52results about How to "Improve the inversion effect" patented technology

The invention discloses a land aerosol optical property retrieval method based on the Gaofen-1 satellite. The land aerosol optical property retrieval method aims to solve the problem that an aerosol optical thickness retrieval method is limited by short wave infrared data, and technical supports are provided for air quality monitoring based on the domestic high-definition satellite. The land aerosol optical property retrieval method comprises the following steps that S1, an atmospheric optical parameter lookup table is generated; S2, a surface reflectance prior database is generated; S3, dark object pixels suitable for retrieval are selected; S4, the observation geometrical conditions of the dark object pixels are calculated; S5, the atmospheric optical parameters of the dark object pixels are obtained in an interpolation mode; S6, aerosol optical thickness retrieval is carried out based on radiation transfer. The method is efficient in operation, retrieval on the regional high-resolution atmospheric aerosol optical property is facilitated, and beneficial data sources are provided for particulate pollution monitoring.

The invention discloses a three-dimensional fine imaging system based on a drilling geological radar technology. The system comprises devices, such as a radar transmitting antenna, wherein the radar transmitting antenna is transmitted into a bore through a radar antenna transmitting device; a radar receiving antenna moves along a measuring line on a tunnel work area surface through a radar antenna driving device; the radar transmitting antenna and the radar receiving antenna are connected with a radar host machine; the radar host machine is respectively connected with a storage battery and a computer; the radar antenna transmitting device and the radar antenna driving device are connected with a control host machine; and the control host machine is connected with the storage battery through a power wire and controls the antenna to transmit and the driving device. The invention also provides a stereo three-dimensional observing method. Three-dimensional detection on surrounding media of the bore is realized by matching the method with a 'transmitting, receiving, circulating and detecting' way based on coverage observation for multiple times, and by a speed imaging method based on constrained reversion and a three-dimensional interpolation method, and the three-dimensional fine detection on the geological status in front of the tunnel work area surface is realized.

The invention discloses an urban impermeable surface extraction method based on a high-resolution satellite remote sensing image. A high-resolution satellite remote sensing image is used as a data source, and a new high-resolution urban impermeable surface comprehensive index is established by combining a normalized building index, a normalized vegetation index, an improved normalized water body index and a normalized soil index. According to the method, the difference between the impermeable surface and vegetation, water and naked bodies is considered, so that the problems of same spectrum offoreign matters and the like when the impermeable surface is extracted by a single index are avoided. Compared with a traditional method of independently using the impermeable surface index to extract the urban impermeable surface, the inversion effect of the model is good, the feature information of the impermeable surface can be mined, and a new technology and means are provided for accuratelyand comprehensively extracting the urban impermeable surface with the high spatial resolution within a large range.

The invention discloses a time frequency electromagnetic and magnetotelluric joint inversion method based on deep learning. The method comprises the following steps: (a) establishing a neural networkN which comprises three deep convolutional sub-networks NMT, NF and NT, combining data of m channels output by the NMT, NF and NT to form joint data, connecting with an intermediate transition layer FA, FA is further connected with a deep convolutional network NL which processes the joint data, and NL is further connected with an output layer; (b) acquiring magnetotelluric and time frequency electromagnetic training and verification data bodies; (c) inputting a magnetotelluric response data body DMT into the NMT, inputting a frequency domain response data body DF into the NF, and inputting a time domain response data body DT into the NT; (d) performing convolution and pooling calculation on the data by the deep convolutional sub-network NMT, the NF and the NT, and then outputting the datato the intermediate transition layer FA; (e) performing full connection or convolution-pooling calculation on the data by the intermediate transition layer FA, and then outputting the data to the deepconvolutional network NL; and (f) performing calculation processing on the data by the deep convolutional network NL, and then outputting m values by the output layer.

The invention discloses a forest-biomass remote sensing inversion method based on spectral curve characteristic differentiation. Landsat 8OLI and actually measured data serve as a data source, and a forest-biomass leaf biomass and above-ground biomass inversion model which takes Landsat 8OLI red-light band reflectivity R, near-infrared band reflectivity NIR, red-light band central wavelength CW<R> and near-infrared band central wavelength CW<NIR> as characterization parameters is constructed. The model is good in inversion effect and provides a new technical method and means for rapidly, accurately and comprehensively estimating above-ground biomass of different forest community types in large stretches of forests.

The application relates to an inversion method and an apparatus for multi-channel transient electromagnetic detection data. The method incudes: multi-channel transient electromagnetic detection is performed at a to-be-detected place so that earth pulse response of the to-be-detected place is obtained; the falling-edge step response of the earth pulse response is obtained; and the inversion of the falling-edge step response is performed. According to the inversion method and the apparatus for the multi-channel transient electromagnetic detection data, the interferences of air waves in the pulse response can be avoided, accurate inversion results can be obtained, and the falling-edge step response obtained by a reverse integral method and the analytical solution are well coincided.

The invention relates to a multi-scale step type layer-stripping full-waveform inversion method for ground penetrating radar data. The method comprises the following steps: firstly, selecting layeringnumber and inversion frequency of a model and a single-frequency and single-layer inversion number of times, and determining an inversion serial number of each layer in each frequency according to anew inversion sequence; secondly, filtering radar observation data and sub-waves to different frequency bands by virtue of a Wiener filter, and adding a space window in a corresponding layer to the calculated gradient by virtue of a dielectric constant gradient formula, so as to obtain a gradient value of the current layer; and finally, calculating the updating amount of the model, and finishing iterative operation layer by layer and frequency by frequency. By utilizing a multi-scale strategy, the independence on an initial model is reduced, and the cycle skip problem is effectively solved; the method is different from an existing offset-time window, and the gradient of a specific layer is acquired by virtue of a space window, so that convergence can be rapidly realized; by utilizing a newstep type inversion sequence, the difference between the inversion frequencies of two adjacent layers is effectively reduced, so that a relatively good inversion effect is achieved, the deep inversion effect of the model is effectively improved, and the stability of full-waveform inversion is improved.

The invention discloses a method and model for calculating canopy reflectivity of broad-leaved vegetation. The method comprises the following steps: S1: inputting parameters, identifying the input parameters, and classifying the input parameters into leaf parameters, canopy parameters and soil parameters; S2: calculating reflectivity and transmissivity of each single leaf according to the leaf parameters; S3: calculating an extinction coefficient and a scattering coefficient of canopies according to the canopy parameters and the leaf parameters in the step S2; S4: calculating related reflection factor and reflectivity of the canopies according to the obtained canopy extinction and scattering parameters; S5: calculating the canopy reflectivity according to the related reflection factor and reflectivity of the canopies. According to the method and the model, a PROSPECT model and an SAIL model are coupled, and a leaf reflectivity and transmissivity input process in a vegetation canopy reflectivity simulation process is canceled under the condition of making full use of available parameters; a parameter acquisition problem in a vegetation canopy spectral information simulation process is effectively simplified, an algorithm is optimized, a calculating process is accelerated, and meanwhile, the coupled model facilitates parametric inversion of the vegetation.

The invention discloses a GPS multi-satellite fusion soil humidity monitoring method based on fuzzy entropy, and belongs to the technical field of soil humidity measurement. The method comprises the steps: extracting oscillation amplitudes of signal-to-noise ratios of three frequency bands of all satellites from a GPS receiver by using the GNSS-IR technology, performing fusion by using a fuzzy entropy method to obtain an optimal multi-satellite three-frequency oscillation amplitude fusion value, and establishing an inversion model with soil humidity by using the fusion value. According to themethod, the difference and complementarity of GPS multi-satellite three-frequency data are effectively utilized, and the inversion effect is obviously improved compared with that of a single-satellitesingle-frequency method; and the problems in the prior art are solved.

The invention discloses a Himawari-8 atmospheric aerosol inversion method based on a deep full connection network, and belongs to the technical field of aerosol remote sensing. The method comprises the following steps: firstly, acquiring remote sensing image data of a Himawari-8 stationary satellite in a NetCDF4 format; obtaining a dependent variable data set required by the inversion model; secondly, acquiring aerosol product data of AERONET site 1.5-level cloud removal and quality control at the time corresponding to the Himawari-8 image; obtaining independent variable data required by the inversion model; then, constructing a deep learning model of the deep full-connection network by using the sorted data set; performing model parameter setting for obtaining the relationship between Himawari-8 remote sensing image data and aerosol data of a ground AERONET station, and constructing an aerosol inversion model; and finally, applying the trained depth full-connection inversion model toa Himawari-8 remote sensing image above the monitoring point to obtain an AOD inversion result corresponding to the monitoring position. Other auxiliary data are not required to be introduced in the aerosol inversion process, so that errors in the inversion process are reduced, and the AOD inversion precision is improved.

The invention provides a three-dimensional wind field inversion method based on a non-uniform sampling correction VAD technology. According to the method, a nonlinear spatial distribution model of a complex wind field is built; an omnibearing non-uniform sampling data model of a radial Doppler speed caused by platform jitter and other factors is built; a nonlinear wind field hypothesis can betterrepresent a real complex wind field structure; non-uniform sampling is carried out on the radial Doppler speed, and therefore, method is more suitable for the conditions that external disturbance conditions such as platform jitter and data abnormal points exist; and therefore, compared with the traditional VAD technology requiring a linear wind field hypothesis and equidirectional interval uniformsampling, the method of the invention has a better inversion effect.

The invention discloses a GNSS-R sea surface wind speed inversion method and system based on a particle swarm algorithm, and belongs to the fields of electronics, information, atmospheric science andthe like. The method comprises the following steps: downloading and storing GNSS-R data and sea surface wind speed data in batches; preprocessing the data, and completing space and time matching of the GNSS-R data and the wind speed data; screening the matched data, reserving high-quality samples, then performing data set file format normalization, and randomly dividing a training set and a test set; taking the DDM characteristic value and the signal incident angle as input, constructing a geophysical mode function, and realizing single characteristic value wind speed inversion; and based on aparticle swarm optimization algorithm, optimizing a combined wind speed inversion model to obtain an optimal combined optimization coefficient, and completing wind speed combined inversion. Accordingto the method, the wind speed combination inversion is carried out by using the particle swarm algorithm, so that the accuracy and efficiency of inversion are greatly improved, and the method has thecharacteristics of simple model, high robustness, high inversion precision and the like.

The invention relates to a wave impedance inversion method using a neural network and a neural network system. The neural network takes N channels of seismic data composed of i-th channel seismic data and multiple channels of seismic data adjacent to the i-th channel seismic data and i-th channel initial model data as input, determines i-th channel wave impedance data as output, and comprises: N parallel feature extraction layers, wherein each feature extraction layer is configured to extract time sequence features of a channel of seismic data input to the feature extraction layer; a merging layer configured to adaptively merge the time sequence features output by the N feature extraction layers to obtain spatial-temporal features of the N channels of seismic data; a regression layer configured to map the spatial-temporal features from a feature domain to a target domain; and an output layer configured to determine the i-th channel wave impedance data according to the output of the regression layer and the i-th channel initial model data. According to the invention, the accuracy of wave impedance inversion is higher, the continuity is stronger, the anti-noise performance is better, and a good inversion effect can still be kept when the initial model is not accurate.

The invention discloses water-based acrylic resin with high distinctness of image. The water-based acrylic resin is prepared from the following raw materials: an acrylic monomer, a styrene monomer, an initiator, a chain transfer agent, a solvent and a mixed solution, wherein the mixed solution is prepared from glycidyl versatate and vinyl versatate in a mass ratio of 2.5: 1. The preparation method comprises the following steps: adding 80% of solvent into a reaction kettle, heating and preserving heat; mixing an acrylic monomer, a styrene monomer, a chain transfer agent and 90% of an initiator, and dropwise adding the mixture into the reaction kettle; dropwise adding a mixed solution composed of glycidyl versatate and vinyl versatate, and carrying out heat preservation; adding the residual solvent and initiator, preserving heat, cooling to 90-95 DEG C, dropwise adding an amine neutralizer solution, dispersing at a high speed, slowly dropwise adding deionized water at 90-95 DEG C, and carrying out reversed-phase emulsification to obtain the water-based acrylic resin. The waterborne acrylic resin provided by the invention can improve the gloss of the resin, reduce the viscosity of the resin, and improve the flexibility, impact resistance and water resistance of a paint film.

The embodiment of the invention discloses a multi-frequency multi-dimensional nuclear magnetic logging method and device. The method comprises the steps of employing a plurality of different frequencies for working when an underground nuclear magnetic logging instrument works, wherein the plurality of different frequencies are different frequencies of which the frequency number is greater than two; respectively acquiring echoes corresponding to a plurality of different frequencies in a plurality of different acquisition time periods; collecting echoes corresponding to one frequency in one collection time period; when the echoes corresponding to any one frequency are collected, enabling the frequencies except the current echo collection frequency in the multiple different frequencies to bein a polarization waiting state; and calculating a predetermined spectrogram according to the acquired echoes of the plurality of frequencies, wherein the predetermined spectrogram comprises one or more of the following spectrograms: a D-T2 spectrogram, a T1-T2 spectrogram and a T1/T2-T2 spectrogram. Through the scheme of the embodiment, the acquisition period is shortened, sufficient polarizationis ensured, and the inversion effects of T1-T2, T1/T2-T2 and D-T2 are improved.

The invention discloses an inversion method for microcystin MC-LR concentration in a water body. The method comprises the steps of establishing a unary linear inversion equation by taking normalized characteristic fluorescence peak intensity of a microcystis aeruginosa extracellular organic matter in a water sample as an independent variable and normalized microcystin MC-LR concentration in the water sample as a dependent variable, wherein the characteristic fluorescence peak intensity and the microcystin MC-LR concentration are normalized through utilization of chlorophyll a concentration inthe water sample; obtaining a to-be-detected water sample, measuring three-dimensional excitation-emission matrix fluorescence spectrum of the microcystis aeruginosa extracellular organic matter in the water sample, and obtaining the characteristic fluorescence peak intensity of the microcystis aeruginosa extracellular organic matter; and substituting the characteristic fluorescence peak intensityin to the unary linear inversion equation, and obtaining the microcystin MC-LR concentration. According to the method, sample pretreatment is avoided, rapid online monitoring of the microcystin MC-LRconcentration can be realized, the characteristic fluorescence peak intensity of the microcystis aeruginosa extracellular organic matter is taken as the variable for establishment of the equation, and a new thought for the inversion method is provided.

The invention provides a corn growth parameter active and passive remote sensing inversion method based on data augmentation and deep learning, and the method comprises the steps: carrying out data preprocessing: respectively preprocessing an optical image and a radar image, obtaining input data, and normalizing the input data; a Beta data mixing method is used for simultaneously carrying out mixing processing on limited actually-measured leaf area indexes, biomass data and corresponding remote sensing data so as to establish a large number of virtual training samples to carry out data augmentation; a leaf area index and biomass inversion model is constructed by combining a twin deep network based on a gating mechanism with optical and radar remote sensing data. The number of field actualmeasurement samples is increased by providing data augmentation, and a twin neural network model of a gating mechanism is constructed, so that a deep learning method with better nonlinear fitting capability is applied to active and passive inversion of corn LAI and biomass, and the inversion precision is improved.

The invention provides a semi-supervised resistivity deep learning inversion method and a semi-supervised resistivity deep learning inversion system based on an adversarial generative network and pseudo-annotation, which can realize a deep learning inversion network under the condition that part of apparent resistivity data lacks a corresponding geological model; vertical features of the electrical method data are extracted, a resistivity model is finally obtained by extracting a feature map, and the mapping relation between the apparent resistivity data and the underground resistivity geologic model is completed; meanwhile, a discriminator is added into the network structure, pseudo labeling is introduced into apparent resistivity data without corresponding geological models, the data size of labeled data is expanded, and a semi-supervised learning strategy is achieved. Through the semi-supervised deep learning electrical method data inversion network, the deep learning network inversion effect is improved when the label data is less.