3results about How to "Narrow down the area" patented technology

A kind of mobile swimming pool efficient water-free detection device, including vacuum box, slider, jacking element and several regional division elements;Slider is set to the bottom of vacuum box, for transporting mobile swimming pool;Jacking element is inserted between slider, and jacking element is used to lift mobile swimming pool, so that mobile swimming pool is contacted with the inner top wall of vacuum box;The top of vacuum box is provided with several inflation valves that are communicated with the inside of hollow box;Inflation valve is connected with inflation equipment;The top surface and side surface of regional division element are fixed to the inner top surface and side wall surface of vacuum box, and regional division element is located directly above slider;Vacuum box is divided into multiple detection areas that are only communicated at bottom. Regional division element is provided, the space in vacuum box is divided into multiple detection areas by regional division element, and air flow meter is provided in each detection area, the space of vacuum box is refined by regional division element, so as to reduce the range of detection area, improve the efficiency of detection.

The invention provides an unmanned aerial vehicle airborne everything recognition method and system based on multi-modal semantic guidance. The method comprises the steps of obtaining a voice signal, a video image and a first text instruction collected by an unmanned aerial vehicle; semantic vectors of the voice signal and the first text instruction are extracted and fused, and a semantic fingerprint vector is generated; performing feature extraction on the video image by adopting a cascade heterogeneous feature extraction network, and determining a visual feature vector; according to the method, the semantic fingerprint vector and the visual feature vector are subjected to semantic alignment, the semantic similarity between the semantic fingerprint vector and the visual feature vector is calculated, the candidate target is determined based on the semantic similarity, target identification is completed, and the semantic fingerprint vector generated through multi-modal fusion can be used for identifying the target without external communication. Local semantic matching and retrieval can be realized on the collected visual feature vectors, so that the problem of target recognition lag caused by remote communication delay is avoided, and the real-time performance is improved; and meanwhile, reasoning is performed through the heterogeneous feature extraction network, so that the energy consumption is remarkably reduced.

This invention provides a method for analyzing the electrochemical performance of solid oxide battery stacks based on a one-dimensional beaded model, comprising: Step S1, discretizing the solid oxide battery stack into multiple beaded units along the flow channel direction and connecting them in series along the height direction to form a one-dimensional beaded model; Step S2, simulating the flow field and thermal field of the inlet region of the solid oxide battery stack using a three-dimensional finite element model and extracting initial boundary conditions; Step S3, obtaining multi-physics coupled solution results by coupling solutions in each beaded unit; Step S4, dividing the solid oxide battery stack into an end plate region and a uniform intermediate region based on the differential decay index threshold and differential decay index; Step S5, calculating the voltage distribution and thermal response characteristics of the solid oxide battery stack under dynamic load. The beneficial effect is that this invention can balance efficiency and accuracy and characterize the multi-physics spatiotemporal evolution characteristics of the solid oxide battery stack under dynamic load.