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 3. Отображено 3.
12-05-2023 дата публикации

Bearing incremental fault diagnosis lifelong learning method based on generated feature replay

Номер: CN116108346A
Автор: SHEN CHANGQING, LIU YAO, DING JIANXIN, WANG SHUBEI, ZHANG FUSHENG, LIU CHENGJIAN, MIAO HONGFANG, WANG DONG, SHI JUANJUAN, HUANG WEIGUO, ZHU ZHONGKUI
Принадлежит:

The invention relates to a bearing incremental fault diagnosis lifelong learning method based on generated feature playback. The method comprises the following steps: dividing a bearing state data set into a plurality of different diagnosis stages; learning a gray image sample in an initial stage, and training a first feature extractor and a first classifier; using the features extracted by the first feature extractor, and utilizing an adversarial generative network to alternately train to obtain a first feature generator; constructing an original fault diagnosis model, and setting the number of neurons in a full connection layer as the number of initial stage fault types; in an incremental learning stage, an original fault diagnosis model in an (n-1) stage is utilized to train and update a fault diagnosis model in an n stage, a feature distillation loss function is utilized to reduce the difference between features extracted by an nth feature extractor and an (n-1) th feature extractor ...

Подробнее
Номер записи: 1
04-04-2023 дата публикации

Machining method of oil nozzle

Номер: CN115890142A
Автор: CHEN GUOJIANG, QIN HONG, LIU CHENGJIAN, PENG YU, NI JIE, ZHANG JIPING, LIU GUOYI
Принадлежит:

The invention relates to a machining method of an oil nozzle. The method comprises the following steps that the appearance, a middle hole, a cooling hole and a cooling cavity of the oil nozzle are machined and formed through additive manufacturing; under the condition that the temperature ranges from 640 DEG C to 680 DEG C, the oil nozzle is subjected to annealing treatment; machining the outer circle, the middle hole and the conical surface of the oil nozzle to set sizes; a cooling cavity and a cooling hole of the oil nozzle are extruded and ground, so that the surface roughness of the cooling cavity and the surface roughness of the cooling hole are Rat; 3.2); a spray hole of the oil nozzle is machined and formed; carburizing, quenching treatment, cold treatment and aging treatment are sequentially conducted on the oil nozzle; and the center hole, the conical surface and the end face of the oil nozzle are precisely ground. According to the method, the problem of low efficiency of design ...

Подробнее
Номер записи: 2
21-07-2023 дата публикации

Bearing small sample fault diagnosis method and system based on meta transfer learning

Номер: CN116465630A
Автор: SHEN CHANGQING, WANG PEIQI, DING JIANXIN, WANG SHUBEI, ZHANG FUSHENG, LIU CHENGJIAN, MIAO HONGFANG, WANG DONG, SHI JUANJUAN, HUANG WEIGUO, ZHU ZHONGKUI
Принадлежит:

The invention relates to a bearing small sample fault diagnosis method and system based on meta transfer learning. The method comprises the following steps: collecting bearing vibration signals under different working conditions; the bearing vibration signals are subjected to fast Fourier transform, the transformed bearing vibration signals are converted into a two-dimensional feature picture set, and the two-dimensional feature picture set comprises a first sample and a second sample; a deep learning convolutional network is trained through the first sample, a feature extractor of the trained network is reserved, a first classifier of the trained network is removed, and the first classifier is used for classifying the first sample; and a model is constructed based on the feature extractor and the second classifier, neuron parameters of a current fault diagnosis task in the model are optimized through meta transfer learning, the feature extractor is enabled to adapt to all other fault diagnosis ...

Подробнее
Номер записи: 3