30results about How to "Identify fit" patented technology

The invention relates to an online identification and visualization method for power grid disturbances based on multi-resolution wavelet analysis, which belongs to the technical field of wide-area dynamic monitoring and visualization of power systems. The method comprises the following steps: carrying out offline simulation analysis on the features of typical dynamic disturbance accidents in power grid, and selecting a wavelet function Psi and the decomposition layer-number I corresponding to the obtained maximum wavelet coefficient energy as online wavelet multi-resolution analysis tools; carrying out regionalization according to the PMU (power management unit) configuration situation of a target power grid, so that each region has one and only one PMU substation for measurement, and transmitting frequency signals measured by each PMU substation to a WAMS (wide area measurement system) master station; acquiring frequency signals of each PMU substation from the WAMS master station, and carrying out I-layer online wavelet multi-resolution analysis on the frequency signals by using the determined wavelet function so as to obtain wavelet coefficients of an I(th) layer; and discriminating disturbance occurrence time, disturbance occurrence potion and disturbance type according to the wavelet coefficients; and carrying out visualized display through taking a map as the background. The method disclosed by the invention can improve the running reliability of an electrical power grid.

The invention discloses an image recognition method based on a mixed model, comprising the following steps: step 1, constructing a feature encoder F from an original image space to a feature space; step 2, constructing a generation model D that fits the probability distribution of the feature space; Step 3, constructing a known class classification model C for K classification of the feature space, where K is the number of known classes; Step 4, jointly training the feature encoder, generation model and classification model to optimize the feature space; Step 5, for The generative model after joint training sets the threshold for identifying unknown classes; step 6, according to steps 1, 2, 3, 4, 5, obtains a K+1 classification hybrid model that can identify images of unknown classes, so as to realize classification and recognition of images to be classified .

The invention discloses a method for identifying the authenticity of the mountain bean root. The authenticity of the mountain bean root medicinal sample is judged by measuring the content of phenylbenzofuran in the mountain bean root medicinal sample. The detection method provided by the invention is accurate and reliable, can effectively eliminate the interference of other impurities in the sample, and can obtain a satisfactory separation and detection effect, and the method is simple, easy to operate, low in cost and strong in practicability, and is very suitable for the detection of product quality.

The invention discloses a method for identifying varieties of camellia flower color buds, using high-performance liquid chromatography-photodiode array detection and ultra-high performance liquid chromatography-quadrupole-time-of-flight mass spectrometry to detect the red petals and pink color of camellia varieties. The composition and content of anthocyanins in the petals of white bud varieties were studied, combined with the analysis of flower color phenotypes, the effective identification of camellia color bud changes was carried out, the relationship between camellia flower color bud changes and anthocyanins was clarified, and the color bud change of camellias was revealed. Formed material basis, thus providing a scientific basis for camellia color bud breeding.

Provided are a motion pedestrian identification method and a device. A video camera shooting direction is taken as a z axis, a pedestrian height direction is taken as a y axis, a plane perpendicular to the ground is taken as an xy axis plane, and then an xyz rectangular coordinate system is established. The motion pedestrian identification method comprises obtaining a candidate pedestrian contour of a time sequence form a depth map of the time sequence, obtaining at least one of x-direction distance distribution, y-direction distance distribution and z-direction distance distribution of the candidate pedestrian contour of the time sequence on the basis of the candidate pedestrian contour of the time sequence, and judging whether a candidate pedestrian contour is belonged to a pedestrian on the basis of at least one of all directions of the candidate pedestrian contour of the time sequence. The distance distribution of a pedestrian contour in all the directions is basically only related to body characteristics and walking characteristic of a person, so the pedestrian contour is unlikely to be influenced by external environment and is suitable for pedestrian identification, the pedestrian can be identified rapidly, and the real-time requirement of many applications of, for example, a vehicle driving assistance system is fitted.

The present invention relates to a multipoint recognition method and system of an infrared touch screen. The present invention comprises: first generating a data structure, corresponding to optical paths on an infrared touch screen in a recognition process, with a two-dimensional matrix arrangement property; and then extracting touch point information according to the data structure. Preferably, in a recognition process, it is determined, according to optical path data, whether a touch point exists on the infrared touch screen; if no, the current recognition process is ended. Preferably, if a touch point exists, an area where the touch point is located is further determined according to the optical path data, and a data structure corresponding to an optical path in the area is generated for the area where the touch point is located; and then touch point information is extracted according to the data structure. The present invention provides a novel multipoint recognition method and device of an infrared touch screen, which are applicable to recognition of multiple touch points, have a high touch response speed and high recognition precision, and have no ghost point or only a few ghost points.

The invention discloses a method for recognizing a speaker based on a multivariate core logistic regression model, comprising the following steps: (A) extracting voice features of the speaker: collecting voice signals of the speaker to be recognized to pre-process, and then extracting mel cepstrum parameters; (B) constructing a speaker model: using a multivariate core logistic regression model asa speaker recognition model; (C) training the speaker recognition model: using the feature vectors extracted from the step A as input training samples, through a minimal sequence optimization algorithm, carrying out an iterative training to optimize the model parameters; (D) recognizing the speaker: extracting the feature vectors of the voice signals of the speaker to be recognized and inputting to the recognition model of the trained speaker, and giving out a posterior probability of each speaker by the multivariate core logistic regression model, wherein the highest probability value is a recognition result. The invention has high rate of recognition, simple model construction and good rapidity.

The invention provides an action recognition method and its neural network generation method, device, and electronic equipment, and relates to the technical field of image recognition. The neural network generation method used for action recognition includes: detecting a target image to obtain key point information of a human body; The convolution kernel bias information is obtained according to the key point information of the human body; the deformed convolution neural network is generated based on the initial convolution neural network according to the convolution kernel bias information, which solves the current image recognition neural network existing in the prior art The technical problem of the network's low recognition ability for action recognition.

The invention provides an action recognition method and its neural network generation method, device, and electronic equipment, which relate to the technical field of image recognition. The neural network generation method for action recognition includes: extracting a target image to obtain optical flow features; The optical flow feature obtains convolution kernel bias information; according to the convolution kernel bias information, a deformed convolution neural network is generated based on the initial convolution neural network, which solves the problem of image recognition neural network in the prior art for action recognition The less effective technical issues.