47results about How to "Save computing space" patented technology

The invention discloses a ResNet-based method for assessing the health status of steam turbine components, which includes: performing a simulation experiment on steam turbine components, and collecting vibration data through multiple measuring points; the vibration data includes fault signal data and normal working condition data; the fault signal The data is tagged; the tag information includes the fault type and remaining usable life; the fault signal data after each tag assignment is segmented and normalized to obtain a sample set; the sample set is divided according to a preset ratio to obtain a training set and verification Set; adopt the strategy of adaptively updating the network learning rate, train the pre-built multi-task learning neural network model based on ResNet through the training set, and obtain the trained evaluation model based on ResNet to the preset convergence condition; realize the steam turbine through the evaluation model Component health assessment. The invention adopts a multi-task learning mechanism, can simultaneously judge the fault type and the health degree of the steam turbine, and has high accuracy.

The present invention provides a multi-satellite mission planning method based on K-means clustering, S1, collecting the user's mission requirements T={t 1 ,t 2 ,t 3 ...t n}, to obtain the set of orbital working hours of each sunlit area corresponding to all currently available satellites O={o 1 ,o 2 ,o 3 ,...o m}. S2, computing task t i to each element o in the set O j The distance Dis ij , forming a task t i The distance set D={d to the orbital set O i1 , d i2 , d i3 ... d in}, the task t i Clustering to the shortest orbital k,Dis ik =Min(D); S3, judging the current clustering scheme s k Whether it belongs to the set S={s 1 ,s 2 ,s 3 ,...s z}, if s k ∈S then output the clustering scheme s k , otherwise the scheme s k Add to the scheme set S, and return to step S2. The present invention quantifies these factors by analyzing the factors affecting multi-satellite task allocation, and combines the K-means clustering algorithm to plan a multi-satellite collaborative task allocation scheme, with fewer iterations and fast calculation speed, which can meet large-scale optimization problems Constrains the time complexity of the algorithm, and greatly improves the quality of imaging and the completion rate of tasks.

The invention relates to a human body communication channel modeling method based on a non-uniform medium and a system thereof. The method comprises the following steps that a human body is divided into multiple structural models according to the structure; the multiple structural models are abstracted into regular geometries; medium layer division is performed in the internal parts of all the structural models, thickness of all medium layers is arranged, and equivalent electrical parameters of all the structural models are computed; and the uniform medium structural models with different electrical parameters are recombined and a human body contour is recovered so that multiple non-uniform medium human body electromagnetic models of the uniform medium structural models with different electrical parameters are constructed. According to the aforementioned human body communication channel modeling method based on the non-uniform medium and the system thereof, the non-uniform medium human body electromagnetic models of the uniform medium structural models with different electrical parameters are constructed so that accuracy is high in comparison with unified uniform medium human body models. Besides, all the structural models are equivalent to the uniform medium structural models so that subsequent computation is simplified and computation space is saved.