91results about How to "Accurate and effective extraction" patented technology

The invention discloses a power system sub-synchronous oscillation mode identification method which is characterized in that time complexity of the matching pursuit search process is reduced by aid of a particle swarm optimization, sub-synchronous oscillation signals are subjected to atomic decomposition in an optimized matching pursuit method, and atomic parameters are converted to sub-synchronous oscillation mode parameters after the optimum atom is searched. The power system sub-synchronous oscillation mode identification method can accurately identify power system sub-synchronous oscillation mode parameters, has good time-frequency characteristic, is favorable for deeply analyzing power system sub-synchronous oscillation characteristic, accordingly effectively restrains oscillation, and ensures stable operation of a unit and a power system.

The invention discloses a malicious file identification method, a device and a storage medium. The method comprises the steps of adopting a preset learning set consisting of a malicious file and a normal file to generate a machine learning model; reading a file to be detected other than the learning set; converting the file to be detected to a vector; and performing the malicious file identification on the file to be detected which is converted to the vector through the machine learning model. The machine learning model is generated by the preset learning set consisting of the malicious file and the normal file, and the generated machine learning model is used for performing the malicious file identification on the file to be detected except the learning set, so that the virus characteristics can be timely, accurately and efficiently extracted, any discovered malicious file can be immediately processed, and the detection efficiency of the malicious file can be greatly improved.

The invention discloses a milling flutter recognition method based on variation modal decomposition and energy entropy, and belongs to the technical field of machine tool machining flutter recognition. The milling flutter recognition method comprises the following steps of S1, establishing a VMD mathematical model; S2, establishing a mathematical model of the energy entropy; S3, conducting three sets of milling machining experiments representing three cutting states, including stable cutting, a weak flutter and a severe flutter, and obtaining three sets of milling force signals through a dynamometer; S4, conducting FFT analysis on the three sets of milling force signals, and proving that the three sets of milling force signals represent the states of stable cutting, the weak flutter and the severe flutter respectively; S5, determining the number K of optimal modals of VMD decomposition and a penalty factor alpha through a VMD parameter automatic selection method based on a kurtosis value; S6, solving instantaneous frequencies of multiple IMF, and determining a milling flutter characteristic frequency band; S7, adopting a hammering experiment to obtain the modal of a knife; S8, extracting a flutter characteristic vector of each IMF based on the energy entropy. According to the milling flutter recognition method based on the variation modal decomposition and the energy entropy, the VDM decomposition effect is improved, and automatic flutter recognition is achieved.

The invention provides a method for extracting finger vein grain by using a directional filtering technique, comprising the steps of positioning finger regions, strengthening filtering and extractinga finger vein mode. The method concretely comprises the following steps: firstly, dividing finger regions on a read-in vein image by using a Kapur entropy threshold value method; secondly, removing burrs for the finger regions by adopting an opening operation in mathematical morphology; thirdly, acquiring a directional diagram of the finger vein region and designing a directional filter by combining the characteristic of the vein grain; fourthly, carrying out filtering enhancement on the image by combining the acquired directional diagram and the directional filter; and finally, carrying outa binaryzaiton operation to extract the finger vein mode by adopting a NiBlack method. The finger vein grain extracted by the method has good connectivity and smoothness and less noise.

The invention provides an image description generation method and device and electronic equipment, and relates to the field of artificial intelligence. The method comprises the following steps: acquiring an input image, and encoding objects contained in each image area in the input image to acquire a first image feature; processing pixels in the first image feature according to a preset rule, anddetermining a second image feature according to the processed pixels; and on the basis of the second image feature and the initial word vector, decoding the region feature corresponding to each imageregion in the first image feature at different moments to obtain a word vector corresponding to each image region, and forming text description corresponding to the input image according to the word vector, the initial word vector being an initial mark of the text description. According to the image description generation method, on one hand, natural language information contained in the input image can be extracted more accurately and effectively; on the other hand, the method can be applied when sentences are long or sentence pattern structures are complex, and the accuracy and fluency of text description are improved.

The invention discloses a fault diagnosis method for an induction motor of an electric automobile based on broad learning. According to the method, an incremental broad learning (IBL) method is adopted to diagnose a fault of the induction motor, the structure of IBL is simple, and the network can be effectively trained and retrained. Feature extraction (empirical mode decomposition and sample entropy), incremental broad learning and singular value decomposition are combined to form an intelligent diagnosis method for the induction motor of the electric automobile. Experimental results show that the method is superior to other algorithms in fault diagnosis for the induction motor of the electric automobile. In addition, the IBL simplified by singular value decomposition (SVD) is smaller inerror and higher in system stability than the previous IBL.

The invention discloses an accurate detection method for the voltage travelling wave of a transmission line fault based on waveform inversion. According to the method, special voltage travelling wave sensors are mounted on grounding wires of capacitive equipment, as capacitive voltage transformers (CVT), of transformer stations at the two ends of a power transmission line; the special voltage travelling wave sensors are utilized to detect a primary voltage traveling wave signal after the transmission line fault and output a secondary voltage traveling wave signal; and a high-frequency transfer function model of the special voltage travelling wave sensors is used to establish a high-frequency inversion function model of the special voltage travelling wave sensors, according to the high-frequency inversion function model, a random resonance-deconvolution inversion algorithm is designed for waveform inversion, the primary voltage traveling wave signal is output and inversed, and accurate inversion from the second voltage traveling wave signal to the primary voltage traveling wave signal is realized. According to the invention, the problem of waveform distortion generated when the primary voltage traveling wave signals is transformed by the special voltage travelling wave sensors, and the primary voltage traveling wave signal can be detected accurately and effectively.

The invention provides a wind power gearbox fault diagnosis method based on the self-adaptive resonance sparse decomposition theory, relates to the wind power gearbox fault diagnosis method based on the self-adaptive resonance sparse decomposition theory, and aims at solving the problems that the early composite fault signals of the existing wind power gearbox have strong background noise and fault characteristic information is weak. The concrete process of the method comprises the steps that step one: the vibration data of the test gearbox are acquired by using a BBM noise vibration detection system so that fault vibration signals of planetary carrier bearing outer ring peeling and planetary gear local peeling are obtained; step two: the quality factor and the proportional coefficient of resonance sparse decomposition are simultaneously optimized so that X* is obtained; step three: X* is substituted in a self-adaptive resonance sparse decomposition method so that the high and low resonance components of the fault vibration signals are obtained; and step four: envelope analysis is performed on the high resonance component of the fault vibration signals so that fault information and the non-fault vibration signals are identified. The wind power gearbox fault diagnosis method based on the self-adaptive resonance sparse decomposition theory is suitable for the field of fault diagnosis.

The invention discloses a quick human face recognition method based on binocular vision measurement technology, which comprises the following steps: effective human face features are extracted in a binocular image, and effective human face regions are divided; the binocular vision measurement technology is then used for carrying out size measurement on extracted human face regions; according to the calculated human face region size, a human face database is divided into multiple human face sub databases; and finally, in a recognition stage, the actually-calculated human face regions are compared with the human face sub databases. Through carrying out comparison recognition on the actually-calculated human face regions and the human face sub databases, the times for comparison recognition are reduced, and the human face recognition speed is greatly enhanced; the actual size of the human face is used for carrying out comparison recognition on the human face sub databases, recognition errors are small, and the human face recognition accuracy rate is effectively improved; and in addition, the human face recognition method also has good universality.

The present invention discloses a base coordinate system calibration method of an aircraft panel horizontal automatic drill riveter. The method comprises the following steps: 1) setting dead center positions of a rammer tight side machine tool and a drilling sprigging side machine tool in the aircraft panel horizontal automatic drill riveter, and drawing a coordinate system definition diagram; 2) respectively installing reflectors for measurement of a laser tracker at the machine tools at two sides of the aircraft panel horizontal automatic drill riveter, and appropriately arranging the laser tracker; 3) controlling translational shafts of two numerical control positioning devices in the aircraft panel horizontal automatic drill riveter to move to assigned positions in order, obtaining position coordinates of the reflectors on the two devices through tracking measurement of the laser tracker, fitting a straight line to express as the coordinate axis direction of each numerical control positioning device base coordinate system, and establishing the real positioning pose model of the base coordinate systems of the two numerical control positioning devices; 4) calculating the rotation matrix and the translation matrix from the base coordinate systems of the two numerical control positioning devices to a measurement coordinate system, and obtaining relative pose transformation matrixes R and T between the base coordinate systems of the two numerical control positioning devices.

The invention discloses a wind power subsynchronous oscillation detection and inhibition method based on empirical mode decomposition, comprising: employing an empirical mode decomposition algorithm to respectively rapidly decompose collected three phase current signals, respectively comparing the amplitude of each decomposed low frequency signal with a set value, and determining occurrence of subsynchronous oscillation according to comparison results. The subsynchronous oscillation decomposition and extraction method based on empirical mode decomposition can perform decomposition completely according to the self time scale characteristic of current signal data; a decomposition process more comprehensively extracts and maintains the main characteristics of an original signal, and meanwhile improves the conciseness and flexibility of signal expression.

The invention provides a vertical domain based micro blog searching ranking method and system. The method comprises obtaining the domain relevancy of each micro blog, the fan number weight of each micro blog user, the concerned weight of each micro blog, the micro blog user type weight and the time weight of each micro blog in prior micro blogs; multiplying the obtained domain relevancies, fan number weights, concerned weights, micro blog user type weights and time weights by pre-assigned weight coefficients respectively, adding the results, and performing searching ranking on the prior micro blogs according to the added results. By means of the method and the system, views and opinions of users towards objects in certain field can be extracted accurately and effectively.

The invention discloses a method for decomposing and reconstructing ISAR (inverse synthetic aperture radar) objects on the basis of an attributed scattering center model. The method is used for extracting and estimating parameters of attributed scattering centers and reconstructing the objects. By the aid of the method, problems that an existing parameter extracting method for dividing image domains is not suitable for environments with low signal-to-noise ratios and objects provided with complicated structures and the signal integrity is destroyed. The scheme includes that the method comprises performing distance imaging, translation compensation, direction imaging and Fourier transformation on signals of the ISAR objects and converting the signals into distance domains and direction angle domains; sequentially estimating parameters of K attributed scattering centers in the spatial objects and iterating and dissolving the parameters of the attributed scattering centers of the spatial objects; superimposing the K attributed scattering centers to reconstruct the ISAR objects and estimate the sizes of the ISAR objects. The method has the advantages that the computational efficiency can be improved by the aid of a quick Fourier transformation algorithm for solving the problem of high computational complexity of an existing method, parameters of the scattering centers of the spatial objects can be accurately extracted and estimated, and the method can be applied to the field of object detection and recognition and ISAR imaging.

The invention discloses a system and a method for carrying out texture feature extraction and structured description on video images. According to the invention, under the control of a dual-core DSP (digital signal processor) chip, a video image acquisition module carries out real-time acquisition on an image, and after the image is processed through an A / D (analog to digital) sampling module, the image is subjected to segmentation so as to obtain a plurality of sub images; a two-dimensional Gabor wavelet module carries out filtering processing on the sub images, and then through carrying out extraction on the sub images by using a texture feature extraction module, the texture features of the sub images are obtained; a classification and discrimination module carries out features matching with a Brodatz standard texture library according to the extracted texture features; and finally, a structured description module carries out structured description on the texture features and the classification and discrimination results concerning the attributes and contents of the image. According to the method provided by the invention, texture features can be extracted from unstructured video image information and subjected to structured description, so that the querying, searching and browsing of the contents of video images can be realized, thereby facilitating the effective application and management of video monitoring.

The invention provides a corn lodging area extraction system and method based on a maximum likelihood method. The lodging area can be simply and quickly extracted. The system comprises an collection unit, a preprocessing unit, a calculation unit, a classification unit and an evaluation unit, wherein the collection unit, the preprocessing unit, the calculation unit, the classification unit and theevaluation unit are connected in sequence; the method comprises the steps that an unmanned aerial vehicle is used for carrying a multispectral sensor to collect field image data of a test area; preprocessing the acquired field image data to obtain a corn canopy multispectral image; calculating textural features and series vegetation index information of lodging corn and normal corn in the corn canopy multispectral image; selecting mean value characteristics of three red, green and near-infrared wave bands, combining with five proper classification characteristics of a ratio vegetation index and a triangular vegetation index, using a maximum likelihood method to supervise and classify the images, and extracting a corn lodging area; and verifying the precision of the classification model through field measurement data.

The invention relates to a schedule creation method and device based on a notification message, a terminal and a storage medium. The method comprises the following steps: acquiring a plurality of notification messages, wherein the notification messages comprise user identifiers; splitting the notification message to obtain a plurality of message words; matching the plurality of message words witha schedule lexicon of the user identifier to obtain schedule keywords corresponding to the plurality of message words; identifying a schedule attribute corresponding to the notification message according to the schedule keyword; when the schedule attribute of the notification message is a to-do schedule, identifying a field sequence corresponding to schedule element fields in the schedule keywords; and generating corresponding schedule information according to the schedule keyword and the corresponding field sequence. By adopting the method, the schedule elements of the notification message can be accurately identified, and the schedule information conforming to the habits of the user can be created, so that manual creation operation is saved, and schedule creation efficiency is effectively improved.

The invention relates to an alternating current electromagnetic positioning device and an alternating current electromagnetic positioning method, and belongs to an electromagnetic positioning device and an electromagnetic positioning method. The alternating current electromagnetic positioning device comprises a data processing and time sequence control unit, a D / A (digital to analog) converting circuit, a sending driving circuit, an overvoltage protection circuit, a three-axis emitting source, a receiving antenna, a multipath selecting circuit, a filter amplifying circuit and an A / D (analog to digital) converting circuit. According to the method, firstly, the data processing and time sequence control unit controls the D / A converting circuit to generate sinusoidal signals with different frequencies, the sinusoidal signals are sent to the three-axis emitting source after being subjected to amplitude and power amplification, the receiving antenna receives electromagnetic signals of space radiation, and the position posture information of a target is solved by a solving algorithm and is transmitted to an upper computer for display. The alternating current electromagnetic positioning device and the alternating current electromagnetic positioning method have the advantages that a hardware processing circuit is simplified, the problems of complicated circuit and low positioning speed caused by the existing time-division system work mode are avoided, the positioning timeliness is ensured, and in addition, the singular point problem of the positioning is solved.

The invention relates to the technical field of remote sensing image information processing and discloses a multi-season crop automatic recognition method based on a time sequential remote sensing image. The method is based on a time sequential remote sensing image, continuous wavelet transform is utilized, a wavelet coefficient spectrum of a time domain and frequency domain is obtained, a characteristic spectrum is generated from the wavelet coefficient spectrum, characteristics lines used for distinguishing vegetation relative preference and deviation period are extracted, and through tracking the value domain distribution of relative preference point variables in an indication growth period in different scaling intervals, the multi-season crop information is effectively extracted. According to the method, the vegetation dynamic change characteristic can be effectively extracted from dimensions of time and frequency, the vegetation change characteristic in low frequency in a relative preference period is extracted for automatic recognition of a multi-season crop, and the method is independent of prior knowledge basically and has the characteristics of good robustness, high classification accuracy, high degree of automation and strong anti-interference ability.

The invention discloses a target-oriented sentiment classification method, which belongs to the technical field of big data and comprises the following steps that: a client server and a central server are established, and the client server is used for collecting text information and sending the text information to the central server; a preprocessing module, a GloVe model module, a position information coding module, an attention encoder and a classifier module are established in a central server; a target sequence to be analyzed is extracted; aiming at the technical problem that target-oriented sentiment analysis is completed by aiming at the target sequence, the method is suitable for a scene needing fine-grained sentiment analysis, the text containing a plurality of targets to be analyzed and different targets with different sentiment colors in the statement can be subjected to sentiment classification, and extraction is more accurate and effective.

The invention provides a knowledge graph construction method and device and electronic equipment, and belongs to the technical field of artificial intelligence. The knowledge graph construction methodcomprises the steps of performing word segmentation and syntactic dependency relationship analysis on each sentence in a to-be-processed text to obtain a word segmentation result and a word sequencelibrary; screening out a frequent sequence of which the length is greater than a preset first threshold from the word sequence library; combining words included in the frequent sequence of which the lifting degree is greater than a preset second threshold value and the frequency is greater than a preset sixth threshold value into a newly added word, and updating the word segmentation result; and establishing a synonym combination according to the updated word segmentation result, updating a word sequence library according to the synonym combination, calculating variant confidence coefficientsamong words in the word sequence, and judging hyponymy concepts among the words according to a calculation result, the variant confidence coefficients representing correlation among the words in the word sequence or among the word sequences. According to the method, the concept and hyponymy relation can be accurately and effectively extracted from the non-defined domain text.

The invention relates to a far-infrared pedestrian detection method based on distinctive features. According to the invention, for an acquired far-infrared image, two distinctive features are found out, wherein two distinctive features comprise a highlight feature corresponding to a human head and a torso perpendicular feature corresponding to a human body; two regions are found out and fused to acquire an interested region; a gradient direction histogram feature is extracted from the interested region; a trained linear support vector machine is used to carry out pedestrian detection; at different temperatures, a pedestrian region can be accurately and efficiently extracted; fast and robust pedestrian detection is realized; and the driving safety factor of a driver at night, in a harsh environment and in the case of low visibility is improved.

The invention discloses a lane line extraction and recognition method based on machine vision. The method comprises the following stepos of collecting image video.; taking image calibration area segmentation and a weighted average method as a grayscale processing algorithm of an interested area, performing smoothing processing on an image by using median filtering, achieving lane line edge detection, image binarization processing and lane line framework extraction by using a canny operator, achieving lane line extraction and identification by using Hough transform based on priori knowledge, combining with the angle range of lane lines and the lateral distance of the lane lines as a screening criteria to eliminate a large number of interference lines during driving and finally achieving accurate extraction of the lane linesd. Accurate and effective extraction of lane lines can be realized, and the early warning accuracy of lane departure is further improved.

The invention discloses a method for on-line monitoring of a mechanical fault of an on-load tap-changer of a transformer. The method comprises the following steps of attaching three vibration sensorsto the top end of a box wall of the on-load tap-changer, and collecting a vibration signal generated in an action process of the on-load tap-changer; carrying out empirical mode decomposition (EMD) onthe collected signal, and computing an improved mask signal through a decomposed intrinsic mode function (IMF); adding the improved mask signal to the original collected vibration signal, and then carrying out empirical mode decomposition; and computing a power characteristic matrix of the intrinsic mode function according to an interval maximum power characteristic formula, comparing the power characteristic matrixes of the OLTC under different working conditions, and carrying out fault diagnosis on the OLTC. According to the method, a working state of the on-load tap-changer of the transformer can be monitored in real time, and the requirement for real-time fault diagnosis of the on-load tap-changer of the transformer can be met. A data support and a theoretical basis are provided for purposeful overhauling, and the waste of manpower, material resources and time is avoided.

The invention relates to a method for identifying a face of a living body. The method for identifying the face of the living body comprises the following steps of (1) collecting the face image information; (2) generating at least one group of face image sequences from the face image information; (3) extracting the human face local features from the at least one group of human face image sequences;(4) screening out a low-dimensional feature vector of the face image information from the face local features; (5) inputting the low-dimensional feature vector into a trained classifier to carry outliving body face image recognition; and (6) outputting an identification result. Correspondingly, the invention further relates to a system for identifying the face of the living body. According to the method and the system provided by the invention, the face of the living body can be quickly and accurately identified, and the high requirement of the embedded equipment for the real-time performance can be met.

The invention discloses an automatic extraction method for a key field in a network academic report. The method is characterized by comprising the steps that 1, academic report sites are collected to form an academic report site database, and the database is used as a crawling seed of a web crawler; 2, the web crawler is utilized to perform report crawling on each report site; 3, each academic report detail page obtained through crawling is subjected to content extraction, wherein extracted content comprises a report title, a report introduction, a reporter, a reporter introduction, report holding time, a report holding place and a report holding organization, and the content is encapsulated and structuralized; 4, the structuralized report content is subjected to data persistence operation; and 5, the steps are repeated till crawling of all the collected sites is completed. According to the method, by sorting out network academic report site information and processing HTML tags in the network academic report content, key information in the network academic report can be effectively extracted.

The invention discloses a cyanobacterial bloom remote sensing monitoring method based on a planktonic algae index and deep learning. The method comprises the steps: (1) extracting and recognizing cyanobacterial bloom in a research region through employing Landsat satellite data and a planktonic algae index (FAI); (2) utilizing a series of data enhancement methods such as rotation, translation, zooming and color transformation to operate cyanobacterial bloom data extracted by a floating algae index (FAI); (3) performing operations such as cutting, resampling and the like on the cyanobacterial bloom data after data enhancement to make a data set suitable for deep learning; (4) adopting a Deeplab deep learning model, and taking cyanobacterial bloom data as a training set for training; and (5)outputting a classification result taking land, a water area and cyanobacterial bloom as main categories. By adopting the method, the cyanobacterial bloom can be accurately and efficiently extractedand identified from the satellite image.

The invention provides a hypnotic state electroencephalogram signal extracting method and system. The method comprises the following steps of collecting a hypnotic state electroencephalogram initial signal, and performing preset-multiplication-frequency trap wave processing; selecting a wavelet function used for extracting a hypnotic state electroencephalogram signal according to the correlation between the wavelet function and the preprocessed electroencephalogram signal; selecting the center frequency-band width ratio of the wavelet function in an adaptive optimization way; and extracting the hypnotic state electroencephalogram signal from the preprocessed electroencephalogram signal through the optimized wavelet function. In the process, the center frequency-band width ratio is optimized in an adaptive way; the self-adapting regulation of the wavelet coefficient is realized; the resolution ratio of the wavelet transform is improved; and the hypnotic state electroencephalogram signal can be effectively and accurately extracted.

The invention relates to the technical field of remote-sensing image information processing, in particular to a farmland cropping index automatic extracting method. The method based on remote-sensing time sequence data includes: continuous wavelet transform is performed on time sequence signals within a year pixel by pixel to obtain a wavelet coefficient spectrum, a series of wavelet coefficient contour lines are extracted from the wavelet coefficient spectrum, a cropping system judging standard is built according to the number of the contour lines of the closed wavelet coefficient which is detected at the same time and the condition whether the intersection of the dimension intervals covered by the contour lines of the closed wavelet coefficient exists or not, and automatic extracting of farmland cropping indexes is achieved. The method has the advantages that the mapping relations of the closed contour lines and crop growing peaks are fully utilized, interference is further removed by setting conditions such as contour line search dimension intervals and initial values, and the method is good in robust, high in categorization precision, high in automation degree and interference resistance, and the like.

The invention belongs to the technical field of microwave testing, and discloses a clamp de-embedding method and system, a storage medium, a computer program and application, and the method comprisesthe steps: employing a clamp to clamp two different transmission lines, and respectively measuring S parameters [St], i.e. [St1] and [St2], of the whole system; calculating S parameters [St01] and [St02] of the two different sections of transmission lines clamped by the clamp according to a formula; setting an appropriate initial value [S], and approximating the S parameter [S] of the clamp. The comparison between the clamp simulation data and the calculation data finds that the clamp S parameter data at the two ends of the tested piece can be accurately and effectively extracted by adopting the clamp de-embedding method of the double transmission lines, and a good foundation is laid for subsequent semiconductor parameter testing. Compared with a traditional calibration method, the methodhas the advantages of being easy to implement and machine and low in requirement for the machining precision of standard parts, and has good application value.

The embodiment of the invention discloses a method and device for determining information of a target object. One specific embodiment of the method for determining the information of the target objectcomprises the following steps: acquiring an image of an identified rectangular frame of the target object; projecting laser point cloud data points into the image based on the external parameters ofa camera and the external parameters of a laser radar; screening point cloud data points projected into the rectangular frame of the target object in the image; determining a filtering window based onthe rectangular frame of the target object; based on the point cloud information in the filtering window, filtering out interference data points with the distance to the laser radar being greater than a preset threshold from the point cloud data points in the rectangular frame of the target object to obtain point cloud on the target object in the rectangular frame of the target object; and determining information of the target object based on the point cloud on the target object and the target object. According to the embodiment, the two-dimensional information of the target object on the image and the three-dimensional information of the target object on the point cloud can be fused, and the extracted information of the target object is accurate and effective.