5results about How to "Break the limits" patented technology

The invention belongs to the technical field of ship control motion prediction, and discloses a ship control motion real-time prediction method and system based on a fusion technology. According to the method, standardization processing is carried out on ship motion data and marine environment data, the marine environment data and the ship motion data are enabled to be input into a constructed ship motion prediction model based on a Transform architecture in the same numerical range, a sliding time window technology is used, online gradient updating is carried out, and the method adapts to new data features; a physical loss item is introduced, and an improved ship motion prediction model based on a Transform architecture is obtained; and carrying out optimization verification to finish nonlinear ship control motion prediction. According to the method, the defects of a traditional model in the aspect of nonlinear processing can be overcome, the reliability of a data driving model can be improved through physical constraints, and powerful technical support is provided for operation control and safe navigation of intelligent ships in the future.

The application discloses an image semantic segmentation model which is composed of a designed mixed attention focusing method, an attention rectification residual module (ARRM) and a mixed feature integration module (MFFM), is trained in a deep supervision mode as a whole, is reasonably provided with deformable convolution, is combined with a constructed multi-scale spatial attention module (MSP) and a double-pooling attention module (DPA) to be jointly optimized, and the problem of small target feature segmentation difficulty is solved. In order to reflect the excellent performance of the model on a downstream task, the training between different data sets is completed in a transfer learning mode, and the application range of the model is expanded. Finally, grouping convolution is added in the backbone network, the calculation cost is greatly reduced, and the problem of high training cost of the segmentation model is solved under the premise of guaranteeing the segmentation effect.

This invention provides a method, apparatus, and device for detecting the physicochemical parameters of oil, relating to the fields of spectral analysis and industrial equipment condition monitoring. The method includes: acquiring hyperspectral data of the oil to be tested; inputting the hyperspectral data into a pre-trained spectral decoupling network to output a base spectral component related to the oil brand, and a common component feature vector unrelated to the oil brand but related to the oil's health status; matching the common component feature vector with a pre-built general reference library to determine the physicochemical parameter values ​​of the oil to be tested. The method of this application achieves efficient and accurate detection of physicochemical parameters of oils from multiple brands, breaking the limitation of "one model per brand," effectively improving the accuracy of the detection results and the generalization ability across brands, significantly reducing the deployment and maintenance costs of the detection system, and significantly improving the practicality of multi-brand oil condition monitoring.

This invention relates to the field of machine vision and industrial defect detection, specifically a metal sign surface inspection system based on machine vision. It includes modules for image acquisition, feature analysis, strategy decision-making, dual detection, and result output. The system extracts surface image features to construct a visual feature space and calculates semantic drift entropy. Its core is to dynamically switch based on the comparison between the entropy value and a preset danger threshold: at low entropy values, a generative vision model is called to extract depth features; at high entropy values, a physical optical morphological filtering algorithm is called to extract basic morphological features. This invention breaks the limitations of a single detection architecture, utilizes adaptive scheduling to solve the problem of environmental light spot noise interference, and balances the ability to identify subtle surface defects under stable lighting with the robustness of structural detection under complex optical interference.

The present application belongs to the field of urban fire safety and artificial intelligence technology, and discloses a kind of urban fire risk and hidden danger intelligent identification analysis system and method.System contains multi-source data acquisition, multi-modal feature fusion, adaptive threshold adjustment, space-time prediction analysis, visualization early warning and fire management terminal.Collection module real-time acquisition monitoring, sensor and BIM data and pre-processing;Fusion module extracts features through convolution network, long short-term memory network, and weights fusion through multi-head attention mechanism;Adaptive module dynamically updates risk determination threshold;Space-time prediction module predicts the space-time evolution of risk;Visualization module displays hidden dangers through BIM three-dimensional display.The present application solves the problems of low detection efficiency, high false alarm and missed detection rate, and lack of dynamic prediction capability, and is suitable for urban fire safety intelligent management.