18 results about "Visibility graph" patented technology

The invention relates to the field of robot path planning, in particular to a multi-robot path planning method based on group control, which comprises the following steps: on the basis of a visibility graph, generating a communication undirected weighted graph, introducing a Laplacian matrix, and reflecting the connectivity of the graph according to a second small feature value of the Laplacian matrix; the arrival time of the robots is controlled by adjusting the side weights, space-time conflicts are avoided, the robot path planning sequence is determined according to the contribution value of the second small feature value, and the overall performance of multi-robot cooperation is optimized; according to the method, data in robot path planning are integrated through the Transform model, the comprehensiveness of environmental understanding is improved, spatial constraints of the path are optimized through a message passing mechanism of the graph neural network, the robot dynamically adapts to environmental changes, the path is generated through the generative adversarial network, diversity and high efficiency are achieved, local optimum is avoided, and the method is suitable for being popularized and applied. And dynamic obstacle avoidance and real-time path optimization are realized through cooperation of multiple models.

The embodiment of the invention provides a wireless charging method for uncertain sensor node positions, and is applied to the technical field of wireless rechargeable sensor networks. The method comprises the following steps: actively detecting sensor node and obstacle distribution based on a transmitting coil array of the mobile charging equipment; a magnetic wave beam vector is formed by independently adjusting the voltage of each coil, and the accurate position of a node and the contour of an obstacle are determined by combining impedance measurement and wave beam intersection positioning; discretizing the region, constructing a vertex visible graph and generating a collision-free path complete graph; and calling a TSP solver to generate an initial charging path by taking minimization of movement consumption as a target, inputting the initial charging path to the trained reinforcement learning model, dynamically optimizing a track and a charging plan through real-time interaction, outputting an optimal path, and issuing the optimal path to the charging equipment. In this way, the problem of selecting a proper charging point in a complex scene with obstacles and uncertain node positions can be solved, accurate positioning of the nodes is achieved, and the charging efficiency is remarkably improved.

The invention discloses a rolling bearing fault diagnosis method, device and equipment and a storage medium, and belongs to the technical field of rolling bearings, and the method comprises the steps: generating a vibration signal sequence according to the vibration data of a rolling bearing; generating energy values of a plurality of local signal segments by using the vibration signal sequence; constructing a coarse-grained sequence by using the energy value of the local signal segment and the vibration sequence number sequence; converting the coarse-grained sequence into a sub-sequence set corresponding to each scale factor by using a preset scale factor set; establishing a weighted natural visibility graph for each sub-sequence in the sub-sequence set, and calculating a graph signal discrete entropy value of each sub-sequence based on the weighted natural visibility graphs; generating a vibration signal multi-scale discrete entropy spectrum feature by using the graph signal discrete entropy value of each sub-sequence; and inputting the vibration signal multi-scale discrete entropy spectrum features into the trained support vector machine model to obtain a fault diagnosis result. And high-precision and high-reliability rolling bearing fault diagnosis can be realized.

The invention discloses a city scene channel gain graph construction method based on mask constraint, which comprises the following steps: firstly, carrying out space division on a whole scene area, setting a minimum sample and a minimum verification / test lattice constraint, then generating a thick / thin wall binary mask based on a top view, carrying out pixel-level uniform sampling along a connecting line of any two points, and finally, carrying out pixel-level sampling on the pixel-level binary mask; on the basis, estimating a local scale by using KDTree, setting an adaptive radius, a maximum connecting edge distance, an out-degree interval and a soft radius bag bottom, constructing a visibility graph, defining a learning target in a residual domain by using FSPL as a baseline, obtaining stable and convergent path loss estimation, carrying out whole graph reasoning on grids, and fusing by using Gaussian / Hanning weight, so as to obtain a path loss estimation result; and then smooth convergence is carried out on results of multiple transmitters to obtain a continuous and consistent full-scene gain graph. According to the method, the high-resolution and high-reliability channel gain graph can be quickly generated in a complex city scene, and a theoretical basis and data support can be provided for network deployment and coverage analysis of a next-generation communication system.

The invention discloses an epileptic seizure detection system based on a double-branch space-time diagram neural network. The epileptic seizure detection system comprises an electroencephalogram signal collector, a signal preprocessor, a double-branch diagram constructor, a double-branch diagram neural network feature learning device, a feature fusion device and a classifier. Wherein the double-branch graph constructor comprises a time branch and a space branch; the time branch maps a down-sampled electroencephalogram sequence into a graph structure for each channel by adopting a horizontal visibility graph algorithm; the spatial branch constructs node features based on multi-scale visual graph network statistical features, and constructs edge weights based on dispersion indexes and cross-channel phase amplitude coupling. The double-branch graph neural network feature learner automatically learns time dynamic features and space correlation features from two types of graph structures, and aggregates multi-channel information through an attention mechanism. And the feature fusion device adopts an attention mechanism to carry out weighted fusion on the double-branch features. According to the invention, the accuracy of epilepsy detection is significantly improved by fusing spatial-temporal features and graph neural network automatic learning through a double-branch architecture.

The invention discloses a video stitching method and device based on optical flow interpolation. The method comprises the following steps: projecting a received multi-view video frame to a public cylindrical coordinate system to obtain a projection image under each view; for overlapped areas of the projection images under two adjacent visual angles, calculating bidirectional optical flow by using the optical flow estimation sub-network and then performing zooming, and performing optical flow interpolation on the two overlapped areas based on the zoomed optical flow to obtain a plurality of slice distorted frames; generating two refined optical flows and a visibility graph by using an optical flow refining sub-network based on the slice distorted frame and the scaled optical flow; calculating a fusion interpolation slice based on the refined optical flow and the visibility graph to generate a fusion image and cover the overlapping region to obtain a spliced video frame; and designing a composite loss function based on unsupervised learning to carry out network training. According to the method, ghosting and fracture artifacts at the video splicing boundary can be effectively eliminated, a high-quality panoramic video with time-space coherence is obtained, and the method is particularly suitable for wide-baseline camera deployment scenes such as automatic driving and security monitoring.

The application discloses a kind of based on multiple visibility map and the fusion of Transformer multi-element time series analysis method, comprising the following steps: S1, using sliding window mechanism converts each channel of multi-element time series into single layer visibility map, combined into multiple visibility map;S2, cross-channel consensus connection is extracted from multiple visibility map by AND aggregation mechanism, and consensus visibility map is generated;S3, with the adjacency relationship of consensus visibility map as constraint, the node embedding is iteratively updated using VG-Attention;S4, the final embedding is input into downstream module, and tasks such as prediction, classification are executed.The method can effectively capture global dependence, improve the performance of multi-element time series analysis.The method can effectively capture global dependence, improve the performance of multi-element time series analysis, and can be widely applied to physiological signal classification in medical diagnosis, weather data prediction, anomaly detection of industrial process monitoring and other multi-element time series analysis scenarios.

The invention discloses a civil aviation bird strike event generation characteristic analysis and prediction method based on a visual graph, and the method comprises the following steps: (1), capturing an original bird strike event sequence generation mechanism based on a visual graph theory, and mapping a bird strike event time sequence into a complex network; (2) identifying a bird strike event occurrence mode and a similarity node based on analysis of network topology characteristics of the bird strike event views; and (3) on the basis of considering bird strike event visibility graph network structure characteristics, proposing an improved gravitation model fusing network high-order structure information and node similarity measurement so as to mine similarities among different bird strike events, and constructing a prediction model fusing similarity information. According to the method, local high-order similarity dependence between time sequences is better captured based on the gravitation model defined by the similarity of the views, the provided prediction model has better interpretability and lower calculation complexity, compared with an existing method, the prediction precision is higher, and effective decision support is provided for analysis and prediction of a bird strike event occurrence mode.

The invention discloses a weak target detection method based on a weighted graph convolutional network, and the method comprises the steps: obtaining a radar echo sequence, and carrying out the preprocessing of the radar echo sequence, and obtaining a phase difference sequence corresponding to a short pulse subsequence; based on a visibility graph theory, constructing a weighted difference visibility graph and a corresponding weighted adjacency matrix and a feature matrix; a graph data set and a graph classification-oriented radar target classification model are constructed, the target classification model comprises a GCN model, a pooling layer of the GCN model is replaced by a graph pooling mechanism based on weighted edge shrinkage, and an output layer of the GCN model is set as a classifier with a controllable false alarm rate; training a radar target classification model by using the graph data set; and for a to-be-detected radar echo sequence, constructing a feature matrix and a weighted adjacency matrix for each distance unit of the to-be-detected radar echo sequence as a detection set, inputting to-be-detected samples in the detection set into the trained radar target classification model, and outputting a classification result of each to-be-detected sample.

The application relates to the technical field of ship path planning, and particularly discloses a ship path planning method and system for exploring potential water surface dangerous areas, which comprises the following steps: acquiring known dangerous areas and potential dangerous areas; generating observation points on the edges of each potential dangerous area to obtain a set of potential observation points; establishing a visibility graph, wherein the nodes of the visibility graph are the observation points, the edges are paths between two observation points which do not pass through the known dangerous areas and the potential dangerous areas, and the edge weight is the length of the path; solving a TSP problem with group constraints by using an improved ant colony algorithm on the visibility graph to obtain a shortest path covering all the potential dangerous areas, which is used as a ship planning path. The application realizes shortest path planning of a ship in a target sea area to bypass the known dangerous areas and traverse the edges of all the potential dangerous areas, and realizes safe and efficient shortest path planning.

To provide a method and a system for classifying a rolling bearing state based on a motif feature of a horizontal visual graph.SOLUTION: Converting the vibration signals of each type of rolling bearing into a horizontal visual graph, labeling nodes of the horizontal visual graph according to a time sequence to obtain n-point motifs, for the n-point motifs with m motif types, calculating an appearance frequency of each type of n-point motif, and obtaining a feature vector based on the appearance frequencies of all types of n-point motifs, extracting a feature element of a bearing vibration signal of a current fault type based on the feature vector, obtaining feature elements of bearing vibration signals of all known rolling bearing fault types, obtaining bearing vibration signals to be classified, and selecting a known rolling bearing fault type corresponding to a maximum value of similarities as a fault type of the bearing vibration signal to be classified.SELECTED DRAWING: Figure 1

The present invention discloses an apparatus for use by a pathfinding device, comprising at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: obtain a first set of nodes of a polygon indicating a first Permitted-Moving-Zone, PMZ, wherein, the first PMZ is not overlapping with any other PMZs; determine a group of Non-permitted-Moving-Zones, NMZs, that lies within the first PMZ or intersects with the first PMZ as its relevant NMZs; and determine a visibility graph for the first PMZ based on its relevant NMZs.

Provided is a method for optimizing searching of a previously viewed scene for a head-mounted display (HMD) device including receiving a plurality of views corresponding to a real-world scene, determining landmarks within each view, determining a co-visibility of the landmarks between different views, generating a co-visibility graph based on the co-visibility of the landmarks between the different views, the co-visibility graph including multiple levels, which include nodes that represent at least one view and edges between the nodes that represent the co-visibility of the landmarks between the at least one view, and detecting a pose of the HMD device based on the co-visibility graph of the real-world scene.

The application discloses a sea surface weak target detection method based on a weighted difference visibility graph feature, and comprises the following steps: collecting a sea surface echo amplitude sequence, converting the sea surface echo amplitude sequence into a phase difference sequence after dividing the sea surface echo amplitude sequence into subsequences, and performing normalization processing on the phase difference sequence; constructing a weighted difference visibility graph network by using the normalized phase difference sequence; extracting features from the weighted difference visibility graph network corresponding to each subsequence, and constructing a sea clutter map feature vector of the subsequence by using the extracted features; taking each sea clutter map feature vector as a training sample point to construct a training sample set; training a target detector by using the training sample set; for a sea surface echo amplitude sequence to be detected, taking a sea clutter map feature vector corresponding to the sequence as a to-be-detected sample point, and determining whether there is a target in the sea surface echo amplitude sequence to be detected by a relative position relationship between the to-be-detected sample point and the target detector.

The invention discloses an occluded fruit detection method based on deep learning and multi-modal fusion, and aims to solve the problem that the complete contour and size of a fruit are difficult to recover due to occlusion. The method comprises the following steps: constructing a three-dimensional Gaussian representation with multispectral and view-angle-related transparency through short-time multi-view acquisition and registration; according to the method, a bladeless view and a visibility graph are generated through semantic hierarchical rendering, and in combination with micro-robust ellipsoid fitting and thin-shell constraint, Gaussian parameters are reversely optimized through re-projection joint loss, so that the technical effects of more accurate estimation of the modal mask and the three-dimensional ellipsoid, less repeated counting, faster convergence and more stable grabbing posture in a re-shielding and backlight scene are achieved.

This invention relates to the technical fields of equipment condition monitoring and fault diagnosis and discloses a method and system for classifying rolling bearing conditions based on motif features of a horizontal visibility graph. The method includes the steps of converting the vibration signals of each type of rolling bearing into a horizontal visibility graph, labeling the nodes of the horizontal visibility graph according to time series to obtain n-point motifs, calculating the occurrence frequency of each n-point motif type for m motif types, and obtaining a feature vector based on the occurrence frequency of all n-point motif types, extracting feature elements of the bearing vibration signals of the current fault type based on the feature vector, obtaining feature elements of the bearing vibration signals of all known rolling bearing fault types, and obtaining the bearing vibration signal to be classified and selecting the known rolling bearing fault type corresponding to the maximum similarity as the fault type of the bearing vibration signal to be classified. The invention can accurately distinguish between single and multiple bearing faults, and different types and severities.

A kind of group vehicle trajectory prediction method based on multi-domain perspective, first, in frequency domain, model utilizes fast Fourier transform to convert historical trajectory into frequency domain representation, selects the frequency of maximum amplitude to capture potential periodic characteristics, combines stacked two-dimensional convolution network and LSTM to extract features in depth;Second, in spatial domain, construct weighted directed spatial topology graph, model dynamic spatial relationship by the relative position, speed and acceleration information between vehicles, use stacked graph convolution network with residual connection to extract spatial features;Then, in time domain, based on circular finite penetration visibility graph, construct time logic graph, by allowing cross-time step connection to capture time-dependent features, use residual stacked GCN to extract time domain features;Finally, by the gate recurrent unit encoding-decoding structure fusion multi-domain feature, realize the trajectory prediction of group vehicle, the future trajectory of group vehicle is high-precision predicted by the present application.

The invention provides a robot navigation method based on an FAR Planner global path planning algorithm, which comprises the following steps: if known map data exist, detecting whether a dynamic obstacle moves away, if so, performing navigation according to a local visual graph, updating the local visual graph to the map data when the navigation is finished, and if not, performing navigation according to the local visual graph; if not, navigating according to the historical trajectory data; if no known map data exist, navigation is carried out according to the local views, and when navigation is finished, the local views are updated to the map data. By adopting the method, the geometrical characteristics of the environment can be extracted in real time in a real environment, the local views are dynamically established to assist path planning and navigation, and meanwhile, the algorithm is only provided with a single lens, does not need extra hardware support and is easy to implement. And the local views and the navigation paths can be dynamically adjusted in real time according to real-time sensor input in an environment without a known map.