Gesture sensing analysis method and system based on gyroscope accelerometer

Abstract

The invention discloses an attitude sensing analysis method and system based on a gyroscope accelerometer, and relates to the technical field of attitude analys.The method comprises the steps that a historical attitude sensing data set is obtained, and an attitude sensing early warning model prototype set is constructed based on the historical attitude sensing data set; acquiring an angular velocity three-axis data sequence and an acceleration three-axis data sequence of the equipment in the preset monitoring window; determining an angular velocity three-axis concentrated screening data set and an acceleration three-axis concentrated screening data set; a pitch angle set, a roll angle set and a yaw angle set are obtained; and according to the pitch angle, roll angle and yaw angle set, matching with the attitude sensing early warning model prototype set, and taking an attitude sensing abnormal type corresponding to a matching result as a target attitude sensing analysis result. The technical problems that in the prior art, attitude sensing analysis needs a long period, and analysis reliability is low are solved, and the technical effect of improving analysis accuracy and analysis efficiency is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of attitude analysis, and specifically relates to an attitude sensing and analysis method and system based on a gyroscope and an accelerometer. Background Art

[0002] In the field of attitude sensing and analysis, traditional methods mainly rely on a single sensor (such as a gyroscope or an accelerometer) for attitude detection, and obtain basic attitude parameters (such as Euler angles) through algorithm processing. A single sensor is easily affected by noise, resulting in poor stability of the attitude analysis result and inability to quickly and timely perform attitude feedback. The prior art has the technical problems of a long period for attitude sensing and analysis and low analysis reliability. Summary of the Invention

[0003] The present application provides an attitude sensing and analysis method and system based on a gyroscope and an accelerometer, which are used to solve the technical problems of a long period for attitude sensing and analysis and low analysis reliability in the prior art.

[0004] In view of the above problems, the present application provides an attitude sensing and analysis method and system based on a gyroscope and an accelerometer.

[0005] In the first aspect of the present application, an attitude sensing and analysis method based on a gyroscope and an accelerometer is provided. The method includes: Obtain a historical attitude sensing data set, and construct a prototype set of attitude sensing warning models based on the historical attitude sensing data set. Each attitude sensing warning model prototype corresponds to an attitude sensing abnormal type.

[0006] Respectively obtain a three-axis data sequence of angular velocity and a three-axis data sequence of acceleration of the device within a preset monitoring window through a gyroscope and an accelerometer disposed on the device.

[0007] Respectively perform data set screening and analysis on the three-axis data sequence of angular velocity and the three-axis data sequence of acceleration to determine a three-axis concentrated screening data set of angular velocity and a three-axis concentrated screening data set of acceleration.

[0008] Perform attitude analysis based on the three-axis concentrated screening data set of angular velocity and the three-axis concentrated screening data set of acceleration to obtain a set of pitch angle, roll angle, and yaw angle.

[0009] Match the set of pitch angle, roll angle, and yaw angle with the prototype set of attitude sensing warning models, and use the attitude sensing abnormal type corresponding to the matching result as the target attitude sensing analysis result.

[0010] In the second aspect of the present application, an attitude sensing and analysis system based on a gyroscope and an accelerometer is provided. The system includes: The prototype set construction module is used to obtain the historical attitude sensing data set and construct a prototype set of the attitude sensing warning model based on the historical attitude sensing data set, where each prototype of the attitude sensing warning model corresponds to an attitude sensing anomaly type.

[0011] The data sequence acquisition module is used to respectively obtain the three-axis data sequence of the angular velocity and the three-axis data sequence of the acceleration of the device within a preset monitoring window through the gyroscope and the accelerometer arranged on the device.

[0012] The centralized screening data acquisition module is used to respectively perform centralized screening analysis on the three-axis data sequence of the angular velocity and the three-axis data sequence of the acceleration to determine the three-axis centralized screening data set of the angular velocity and the three-axis centralized screening data set of the acceleration.

[0013] The yaw angle set acquisition module is used to perform attitude analysis based on the three-axis centralized screening data set of the angular velocity and the three-axis centralized screening data set of the acceleration to obtain the pitch angle, roll angle and yaw angle set.

[0014] The target attitude sensing analysis result acquisition module is used to match the pitch angle, roll angle and yaw angle set with the prototype set of the attitude sensing warning model, and use the attitude sensing anomaly type corresponding to the matching result as the target attitude sensing analysis result.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages: In this application, by obtaining the historical attitude sensing data set and constructing a prototype set of the attitude sensing warning model based on the historical attitude sensing data set, where each prototype of the attitude sensing warning model corresponds to an attitude sensing anomaly type, and then respectively obtaining the three-axis data sequence of the angular velocity and the three-axis data sequence of the acceleration of the device within a preset monitoring window through the gyroscope and the accelerometer arranged on the device, and then respectively performing centralized screening analysis on the three-axis data sequence of the angular velocity and the three-axis data sequence of the acceleration to determine the three-axis centralized screening data set of the angular velocity and the three-axis centralized screening data set of the acceleration, and then performing attitude analysis based on the three-axis centralized screening data set of the angular velocity and the three-axis centralized screening data set of the acceleration to obtain the pitch angle, roll angle and yaw angle set, and matching the pitch angle, roll angle and yaw angle set with the prototype set of the attitude sensing warning model, and using the attitude sensing anomaly type corresponding to the matching result as the target attitude sensing analysis result. It achieves the technical effect of improving the reliability of the analysis data by performing data screening, and further improving the accuracy of the attitude sensing analysis. Description of the Drawings

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for description in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0017] Figure 1 Schematic flow diagram of the attitude sensing analysis method based on gyroscope and accelerometer provided by the embodiments of the present application; Figure 2 Schematic structural diagram of the attitude sensing analysis system based on gyroscope and accelerometer provided by the embodiments of the present application.

[0018] Explanation of reference numerals: prototype set construction module 11, data sequence acquisition module 12, centralized screening data acquisition module 13, yaw angle set acquisition module 14, target attitude sensing analysis result acquisition module 15. Detailed implementation manners

[0019] The present application provides an attitude sensing analysis method and system based on gyroscope and accelerometer, which are used to solve the technical problems of long cycle and low analysis reliability in attitude sensing analysis in the prior art.

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0021] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices.

[0022] Embodiment 1, as Figure 1 shown, the present application provides an attitude sensing analysis method based on gyroscope and accelerometer, wherein the method includes: Step S100: Obtain a historical attitude sensing data set, and construct a prototype set of attitude sensing early warning models based on the historical attitude sensing data set, where each prototype of the attitude sensing early warning model corresponds to an attitude sensing abnormal type.

[0023] In a possible embodiment, the historical attitude sensing data set is the analysis result of the data collected by the gyroscope and accelerometer of the device within the historical time, as well as the obtained attitude sensing result. The historical attitude sensing data set reflects the attitude changes of the device in various environments and working states, and is the basic data for training the attitude warning model. The attitude sensing warning model prototype set is a set containing multiple model prototypes, and each prototype represents a specific type of attitude anomaly, such as excessive vibration, excessive tilt, etc. By performing clustering learning analysis on the historical attitude sensing data set, the constructed attitude sensing warning model prototype set is obtained, achieving the technical effect of providing data support for the rapid attitude sensing analysis of subsequent real-time attitude data. The attitude sensing anomaly type refers to various abnormal states that may occur during the use of the device, such as the tilt angle of the device exceeding a certain threshold, abnormal vibration of the device, etc.

[0024] Preferably, by integrating and analyzing the attitude data of a type of attitude sensing anomaly in the historical state, the attitude sensing warning model prototype of this type of attitude sensing anomaly is determined, thereby providing a prototype comparison basis for subsequent feature matching.

[0025] Further, to obtain the historical attitude sensing data set and construct the attitude sensing warning model prototype set based on the historical attitude sensing data set, step S100 of the embodiment of the present application further includes: Aggregating the historical attitude sensing data set for the same type of attitude sensing anomaly to obtain L aggregated historical attitude sensing data sets, where L is a positive integer; Extracting the pitch angle, roll angle, and yaw angle sets respectively from the L aggregated historical attitude sensing data sets to obtain a set of L aggregated historical pitch-roll-yaw angle set data groups; Using the prototype network to learn from the set of L aggregated historical pitch-roll-yaw angle set data groups respectively to construct the attitude sensing warning model prototype set.

[0026] Further, using the prototype network to learn from the set of L aggregated historical pitch-roll-yaw angle set data groups respectively to construct the attitude sensing warning model prototype set, step S100 of the embodiment of the present application further includes: Extracting the first aggregated historical pitch-roll-yaw angle set data group set from the set of L aggregated historical pitch-roll-yaw angle set data groups; Inputting the first aggregated historical pitch-roll-yaw angle set data group set into the prototype network construction function for metric space learning to obtain the first attitude sensing warning model prototype; Performing metric space learning on the set of L aggregated historical pitch-roll-yaw angle set data groups to obtain the attitude sensing warning model prototype set.

[0027] Furthermore, the prototype network construction function is as follows: ; wherein, is the first aggregated historical pitch-roll-yaw angle set data group set, is the number of data groups in the first aggregated historical pitch-roll-yaw angle set data group set, is a positive integer, is the th first aggregated historical pitch-roll-yaw angle set data group, is an embedding function, which is used to map the input first aggregated historical pitch-roll-yaw angle set data group set to a three-dimensional embedded feature space through a prototype network, are the learning parameters of the prototype network model.

[0028] In a possible embodiment, taking the posture sensing anomaly type as an index, the anomaly types existing in each historical posture sensing data in the historical posture sensing data set are extracted, and the historical posture sensing data belonging to the same posture sensing type (such as angle offset, excessive tilt, excessive vibration, static drift, etc.) are divided into the same set, obtaining L aggregated historical posture sensing data sets. Wherein, L is the total number of all posture sensing anomaly types in the historical posture sensing data set.

[0029] Furthermore, taking the pitch angle, roll angle and yaw angle set as an index, the data in the L aggregated historical posture sensing data sets are retrieved to determine the posture situation when an anomaly occurs in each aggregated historical posture sensing data, obtaining L aggregated historical pitch-roll-yaw angle set data group sets. Wherein, each aggregated historical pitch-roll-yaw angle set data group reflects the overall posture situation of the device when an anomaly occurs. The L aggregated historical pitch-roll-yaw angle set data group sets are subjected to feature mean recognition through a prototype network to determine the general situation of the posture features conforming to each posture sensing anomaly type, thereby constructing a prototype of the posture sensing early warning model for each posture sensing anomaly type.

[0030] Preferably, the prototype network learns the distance between different first aggregated historical pitch-roll-yaw angle set data groups within the first aggregated historical pitch-roll-yaw angle set data group set, and uses the prototype network construction function to obtain the centroid of the data within the set (the most representative data feature, that is, the prototype). That is, the prototype network construction function is used to learn the attitude features within the first aggregated historical pitch-roll-yaw angle set data group set and balance the similarity between the features, thereby constructing the prototype of the first attitude sensing and warning model. When the real-time feature matches the prototype of the first attitude sensing and warning model successfully, it indicates that the device is in the attitude sensing abnormal type corresponding to the prototype of the first attitude sensing and warning model.

[0031] Preferably, the embedding function is the core of the prototype network and is used to map the historical pitch angle, historical roll angle, and historical yaw angle sets in the first aggregated historical pitch-roll-yaw angle set data group to a three-dimensional feature space, thereby facilitating subsequent distance measurement. is the learning parameter automatically obtained during the training process when the prototype network learns the mapping relationship between features and attitude sensing abnormal types. The prototype network construction function is used to learn the general correspondence between the data features and attitude sensing abnormal types in the first aggregated historical pitch-roll-yaw angle set data group set, thereby obtaining the prototype of the first attitude sensing and warning model.

[0032] Based on the same principle as obtaining the prototype of the first attitude sensing and warning model, similar learning is performed on L aggregated historical pitch-roll-yaw angle set data group sets to construct a complete set of attitude sensing and warning model prototypes. By providing the prototype features of each attitude sensing abnormal type, a benchmark is provided for the real-time attitude data analysis of the device. Through the matching of the real-time attitude data of the device with these prototypes, it can be identified whether the device has abnormalities and warning information can be sent in a timely manner, thereby improving the operation safety and reliability of the device.

[0033] Step S200: Respectively obtain the angular velocity three-axis data sequence and the acceleration three-axis data sequence of the device within a preset monitoring window through the gyroscope and accelerometer arranged on the device.

[0034] Furthermore, each angular velocity three-axis data includes three angular velocity data of the device on the x-axis, y-axis, and z-axis respectively, and each acceleration three-axis data includes three acceleration data of the device on the x-axis, y-axis, and z-axis respectively.

[0035] In one embodiment, the preset monitoring window is a sampling time period preset by those skilled in the art. Data of the gyroscope and the accelerometer are collected at a preset sampling frequency within the preset monitoring window, so as to obtain a three-axis angular velocity data sequence and a three-axis acceleration data sequence of the device. Each three-axis angular velocity data includes three angular velocity data of the device on the x-axis, y-axis, and z-axis respectively, and each three-axis acceleration data includes three acceleration data of the device on the x-axis, y-axis, and z-axis respectively. The three-axis angular velocity data sequence reflects the change of the angular velocity of the device on the three axes. The three-axis acceleration data sequence reflects the change of the acceleration of the device on the three axes. The technical effect of providing data support for subsequent device attitude analysis is achieved.

[0036] Step S300: Perform centralized screening and analysis on the three-axis angular velocity data sequence and the three-axis acceleration data sequence respectively to determine a three-axis angular velocity centralized screening data set and a three-axis acceleration centralized screening data set.

[0037] In one embodiment, after obtaining the three-axis angular velocity data sequence and the three-axis acceleration data sequence, it is necessary to perform a general situation analysis on the angular velocity data and the acceleration data of the device within the preset monitoring window, that is, to screen out the data that can best represent the device attitude during this period, and obtain the three-axis angular velocity centralized screening data set and the three-axis acceleration centralized screening data set. The three-axis angular velocity centralized screening data set includes a three-axis angular velocity centralized screening data set on the x-axis, a three-axis angular velocity centralized screening data set on the y-axis, and a three-axis angular velocity centralized screening data set on the z-axis. The three-axis acceleration centralized screening data set includes a three-axis acceleration centralized screening data set on the x-axis, a three-axis acceleration centralized screening data set on the y-axis, and a three-axis acceleration centralized screening data set on the z-axis.

[0038] Further, when performing centralized screening and analysis on the three-axis angular velocity data sequence and the three-axis acceleration data sequence respectively to determine a three-axis angular velocity centralized screening data set and a three-axis acceleration centralized screening data set, step S300 of the embodiment of the present application further includes: Extract the mode of the three-axis angular velocity data sequence and the three-axis acceleration data sequence respectively to obtain a three-axis angular velocity mode data set and a three-axis acceleration mode data set with timestamp identifiers; Extract the x-axis angular velocity mode set from the three-axis angular velocity mode data set with timestamp identifiers, and extract the x-axis angular velocity data sequence from the three-axis angular velocity data sequence; Combined with the timestamp identifier, perform two-dimensional constraint centralized screening of data similarity and time approximation on the x-axis angular velocity mode set in the x-axis angular velocity data sequence to obtain the x-axis angular velocity centralized screening data; In the angular velocity three-axis data sequence and the acceleration three-axis data sequence, perform data similarity and time approximation two-dimensional constraint centralized screening on the angular velocity three-axis mode data set and the acceleration three-axis mode data set with timestamp identifiers to obtain the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set.

[0039] Further, in combination with the timestamp identifier, perform data similarity and time approximation two-dimensional constraint centralized screening on the angular velocity x-axis mode set in the angular velocity x-axis data sequence to obtain the angular velocity x-axis centralized screening data. Step S300 of the embodiment of the present application further includes: In combination with the timestamp identifier, use the two-dimensional approximation analysis function to traverse and perform approximation analysis on the angular velocity x-axis data in the angular velocity x-axis data sequence and the angular velocity x-axis mode set to obtain a two-dimensional approximation coefficient sequence; Add the angular velocity x-axis data in the two-dimensional approximation coefficient sequence that is greater than or equal to the preset two-dimensional approximation coefficient threshold to the x-axis centralized screening set; Calculate the mean value of the x-axis centralized screening set to obtain the angular velocity x-axis centralized screening data.

[0040] Further, the two-dimensional approximation analysis function is: ; Wherein, is the two-dimensional approximation coefficient, is the number of angular velocity x-axis data in the angular velocity x-axis data sequence, is a positive integer, is any angular velocity x-axis data in the angular velocity x-axis data sequence, is the th angular velocity x-axis mode in the angular velocity x-axis mode set, is the timestamp corresponding to any angular velocity x-axis data in the angular velocity x-axis data sequence, is the th timestamp corresponding to the angular velocity x-axis mode in the angular velocity x-axis mode set, is the weight for balancing data similarity and time approximation.

[0041] In a possible embodiment, the mode is the data with the highest frequency of occurrence in a set of data. That is to say, compared with other data in a set of data, the mode is more representative. Therefore, it is necessary to extract the mode from the angular velocity three-axis data sequence and the acceleration three-axis data sequence respectively, and in order to improve the reliability of subsequent nearest neighbor analysis, synchronously extract the time points where the extracted modes are located to obtain the angular velocity three-axis mode data set and the acceleration three-axis mode data set with timestamp identifiers. Among them, the timestamp identifier is used to describe the data acquisition time.

[0042] Extract the x-axis mode set of angular velocity from the three-axis mode data set of angular velocity with timestamp identification, and extract the x-axis data sequence of angular velocity from the three-axis data sequence of angular velocity, so as to screen the angular velocity data on the x-axis. During the centralized screening process, screen and analyze the representative data other than the mode from two dimensions of data similarity and time approximation, so as to obtain the centralized screening data of the x-axis of the angular velocity.

[0043] Preferably, the two-dimensional approximation analysis function is used to analyze the overall approximation of each x-axis angular velocity data in the x-axis data sequence of angular velocity to the x-axis mode set of angular velocity from two dimensions of data similarity and time approximation, and obtain the two-dimensional approximation coefficient sequence. Among them, the two-dimensional approximation coefficient reflects the degree to which the x-axis angular velocity data can be used as representative data. The larger the two-dimensional approximation coefficient, the higher the quality of the corresponding x-axis angular velocity data. After performing approximate analysis on all the x-axis angular velocity data in the x-axis data sequence of angular velocity, the two-dimensional approximation coefficient sequence is obtained.

[0044] The preset two-dimensional approximation coefficient threshold is the lowest two-dimensional approximation coefficient that can be used as reliable data preset by those skilled in the art. Add the x-axis angular velocity data in the two-dimensional approximation coefficient sequence that is greater than or equal to the preset two-dimensional approximation coefficient threshold into the x-axis centralized screening set, and perform mean calculation to obtain the centralized screening data of the x-axis of the angular velocity.

[0045] Based on the same principle as obtaining the centralized screening data of the x-axis of the angular velocity, screen the three-axis data sequence of angular velocity and the three-axis data sequence of acceleration respectively to obtain the three-axis centralized screening data set of angular velocity and the three-axis centralized screening data set of acceleration.

[0046] Step S400: Perform attitude analysis based on the three-axis centralized screening data set of angular velocity and the three-axis centralized screening data set of acceleration to obtain the pitch angle, roll angle and yaw angle set.

[0047] Furthermore, step S400 of the embodiment of the present application further includes: Analyze the three-axis centralized screening data set of acceleration by using the pitch angle calculation formula to obtain the pitch angle; Analyze the three-axis centralized screening data set of acceleration by using the roll angle calculation formula to obtain the roll angle; Analyze the three-axis centralized screening data set of angular velocity by using the yaw angle calculation formula to obtain the yaw angle set.

[0048] Preferably, after obtaining the angular velocity three-axis concentrated screening data set and the acceleration three-axis concentrated screening data set, that is, the attitude analysis basic data, the pitch angle calculation formula, the roll angle calculation formula, and the yaw angle calculation formula are respectively used to calculate each angle.

[0049] Among them, the pitch angle calculation formula is: ; The roll angle calculation formula is: ; Among them, is the pitch angle, is the roll angle, is the acceleration x-axis concentrated screening data, is the acceleration y-axis concentrated screening data, is the acceleration z-axis concentrated screening data.

[0050] The yaw angle calculation formula is: ; Among them, is the z-axis yaw angle of the device, is the z-axis yaw angle of the device at the previous attitude analysis, is the gyroscope angular velocity z-axis concentrated screening data.

[0051] Based on the same principle of obtaining the z-axis yaw angle of the device, the x-axis yaw angle and the y-axis yaw angle of the device are respectively calculated. The x-axis yaw angle, the y-axis yaw angle, and the z-axis yaw angle are summarized to obtain the yaw angle set.

[0052] Step S500: Match the pitch angle, roll angle, and yaw angle set with the attitude sensing early warning model prototype set, and use the attitude sensing abnormal type corresponding to the matching result as the target attitude sensing analysis result.

[0053] Using the embedding function Perform feature mapping on the pitch angle, roll angle, and yaw angle set to obtain real-time attitude features. Calculate the cosine similarity between the real-time attitude features and the attitude sensing early warning model prototype set respectively to obtain a similarity set. Use the attitude sensing early warning model prototype corresponding to the maximum value in the similarity set as the matching result, and use the attitude sensing abnormal type corresponding to the matching result as the target attitude sensing analysis result.

[0054] By performing prototype matching, the response speed and accuracy to the attitude change of the device are achieved, ensuring the timely discovery of potential faults or abnormalities and guaranteeing the safe and stable operation of the device.

[0055] In summary, the embodiments of the present application at least have the following technical effects: In the present application, a historical attitude sensing data set is obtained, and an attitude sensing early warning model prototype set is constructed based on the historical attitude sensing data set. Each attitude sensing early warning model prototype corresponds to an attitude sensing abnormal type. Then, the angular velocity three-axis data sequence and the acceleration three-axis data sequence of the device within a preset monitoring window are respectively obtained through the gyroscope and the accelerometer disposed on the device. Furthermore, data centralized screening and analysis are respectively performed on the angular velocity three-axis data sequence and the acceleration three-axis data sequence to determine the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set. Then, attitude analysis is performed based on the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set to obtain the pitch angle, roll angle, and yaw angle sets. According to the pitch angle, roll angle, and yaw angle sets and the attitude sensing early warning model prototype set, the attitude sensing abnormal type corresponding to the matching result is used as the target attitude sensing analysis result. The technical effect of improving the reliability of the analysis data and further improving the accuracy of the attitude sensing analysis is achieved by performing data screening.

[0056] Embodiment 2, based on the same inventive concept as the attitude sensing analysis method based on a gyroscope and an accelerometer in the foregoing embodiment, as Figure 2 shown, the present application provides an attitude sensing analysis system based on a gyroscope and an accelerometer. The system in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the system includes: A prototype set construction module 11, configured to obtain a historical attitude sensing data set and construct an attitude sensing early warning model prototype set based on the historical attitude sensing data set, where each attitude sensing early warning model prototype corresponds to an attitude sensing abnormal type; A data sequence obtaining module 12, configured to respectively obtain the angular velocity three-axis data sequence and the acceleration three-axis data sequence of the device within a preset monitoring window through the gyroscope and the accelerometer disposed on the device; A centralized screening data obtaining module 13, configured to respectively perform data centralized screening and analysis on the angular velocity three-axis data sequence and the acceleration three-axis data sequence to determine the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set; A yaw angle set obtaining module 14, configured to perform attitude analysis based on the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set to obtain the pitch angle, roll angle, and yaw angle sets; A target attitude sensing analysis result obtaining module 15, configured to match the pitch angle, roll angle, and yaw angle sets with the attitude sensing early warning model prototype set, and use the attitude sensing abnormal type corresponding to the matching result as the target attitude sensing analysis result.

[0057] Further, the prototype set construction module 11 is used to execute the following steps: Aggregate the same type of attitude sensing anomaly types in the historical attitude sensing data set to obtain L aggregated historical attitude sensing data sets, where L is a positive integer; Extract the pitch angle, roll angle, and yaw angle sets from the L aggregated historical attitude sensing data sets respectively to obtain a set of L aggregated historical pitch-roll-yaw angle set data groups; Use the prototype network to learn from the set of L aggregated historical pitch-roll-yaw angle set data groups respectively to construct a prototype set of the attitude sensing early warning model.

[0058] Further, the prototype set construction module 11 is used to execute the following steps: Extract the first aggregated historical pitch-roll-yaw angle set data group set from the set of L aggregated historical pitch-roll-yaw angle set data groups; Input the first aggregated historical pitch-roll-yaw angle set data group set into the prototype network construction function for metric space learning to obtain the first attitude sensing early warning model prototype; Perform metric space learning on the set of L aggregated historical pitch-roll-yaw angle set data groups to obtain a prototype set of the attitude sensing early warning model.

[0059] Further, the prototype network construction function is: ; Where is the first aggregated historical pitch-roll-yaw angle set data group set, is the number of data groups in the first aggregated historical pitch-roll-yaw angle set data group set, is a positive integer, is the th first aggregated historical pitch-roll-yaw angle set data group, is the embedding function, which is used to map the input first aggregated historical pitch-roll-yaw angle set data group set into a three-dimensional embedded feature space through a prototype network, is the learning parameter of the prototype network model.

[0060] Further, the centralized screening data acquisition module 13 is used to execute the following steps: Extract the mode of the angular velocity three-axis data sequence and the acceleration three-axis data sequence respectively to obtain an angular velocity three-axis mode data set with timestamp identification and an acceleration three-axis mode data set; Extract the angular velocity x-axis mode set from the angular velocity three-axis mode data set with timestamp identification, and extract the angular velocity x-axis data sequence from the angular velocity three-axis data sequence; Combined with the timestamp identification, perform data similarity and time approximation two-dimensional constraint centralized screening on the angular velocity x-axis mode set in the angular velocity x-axis data sequence to obtain the angular velocity x-axis centralized screening data; Perform data similarity and time approximation two-dimensional constraint centralized screening on the angular velocity three-axis mode data set and the acceleration three-axis mode data set with timestamp identification in the angular velocity three-axis data sequence and the acceleration three-axis data sequence to obtain the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set.

[0061] Further, the centralized screening data acquisition module 13 is used to execute the following steps: Combined with the timestamp identification, use the two-dimensional approximation analysis function to traverse and perform approximation analysis on the angular velocity x-axis data in the angular velocity x-axis data sequence and the angular velocity x-axis mode set to obtain the two-dimensional approximation coefficient sequence; Add the angular velocity x-axis data in the two-dimensional approximation coefficient sequence that is greater than or equal to the preset two-dimensional approximation coefficient threshold to the x-axis centralized screening set; Calculate the mean value of the x-axis centralized screening set to obtain the angular velocity x-axis centralized screening data.

[0062] Further, the two-dimensional approximation analysis function is: ; Among them, is the two-dimensional approximation coefficient, is the number of angular velocity x-axis data in the angular velocity x-axis data sequence, is a positive integer, is any angular velocity x-axis data in the angular velocity x-axis data sequence, is the th angular velocity x-axis mode in the angular velocity x-axis mode set, is the timestamp corresponding to any angular velocity x-axis data in the angular velocity x-axis data sequence, is the th timestamp corresponding to the angular velocity x-axis mode in the angular velocity x-axis mode set, is the weight for balancing data similarity and time approximation.

[0063] Further, the yaw angle set acquisition module 14 is used to execute the following steps: Use the pitch angle calculation formula to analyze the acceleration three-axis centralized screening data set to obtain the pitch angle; Analyze the acceleration three-axis concentrated screening data set using the roll angle calculation formula to obtain the roll angle; Analyze the angular velocity three-axis concentrated screening data set using the yaw angle calculation formula to obtain the yaw angle set.

[0064] Furthermore, each angular velocity three-axis data includes three angular velocity data of the device on the x-axis, y-axis, and z-axis respectively, and each acceleration three-axis data includes three acceleration data of the device on the x-axis, y-axis, and z-axis respectively.

[0065] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0066] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0067] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A posture sensing analysis method based on a gyroscope accelerometer, characterized in that: The method comprises: Acquire a historical posture sensing data set, and construct a posture sensing early warning model prototype set based on the historical posture sensing data set, wherein each posture sensing early warning model prototype corresponds to a posture sensing abnormality type; The gyroscope and accelerometer disposed on the device are used to obtain the angular velocity triaxial data sequence and the acceleration triaxial data sequence of the device within the preset monitoring window respectively; Performing data concentration screening and analysis on the angular velocity three-axis data sequence and the acceleration three-axis data sequence respectively, to determine an angular velocity three-axis concentrated screening data set and an acceleration three-axis concentrated screening data set; Performing attitude analysis based on the angular velocity three-axis concentrated screening data set and the acceleration three-axis concentrated screening data set to obtain a pitch angle, a roll angle, and a yaw angle set; The pitch angle, roll angle and yaw angle set are matched with the attitude sensing warning model prototype set, and the attitude sensing abnormality type corresponding to the matching result is used as the target attitude sensing analysis result.

2. The gyro-accelerometer-based attitude sensing analysis method according to claim 1, characterized in that: Acquiring a historical posture sensing data set, and constructing a posture sensing warning model prototype set based on the historical posture sensing data set, including: Aggregating the same type of anomaly types of posture sensing on the historical posture sensing data sets to obtain L aggregated historical posture sensing data sets, where L is a positive integer; Extracting pitch angle, roll angle and yaw angle sets from the L aggregated historical attitude sensing data sets respectively to obtain L aggregated historical pitch angle-roll angle-yaw angle set data set sets; The prototype network is used to learn the L aggregated historical pitch angle-roll angle-yaw angle set data sets respectively to construct a prototype set of attitude sensing warning model.

3. The gyro-accelerometer-based attitude sensing analysis method according to claim 2, characterized in that: The prototype network is used to learn the L aggregated historical pitch angle-roll angle-yaw angle set data sets respectively to construct a prototype set of attitude sensing warning models, including: Extracting a first aggregated historical pitch angle-roll angle-yaw angle set data set from the L aggregated historical pitch angle-roll angle-yaw angle set data set sets; Inputting the first aggregated historical pitch angle-roll angle-yaw angle set data set into the prototype network construction function to perform metric space learning, and obtaining a first attitude sensing warning model prototype; Metric space learning is performed on the L aggregated historical pitch angle-roll angle-yaw angle set data group sets to obtain a prototype set of attitude sensing warning models.

4. The gyro-accelerometer-based attitude sensing analysis method according to claim 3, characterized in that: The prototype network construction function is: ； in, is the first aggregated historical pitch angle-roll angle-yaw angle set data set, is the number of data sets in the first aggregated historical pitch angle-roll angle-yaw angle set data set, is a positive integer, For the The first aggregated historical pitch-roll-yaw angle set data set, is the embedding function, For mapping the input first aggregated historical pitch angle-roll angle-yaw angle set data set into a three-dimensional embedding feature space through a prototype network, are the learning parameters of the prototype network model.

5. The gyro-accelerometer-based attitude sensing analysis method according to claim 1, characterized in that: The angular velocity three-axis data sequence and the acceleration three-axis data sequence are respectively subjected to data concentration screening and analysis to determine an angular velocity three-axis concentrated screening data set and an acceleration three-axis concentrated screening data set, including: Performing mode extraction on the angular velocity three-axis data sequence and the acceleration three-axis data sequence respectively to obtain an angular velocity three-axis mode data set and an acceleration three-axis mode data set with a timestamp; Extracting an angular velocity x-axis mode set from an angular velocity three-axis mode data set with a timestamp, and extracting an angular velocity x-axis data sequence from the angular velocity three-axis data sequence; Combined with the timestamp identifier, the angular velocity x-axis mode set is subjected to data similarity and time approximation dual-dimensional constraint centralized screening in the angular velocity x-axis data sequence to obtain angular velocity x-axis centralized screening data; In the angular velocity three-axis data sequence and the acceleration three-axis data sequence, the angular velocity three-axis mode data set and the acceleration three-axis mode data set with timestamp identification are subjected to data similarity and time approximation two-dimensional constraint centralized screening to obtain the angular velocity three-axis centralized screening data set and the acceleration three-axis centralized screening data set.

6. The gyro-accelerometer-based attitude sensing analysis method according to claim 5, characterized in that: Combined with the timestamp identifier, the angular velocity x-axis mode set is subjected to data similarity and time approximation dual-dimensional constraint centralized screening in the angular velocity x-axis data sequence to obtain angular velocity x-axis centralized screening data, including: In combination with the timestamp identifier, a two-dimensional approximate analysis function is used to traverse the angular velocity x-axis data in the angular velocity x-axis data sequence and the angular velocity x-axis mode set for approximate analysis to obtain a two-dimensional approximate coefficient sequence; Adding the angular velocity x-axis data in the two-dimensional approximation coefficient sequence that is greater than or equal to a preset two-dimensional approximation coefficient threshold into the x-axis concentrated screening set; The mean of the x-axis concentrated screening set is calculated to obtain the angular velocity x-axis concentrated screening data.

7. The gyro-accelerometer-based attitude sensing analysis method according to claim 6, characterized in that: The two-dimensional approximate analysis function is: ； in, is the two-dimensional approximation coefficient, is the number of angular velocity x-axis data in the angular velocity x-axis data sequence, is a positive integer, is any angular velocity x-axis data in the angular velocity x-axis data sequence, is the mode set of angular velocity x-axis The x-axis mode of the angular velocity, is the timestamp corresponding to any angular velocity x-axis data in the angular velocity x-axis data sequence, is the mode set of angular velocity x-axis The timestamp corresponding to the mode of the angular velocity x-axis, The weight is to balance data similarity and time proximity.

8. The gyro-accelerometer-based attitude sensing analysis method according to claim 1, characterized in that: include: The pitch angle calculation formula is used to analyze the acceleration three-axis concentrated screening data set to obtain the pitch angle; The roll angle calculation formula is used to analyze the acceleration triaxial concentrated screening data set to obtain the roll angle; The yaw angle calculation formula is used to analyze the angular velocity three-axis concentrated screening data set to obtain a yaw angle set.

9. The gyro-accelerometer-based attitude sensing analysis method according to claim 1, characterized in that: Each angular velocity triaxial data includes three angular velocity data of the device on the x-axis, y-axis and z-axis respectively, and each acceleration triaxial data includes three acceleration data of the device on the x-axis, y-axis and z-axis respectively.

10. The attitude sensing and analysis system based on gyroscope and accelerometer is characterized in that: For implementing the attitude sensing analysis method based on a gyro accelerometer according to any one of claims 1 to 9, the system comprises: A prototype set construction module is used to obtain a historical posture sensing data set, and to construct a posture sensing warning model prototype set based on the historical posture sensing data set, wherein each posture sensing warning model prototype corresponds to a posture sensing anomaly type; The data sequence acquisition module is used to respectively acquire the angular velocity triaxial data sequence and the acceleration triaxial data sequence of the device within a preset monitoring window through the gyroscope and accelerometer arranged on the device; A centralized screening data acquisition module is used to perform data centralized screening analysis on the angular velocity three-axis data sequence and the acceleration three-axis data sequence, respectively, to determine an angular velocity three-axis centralized screening data set and an acceleration three-axis centralized screening data set; A yaw angle set acquisition module, used for performing attitude analysis based on the angular velocity three-axis concentrated screening data set and the acceleration three-axis concentrated screening data set to obtain a pitch angle, a roll angle and a yaw angle set; The target attitude sensing analysis result acquisition module is used to match the pitch angle, roll angle and yaw angle set with the attitude sensing warning model prototype set, and use the attitude sensing abnormality type corresponding to the matching result as the target attitude sensing analysis result.

Images