Настройки

Укажите год
-

Небесная энциклопедия

Космические корабли и станции, автоматические КА и методы их проектирования, бортовые комплексы управления, системы и средства жизнеобеспечения, особенности технологии производства ракетно-космических систем

Подробнее
-

Мониторинг СМИ

Мониторинг СМИ и социальных сетей. Сканирование интернета, новостных сайтов, специализированных контентных площадок на базе мессенджеров. Гибкие настройки фильтров и первоначальных источников.

Подробнее

Форма поиска

Поддерживает ввод нескольких поисковых фраз (по одной на строку). При поиске обеспечивает поддержку морфологии русского и английского языка
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Укажите год
Укажите год
Применить Всего найдено 12. Отображено 12.
27-06-2023 дата публикации

Actuator control method and controller based on online nerve prediction

Номер: CN116339136A
Автор: WANG YICHUAN, ZHAO CHENGCHENG, SHI ZHIGUO, CHEN JIMING
Принадлежит:

The invention discloses an actuator control method based on online nerve prediction and a controller, and the control method comprises the steps: selecting a hyper-parameter with the upper bound of the maximum confidence interval as an optimal hyper-parameter, thereby maximizing an accumulated reward; the system is identified through online learning of a neural network, so that the dynamic state of the time-varying electro-hydraulic actuator system is learned; the actuator displacement amount of the next sampling time step is predicted through learning the dynamic neural network of the time-varying electro-hydraulic actuator system, and the optimal control amount is obtained through iterative updating, so that the obtained control amount can minimize the displacement tracking error. A controller designed based on a Zynq system-on-chip is used as a device carrier of the control method, and the controller is used for hardware acceleration and implementation of the control method. According ...

Подробнее
Номер записи: 1
30-06-2023 дата публикации

OPC UA information model construction method based on unstructured text data

Номер: CN116362245A
Автор: LIU YANG, SHI ZHIGUO, HE SHIBO, GU CHAOJIE, CHEN CAILIAN
Принадлежит:

The invention discloses an OPC UA information model construction method based on unstructured text data, which mainly solves the problems that the existing OPC UA information model modeling often depends on structured data and lacks description information from text data, and the like, and comprises the following implementation steps: designing an annotation label which is based on corpus text characteristics and meets industrial field modeling requirements, designing an organizational relationship of the labels; performing BIO serialization labeling on the extracted sample text material to obtain a training sample; constructing a Chinese pre-training model based on a BERT model to extract word vectors; based on the CRF technology, word vectors are processed, classified and labeled; and extracting effective entities in the text by utilizing the trained model, and performing model organization according to a predefined organization rule. According to the method, industrial unstructured text ...

Подробнее
Номер записи: 2
30-05-2023 дата публикации

Cascade sparse multi-polarization area array-based two-dimensional direction of arrival and polarization parameter closed joint estimation method

Номер: CN116186474A
Автор: YUE YAXING, YANG TIANZHUO, ZHOU CHENGWEI, SHI ZHIGUO, CHEN JIMING
Принадлежит:

The invention discloses a two-dimensional direction of arrival and polarization parameter closed joint estimation method based on a cascade sparse multi-polarization area array, which mainly solves the problems of higher calculation complexity and limited degree of freedom of the existing method, and comprises the following implementation steps of: constructing the cascade sparse multi-polarization area array; modeling a cascade sparse multi-polarization area array receiving signal model and separating and decoupling multi-dimensional parameters; carrying out block smoothing processing on the covariance matrix of the received signal; constructing a virtual cascade uniform multi-polarization area array receiving signal covariance matrix; characteristic decomposition is carried out on the virtual covariance matrix, and estimation of an included angle between the direction of arrival of each signal source and a single coordinate axis is solved; and solving the estimation of the included angle ...

Подробнее
Номер записи: 3
14-07-2023 дата публикации

Face recognition method and device based on millimeter wave radar phase characteristics

Номер: CN116434301A
Автор: JIANG DECHEN, WANG YONG, SHI ZHIGUO, WANG ZHIMING
Принадлежит:

The invention discloses a face recognition method and device based on millimeter wave radar phase characteristics, and the method comprises the steps: collecting a face radar echo signal, selecting a target point in a frequency domain, and taking the data of the target point as to-be-processed data containing face information; extracting a phase feature in the to-be-processed data, and performing unwrapping operation on the phase feature to obtain a phase change difference of the target point; performing 3D convolution and maximum pooling on the phase change differences of all channel target points to obtain phase features after dimension reduction; carrying out downsampling on the phase features after dimension reduction by using a convolution auto-encoder; and inputting the downsampled data into the convolutional neural network to realize face recognition. Different from face recognition based on vision, the millimeter wave radar is used, the resolution ratio is high, the method is not ...

Подробнее
Номер записи: 4
25-07-2023 дата публикации

MIMO radar multi-dimensional parameter closed estimation method based on polarization subspace orthogonal principle

Номер: CN116482632A
Автор: YUE YAXING, LI TIANYU, SHI ZHIGUO, MIAO JIAHAO, CHEN JIMING
Принадлежит:

The invention discloses an MIMO radar multi-dimensional parameter closed estimation method based on the polarization subspace orthogonality principle, which mainly solves the problem of multi-dimensional parameter estimation under the error of a receiving and transmitting array model of the existing method based on the polarization MIMO radar, and comprises the following implementation steps of: obtaining horizontal and vertical polarization sampling matrixes of a receiving and transmitting array based on array measurement; modeling a signal model of the polarized MIMO radar based on the measured transceiving array sampling matrix; performing matched filtering processing on the received signal; performing multi-dimensional parameter separation and decoupling on the signal after the matched filtering processing; reducing the dimension of the multi-dimensional parameter estimation problem into a direction of arrival estimation problem, and solving the direction of arrival based on a covariance ...

Подробнее
Номер записи: 5
02-05-2023 дата публикации

Static time sequence analysis method and device capable of sensing dynamic power supply noise

Номер: CN116050332A
Автор: CHEN YUFEI, ZHUO CHENG, YIN XUNZHAO, SHI ZHIGUO
Принадлежит:

The invention discloses a static time sequence analysis method and device capable of sensing dynamic power supply noise. The static time sequence analysis method comprises the following steps: modeling a rising edge and a falling edge of the dynamic power supply noise by using an accumulative probability distribution function of deformed Weber distribution; errors caused by misalignment of noise and signal edges are eliminated through a time dislocation factor; calculating time delay and output signal conversion time of each circuit basic unit under different conditions by using SPICE to generate a data set; constructing a multi-layer perceptron, and training by using the previously obtained data set to obtain a trained model; and predicting the time delay and the output signal conversion time of the circuit standard unit under different dynamic noises, different input conversion times and different output load capacitors by using the neural network model so as to obtain a time sequence ...

Подробнее
Номер записи: 6
02-06-2023 дата публикации

Human body gait recognition method based on millimeter wave radar and dynamic sampling neural network

Номер: CN116206359A
Автор: HUANG YANYAN, WANG YONG, SHI ZHIGUO, ZHUO CHENG
Принадлежит:

The invention discloses a human body gait recognition method based on a millimeter wave radar and a dynamic sampling neural network. According to the method, radar point cloud data of human body walking are obtained by using a millimeter wave radar, and point cloud dynamic information is obtained by performing differential processing on adjacent point cloud data frames; the original point cloud data and the processed point flow data are processed through a point cloud feature extraction network to obtain a feature sequence of the point cloud and the point flow; sampling through a dynamic frame sampling module to obtain point cloud and point flow feature subsequences containing rich gait information; and performing feature fusion on the point cloud and the point flow feature subsequences through a time sequence feature aggregation module, and finally outputting a final human body gait recognition result through an average pooling layer and a multi-layer perceptron. A novel radar point cloud ...

Подробнее
Номер записи: 7
04-08-2023 дата публикации

Direction of arrival estimation method based on tensor compression neural network

Номер: CN116541690A
Автор: ZHENG HANG, SHI ZHIGUO, ZHOU CHENGWEI, CHEN JIMING
Принадлежит:

The invention discloses a direction-of-arrival estimation method based on a tensor compression neural network, which mainly solves the problems of huge neural network weight parameter scale, large computing resource loss, long training time and the like in the existing method, and comprises the implementation steps of: carrying out tensor modeling on a uniform planar array receiving signal; carrying out five-dimensional covariance tensor derivation; generating a state space tensor based on inverse Tucker decomposition; constructing a tensor compression neural network and carrying out back propagation on an output layer loss function; and carrying out tensor compression neural network acceleration training and direction of arrival estimation result output. According to the method, on the premise that the structural characteristics of the multi-dimensional received signals are reserved, the state space generation of the neural network layer is mapped into the inverse tensor decomposition ...

Подробнее
Номер записи: 8
30-05-2023 дата публикации

Multi-target unmanned aerial vehicle identification device and method based on radio frequency array

Номер: CN116184322A
Автор: WANG YUANLIANG, CHEN JIMING, SHI ZHIGUO, ZHOU CHENGWEI
Принадлежит:

The invention discloses a multi-target unmanned aerial vehicle identification device and method based on a radio frequency array. The system comprises a radio frequency antenna array, a receiver, an AD acquisition module, an FPGA data processing board, an FPGA data conversion board, a signal processor, and a Zynq development board supporting PYNQ. The radio frequency antenna array receives airspace unmanned aerial vehicle signals, the receiver and the AD module are configured to acquire multichannel digital signals, a beam forming algorithm is realized in the FPGA data processing board to complete signal acquisition of airspace filtering, and the acquired signals are uploaded to the signal processor through the FPGA data conversion board. And designing a deep learning scheme of secondary target auxiliary identification to identify the unmanned aerial vehicle of each suspected target, and finally deploying a deep neural network on a Zynq board to realize off-line real-time identification ...

Подробнее
Номер записи: 9
22-08-2023 дата публикации

Sparse array DOA (direction of arrival) estimation method based on CNN (convolutional neural network) convolution kernel tensor decomposition

Номер: CN116629326A
Автор: ZHENG HANG, ZHOU CHENGWEI, SHI ZHIGUO, WANG QING, CHEN JIMING
Принадлежит:

The invention discloses a sparse array direction-of-arrival estimation method based on CNN convolution kernel tensor decomposition, mainly solving the problems of high network training cost and low efficiency of the existing sparse array direction-of-arrival estimation method based on a convolutional neural network, and the implementation steps are as follows: constructing a two-dimensional nested co-prime plane array; modeling a sparse tensor signal of the nested co-prime planar array and deducing a cross-correlation tensor; constructing virtual domain tensor input of the convolutional neural network; carrying out CNN (Convolutional Neural Network) convolution kernel decomposition and compression convolution calculation; and estimating the two-dimensional direction of arrival based on the fast training of the compressed convolutional neural network. According to the method, tensor decomposition is carried out on a high-dimensional CNN convolution kernel, a corresponding sparse tensor statistic ...

Подробнее
Номер записи: 10
25-07-2023 дата публикации

Image recognition method and device and electronic equipment

Номер: CN116486236A
Автор: WANG SHUYI, ZHOU TAO, CHEN JIMING, SHI ZHIGUO
Принадлежит:

The invention provides an image recognition method and device and electronic equipment. The method comprises the following steps: acquiring a to-be-recognized image; extracting a plurality of different levels of pixel-level features of the to-be-recognized image through a pixel-level feature extraction backbone network of the image recognition model; fusing the plurality of different levels of features under different scales by using a feature pyramid network of an image recognition model to obtain a plurality of fused features under different scales; the convolution kernel branch network of the detection head is utilized to process the fusion features under multiple different scales so as to acquire a convolution kernel and a classification result; and processing the fusion features under the plurality of different scales by using the mask branch network of the detection head to obtain mask features, according to the method, the mask features, the convolution kernel and the classification ...

Подробнее
Номер записи: 11
02-06-2023 дата публикации

Airspace small target detection method based on deep learning semantic feature separation

Номер: CN116206219A
Автор: LI MIAO, WANG XINGJIAN, ZENG GUANGYANG, SHI ZHIGUO, CHEN JIMING
Принадлежит:

The invention discloses an airspace small target detection method based on deep learning semantic feature separation. The method comprises the steps that an input picture is preprocessed, noise interference is eliminated, and the learning ability of a small target is improved; processing the target picture by adopting a neural network, and extracting semantic information and texture information of the neural network through deconvolution operation; using a detector to generate a target area existence probability graph, using probability graph filtering to process target semantic and texture features, eliminating background feature interference, retaining target foreground features, and sparsifying a target feature representation space; adopting a Transform structure network to query and process target semantic information and texture feature information, extracting relevance among target features, and generating a plurality of target prediction frame sets; a Hungary algorithm is adopted ...

Подробнее
Номер записи: 12