32results about How to "Add nonlinearity" patented technology

The invention discloses a lightweight unmanned aerial vehicle target tracking method and system based on dual cross-correlation fusion. The method comprises the following steps: acquiring a video stream of a target area; adding an initial target frame to a first frame image of the video stream to calibrate the initial position of a target, taking the first frame image as a template image, and taking other frame images of the video stream as search images; preprocessing the template image and the search image; inputting the preprocessed template image and the search image into a pre-trained target tracking model to obtain a predicted position of the target in a certain frame of image of the video stream, thereby realizing target tracking; a twin network is adopted as an overall framework, a backbone network is a lightweight network, the real-time performance of an algorithm is improved while the network parameter quantity and the calculation quantity are reduced, a dual cross-correlation fusion module is adopted, the nonlinearity of fusion features is enhanced, target feature information is highlighted, an anchor-free position prediction head is used, and the tracking speed is improved. And the network parameter quantity is reduced.

The invention discloses an optical frequency comb generation device and method based on lithium niobate film waveguide series connection double micro-rings. The device comprises a first bus waveguide, a first annular waveguide, a second annular waveguide and a second bus waveguide which are etched on a lithium niobate film. Wherein one end of the first bus waveguide is a pump light input end, the other end of the first bus waveguide is an auxiliary light input end, the first bus waveguide and the first annular waveguide are mutually coupled, the first annular waveguide and the second annular waveguide are mutually coupled, the second annular waveguide is a soliton cavity, and a temperature regulation and control device is arranged below the soliton cavity. The temperature of the soliton cavity is regulated and controlled to enable the harmonic peak of the soliton cavity to drift, so that the pump light enters a soliton state, the auxiliary light performs power compensation on the soliton state of the pump light, and a soliton optical frequency comb is generated. In combination with auxiliary photo-thermal modulation, the micro-cavity thermal effect is compensated, the soliton stability is improved, and finally a comb-shaped structure spectrum with stable performance, high coherence and high flatness is output from a soliton cavity port.

According to the generative image information steganography method provided by the invention, a memristor is implanted into an existing chaotic system, a new memristor chaotic system NCSM is constructed, the system can generate a chaotic sequence with higher randomness, and when the system is applied to image encryption, the security of an encrypted image can be improved; in the aspect of image steganography, a ciphertext image is converted into a generative steganography image through a generative reversible neural network model Glow, so that the information security of a plaintext image is ensured, and the visual security of the ciphertext image is also ensured; different from a current model based on irreversible adversarial generation, the generative reversible neural network does not need to train a decoder, so that the complexity of the model and the training difficulty of the model are reduced; besides, by optimizing the input of the reversible generation model, the content of the generative steganographic image can be controlled to a certain extent, and the method can adapt to wider application scenes.

The invention relates to the field of monitoring, in particular to a tailing paste filling concentration online monitoring method. Comprising the following steps: acquiring sound wave signals and micro-seismic signals, and performing feature extraction on the acquired sound wave signals and micro-seismic signals by using a frequency feature extraction algorithm to obtain frequency features; calculating the comprehensive propagation velocity of the tailing paste through a comprehensive propagation velocity calculation algorithm; calculating the dynamic density of the tailing paste by using a dynamic density calculation algorithm in combination with the physical characteristics of the paste particles; the dynamic mass concentration of the tailing paste is calculated by obtaining the dynamic density, the solid density and the water density of the tailing paste and combining the concentration correction factor; and monitoring the dynamic mass concentration change of the tailing paste in real time by using a dynamic concentration fluctuation range monitoring algorithm. The technical problem that in the tailing paste filling process, the influence of multiple factors such as the paste particle size and the friction angle is not fully considered in the prior art, and consequently the precision of dynamic density calculation is low is solved.

The invention relates to the field of artificial intelligence, in particular to a mineral resource recognition model training method based on data collected by an unmanned aerial vehicle. The training method comprises the steps of collecting real geological data, training a data generation model, generating simulated geological data by adopting the trained data generation model, and training a feature extraction model and a classification model by taking the simulated geological data and the real geological data as training samples together. An existing mineral resource identification model training method based on data acquired by an unmanned aerial vehicle has the problems of low model precision and low calculation speed of a mineral resource identification model. According to the mineral resource identification model training method based on the data acquired by the unmanned aerial vehicle provided by the invention, the model precision of the mineral resource identification model is relatively high, and the calculation speed is relatively high.

A single-mode high-nonlinearity photonic crystal fiber capable of maintaining circular polarization transmission comprises a solid fiber core and a porous cladding structure arranged periodically, and the nonlinearity of the fiber is improved by optimizing the size of the fiber core and the duty ratio of cladding air holes. The torsion rate alpha introduced in the axial direction of the optical fiber is 0-8 rad / mm to introduce 10 <-3 >-order circular birefringence to maintain the circular polarization state, and effective refractive index change curves in different modes are changed. A plurality of cladding air holes are selected on the cross section of the optical fiber for selective filling or doping to realize phase matching of a fiber core high-order mode and a plurality of cladding modes, so that the fiber core high-order mode with the refractive index changing under torsion is filtered out by means of intermode coupling, the loss of the filtered high-order mode is increased by more than two orders of magnitude, and the high-order mode loss is reduced by more than two orders of magnitude. And single-mode transmission is effectively realized. Meanwhile, the characteristics of circular polarization state transmission and high nonlinearity are kept, and the optical fiber can be applied to application research of high-precision optical fiber sensing, optical fiber lasers with special circular polarization state output and the like.

The invention discloses a wave pitching motion control method and system for an offshore floating type wind turbine generator, and belongs to the technical field of offshore wind power equipment control. The method comprises the following steps: signal acquisition: acquiring unit operation state, marine environment and pitching motion state parameters in real time; wave load prediction: predicting a future wave load based on the improved LSTM neural network model; evaluating a pitching motion state, pre-judging a motion trend through a kinetic model, and judging whether the motion trend exceeds a safety threshold value or not; cooperative control instruction generation: generating cooperative control instructions for fine tuning of a variable pitch angle, a generator torque and a cabin yaw angle based on an MPC algorithm; and control execution and feedback correction: driving an execution mechanism to act and dynamically correcting a control instruction. By accurately predicting the wave load, cooperatively adjusting multiple controllable components and dynamically correcting, efficient and accurate suppression of pitching motion is achieved, and the unit operation stability, the wind energy capturing efficiency and the equipment safety are improved.

The invention discloses an unmanned aerial vehicle image transmission signal modulation mode identification system and method based on multi-dimensional feature cooperation. The method comprises the following steps: (S100) collecting and preprocessing unmanned aerial vehicle communication signal data; (S200) the one-dimensional features are extracted and remodeled into a two-dimensional feature map, and normalization processing is carried out; estimating a two-dimensional time-frequency diagram of the image transmission signal; (S300) feature combination: combining the obtained remolded graph and the time-frequency graph as four channels to form a multi-channel feature graph; (S400) generating a training set and a test set; (S500) constructing a VGGR16ChannelAttention deep learning network, and carrying out the training of the VGGR16ChannelAttention deep learning network; and (S600) identifying the image transmission signal modulation mode of the unmanned aerial vehicle to be identified. According to the method, the time-frequency analysis technology and the deep learning technology are fused, the method surpasses a traditional method through time-frequency synchronous optimization transformation, and complex signals can be accurately analyzed.

The invention relates to a data desensitization method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of data processing. The data desensitization method specifically comprises the following steps: carrying out slot division processing on to-be-desensitized data according to a rule that the data length meets an incremental rule to obtain at least one group of slot division data; determining a corresponding initial desensitization rule according to the data length of the sub-slot data, desensitizing the sub-slot data according to the initial desensitization rule to obtain desensitization data of each group of sub-slot data, and connecting the desensitization data of each group of sub-slot data in series to obtain first desensitization data; a pre-trained clustering model is adopted to determine a data label of each group of slotted data, wherein the data labels comprise a first label representing a large amount of data and a second label representing a small amount of data; and determining a secondary desensitization rule based on the distribution condition of the data labels of all the slotted data, and processing the first desensitization data and the to-be-classified data according to the secondary desensitization rule to obtain second desensitization data.

This invention provides an audio dynamic range control circuit and an audio device. The audio dynamic range control circuit includes an audio zero-crossing detection circuit, a preprocessing circuit, a first control circuit, a first gain selection circuit, a gain multiplication processing circuit, a second control circuit, a third control circuit, and a second gain selection circuit. The audio dynamic range control circuit provided by this invention achieves an effect similar to the existing RMS algorithm by performing absolute value processing on the linear domain sampling data of the audio signal, while simplifying the hardware circuit implementation. Simultaneously, by detecting the zero-crossing position of the audio signal through the audio zero-crossing detection circuit, and updating the gain value at subsequent zero-crossing positions when a sampling point with high energy is detected, the nonlinearity of the output signal can be greatly improved and auditory discomfort can be eliminated, effectively improving audio quality.

The present invention discloses a method for determining rock damage by using crack closure energy. Crack closure energy is defined as the energy required for the rock to effectively close the initial microcracks inside the rock under the action of a compressive load. To determine rock damage by using this method, firstly, a uniaxial compression test is performed on the rock, and the corresponding stress-strain curve is recorded; then, the crack closure stress of the rock is determined according to the stress-strain curve; the crack closure energy is marked in the stress-strain curve according to the crack closure stress; equidistant vertical auxiliary lines are drawn within the crack closure energy area to divide it into a number of small trapezoids; the area of ​​each small trapezoid is calculated and then summed; the total area obtained is the crack closure energy of the rock. The method of the present invention uses crack closure energy to determine the initial microcrack damage of the rock more objectively, is easy to program to realize batch operation, and can effectively evaluate the initial microcrack damage inside the rock.

The invention relates to establishment of a rotary table servo system model, in particular to an iterative identification method and a nonlinear disturbance data acquisition method of a linear model of a rotary table servo system. In order to solve the defects that in the prior art, data of a non-linear disturbance part is difficult to obtain and even cannot be measured, the iterative identification method for the linear model of the rotary table servo system comprises the steps that (1) an input signal of a first amplitude is used for exciting the rotary table servo system to identify and obtain the linear model; 2) respectively exciting the turntable servo system and the linear model by using input signals with the same frequency and the amplitude smaller than the first amplitude to obtain two output signals, and the difference value of the two output signals is a non-linear compensation value; the method comprises the steps of (1) obtaining a nonlinear compensation value, (2) obtaining a nonlinear compensation value, (3) obtaining a linear model through re-identification according to the difference value between the output signal in the step (1) and the nonlinear compensation value, (4) repeating the step (2) to obtain a new nonlinear compensation value, and (5) repeating the step (3) and the step (4) until the error of the linear model meets a set threshold value, and outputting the linear model.

This application provides a deep learning-based method and system for evaporator defect detection, relating to the field of anomaly detection. The method includes: acquiring multi-view image data of the evaporator; extracting multi-scale features, including the number of fins, their arrangement direction, and the spatial location of notches, through the backbone network layer of a deep learning target detection model; nonlinearly enhancing the multi-scale features through the model's feature enhancement network layer to obtain enhanced multi-scale features; and then extracting reconstructed features through a hybrid attention mechanism layer; based on the reconstructed features, obtaining the detection results of evaporator fins and notches through the model's detection layer, and comparing them with preset structural parameters to obtain the defect detection results. This application, used in evaporator defect detection, solves the technical problem of existing technologies being unable to stably identify evaporator defects under changing environmental and evaporator structural parameters.

This invention belongs to the field of low-light image enhancement, specifically relating to a low-light image enhancement method based on spectral interaction. The method includes constructing a low-light image enhancement model, which comprises a spatial feature extraction module, a spectral real part interaction module, and a brightness-perceived fusion module. The model is trained, and the trained low-light image enhancement model is used to enhance low-light images. This invention employs a two-branch fusion strategy, which effectively utilizes the advantages of both branches while avoiding their respective shortcomings, further improving image quality and clarity.

This invention discloses an optical-SAR image translation method based on a scattering feature enhancement strategy. The method proposes an image translation framework, SFEG, which mainly includes an SCG module and a GFE module. The SCG module first extracts edges from the input image, generates an edge mask by combining target annotation information, and models the scattering field using prior knowledge of the SAR target scattering characteristics. Finally, the generated scattering field and the corresponding optical image are input into the generator. This module can effectively restore the scattering characteristics of the target in the input image. The GFE module includes a channel attention branch and a spatial attention branch, which can enhance the input features from the channel and spatial dimensions respectively, and combine nonlinear functions to achieve feature recombination, further enhancing the nonlinear and refined output capabilities of SFEG. This module can improve the actual effect of generating SAR images.

The application provides a filter, a duplexer, a multiplexer and a communication device. The filter comprises an input terminal, an output terminal, one or more series resonators and one or more parallel resonators, wherein at least one resonator in the one or more series resonators and / or the one or more parallel resonators is split into two series split resonators, and a phase delay factor PDC of the series split resonators satisfies a predetermined range. Through the processing scheme of the present disclosure, the non-linear characteristic of the filter is improved.

The invention discloses a calculation method for analyzing the tensile bearing characteristic of an anchor rod, and belongs to the field of numerical calculation of roadway support. The method comprises the seven steps of generating an anchor unit, defining a large deformation calculation mode, defining an attribute parameter, defining a fish function'inisise ', executing the fish function'inisise', defining a fish function'modifying ', and executing the fish function'modifying'. The bearing process after the anchor unit is pulled is divided into an elastic stage, a yield stage, a strain hardening stage, a strain softening stage and a fracture stage. According to the method, a yield stage formula, a strain hardening stage formula and a strain softening stage formula are provided, and the formulas are embedded into a fish function'modifying '. The yield characteristic, the strain hardening characteristic and the strain softening characteristic of the anchor unit can be simulated based on the fish function'modifying '. According to the invention, a fracture stage is designed for the anchor unit. And when the anchor unit is pulled and enters a fracture stage, the bearing capacity of the anchor unit is instantaneously reduced to zero. Compared with original anchor unit calculation, the method can better simulate the tensile bearing characteristics of the anchor rod.

The present application discloses an encoding method and a phase interpolator of a phase interpolator, aiming to solve the problem of large nonlinearity of the output phase of the existing phase interpolation structure. Two channels are selected by digital signals, and the phases are summed according to different weights to obtain an output phase signal. The phase interpolator includes a phase input signal module, a weight selection module and an output phase signal module connected in sequence, and the weight selection module includes a plurality of weighted channel units of the P channel, a plurality of weighted channel units of the N channel, and a decoding circuit; and also includes: a channel exchange module, which is arranged between the phase input signal module and the weight selection module; when the code value Code reaches the middle value of its entire variation range, the decoding circuit exchanges the connected P channel and N channel to improve the nonlinearity of the output phase.

The present disclosure relates to a 5th generation (5G) or pre-5G communication system for supporting a higher data transfer rate than that of a 4th generation (4G) communication system such as long term evolution (LTE). According to embodiments of the present disclosure, a method performed by a base station in a wireless communication system includes receiving an uplink signal from a terminal configured with timing advance (TA) in an uplink period according to a time division duplex (TDD) configuration, transmitting an advance signal for a specific duration between an end point of receiving the uplink signal and a start point of a downlink period of the TDD configuration, through a transmission path of the base station having a power amplifier as a transmission end and transmitting a downlink signal to the terminal in the downlink period.

The present invention discloses a Y-type benzimidazole third-order nonlinear optical material and its preparation method and application. The benzimidazole molecule has a conjugated large π plane structure and also has a plurality of modifiable active sites. The present invention uses benzimidazole as a matrix, introduces an electron-withdrawing group at one end to increase the electron-withdrawing property of the matrix, and connects different electron-donating groups at the other end through Suzuki coupling to form a Y-type, thereby enhancing the charge transfer capability within the molecule and producing better nonlinear optical performance. The technical solution adopted by the present invention is: the present invention provides a type of formula (DEN): #imgabs0# In the formula (DEM): R1 is a nitro group or a cyano group, and R2 is a triarylamine group or a phenyl group.

The invention discloses an asymmetric thrust-disk-free three-degree-of-freedom axial-radial double-piece hybrid magnetic bearing, wherein a rotor outer shell is provided with two three-pole radial stators with the same structure respectively located on both sides of a rotor center point and two three-pole axial stators with the same structure respectively located on both sides of a rotor center point, a circular permanent magnet ring is fixedly embedded between the two stator yokes, projections of a total of six three-pole radial stator magnetic poles of the two three-pole radial stators on a radial cross section are evenly arranged in a circumferential direction, and three axial stator magnetic poles and three radial stator magnetic poles of the three-pole axial stator on the same side of the rotor center point are evenly arranged in a circumferential direction; a set of circular bias flux regulating coils is wound between the permanent magnet ring and the rotating shaft, and a variable topology six-pole structure of two three-pole bearings with a rotation direction difference of 60 degrees is adopted, so that the structure is more compact and the critical speed is improved.

This application relates to the field of wind power generation and discloses a micro-site selection method for polar wind farms based on the wake and icing characteristics of complex terrain. The method includes: modeling the terrain scene using spatial environmental parameters for each meteorological time zone, and initializing the meteorological environmental parameters; discretizing the wind farm area into a grid, and using mixed-integer linear programming for global optimization to determine the coarsely selected grid area where the wind turbines are located; and using a multi-stage adaptive hybrid heuristic algorithm that integrates multiple population genetic algorithms and particle swarm optimization to search in the continuous coordinate space corresponding to the coarsely selected grid area, and optimizing the wind turbine layout scheme by solving the fitness function. This eliminates the overestimation of the actual wind farm capacity under extreme blizzard conditions by traditional general models, ensuring the reliability of polar energy system planning.

A crude oil futures price prediction method based on Dens-TCN and CNN comprises the following steps: step 1, obtaining crude oil time series data and related macroscopic data, constructing a comprehensive data set, and preprocessing the data; 2, decomposing the preprocessed data by adopting EEMD (Ensemble Empirical Mode Decomposition), and extracting fine price fluctuation characteristics; 3, constructing a comprehensive prediction model based on the Dens-TCN and the CNN; 4, training a comprehensive prediction model, and performing parameter optimization on the model; and 5, combining the prediction results of the Dens-TCN and the CNN by adopting an equal weight method to generate a high-precision prediction result of the future crude oil future price. According to the method, the advantages of the Dens-TCN and the CNN are fused, the multi-level characteristics of the crude oil price can be effectively captured, emergencies and nonlinear fluctuation in a complex market environment can be dealt with, high prediction precision and stability are achieved, and reliable decision support is provided for investors, enterprises and policy formulators.

The present disclosure relates to the field of quantum model technology, and proposes a quantum-derived pulse convolutional neural network, an image processing method and a system. The pulse convolutional neural network includes a preprocessing module, a pulse generation layer, a pulse neuron and a pulse counter connected in sequence; the pulse neuron includes one or more convolution groups connected in sequence, one or more quantum-derived convolution operation layers and a quantum-inspired neuron; a convolution group includes a convolution layer, a ReLU activation layer and a pooling layer; a quantum-derived convolution operation layer is used to replace one or more convolution layers of the pulse convolutional neural network, and the Relu activation layer after the quantum-derived convolution operation layer is removed, and the quantum-derived convolution operation layer performs a quantum-derived convolution operation. The present disclosure introduces quantum-derived technology into the pulse convolutional neural network, which can effectively improve the efficiency of the pulse convolutional deep neural network and enhance the robustness of the pulse convolutional deep neural network.

The invention belongs to the field of optical communication security equipment, and particularly relates to a system and a method for generating electro-optical chaos based on a mutual injection feedback loop, the system comprises a first feedback branch and a second feedback branch, the first feedback branch comprises a Mach-Zehnder modulator 1 and an optical fiber delay line 1; the second feedback branch corresponding to the first feedback branch comprises a Mach-Zehnder modulator 2 and an optical fiber delay line 2, a chaotic signal output port is arranged between the Mach-Zehnder modulator 1 and the optical fiber delay line 1 or between the Mach-Zehnder modulator 2 and the optical fiber delay line 2, and the delay time of the optical fiber delay line 1 and the delay time of the optical fiber delay line 2 are different. The initial phase of the Mach-Zehnder modulator 1 and the initial phase of the Mach-Zehnder modulator 2 are different, the system has the advantages of being simple in structure, good in delay characteristic suppression effect and higher in chaos safety, a chaos synchronization system corresponding to the system is established at the receiving end, and the secrecy capacity of the whole communication system is high.

This invention relates to the field of precision drive and positioning technology, and discloses a three-dimensional macro-micro composite positioning stage, which includes a stage body, an X-axis motion component, a Y-axis motion component, a Z-axis motion component, and a mobile terminal platform connected to the X-axis motion component, Y-axis motion component, and Z-axis motion component respectively. The X-axis motion component includes an X-axis macro-drive and an X-axis micro-drive; the Y-axis motion component includes a Y-axis macro-drive and a Y-axis micro-drive; the Z-axis motion component includes a Z-axis macro-drive and a Z-axis micro-drive, used to jointly drive the mobile terminal platform to linear motion along the X-axis, Y-axis, and Z-axis. This invention balances large stroke and high precision, achieving three-dimensional positioning with millimeter-level stroke and nanometer-level precision. The X-axis, Y-axis, and Z-axis micro-drives use piezoelectric ceramic direct drive, resulting in high positioning accuracy and fast response speed. The overall structure of this invention is reasonable and compact, and its operation is reliable.

The invention is suitable for the field of power load prediction, and discloses a load time sequence mixed quantum fusion prediction method, system, device and medium, and the method comprises the steps: obtaining multi-channel historical load time sequence data, carrying out the sequence embedding processing, obtaining an initial sequence, and inputting the initial sequence to a plurality of layers of star aggregation-distribution modules which are sequentially connected in series; wherein the star aggregation-distribution module adopts a first quantum parallel hybrid network to carry out network processing on an initial sequence to obtain a multi-sequence representation; performing random pooling operation through multi-sequence representation to obtain core representation; the core representation and the multi-sequence representation are spliced, a second quantum parallel network is adopted for fusion, and an output sequence is obtained; the output sequence of the previous module is used as the input of the next module, and the final output sequence is used as a power load prediction result. According to the method, the parallel mixed classical quantum regression network is introduced into the star aggregation-distribution module, so that complex time sequence data is effectively processed, and the time sequence prediction performance is improved.

The present application relates to an image restoration method, device, computer equipment, storage medium and program product. It can be used in the field of financial technology or other related fields. The method comprises: obtaining a target defective image; inputting the target defective image into a pre-trained image restoration model, performing feature extraction processing on the target defective image through a downsampling module and a feature extraction module in the image restoration model respectively, fusing the feature maps output by the downsampling module and the feature extraction module and then performing upsampling processing to output the target restoration image; wherein the feature extraction module comprises a cascaded first feature extraction layer, a first slicing layer and a first connection layer, the first feature extraction layer is used for spatial feature extraction, the slicing layer is used for slicing the feature map output by the first feature extraction layer, and the first connection layer is used for connecting and outputting multiple feature maps. The use of this method can improve the quality of image restoration.

The application discloses a near-transform limit nanosecond pulse mode-locked fiber laser with periodic bifurcation and a producing method. The laser is a fiber ring cavity which comprises a wavelength division multiplexer, an erbium-doped fiber, a first polarization controller, a polarization detector, a dispersion control unit, an output coupler, a fiber isolator and a second polarization controller which are connected in sequence in a closed loop. The method comprises the following steps: coupling pump continuous light into the laser through the wavelength division multiplexer; absorbing the pump continuous light by the erbium-doped fiber and radiating gain pulses in a long-wave band; oscillating the generated pulses in the fiber laser cavity; through the cooperation of the first polarization controller, the polarization detector and the second polarization controller, generating equivalent saturated absorption effect by using the nonlinear polarization rotation generated when the gain pulses propagate in the fiber, and realizing pulse mode-locking; and making the fiber laser work in an abnormal dispersion region so that the generated pulses satisfy the soliton area theorem and near-transform limit pulses are generated. The application can realize nanosecond pulse periodic bifurcation output of a short-cavity fiber laser.

The invention provides a pressure sensing element and a manufacturing method thereof. A pressure sensing element (100) for a pressure sensor device includes a substrate portion (102) including a cavity (104) formed therein, a mouth (184) of the cavity (104) having an open periphery (186). A film layer (106) covers an open periphery (186) of the cavity (104), and a strain gauge element (166, 168, 170, 172) is disposed on the film layer (106) over the cavity (104). The cavity (104) includes a main cavity (118) and a first accessory (120) integrally formed with the main cavity (118).