10results about How to "Reasonable" patented technology

The application discloses a deepwater oilfield development yield prediction method, an electronic device, an apparatus and a storage medium, wherein the method comprises: predicting the natural yield of old wells, the yield increasing potential of old wells, the yield of new wells, planned yield loss and non-planned yield loss for each prediction year respectively; wherein predicting the non-planned yield loss comprises: determining the development stage to which a well area belongs, wherein for the well area in the aging stage, a historical curve graph of the non-planned loss rate and time relationship is determined first, the non-planned loss rate in the prediction year is calculated by fitting the relationship between the non-planned yield loss rate and the production time of the aging stage in the curve graph, the non-planned loss rate in the prediction year is corrected by using a correction factor to determine the final non-planned loss rate in the prediction year, so as to determine the non-planned yield loss in the prediction year; and the deepwater oilfield development yield is calculated according to the above prediction results.

The application discloses strigasiaticaside A and an extraction method and application thereof, and finds, through chemical component research on the whole grass of striga asiatica, a new synbinaphthyl tetrahydrofuran lignan glycoside, that is, strigasiaticaside A, identifies the chemical structure of the new compound through nuclear magnetic resonance, mass spectrometry, ultraviolet, infrared and circular dichroism spectrum technology.The structure of the new compound is identified as (-) sesamin-5-O-beta-D-glucopyranoside.The anti-hepatitis activity of the new natural product is evaluated by using LPS stimulated HepG2 cells.The results show that when the concentration is 100 muM, the compound has a significant inhibitory effect on the production of inflammatory factors IL-10 and NF-kappa B protein, and has no cytotoxicity, indicating that the compound may have potential anti-hepatitis activity.

The present disclosure provides a kind of teaching model training based on federal learning, teaching task processing method and device, it is related to artificial intelligence, natural language processing, big language model, deep learning and other technical fields.The specific implementation scheme is: receiving the initial teaching model output by central server, initial teaching model is used to generate or execute teaching task;Using local data sample set and composite loss function, the initial teaching model is trained, and the teaching model that training is completed is obtained, the composite loss function includes: the task loss function of teaching task and the atlas constraint loss function for making the internal representation generated by teaching model when processing text and the knowledge concept involved in text in the structured representation of knowledge graph in alignment;Based on the model update amount of the teaching model that training is completed and initial teaching model;Based on knowledge graph, obtain entity embedding matrix;The model update amount and entity embedding matrix are sent to central server.

The application discloses a kind of based on multi-dimensional data fusion's campus intelligent consumption analysis system and method, it is related to electric digital data processing technical field.The system includes terminal feature fusion, abnormality identification analysis processing, classification verification and synchronous three big modules.Terminal feature fusion module fuses terminal feature data and time feature data in original multidimensional data, generates behavior correlation feature, reduces data dispersion;Abnormality identification analysis processing module carries out abnormality identification and processing to behavior correlation feature, forms feature data set and verifies validity, improves detection accuracy;Classification verification and synchronous module uploads the classification result after verification as core data set to specified service platform, and synchronizes to its database, improves platform processing efficiency.Through multi-module cooperation, the efficient fusion of campus consumption data, accurate anomaly detection and data synchronization are realized, and the comprehensiveness of campus consumption analysis and service platform processing efficiency are improved.

The application provides a method for optimizing deployment data of an unmanned aerial vehicle (UAV) assisted by a reflecting surface and related equipment. The method comprises obtaining deployment data corresponding to an UAV in a predetermined area; based on the deployment data, constructing an objective function for optimizing the deployment data, with the goal of maximizing the ratio of total safe communication rate to total communication energy consumption, and constructing constraint conditions corresponding to the objective function; and based on the constraint conditions, solving the objective function to obtain the maximized ratio and the optimized deployment data. The method solves the technical problem that, in the prior art, the deployment data corresponding to a network will cause a conflict between the total safe communication rate and the total communication energy consumption of the network when the network is deployed by using an intelligent reflecting surface and an UAV, and achieves the goal of achieving a good balance between safety performance and energy consumption in a cell-free network.

The invention discloses a crop remote sensing sample automatic construction method based on time-space-spectrum combined characteristics. The method comprises the following steps: acquiring multi-period optical remote sensing images of a target year in a research area; traversing unmarked cultivated land pixels in the research area, and calculating the similarity between the time sequence feature, the spectral feature and the spatial feature of each cultivated land pixel and the corresponding crop type feature signature library in the time sequence dimension, the spectral dimension and the spatial dimension; and performing spatial connectivity verification on the candidate amplified sample points by judging the comprehensive similarity, and combining the final amplified sample points with the core sample library to form a massive enhanced sample library. Through a step-by-step filtering mechanism of time sequence consistency screening, spectrum consistency screening, space purity screening and neighborhood verification, multi-source feature information of a time dimension, a spectrum dimension and a space dimension is fused, and a high-purity core sample library is constructed.

The invention discloses a spectrum detection method and device based on a spiral microchannel flow cell. The method comprises the following steps: collecting a sample fluid, carrying out multi-inlet spiral channel mixing strengthening treatment to generate a mixing strengthening characteristic, and generating mass transfer efficiency distribution based on the mixing strengthening characteristic; determining a key mass transfer enhanced area through mass transfer efficiency distribution, and extracting concentration gradient distribution characteristics to determine an in-situ detection sensitive area; carrying out multi-parameter detection on the in-situ detection sensitive area and fusing to determine an effective acquisition interval to obtain original spectral data; determining a response precision threshold value based on the effective acquisition interval and the mixed strengthening characteristics, and performing flow state stability evaluation on the key mass transfer strengthening region to form a flow state stability index to construct partition monitoring configuration; and identifying an abnormal response area to generate a flow state correction coefficient, and correcting the original spectrum data to generate a spectrum detection result, thereby realizing self-adaptive spectrum detection in the spiral micro-channel.

The application discloses a kind of aluminum alloy variable polarity TIG deep penetration welding penetration prediction method and system, comprising: collecting welding pool image, input pre-trained type convolutional neural network to carry out pool state identification;If it is positive polarity stage, then the image is input to pre-trained semantic segmentation type convolutional neural network, the shadow area caused by pool area and small hole is marked, and the segmentation result image is obtained;Approximate depth of pool width and small hole is calculated based on the segmentation result image;The pool width and the approximate depth of small hole are input into the pre-trained integrated learning prediction model to predict the current penetration;The system includes image acquisition device, pool state identification module, feature region segmentation module, feature size calculation module and penetration prediction module;The application can realize the real-time prediction of aluminum alloy blind hole type deep penetration welding weld penetration, and provide support for energy control and process optimization of welding process.

The application discloses an ancient village landscape evaluation method and system based on visual field analysis, which comprises the following steps: collecting data of a target area and preprocessing the data; constructing a visual field analysis optimization model based on the preprocessed data; obtaining visible areas in the target area based on the visual field analysis optimization model; combining water resource flow, landscape grades and visible areas in the target area to construct an ancient village landscape comprehensive evaluation method; obtaining a comprehensive evaluation index score of an observation point based on the ancient village landscape comprehensive evaluation method; judging whether the comprehensive evaluation index score of the observation point is greater than a set threshold value, and if yes, taking the observation point as a predicted viewing platform site; judging whether the distance between the predicted viewing platform site and an original viewing platform address is greater than an error range, and obtaining the accuracy of the predicted viewing platform site. The application is more scientific, reasonable and operable for the selection of the ancient village viewing platform.

The invention relates to the field of image recognition, in particular to an offset paper defect detection system based on image recognition, comprising: a defect acquisition unit used for recognizing defect areas of target images and determining defect categories corresponding to the defect areas; the acquisition optimization unit is used for judging whether to adjust the acquisition frequency or not according to the defect category stability and the defect category uniformity; the region division unit is used for determining a calibration region and a polymerization region of the target image and determining whether calibration correction is carried out on the calibration region or not according to the co-occurrence frequency of the free defect point and the calibration region; the data storage unit is used for determining a storage mode according to the characteristic difference degree and the characteristic cardinal number; according to the invention, the selection efficiency of defect sample data is improved.