243 results about "Eigenvalue computation" patented technology

The invention relates to an optical film detection method, device and equipment based on image processing. The method comprises the following steps: collecting multiple frames of images of a multilayer optical film under different illumination conditions, and carrying out preprocessing and characteristic value calculation on the multiple frames of images to obtain a characteristic value response matrix; calculating an optimal weight coefficient of each frame of image according to the feature value response matrix, and performing multi-frame image feature fusion to obtain integrated feature representation; performing multi-scale defect feature extraction on the integrated feature representation to obtain a defect candidate region; spectral response characteristics are extracted based on the defect candidate area, matching degree calculation is carried out on the spectral response characteristics and a preset wavelength-film thickness mapping relation matrix, and a multilayer film defect distribution diagram is obtained; and performing autoregression prediction and processing control parameter optimization on the multilayer film defect distribution diagram to obtain a closed-loop control scheme. According to the invention, accurate identification and classification of different film layer defects are realized, and closed-loop control of defect detection results and manufacturing parameter optimization is further realized.

The invention relates to the technical field of energy storage battery diagnosis, and discloses an energy storage battery fault diagnosis method and system based on a data fusion algorithm. The method comprises the following steps: acquiring historical operation data and real-time operation data of an energy storage battery in a preset operation period, and generating a historical fault data set according to the historical operation data; performing multi-source feature fusion processing on the historical fault data set to generate a fusion feature parameter set integrating voltage, current, temperature and impedance parameter joint change features; a multi-dimensional fault space is constructed based on the parameter set, coordinate axes of the multi-dimensional fault space correspond to different parameter dimensions, and spatial position coordinates represent parameter change characteristic values; calculating the fault correlation degree of the historical fault event in the multi-dimensional fault space, and determining a fault early warning index set; and extracting real-time characteristic parameters based on the real-time operation data to form a state vector, carrying out space mapping correlation calculation on the state vector and the fault early warning index set in a multi-dimensional fault space, and outputting a real-time fault correlation factor.

The present invention discloses a method for diagnosing micro-short circuit fault of a battery, comprising: acquiring real-time voltage data of each battery in battery module of electric vehicle, obtaining voltage matrix after noise elimination; calculating dynamic reference voltage sequence of the each battery in the voltage matrix within sliding window length, extracting eigenvalues of the each battery at different times; calculating correlation coefficients between voltages of the each battery at different times and reference voltage within the sliding window; forming feature point matrix of the battery module based on the eigenvalues and the correlation coefficients; calculating dynamic reference feature point sequence based on the feature point matrix, and calculating improved Frechet distances between feature point sequences of the each battery and the dynamic reference feature point sequence, using Frechet distance values as anomaly scores of the each battery; thus, judging whether a short-circuit fault occurs in the each battery.

An effective prestress estimation and evaluation characteristic value calculation method for a concrete structure, wherein the method comprises: S1, calculating effective prestress probability distribution of prestressed tendons of the concrete structure; S2, establishing Gaussian mixture models of effective prestress of the concrete structure; S3, performing normalization processing and normal significance determination on the Gaussian mixture models; S4, performing sampling estimation on structure's effective prestress probability distribution with theoretical distribution being normal distribution; S5, calculating an effective prestress evaluation characteristic value of a component under the normal distribution; S6, performing grouping processing and normal significance determination on the Gaussian mixture models; S7, performing sampling estimation on structure's effective prestress probability distribution with theoretical distribution being N sub-distribution Gaussian mixture models; and S8, calculating the effective prestress evaluation characteristic value of the component under the N sub-distribution Gaussian mixture models. According to the method, the evaluation characteristic value of effective prestress can be scientifically determined based on the actually measured sampling data of the effective prestress.

The invention relates to a BIM-based deep foundation pit construction dynamic monitoring method and system, and specifically, the method comprises the steps: dividing a BIM design model into a plurality of stages according to a construction stage, and obtaining deep foundation pit site point cloud data and a current construction stage; constructing a point cloud picture based on the affiliation consistency probability and boundary conformity of the points in the point cloud in the BIM design model at the current construction stage, and segmenting the point cloud picture by adopting minimum segmentation with hierarchical constraint to obtain a plurality of point cloud clusters; calculating the internal geometrical morphology of the point cloud cluster based on the characteristic value of the covariance matrix of the point cloud cluster, and associating the point cloud cluster with the construction member in the BIM design model at the current construction stage by using the position relationship between the geometrical morphology and the point cloud cluster; and performing three-dimensional reconstruction on the point cloud cluster by using construction components associated with the point cloud cluster, and updating and visually displaying the current construction progress state of each component in the BIM design model.

The invention discloses a signal acquisition and processing method and system based on a multifunctional radar, and relates to the technical field of signal acquisition and processing, and the method comprises the steps: employing a quantum genetic algorithm to optimize radar transmission waveform parameters, including center frequency, bandwidth and frequency modulation slope, and generating a nonlinear frequency modulation waveform through FPGA hardware; a non-uniform sparse array is adopted to receive a target echo signal, time domain random interval sampling and frequency domain pseudo-random frequency point selection are synchronously implemented, and time-space-frequency three-dimensional compressed sensing observation data are formed; performing trilinear tensor joint sparse reconstruction on time-space-frequency three-dimensional compressed sensing observation data, decomposing a polarization scattering matrix eigenvalue from a reconstructed signal, and calculating a coupling characteristic quantity of eigenvalue entropy and micro-Doppler frequency; inputting the coupling characteristic quantity into a deep reinforcement learning model, and dynamically outputting a constant false alarm detection threshold, a moving target display filter order and a resource allocation weight; and detecting a threshold based on a constant false alarm rate.

The invention discloses a characteristic value calculation method based on electrocardio data, belongs to the field of medical data analysis, and provides an electrocardio data analysis and processing method. Comprising the following steps: decomposing different frequency components of an electrocardiosignal by using wavelet transform, and analyzing on different time scales. Wavelet transformation is relatively suitable for detecting QRS wave groups, and due to the fact that the wavelet transformation can highlight transient characteristics of signals on different frequency bands, complex electrocardiosignal processing is more efficient. In addition, wavelet transform is not only suitable for wave crest detection, but also can help to remove signal noise and improve signal definition. The deep learning technology is used for automatically extracting the features from the electrocardiosignals, a feature extractor does not need to be manually designed, related medical professional knowledge and the artificial intelligence technology are combined, wave crests in the electrocardiosignals can be more accurately detected, and meanwhile anomaly detection can be carried out, such as early signs of heart diseases such as sinus arrhythmia and the like.

The invention relates to a data processing method and device, computer equipment, a computer readable storage medium and a computer program product. After first matrix data is obtained, tensor data to be processed are obtained, and the tensor data are distributed to a plurality of calculation units; calculating a local feature value corresponding to the target calculation unit according to the local tensor data allocated to the target calculation unit; dividing the target calculation unit into data blocks corresponding to the number of the plurality of calculation units; according to the serial number of the data block, mapping the local feature value recorded in each data block to a target storage block of a target memory until the local feature value in each calculation unit is distributed to the corresponding storage block in the target memory, and obtaining a global feature value table; according to the global feature value table, feature value calculation is carried out to determine the target feature value, conflict-free parallel calculation is achieved, the utilization rate of calculation resources is increased, and the data processing efficiency is improved.

The invention provides a pressure sore detection method and system based on a pressure sensor. The method comprises the following steps: acquiring pressure data of a to-be-detected part of a user through the pressure sensor; preprocessing the collected pressure data to form a pressure matrix; extracting at least one basic pressure characteristic value based on the pressure matrix; calculating characteristic values of any basic pressure characteristic value in a time domain, a frequency domain and a time-frequency domain, and constructing a multi-level pressure characteristic matrix; and inputting the multi-level pressure feature matrix into a pressure sore state detection model to obtain the pressure sore state of the user, through the method, pressure data acquisition can be performed on the body part of the user in a real-time and non-inductive manner, pressure sore detection is performed, autonomous monitoring of the body part of the user is realized by using an artificial intelligence algorithm, and the user experience is improved. The detection process is flexible and efficient, and the privacy of the user is effectively protected.

The invention discloses a distribution network fault diagnosis method based on harmonic spectrum analysis and a related device, and belongs to the technical field of distribution network fault diagnosis. According to the method, when a power distribution network fault occurs, an abnormal signal is collected and preprocessed, and a harmonic spectrum is obtained; performing feature extraction on the harmonic spectrum to obtain a plurality of feature values; calculating a distribution network fault comprehensive research and judgment index through the plurality of characteristic values; and inputting the distribution network fault comprehensive research and judgment index into a pre-trained SVM classification model to obtain a fault diagnosis classification result. According to the method, a multi-dimensional spectrum analysis method is introduced, and mutual influence of various fault factors is comprehensively considered by extracting multi-level harmonic characteristics of fault signals. In combination with advanced mode recognition and machine learning technologies, the fault type classification can be efficiently and accurately carried out, the problems of misjudgment and blind spots caused by single feature extraction in a traditional method are avoided, and the fault diagnosis precision is improved.

The present invention relates to the field of image analysis technology, specifically a tunnel vehicle-mounted leakage detection method based on infrared thermal imaging technology, comprising: acquiring an infrared image frame sequence and performing preprocessing and feature enhancement processing; extracting gradient eigenvalues ​​and outlier eigenvalues ​​from the enhanced infrared image to identify initial leakage areas; tracking each initial leakage area in a continuous enhanced infrared image frame sequence to obtain a regional spatial sequence, and extracting the spatial gradient eigenvalues ​​of the initial leakage area; calculating a multi-feature comprehensive leakage index based on the gradient eigenvalues, outlier eigenvalues, and spatial gradient eigenvalues, processing the initial leakage area to obtain a leakage area; performing feature analysis on each leakage area to obtain regional multidimensional features, and performing a severity assessment based on the regional multidimensional features. The present invention improves the reliability of leakage analysis and assessment.

The invention belongs to the technical field of phase modifier voiceprint construction, and particularly relates to a voiceprint library construction method and device for phase modifier rotating equipment. Comprising the following steps: acquiring operation sound data of equipment in an initial state, extracting initial voiceprint feature information, acquiring equipment sound data and extracting monitored voiceprint feature information according to a certain monitoring period in an equipment operation process, acquiring voiceprint feature change information through comparison, judging whether the voiceprint feature change information is abnormal or not, and if so, judging whether the voiceprint feature change information is abnormal or not. And if not, extracting different voiceprint feature information, constructing an analysis time period, combining with equipment operation state data, constructing a multi-modal feature tag set by using the different voiceprint feature information, combining the multi-modal feature tag set with the initial voiceprint feature information, generating a feature set tag set, performing dynamic feature value calculation, and finally constructing a voiceprint library. According to the method, the constructed voiceprint library is not only based on the sound features, but also associated with the multi-dimensional operation state of the equipment, so that the applicability and accuracy of the voiceprint library are improved.

The invention relates to the technical field of monocrystalline silicon manufacturing, in particular to a silicon dopant volatilization compensation evaluation method and system based on process analysis, and the method comprises the steps: obtaining a doping target instruction in a monocrystalline silicon drawing process, analyzing the doping target instruction, and obtaining a dopant target concentration value and a dopant target uniform characteristic value of each drawing stage; aiming at each drawing stage, setting a corresponding monitoring frequency, and monitoring drawing process parameters according to the monitoring frequency to obtain a real-time process parameter set; performing doping volatilization analysis on the real-time process parameter set to obtain an actual concentration value and an actual uniform characteristic value of the dopant under the process condition; based on the dopant target concentration value, the dopant actual concentration value, the dopant target uniform characteristic value and the dopant actual uniform characteristic value, calculating concentration deviation and uniform characteristic deviation; the device can be dynamically adjusted along with the propulsion of the drawing process and the change of technological parameters, so that the high-efficiency compensation on the volatilization of the dopant is realized.

The embodiment of the invention discloses a point cloud filtering method and device, electronic equipment and a storage medium, and relates to the technical field of computer vision, and the method comprises the steps: obtaining point cloud data, dividing a point cloud into a plurality of voxels through employing an octree structure according to the coordinate range of the point cloud and a preset voxel side length, calculating the mass center of each point in each voxel, and calculating the mass center of each point in each voxel; circumscribed balls are obtained through calculation with the centroid as the center of sphere and the diagonal length of voxels as the diameter, points except the center of sphere in the circumscribed balls are filtered out, coordinates of the center of sphere of the circumscribed balls serve as feature values, the distance between the feature values is calculated, part of voxels are reserved according to the sampling rate and the distance, and the coordinates and the feature values of the reserved voxels are added to point cloud data. And deleting the other part until the target sampling rate is reached, and obtaining the filtered point cloud data. The method solves the problems that in the prior art, the inherent geometric structure of the point cloud cannot be kept unaffected, the calculation amount is large, the processing efficiency is low, and the requirements for real-time performance and practicability are difficult to meet.

The invention discloses a deep foundation bearing capacity characteristic value calculation method and system based on data fusion. The method comprises the steps that data collection is completed through a deep load test, and a load settlement curve and a settlement time logarithmic curve are formed; generating a smooth curve; identifying a slope mutation point of the curve by adopting a first-order derivative mutation method, wherein a load corresponding to the mutation point is a proportional limit load; if the slope abrupt change point is not recognized, the curve is set as a slowly changing curve; extracting the slope and the slope change rate of the curve under each level of load, and completing the judgment and correction of the limit load by fusing the multi-dimensional data of the two curves and combining preset thresholds including the settlement amount and the settlement stabilization time; and further calculating the bearing capacity characteristic value of the deep foundation by combining the numerical relationship between the ultimate load and the proportional limit load.

The invention provides a method and a system for selecting an input signal of a damping controller of a synchronous phase modifier. The method comprises the following steps: establishing a differential equation and an algebraic equation of the whole system based on a dynamic model of a grid-connected system; linearizing a differential equation and an algebraic square of the whole system at a stable operation point, and establishing a small-signal linearization model of the whole system; solving a system state matrix based on the small signal linearization model, calculating a characteristic value, a left characteristic vector and a right characteristic vector of the system state matrix, and determining different oscillation modes of the system according to a characteristic value calculation result of the system state matrix; calculating participation factors in different oscillation modes by combining the left feature vector and the right feature vector of the system, and constructing a participation factor matrix; and in combination with the participation factor matrix, aiming at a certain oscillation mode of the system, the size of the participation factors between the oscillation mode and different state variables is compared and analyzed, and the state variable with the maximum participation factor is used as an input signal of the damping controller of the synchronous phase modifier.

The invention discloses a storage and identification method and system based on multi-dimensional personnel identification data, and the method comprises the steps: obtaining the effective feature data of a to-be-identified personnel, determining the feature value, the initial weight value and the interference quantized value corresponding to each dimension feature vector in the effective feature data, calculating a target weight value corresponding to each dimension feature vector based on the initial weight value of each dimension feature vector and the corresponding interference quantized value, calculating a fusion value corresponding to the effective feature data based on the target weight value and the feature value of each dimension feature vector, and performing feature vector mapping on the fusion value to obtain the effective feature data. And obtaining and storing a target feature description vector corresponding to each dimension feature vector, inputting the target feature description vector into a pre-training recognition model, and outputting a recognition result of the to-be-recognized person. Through the above method, the dimension feature vectors of different dimensions are fused, the accuracy of the target feature description vector is ensured, and the accuracy of the recognition result is improved.

The invention discloses a target knowledge data recognition and extraction method optimized by a depth algorithm, relates to the technical field of knowledge extraction, and solves the problem of overlong subsequent content output time caused by relatively slow target data comparison process. Text semantics are converted into quantifiable geometric features, compared with traditional keyword matching, implicit association (such as synonyms and context dependence) between semantics can be captured, recognition features are mapped to a two-dimensional coordinate system, a target object is rapidly matched through the Euclidean distance, the efficiency is improved by 10-20 times compared with full-amount traversal query, clustering analysis based on historical extracted data is achieved, and the query efficiency is improved. Through calculation of a mean value time column and a feature value TT, the optimal computing power configuration is automatically identified, and for time feature sorting of different extraction objects, tasks which are short in time consumption are preferentially processed, so that the queue is prevented from being blocked by long tasks.

The invention relates to the technical field of data processing, in particular to a digital archive information data processing method, which comprises the following steps of: 1, splitting archive information into a plurality of single-page archives, calculating regional characteristic values by combining capacity values and angle characteristic values of each content region, and arranging the regional characteristic values in sequence to form a content characteristic sequence; 2, constructing a text feature sequence in combination with the three feature values of the content area; 3, splitting the text content into a plurality of unit areas, and combining the two feature sequences of each unit area to obtain a unit feature sequence; and 4, when the archive information is sent, obtaining the initial binding value of the archive information, when the archive information is received, initially and sequentially checking the initial binding value, and determining an abnormal single page based on a checking result and performing unit checking. The method has confidentiality and safety, can identify and record tampering and modification in the archive transmission process, and is beneficial to improving the transmission efficiency in the archive information data processing process.

For maintaining precision in processing of generating information relating to an object by using point cloud data, an image processing apparatus includes a data acquisition unit, a feature value computation unit, and an area correction unit. The data acquisition unit acquires point cloud data generated by scanning a space to be measured. The feature value computation unit sets a plurality of partial areas with a first boundary of a processing target area in the point cloud data as a reference, and computes a feature value for each of the plurality of partial areas. The area correction unit corrects the processing target area by setting a new second boundary, based on a position of a partial area satisfying a reference relating to the feature value.

The invention belongs to the technical field of data acquisition, and particularly relates to a signal sampling method and system for dynamic frequency optimization. The method comprises the following steps: presetting a basic sampling frequency based on target signal characteristics, sampling an analog signal at a frequency f1 through an analog-to-digital converter (ADC) to generate a first digital signal sequence, executing a feature extraction algorithm on the first digital signal sequence in an MCU to obtain an actual signal feature value, and calculating a difference value between the actual feature value and a preset target feature value, according to the method, signal features are extracted in real time, the sampling frequency is adjusted in a closed-loop mode through PI D, power consumption and resource occupation are reduced, meanwhile, it is ensured that transient events are completely captured, the power consumption is reduced in a self-adaptive mode along with signal complexity, the capture rate of the transient events is high, and the method is suitable for large-scale popularization and application. Missing acquisition of a traditional fixed low-frequency scheme is avoided; the method is suitable for wearable, industrial monitoring, automotive electronics and other scenes.

A water meter bearing wear state monitoring method and system based on a flow rate working condition relates to the field of electric digital data processing, and the method comprises the following steps: monitoring the generation frequency of a metering pulse on a target water meter, and determining whether the current metering working condition of the target water meter is a high flow rate working condition or a low flow rate working condition based on the generation frequency; counting a first pulse time interval sequence under the low flow speed working condition, and calculating a dispersion characteristic value of the first pulse time interval sequence as a first state characteristic value; counting a second pulse time interval sequence under the high flow rate working condition, and calculating a dispersion characteristic value of the second pulse time interval sequence as a second state characteristic value; based on the first state characteristic value and the second state characteristic value, calculating a working condition response index representing the wear degradation effect of the water meter bearing on the target water meter; when the working condition response index is larger than a preset abrasion judgment threshold value, abrasion early warning information corresponding to the target water meter is generated. By implementing the application, the water meter maintenance cost can be reduced.

The invention relates to the technical field of trajectory tracking, in particular to a campus personnel trajectory tracking system based on big data, which comprises a multi-channel video acquisition module, a target personnel identification module, a preliminary trajectory construction module, a blind area trajectory prediction module and a display terminal, according to the invention, the monitoring video stream of the campus key area is called in real time and analyzed, the image sequence of the target person is constructed, the image sequence is analyzed, and if the face image cannot be extracted, the identity similarity value is calculated based on the clothing color feature value and the body feature value to determine the target person. The method comprises the following steps: constructing a preliminary track according to spatio-temporal information of a target person at different monitoring terminals, identifying passing paths corresponding to missing segments of the preliminary track, if the passing paths are not unique, calculating a comprehensive confidence evaluation value of each passing path, selecting the passing path with the maximum value as a complemented path, connecting front and back nodes to generate a complete track, and completing the complemented path. Therefore, high-precision identification and continuous trajectory tracking of non-school personnel are realized.

A compute instance stores a programmable feature extractor associated with a machine learning model maintained by a server-based computing system configured to communicate with the compute instance by way of a network. The machine learning model is based on a feature set that includes a plurality of features. The compute instance executes the programmable feature extractor to generate a feature vector corresponding to a data instance accessed by the compute instance, where the feature vector includes a feature value specific to the data instance for each feature included in the feature set. The compute instance transmits the feature vector corresponding to the data instance to the server-based computing system for use as a training input to the machine learning model.

The invention discloses an improved principal component analysis efficiency evaluation method. The method comprises the following steps: (1) carrying out standardized preprocessing on data; (2) calculating an index correlation matrix of all samples, and improving the nonlinear performance of the algorithm through a kernel function; (3) solving a characteristic value of the correlation matrix; (4) calculating the contribution rate of each component; (5) calculating a coefficient corresponding to the original index; (6) normalizing the coefficient to obtain an index weight; and (7) calculating linear summation to obtain an evaluation value. The method can realize data dimension reduction, is suitable for multi-scheme data evaluation, can analyze and calculate the evaluation index weight, does not need human intervention, and has strong objectivity.

The present application provides an image segmentation method, apparatus, electronic device and storage medium, comprising: inputting an image to be segmented into the first two first convolutional layers of a feature extraction network and performing convolution processing in sequence to obtain an original feature map output by the second convolutional layer of the feature extraction network; performing noise separation on the original feature map to obtain an initial noise map, and reconstructing the initial noise map to obtain a noise distribution map; calculating the feature value of each optimized pixel point based on the feature value of each pixel point in the noise distribution map and the original feature map; generating an optimized feature image based on the feature value of each optimized pixel point; performing portrait segmentation on the optimized feature image to obtain a portrait area and a background area, with the aim of improving the accuracy of image segmentation through a deep convolutional neural network.

The present invention relates to the technical field of traffic analysis, and discloses an online traffic data analysis method and system, including: collecting traffic set data of each zone of the platform within a detection period, and obtaining a first traffic characteristic value of each zone; collecting historical traffic set data of each zone in the detection period and historical target conversion data of each zone, and calculating the historical target conversion value of each zone; obtaining the second traffic characteristic value of each zone, calculating the comprehensive traffic value of the platform, and obtaining the comprehensive conversion results analysis result; comparing with the traffic threshold, and combining the traffic risk characteristics of each zone, generating a traffic anomaly report. The present invention can accurately evaluate the traffic value and conversion efficiency, comprehensively consider the traffic scale, historical conversion capacity and real-time user behavior, accurately evaluate the actual value of each partition traffic and the overall conversion efficiency of the platform, can effectively identify traffic anomalies and risks, can timely discover abnormal traffic fluctuations, and accurately locate the source of risks.

The purpose of the present invention is to provide a technique for diagnosing, early, efficiently, and with minimal false alarms, an abnormality caused by aging deterioration of a device in a semiconductor manufacturing apparatus, on the basis of a change in a feature quantity trend obtained from a signal waveform acquired by a sensor mounted on the device.　To this end, a diagnostic apparatus for a semiconductor manufacturing apparatus according to the present invention comprises: a feature quantity computation unit that computes feature quantity from sensor waveform data corresponding to a process ID associated with a sensor item indicating a state of a device to produce a feature quantity trend in a time series; a trend type classification unit that classifies the feature quantity trend by trend type; a trend change amount computation unit that classifies a change in the feature quantity trend by trend change type, applies a trend change amount computation method corresponding to the trend change type on the basis of the trend type, and calculates a trend change amount; and a trend change abnormality diagnosis unit that determines an abnormality of the feature quantity trend in diagnosis from the feature quantity trend in diagnosis and the feature quantity trend in normal condition.

The present invention relates to a dynamic monitoring method and system for deep foundation pit construction based on BIM. Specifically, according to the construction stages, the BIM design model is divided into multiple stages. The on-site point cloud data of the deep foundation pit and the current construction stage are obtained. Based on the attribution consistency probability and boundary compliance degree of the points in the point cloud in the BIM design model of the current construction stage, a point cloud map is constructed. The point cloud map is segmented using minimum segmentation with hierarchical constraints to obtain multiple point cloud clusters. The intrinsic geometric shape of the point cloud clusters is calculated based on the eigenvalues of the covariance matrix of the point cloud clusters. Using the geometric shape and the positional relationship of the point cloud clusters, the point cloud clusters are associated with the construction components in the BIM design model of the current construction stage. The point cloud clusters are three-dimensionally reconstructed using the construction components associated with the point cloud clusters, and the current construction progress status of each component is updated and visually displayed in the BIM design model.