Transportation product safety assessment method based on flight attitude

Abstract

The invention discloses a transportation product safety assessment method based on a flight attitude, and relates to the technical field of data processing, and the method comprises the steps: employing a flight attitude sensing array and an environment sensing array to carry out the synchronous real-time sensing of the attitude change and environment state of a target unmanned plane in a preset monitoring window according to a preset monitoring interval, obtaining a flight attitude sensing data set sequence and an environment sensing data set sequence; constructing a preset flight attitude framework sequence; obtaining a flight attitude sensing frame sequence; determining a flight attitude frame deviation degree sequence; determining a flight attitude deviation influence coefficient; determining an environment deviation influence coefficient; and carrying out comprehensive analysis to obtain a transportation product safety assessment result. The technical problem that in the prior art, the unmanned aerial vehicle flight attitude is lack of attitude correlation analysis, so that the safety evaluation accuracy of the transportation product is low is solved, and the technical effect of reliably evaluating the safety of the transportation product according to the unmanned aerial vehicle flight attitude is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for evaluating the safety of a transportation product based on flight attitude. Background Art

[0002] At present, the analysis of UAV flight attitude often starts from a single dimension, such as analyzing individual factors based on the pitch angle, roll angle of the UAV or the temperature and air pressure in the environment. The complex interaction between these factors during the flight process is ignored. For example, in some cases, a sudden change in flight attitude may be caused by environmental factors (such as changes in wind speed or temperature), and the normal flight attitude will be restored at the next time point. If the flight attitude or environmental factors are considered separately, it may lead to misjudgment or underestimate the actual safety risks.

[0003] The existing technology has the technical problem of lack of posture correlation analysis of UAV flight posture, resulting in low accuracy of safety assessment of transportation products. Summary of the invention

[0004] The present application provides a method for transport product safety assessment based on flight attitude, which is used to solve the technical problem in the prior art that there is a lack of attitude correlation analysis on the flight attitude of unmanned aerial vehicles, resulting in low accuracy of transport product safety assessment.

[0005] In view of the above problems, the present application provides a method for safety assessment of transportation products based on flight attitude, the method comprising: Using the flight attitude sensor array and the environment sensor array to synchronously sense the attitude changes and environmental conditions of the target UAV in a preset monitoring window in real time according to the preset monitoring interval, and obtain a flight attitude sensor data set sequence and an environment sensor data set sequence; Acquire the preset flight path and preset flight parameters of the target UAV, and construct a preset flight attitude frame sequence in combination with the preset monitoring interval and the preset monitoring window; Traversing the flight attitude sensing data set sequence to perform attitude analysis and obtain a flight attitude sensing frame sequence; Performing mapping deviation analysis on the preset flight attitude frame sequence and the flight attitude sensing frame sequence to determine a flight attitude frame deviation degree sequence; Performing a comprehensive analysis of the dual effects of deviation superposition effect and concentrated deviation degree on the flight attitude frame deviation degree sequence to determine the flight attitude deviation influence coefficient; Performing a neighbor interaction correlation trend analysis on the environmental sensor data set sequence from front to back to determine a target environmental interaction correlation sensor data set, performing an environmental deviation analysis on a preset environmental data set corresponding to the preset flight parameters to determine an environmental deviation impact coefficient; The product safety assessment network layer is used to conduct a comprehensive analysis on the flight attitude deviation influence coefficient and the environment deviation influence coefficient to obtain a transportation product safety assessment result.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: The present application uses a flight attitude sensor array and an environment sensor array to synchronously sense the attitude changes and environmental states of a target UAV in a preset monitoring window in real time according to a preset monitoring interval, obtains a flight attitude sensor data set sequence and an environment sensor data set sequence, then obtains a preset flight path and preset flight parameters of the target UAV, combines the preset monitoring interval and the preset monitoring window, constructs a preset flight attitude frame sequence, and then traverses the flight attitude sensor data set sequence to perform attitude analysis to obtain a flight attitude sensor frame sequence, determines a flight attitude frame deviation degree sequence by performing mapping deviation analysis on the preset flight attitude frame sequence and the flight attitude sensor frame sequence, and then performs a comprehensive analysis of the double effects of the deviation superposition effect and the concentrated deviation degree on the flight attitude frame deviation degree sequence to determine the flight attitude deviation influence coefficient, and then performs a neighbor interaction correlation trend analysis on the environment sensor data set sequence in a front-to-back order to determine the target environment interaction correlation sensor data set, performs an environmental deviation analysis in combination with the preset environment data set corresponding to the preset flight parameters to determine the environmental deviation influence coefficient, and uses a product safety assessment network layer to perform a comprehensive analysis on the flight attitude deviation influence coefficient and the environmental deviation influence coefficient to obtain a transportation product safety assessment result. The technical effect of improving the reliability of transportation product safety assessment has been achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0008] Figure 1 A schematic diagram of a flow chart of a method for safety assessment of a transportation product based on flight attitude provided in an embodiment of the present application; Figure 2 A schematic diagram of a process for determining concentrated deviations from a straight line in a method for safety assessment of a transportation product based on flight attitude provided in an embodiment of the present application. DETAILED DESCRIPTION

[0009] The present application provides a transportation product safety assessment method based on flight attitude, which is used to solve the technical problem that the prior art lacks attitude correlation analysis of UAV flight attitude, resulting in low accuracy of transportation product safety assessment.

[0010] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0011] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, method, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0012] Examples, such as Figure 1 As shown, the present application provides a method for safety assessment of transportation products based on flight attitude, wherein the method comprises: S100: using the flight attitude sensor array and the environment sensor array to synchronously and real-time sense the attitude change and environment state of the target UAV in a preset monitoring window according to a preset monitoring interval, and obtaining a flight attitude sensor data set sequence and an environment sensor data set sequence; In a possible embodiment, the target UAV is any type of UAV for product transportation. The preset monitoring interval is the time interval between two adjacent monitorings when monitoring the flight attitude and flight environment of the target UAV, which is preset by a person skilled in the art, and can be 3 seconds, 5 seconds, etc. Preferably, the preset monitoring interval should be set according to factors such as the flight speed of the UAV and the accuracy of the sensor. The preset monitoring time is the time period for monitoring the flight attitude and flight environment of the target UAV, which is preset by a person skilled in the art, and can be 3 minutes, 6 minutes, 10 minutes, etc. The flight attitude sensor array is an array composed of sensors for monitoring the flight attitude of the UAV, including gyroscopes, accelerometers, magnetometers, etc. The parameters of each sensor include but are not limited to measurement range, sensitivity, resolution, etc. The environment sensor array is an array composed of sensors for monitoring the flight environment of the UAV, including temperature sensors, air pressure sensors, humidity sensors, light sensors, wind speed sensors, etc.

[0013] In one embodiment, the flight attitude sensor array and the environmental sensor array are started to ensure that they work normally and are synchronized. The data of each sensor is connected to the data acquisition component through the hardware interface, and each sensor is calibrated to ensure the accuracy of the measurement. Within the preset monitoring interval, data is collected from the flight attitude sensor and the environmental sensor array at the same time. During data collection, the time synchronization of the flight attitude data and the environmental data should be ensured, that is, they have the same timestamp, and the flight attitude sensor data set sequence and the environmental sensor data set sequence are obtained. Among them, the flight attitude sensor data set sequence reflects the change of the flight attitude of the target drone within the preset monitoring window over time. The environmental sensor data set sequence reflects the change of the flight environment of the target drone within the preset monitoring window over time.

[0014] Preferably, each flight attitude sensor data set includes the attitude angle (such as pitch angle, roll angle, yaw angle), acceleration, angular velocity and other data of the drone. Each environmental sensor data set includes environmental variables such as air pressure, temperature, humidity, wind speed, etc. By obtaining the flight attitude sensor data set sequence and the environmental sensor data set sequence, the technical effect of providing basic analysis data for subsequent flight attitude deviation analysis and environmental state development trend analysis is achieved.

[0015] For example, suppose that the target UAV flies from point A to point B according to the preset flight path and preset flight parameters in a certain flight mission. During this process, the real-time flight attitude of the UAV is collected through the flight attitude sensor array (such as recording the pitch angle, roll angle and yaw angle of the UAV once a second), and the temperature, humidity, air pressure and other data of the environment in which the UAV is located during the flight are recorded through the environmental sensor array. The whole process is carried out synchronously according to the preset monitoring interval (for example, collecting data once a second), and the flight attitude sensor data set sequence and environmental sensor data set sequence finally obtained can provide a data basis for the subsequent flight attitude analysis and environmental deviation analysis.

[0016] S200: Acquire a preset flight path and preset flight parameters of the target UAV, and construct a preset flight attitude frame sequence in combination with the preset monitoring interval and the preset monitoring window; In one possible embodiment, the preset flight path is the flight trajectory pre-planned by the target UAV during the execution of the product transportation mission. The path can be obtained through the UAV flight plan or mission instructions, and usually includes information such as the starting point, end point, flight altitude, heading, etc. The preset flight parameters include the UAV's flight mode (such as automatic flight, manual flight, etc.), flight speed, flight altitude, pitch angle, roll angle, yaw angle, etc. These parameters are usually defined in the flight plan and associated with the flight path. The preset flight attitude frame sequence refers to a sequence that predicts each attitude change (pitch angle, roll angle, yaw angle, etc.) that the UAV may experience during the flight based on the flight path and flight parameters. Optionally, according to the preset monitoring interval (such as collecting flight data once per second) and the preset monitoring window (such as the specific time period of the flight mission), the flight attitude change and related flight parameters of the UAV are calculated at each collection moment. That is, at each time node of the preset flight path, the corresponding flight attitude is collected. In combination with the flight path and flight parameters, the flight attitude of the UAV at each moment in the preset flight path is gradually calculated according to the monitoring interval and flight time. Each flight path point has a corresponding flight attitude framework. By obtaining the preset flight attitude framework sequence, the technical effect of providing comparison data for the subsequent analysis of the attitude deviation of the target UAV within the preset monitoring window is achieved.

[0017] S300: traversing the flight attitude sensing data set sequence to perform attitude analysis and obtain a flight attitude sensing frame sequence; S400: performing mapping deviation analysis on the preset flight attitude frame sequence and the flight attitude sensing frame sequence to determine a flight attitude frame deviation degree sequence; In a possible embodiment, each attitude data in the flight attitude sensing data set sequence is parsed to extract the required flight attitude parameters (such as pitch angle, roll angle, yaw angle, etc.). Optionally, the pitch angle (Pitch), roll angle (Roll), yaw angle (Yaw), etc. are extracted from the sensor data. And the instantaneous attitude change of the drone is calculated based on each attitude data. For example, the change in pitch angle indicates the up and down tilt of the drone, the change in roll angle indicates the side flip of the drone, and the change in yaw angle indicates the change in the heading of the drone. The parsed attitude data is synchronized with the timestamp to ensure that each attitude data can correctly correspond in the time series. Then, the flight attitude sensing frame sequence is constructed based on the parsed attitude data. The flight attitude sensing frame sequence reflects the change of the attitude of the target drone over time during the flight process of the preset monitoring window.

[0018] In one embodiment, the preset flight attitude frame and the flight attitude sensing frame at the same time point are extracted from the preset flight attitude frame sequence and the flight attitude sensing frame sequence, and the corresponding flight attitude frame deviation is obtained by using a mapping deviation analyzer. The flight attitude frame deviation reflects the degree of deviation between the flight attitude of the target UAV and the preset flight attitude at that time point. The greater the flight attitude frame deviation, the greater the deviation of the flight attitude of the target UAV, and the greater the possibility of the transported product being overturned.

[0019] Optionally, multiple sample preset flight attitude frames and multiple sample flight attitude sensing frames, as well as corresponding multiple sample flight attitude frame deviations are obtained as training data sets. The training data sets are used to perform supervised training on the framework constructed based on the feedforward neural network, and the mapping relationship between the preset flight attitude frame and the flight attitude sensing frame and the flight attitude frame deviation is learned until the training converges, and the trained mapping deviation analyzer is obtained. The technical effect of intelligently analyzing the flight attitude frame deviation and improving the analysis efficiency is achieved.

[0020] S500: Performing a comprehensive analysis of the dual effects of deviation superposition effect and concentrated deviation degree on the flight attitude framework deviation degree sequence to determine a flight attitude deviation influence coefficient; Further, the flight attitude frame deviation degree sequence is subjected to a comprehensive analysis of the dual effects of deviation superposition effect and concentrated deviation degree to determine the flight attitude deviation influence coefficient. Step S500 of the embodiment of the present application further includes: Based on the monitoring time of the flight attitude frame deviation in the flight attitude frame deviation sequence, weight allocation is performed in sequence to determine the allocation weight value sequence; Performing weighted calculation on the flight attitude frame deviations in the flight attitude frame deviation sequence according to the distribution weight value sequence to determine a weighted flight attitude frame deviation; Performing a concentrated deviation degree analysis on the flight attitude frame deviation degree sequence to determine the flight attitude frame concentrated deviation degree; The weighted flight attitude framework deviation and the flight attitude framework concentrated deviation are averaged to obtain the flight attitude deviation influence coefficient.

[0021] In a possible embodiment, after obtaining the flight attitude framework deviation degree sequence, a comprehensive impact analysis is performed from two dimensions: the superposition impact of each deviation and the general deviation situation within a preset monitoring window, to obtain the flight attitude deviation impact coefficient. The flight attitude deviation impact coefficient reflects the comprehensive deviation analysis of the deviation flight attitude of the target UAV within the preset monitoring window.

[0022] In one embodiment, weight distribution is performed successively according to the monitoring time of the flight attitude framework deviation in the flight attitude framework deviation sequence, and different weights are given to the deviations at different time points. Preferably, the flight attitude framework deviation closer to the end of the preset monitoring window should be given a higher weight. The ratio of the rank of each flight attitude framework deviation in the flight attitude framework deviation sequence to the total number of the flight attitude framework deviation sequence is used as the corresponding distribution weight value, thereby obtaining the distribution weight value sequence.

[0023] Optionally, the flight attitude frame deviation at each moment in the flight attitude frame deviation sequence is multiplied by the corresponding allocated weight value according to the allocated weight value sequence, and then the products are added to obtain the weighted flight attitude frame deviation.

[0024] Optionally, the general distribution of multiple flight attitude frame deviations in the flight attitude frame deviation sequence is analyzed, concentrated distribution data is extracted, and the flight attitude frame concentrated deviation is determined. The flight attitude frame concentrated deviation reflects whether the attitude change of the UAV during flight is stable or too drastic, which is the general attitude deviation of the target UAV.

[0025] Then, the weighted flight attitude framework deviation and the flight attitude framework concentration deviation are comprehensively analyzed, and the influence coefficient of the flight attitude deviation is evaluated by calculating the mean of the two. The mean calculation result is used as the flight attitude deviation influence coefficient. The flight attitude framework concentration deviation reflects the overall deviation of the attitude fluctuation of the target UAV within the preset monitoring window.

[0026] By weighted calculation of the flight attitude frame deviation sequence and analysis of the two dimensions of concentrated deviation, the stability and deviation of the UAV's flight attitude can be evaluated in real time and accurately. This provides more accurate flight attitude feedback and improves the reliability of subsequent analysis of the safety of transportation products.

[0027] Furthermore, step S500 of the embodiment of the present application further includes: Based on the flight attitude framework deviation sequence, a concentrated deviation analysis space is constructed, wherein the concentrated deviation analysis space is a two-dimensional analysis space, the horizontal axis is time, the vertical axis is the flight attitude framework deviation, and the concentrated deviation analysis space includes a plurality of analysis space points, each analysis space point corresponds to the flight attitude framework deviation at a time point; Calculating the mean ordinate values ​​of the plurality of analysis space points, and taking a straight line passing through the mean ordinate value and parallel to the abscissa axis of the concentrated deviation analysis space as a starting straight line; Iterate the starting straight line in the concentrated deviation analysis space according to a preset moving step length to determine a concentrated deviation straight line; A concentrated deviation neighborhood is constructed based on the concentrated deviation straight line and the preset moving step length, and the mean values ​​of the longitudinal coordinates of multiple analysis space points in the concentrated deviation neighborhood are calculated to obtain the concentrated deviation degree of the flight attitude framework.

[0028] Further, such as Figure 2 As shown, the starting straight line is iterated in the concentrated deviation analysis space according to a preset moving step length to determine a concentrated deviation straight line. Step S500 of the embodiment of the present application further includes: Constructing a starting neighborhood of the starting straight line in the concentrated deviation analysis space according to the preset moving step length, wherein the starting neighborhood includes a plurality of analysis space points whose distance to the starting straight line is equal to the preset moving step length; Counting the number of analysis space points in the starting neighborhood to obtain starting neighborhood statistics; According to a first direction and a second direction, the starting straight line is moved according to the preset moving step length to obtain a first iterative straight line and a second iterative straight line, wherein the first direction is above the starting straight line and the second direction is below the starting straight line; Constructing a first iteration neighborhood of the first iteration straight line and a second iteration neighborhood of the second iteration straight line, and obtaining first iteration neighborhood statistics and second iteration neighborhood statistics; Determine whether the first iterative neighborhood statistic and / or the second iterative neighborhood statistic is greater than or equal to the initial neighborhood statistic. If so, iterate the first iterative straight line and / or the second iterative straight line along the first direction and / or the second direction respectively until a preset number of iterations is met, and take the iterative straight line corresponding to the maximum value of the iterative neighborhood statistic during the iteration process as the concentrated deviation straight line.

[0029] Further, when the first iterative neighborhood statistic and the second iterative neighborhood statistic are both less than or equal to the starting neighborhood statistic, the starting straight line is used as the concentrated deviation straight line.

[0030] In a possible embodiment, a two-dimensional analysis space is constructed, wherein the horizontal axis of the two-dimensional analysis space is time and the vertical axis is the flight attitude frame deviation. Each flight attitude frame deviation and the corresponding time point in the flight attitude frame deviation sequence are input into the two-dimensional analysis space to obtain a plurality of analysis space points. Each analysis space point corresponds to the flight attitude frame deviation at a time point. Further, the concentrated deviation analysis space is constructed based on the plurality of analysis space points and the two-dimensional analysis space.

[0031] Optionally, the mean of the ordinates of the multiple analysis space points is calculated, and a straight line passing through the mean of the ordinates and parallel to the abscissa axis of the concentrated deviation analysis space is used as the starting straight line. The starting straight line is then iterated in the concentrated deviation analysis space according to a preset moving step length to determine the concentrated deviation straight line. The preset moving step length is a deviation value pre-set by a person skilled in the art for two adjacent iterations. The concentrated deviation straight line is the straight line with the most densely distributed analysis space points around it, and the concentrated deviation neighborhood obtained based on the concentrated deviation straight line can best reflect the concentrated distribution of the flight attitude framework deviation sequence.

[0032] Optionally, a concentrated deviation neighborhood is constructed based on the concentrated deviation straight line and the preset moving step length. The concentrated deviation neighborhood includes a plurality of analysis space points whose distance to the concentrated deviation straight line is the preset moving step length. Then, the mean of the ordinates of the plurality of analysis space points in the concentrated deviation neighborhood is calculated to obtain the concentrated deviation degree of the flight attitude framework.

[0033] In one embodiment, a starting neighborhood of the starting straight line is constructed in the concentrated deviation analysis space according to the preset moving step, wherein the starting neighborhood includes a plurality of analysis space points whose distance to the starting straight line is the preset moving step. The number of analysis space points in the starting neighborhood is counted to obtain a starting neighborhood statistic. The starting neighborhood statistic reflects the density of the analysis space points distributed around the starting straight line.

[0034] Then, the starting straight line is moved according to the preset moving step in a first direction and a second direction to obtain a first iterative straight line and a second iterative straight line, wherein the first direction is above the starting straight line and the second direction is below the starting straight line.

[0035] Based on the same construction principle as the initial neighborhood, the first iteration neighborhood of the first iteration line and the second iteration neighborhood of the second iteration line are obtained, and the number of analysis space points in the first iteration neighborhood and the second iteration neighborhood are counted respectively to obtain the first iteration neighborhood statistics and the second iteration neighborhood statistics.

[0036] Optionally, determine whether the first iterative neighborhood statistic and / or the second iterative neighborhood statistic is greater than or equal to the starting neighborhood statistic. If so, it indicates that there is a straight line whose number of surrounding distribution analysis space points is greater than the number of surrounding distribution analysis space points of the starting straight line. At this time, the first iterative straight line and / or the second iterative straight line are iterated along the first direction and / or the second direction respectively until a preset number of iterations (the maximum number of iterations pre-set by a person skilled in the art) is met, and the iterative straight line corresponding to the maximum value of the iterative neighborhood statistic during the iteration process is taken as the concentrated deviation straight line.

[0037] When both the first iterative neighborhood statistic and the second iterative neighborhood statistic are less than or equal to the starting neighborhood statistic, it indicates that the number of analysis space points distributed around the starting straight line is the largest. At this time, the iteration is stopped and the starting straight line is used as the concentrated deviation line.

[0038] S600: performing a neighbor interaction correlation trend analysis on the environmental sensor data set sequence in order from front to back, determining a target environmental interaction correlation sensor data set, performing an environmental deviation analysis in combination with a preset environmental data set corresponding to the preset flight parameter, and determining an environmental deviation influence coefficient; S700: Utilizing the product safety assessment network layer, comprehensively analyzing the flight attitude deviation influence coefficient and the environment deviation influence coefficient to obtain a transportation product safety assessment result.

[0039] Further, the neighbor interaction correlation trend analysis is performed on the environmental sensor data set sequence in order from front to back to determine the target environmental interaction correlation sensor data set. Step S600 of the embodiment of the present application further includes: Extracting a first environmental sensor data set and a second environmental sensor data set of the environmental sensor data set sequence; Performing inner product calculation and normalized matrix processing on the first environmental sensor data set and the second environmental sensor data set to obtain a first nearest neighbor interaction correlation matrix; Performing convolution calculation on the first neighbor interaction correlation matrix and the second environment sensor data set to determine a first environment interaction correlation sensor data set; Extracting a third environment sensor data set again, performing a neighbor interaction correlation trend analysis on the third environment sensor data set and the first environment interaction correlation sensor data set, and determining a second environment interaction correlation sensor data set; Extract the Nth environmental sensor data set again, perform neighbor interaction correlation trend analysis on it and the N-2th environmental interaction correlation sensor data set, determine the N-1th environmental interaction correlation sensor data set, and take the N-1th environmental interaction correlation sensor data set as the target environmental interaction correlation sensor data set, wherein N is the number of environmental sensor data sets in the environmental sensor data set sequence, and N is an integer greater than or equal to 1.

[0040] Furthermore, step S600 in the embodiment of the present application further includes: Using a cosine similarity formula, similar inner product calculation is performed on the first environmental sensor data set and the second environmental sensor data set to obtain a first data similarity coefficient set; The first data similarity coefficient set is traversed by a similarity coefficient processor for normalization processing, and a matrix is ​​constructed for the processing result to obtain the first neighbor interaction correlation matrix.

[0041] Furthermore, step S600 in the embodiment of the present application further includes: The similarity coefficient processor includes a similarity coefficient processing formula, and the similarity coefficient processing formula is: ; in, is the normalized value of the ith data similarity coefficient set in the first data similarity coefficient set, e is the base of the natural logarithm, q is the total number of data similarity coefficients in the first data similarity coefficient set, is the data similarity coefficient of the i-th type of first environmental sensor data and second environmental sensor data in the first environmental sensor data set and the second environmental sensor data set.

[0042] In one possible embodiment, by performing a neighbor interaction correlation trend analysis on the sequence of the environmental sensor data set, a layer-by-layer analysis of the environmental sensor data set is achieved, thereby determining the trend of environmental changes within a preset monitoring window. The target environment interaction correlation sensor data set obtained after considering the correlation trend is subjected to a deviation analysis with the preset environment data set corresponding to the preset flight parameters, and finally an environmental deviation impact coefficient is obtained. The environmental deviation impact coefficient reflects the degree of deviation of the environment when the target UAV transports products within the preset monitoring window. Optionally, the environmental data similarity of the target environment interaction correlation sensor data set and the preset environment data set is calculated using the cosine similarity formula, and then the difference between the environmental data similarity and 1 is used as the environmental deviation impact coefficient.

[0043] Then, the sample flight attitude deviation influence coefficient, the sample environment deviation influence coefficient and the sample transportation product safety assessment result are obtained as training data, and the framework built based on the convolutional neural network is supervised and trained using the training data. During the training, the network parameters are updated according to the training situation until the training converges, and the trained product safety assessment network layer is obtained. The flight attitude deviation influence coefficient and the environment deviation influence coefficient are transmitted to the product safety assessment network layer for analysis to obtain the transportation product safety assessment result. The technical effect of reliably assessing the safety of transportation products based on the flight attitude of the drone is achieved.

[0044] In a possible embodiment, a first environmental sensor data set at the first position and a second environmental sensor data set at the second position are extracted from the sequence of environmental sensor data sets in chronological order. These sets represent the flight environment data of the target UAV at the first time point and the second time point, respectively, and may include multiple environmental parameters such as temperature, humidity, air pressure, wind speed, etc.

[0045] Optionally, the inner product calculation is performed on the extracted first environmental sensor data set and the second environmental sensor data set using the cosine similarity formula. The inner product calculation is used to quantify the correlation and similarity between the two data sets, which can help reveal the potential association between them. And the inner product calculation result is processed by normalizing the matrix using the similarity coefficient processing formula to obtain the first neighbor interaction correlation matrix. That is, the inner product result is normalized to eliminate the differences in different dimensions and data scales, and ensure the comparability between different data sets. Among them, the first neighbor interaction correlation matrix reflects the implicit association relationship between the first environmental sensor data set and the second environmental sensor data set.

[0046] Then, the calculated first nearest neighbor interaction correlation matrix is ​​convolved with the second environmental sensor data set. The convolution operation can further explore the deep correlation relationship between the first environmental sensor data set and the second environmental data set, thereby obtaining the first environmental interaction correlation sensor data set. The main purpose of this stage is to extract important correlation patterns through convolution, making the analysis results more accurate and capturing more subtle trends in environmental changes.

[0047] Preferably, multiple sample neighbor interaction association matrices, multiple sample environment sensor data sets, and multiple sample environment interaction association sensor data sets are obtained as convolutional network training data, and the convolutional network layer is supervised and trained using the convolutional network training data until the training converges to obtain a trained convolutional network layer. The first neighbor interaction association matrix and the second environment sensor data set are convoluted using the convolutional network layer to obtain the first environment interaction association sensor data set.

[0048] Then, the third environment sensor data set is extracted, and according to the same principle as that of obtaining the first environment interaction-related sensor data set, a neighbor interaction-related trend analysis is performed on it and the first environment interaction-related sensor data set, thereby determining the second environment interaction-related sensor data set.

[0049] Based on the same principle as that of obtaining the first environmental interaction-related sensor data set, the Nth environmental interaction-related sensor data set is further extracted, and the neighbor interaction-related trend analysis is performed on it and the previous environmental interaction-related sensor data set (N-2), to obtain the N-1th environmental interaction-related sensor data set. Finally, the N-1th environmental interaction-related sensor data set is used as the target environmental interaction-related sensor data set. This set contains key environmental features that have been refined through multiple layers of analysis, and can effectively reflect the impact of environmental changes on flight missions.

[0050] Optionally, the cosine similarity formula is used to perform similar inner product calculations on the first environmental sensor data set and the second environmental sensor data set. Cosine similarity can measure the similarity between two data sets and help screen out data points with a high degree of similarity. Then, the similarity coefficient processor is used to traverse the first data similarity coefficient set and perform normalization processing. The normalized similarity coefficient can remove the deviation of the data, making the measurement of similarity more fair. The normalized result is matrix constructed to generate a first nearest neighbor interaction correlation matrix to provide a quantitative basis for subsequent analysis.

[0051] Optionally, obtaining the similarity coefficient processor includes a similarity coefficient processing formula, and the similarity coefficient processing formula is: ; in, is the normalized value of the ith data similarity coefficient set in the first data similarity coefficient set, e is the base of the natural logarithm, q is the total number of data similarity coefficients in the first data similarity coefficient set, is the data similarity coefficient of the i-th type of first environmental sensor data and second environmental sensor data in the first environmental sensor data set and the second environmental sensor data set.

[0052] Optionally, the similarity coefficient processing formula uses an exponential function to perform weighted processing on the similarity, ensuring that a greater weight is given to data points with higher similarity, thereby further improving the accuracy of the analysis.

[0053] Through inner product calculation and cosine similarity, the similarities between different environmental data sets can be quickly identified, and multi-level neighbor interaction correlation trend analysis can be performed on this basis. Through convolution calculation and multi-level analysis, the system can deeply explore the subtle correlations between environmental data sets, ensuring that the target environmental interaction correlation sensor data set obtained in the end can accurately reflect environmental changes. The technical effect of accurately assessing the impact of environmental deviation and providing solid data support for analyzing the safety assessment of transportation products has been achieved.

[0054] In summary, the embodiments of the present application have at least the following technical effects: The present application realizes the full data collection of the flight process by synchronously monitoring the flight attitude changes and environmental conditions of the target UAV within a preset monitoring window in real time through the flight attitude sensor array and the environmental sensor array. Furthermore, by gradually analyzing the flight attitude data set sequence and the environmental data set sequence, the potential risks in the flight process can be deeply explored. Through the analysis of the flight attitude parsing and mapping deviation, the degree of deviation of the flight attitude can be accurately evaluated, and the impact of the environmental deviation can be analyzed. A comprehensive analysis is performed based on the flight attitude deviation influence coefficient and the environmental deviation influence coefficient obtained by the analysis, taking into account the mutual influence of the flight attitude and environmental changes, effectively improving the evaluation capability of complex risk factors in the flight process, and achieving the technical effect of improving the reliability of the safety evaluation of transportation products.

[0055] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0056] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0057] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations 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 equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for safety assessment of transportation products based on flight attitude, characterized in that: The method comprises: Using the flight attitude sensor array and the environment sensor array to synchronously sense the attitude changes and environmental conditions of the target UAV in a preset monitoring window in real time according to the preset monitoring interval, and obtain a flight attitude sensor data set sequence and an environment sensor data set sequence; Acquire the preset flight path and preset flight parameters of the target UAV, and construct a preset flight attitude frame sequence in combination with the preset monitoring interval and the preset monitoring window; Traversing the flight attitude sensing data set sequence to perform attitude analysis and obtain a flight attitude sensing frame sequence; Performing mapping deviation analysis on the preset flight attitude frame sequence and the flight attitude sensing frame sequence to determine a flight attitude frame deviation degree sequence; Performing a comprehensive analysis of the dual effects of deviation superposition effect and concentrated deviation degree on the flight attitude frame deviation degree sequence to determine the flight attitude deviation influence coefficient; Performing a neighbor interaction correlation trend analysis on the environmental sensor data set sequence from front to back to determine a target environmental interaction correlation sensor data set, performing an environmental deviation analysis on a preset environmental data set corresponding to the preset flight parameters to determine an environmental deviation impact coefficient; The product safety assessment network layer is used to conduct a comprehensive analysis on the flight attitude deviation influence coefficient and the environment deviation influence coefficient to obtain a transportation product safety assessment result.

2. A method for transport product safety assessment based on flight attitude as claimed in claim 1, characterized in that: The flight attitude frame deviation degree sequence is subjected to a comprehensive analysis of the dual effects of deviation superposition effect and concentrated deviation degree to determine the flight attitude deviation influence coefficient, including: Based on the monitoring time of the flight attitude frame deviation in the flight attitude frame deviation sequence, weight allocation is performed in sequence to determine the allocation weight value sequence; Performing weighted calculation on the flight attitude frame deviations in the flight attitude frame deviation sequence according to the distribution weight value sequence to determine a weighted flight attitude frame deviation; Performing a concentrated deviation degree analysis on the flight attitude frame deviation degree sequence to determine the flight attitude frame concentrated deviation degree; The weighted flight attitude framework deviation and the flight attitude framework concentrated deviation are averaged to obtain the flight attitude deviation influence coefficient.

3. A method for safety assessment of transportation products based on flight attitude as claimed in claim 2, characterized in that: include: Based on the flight attitude framework deviation sequence, a concentrated deviation analysis space is constructed, wherein the concentrated deviation analysis space is a two-dimensional analysis space, the horizontal axis is time, the vertical axis is the flight attitude framework deviation, and the concentrated deviation analysis space includes a plurality of analysis space points, each analysis space point corresponds to the flight attitude framework deviation at a time point; Calculating the mean ordinate values ​​of the plurality of analysis space points, and taking a straight line passing through the mean ordinate value and parallel to the abscissa axis of the concentrated deviation analysis space as a starting straight line; Iterate the starting straight line in the concentrated deviation analysis space according to a preset moving step length to determine a concentrated deviation straight line; A concentrated deviation neighborhood is constructed based on the concentrated deviation straight line and the preset moving step length, and the mean values ​​of the longitudinal coordinates of multiple analysis space points in the concentrated deviation neighborhood are calculated to obtain the concentrated deviation degree of the flight attitude framework.

4. A method for safety assessment of transportation products based on flight attitude as claimed in claim 3, characterized in that: Iterating the starting straight line in the concentrated deviation analysis space according to a preset moving step length to determine the concentrated deviation straight line includes: Constructing a starting neighborhood of the starting straight line in the concentrated deviation analysis space according to the preset moving step, wherein the starting neighborhood includes a plurality of analysis space points whose distance to the starting straight line is equal to the preset moving step; Counting the number of analysis space points in the starting neighborhood to obtain starting neighborhood statistics; According to a first direction and a second direction, the starting straight line is moved according to the preset moving step length to obtain a first iterative straight line and a second iterative straight line, wherein the first direction is above the starting straight line and the second direction is below the starting straight line; Constructing a first iteration neighborhood of the first iteration straight line and a second iteration neighborhood of the second iteration straight line, and obtaining first iteration neighborhood statistics and second iteration neighborhood statistics; Determine whether the first iterative neighborhood statistic and / or the second iterative neighborhood statistic is greater than or equal to the initial neighborhood statistic. If so, iterate the first iterative straight line and / or the second iterative straight line along the first direction and / or the second direction respectively until a preset number of iterations is met, and take the iterative straight line corresponding to the maximum value of the iterative neighborhood statistic during the iteration process as the concentrated deviation straight line.

5. A method for transport product safety assessment based on flight attitude as claimed in claim 4, characterized in that: When both the first iterative neighborhood statistic and the second iterative neighborhood statistic are less than or equal to the initial neighborhood statistic, the initial straight line is used as the concentrated deviation straight line.

6. A method for safety assessment of transportation products based on flight attitude as claimed in claim 1, characterized in that: The environmental sensor data set sequence is subjected to neighbor interaction correlation trend analysis in order from front to back to determine the target environmental interaction correlation sensor data set, including: Extracting a first environmental sensor data set and a second environmental sensor data set of the environmental sensor data set sequence; Performing inner product calculation and normalized matrix processing on the first environmental sensor data set and the second environmental sensor data set to obtain a first nearest neighbor interaction correlation matrix; Performing convolution calculation on the first neighbor interaction correlation matrix and the second environment sensor data set to determine a first environment interaction correlation sensor data set; Extracting a third environment sensor data set again, performing a neighbor interaction correlation trend analysis on the third environment sensor data set and the first environment interaction correlation sensor data set, and determining a second environment interaction correlation sensor data set; Extract the Nth environmental sensor data set again, perform neighbor interaction correlation trend analysis on it and the N-2th environmental interaction correlation sensor data set, determine the N-1th environmental interaction correlation sensor data set, and take the N-1th environmental interaction correlation sensor data set as the target environmental interaction correlation sensor data set, wherein N is the number of environmental sensor data sets in the environmental sensor data set sequence, and N is an integer greater than or equal to 1.

7. A method for safety assessment of transportation products based on flight attitude as claimed in claim 6, characterized in that: include: Using a cosine similarity formula, similar inner product calculation is performed on the first environmental sensor data set and the second environmental sensor data set to obtain a first data similarity coefficient set; The first data similarity coefficient set is traversed by a similarity coefficient processor for normalization processing, and a matrix is ​​constructed for the processing result to obtain the first neighbor interaction correlation matrix.

8. A method for transport product safety assessment based on flight attitude as claimed in claim 7, characterized in that: include: The similarity coefficient processor includes a similarity coefficient processing formula, and the similarity coefficient processing formula is: ； in, is the normalized value of the ith data similarity coefficient set in the first data similarity coefficient set, e is the base of the natural logarithm, q is the total number of data similarity coefficients in the first data similarity coefficient set, is the data similarity coefficient of the i-th type of first environmental sensor data and second environmental sensor data in the first environmental sensor data set and the second environmental sensor data set.

Images