64 results about "Model synthesis" patented technology

A system and method are provided which dynamically adapts to a changing economic environment by selecting or synthesizing an economic model from a set of economic models based on the selected model's ability to make accurate predictions about an actual economic market. The method and system each forms a space of different economic models, forms a behavioral landscape by extracting observables from executions of the economic models, and performs model selection and composite model synthesis through optimization over the behavioral landscape.

The invention discloses a multi-model integrated flood forecasting method based on propagation time clustering analysis and belongs to the technical field of hydrologic forecasting. The method includes adopting a derived dynamic time warp matching method for floor process similarity analysis; estimating flow propagation time of each station in the upstream and the downstream; decomposing a sample into a plurality of clusters by performing clustering analysis on the flow propagation time; respectively building an SVM regression model for each sub-flow sequence to simulate the flood forming process; combining the sub-models into a comprehensive model. Comparing a comprehensive forecasting result acquired by the method with a model forecasting result acquired by a single model under conventional conditions and on the basis of flow clustering, the comparison result shows that the comprehensive model is better in comprehensive performance.

An approach and tool integrate cyber-physical systems design based on the function-behavior-state (FBS) methodology where multi-domain simulation models capturing both the behavioral and structural aspects of a system are automatically generated from its functional description. The approach focuses on simulation-enabled FBS models using automatic and context-sensitive mappings of functional basis elementary functions to simulation components described in physical modeling languages. Potentially beneficial process loops are recognized and inserted in the functional model.

The invention discloses a comprehensive optimization method of a simulation credibility evaluation model of an electric propulsion system. The comprehensive optimization method comprises the following steps: decomposing the electric propulsion system into multiple subsystems; decomposing the subsystems under various standards of a simulation system with an establishment process of system simulation as an evaluation standard until obtaining a minimum unit; carrying out topology modeling on the electric propulsion system; collecting branch voltages, a phase angle, active power and passive power of the electric propulsion system; adopting corresponding intelligent algorithms to optimize different attributes of the electric propulsion system to obtain a credibility evaluation optimization model; collecting data of an actual system under the same working condition, and using the credibility evaluation optimization model to test the actually collected data to obtain credibility; if the credibility is smaller than a credibility index n, carrying out topology modeling on the electric propulsion system again; and if the credibility is larger than the credibility index n, the credibility evaluation is terminated. The comprehensive optimization method disclosed by the invention is high in stability and is high in reliability.

The invention discloses a fire image detection method based on a mixed color model and a neural network. The method comprises the following steps: firstly, performing an inter-frame difference methodon a monitoring video image to obtain a moving foreground region; extracting an area conforming to flame color characteristics through the mixed color model, and performing secondary color area expansion to obtain candidate areas filtered by the color model; taking a union set of the motion foreground area and the color candidate area to obtain a suspected flame area; and finally, constructing a dense connection convolutional neural network, and comprehensively judging whether a suspected flame region has a fire behavior or not through the trained network model and the secondary mixed color model. According to the method, the flame candidate area with high precision is obtained through motion detection and the mixed color model, the detection speed is high, the deep neural network guarantees the flame detection accuracy and the generalization capability, the method can be widely deployed in an actual intelligent security system, fire hazards are early warned, and losses caused by the fire hazards are reduced.

A speaker recognition method through emotional model synthesis based on Neighbors Preserving Principle is enclosed. The methods includes the following steps: (1) training the reference speaker's and user's speech models; (2) extracting the neutral-to-emotion transformation/mapping sets of GMM reference models; (3) extracting the emotion reference Gaussian components mapped by or corresponding to several Gaussian neutral reference Gaussian components close to the user's neutral training Gaussian component; (4) synthesizing the user's emotion training Gaussian component and then synthesizing the user's emotion training model; (5) synthesizing all user's GMM training models; (6) inputting test speech and conducting the identification. This invention extracts several reference speeches similar to the neutral training speech of a user from a speech library by employing neighbor preserving principles based on KL divergence and combines an emotion training speech of the user using the emotion reference speech in the reference speech, improving the performance of the speaker recognition system in the situation where the training speech and the test speech are mismatched, and the robustness of the speaker recognition system is increased.

The invention relates to the field of software monitoring of rack operation and maintenance, in particular to a model fault detection method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a stability index and a correctness index of a model to be evaluated and a stability index and a correctness index of a model variable in the model to be evaluated; Obtaining a model score according to a to-be-evaluated model, the stability index and the correctness index of the model variable and a preset fuzzy judgment matrix; Obtaining a model scoring weight parameter; And obtaining a model comprehensive score according to the model score and the model score weight parameter, and when the model comprehensive score is lower than a preset threshold, determining that the model has a fault. According to the model evaluation, the score of the overall health degree of the model is obtained by comprehensively considering various parameters such as the stability and correctness of the model and the stability and correctness of variables. a comprehensive evaluation result of the model health degree is output based on the comprehensive evaluation system so as to judge whether the model is in a normal state.

The invention relates to a high-intensity focused ultrasound spherical phased array with array elements annually and intensively distributed, comprising a rigid spherical crown body and a plurality of array elements, wherein one array element is independently arranged in the center of the spherical crown body, and the other array elements are evenly and dispersedly distributed on a plurality of layers of concentric rings which take the center of the spherical crown body as an original point; the angular spacings among all ring layers are not equal, and the spacings among all ring layers and the spacing between adjacent array elements in the same ring are set according to an array element intensive distribution principle; the position of the original array element of each layer of concentric rings is randomly set, the other array elements are subsequently, intensively and evenly distributed on the concentric rings according to equiangular spacing, and each array element has independent electric excitation signal feeder lines which are respectively connected to a phased signal excitation system. The invention calculates the phase position and the voltage amplitude distribution of the array elements by adopting a focus model synthesis method based on a pseudo-inverse matrix theory, realizes acoustic beam electron focusing and scanning, generates required single-focus or multi-focus sound field control modes, and the like and is applied to heating and fusing target tumor focuses in the human body.

The invention discloses a method for modeling an overload performance evaluation model of a transformer in consideration of multi-factor, comprising the following steps of collecting data based on a scheduling operation and production management information platform; establishing a model based on a multi-factor index; evaluating the performance of the model; utilizing a small value K between a determined maximum load rate K1 and a determined maximum load rate K2, and inputting K as an independent variable into economic loss and economic income formulae in an economic profit and loss model to realize overload performance evaluation of the transformer from the angle of economic profit and loss. The model established by the invention comprehensively considers the multi-factors, which influence the overload performance of the transformer, such as hot-spot temperature limitation, failure rate limitation and economic profit and loss indexes; and compared with an existing overload evaluation model, the constraint conditions are comprehensive. Compared with a traditional heat conduction model, the precision for evaluating the hot-spot temperature aspect of the transformer can be effectively improved in the improved hot-spot temperature model.

The invention relates to a power system transient stability evaluation method based on a short-time disturbed trajectory after fault removal. The method is characterized by comprising the following steps: firstly, selecting the short-time disturbed trajectory of the electric quantity at the end of the generator after the fault is removed, and secondly, searching the characteristic arrangement modeof the short-time trajectory when the evaluation accuracy is optimal, so that the local characteristic extracted by the model is more robust, and optimizing the parameters of the network window by taking the optimal comprehensive evaluation index of the model as the target, so that the generalization ability of the model is enhanced, and finally establishing the mapping relation between the short-time trajectory and the transient stability, so that rapid and accurate transient stability rapid evaluation of the power system can be realized. According to the method, misjudgment missing samplesof model evaluation can be effectively reduced, and compared with a traditional machine learning evaluation method, the method is more accurate and efficient, and the transient stability can be evaluated based on the short-time disturbed trajectory, and a certain time margin is reserved for a dispatcher to take a control measure.

Converting from transaction level model to register transfer level. A method comprises accessing a transaction level model description. Function calls in the transaction level model description are automatically converted to signal accesses in a register transfer level description. Interlace logic is automatically generated in the register transfer level description. The interface logic provides an interface between a bus interface and the signal accesses. The register transfer level description is stored in a computer readable medium.

The invention discloses an augmented reality-based dressing-up and try-on system used for game role play. The augmented reality-based dressing-up and try-on system comprises a camera shooting unit, an equipment costume model forming unit, a live model forming unit, a model synthesis unit, a game scene rendering unit and a display unit. The augmented reality-based dressing-up and try-on system used for game role play has the advantages that an equipment costume three-dimensional model and a human body three-dimensional model are synthesized into a three-dimensional effect model, a game player can beneficially learn the effect after game costume is tried on and game equipment is equipped, and an interest of a game player on a reality simulation role in a game playing process is raised; and by combining the game scene rendering effect with the three-dimensional effect model together, the simulation dressing-up and try-on three-dimensional effect is enhanced and is displayed by virtue of the display unit, and the game player can beneficially and visually know and simulate the real effect of a corresponding game role, so that the game player has more immersive game experience feeling, game entertainment excitement can be more beneficially improved, and higher requirements of the game player on game experience can be met.

The invention discloses a virtual fitting system based on augmented reality, comprising a first camera unit, a clothes model forming unit, a human body model forming unit, a model synthesis unit, a live-action rendering unit, a second camera unit, and a display unit. The output end of the first camera unit is connected to the input ends of the clothes model forming unit and the human body model forming unit. The output ends of the clothes model forming unit and the human body model forming unit are connected to the input end of the model synthesis unit. The output ends of the model synthesis unit and the second camera unit are connected to the input end of the live-action rendering unit. The output end of the live-action rendering unit is connected to the input end of the display unit. By synthesizing a clothes 3D model and a human body 3D model into a 3D effect model, a person can know the corresponding effect of clothes that the person tries on, and the fitting time is reduced. By combining the live-action rendering effect and the 3D effect model, the reality effect of virtual fitting is enhanced.

The invention relates to a model synthesis method, in particular to a global optimization-based three-dimensional model synthesis method, which utilizes the structural information of a three-dimensional model sample and a global optimization algorithm to synthesize various target three-dimensional models similar to the initial model. The method first utilizes a model structure analysis method to split the three-dimensional model sample and obtain the class number of each spatial block, then establishes a global model synthesis energy equation on the basis of the constraint relation between the class numbers obtained from the three-dimensional model sample, and adopts an expectation maximization algorithm for iterative optimization to work out a global energy minimum, and the final result is a synthesized model. The method can effectively keep the continuity of target three-dimensional model structures and the integrity of the visual effect.

The invention discloses a traffic flow prediction method fusing spatial and temporal features. The method comprises the following steps: step 1, preprocessing data; step 2, introducing an automatic encoder to obtain data characteristics; step 3, introducing an SAEs model and acquiring spatial feature ; step 4, introducing an LSTM model, and obtaining time features; step 5, synthesizing the SAEs model and the LSTM model to obtain an ideal hybrid model, and establishing a hybrid deep learning model SAES-LSTM to predict the traffic flow of an urban expressway. Time and space information is comprehensively utilized. The collected information of a database is analyzed and utilized more fully, and therefore a prediction result can be more accurate.

The two-dimensional coordinates of a vertex are extracted in each of a top view and front view and, if their X-coordinates are equal to each other, the combination of their Y-coordinate values is determined to be the two-dimensional coordinates of a candidate vertex in a side view. Then, candidate line segments for the side view are extracted from the line segments connecting two candidate vertices, excepting not only those line segments for which no corresponding line segment exists in the top and front views, but a so those line segments for which corresponding horizontal or vertical line segments exist in the top and front views, and which a e not horizontal or vertical in the side view.

The invention provides a human body scene image eigen-decomposition and relighting method and device, and the method comprises the steps: making a human body model through employing Clo3d, carrying out the light tracing through employing blender, and carrying out the rendering, and obtaining a training data set; pre-training a normal prediction network by using a single image; taking the original image and the normal graph obtained through prediction as input of an image eigen-decomposition network, and obtaining a reflectivity graph, a shadow graph and spherical harmonic illumination parameters through prediction; re-illuminating the original image by using the new spherical harmonic illumination parameter; and finally, decomposing a single picture into a reflectivity picture and a shadow picture, so the application of picture relighting, 3D model synthesis, model surface texture reconstruction and the like can be achieved. The shadow of the input picture can be removed in the field of three-dimensional reconstruction, and the surface texture quality of a three-dimensional reconstruction model is improved.

A system and method for predicting passive, high-frequency sonar array performance at any underwater location, implemented as a software package running on a computer. An operator selects an arbitrary geographical location, specifies a sonar array and defines a target. The necessary bathymetry, and associated oceanographic, environmental and meteorological data are automatically acquired or generated, along with appropriate models for ocean surface and bottom reflectivity, and vertical noise directionality. A Comprehensive Acoustic System Simulation (CASS) software module performs appropriate ray tracing, which is used to produce Cross-Spectral Density (CSD) matrices for the target and the anisotropic ambient noise. These are used to generate graphic and numeric representations of the array's performance at the selected location. A time of year may also be specified and the appropriate anisotropic noise field for conditions expected at that location at that time of year will be generated. Shipping realizations may be incorporated.

The invention discloses a four-step-coupled middle and long term hydrologic forecasting method. The four-step-coupled middle and long term hydrologic forecasting method comprises the following steps:step (1) forecasting: carrying out middle and long term hydrologic forecasting by adopting various models according to watershed hydrologic and meteorological data; step (2) carrying out error correction: carrying out the error correction on a forecasting result by adopting an AR (Augmented Reality) model, so as to obtain a corrected forecasting result; step (3) integrating various models: carrying out weighting processing on the forecasting result corrected by the various models according to errors of the corrected forecasting result of all the models, so as to obtain a comprehensive forecasting result; step (4) evaluating the reliability: based on the comprehensive forecasting result, quantitatively evaluating uncertain factors in a forecasting process by adopting a hydrologic uncertainty processor (HUP), so as to obtain an interval forecasting result under a certain confidence degree. The forecasting result can provide abundant uncertainty information, and more scientific and reasonable decision-making evidences are provided for water resource dispatching and configuration; the forecasting result has important actual meaning.

The invention discloses a time-based microblog document expansion method oriented to microblog retrieval, and relates to the technical field of microblog retrieval. The microblog document expansion method aims to solve the problem that the improvement of microblog retrieval performance is limited since a function on estimating a microblog document model by the document expansion can be weakened due to adverse effects brought to the document expansion by the short microblog. The time-based microblog document expansion method utilizes the time characteristic expansion document of a relevant microblog, and puts forward the time-based microblog document model. The model comprehensively considers the characteristics of the explosiveness, which is integrally presented on an aspect of time, of the relevant microblog and the adjacency, which is presented on an aspect of time, of an individual, utilizes the distribution of words on a microblog in an outbreak period and the microblog which is in near neighbor with the microblog in the outbreak period to obtain the weight of a document expansion word, and selects a query expansion word by a method based on machine learning so as to estimate an accurate document model. The method can better avoid influence brought to the document expansion by the short microblog.