation in the current environment.

[0024] If the disturbance threshold is set , the current D value exceeds the threshold, indicating that the environment has changed significantly, and the corresponding disturbance parameters should be recorded for further analysis.

[0025] The benefit of this formula is that by simultaneously considering the fluctuations of acceleration, magnetic field, temperature and humidity, it can comprehensively evaluate environmental disturbances, making it suitable for complex industrial environments, laboratory monitoring, and equipment health monitoring scenarios.

[0026] See also Figure 3 , the steps to obtain the balance value of posture data are: S201: calling environmental disturbance parameters, collecting angular velocity data sequences, and calculating the angular velocity change rate. The calculation method is to obtain the first-order difference of the angular velocity data within a fixed time window, and average the absolute values ​​thereof to obtain the average change rate of the angular velocity, thereby obtaining the angular velocity change rate. Collect angular velocity data sequence and calculate angular velocity change rate. When collecting angular velocity data, the device needs to use a gyroscope sensor to obtain three-axis angular velocity data at a fixed frequency, such as sampling once every 10ms, and continuously obtaining data within 1000ms to form a time series containing 100 data points. Each data point contains angular velocity information around the X, Y, and Z axes. For the calculation of the angular velocity change rate, it is necessary to perform a first-order difference calculation on the time series, that is, to calculate the change in angular velocity between adjacent data points to reflect the change trend of the angular velocity. For example, if the angular velocity of a device at t1=10ms is (0.2,-0.1,0.3)rad / s, and the angular velocity at t2=20ms becomes (0.25,-0.15,0.32)rad / s, then the angular velocity change at this moment is calculated to be (0.05,-0.05,0.02)rad / s. In the entire time series, adjacent differences are calculated for each time point, and the average of the absolute values ​​is calculated to obtain the angular velocity change rate within the time window. The angular velocity change rate can be used for subsequent attitude stability analysis to obtain the angular velocity change rate.

[0027] S202: Calculate the attitude angle change rate based on the angular velocity change rate, obtain the attitude angle change amount per unit time by performing time differentiation processing on the attitude angle data, and calculate the average value thereof to reflect the attitude change trend and obtain the attitude angle change rate; Calculate the attitude angle change rate. During the calculation process, first interpolate the sensor data to ensure that the time interval of the data points is consistent, and then use the numerical integration method to calculate the attitude angle, that is, by accumulating angular velocity data, obtain the rotation angle change of the device in three-dimensional space. For example, if the device rotates around the X-axis, the angular velocity is 0.2, 0.22, 0.25, and 0.28 rad / s in 100ms, then the attitude angle change in this period can be calculated by accumulating angular velocity, that is, ,in For the The angular velocity at time, is the sampling time interval, and the attitude angle change in the time window is obtained by cumulative calculation. The attitude angle change rate is defined as the attitude angle change per unit time. For example, if the attitude angle change in 1000ms is 5°, then the attitude angle change rate in the time window is 5° / s. This rate can be used for subsequent attitude stability judgment to obtain the attitude angle change rate.

[0028] S203: Call the attitude angle change rate to determine whether it exceeds the set attitude stability threshold. If it exceeds, use a sliding window to smooth the attitude data stream and adjust the weight of the abnormal frame data using the formula: ; Calculate and obtain the balance value of posture data; in, Represents the balance value of posture data, Represents the first The weight of the frame, Represents the attitude angle data of the corresponding frame, Represents the window frame number.

[0029] Call the attitude angle change rate to determine whether it exceeds the set attitude stability threshold. If it exceeds, use the sliding window to smooth the attitude data stream and adjust the weight of the abnormal frame data. In the calculation process, you first need to set the sliding window size. For example, set the sliding window to include the most recent 10 frames of data, and the attitude angle data of each frame is recorded as , the weight setting is based on the degree of abnormality. If the attitude angle change rate of a frame exceeds the threshold, the weight of the frame is reduced to reduce its impact on the final result. The formula is: ; in, Represents the first The frame's attitude angle data, represents the weight of the corresponding frame, Represents the number of frames in the window. When setting the weight, the weight of the stable frame is set to 1, and the weight of the abnormal frame is set to 0.5. For example, in 10 frames of data, if the attitude angle change rate of the 3rd and 7th frames exceeds the threshold, the weight is set to , the weights of the other frames are 1. Substituting them into the formula, we can get the balance value of the posture data after the sliding window is smoothed, and finally get the balance value of the posture data; Calculate the values ​​of each parameter: Assume that the sliding window contains Frame data: (Unit: degree); The corresponding weights are: ; That is, the third frame is detected as an abnormal frame, and its weight is reduced to 0.5.

[0030] Compute the sliding window weighted average: First calculate the numerator: ; Calculate the denominator: ; Calculate the balance value of posture data: ; Result analysis: Calculated balance value of posture data , which represents the smoothed attitude angle within the current sliding window.

[0031] If the stability threshold is set , the current calculation result is lower than the threshold, indicating that the data tends to be stable after adjusting the abnormal frame weight.

[0032] The innovation of this formula is that by setting the weight of abnormal frames, the influence of abnormal data on the overall attitude angle balance value is reduced, the reliability of attitude data is improved, and at the same time, the sliding window is used to ensure the continuity of the data stream, making the attitude adjustment smoother and more accurate.

[0033] See also Figure 4 , the steps to obtain the trend optimization weight parameters are: S301: calling the attitude data balance value, calculating the short-term, medium-term and long-term attitude change trends, calculating the short-term, medium-term and long-term attitude angle change averages in a fixed time window, calculating the attitude change rate in each time period through a sliding window, and storing the calculation results as attitude trends; First, it is necessary to analyze the short-term (within 1 second), medium-term (within 5 seconds), and long-term (within 30 seconds) attitude data, which involves collecting continuous attitude data from sensors, such as angle changes. For example, if the flight data of the aircraft is used, the attitude angle can be obtained every second, and then the average attitude angle change every second, every five seconds, and every thirty seconds can be calculated. In this way, the changing trend of the attitude on different time scales can be clearly seen, which is crucial for understanding the rapid response of the aircraft in the short term and the long-term flight stability. These trend data are finally summarized into attitude trends, which are critical for subsequent flight control system adjustments.

[0034] S302: calling the posture trend, calculating the trend matching degree of the current posture data stream, using the trend similarity calculation method, calculating the Euclidean distance between the short-term, medium-term, and long-term trend data and the current data stream, and normalizing them to obtain the trend matching value, which is used to evaluate the similarity between the current posture and the historical trend and obtain the trend matching degree; After analyzing the attitude trend, the next step is to calculate the trend match with the current attitude data stream, which involves comparing the real-time data with the short-term, medium-term, and long-term trends obtained from the previous analysis. For example, if at a certain moment, the real-time attitude data of the aircraft suddenly deviates from the long-term trend, this may indicate that the aircraft has encountered an unexpected situation or a system failure. By calculating the Euclidean distance between the real-time data and the trend data and normalizing it, a trend match index can be obtained. This value reflects the consistency of the current data with the historical trend. The final trend match can help the flight control system determine whether adjustments are needed to return to normal flight mode.

[0035] S303: Call the trend matching degree to determine the deviation between the short-term trend and the long-term trend. If the deviation exceeds the set threshold, adjust the calculation parameters of the inertial sensor characteristic items using the formula: ; Calculate and obtain trend optimization weight parameters; in, represents the trend optimization weight parameter, Representative The short-term attitude trend value of the frame, Representative The long-term attitude trend value of the frame, Represents the total number of data frames used in the calculation, Adjustment factor for trend matching.

[0036] here, and The short-term and long-term trends are frame data, and is the adjustment coefficient, which is used to adjust the sensitivity of the trend optimization weight parameter. The trend optimization weight parameter calculated according to this formula can be used to fine-tune the output of the sensor to ensure that the aircraft can maintain the correct attitude response when encountering unstable airflow, further ensuring flight safety; Parameter acquisition process and calculation: Calculate short-term and long-term trend data: Assume that the attitude angle data (unit: degree) of the aircraft is as follows, and the short-term trend (1 second window) and long-term trend (30 second window) data are as follows: ; ; Total number of data frames .

[0037] Calculate the trend deviation: Calculate the absolute difference between the short-term trend value and the long-term trend value: ; Calculate the trend deviation from the mean: Compute the mean trend deviation for all frames: ; Calculate trend optimization weight parameters: Assuming trend matching adjustment factor ,but: ; ; Result analysis: Calculated trend optimization weight parameters This value is used to adjust the calculation parameters of the inertial sensor to make it more consistent with the current attitude trend change.

[0038] If the trend optimization weight parameter threshold is set to , then the current If the threshold is exceeded, it indicates that there is a large deviation between the short-term trend and the long-term trend, and the sensor's calculation parameters need to be adjusted to make them more consistent with the long-term posture trend changes.

[0039] The innovation of this formula is that by calculating the deviation between short-term and long-term trends and introducing adjustment coefficients, the calculation parameters of the inertial sensor are dynamically optimized to improve the stability of the attitude data and make it more consistent with actual flight conditions.

[0040] See also Figure 5 , the steps to obtain the inertia deviation correction are: S401: calling trend optimization weight parameters, calculating instantaneous angular velocity mutation and acceleration offset direction, collecting angular velocity data stream and calculating angular velocity change values ​​between adjacent frames, calculating instantaneous angular velocity mutation through time difference normalization, obtaining changes in acceleration three-axis components at the same time, and calculating the offset angle of the acceleration vector direction relative to the previous frame, generating angular velocity mutation and acceleration offset direction; Call the trend optimization weight parameter, obtain the angular velocity data stream, calculate the instantaneous angular velocity mutation, and calculate the acceleration offset direction. The angular velocity data stream is obtained from the inertial sensor, and its sampling frequency is set to 100Hz, that is, the angular velocity information is recorded every 10ms. The sampled data contains the angular velocity components around the X, Y, and Z axes. In order to calculate the instantaneous angular velocity mutation, it is necessary to calculate the angular velocity difference between adjacent data frames and divide it by the time interval for normalization. For example, if the angular velocity at a certain moment is (0.5, -0.3, 0.2) rad / s, and the angular velocity of the next frame is (0.8, -0.2, 0.3) rad / s, then the instantaneous angular velocity mutation is calculated as Similarly, the entire data stream is traversed, calculated and stored. In addition, it is necessary to obtain the three-axis data of the acceleration sensor and calculate the angle change of the acceleration vector relative to the previous frame to analyze the change trend of the motion direction. The specific calculation method is to calculate the cosine similarity of the acceleration vectors of the previous and next two frames, such as , obtain the acceleration offset direction, and finally form the angular velocity mutation amount and acceleration offset direction.

[0041] S402: calling the angular velocity mutation amount and the acceleration offset direction to determine whether the current motion state meets the high-speed motion condition, comparing the angular velocity threshold and the acceleration change threshold, if either one exceeds the set range, the current state is determined to be a high-speed motion state, otherwise it is marked as a low-speed or stable motion state, and a high-speed motion determination result is generated; The angular velocity mutation amount and acceleration offset direction are called to determine whether it is in a high-speed motion state. First, the high-speed motion thresholds of the angular velocity and acceleration are set. For example, the angular velocity threshold is set to 3rad / s, and the acceleration change threshold is set to 2m / s2. When the currently calculated angular velocity mutation amount is greater than 3rad / s or the angle change in the acceleration offset direction is greater than 30°, it is judged to be a high-speed motion state, otherwise it is marked as a low-speed or stable motion state. The specific judgment method uses threshold comparison. For example, at t=5s, if the angular velocity mutation amount is 3.5rad / s and the acceleration offset angle is 25°, the high-speed motion standard is not met. If at t=10s, the angular velocity mutation amount increases to 4.1rad / s and the acceleration offset angle reaches 35°, the high-speed motion condition is met, the state is recorded as a high-speed motion state, and finally a high-speed motion judgment result is generated.

[0042] S403: Call the high-speed motion judgment result. If the high-speed motion condition is met, adjust the attitude integration step length and calculate the rotational inertia compensation value using the formula: ; Calculate and obtain the inertia deviation correction value; in, represents the inertia deviation correction, Representative The instantaneous angular velocity mutation of the frame, Represents the acceleration offset direction of the corresponding frame, Represents the total number of data frames used in the calculation, is the inertia compensation adjustment coefficient.

[0043] Call the high-speed motion judgment result. If the high-speed motion condition is met, adjust the attitude integral step and calculate the rotational inertia compensation value. The integral step is adjusted according to the size of the current angular velocity mutation. If the angular velocity mutation exceeds the threshold of 3rad / s, adjust the integral step to 1.5 times the original value. At the same time, in order to compensate for the rotational inertia effect, it is necessary to calculate the inertia deviation correction value. The formula is: ; Assuming that the sampling window Frame, the angular velocity mutation data is rad / s, the acceleration offset direction is m / s2, calculate the numerator: ; Calculate the inertia deviation correction: ; Assumed inertia compensation adjustment factor ,but: ; Finally, the inertia deviation correction amount is obtained.

[0044] See also Figure 6 , the steps to obtain the optimized posture data stream are: S501: Call the inertia deviation correction value, adjust the rotation matrix parameters of the attitude calculation model, obtain the three-axis angular velocity and linear acceleration data based on the inertial sensor data stream, and calculate the compensation term of the rotation matrix. According to the inertia deviation correction value, the quaternion parameters in the rotation matrix are corrected to make it more consistent with the current attitude change, and finally the rotation matrix optimization parameters are obtained; Call the inertial deviation correction amount, adjust the rotation matrix parameters of the attitude calculation model, obtain the three-axis angular velocity and linear acceleration data based on the inertial sensor, and use the quaternion representation to establish the rotation matrix. First, the angular velocity data collected from the sensor Perform discrete integration to calculate the rotation angle increment , correct the current posture through the quaternion update formula, if the inertial deviation correction amount If the value exceeds the set threshold, the weighted mean method is used to adjust the rotation matrix. For example, for the quaternion at a certain moment , updated according to the inertia correction calculation , the updated posture is ,in Represents quaternion multiplication, and finally obtains the rotation matrix optimization parameters.

[0045] S502: Calling the rotation matrix optimization parameters, calculating the smoothness of the posture data curve, using the sliding window filtering method to smooth the posture data, evaluating the smoothness of the posture curve by calculating the mean square error of the posture angle change in the window, and obtaining the smoothness of the posture data; Call the rotation matrix optimization parameters, calculate the smoothness of the posture data curve, use the sliding window filtering method to smooth the posture data stream, and set the window size , extract the attitude angle data within the window , calculate the mean square error: ; in, is the mean of the data in the window, if If the smoothness is lower than the set smoothness threshold, the data in this interval is considered smooth, otherwise secondary filtering is performed. At the same time, in order to reduce the influence of noise, the adaptive weighted average method is used to calculate the optimized angle , where the weight Adaptive adjustment is made based on data deviation to finally obtain the smoothness of the posture data.

[0046] S503: Call the smoothness of the posture data, correct the posture angle calculation error, calculate the average error between the smoothed posture data and the original data, and dynamically adjust the posture angle using the formula: ; Calculate and obtain the optimized posture data stream; in, Represents the posture data stream, is the original attitude angle, is the attitude angle after smoothing, Represents the total number of calculation frames, is the attitude correction factor.

[0047] Call the smoothness of the attitude data, correct the attitude angle calculation error, calculate the average error between the smoothed attitude data and the original data, and dynamically adjust the attitude angle using the formula: ; in, is the original attitude angle, is the attitude angle after smoothing, Represents the total number of calculation frames, is the attitude correction factor. For example, if the sampling window size , the error term is calculated as , then the error mean is: ; Assumptions , then the final error correction value is: ; The final operation obtains the optimized posture data stream.

[0048] A posture data intelligent analysis system, comprising: The environmental disturbance calculation module obtains acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data, calculates the attitude disturbance intensity value and environmental fluctuation amplitude value, and weights them to obtain the environmental disturbance value; The attitude data balance module calls the environmental disturbance value, calculates the angular velocity change rate and the attitude angle change rate, filters and processes the attitude data stream, and obtains the attitude data balance value; The trend matching adjustment module calls the attitude data balance value, calculates the short-term attitude trend, the medium-term attitude trend, and the long-term attitude trend, calculates the trend deviation, and determines whether it exceeds the trend deviation threshold. If it exceeds, it adjusts the calculation parameters of the inertial sensor feature items, calculates the trend optimization weight parameters, corrects the trend fitting degree, and obtains the trend matching adjustment value; The inertia deviation correction module calls the trend matching adjustment value, calculates the instantaneous angular velocity mutation amount and the acceleration offset direction, calculates the difference between the two, and obtains the inertia deviation correction amount; The attitude output optimization module calls the inertia deviation correction, adjusts the rotation matrix parameters, corrects the attitude angle calculation error, smoothes the optimized attitude data stream, and obtains the attitude data analysis results.

[0049] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for intelligent analysis of posture data, characterized in that: The following steps are involved: S1: Obtain the acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, calculate the environmental disturbance parameters, and obtain the environmental disturbance value; S2: calling the environmental disturbance value, calculating the angular velocity change rate and the attitude angle change rate, smoothing the attitude data stream, and obtaining the attitude data balance value; S3: calling the attitude data balance value, calculating the short-term, medium-term and long-term attitude change trends, obtaining the corresponding attitude trend, calculating the trend matching degree, screening the trend fitting parameters, judging the deviation between the short-term trend and the long-term trend, and if the deviation exceeds the threshold, adjusting the calculation parameters of the inertial sensor characteristic items, obtaining the trend optimization weight parameters, and obtaining the trend matching adjustment value; S4: calling the trend matching adjustment value, calculating the instantaneous angular velocity mutation amount and acceleration offset direction, adjusting the attitude integration step, and obtaining the inertia deviation correction amount; S5: calling the inertia deviation correction amount, adjusting the rotation matrix parameters, correcting the attitude angle calculation error, obtaining the optimized attitude data stream, and obtaining the attitude data analysis result.

2. The method for intelligent analysis of posture data according to claim 1, characterized in that: The environmental disturbance parameters include motion state influencing factors, magnetic field interference coefficients, and temperature and humidity variation coefficients; the attitude data balance value includes angular velocity smoothing parameters, attitude angle weight adjustment factors, and abnormal frame error correction amounts; the trend optimization weight parameters include trend matching adjustment coefficients, short-term offset correction factors, and long-term attitude fitting weights; the inertia deviation correction amounts include rotation compensation adjustment values, motion inertia correction factors, and acceleration error balance amounts; the optimized attitude data stream includes attitude angle dynamic balance values, inertia compensation calculation amounts, and attitude matrix calibration parameters.

3. The method for intelligent analysis of posture data according to claim 1, characterized in that: The steps for obtaining the environmental disturbance parameters are: S101: Acquire acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data of the attitude sensor, calculate the fluctuation rate of the acceleration data, obtain the standard deviation of the acceleration data within the time window, and obtain the acceleration fluctuation rate; S102: Calculating a magnetic field offset value based on the acquired magnetic field strength data, by calculating the root mean square value of the three-axis components of the magnetic field strength and comparing it with the magnetic field reference value to obtain a magnetic field offset amplitude; S103: calling the acceleration fluctuation rate, the magnetic field offset amplitude and the temperature and humidity change rate, and comparing them with the set threshold value to determine whether they exceed the set threshold value. If any parameter exceeds the threshold value, the corresponding environmental disturbance parameter is recorded, using the formula: ； The environmental disturbance parameters are obtained by calculation, where: represents the environmental disturbance parameter, represents the acceleration volatility, represents the magnetic field deviation amplitude, Represents the rate of change of temperature and humidity.

4. The method for intelligent analysis of posture data according to claim 1, characterized in that: The steps for obtaining the posture data balance value are: S201: calling the environmental disturbance parameter, collecting the angular velocity data sequence, calculating the angular velocity change rate, obtaining the first-order difference of the angular velocity data within a fixed time window, averaging the first-order difference of the angular velocity data, identifying the average change rate of the angular velocity, and obtaining the angular velocity change rate; S202: Calculating the attitude angle change rate based on the angular velocity change rate, identifying the attitude angle change amount per unit time by performing time differentiation processing on the attitude angle data, and calculating the average value thereof to obtain the attitude angle change rate; S203: calling the attitude angle change rate to determine whether it exceeds the set attitude stability threshold. If it exceeds, the attitude data stream is smoothed using a sliding window, and the weight of the abnormal frame data is adjusted using the formula: ； Calculate and obtain the balance value of posture data; in, Represents the balance value of posture data, Represents the first The weight of the frame, Represents the attitude angle data of the corresponding frame, Represents the window frame number.

5. The method for intelligent analysis of posture data according to claim 1, characterized in that: The steps for obtaining the trend optimization weight parameter are as follows: S301: calling the posture data balance value, calculating the short-term, medium-term and long-term posture change trends, respectively calculating the short-term, medium-term and long-term posture angle change averages within a fixed time window, calculating the posture change rate of the time period through a sliding window, and storing the calculation results as posture trends; S302: calling the posture trend, calculating the trend matching degree of the current posture data stream, using trend fitting calculation, calculating the Euclidean distance between the short-term, medium-term, and long-term trend data and the current data stream, and normalizing them to obtain a trend matching value, thereby obtaining the trend matching degree; S303: calling the trend matching degree to determine the deviation between the short-term trend and the long-term trend. If the deviation exceeds the set threshold, adjusting the calculation parameters of the inertial sensor characteristic items, using the formula: ； Calculate and obtain trend optimization weight parameters; in, represents the trend optimization weight parameter, Representative The short-term attitude trend value of the frame, Representative The long-term attitude trend value of the frame, Represents the total number of data frames used in the calculation, Adjustment factor for trend matching.

6. The method for intelligent analysis of posture data according to claim 1, characterized in that: The steps for obtaining the inertia deviation correction amount are as follows: S401: calling the trend optimization weight parameter, calculating the instantaneous angular velocity mutation amount and the acceleration offset direction, collecting the angular velocity data stream and calculating the angular velocity change value between adjacent frames, calculating the instantaneous angular velocity mutation amount by time difference normalization, obtaining the change of the three-axis components of acceleration at the same time, and calculating the offset angle between the acceleration vector direction and the previous frame, generating the angular velocity mutation amount and the acceleration offset direction; S402: calling the angular velocity mutation amount and the acceleration offset direction to determine whether the current motion state meets the high-speed motion condition, comparing the angular velocity threshold and the acceleration change threshold, if either one exceeds the set range, then the current state is determined to be a high-speed motion state, otherwise it is marked as a low-speed or stable motion state, and a high-speed motion determination result is generated; S403: calling the high-speed motion determination result, if the high-speed motion condition is met, adjusting the posture integration step length, calculating the rotational inertia compensation value, using the formula: ； Calculate and obtain the inertia deviation correction value; in, represents the inertia deviation correction, Representative The instantaneous angular velocity mutation of the frame, Represents the acceleration offset direction of the corresponding frame, Represents the total number of data frames used in the calculation, is the inertia compensation adjustment coefficient.

7. The method for intelligent analysis of posture data according to claim 1, characterized in that: The steps for obtaining the optimized posture data stream are: S501: calling the inertia deviation correction amount, adjusting the rotation matrix parameters of the attitude calculation model, acquiring the three-axis angular velocity and linear acceleration data based on the inertial sensor data stream, and calculating the compensation term of the rotation matrix, and correcting the quaternion parameters in the rotation matrix according to the inertia deviation correction amount to obtain the rotation matrix optimization parameters; S502: calling the rotation matrix optimization parameters, calculating the smoothness of the posture data curve, smoothing the posture data using a sliding window filtering method, evaluating the smoothness of the posture curve by calculating the mean square error of the posture angle change in the window, and obtaining the smoothness of the posture data; S503: calling the smoothness of the posture data, correcting the posture angle calculation error, calculating the mean error between the smoothed posture data and the original data, and dynamically adjusting the posture angle, using the formula: ； Calculate and obtain the optimized posture data stream; in, Represents the posture data stream, is the original attitude angle, is the attitude angle after smoothing, Represents the total number of calculation frames, is the attitude correction factor.

8. A posture data intelligent analysis system, characterized in that: The system is used to execute a posture data intelligent analysis method according to any one of claims 1 to 7, comprising: The environmental disturbance calculation module obtains acceleration, angular velocity, magnetic field strength, temperature and humidity, and vibration data, calculates the attitude disturbance intensity value and environmental fluctuation amplitude value, and weights them to obtain the environmental disturbance value; The attitude data balance module calls the environmental disturbance value, calculates the angular velocity change rate and the attitude angle change rate, filters and processes the attitude data stream, and obtains the attitude data balance value; The trend matching adjustment module calls the attitude data balance value, calculates the short-term attitude trend, the medium-term attitude trend, and the long-term attitude trend, calculates the trend deviation, and determines whether it exceeds the trend deviation threshold. If it exceeds, the calculation parameters of the inertial sensor feature items are adjusted, the trend optimization weight parameters are calculated, the trend fitting degree is corrected, and the trend matching adjustment value is obtained; The inertia deviation correction module calls the trend matching adjustment value, calculates the instantaneous angular velocity mutation amount and the acceleration offset direction, calculates the difference between the two, and obtains the inertia deviation correction amount; The attitude output optimization module calls the inertia deviation correction amount, adjusts the rotation matrix parameters, corrects the attitude angle calculation error, smoothes the optimized attitude data stream, and obtains the attitude data analysis result.

Images