826 results about "Component analysis" patented technology

The present invention is a method for detecting an abnormal event for process units of a hydrocracking unit. The method compares the operation of the process units to a model developed by principle components analysis of normal operation for these units. If the difference between the operation of a process unit and the normal operation indicates an abnormal condition, then the cause of the abnormal condition is determined and corrected.

The invention relates to a photochemical colorimetric sensor array, and concretely relates to a method for detecting various heavy metal ions with the photochemical colorimetric sensor array. The method comprises the following steps: 1, fixing several heavy metal ion indicators on a porous sensitive film to form an array; 2, carrying out multichannel analysis by means of a plurality of single channel flow cell systems which are mutually independent with each other; and 3, rapidly determining whether contents of various trace heavy metal ions of Hg, Pb, Cd, Ag, As, Ni, Cu, Zn and the like in water exceed sewage discharge standards prescribed by national standards within 10min. According to the sensor array, a color imaging device acquires changes of colors before and after the exposure of the indicators in various different heavy metal ions and constructs RGB (red, green, blue) spectrum values corresponding with the color changes into the "fingerprint" of each heavy metal ions. In the sample test process, mathematical statistic methods of cluster analysis, main component analysis and the like are adopted to contrast the fingerprint of the sample with a "fingerprint" database, so unknown heavy metal ions can be qualitatively or semi-quantitatively analyzed.

The invention discloses a near infrared spectrum analyzing method based on the isolated component analysis and genetic neural network, which comprises the following steps for the acquired near infrared spectrum: firstly, effectively compressing spectrum data by using wavelet transform; secondly, extracting an independent component and a corresponding mixed coefficient matrix of a near infrared spectrum data matrix by using an isolated component analysis method; thirdly, building a three-layer BP neutral network, using the mixed coefficient matrix of a training sample as the input and correspondingly measured component concentration matrix as the output, and optimizing a neutral network structure by adopting a genetic algorithm, and obtaining a GA-BP neutral network by the training of the training sample; fourthly, predicting and analyzing the measured component concentration of the predicted set sample by using the GA-BA neutral network. The method enriches the chemical measurement method, widens the application range of the isolated component analysis and has favorable application prospect.

A tilled land quality evaluation method based on GIS for province-wide land comparison involves province-level parameter macrocontrol and united standard index system, county-level investigated basic data, united modeling and computation of city-level and four-hierarchy quality inspection mode. The concrete step are as follows: dividing index control area by GIS space superposition technology, hierarchical clustering method, determining evaluation factor by main component analysis method, determining weight by fuzzy hierarchy method, standardizing the factor index by segmental interpolation functions, computing index as natural quality etc. by modular design, constructing combined analog system for tilled land quality factor, dividing tilled land quality level by frequency method and uniformly-spaced method. The invention provides scientific basis for differential precise management for tilled land of province-area even whole country, and has significance to promoting reasonable utilization and protection for tilled land resource, ensuring grain safety of country.

The invention discloses a BP neural network construction and identification method for identifying different brown rice grains, adopting the following technical scheme: including the following steps: 1) acquiring images: 2) image preprocessing: 3) extracting image feature information of different brown rice grains: 4) main Component analysis reduces dimensionality of image feature information; 5) Design BP neural network structure: 6) Train neural network, and use the BP network neural network constructed by any of the above-mentioned construction methods to identify different brown rice grains. The method proposed by the present invention obtains images of different types of brown rice grains through black as the background, uses image processing technology to obtain its characteristic information, and uses principal component analysis to reduce the dimensionality of the characteristic information, and finally uses BP neural network to perform different types of brown rice grains. identify. The method can identify different types of grains objectively, accurately and quickly, and overcomes the disadvantages of traditional manual detection.

The invention discloses a flood forecasting method based on cluster analysis and real time correction, which comprises the following steps: 1) using PCA(Principal Component Analysis) to perform dimensionality reduction to the input of a model; 2) using the K-means clustering method to conduct clustering analysis on original data; dividing the flood data into different classifications; and then training different SVM models; when a testing sample is inputted, using the clustering center to determine the classification of the test sample and predicting the corresponding model to obtain a predicted value q; and 3) using a BP neural network for real time correction; calculating the error sequence between the predicated value and the actual value; using the error sequence data to train the BP neural network error correction model to obtain the error correction value qe. The final forecasting result is the model predicted value q plus the error correction value qe. According to the invention, the original hydrological data are divided into several classifications by cluster analysis, and through the training of the models, forecasting can be available by the multiple models. Then, real-time correction is achieved by the BP neural network to improve the forecasting accuracy for the time of flood peak.

The invention provides an industrial park atmospheric pollutant diffusion simulating and tracing method. The industrial park atmospheric pollutant diffusion simulating and tracing method specificallycomprises the following steps: (1) based on multi-point pollutant component analysis and concentration real-time monitoring data, analyzing atmospheric pollutant emission characteristics of an emission source, and calculating dynamic emission source intensity; (2) constructing an atmospheric pollutant diffusion model based on a Gaussian smoke plume diffusion theory and a smoke mass diffusion theory, calculating pollutant diffusion distribution characteristics of an emission source according to real-time meteorological data, and predicting the pollutant concentration of sensitive points; and (3) simulating and calculating the influence of the low-rise building within the factory boundary distance of the industrial park on airflow movement, determining a transmission path of emission sourcepollutants, and performing tracing analysis on sensitive point pollutants. The theories of Gaussian diffusion, smoke mass diffusion and computational fluid mechanics are integrated, the problem that the industrial park atmospheric pollutant emission traceability is difficult is solved, and the effective supervision of the industrial park ambient air quality is realized.

The invention discloses a noise diagnosis algorithm for rolling bearing faults of rotary equipment. Firstly, a sound pick-up device collects running noise signals of a rolling bearing, and the signalsare subjected to preliminary fault judgment through a bearing normality and anomaly pre-classification model based on an anomaly detection algorithm; secondly, according to a fault pre-judgment result, the abnormal signals (the faults occur) pass through a neural network filter to filter normal components in the signals of the bearing, the output net abnormal signals are connected to a subsequentfeature extraction module, and the normal signals (no faults occur) are directly connected to the feature extraction module; the feature extraction module extracts Mel-cepstrum coefficients (MFCC) ofthe signals to serve as eigenvectors, feature reconstruction is carried out by utilizing a gradient boosted decision tree (GBDT) to form composite eigenvectors, and principal component analysis (PCA)is used for carrying out dimensionality reduction on features; and finally, feature signals are input into an improved two-stage support vector machine (SVM) ensemble classifier for training and testing, and at last, high-accuracy fault type diagnosis is achieved. According to the algorithm, the bearing faults can be effectively detected and relatively high fault identification accuracy is kept;and the algorithm has relatively high effectiveness and robustness for detection and classification of the bearing faults.

Change in a torque waveform is monitored while removing continuously-varied periodic noise in real time, and the change in the torque waveform caused purely by the wafer state is detected by separating noise components while removing the noise not caused by the wafer state such as drift noise cased by dressing conditions and the polishing pad state, thereby reliably detecting a polishing end point at high precision in the wafer state when polishing is finished. A polishing end point detection device utilizing torque change having a torque measurement means for measuring torque, a periodic component analysis means for analyzing periodic components in data by subjecting Fourier transformation to the measured data, an average processing time calculation means for calculating moving average processing time for removing periodic noise components based on the analyzed periodic components, an average processing means for correcting the waveform by performing averaging process based on the moving average processing time calculated in real time for the data, and an end point detection means for detecting the polishing end point of a predetermined film based on change in the corrected torque waveform is provided.

A method of isolating sources of variance in parametric data includes steps of: (a) cleaning a data set of measurements for a plurality of parameters; (b) generating a principal component analysis basis from the cleaned data set; (c) estimating an independent component analysis model from the principal component analysis basis; (d) calculating percentages of variance for the plurality of parameters explained by each component in the estimated independent component analysis model; (e) if the calculated percentages of variance indicate that a component is a minor component, then transferring control to step (f), else transferring control to step (g); (f) removing the minor component from the principal component analysis basis and transferring control to step (c); and (g) generating as output the estimated independent component analysis model wherein no component of the independent component analysis model is a minor component.

The invention provides a method for conducting statistics on elevator visitor flow based on intelligent visual perception. The method comprises the steps of S1. establishing a head-and-shoulder model sample database; S2. conducting feature extraction and model training, wherein samples in S1 are subjected to principle component analysis (PCA) feature extraction, and a support vector machine (SVM) trainer is used for training model generating; S3. detecting targets, wherein matching calculation is conducted on images collected in real time according to human body head-and-shoulder model data obtained in S2, and targets in current images are obtained through detection; S4. tracking targets, wherein targets detected in S3 are tracked; and S5. conducing statistics on the visitor flow, wherein corresponding counters are operated according to incoming and outgoing conditions of targets when targets tracked in S4 go cross a crossing line. The method has the advantages that real-time intelligent analyses are conducted on images collected in real time, so that elevator visitor flow data can be obtained, accurate evidences are provided for establishing of effective and energy-saving elevator dispatching strategies, the problem that prediction models of the elevator visitor flow in traditional methods are complex and difficult to establish is solved, and the imprecise prediction caused by special events are prevented.

The invention relates to a method for measuring a particle accelerator beam position. The method for measuring the particle accelerator beam position includes: step S1, sampling N electrode signals of N electrodes of a stripe beam position probe through N channels when the beam passes through the probe; step S2, making up an M*N matrix X for the N electrode signals, and analyzing the principal components of the matrix X; step S3, calculating to obtain beam signal induction amplitudes which are corresponding to the N electrodes one by one, and outputting the N beam signal induction amplitudes through the N channels; step S4, comparing the N beam signal induction amplitudes output from the N channels and calculating to obtain a position signal of the beam. According to the method for measuring the particle accelerator beam position, the electrode signals of the electrode probes are formed into the matrix and are processed simultaneously through a principal component analysis method, and accordingly the noise of each electrode signal is eliminated effectively, and the beam position measuring precision is effectively improved.

The invention relates to an ultrasonic scattering coefficient optimal computation method for crack direction recognition, belonging to the field of nondestructive examination. According to the method, an ultrasonic phased array detection system is used for acquiring full-matrix data of a crack defect; firstly, the acquired data are used for performing full-focusing imaging on the defect to determine the position of the defect and then a scattering coefficient space distribution of the crack defect is computed to determine an angle of the crack; the quantity of wafers included in a sub array and the quantity of wafers between adjacent sub arrays greatly affect the measurement precision of the crack angle. A plurality of evaluation indexes are used for evaluating the quantity of a crack angle measurement result according to the quantities of the wafers included in different sub arrays and the wafers between the adjacent sub arrays, and the measurement result is comprehensively evaluated by a main component analysis method to obtain an optimal measurement result; corresponding parameters, namely the quantities of the wafers included in the sub arrays and the quantity of the wafers between the adjacent sub arrays, are optimal detection parameters.

The invention relates to an intrusion detection method and device for an industrial control system, computer equipment and a computer-readable storage medium. The method comprises the steps: extracting an original data set of the industrial control system from a network data set of a communication protocol of the industrial control system; obtaining a training data set and a test data set from theoriginal data set; performing feature dimension reduction on the training data set and the test data set by using a principal component analysis method to obtain a training data set subjected to feature dimension reduction and a test data set subjected to feature dimension reduction; training the training data set subjected to dimension reduction based on an intrusion detection model to obtain aclassification model; and inputting the test data set subjected to feature dimension reduction into the classification model for classification processing to obtain an intrusion detection result of the industrial control system. According to the scheme, the feature dimension reduction is carried out through a principal component analysis method, redundant information is removed, the calculated amount is reduced, and therefore the technical problem that a traditional technology industrial control system intrusion detection method has a long training time is solved.

The invention provides a rapid nondestructive apple wormhole testing method based on a hyperspectral imaging technology. The rapid nondestructive testing method comprises the following steps: collecting information of an apple spectrum by using a hyperspectral imaging system; analyzing the characteristics of an apple spectrum curve; carrying out main component analysis on spectrum data; meanwhile, determining a single-waveband characteristic image according to a characteristic wavelength; separating to extract apples and wormholes in sequence by using a maximum entropy classification threshold value separation method; extracting an interested area of a PC1 (Principal Component 1) image in sequence; extracting textural features and spectrum characteristics of the interested area; and finally, rapidly and nondestructively detecting apple wormholes by using a BP (Back Propagation) neural network. The method has the advantages of high detection speed, and simplicity and convenience in operation and the precision of a detection result is high; furthermore, the apples can be prevented from being damaged.

A multicomponent fluid composition monitoring and compositional control system, in which a component analysis is effected by titration or other analytical procedure, for one or more components of interest, and a computational means then is employed to determine and responsively adjust the relative amount or proportion of the one or more components in the multicomponent fluid composition, to maintain a predetermined compositional character of the multicomponent fluid composition. The system is usefully employed in semiconductor manufacturing photoresist and post-etch residue removal, in which the cleaning medium is a semi-aqueous solvent composition, and water is the monitored and responsively adjusted component.

The invention provides an earthquake early warning method based on Internet of Things multi-sensor information fusion and a neutral network technology, which comprises the concrete steps of firstly performing networking fusion on the multi-sensor information of multiple earthquake parameter monitoring points by the Internet of Things technology, wherein the information comprises soil radon alpha energy spectrum, well radon alpha energy spectrum, temperature in soil, humidity in the soil, air pressure in the soil, surface temperature, surface humidity, surface air pressure, surface wind speed, surface rain fall, well water level, well water temperature, well water turbidity, well water mercury, well water carbon dioxide, earthquake sound, earthquake tilting and the like; then obtaining abnormal information feature vectors by wavelet analysis, a correlational analysis method, a feature tree search method, a non-linear quantifying and processing method and a main component analysis method; and finally obtaining the relation between the multi-sensor information abnormality and an earthquake by the neutral network technology, thus achieving the aim of predicting the earthquake. The technical scheme of earthquake early warning method based on the Internet of Things multi-sensor information fusion and the neutral network technology is feasible and convenient to implement and can achieve the aim of more accurately predicting the earthquake.

The invention discloses a planar array electrode capacitance sensor. A bottom shielding plate, edge shielding frames and inter-electrode shielding strips are arranged at the bottom so that influences, caused by environment noises, on a measurement result are greatly reduced, and the measurement precision is easy to improve. The invention further discloses a planar array electrode capacitance imaging system and imaging method based on the planar array electrode capacitance sensor; a data fusion technology is applied to image processing of a reconstructed image; a detected material is divided into 8 layers by utilizing ANSYS finite element simulation software aiming at the 3*4 planar array electrode capacitance sensor and a sensitivity matrix of each layer is calculated; image reconstruction is carried out by adopting a Landwerb iterative algorithm; in order to fuse and reconstruct a layered image, a main component analysis method is used for calculating main component vectors of the reconstructed layered image and main components are fused according to percentage of a contribution rate; finally, the fused image is obtained; image errors and related coefficients of the image obtained by data fusion are improved when being compared with that of an original image.

The invention discloses a multi-sensor numerical control machine tool cutter wear monitoring method based on deep learning. The method comprises the following steps that cutting force and vibration signals in a machining process of a numerical control machine tool are collected, and a plurality of sections of signals of the same time intervals are intercepted; the intercepted signals of all sections are normalized and subjected to reduction using a principal component analysis method; a numerical control machine tool cutter wear identification model is built based on a convolutional autocoder,and is trained, and the numerical control machine tool cutter wear identification model comprises a convolution layer, a linear rectification layer, a pooling layer and an auxiliary layer; and the cutting force and vibration signals collected in real time are subjected to normalization treatment and are subjected to reduction by using the principal component analysis method, and then are input into the trained numerical control machine tool cutter wear identification model to obtain a tool wear state identification result. According to the method, wear states of various cutters under different working conditions can be accurately identified in real time.

The invention relates to a method for detecting the pulmonary artery blood pressure by using a heart sound analysis method of a multilayer feedforward network. The method comprises the steps of: extracting a heart sound signal; preprocessing the heart sound signal by fast Fourier transform and normalized average Shannon energy distribution to obtain heart sound signal features; filtering the heart sound signal features by using a principal component analysis method; and performing training learning on the filtered hear sound signal features by using a perceptron neural network or a multilayer perceptron feedforward neural network to obtain an optimal neural network between the heart sound signal features and a pulmonary artery blood pressure value. The heart sound signal features are signal features of a first heart sound and a second heart sound; the heart sound signal features comprise crest frequency, average frequency, crest amplitude, average amplitude, duration and time interface of heart sounds. By adopting the method for detecting the pulmonary artery blood pressure by using the heart sound analysis method of the multilayer feedforward network, provided by the invention, the pulmonary artery blood pressure can be easily and conveniently detected at low cost in a non-intrusive and no risk manner.