74 results about "Similarity model" patented technology

The invention discloses a multi-parameter distortion compensation method for a fractured rock mass seepage similar model test, and relates to the technical field of rock mass engineering physical model tests. The method comprises the following steps of: firstly, determining two constraint conditions that the fluid property is invariable and the gravitational acceleration is invariable, correspondingly selecting a basic physical quantity and extracting a core criterion; then analyzing three types of distortion sources of geometric dimensions, dielectric materials and boundary conditions, and establishing a distortion function according to result parameters of seepage flow velocity, seepage flow and seepage pressure; secondly, parameters difficult to control are selected as distortion items, parameters easy to regulate and control are selected as compensation items, the distortion items are substituted into a distortion function to obtain a distortion ratio, the optimal distortion ratio of the compensation items is solved based on a compensation balance formula, and the parameters are adjusted; and finally verifying the interval range of the actually measured distortion ratio. According to the method, the multi-source distortion coupling influence is accurately quantified, the compensation pertinence and precision are improved, operation is easy and convenient, universality is high, and reliable support is provided for underground engineering seepage research.

The invention discloses a rainfall image sequence similarity search method based on a twin neural network. The rainfall image sequence similarity search method comprises the steps of obtaining a target rainfall image sequence and a target rainfall image library; constructing a rainfall image similarity model; based on the target rainfall image sequence and the target rainfall image library, performing training optimization on the constructed rainfall image similarity model to obtain a target rainfall image similarity model; and based on the target rainfall image sequence and the feature vector of the corresponding frame in the target rainfall image sequence extracted by the target rainfall image similarity model, carrying out rainfall image sequence similarity search on the target rainfall image library to obtain a rainfall image sequence similar to the target rainfall image sequence. According to the method, deep learning and a meteorological mechanism are fused, efficient similarity search of a rainfall image sequence is realized by constructing a rainfall image sequence similarity search model of a feature extraction network and twin neural network architecture, and three bottlenecks of feature robustness, time sequence modeling and calculation efficiency are broken through.

The invention discloses a tobacco marketing abnormal address management and control method, system and device based on GIS and multi-dimensional data fusion and a medium, and belongs to the technical field of tobacco industry informatization and internet marketing. The technical problems that in the prior art, address risk control dimensionality is single, query precision is insufficient, data linkage is lacked and visual analysis capacity is weak are solved. According to the technical scheme, the method comprises the steps that abnormal address data is constructed and queried based on a multi-dimensional label; gIS visualization and correlation analysis based on a mixed similarity model: GIS positioning and range loading, spatial candidate set filtering, text similarity fine screening, mixed weighted scoring and visualization rendering and statistics are realized through the mixed similarity model of spatial index coarse screening and text algorithm fine screening, and correlation order detail form display based on GIS range triggering is carried out.

The invention discloses a denture three-dimensional model feature construction and matching method based on curvature adaptive sampling, which comprises the following steps: S1, acquiring denture surface point cloud data through three-dimensional scanning, converting the data into an STL model, and comparing a historical STL model pre-stored in a folder; s2, constructing a self-adaptive sampling weight, selecting a triangular patch through replacement random sampling, and generating a sampling point; s3, carrying out normalization processing on the sampling points; s4, calculating the distribution of D2 and A3 by using the normalized sampling points, and obtaining corresponding shape features; s5, splicing the features to form a joint geometric feature vector; and S6, performing S1-S5 on the to-be-matched false tooth to obtain a matching vector, performing S2-S5 on the historical models one by one to obtain comparison vectors, calculating cosine similarity between the matching vector and each comparison vector, and outputting a Top-K similarity model. According to the method, key geometric structure feature expression is enhanced, robustness to point cloud density change, local defects and the like is high, and matching is ensured to be reliable and consistent.

The invention discloses a remote bank fuzzy question-answer accurate matching method based on deep learning. The method comprises the following steps: firstly, constructing a business semantic system (clear business class-scene-specific intention), a term and variant mapping library and a sentence pattern template containing a replaceable entity slot position; extending the original similar question pairs to generate derivative sentences and candidate similar sentences, filtering, clustering to generate cross-sentence pattern similar pairs, and selecting negative samples to control the literal overlapping rate; then constructing a twin BERT network, mixing original and enhanced samples into a training set, and training in two stages by using improved triple loss; finally, sentence vectors are generated through the trained twin network, semantic similarity is judged according to cosine similarity, question pairs which are judged to be similar through a model are subjected to secondary verification through a service rule base, and score reduction is carried out if the question pairs do not pass verification.

The invention relates to the technical field of aero-engines, and discloses a dynamic tester design method and system for simulating a complex rotor system, and the method comprises the steps: constructing a three-dimensional entity similar model of a dual-rotor supporting system, and analyzing the modal shape vector of each order of a one-dimensional dynamic parameterized model of the dual-rotor supporting system. Calculating the parameter modal confidence coefficient and the associated modal confidence coefficient of the similar model according to the modal shape vector of the similar model and the modal shape vector of each order of the similar model, and adjusting the similar model by taking the two modal confidence coefficients as evaluation criteria to obtain a similar model with high confidence coefficient; according to the total rigidity, the supporting rigidity, the squirrel cage rigidity and the end seal rigidity of the similar model, an oil film damper rigidity reference value is obtained through analysis, the oil film damper structure is adjusted, and therefore the oil film damper structure meeting the design requirement is output. According to the invention, the bearing environment of a real aero-engine rotor can be simulated with high fidelity, and the problem of insufficient measurement dimension in a rotor synchronous monitoring scheme of a double-rotor tester is solved.

The application relates to a method for training a text similarity model, comprising: obtaining one or more original offer text pairs; performing the following operations on each of the one or more original offer text pairs: generating a first similarity score of the original offer text pair based on a first text similarity model; extracting an offer feature parameter corresponding to the original offer text pair; correcting the first similarity score based on the offer feature parameter to generate a second similarity score by using a preset correction formula; performing floating sampling on the offer feature parameter to generate a plurality of enhanced offer feature parameters, and restoring each of the plurality of enhanced offer feature parameters to an enhanced offer text pair; and performing model training by using an enhanced training set to generate a second text similarity model, wherein the enhanced training set comprises the one or more original offer text pairs and the enhanced offer text pair generated based on each original offer text pair.

The invention provides a method for constructing a bearing test distortion model of an ice-resistant structure under the action of a dynamic ice load, and belongs to the field of test similar reduced scale model design under an ice breaking load of the ice-resistant structure. The problem that in the prior art, the strain rate of a fully-similar model is dissimilar due to the scale effect is solved; a traditional distortion model design method has the problem of dissimilar energy distribution. The method comprises the following steps: establishing a fully similar model from an ice-resistant structure prototype to a geometric and plate thickness equal ratio reduced scale; establishing a plate thickness distortion reduced scale model by using the same material on the basis of the geometric similar reduced scale model; preliminarily determining a collision contact position of the thickness distortion reduced scale model; a dynamic ice load similarity criterion is applied, correction factors are deduced through dimensional analysis, and ice body parameters and collision speed of a distortion model system are adjusted, so that the ice breaking capacity and dynamic structure response are similar to those of a prototype system. The device is mainly used in the field of polar region icebreaking.

A URL is input into a virtual browser of a crawler which requests that URL and all of its resource URLs in order to form a network graph. Network features are extracted. An identity stage uses an identity model based upon lexical and host features to determine if the input URL and resource URLs are malicious or benign. A similarity stage uses a similarity model and similarity functions to produce a similarity score between each pair of the input URL and a resource URL. An activity stage uses an activity model to determine if the input URL's network is malicious or benign based upon the network graph and extracted network activity features. Edges of the graph are weights, similarity scores, attributes or network features. A finality stage concludes malicious if the identity stage concluded that the URL was malicious or the activity stage concluded that the URL's network was malicious.

The application discloses a kind of long approach in paste flow slope angle prediction method, belong to filling technical field, solve the prediction of filling slope angle in prior art is based on experience formula or static homogeneous model, unable to consider the influence of waste rock dynamic settlement on slurry flow resistance and slope form Problem, the method includes based on paste raw material basic parameter construction paste similarity model, based on range analysis combined with interaction analysis influencing factor on the influence degree of flow slope angle, convert flow slope angle calculation model into slope angle satisfaction function, obtain slope angle optimal filling parameter;The present application can comprehensively study the influence of multiple influencing factors and their interaction on paste flow slope angle through the pre-design and implementation of orthogonal experiment scheme, avoids the one-sidedness that single factor analysis may bring, makes the influence law obtained more comprehensive and in-depth, and is different from the homogeneous flow model or simplified discrete element model that the existing numerical simulation method is mostly used.

This invention discloses a method and system for identifying cross-connection issues in low-voltage distribution areas. It acquires voltage and current time-series data of users in the low-voltage distribution area, cleans the data using a mask interpolation strategy, and adaptively extracts voltage drop and current surge features based on the median absolute deviation algorithm. It constructs a voltage trend similarity model based on local regularization kernels and a mutation co-occurrence similarity model based on event level weights, employing a dual-track strategy combining daily rigorous assessment and feature revival to generate suspected voltage correlation pairs. A density-based clustering algorithm is used to generate an initial population, and a voltage similarity-weighted centroid screening mechanism is introduced to remove loose outliers in the feature space. A pairwise cross-validation model is constructed based on electrical causality, and the population is topologically refined using relative voltage response criteria and a maximum fully connected clique extraction algorithm to determine the final cross-connection population. This invention significantly improves the accuracy and robustness of cross-connection identification.

The invention provides a near-infrared displacement monitoring method and system based on multi-reference point correction, and the method comprises the steps: obtaining a near-infrared target image sequence, and extracting a candidate reference point set and a measuring point set; performing multi-dimensional feature extraction on the two to obtain a feature vector set; inputting the feature vectors into a preset multi-dimensional similarity model for similarity evaluation, and generating a comprehensive similarity matrix; self-adaptive matching is carried out based on the matrix and a similarity threshold value, and a reference point-measuring point matching relation table is established; and finally, performing differential correction on the measuring point set according to the matching relation to obtain a measuring point displacement value set. According to the near-infrared displacement monitoring method and system based on multi-reference point correction provided by the invention, the reference stability is improved through multi-reference point screening and multi-dimensional feature extraction, and adaptive matching is performed through the preset multi-dimensional similarity model to ensure the matching accuracy, so that the displacement data accurately reflects the real deformation of the structure, and the accuracy of the real deformation of the structure is improved. And the accuracy and robustness of a monitoring result are improved.

The invention relates to an engineering site residual deformation prediction method and device based on a physical similarity model. The method comprises the following steps: acquiring earth surface residual deformation time of a target monitoring area and observation time of a preset physical similarity model experiment; determining a time similarity ratio and a first geometric similarity ratio based on the earth surface residual deformation time and the observation time; obtaining prototype residual deformation of the target monitoring area and model residual deformation obtained by monitoring through monitoring equipment; determining a second geometric similarity ratio based on the prototype residual deformation and the model residual deformation; selecting a minimum value in the first geometric similarity ratio and the second geometric similarity ratio as a target geometric similarity ratio; and performing amplification processing on the residual deformation simulation value by using the target geometric similarity ratio to obtain a residual deformation prediction value, and performing amplification processing on the residual deformation simulation time by using the time similarity ratio to obtain a residual deformation duration prediction value. According to the scheme, long-term prediction of the earth surface residual deformation of the target monitoring area is realized.

The invention discloses an intelligent determination method and device of a network strategy and a storage medium, relates to the field of communication, is used for accurately matching the network strategy for network equipment, and comprises the following steps: receiving strategy request information; the strategy request information comprises resource demand information; determining at least one candidate network strategy from a strategy knowledge base based on the strategy request information and a similarity model; the strategy knowledge base comprises a plurality of network strategies, and the similarity between each candidate network strategy in the at least one candidate network strategy and the strategy request information is greater than a first threshold value; determining a policy score of each candidate network policy in the at least one candidate network policy based on the policy request information, the at least one candidate network policy and a reordering model; based on the strategy score of each candidate network strategy, determining a target network strategy meeting a matching degree requirement from the at least one candidate network strategy; the target network strategy comprises a strategy name and API information. The method and the device are applied to a network policy matching process.

The present application relates to the field of artificial intelligence, in particular to a case retrieval method based on multiple models. Various multi-source data are collected and integrated; the semantic dependency relationship between the text fragments of the contradiction is captured by using Bert, and the similarity of the case data is calculated by using the BM25 algorithm; a model is trained by Bert according to twelve types of classifiers, and at the same time, a model is retrained for the overall similarity model training. The similarity model constructed by combining the case classification model in the prior art and the present application completes the accurate case retrieval of diversified and long-length contradiction dispute data. The legal case data of the present application is more specific and perfect, contains sufficient legal knowledge, can cope with the change of legal rules, the unification of diversified legal data, accurate collection, efficient auxiliary analysis of legal cases and resolution work. The present application is more accurate in calculating the similarity, and the relevant category of legal cases is recommended accurately.

This invention discloses an image restoration method, comprising: acquiring a target image, a set of reference images, and a boundary intensity map; extracting target features, a list of reference features, and boundary features; based on the extracted features, performing similarity modeling using a learnable spiral path to generate a similarity model result; based on the similarity model result, performing geometric alignment and dual-domain fusion on the list of reference features to generate initial fusion features and alignment parameters; calculating an error pattern map based on the initial fusion features; performing closed-loop feedback according to the error pattern map to iteratively update and generate final fusion features; and decoding the final fusion features to obtain the restored image. This invention improves the structural consistency and detail fidelity of the restored image under non-rigid deformation through boundary-aware spiral sampling, non-folded alignment, and closed-loop feedback.

The application discloses a reinforcement learning system and method for high dynamic decision scene agent exploration, which comprises a positive actor-critic model, a negative actor-critic model, an actor-critic model, a strategy similarity model and a strategy action selection model. The learning of a main strategy is guided by using a positive strategy and a negative strategy, and the closeness of the main strategy to the positive strategy is minimized and the closeness of the main strategy to the negative strategy is maximized, so that the main strategy generates high-value actions with a high probability and generates low-value actions with a low probability, and the overall performance of the method is improved. Meanwhile, in the training process, the strategy action selection model is used to select the main strategy action, the positive strategy action and the negative strategy action in the form of probability, the diversity of action selection is increased, the diversity of the exploration environment is indirectly improved, and finally the performance is improved.

The application belongs to the field of tourism recommendation, and particularly relates to a similarity measurement method and system for complex landscapes, which comprises the following steps: obtaining multi-modal data time series of garden landscapes from a landscape training data set and converting the multi-modal data time series into time series vectors; training a feature extraction model according to a multi-modal time series data matrix to obtain an optimal feature extraction model; obtaining an optimal global vector by using the optimal feature extraction model and the multi-modal time series data matrix, and performing decoupling mapping to obtain an attribute vector; inputting the attribute vector and the training data set into a similarity model to optimize the similarity model and obtain an optimal similarity model; inputting actual landscape data into the optimal feature extraction model to obtain an actual global vector; and inputting the actual global vector into the optimal similarity model to obtain the similarity of different landscapes. The application can obtain more accurate similarity between gardens, provide reasonable basis for tourists' travel route planning, and thus better plan the travel route.

The invention relates to a similarity calculation method of a affair graph and a related product. The method comprises the following steps: acquiring a plurality of affair maps; based on different types of information similarity models, respectively analyzing the similarity of the target information between the affair maps to obtain different types of information similarity features; on the basis of domain weights, the information similar features are fused, associated features are obtained, and the domain weights are determined according to domain knowledge; fusing the weight of each matching edge under the maximum weight matching of the bipartite graph to obtain a graph similarity; the bipartite graph is determined according to the associated features. According to the scheme provided by the invention, the domain weight is determined through the domain knowledge, and the fusion process of the information similar features is guided to improve the influence of key elements in the information similar features on the associated features, so that the related parameters of the bipartite graph more accurately reflect the similarity degree between the two affair maps, and accurate calculation of the map similarity is facilitated.

The invention relates to a sensitive information identification method and device, equipment, a storage medium and a program product. The method comprises the steps of obtaining a to-be-identified field, and performing first similarity calculation on the to-be-identified field and a plurality of pre-stored sensitive label vectors to obtain candidate sensitive label vectors; and inputting the candidate sensitive tag vector and the to-be-identified field into a preset sub-domain similarity model for second similarity calculation to obtain an identification result of the to-be-identified field. The high-dimensional vector representation formed by different domain vectors of the sensitive tag is pre-stored, and the first similarity calculation is performed on the to-be-identified field and the pre-stored sensitive tag vectors, so that the identification efficiency of the sensitive information is improved; according to the method, the candidate sensitive label vector and the to-be-identified field are simultaneously input into the preset domain identification model for similarity calculation, and the candidate sensitive label vector is used as a vector form, so that a coding processing flow is avoided, the complexity of online calculation can be reduced, and the identification accuracy of sensitive information is improved.

This invention discloses a novel intelligent discovery and evaluation method and system for sensitive information based on multi-dimensional features, belonging to the field of smart city network security technology. The invention employs unsupervised new word mining with dual feature filtering, dynamic thresholding, and long word segmentation verification, combined with AC automata and three-level quality grading. Sensitive word variant determination constructs a five-dimensional fusion representation of characters and components, overcoming the limitations of single matching. Sensitivity analysis builds a two-layer labeling system based on fine-tuned RoBERTa, achieving fine-grained quantification through dual-task training and triple dynamic correction. Sensitive topic mining introduces seed word-guided clustering and incremental update algorithms. Phonetic similarity matching constructs a multi-granularity speech similarity model and a dynamic fuzzy sound rule library. AI comprehensive discrimination integrates features from the entire process, constructing dynamic weights and a structured Prompt framework to achieve interpretable judgment and cross-module feedback optimization.

This invention proposes a parametric design method for escalator truss structures, comprising the following steps: Step S1, establishing a parametric truss model; Step S2, writing an automated modeling program; Step S3, creating a parameter input interface. This invention improves the modeling efficiency of escalator trusses, reduces manual workload in the modeling process, and simultaneously meets the need for efficiently establishing a large number of similar models in the truss optimization design process.

The present application relates to a kind of method and device suitable for text recognition error correction, the method is by obtaining first text list;The target text to be identified is input to text detection model, and the target text recognition result is obtained;The first text content in the target text recognition result with confidence higher than the preset threshold is supplemented to the first text list, and second text list is obtained;The target text recognition result is input to training glyph similarity model, and glyph similarity function is obtained;Second text content in the target text recognition result with confidence lower than the preset threshold is corrected, can the recognition result with confidence higher than threshold be added to existing non-full text list to obtain full text list, can the error generated by text recognition be accurately corrected.

The application discloses a part model matching method and device, computer equipment and a storage medium. The method comprises the following steps: obtaining a to-be-matched contour machining time length corresponding to a to-be-matched part model and a to-be-matched element parameter; performing similar element processing on the to-be-matched element parameter and an element parameter corresponding to a part model to determine a similar element value corresponding to the part model; performing similarity processing based on the to-be-matched contour machining time length, the to-be-matched element parameter and the similar element value to determine a similarity value corresponding to the part model; and determining a target part model matched with the to-be-matched part model based on a preset rule according to the similarity value. The method of the application can accurately calculate the matching similarity value between part models, solves the problem that it is difficult to check similar models in a large number of models, and can quickly match and calculate the part model instance with the highest part similarity from the part model library.

The application discloses a large-aspect-ratio flying wing layout aircraft rigid-elastic coupling aeroelastic scaling model design method. The method is based on the rigid-elastic coupling aeroelastic equation, the key parameters to be matched are determined through dimensionless, including inertia ratio, reduced frequency and Froude number, and length, speed and density are used as basic scales to derive target scaling parameters such as mass / inertia, structural stiffness and modal frequency. When the flow similarity is difficult to meet, the iterative optimization strategy of lift and moment coefficient error minimization is introduced, and the speed ratio and mass / inertia scaling constraints are coupled into the structural arrangement variable optimization to realize the coordination of static / dynamic aeroelastic compatibility. The model design adopts the spar-pod structure system of'stiffness beam + external shape', and the mass / inertia and modal matching are completed by adjusting the beam section and counterweight, and the flutter critical speed, frequency and shape are used as the similarity verification indexes. The method can be used for early evaluation and verification of large-aspect-ratio flying wing aeroelastic phenomena.

This invention provides a dynamic load similarity model test system and related methods and equipment, belonging to the field of civil engineering testing technology. The system includes: a box frame structure, a pavement-underpass tunnel similarity model, a vibrator module, and a data monitoring and acquisition module. The pavement-underpass tunnel similarity model is constructed from similar materials within the box frame structure based on similarity theory. The vibrator module is used to simulate and apply dynamic loads on the pavement and / or vibration loads from the underpass tunnel. The data monitoring and acquisition module is used to collect dynamic response data of the structure under the relevant loads and load value information of the two types of loads. This invention can provide a reliable physical test verification method for studying the dynamic response characteristics of pavement structures and underpass tunnel structures under the coupled influence of pavement dynamic loads and underpass tunnel vibration loads, while taking into account ease of implementation and cost.

The application discloses a kind of coal seam roof well upper and lower area fracturing physical similarity simulation experiment method, comprising the following steps: one, according to the field geological conditions and experimental purpose to build physical similarity model;Two, according to coal seam roof well upper and lower area fracturing engineering practice, thick hard target disaster-causing rock layer is expanded and fractured using static breaking agent;Three, according to the similarity ratio of experimental parameters, excavate area fracturing physical similarity model and carry out data monitoring.The experimental method of the application is mainly based on the similarity of rock mass cracking mechanism and fracture propagation morphology after fracturing, without considering the hydration reaction of rock mass, the target disaster-causing rock layer is expanded and fractured using static breaking agent, and the fracturing weakening effect of fracturing on rock layer is simulated, which has good implementability in laboratory, and the simulation is closer to the real working condition, which improves the accuracy of the experiment.

The method for generating failure prediction data of an X-ray tube includes S10 to S40. And S10, collecting the time sequence data of the referenced X-ray tube, and extracting the extracted data of the referenced X-ray tube from the time sequence data of the referenced X-ray tube. And S20, training a similarity model, wherein training features are extracted data of the referenced X-ray tube. And S30, inputting the extraction data of the to-be-predicted X-ray tube obtained by adopting the same method as the S10 into the model trained in the S20 to obtain a model output result. And S40, analyzing according to a result output by the model in the S30 to obtain failure prediction data of the to-be-predicted X-ray tube. The method for generating the failure prediction data of the X-ray tube is beneficial to improving the accuracy of failure prediction of the X-ray tube. In addition, the invention also provides a system for generating the failure prediction data of the X-ray tube and a computer readable storage medium.

This invention belongs to the field of sensor and machine learning technology. It proposes a simulation test system and method for the mechanical properties of rock joints based on the principle of equivalent body force field. The main components include: constructing a geomechanical similarity model based on the principle of equivalent body force field, including a base and a rock mass physical model. When the base moves relative to the interface of the rock mass physical model, the friction between the base and the bottom surface of the rock mass physical model generates uniformly distributed shear stress, which generates equivalent body force intensity inside the rock mass physical model; collecting multidimensional sensing data when the base moves relative to the interface of the rock mass physical model; automatically identifying the rock mass joint extension trajectory based on the multidimensional sensing data and using an improved U-Net machine learning model; obtaining displacement field data through an improved particle image velocimetry combined with an optical flow algorithm; using a Bayesian inference inversion model to invert the mechanical parameters of the joint surface; and evaluating the uncertainty of the inversion results using the Markov chain Monte Carlo method.

The invention discloses a runoff forecasting method and system based on signal decomposition and similarity analysis, a medium and equipment in the technical field of runoff forecasting. The method comprises the following steps: performing signal decomposition on an obtained original runoff time sequence to obtain a signal component set; according to the signal component set, runoff forecasting is carried out by utilizing a trained similarity model, and a runoff forecast value of each signal component is obtained; and performing linear superposition on the runoff forecast values of the signal components to obtain a final runoff forecast value. According to the method, similarity forecasting and linear superposition are carried out after modal decomposition is carried out on the obtained original runoff time sequence, and the forecasting precision is improved compared with a traditional similarity forecasting method.