33results about How to "Reduce the amount of real-time calculation" patented technology

The invention discloses an optimal two-vector combination-based model predictive control method and system. The method is applied to a permanent-magnet synchronous motor control system driven by a three-phase two-level inverter. A model predictive current control strategy is adopted, all two-vector combinations and resultant vector sets obtained under action time are considered at the same time, and a cost function is inspected and an optimal resultant vector is selected from all to-be-selected sets. In order to simplify the optimization process, an equivalent voltage equation is given, a sector transformation method is provided and to-be-selected two-vector combination sets are transformed into multiple fixed line segments; and a fast algorithm is given and partial complicated calculation is transformed to offline execution, so that the real-time calculated amount of the novel method is effectively reduced. The model predictive control method disclosed by the invention is simple in structure, small in real-time calculated amount and easy to implement; and the response speed of a motor is high, current ripples and distortion are small, the switching frequency is low and the dynamic and static performance of the system is excellent.

The invention provides a human bod motion animation generating method of a two-dimensional virtual figure. The human body motion animation generating method comprises the steps of shooting a real person motion video; extracting an initial frame image of the video; detecting the angular point of the initial frame image; calculating a sparse characteristic point light stream frame by frame; calculating 2D grid deformation; calculating a modal grid and a weight sequence; editing the modal grid and the weight sequence; using an isomorphic 2d grid to conduct animation rendering on the virtual figure. According to the technical scheme, compared with a purely manually made animation, since motions come from the real person video, the animation is more vivid, and particularly, it is very difficultto achieve the authenticity of the deformation and the rhythm randomness of the motions through a manual method; the 2d figure has a certain stereoscopic effect, the animation effect can be adjustedaccording to requirements through manual modification simply with the isomorphic grid, the same motion data can adapt to different figures of different body types, different skins, different clothes and different icons, and when the animation is applied, the real-time calculation amount is small.

The invention discloses a method for displaying a real-time shadow in virtual reality. The method comprises the following steps: key time points are selected according to illumination change of a scene; shadow values are calculated for the key time points, a shadow map is obtained, and frames containing the shadow map are taken as key frames corresponding to the key time points; the current time point is determined when the real-time shadow is displayed, two key frames corresponding to the two key time points before and after the current time point are read, interpolating calculation is performed on the shadow map in the two key frames, and the real-time shadow map of the frame corresponding to the current time point is obtained, wherein an interpolation for the interpolating calculation is determined according to the interval between the current time point and the key time points before and after the current time point. Consumption of computing resources is reduced while the real-time shadow effect can be realized.

The invention relates to a simulation method of a double four-quadrant converter. The double four-quadrant converter comprises a converter and a middle circuit, wherein the converter consists of four parallel branch circuits. The method comprises the following steps of: constructing a corresponding state equation according to different conducting states of switch elements in the four-quadrant converter; constructing a state equation judging module of a simulation model of the double four-quadrant converter by using a Simulink module; constructing an adams resolving module of the simulation model of the double four-quadrant converter by using the Simulink module; and performing simulation calculation by using the double four-quadrant converter. A modeling method has an intuitional principle, simple implementation steps, small real-time calculated amount and more correct and quick simulation, is easy to understand and is beneficial to instantaneity of simulation, and the convergence during zero pulse is effectively ensured.

The invention discloses an automatic placement method for building annotations in a three-dimensional scene. According to the method, a candidate annotation set containing fixed-orientation annotations and dynamic-orientation annotations is constructed for each building model in a preprocessing stage according to geometric morphological characteristics of a three-dimensional building, and optimalannotations are screened according to an annotation evaluation algorithm related to viewpoints in an interactive browsing stage. And when the annotations are projected to the screen space conflict processing, the annotations are sequentially placed by adopting a greedy algorithm according to the global priorities of the annotations. Compared with the prior art, the method has the advantages that the real-time calculation amount is small, the dynamic change of the visual angle, the visual range and the visibility can be adapted, and the relevance between annotations and building elements is effectively kept. Besides, the annotations are sequentially placed according to the priority order through the greedy algorithm, the annotations which are close to the viewpoint, low in conflict degree and high in building significance can be reserved while screen space annotation conflicts are efficiently processed, and the readability of global annotations is improved.

The embodiment of the invention provides a video signal processing method and device and a storage medium, and the method comprises the steps: carrying out the preset processing of a to-be-processed video signal in an original format, and obtaining a first linear RGB signal; obtaining a target pixel value corresponding to a pixel value of each pixel point in the first linear RGB signal according to a preset mapping table, and obtaining a second linear RGB signal according to the target pixel value, wherein the mapping table comprises a mapping relation between the pixel value based on the original format and the pixel value based on the target format; and performing inverse processing opposite to the preset processing on the second linear RGB signal to obtain a target video signal in a target format. According to the technical scheme, the mapping table is created in advance, in the process of converting the to-be-processed video signal from the original format to the target format, the real-time calculation amount is small, the processing time of the video signal is short, and the real-time conversion effect can be achieved.

The invention belongs to the field of power transmission line maintenance, and particularly relates to a method for determining the division number of visual alarm regions of a power transmission linechannel, which comprises the following steps of: a, performing offline training on a model based on visual alarm data of the power transmission line channel to obtain a suggested value; b, determining a determination value of the power transmission line channel visual alarm area division number through online decision. According to the method, the division number of the visual alarm regions of the power transmission line channel can be automatically determined, the visual alarm high-incidence regions of the power transmission line channel are identified based on the determined value, the coverage radius of the found visual alarm high-incidence regions conforms to the inspection radius of power transmission line maintainers, and convenience is brought to the power transmission line maintainers.

The invention discloses a geomagnetic indoor positioning method based on a gated recurrent neural network and particle filtering. According to the method of the invention, the gated recurrent neural network is trained through a built geomagnetic indoor database to perform matching and positioning of geomagnetic track signals;the trained gated recurrent neural network is used for matching and positioning, so that better positioning precision can be brought to matching and positioning of the geomagnetic track signals. Compared with a common geomagnetic track signal matching algorithm based on dynamic time planning, the trained model reduces the real-time calculation amount in the matching and positioning process. According to the method of the invention, a system for performing real-time positioning is designed while a particle filtering algorithm is adopted on the basis of performing matching and positioning on the geomagnetic track signals by the neural network model. According to the system, the advantage of extracting geomagnetic track signal features by the gated recurrent neural network is effectively utilized, and better real-time positioning precision is brought to the particle filtering algorithm.

The invention relates to a simulation method for a single four-quadrant converter, which comprises a converter and an intermediate circuit, wherein the converter consists of four branch circuits which are connected in parallel. The method comprises the following steps of: constructing corresponding state equations according to different conducting states of each switch element in the four-quadrant converter; constructing a state equation judgment module of a simulation model of the single four-quadrant converter by adopting a Simulink module; constructing an adams resolving module of the simulation model of the single four-quadrant converter by adopting the Simulink module; and performing simulation calculation by utilizing the simulation model of the single four-quadrant converter. The modeling method of the invention has visual principle, simple realization step, small real-time calculated amount and accurate and quick simulation, is easy to understand, is favorable for the instantaneity of the simulation, and effectively ensures the convergence during zero pulse.

The invention discloses a method for implementing CIM (common information model) system development by combining BIM (building information modeling) and GIS (geographic information system) technologies. City-level standardized coding is applied in the design stage of a BIM model, data containing GIS information is coded into the BIM model, and model component set information is uploaded to various data storage interfaces on an online storage platform through a network in a data analysis mode when the BIM model is delivered. The information is stored in a database storage mode according to the unique identifier of the BIM model, GIS grid coordinates and other information in different regions and types, and meanwhile, a query service interface is provided for a GIS software system. The engineering data obtained by splitting is stored in a classified manner by using a database method, and can be provided for a third-party CIM system to carry out engineering visual presentation and data reutilization based on geographic position information, so that the technical barrier between the GIS and the BIM is broken through, and the technical threshold of the industry is reduced.

The invention provides a train axle box bearing health state early warning method and system. The method comprises the steps that at least two characteristic parameters of m axle box bearings in the same train are collected in real time, and m is a positive integer; calculating a characteristic index of at least one characteristic parameter of each axle box bearing; respectively calculating a mean value of all feature indexes and/or feature parameters at the current moment; calculating a deviation value of each characteristic index and/or characteristic parameter of each axle box bearing at the current moment according to the mean value; judging whether the deviation value of the characteristic index and/or the characteristic parameter of each axle box bearing exceeds the deviation threshold value of the corresponding characteristic index or the characteristic parameter or not; if a certain axle box bearing exceeds the characteristic index of the deviation threshold value and/or the variety number of the characteristic parameters reaches a preset value, health early warning is sent to the axle box bearing. The method effectively solves the problem that accurate early warning of the train axle box bearing health state is difficult.

The invention relates to the technical field of robot vision, and particularly discloses a data fusion method, device and equipment for an event camera and a standard camera, and a medium, and the method comprises the following steps: dividing depth image information into a plurality of pieces of point cloud block information according to the resolution of event information; obtaining forward probability information and backward probability information according to the probability density of the point cloud block information and the event information, wherein the forward probability information and the backward probability information are the forward probability and the backward probability of the event information migrated to the position of the point cloud block information respectively; obtaining correlation information of the point cloud block information and the event information according to the forward probability information and the backward probability information; obtaining conditional probabilities of the event information and the point cloud block information according to the correlation information and a total probability formula; fusing the point cloud block information and the event information according to the maximized conditional probability; the method has the characteristics of high matching precision, small calculation amount and high fusion efficiency.