Sensitivity analysis method and device for maneuvering observation field of unmanned aerial vehicle, electronic equipment and storage medium

Abstract

The invention provides a sensitivity analysis method and device for a maneuvering observation field of an unmanned aerial vehicle, electronic equipment and a storage medium, and relates to the technical field of meteorological observation. According to a complete observation operator required by unmanned aerial vehicle observation assimilation, an assimilation universality module adaptive to unmanned aerial vehicle observation is built. According to the background forecasting field, the analysis forecasting field, the real field of the weather forecasting area and the innovation vector of the unmanned aerial vehicle maneuvering observation field, the influence of the observation field of the observation type on numerical forecasting is obtained, the process of calculating the first observation influence is simplified, efficient utilization and accurate evaluation of the unmanned aerial vehicle maneuvering observation field in a forecasting system are ensured, and the forecasting efficiency is improved. And the accuracy of determining the influence of the unmanned aerial vehicle maneuvering observation field on the weather forecast is improved. In addition, a compatible mechanism with an existing forecast-to-observation sensitivity analysis system is considered, and transverse comparison of sensitivities of different observation means of the same observation type is realized by setting observation type and observation means labels for a maneuvering observation field.

Description

Technical Field

[0001] The present invention relates to the field of meteorological observation technology, and in particular to a sensitivity analysis method, device, electronic equipment and storage medium for a mobile observation field of an unmanned aerial vehicle. Background Art

[0002] With the development of meteorological detection capabilities, the means, types, quantity and resolution of meteorological observations have been continuously improved, and various new observation technologies have been continuously emerging. Observational data are of great significance to forecasting. Tens of thousands of observations are used in assimilation analysis to improve the initial field of numerical models, laying a good data foundation for improving the quality of numerical forecasting. However, the actual number of observations is large, and the number of observations that can be used for assimilation analysis is very large, reaching millions or even tens of millions. Although assimilating a large number of observational data can improve the results of analysis and forecasting, the types, elements, spatial locations and combinations of different observations in assimilation analysis may have different contributions to forecasting, and the sensitivity of numerical forecast quality to different observational data is not consistent. Not all observations assimilated into the analysis will improve forecast performance. Therefore, monitoring the application of observational data in assimilation analysis and numerical forecasting, objectively and quantitatively evaluating the sensitivity of numerical weather forecasting to multi-source observational data, and accurately understanding the contribution of observational data assimilation to numerical forecasting are the basis for giving full play to the role of observational data, and are of great scientific significance for improving the quality of numerical forecasting, the effective application of observational data and the further improvement of the observation system.

[0003] In recent years, the forecast sensitivity to observations (FSO) verification method based on adjoint observations has gradually matured and become a flexible and efficient means to evaluate the impact of observations on forecast quality. Compared with traditional observation verification methods, the advantage of the FSO method is that it directly evaluates the impact of any observation assimilated into the analysis on the short-term forecast error, can be used to diagnose the effectiveness of different types of observations on forecast quality, and has high computational efficiency.

[0004] With the advancement of observation technology, the types of observation are becoming increasingly rich. Among them, large-scale UAV meteorological detection systems, as the forefront of the development of observation technology, can drop sounding instruments from the air and obtain more detailed three-dimensional temperature, humidity, wind direction, wind speed, air pressure and other meteorological data in a wide range. This detection method improves the time and horizontal resolution of observation while ensuring the reliability of data, and can carry out targeted real-time encrypted observations according to the development and movement of large-scale weather systems. However, my country is still in its infancy in the field of large-scale UAV meteorological detection. There is a serious lack of understanding of the role of UAV maneuvering observations in numerical forecasting, and there is still a lack of systematic evaluation of its contribution to numerical forecasting. Therefore, there is an urgent need for reliable evaluation and analysis methods for large UAV dropsonde observation data, so as to recognize its value in improving the performance of numerical weather forecasting. In addition, considering that the opportunity cost and economic cost of large UAV flights are both high, whether the estimated analysis observations (including the elements, location, altitude, etc. of the observations) in the early stage of flight have a positive contribution to numerical forecasting is a necessary prerequisite for guiding the planning and design of UAV routes.

[0005] Faced with new mobile observation methods of airborne platforms such as drones, existing observation data cannot be directly input into the data assimilation system; the assimilation system still lacks the complete observation operators and tangent / adjoint operators required for airborne sounding assimilation; faced with sounding data of different observation methods of the same observation type (for example, sounding observation types can be subdivided into balloons, drones, airships, radars and other observation methods), the existing FSO system can only calculate the sensitivity of different methods separately by replacement, but cannot achieve a horizontal comparison of the sensitivity differences of different methods, ignoring the mutual influence of sounding data of different methods, and the judgment of the impact of meteorological observation data on numerical forecasts is not accurate. Summary of the invention

[0006] The present invention provides a sensitivity analysis method, device, electronic device and storage medium for a UAV mobile observation field, so as to solve the defects in the prior art that the influence of UAV mobile observation data on numerical forecast cannot be judged, and the horizontal comparison of the influence of different observation means of the same observation type cannot be completed, so as to improve the comprehensiveness and accuracy of judging the influence of meteorological observation data on numerical forecast.

[0007] The present invention provides a sensitivity analysis method for a UAV mobile observation field, comprising: obtaining the UAV mobile observation field of a weather forecast area according to a UAV mobile observation means, obtaining a background field of the weather forecast area, obtaining the analysis field of the weather forecast area based on assimilation analysis of the UAV mobile observation field and the background field; obtaining the background forecast field based on the background field, and obtaining the analysis forecast field based on the analysis field; obtaining a first observation influence of the UAV mobile observation field based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, and when the first observation influence is less than zero, determining that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast, and when the first observation influence is greater than zero, determining that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast; obtaining a second observation influence of other observation means, and obtaining a comparison result of the UAV mobile observation means and other observation means based on the first observation influence and the second observation influence.

[0008] According to the sensitivity analysis method of the UAV mobile observation field provided by the present invention, based on the assimilation analysis of the UAV mobile observation field and the background field, the analysis field of the weather forecast area is obtained, including: according to the conversion of the pattern space variable and the observation space variable, the nonlinear observation operator of the analysis field variable is obtained; the covariance matrix of the background error of the background field is obtained, and the covariance matrix of the observation error of the UAV mobile observation field is obtained; based on the nonlinear observation operator of the analysis field variable, the covariance matrix of the background error, the covariance matrix of the observation error, the UAV mobile observation field and the background field, the objective function of the analysis field variable is constructed; when the objective function reaches the minimum value, the value of the analysis field variable is used as the analysis field.

[0009] According to the sensitivity analysis method of the UAV maneuvering observation field provided by the present invention, based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV maneuvering observation field, the first observation influence of the UAV maneuvering observation field is obtained, including: obtaining the adjoint mode of the first trajectory based on the background forecast field, and obtaining the adjoint mode of the second trajectory based on the analysis forecast field; performing a three-dimensional variational assimilation analysis operation on the UAV maneuvering observation field to obtain the new information vector; determining a diagonal matrix based on the weighting coefficients of the forecast error components; obtaining a linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error and the nonlinear observation operator of the analysis field; and obtaining the linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error and the nonlinear observation operator of the analysis field. The post-assimilation forecast error is obtained based on the difference between the analysis forecast field and the real field in the weather forecast area, and the sensitive value of the post-assimilation forecast error is obtained based on the product of the post-assimilation forecast error, the diagonal matrix and the adjoint pattern of the second trajectory; the pre-assimilation forecast error is obtained based on the difference between the background forecast field and the real field, and the sensitive value of the pre-assimilation forecast error is obtained based on the product of the pre-assimilation forecast error, the diagonal matrix and the adjoint pattern of the first trajectory; the sensitive value of the comprehensive forecast error is obtained based on the sum of the sensitive value of the post-assimilation forecast error and the sensitive value of the pre-assimilation forecast error, and the first observation impact is obtained based on the product of the sensitive value of the comprehensive forecast error, the new information vector and the transposed matrix of the linear Kalman gain matrix.

[0010] According to the sensitivity analysis method of the UAV maneuverable observation field provided by the present invention, the adjoint mode of the first trajectory is obtained based on the background forecast field, and the adjoint mode of the second trajectory is obtained based on the analysis forecast field, including: obtaining the first tangent line mode based on the tangent line integral of the background field in the reverse time delay, transposing the first tangent line mode to obtain the adjoint mode of the first trajectory; obtaining the second tangent line mode based on the tangent line integral of the analysis field in the reverse time delay, transposing the second tangent line mode to obtain the adjoint mode of the second trajectory.

[0011] According to the sensitivity analysis method of the UAV maneuverable observation field provided by the present invention, based on the first observation influence and the second observation influence, a comparison result of the UAV maneuverable observation means and other observation means is obtained, including: when the first observation influence is less than the second observation influence, the difference between the first observation influence and the second observation influence is obtained, and the smaller the difference is, the better the UAV maneuverable observation means is relative to other observation means.

[0012] According to the sensitivity analysis method of the UAV maneuvering observation field provided by the present invention, the objective function is as follows.

[0013] ; in, is the objective function, To analyze the field, For the background field, is the nonlinear observation operator for analyzing field variables, is the covariance matrix of the background error, is the covariance matrix of the observation error, It is a mobile observation site for drones.

[0014] According to the sensitivity analysis method of the UAV maneuverable observation field provided by the present invention, a background forecast field is obtained based on the background field, and an analysis forecast field is obtained based on the analysis field, including: based on the background field, forward integration is performed in the Weather Research and Forecasting Non-hydrodynamics Forecasting System WRFNL nonlinear model to obtain the background forecast field; based on the analysis field, forward integration with the same time effect is performed in the WRFNL nonlinear model to obtain the analysis forecast field.

[0015] The present invention also provides a sensitivity analysis device for a UAV mobile observation field, comprising: an assimilation analysis module, used to obtain the UAV mobile observation field of the weather forecast area according to the UAV mobile observation means, obtain the background field of the weather forecast area, and obtain the analysis field of the weather forecast area based on the assimilation analysis of the UAV mobile observation field and the background field; a forecast module, used to obtain the background forecast field based on the background field, and obtain the analysis forecast field based on the analysis field; an observation influence determination module, used to obtain the first observation influence of the UAV mobile observation field based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, when the first observation influence is less than zero, determine that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast, and when the first observation influence is greater than zero, determine that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast; a comparison module, used to obtain the second observation influence of other observation means, and obtain the comparison result of the UAV mobile observation means and other observation means based on the first observation influence and the second observation influence.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned sensitivity analysis methods for the maneuverable observation field of an unmanned aerial vehicle.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned sensitivity analysis methods for the unmanned aerial vehicle maneuvering observation field.

[0018] The sensitivity analysis method, device, electronic device and storage medium of the UAV mobile observation field provided by the present invention builds a universal assimilation module adapted to UAV observations based on the complete observation operator required for UAV observation assimilation. According to the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, the influence of the observation field of the observation type on the numerical forecast is obtained, which simplifies the process of calculating the first observation influence, ensures the efficient use and accurate evaluation of the UAV mobile observation field in the forecast system, and improves the accuracy of determining the influence of the UAV mobile observation field on the weather forecast. In addition, the compatibility mechanism with the existing forecast observation sensitivity analysis system is considered, and by setting the observation type and observation means labels for the mobile observation field, a horizontal comparison of the sensitivity of different observation means of the same observation type is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is one of the flow charts of the sensitivity analysis method of the UAV mobile observation field provided by the present invention.

[0021] Figure 2 This is the second flow chart of the sensitivity analysis method of the UAV mobile observation field provided by the present invention.

[0022] Figure 3 It is a structural schematic diagram of the sensitivity analysis device of the UAV mobile observation field provided by the present invention.

[0023] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The Forecast Sensitivity to Observations (FSO) verification method is an important tool for evaluating the impact of observational data on numerical weather forecast results. This method has shown significant advantages in improving forecast accuracy, increasing verification speed, and understanding the role of observational data. However, in the face of new mobile observations from airborne platforms such as drones, the existing FSO system cannot support the scientific analysis of forecast sensitivity. Specifically, FSO is a forecast sensitivity to observation system developed under the framework of the Weather Research and Forecasting Data Assimilation (WRFDA) variational assimilation system, but WRFDA does not support the assimilation of new airborne observations, nor does it support the direct horizontal comparison of different observation methods of the same observation type. In fact, for new observation methods, existing research usually divides its sensitivity into two schemes. The most commonly used one is to compare the forecast results of assimilating conventional observations with the forecast results of assimilating new observations at the same observation location, and then evaluate and compare the impact and improvement of new observations on the forecast, but this method is computationally intensive and ignores the interrelationships between various types of observations. Another method is to disguise the new type of observation as an observation type that can be read by the assimilation system, but the existing assimilation system cannot distinguish between different observation methods of the same observation type. For example, different sounding methods such as balloon sounding, drone sounding, airship sounding, radar sounding, and laser sounding will only be regarded as one type of sounding in the assimilation system. Therefore, if a similar strategy is adopted in the FSO method, it will not be possible to distinguish the sensitivity differences between the new type of sounding and other sounding methods. This will not only affect the role of the original conventional sounding observations, but also make it impossible to measure the horizontal comparison of the objective contribution of observation data.

[0026] Large-scale unmanned aerial vehicle meteorological mobile observation is one of the development frontiers in the field of meteorological observation. However, as mentioned above, in the face of this type of new mobile observation of airborne platforms, the existing observation-forecast sensitivity test system cannot support the scientific analysis of the contribution of this type of observation to the forecast. In order to solve the above problems, the present invention is innovative on the basis of the idea of ​​FSO method, and establishes a new forecast-to-observation sensitivity analysis system compatible with airborne platform observations, and realizes the establishment of an independent sensitivity analysis module for airborne platform observations, including a series of links such as observation quality control, observation preprocessing, assimilation analysis, nonlinear forecasting, and adjoint model application, ensuring the independent, flexible, efficient utilization and accurate evaluation of airborne observation data in the forecast system. In addition, the present invention also considers the compatibility mechanism with the existing forecast-to-observation sensitivity analysis system, and expands the horizontal comparison of the contribution of different observation methods of the same observation type. Specifically, for sounding data, a horizontal comparison of the forecast contribution of multiple sounding methods such as airborne sounding and conventional sounding, wind profile radar, and laser wind radar is realized. It not only strengthens the independent analysis capability of airborne platform observations, but also realizes comprehensive analysis with diversified space-air-ground observation platforms such as ground observations and satellite observations, and constructs a comprehensive, multi-level forecast sensitivity analysis framework for fully understanding the value of UAV observation data, recognizing the interaction between UAV observations and other observation methods, and rationally planning UAV flight routes. It provides strong technical support.

[0027] Combine the following Figure 1-Figure 4 The present invention describes the sensitivity analysis method, device and electronic equipment of the unmanned aerial vehicle maneuvering observation field.

[0028] Figure 1 This is one of the flow charts of the sensitivity analysis method of the UAV mobile observation field provided by the present invention, such as Figure 1 As shown, the sensitivity analysis method of the UAV mobile observation field includes S100 to S300, and each step is specifically as follows.

[0029] S100: According to the UAV mobile observation method, obtain the UAV mobile observation field of the weather forecast area, obtain the background field of the weather forecast area, and obtain the analysis field of the weather forecast area based on the assimilation analysis of the UAV mobile observation field and the background field.

[0030] Acquire and organize the UAV dropsonde observation data (UAV mobile observation field). Preprocess the original UAV observations to complete observation quality control, sparseness, observation error statistics and observation format conversion.

[0031] Obtain different types of observation data, including observation data from airborne sounding, conventional sounding, wind profiler radar, laser wind radar and other multi-source sounding methods. Mark data of different types and different observation methods so that the method can be expanded later to become a test platform for various types of observations. Obtain UAV mobile observation data based on UAV mobile observation methods, and obtain UAV mobile observation fields based on UAV mobile observation data.

[0032] Read in the background field information, initialize the background information, and obtain the background field.

[0033] Based on the background error calculation model (Generator Baseband Environment, GEN_BE V2.0) and the forecast data without assimilation, the background error characteristics of atmospheric variables are statistically analyzed to obtain the covariance matrix of the background error. Furthermore, the covariance matrix of the observation error is obtained.

[0034] The nonlinear observation operator and tangent / adjoint operator of the complete analysis field variables required for airborne sounding assimilation are constructed within WRFDA, and the analysis field is obtained using WRFDA assimilation technology based on the UAV maneuver observation field and the background field. Figure 2 As shown, the present invention uses the three-dimensional variational assimilation analysis (3DVAR) of WRFDA, and 3DVAR obtains the maximum likelihood estimate of the true state of the atmosphere by solving the minimum value of the objective function. When the objective function reaches the minimum value, the corresponding analysis field is obtained.

[0035] S200: Acquire a background forecast field based on the background field, and acquire an analysis forecast field based on the analysis field.

[0036] The background forecast field is obtained based on the background field, and the analysis forecast field is obtained based on the analysis field. Specifically, based on the background field, forward integration is performed in the Weather Research and Forecasting Non-hydrodynamics Forecast System WRFNL nonlinear model to obtain the background forecast field; based on the analysis field, forward integration with the same time effect is performed in the WRFNL nonlinear model to obtain the analysis forecast field.

[0037] Starting from the background field, forward integration is performed in the Weather Research and Forecasting Nonhydrostatic Forecasting System (WRFNL) nonlinear model to obtain the background forecast field. Starting from the analysis field, forward integration with the same time effect is performed in the WRFNL nonlinear model to obtain the analysis forecast field.

[0038] S300: Based on the new information vector of the background forecast field, the analysis forecast field, the real field of the weather forecast area and the UAV mobile observation field, the first observation influence of the UAV mobile observation field is obtained. When the first observation influence is less than zero, it is determined that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast. When the first observation influence is greater than zero, it is determined that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast.

[0039] Construct the innovation vector of the background forecast field, analysis forecast field, real field, and UAV maneuver observation field and the conversion formula of the first observation impact.

[0040] According to the conversion formula, the first observation impact can be calculated when the new information vectors of the background forecast field, analysis forecast field, real field, and UAV mobile observation field are known. The first observation impact is used to characterize the impact of the UAV mobile observation field on the weather forecast. If the calculated first observation impact is less than 0, it means that the assimilated observations of the UAV mobile observation field reduce the forecast error and make a positive contribution to the weather forecast. If the calculated first observation impact is greater than 0, it means that the assimilated observations of the UAV mobile observation field increase the forecast error and make a negative contribution to the weather forecast.

[0041] S400: Obtain a second observation influence of other observation means, and obtain a comparison result of the UAV maneuverable observation means and other observation means based on the first observation influence and the second observation influence.

[0042] Based on the first observation influence and the second observation influence, the comparison results of the UAV mobile observation means and other observation means are obtained. Specifically, when the first observation influence is smaller than the second observation influence, the difference between the first observation influence and the second observation influence is obtained. The smaller the difference is, the better the UAV mobile observation means is relative to other observation means.

[0043] For data of different types and different observation methods, the corresponding second observation influence is obtained respectively, and the contribution of observation data of different observation types, elements, heights, and positions to the forecast can be calculated. For example, the first observation influence is -3, and the second observation influence is -1. This means that the first observation influence is better than the second observation influence. Further obtain the difference between the first observation influence and the second observation influence. If the difference is smaller, it means that the UAV mobile observation method is better than other observation methods.

[0044] This application achieves a horizontal comparison between UAV mobile observation means and other observation means by comparing the first observation impact and the second observation impact.

[0045] Furthermore, for each type of observation data, the corresponding maximum, minimum, average and root mean square error statistics of the observation field minus the background field (OMB) are obtained, and the maximum, minimum, average and root mean square error statistics of the observation field minus the analysis field (OMA) are obtained to further analyze the stability of the impact of the observation data on the forecast.

[0046] According to the label information of the previous observation type, the sensitivity test results of the UAV mobile sounding observation are analyzed, and the contribution of this type of data to numerical weather forecasting is analyzed.

[0047] The present invention realizes a compatible mechanism with the existing forecast-observation sensitivity analysis system, which not only strengthens the independent analysis capability of airborne platform observations, but also realizes comprehensive analysis with diversified sky-air-ground observation platforms such as ground observations, satellites, and radar observations, and constructs a comprehensive, multi-level forecast-observation sensitivity analysis framework to provide technical support for improving the accuracy and timeliness of numerical weather forecasts. Based on the sensitivity test results of independent forecasts to the types, positions, altitudes, and elements of airborne platform observations, the present invention guides the design of drone flight routes, detector placement points, altitudes, and key observation elements, and combines the test results of diversified observation platforms to guide the optimization research of the layout of comprehensive meteorological station networks.

[0048] The sensitivity analysis method of the UAV mobile observation field provided in the embodiment of the present invention builds a universal assimilation module adapted to UAV observations based on the complete observation operators required for UAV observation assimilation. According to the new information vector of the background forecast field, the analysis forecast field, the real field of the weather forecast area and the UAV mobile observation field, the influence of the observation field of the observation type on the numerical forecast is obtained, which simplifies the process of calculating the first observation influence, ensures the efficient use and accurate evaluation of the UAV mobile observation field in the forecast system, and improves the accuracy of determining the influence of the UAV mobile observation field on the weather forecast. In addition, the compatibility mechanism with the existing forecast observation sensitivity analysis system is considered, and by setting the observation type and observation means labels for the mobile observation field, a horizontal comparison of the sensitivity of different observation means of the same observation type is achieved.

[0049] Based on the above embodiment, based on the assimilation analysis of the UAV mobile observation field and the background field, the analysis field of the weather forecast area is obtained, including S110 to S140, and each step is specifically as follows.

[0050] S110: Obtaining a nonlinear observation operator of the analysis field variable according to the conversion between the pattern space variable and the observation space variable.

[0051] S120: Obtain the covariance matrix of the background error of the background field, and obtain the covariance matrix of the observation error of the UAV maneuvering observation field.

[0052] S130: Constructing an objective function of the analysis field variables based on the nonlinear observation operator of the analysis field variables, the covariance matrix of the background error, the covariance matrix of the observation error, the UAV maneuvering observation field and the background field.

[0053] S140: When the objective function reaches a minimum value, the value of the analysis field variable is used as the analysis field.

[0054] The formula of the objective function is as follows.

[0055] ; in, is the objective function, To analyze the field, For the background field, is the nonlinear observation operator for analyzing field variables, is the covariance matrix of the background error, is the covariance matrix of the observation error, It is a mobile observation site for drones.

[0056] It is a nonlinear observation operator that converts pattern space variables into observation space variables, making the pattern field and the UAV maneuvering observation field computationally consistent. The covariance matrix of the background error and the covariance matrix of the observation error determine the relative importance of the UAV maneuvering observation field information and the background field information, the mutual conversion between variables, and the propagation of variables in space.

[0057] According to the nonlinear observation operator of the analysis field variables, the covariance matrix of the background error, the covariance matrix of the observation error, the UAV maneuvering observation field and the background field, the objective function of the analysis field variables is constructed. The minimum value of the objective function is solved. When the objective function reaches the minimum value, the corresponding analysis field variable is taken as the obtained analysis field.

[0058] Solve the minimum value of the objective function according to The gradient of L can be obtained, and according to the cost function iteration, Gradually decrease (the ideal value is ), the corresponding This is the analysis field after assimilation.

[0059] The present invention obtains the analysis field based on the assimilation analysis of the UAV maneuver observation field and the background field, thereby simplifying the calculation amount of obtaining the analysis field. The analysis field is obtained by obtaining the minimum value of the objective function, thereby improving the accuracy of determining the analysis field.

[0060] Based on the above embodiment, based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV maneuver observation field, the first observation impact of the UAV maneuver observation field is obtained, including S310 to S370, and each step is specifically as follows.

[0061] S310: Acquire an accompanying pattern of the first track based on the background forecast field, and acquire an accompanying pattern of the second track based on the analysis forecast field.

[0062] S320: Perform three-dimensional variational assimilation analysis on the UAV maneuver observation field to obtain the new information vector.

[0063] S330: Determine a diagonal matrix based on the prediction error component weighting coefficients.

[0064] S340: Obtain a linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error, and the nonlinear observation operator of the analysis field.

[0065] S350: Obtain the assimilated forecast error based on the difference between the analysis forecast field and the real field of the weather forecast area, and obtain the sensitive value of the assimilated forecast error based on the product value of the assimilated forecast error, the diagonal matrix and the adjoint mode of the second trajectory.

[0066] S360: Obtain the pre-assimilation forecast error based on the difference between the background forecast field and the true field, and obtain the sensitive value of the pre-assimilation forecast error based on the product value of the pre-assimilation forecast error, the diagonal matrix and the adjoint mode of the first trajectory.

[0067] S370: Based on the sum of the sensitivity value of the forecast error after assimilation and the sensitivity value of the forecast error before assimilation, obtain the comprehensive forecast error sensitivity value, and based on the product of the comprehensive forecast error sensitivity value, the new information vector and the transposed matrix of the linear Kalman gain matrix, obtain the first observation impact.

[0068] Real fields (usually actual observations and global reanalysis data) are introduced to characterize the real conditions of the atmosphere. The background forecast field error and the analysis forecast field error are calculated based on the real fields. Based on the difference between the analysis forecast field error and the background forecast field error, the reduction in forecast error under the nonlinear mode, that is, the forecast accuracy, is calculated.

[0069] ; in, is the background forecast field error, To analyze the forecast field error, For the real field, To analyze the forecast field, For the background forecast field, is a diagonal matrix with weighting coefficients of the forecast error components, For forecast accuracy.

[0070] The actual atmospheric motion has highly complex and strong nonlinear motion characteristics. The nonlinear prediction model of atmospheric motion in the numerical prediction model is as follows.

[0071] ; in, For the forecast field, is the initial field, is a nonlinear numerical prediction. is the nonlinear integral of the initial field delay time in the forward direction.

[0072] is a function of the model forecast field (forecast accuracy function) and changes with the change of the model initial field, that is, . Forecast accuracy The gradients of the background forecast field variables and the analysis forecast field variables are and , and used as the initial field of the Weather Research and Forecasting Parallel Unstructured Grid System (WRFPLUS) adjoint model, and reverse integration was performed along the reverse trajectory of the forecast to obtain the forecast accuracy Sensitivity to initial field.

[0073] ; in, is the prediction accuracy function, is the initial field, is the nonlinear integral of the initial field delay time in the forward direction, is the tangent linear integral of the initial field in the reverse direction of time delay, for The corresponding trajectory of the accompanying mode, For the forecast field.

[0074] The companion pattern of the first trajectory is obtained based on the background forecast field, and the companion pattern of the second trajectory is obtained based on the analysis forecast field. Specifically, the first tangent line pattern is obtained based on the tangent line integral of the background field in the reverse time delay, and the first tangent line pattern is transposed to obtain the companion pattern of the first trajectory; the second tangent line pattern is obtained based on the tangent line integral of the analysis field in the reverse time delay, and the second tangent line pattern is transposed to obtain the companion pattern of the second trajectory.

[0075] Based on the adjoint model of WRFDA, the sensitivity of the WRF grid points is converted to the observation space to obtain the sensitivity of the forecast error to the observation point. Finally, the and The dot product of the first observation influences . is the new information vector, which is calculated by the 3DVAR system. In the specific solution of the present invention, a Taylor approximation scheme of the third-order linear estimation which has the highest accuracy in representing the change of the prediction accuracy with the observation is adopted.

[0076] ; in, is the first observed impact, is the innovation vector, To analyze the forecast field, For the background forecast field, is a diagonal matrix with weighting coefficients of the forecast error components, For forecast accuracy, For the real field, is the companion mode of the second trajectory, is the adjoint mode of the first trajectory, is the linear Kalman gain matrix, is the forecast error after assimilation, is the sensitivity value of the forecast error after assimilation, is the forecast error before assimilation, is the sensitivity value of the forecast error before assimilation.

[0077] The calculation formula of the linear Kalman gain matrix is ​​as follows.

[0078] ; in, is the linear Kalman gain matrix, is the covariance matrix of the background error, is the nonlinear observation operator of the analytical field, is the covariance matrix of the observation errors.

[0079] The present invention obtains the first tangent line pattern according to the inverse tangent line integral of the background field delay time, and then obtains the companion pattern of the first trajectory; obtains the second tangent line pattern according to the inverse tangent line integral of the analysis field delay time, and then obtains the companion pattern of the second trajectory, which is conducive to simplifying the calculation of the first observation impact. The present invention obtains the observation impact of the observation type according to the new information vector of the background forecast field, the analysis forecast field, the real field of the weather forecast area and the unmanned aerial vehicle maneuvering observation field, simplifies the process of calculating the first observation impact, ensures the efficient use and accurate evaluation of the unmanned aerial vehicle maneuvering observation field in the forecast system, and improves the accuracy of determining the impact of the unmanned aerial vehicle maneuvering observation field on the weather forecast.

[0080] The present invention realizes the sensitivity test of the forecast based on the FSO system to the unmanned aerial vehicle maneuvering sounding, and realizes the sensitivity test algorithm from a series of links such as observation quality control, observation preprocessing, assimilation analysis, data assimilation, nonlinear prediction, and adjoint model calculation. The process includes one analysis assimilation, two nonlinear prediction model integrations, two tangent linear adjoint model integrations, and one adjoint analysis assimilation. The calculation cost is about 10 to 15 times that of the ordinary single model forecast sensitivity test, and the calculation efficiency is high.

[0081] The present invention expands the horizontal comparison of the contribution of data from different observation methods to forecasts, and extends the verification system to multi-source data beyond UAV mobile observations. It not only strengthens the independent analysis capability of airborne platform observations, but also realizes the comprehensive analysis of sky-air-ground multi-platform observations such as ground observations and satellite observations, and constructs a comprehensive and multi-level forecast sensitivity analysis framework for fully understanding the value of UAV observation data and rationally planning UAV flight routes.

[0082] The sensitivity analysis device for the unmanned aerial vehicle maneuverable observation field provided by the present invention is described below. The sensitivity analysis device for the unmanned aerial vehicle maneuverable observation field described below and the sensitivity analysis method for the unmanned aerial vehicle maneuverable observation field described above can be referenced to each other.

[0083] A sensitivity analysis device for a UAV mobile observation field includes: an assimilation analysis module 301, which is used to obtain the UAV mobile observation field of a weather forecast area according to the UAV mobile observation method, obtain the background field of the weather forecast area, and obtain the analysis field of the weather forecast area based on the assimilation analysis of the UAV mobile observation field and the background field.

[0084] The forecast module 302 is used to obtain the background forecast field based on the background field and to obtain the analysis forecast field based on the analysis field.

[0085] The observation impact determination module 303 is used to obtain the first observation impact of the UAV mobile observation field based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field. When the first observation impact is less than zero, it is determined that the assimilated observations of the UAV mobile observation field have a positive contribution to the weather forecast. When the first observation impact is greater than zero, it is determined that the assimilated observations of the UAV mobile observation field have a negative contribution to the weather forecast.

[0086] The comparison module 304 is used to obtain a second observation influence of other observation means, and obtain a comparison result between the UAV maneuvering observation means and other observation means based on the first observation influence and the second observation influence.

[0087] The sensitivity analysis device for the UAV mobile observation field provided in the embodiment of the present invention builds a universal assimilation module adapted to UAV observations based on the complete observation operators required for UAV observation assimilation. According to the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, the influence of the observation field of the observation type on the numerical forecast is obtained, which simplifies the process of calculating the first observation influence, ensures the efficient use and accurate evaluation of the UAV mobile observation field in the forecast system, and improves the accuracy of determining the influence of the UAV mobile observation field on the weather forecast. In addition, the compatibility mechanism with the existing forecast observation sensitivity analysis system is considered, and by setting the observation type and observation means labels for the mobile observation field, a horizontal comparison of the sensitivity of different observation means of the same observation type is achieved.

[0088] In one embodiment, the assimilation analysis module 301 is used to: obtain the nonlinear observation operator of the analysis field variables according to the conversion of the pattern space variables and the observation space variables; obtain the covariance matrix of the background error of the background field, and obtain the covariance matrix of the observation error of the UAV maneuvering observation field; construct the objective function of the analysis field variables based on the nonlinear observation operator of the analysis field variables, the covariance matrix of the background error, the covariance matrix of the observation error, the UAV maneuvering observation field and the background field; when the objective function reaches the minimum value, the value of the analysis field variable is used as the analysis field.

[0089] In one embodiment, the observation impact determination module 303 is used to: obtain the adjoint mode of the first trajectory based on the background forecast field, and obtain the adjoint mode of the second trajectory based on the analysis forecast field; perform three-dimensional variational assimilation analysis on the UAV maneuvering observation field to obtain the new information vector; determine the diagonal matrix based on the weighting coefficients of the forecast error components; obtain the linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error and the nonlinear observation operator of the analysis field; obtain the assimilated forecast error based on the difference between the analysis forecast field and the real field of the weather forecast area, and obtain the forecast error based on the forecast error component weighting coefficient; obtain the linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error and the nonlinear observation operator of the analysis field; obtain the forecast error after assimilation based on the difference between the analysis forecast field and the real field of the weather forecast area, and obtain the forecast error after assimilation ... analysis forecast field The sensitive value of the post-assimilation forecast error is obtained based on the product of the post-assimilation forecast error, the diagonal matrix and the adjoint pattern of the second trajectory; the pre-assimilation forecast error is obtained based on the difference between the background forecast field and the true field, and the sensitive value of the pre-assimilation forecast error is obtained based on the product of the pre-assimilation forecast error, the diagonal matrix and the adjoint pattern of the first trajectory; the sensitive value of the comprehensive forecast error is obtained based on the sum of the sensitive value of the post-assimilation forecast error and the sensitive value of the pre-assimilation forecast error, and the first observation impact is obtained based on the product of the sensitive value of the comprehensive forecast error, the new information vector and the transposed matrix of the linear Kalman gain matrix.

[0090] In one embodiment, the observation influence determination module 303 is used to: obtain a first tangent line pattern based on the tangent line integral of the background field delay time inverse, transpose the first tangent line pattern to obtain a companion pattern of the first trajectory; obtain a second tangent line pattern based on the tangent line integral of the analysis field delay time inverse, transpose the second tangent line pattern to obtain a companion pattern of the second trajectory.

[0091] In one embodiment, the comparison module 304 is used to: when the first observation influence is smaller than the second observation influence, obtain the difference between the first observation influence and the second observation influence, and the smaller the difference is, the better the UAV maneuver observation method is compared with other observation methods.

[0092] In one embodiment, the objective function is: ; in, is the objective function, To analyze the field, For the background field, is the nonlinear observation operator for analyzing field variables, is the covariance matrix of the background error, is the covariance matrix of the observation error, It is a mobile observation site for drones.

[0093] In one embodiment, the forecast module 302 is used to: based on the background field, perform forward integration in the Weather Research and Forecasting Non-hydrodynamics Forecasting System WRFNL nonlinear model to obtain the background forecast field; based on the analysis field, perform forward integration with the same time effect in the WRFNL nonlinear model to obtain the analysis forecast field.

[0094] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor (processor) 410 , a communication interface (Communications Interface) 420 , a memory (memory) 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other through the communication bus 440 . The processor 410 can call the logic instructions in the memory 430 to execute the sensitivity analysis method of the UAV maneuverable observation field, which includes: according to the UAV maneuverable observation means, obtaining the UAV maneuverable observation field of the weather forecast area, obtaining the background field of the weather forecast area, and obtaining the analysis field of the weather forecast area based on the assimilation analysis of the UAV maneuverable observation field and the background field; obtaining the background forecast field based on the background field, and obtaining the analysis forecast field based on the analysis field; based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV maneuverable observation field, obtaining the first observation influence of the UAV maneuverable observation field; when the first observation influence is less than zero, determining that the assimilated observation of the UAV maneuverable observation field has a positive contribution to the weather forecast, and when the first observation influence is greater than zero, determining that the assimilated observation of the UAV maneuverable observation field has a negative contribution to the weather forecast; obtaining the second observation influence of other observation means, and obtaining the comparison result of the UAV maneuverable observation means and other observation means based on the first observation influence and the second observation influence.

[0095] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0096] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the sensitivity analysis method of the UAV mobile observation field provided by the above-mentioned methods, the method comprising: obtaining the UAV mobile observation field of the weather forecast area according to the UAV mobile observation means, obtaining the background field of the weather forecast area, and obtaining the analysis field of the weather forecast area based on the assimilation analysis of the UAV mobile observation field and the background field; obtaining the background forecast field based on the background field, and obtaining the analysis forecast field based on the analysis field; obtaining the first observation influence of the UAV mobile observation field based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, and when the first observation influence is less than zero, determining that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast, and when the first observation influence is greater than zero, determining that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast; obtaining the second observation influence of other observation means, and obtaining the comparison result of the UAV mobile observation means and other observation means based on the first observation influence and the second observation influence.

[0097] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0098] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sensitivity analysis method for unmanned aerial vehicle maneuvering observation field, characterized in that: include: According to the UAV mobile observation means, the UAV mobile observation field of the weather forecast area is obtained, the background field of the weather forecast area is obtained, and the analysis field of the weather forecast area is obtained based on the assimilation analysis of the UAV mobile observation field and the background field; Acquire a background forecast field based on the background field, and acquire an analysis forecast field based on the analysis field; Based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, a first observation influence of the UAV mobile observation field is obtained, and when the first observation influence is less than zero, it is determined that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast, and when the first observation influence is greater than zero, it is determined that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast; Obtain a second observation influence of other observation means, and based on the first observation influence and the second observation influence, obtain a comparison result between the UAV mobile observation means and the other observation means.

2. The sensitivity analysis method of the UAV mobile observation field according to claim 1 is characterized in that: The obtaining of the analysis field of the weather forecast area based on the assimilation analysis of the UAV mobile observation field and the background field includes: According to the conversion between pattern space variables and observation space variables, the nonlinear observation operator of the analysis field variables is obtained; Obtaining a covariance matrix of background errors of the background field, and obtaining a covariance matrix of observation errors of the UAV maneuvering observation field; Constructing an objective function of the analysis field variables based on a nonlinear observation operator of the analysis field variables, a covariance matrix of the background error, a covariance matrix of the observation error, the UAV maneuvering observation field, and the background field; When the objective function reaches a minimum value, the value of the analysis field variable is used as the analysis field.

3. The sensitivity analysis method of the UAV mobile observation field according to claim 2 is characterized in that: The obtaining, based on the new information vector of the background forecast field, the analysis forecast field, the real field of the weather forecast area and the UAV maneuver observation field, a first observation influence of the UAV maneuver observation field comprises: Acquire an adjoint pattern of a first track based on the background forecast field, and acquire an adjoint pattern of a second track based on the analysis forecast field; Performing a three-dimensional variational assimilation analysis operation on the UAV mobile observation field to obtain the new information vector; Determining a diagonal matrix based on the weighting coefficients of the forecast error components; Obtaining a linear Kalman gain matrix based on the covariance matrix of the background error, the covariance matrix of the observation error and a nonlinear observation operator of the analysis field; Obtaining an assimilated forecast error based on a difference between the analysis forecast field and a real field of the weather forecast area, and obtaining a sensitive value of the assimilated forecast error based on a product value of the assimilated forecast error, the diagonal matrix, and an adjoint mode of the second trajectory; Obtaining a pre-assimilation forecast error based on a difference between the background forecast field and the true field, and obtaining a sensitive value of the pre-assimilation forecast error based on a product value of the pre-assimilation forecast error, the diagonal matrix, and an adjoint mode of the first trajectory; Based on the sum of the sensitivity value of the forecast error after assimilation and the sensitive value of the forecast error before assimilation, a comprehensive forecast error sensitive value is obtained; based on the product of the comprehensive forecast error sensitive value, the new information vector and the transposed matrix of the linear Kalman gain matrix, the first observation impact is obtained.

4. The sensitivity analysis method of the UAV mobile observation field according to claim 3 is characterized in that: The step of acquiring the companion mode of the first track based on the background forecast field and acquiring the companion mode of the second track based on the analysis forecast field comprises: Based on the tangent linear integral of the background field in the reverse direction of time delay, a first tangent linear pattern is obtained, and the first tangent linear pattern is transposed to obtain an adjoint pattern of the first trajectory; Based on the time-delayed inverse tangent linear integral of the analysis field, a second tangent linear mode is obtained, and the second tangent linear mode is transposed to obtain the adjoint mode of the second trajectory.

5. The sensitivity analysis method of the UAV mobile observation field according to claim 1 is characterized in that: The obtaining, based on the first observation influence and the second observation influence, a comparison result between the UAV mobile observation means and the other observation means, comprises: When the first observation influence is smaller than the second observation influence, the difference between the first observation influence and the second observation influence is obtained. The smaller the difference is, the better the UAV mobile observation method is relative to the other observation methods.

6. The sensitivity analysis method of the UAV mobile observation field according to claim 2 is characterized in that: The objective function is: ； in, is the objective function, is the analysis field, is the background field, is the nonlinear observation operator of the analysis field variable, is the covariance matrix of the background error, is the covariance matrix of the observation error, It is the mobile observation field for the UAV.

7. The sensitivity analysis method of the UAV mobile observation field according to claim 1 is characterized in that: The acquiring of the background forecast field based on the background field and the acquiring of the analysis forecast field based on the analysis field comprises: Based on the background field, forward integration is performed in a nonlinear model of the Weather Research and Forecasting Non-hydrodynamics Forecasting System WRFNL to obtain the background forecast field; Based on the analysis field, forward integration with the same time effect is performed in the WRFNL nonlinear model to obtain the analysis forecast field.

8. A sensitivity analysis device for a UAV mobile observation field, characterized in that: include: An assimilation analysis module, used to obtain a UAV mobile observation field of a weather forecast area according to a UAV mobile observation method, obtain a background field of the weather forecast area, and obtain an analysis field of the weather forecast area based on assimilation analysis of the UAV mobile observation field and the background field; A forecast module, used for acquiring a background forecast field based on the background field, and acquiring an analysis forecast field based on the analysis field; an observation impact determination module, configured to obtain a first observation impact of the UAV mobile observation field based on the background forecast field, the analysis forecast field, the real field of the weather forecast area and the new information vector of the UAV mobile observation field, and when the first observation impact is less than zero, determine that the assimilated observation of the UAV mobile observation field has a positive contribution to the weather forecast; and when the first observation impact is greater than zero, determine that the assimilated observation of the UAV mobile observation field has a negative contribution to the weather forecast; A comparison module is used to obtain a second observation influence of other observation means, and based on the first observation influence and the second observation influence, obtain a comparison result between the UAV mobile observation means and the other observation means.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the sensitivity analysis method of the unmanned aerial vehicle maneuvering observation field as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the sensitivity analysis method of the UAV maneuverable observation field as described in any one of claims 1 to 7 is implemented.

Images