34 results about "Standard normal variate" patented technology

The invention discloses a method for discriminating fermentation quality of congou black tea based on a near-infrared-spectroscopy-combined amino acid analysis technology. The method comprises: selecting a sample and performing pre-processing; using high performance liquid chromatograph to determine the content of amino acids in the sample; acquiring the spectrum of the sample, utilizing synergy interval partial least square to establish a near-infrared-spectroscopy quantitative discrimination model for amino acids, finding amino acid variation distribution, and discriminating the fermentation quality of congou black tea. According to the method for discriminating the fermentation quality of congou black tea based on the near-infrared-spectroscopy-combined amino acid analysis technology, pretreatment is performed on an acquired original spectrum by utilizing standard normal variable transformation (SNVT), and the amino acid near-infrared discrimination model is constructed by employing synergy interval partial least square (SiPLS). The invention provides the quantitative determining method for scientifically accurately discriminating the fermentation quality congou black tea.

The invention discloses a near infrared spectrum based discrimination method for Zhongning fructus lycii. The method includes: taking a near infrared spectrometer as a detecting tool, scanning the near infrared spectrum of a fructus lycii sample, conducting distance discriminant analysis on the near infrared spectrum of the fructus lycii sample, in a spectral wavelength range of 6500-5200cm<-1>, by adopting a pretreatment method combining an original spectrum with a multiplicative scatter correction method and the original spectrum with a standard normal variate transformation method, establishing a fructus lycii production place discriminant analysis model and carrying out discriminant analysis on the production place of the fructus lycii sample. The method can reach a recognition rate up to over 90%. During operation, the fructus lycii sample has no need for any treatment. Being nondestructive, rapid and real-time, the method does not employ any reagent, has no harm to human body and environment, and can replace traditional high cost, time-consuming, and laboursome production place discrimination method, thus being a near infrared spectrum based new method for rapid discrimination of the fructus lycii production place.

The present invention provides a probabilistic reliability estimation method and device of an active power distribution network. The method includes the following steps that: the load power and / or power supply power at each node of a power distribution network system are obtained; random variables composed of the load power and / or power supply power are transformed into first standard normal variables which are unrelated to one another; a point estimation method is adopted to construct a sample matrix corresponding to a random variable space based on the first standard normal variables; and failure mode influence analysis is carried out for the sample matrix, so that the probability density function of a reliability index is obtained. According to the method and device of the invention, a condition that the point estimation method can convert influences on reliability caused by the power supply power of DG (distributed generation) and the load power at each node into a deterministic problem so as to solve the deterministic problem is considered, and therefore, calculation is simple; a condition that Nataf transformation is not limited by the type of input variables, so that the relativity problem of the random variables can be effectively processed; and the point estimation method and the Nataf transformation are combined together, and therefore, the accurate estimation of the probabilistic reliability of the active power distribution network can be realized.

Disclosed is a scheme based on near infrared technology and used for detecting phosphorous in water. The scheme includes: utilizing near infrared spectroscopy to measure 0mg / L-50mg / L of phosphorous solutions different in concentration by utilizing NIR equipment to acquire spectra, and storing the spectra as CSV data; utilizing TQ analyst for primary analysis; utilizing a derivative algorithm (Derivatives norris gap), a multiplicative scatter correction (Multiplicative scatter correction, MSC) algorithm, a smoothing (Smoothing) algorithm and a standard normal variate transformation (Standard normal variate transformation, SNV) method to pre-process near infrared spectrum data respectively; utilizing MATLAB for simulation, selecting a simulation model PLSR, utilizing a concentration gradient classifying method for simulation, and building a mathematic model; acquiring simulation results, optimizing the same, and integrating to acquire a final result. By the scheme, quick, environment-friendly, nondestructive and online detection of the phosphorous solutions different in concentration can be realized.

The invention discloses a pit mud total acid prediction model building method based on a hyperspectral image technology. Firstly, hyperspectral image information of infrared and visible light wave bands is acquired, average spectral information of an ROI (region of interest) is extracted through ENVI software, errors caused by non-uniform illumination are eliminated through black and white correction, the average spectral reflectivity of each pixel in an image is estimated, quantitative predication models based on full wave bands and characteristic spectra are built respectively through no processing, SNV (standard normal variate) and different modelling methods, and the best model for quantitative predication of pit mud total acid is screened by comparing and analyzing determination coefficients and root-mean-square errors of a training set and a testing set of each fitting model. The method provides powerful technique support for transformation and upgrading of baijiu brewing industrialization and digital and intelligent online real-time monitoring.

The invention belongs to the field of geological prospecting and specifically relates to a mineral identification method based on laser-induce breakdown spectroscopy and linear discriminant analysis. The method comprises the following steps: 1) collecting data on the basis of an LIBS (Laser-induced Breakdown Spectroscopy) optical system; 2) adopting a cubic spline interpolation algorithm for correcting various spectrogram wavelengths, thereby causing the collected data be consistent; 3) performing standard normal variable correction on the smoothed spectroscopic data, thereby acquiring a spectrum after the standard normal variable correction; 4) utilizing a linear discriminant analysis method to perform feature extraction and clustering analysis on the spectroscopic data after the standard normal variable correction. According to the method, the quick and accurate analysis and discrimination for a mineral sample can be realized in the manner of collecting a certain quantity of analysis samples for forming a training sample and adopting a statistic mode algorithm for acquiring the inherent space structure information of the sample.

The invention relates to a quantitative analysis method for dual adulterated sanqi based on near infrared spectroscopy and stoichiometry. The method comprises: preparing an experimental sample and collecting near infrared spectroscopy of the sample; adopting a grouping manner of integral sequence and local random to divide a data set to a training set and a prediction set; determining an optimum factor number according to variation of RMSECV with the factor number; investigating the pretreatment effect of a SG smoothing method, multiplicative scatter correction, standard normal variables, a first derivative, a second derivative, continuous wavelet transform, and a combination thereof, to obtain an optimal pretreatment method; and adopting the optimal pretreatment method to pre-treat the spectrum, establishing a PLSR model, and predicting an unknown sample. Based on the near infrared spectroscopy and stoichiometry, the method is efficient and lossless, and is fast in detection and highin accuracy. The method is suitable for quantitative analysis of a dual adulterated sanqi sample.

The invention provides a surface enhanced Raman spectroscopy detection method for total arsenic content in food. The surface enhanced Raman spectroscopy detection method comprises the following steps of: preparing Cu2O nanoparticles; synthesizing Cu2O / Ag nanoparticles; establishing a total arsenic content detection standard curve; preprocessing original SERS spectral data through standard normal variable transformation (SNV), and establishing a quantitative detection model on the basis of partial least squares (PLS); and after pretreating a to-be-detected food sample, adding the to-be-detected food sample into a Cu2O / Ag surface enhanced Raman substrate, performing uniform mixing so as to obtain a mixed solution, measuring a Raman spectrum after the mixed solution is dried, and calculating the total arsenic content in the to-be-detected food sample through the quantitative detection model. The Cu2O / Ag surface enhanced Raman substrate prepared by the invention can realize label-free, rapid and high-sensitivity detection of total arsenic in food based on an SERS technology. The quantitative detection model is established based on the standard normal variable transformation pretreatment and the partial least square method, so that the total arsenic content in the food is accurately predicted, and rapid evaluation of heavy metals in the food is realized.

The invention discloses an electronic nose vinegar variety identification method through a fuzzy covariance learning network. The method concretely comprises the following steps that firstly the electronic nose is applied to acquire different varieties of vinegar data, then the acquired vinegar data are preprocessed through a standard normal variable transformation (SNV), then data compression isperformed by using principal component analysis (PCA) and the identification information of training samples is extracted by using linear discriminant analysis (LDA), and finally the vinegar variety is identified according to the fuzzy covariance learning network method. The electronic nose vinegar variety identification method through the fuzzy covariance learning network is high in classification accuracy, simple and convenient, lossless, low in cost and easy to implement and popularize and use.

The invention provides a method for detecting the moisture content of dry powder extinguishing agents based on near infrared spectroscopy analysis. The method comprises the following steps of: 1) establishing a calibration set and a validation set sample spectrum; 2) preprocessing original spectral information through a second derivative, standard normal variate and KarlNorris derivate smoothing filter; 3) establishing a calibration model for the functional relation between a near infrared spectrum of the calibration set and the true value of the moisture content of a corresponding dry powder extinguishing agent sample by using a partial least squares method; 4) validating and optimizing the calibration model; and 5) inputting the near infrared spectrum of the dry powder extinguishing agent sample to be detected into the optimized calibration model, so as to obtain the moisture content value. The method is a method for quickly detecting the moisture content of the dry powder extinguishing agents based on the near infrared spectroscopy analysis, is established based on the analysis of the near infrared spectrums of the moisture content values of different dry powder extinguishing agents; and the detection method has the advantages of being quick in analysis, high in efficiency, simple and convenient to operate, low in cost, and free from pollution to the environment.

The invention discloses a flour quality detection method based on hybrid simulated annealing and genetic algorithms. The flour quality detection method comprises main steps as follows: scanning flourby an infrared spectrometer to obtain spectral information, performing standard normal variable transformation on a flour spectrum to eliminate solid particles and surface scattering; globally searching the optimal spectral signal characteristic with the genetic algorithm, and searching the optimal individual in the genetic algorithm with the simulated annealing algorithm to realize combination ofglobal search and local search; preprocessing characteristic vectors, and constructing a classifier by a radial neural network to classify the processed characteristic vectors, so as to finish flourquality detection. The method has better robustness, by means of spectral preprocessing, spectral noise is removed, model complexity is reduced, and computation efficiency is improved; by combinationof the genetic algorithm and the simulated annealing algorithm, global and local optimization capabilities of a model are enhanced, accuracy of the flour quality detection is improved by the radial neural network, and nondestructive testing is realized.

The invention discloses a method for determining the yield of gellan gum in gellan gum fermentation liquor by utilizing near infrared spectroscopy, belongs to the field of microbial fermentation, andrelates to the technical field of near infrared spectroscopy detection. After the fermentation liquid is sampled at different time points, the yield of gellan gum in the fermentation liquid is analyzed by utilizing a traditional method; meanwhile, a near infrared spectrogram of the fermentation liquid is obtained by scanning of a near infrared spectrograph; an original spectrogram is preprocessedthrough methods such as standard normal variable transformation (SNV), multivariate scattering correction (MSC), smooth filtering (S-G), a normalization method and derivative processing, a mathematical model is established by combining an interval partial least squares method (siPLS), and the model is evaluated through parameters such as correlation coefficients and standard root mean square errors. The method is simple and quick to operate and pollution-free, can greatly save the experimental time and cost in the screening of high-yield gellan gum strains, and can also be used for laboratoryor industrial production of gellan gum.