77results about How to "Effective refactoring" patented technology

The invention aims at solving the problem that in the existing motors which operat at high speed and adopt a single-current sampling method to reconstruct phase current, and space vector pulse width modulation (SVPWM) can not finish phase current reconstruction in a non-observation area at the position of a sector boundary and providing a motor phase current reconstruction method based on symmetric pulse width modulation (PWM) carrier wave, wherein algorithm of the method is easy to achieve, and a control effect of the method is relatively better. The method decomposes a modulation vector into two non-zero basic space vectors adjacent to a non-zero basic space vector in a sector boundary area, and the single-current sampling method is adopted to reconstruct motor phase current in an acting process of the adjacent two non-zero basic vectors. The method enables the non-observation area in the sector boundary area to adopt a symmetrical PWM method to modulate, converts the non-observation area into an observation area, and can reconstruct the motor phase current in the non-observation area in single-current sampling method.

The invention provides a time sequence signal efficient denoising and high-precision reconstruction modeling method and system. The method comprises: carrying out data preprocessing on original pulsewave signals; selecting a preset signal duration, and dividing the pulse wave signals after data preprocessing into a prediction set, a training set and a test set; selecting a convolutional neural network as a basic model of the deep convolutional noise reduction auto-encoder, and obtaining a deep convolutional noise reduction auto-encoder model according to a signal denoising requirement; inputting the training set into a deep convolution noise reduction auto-encoder model for training, and optimizing and selecting parameters of the deep convolution noise reduction auto-encoder model by using the regularization parameters and the test set to obtain an optimal deep learning model; and inputting the noisy pulse wave signal prediction set into the optimal deep learning model to obtain deepstructure features, performing signal reconstruction and denoising processing, and evaluating model performance. According to the method, denoising and reconstruction of the pulse wave signals are effectively carried out, and a new thought is provided for filtering same-frequency noise interference in the pulse wave signals.

The invention discloses a zero sample learning method based on a self-coding generative adversarial network. The method comprises the following steps: inputting visual features of visible category samples and corresponding category semantic features; Inputting a specific balance parameter Lambda and alpha numerical value; setting Initial values and learning rates of the parameters are set, using an Adam optimizer is to train the self-encoding confrontation generation network provided by the invention, and obtiaining model parameters of an encoder and a decoder; Inputting the semantic featuresof which the categories are not seen, and synthesizing the visual features of the corresponding categories by utilizing the trained model parameters; And classifying the test samples of which the categories are not seen. According to the invention, the semantic relationship between the visual modality and the category semantic modality can be effectively aligned. Visual information and category semantic information are fully fused together, semantic association between two modes can be more effectively mined, and more effective visual features are synthesized.

The invention discloses an abnormal behavior detection method based on video monitoring, and the method comprises the steps: quickly and accurately detecting a foreground target object in a video frame image through a YOLOv3 target detection algorithm, removing the impact from background noise, and meeting the real-time requirements of abnormal detection; extracting features of a target object inthe video frame image; firstly, features are clustered; inputting the features into an SVM classifier, obtaining an abnormal score with the highest score as the target object, finally obtaining the highest value in the abnormal scores of all the target objects in the video frame image as the abnormal score of the frame image, carrying out the quick and accurate classification through the SVM classifier, and meeting the real-time requirements. According to the method, the deep learning method and the machine learning method are adopted, the occurrence of the abnormal event can be effectively detected, the requirement of real-time detection can be met, and the accuracy of abnormal event detection is improved.

The invention relates to a vehicle foresight cruise control method based on a high-precision map. The vehicle foresight cruise control method includes the following steps of vehicle positioning, map transmission, map reconstruction and foresight cruise vehicle speed planning. According to the vehicle foresight cruise control method, ahead road condition information is predicted in real time basedon a GPS and the high-precision map, the cruise vehicle speed is adaptively adjusted under different working conditions, the relationship between low fuel consumption and high time efficiency is balanced, the transportation cost is saved, the transportation efficiency is improved, and compared with ordinary cruise control, the fuel economy of a vehicle is greatly improved on the basis of ensuringthe time efficiency of the vehicle.

The invention belongs to the field of radar signal processing, and particularly relates to an SA-ISAR (Sparse Aperture-Inverse Synthetic Aperture Radar) self focusing method based on structure sparsity and entropy joint constraints. The method comprises the following steps of Step 1, performing echo modeling on radar echo subjected to envelope alignment; Step 2, applying layered structured sparseprior to an ISAR image; Step 3, updating the ISAR image and an upper layer variable by a relax variational bayes method; Step 4, updating a phase error through a minimum entropy method based on fixedpoints; and Step 5, judging whether a termination condition is reached or not, stopping iterative loop if the termination condition is reached, returning to the Step 3 if the termination condition isnot reached, and outputting an image subjected to self focusing after the termination condition is reached. The SA-ISAR self focusing method has the advantages that the self focusing precision of theISAR images at the sparse aperture can be improved, so that the formed ISAR images are clearer; the calculation complexity is lower; the iterative convergence speed is faster; and the sparse apertureresistance capability is high.

The invention relates to a multispectral and panchromatic image fusion method based on a dense and jump connection deep convolutional network. The method comprises two parts of model training and image fusion, and is characterized by at the model training stage, firstly, performing the down-sampling on an original clear multispectral image and a panchromatic image to obtain a simulation training image pair; secondly, extracting the characteristics of the simulated multispectral and panchromatic images, fusing the characteristics by utilizing a dense connection network, and reconstructing a high-spatial-resolution multispectral image by utilizing jump connection; and finally, adjusting parameters of the model by using an Adam algorithm; at the image fusion stage, firstly, extracting the features of multispectral and panchromatic images, fusing the features by utilizing a dense connection network, and reconstructing a high-spatial-resolution multispectral image in combination with jump connection, wherein the two feature extraction sub-networks are responsible for extracting the features of the input image pair, and the three dense connection networks are responsible for fusing the features, the jump connection and two transposed convolutions are responsible for reconstructing the high-spatial-resolution multispectral image.

The invention provides a DRAM (Dynamic Random Access Memory) source synchronization test method and circuit solving the technical problems of complexity and low reliability of a test way in the prior art. The circuit is integrated in a DRAM and can precisely measure source synchronization time parameters, and whether the parameters accord with a code standard can be tested through running once. When the test circuit is activated, driving circuits and receiving circuits of a DQ (Data Strobe) base pin and a DQS (Data Strobe Signal) base pin are opened simultaneously. The test circuit comprises an adjustable delay unit, a DQS edge can be moved relative to a DQ edge, and the adjustable delay unit is arranged before the receiving circuit of the DQS base pin; the adjustable delay unit can also be placed behind the receiving circuit of the DQS base pin, and the DQ base pin receives a DQS in a delay way. By using the DRAM source synchronization test method and circuit, a source synchronization test function which can not be realized by a test machine can be restructured effectively; and the advantages of simplicity and convenience in test operation and better precision and reliability areachieved.

The invention discloses a broadband frequency spectrum monitoring system and a high-speed pseudorandom sequence signal generating method thereof. The broadband frequency spectrum monitoring system comprises a signal preconditioning module, a power divider, a mixer, a low-pass filter, an intermediate frequency amplifier, an FPGA module and a reconstitution algorithm device, wherein the output end of the signal preconditioning module is connected with the input end of the power divider, high-speed pseudorandom sequence signals generated by the FPGA module and signals output by four output ends of the power divider are input into the input end of the mixer, four output ends of the mixer are connected with the input end of the intermediate frequency amplifier through the low-pass filter, the four output ends of the intermediate frequency amplifier are connected with the FPGA module, and the FPGA module is connected with the reconstitution algorithm device. Noise interference is reduced, system stability is improved, and the requirement for processing broadband sparse signals of a modulating broadband converter system is met.

The invention discloses a mesh reconstruction method of the direct discrete solution solving of an optimal control problem, and aims at solving the problems that the optimizing computation consumes alarge amount of time due to excessively large time amount given by an existing mesh reconstruction method or too many iteration times, or an optimizing result is not an ideal result because the discrete accuracy cannot be ensured, and an existing method normally fails to accurately find the structural switching point of a system. The method has the advantages that the direct discrete solution solving variable scale of the complicated optimal control problem can be reduced, the computational cost is lower, the iteration time is lesser, and a high quality solution is obtained through lesser parameters. The method is suitable for the on-line optimization of a large-scale complicated dynamic optimization problem. The method is rapid and effective, the scale of a discrete nonlinear programmingproblem can be reduced to the maximum extent to reduce the iteration time, and the structural switching point of the system can be rapidly and accurately positioned.

The invention discloses a building construction quality actual measurement data acquisition device. The device comprises a vehicle body, a control mechanism and a data acquisition mechanism, the vehicle body comprises a rear suction cup set, a rear vehicle frame base, a steering engine, a gear, a rack, a front suction cup set, a front vehicle frame base, a spherical hinge and an electric push rod.The control mechanism comprises a single-chip microcomputer, a controller, a battery and a power switch; the data acquisition mechanism comprises an ultrasonic sensor and a three-dimensional laser scanner. The building construction quality actual measurement data acquisition device can rapidly, effectively, completely and automatically move and can achieve the building construction quality actualmeasurement data acquisition with high-precision so as to achieve the complete and high-precision measurement for the whole object to be measured and directly perform rapid reverse three-dimensionaldata acquisition and model reconstruction from the measured object, and each piece of three-dimensional data in the laser point cloud is the real data of the directly measured object, so that the acquired actual measurement data is real and reliable, the traditional single-point measurement method is broken through, and the labor and time are effectively reduced.

The invention provides an interferometric three-dimensional imaging method for space high-speed moving targets. The method comprises the following steps of 1, subjecting the echo signals of three antennas to translational-motion compensation and dechirping treatment, and constructing a joint parameterization sparse representation model with an unknown speed; 2, by using an improved OMP algorithm, solving the joint parameterization sparse representation model to obtain a moving speed, the location and the distance of each scattering point, and the interferometric phase information; 3, by using the obtained interferometric phase information, conducting the interference avoidance and obtaining the azimuth and the elevation position of each scattering point. Therefore, the influence of the high-speed movement of a space target is compensated. Meanwhile, the three-dimensional imaging of the space target is realized effectively.

The invention discloses a non-convex compression preception optimization reconstruction method based on sketch representation and structured clustering. The method mainly settles a problem of inaccurate compression image reconstruction on the condition of low sampling rate. The method comprises the following steps of according to a sketch of an image, defining a sketchable block and a non-sketchable block, wherein the non-sketchable block comprises a smooth block and a patterned block, and the sketchable block comprises a unidirectional block and a multidirectional block; performing clustering based on sketching direction guidance on the unidirectional block; performing clustering based on a direction distribution characteristic on the multidirectional block; performing gray scale clustering on the smooth block and the pattern block; performing multi-measuring-vector observation on each kind of image blocks; and in reconstruction, executing a particle swarm optimization algorithm based on crossing and atom direction restraint according to multiple measuring matrixes, kind index and direction information of each kind of image blocks for obtaining a final reconstructed image. Compared with a TS-RS method and an NR-DG method, the non-convex compression perception optimization reconstruction method has advantages of high quality of the reconstructed image, high robustness and high suitability for reconstruction of a natural image.

The invention discloses a short-term power load prediction model establishment method based on EMD-VMD-PSO-BPNN. The short-term power load prediction model is applied to power load prediction of a papermaking enterprise, and comprises the following steps: firstly, obtaining data of a total effective load with qualified data quality of the papermaking enterprise; performing sequence decomposition on the total effective load by adopting an EMD-VMD combination algorithm; reconstructing the decomposed sequence by adopting approximate entropy; selecting a model to input by using a lagging autocorrelation method; adopting PSO-BPNN to model the reconstructed sequence; and training the PSO-BPNN model by adopting the training sample, establishing a prediction model, predicting the power consumptionload of the papermaking enterprise, and finally analyzing the prediction effect. The short-term power load prediction model is established based on the EMD-VMD-PSO-BPNN algorithm, and the method hasthe advantages of being fast in model convergence, high in prediction result precision, free of lag and the like.

The invention relates to a three-dimensional curved surface reconstruction method suitable for a biological membrane, and belongs to the technical field of biology. The main technical scheme is as follows: cleaning and sampling, fixing, rinsing, dehydrating, permeabilizing, waxing, embedding, slicing, sticking and baking, dewaxing, dyeing and sealing; adjusting the wavelength of emitted light, adjusting the amplification factor, photographing and storing images; and building a three-dimensional model. The three-dimensional curved surface reconstruction method suitable for the biological membrane provided by the invention is proposed for a biological membrane with the thickness greater than 0.1 mm, and in order to avoid the problem that a dye and a laser light beam cannot effectively enterthe interior of the biological membrane due to the fact that the biological membrane is excessively thick, thus the characterization is difficult. By using the technology, the complex physical form ofthe biological membrane can be effectively reconstructed, meanwhile, a relatively high reduction degree is ensured, and a feasible means is provided for researching the form change of the biologicalmembrane and influence factors.

The embodiment of the invention discloses a power signal filtering method and system using shrinkage gradient. The method comprises the following steps: step 101, acquiring a signal sequence S acquired according to a time sequence; 102, calculating a shrinkage factor beta; 103, solving an initial value x0<SM> of the shrinkage sequence x<SM>; 104, creating an iterative control parameter k and assigning 0, and creating a contraction step length tk and assigning delta T; step 105, solving a gradient vector shown in the specification; step 106, solving a shrinkage gradient function gk; 107, judging whether the shrinkage gradient function gk is smaller than an absolute value shown in the specification or not and obtaining a first judgment result; step 108, adding 1 to the value of the iterativecontrol parameter k, and calculating the kth step value of the shrinkage vector x<SM>; step 109, solving an adjacent error e; 110, judging whether the adjacent error e is greater than or equal to a preset threshold epsilon 0 or not to obtain a second judgment result; 111, recording a signal sequence SNEW from which noise is filtered, specifically, SNEW=xk<SM>;

The invention discloses a mine low-voltage cable insulation fault detection device and method based on an infrared technology. The detection device comprises an infrared detection safety cap and a bracelet display device. The detection method comprises a first step of acquisition of an infrared radiation signal, primary filtering de-noising and analog-to-digital conversion processing, a second step of secondary filtering de-noising of a digital current signal, a third step of obtaining and display output of a mining low-voltage cable temperature value, a fourth step of output of a mining low-voltage cable infrared thermogram, and a fifth step of repetition from the first step to the fourth step for multiple times until completion of fault detection of a whole mining low-voltage cable. The insulation fault detection of the mining low-voltage cable can be completed simply and rapidly, detection efficiency is high, and the existing problems of complex use operation, low detection efficiency, low testing precision and the like which exist to different extents on the basis of an additional power supply detection method are solved.

The invention discloses a multimedia data time correction method, a computer device and a computer readable storage medium, the method comprises the steps that time data of multimedia data are obtained, and whether the time data are continuous time data is judged; if it is determined that the time data are not continuous time data, first fitting straight lines of the time data are calculated according to the time data in the historical data of the multimedia data, the distances between the time data of the current multimedia data and the first fitting straight lines are calculated, if the distances are larger than a preset distance threshold value, the time data in the historical data of the multimedia data are cleared, the time data of current and follow-up received multimedia data are used for calculating second fitting straight lines, and the second fitting straight lines are used for calculating the correction time data of the current and follow-up received multimedia data. The multimedia data time correction method can be realized by the computer device. According to the multimedia data time correction method, the computer device and the computer readable storage medium, the situation that the video stops or the video breaks down when the multimedia data are subjected to time reconstruction can be avoided.

The embodiment of the invention discloses a power signal filtering method and system using Dantzig total gradient minimization. The method comprises the following steps: step 101, acquiring a signal sequence S acquired according to a time sequence; step 102, solving the total gradient sparsity p of the Dantzig; 103, solving a delay matrix D; step 104, solving a Dantzig total gradient vector gamma;and step 105, solving a signal sequence Snew after the noise is filtered out.

The invention discloses an electromechanical equipment few-sample degradation trend prediction method of an unsupervised meta learning network, relates to the technical field of service performance evaluation and prediction of electromechanical equipment, and solves the technical problem that an existing meta learning method generally depends on label sample support and is difficult to be directly applied to historical data with scarce labels. According to the technical scheme, the method is characterized in that by aggregating the training process of each inner loop, cross-task outer loop optimization and training are carried out on model parameters obtained by training a support set of each training set through a support set of a test set, and finally an unsupervised meta-learning agent model is generated; and the classic deep circulation network is effectively reconstructed, the classic deep circulation network has remarkable generalization ability under excitation of few samples, connection is established between historical large sample data and insufficient prediction samples, and the problem of labeling of historical label-free data is effectively solved.

The invention relates to the technical field of automobile production, in particular to a vehicle VIN character recognition and character carving depth detection system and detection method. The method comprises: enabling a to-be-measured workpiece to be located at an initial scanning position of the measuring device; photographing the VIN code for two-dimensional recognition of characters; scanning the obvious defects; carrying out scanning by moving a workpiece to be detected or a sensor, so that a laser strip connecting line of the sensor scans a VIN code character area, and carrying out continuous photographing to obtain an image; calculating three-dimensional information; reconstructing characters according to the three-dimensional result, and comparing the reconstructed characters with a two-dimensional recognition result; and if the three-dimensional identification result is consistent with the two-dimensional identification result, continuing to output the detection result. According to the invention, the accuracy of the identification effect is improved, the measurement convenience is improved, the high precision and repeatability of the measurement are ensured, the measurement can be realized, and the measurement result can be effectively analyzed and evaluated.

The invention relates to an improved SAMP underwater acoustic channel estimation algorithm. The invention aims at a pilot signal of a subcarrier in an MIMO-OFDM underwater acoustic communication system, based on the improvement of an SAMP (Sparsity Adaptive Matching Pursuit) algorithm, the original iteration condition is replaced by setting the second-order difference of adjacent residual errors,and clutters are further filtered according to the channel pulse energy ratio, so that the system performance can be effectively improved. Figure 1 in the abstract drawing of the specification is a specific implementation flow chart of the invention.

The embodiment of the invention discloses a transformer operation state vibration detection signal reconstruction method and system using data regularization. The method comprises the following steps:step 1, inputting a measured vibration sound signal sequence S; step 2, carrying out data conversion on the vibration sound signal sequence S to obtain a vibration sound signal segment sequence si, i= 1, 2 - I, wherein I represents the number of the vibration sound signal segment sequences; step 3, carrying out regularization processing on the vibration sound signal segment sequences si, i = 1,2 - I, and obtaining a vibration sound signal segment sequence with noise filtered: FORMULA, wherein formula is the square of a l2 model, x is a first temporary vector, and omega is a regularization vector, wherein the value is as follows: omega = [ 0 1 1 0 1 1 0 ] T; and step 4, rearranging the vibration sound signal segment sequence si, i = 1, 2 - I, and obtaining a reconstructed vibration soundsignal sequence SNEW, wherein t is a second temporary vector, and the value of the second temporary vector is as follows: if 7I = N, t = SNEW, otherwise, FORMULA, wherein the formula represents elements from No. 7 (I-1) + 1 to No. N in the Ith vibration sound signal segment sequence with noise filtered.

The embodiment of the invention discloses a power signal filtering method and system utilizing an energy scale. The method comprises the following steps: step 101, acquiring a signal sequence S acquired according to a time sequence; step 102, solving an energy scale coefficient sequence; step 103, solving an edge energy scale coefficient; step 104, solving a direction energy scale coefficient; step 105, solving a direction scale threshold value; step 106, solving a high-density energy scale coefficient; and step 107, solving a signal sequence after noise filtering.

An embodiment of the invention discloses a method and system for reconstructing power signals based on a prediction matrix, and the method comprises the following steps of step 1, inputting a measuredpower signal sequence S; and step 2, performing data reconstruction on the power signal sequence S, wherein the reconstructed power signal sequence is SNEW, and specifically SNEW=W*SSEG, wherein W represents the prediction matrix, W* represents a pseudo-inverse of the prediction matrix, and SSEG represents a set of segment sequences.

The invention relates to a vehicle foresight cruise control method based on a high-precision map. The vehicle foresight cruise control method includes the following steps of vehicle positioning, map transmission, map reconstruction and foresight cruise vehicle speed planning. According to the vehicle foresight cruise control method, ahead road condition information is predicted in real time basedon a GPS and the high-precision map, the cruise vehicle speed is adaptively adjusted under different working conditions, the relationship between low fuel consumption and high time efficiency is balanced, the transportation cost is saved, the transportation efficiency is improved, and compared with ordinary cruise control, the fuel economy of a vehicle is greatly improved on the basis of ensuringthe time efficiency of the vehicle.

An attack packet source tracing method and system based on dynamic probabilistic dual-domain packet marking, the source tracing method includes the following steps: setting the marking space of the dual-domain marking domain; marking the two marking domains with opposite marking strategies; After the attack, the victim host counts all attack packets, and reconstructs the attack path in two steps: (1) all the tag information that appears in the tag field is the router that constitutes the attack path; (2) count the number of occurrences of the tag information; m 1 、m 2 Denote the two label domains in the above label space respectively; m 1 The router with more appearance times is closer to the victim host, and the number of times decreases along the attack path to the attack host; m 2 The router with more occurrences in , the closer to the attacking host, and the number of times decreases along the attack path to the victim host. The invention also provides a system for realizing the above method, which satisfies the requirement of high security in the autonomous local area network.