21 results about "Pattern space" patented technology

The invention relates to the technical field of industrial equipment fault monitoring and diagnosis, in particular to an industrial equipment fault prediction and diagnosis system based on deep learning, which comprises a dynamic data alignment module, a multi-dimensional feature fusion module, a fault mode decoupling module and an interpretable diagnosis mapping module. Wherein the dynamic data alignment module is used for constructing a multi-dimensional equipment state data stream under a unified time axis; the multi-dimensional feature fusion module is used for performing orthogonal decomposition in a feature space and extracting an independent feature vector set; the fault mode decoupling module is used for constructing a fault mode space projection template; and the interpretable diagnosis mapping module carries out constraint matching to generate a visual diagnosis report. According to the method, through dynamic alignment and multi-dimensional feature fusion of multi-source heterogeneous sensor data, fault mode decoupling and interpretable diagnosis mapping are combined, so that feature redundancy and noise interference are reduced, and meanwhile, accurate fault positioning and residual life prediction are achieved.

The invention provides a large-scale-oriented IPv6 target address generation method and a large-scale-oriented IPv6 target address generation system. The method comprises the following steps: constructing an IPv6 seed address set based on an IPv6 capacity expansion strategy of sampling distribution and an IPv6 classification strategy based on an address structure; clustering the IPv6 seed address set to construct an IPv6 address space tree so as to generate a low-dimension IPv6 address mode set and a high-dimension IPv6 address mode set; generating an IPv6 target address set in the low-dimension IPv6 address mode space, and removing IPv6 alias addresses in the IPv6 target address set to obtain a non-alias target address set; a non-alias target address set in a low-dimension IPv6 address mode space is detected, an active high-dimension IPv6 address mode is screened out from a high-dimension IPv6 address mode set according to the detected IPv6 active address, an IPv6 target address set is generated in a corresponding space of the active high-dimension IPv6 address mode, IPv6 alias addresses are removed, and the IPv6 address mode is obtained. After a non-alias target address set is obtained, detecting the non-alias target address set and collecting responded IPv6 active addresses; and combining the IPv6 active addresses detected in all the seed address sets to form a final IPv6 active address set.

The application provides a stereoscopic well pattern space stress field active utilization method and device, including obtaining a plurality of fracturing schemes of a target well pattern; performing fracturing simulation on the target well pattern according to the fracturing schemes to determine a fracturing structure of the target well pattern, wherein the fracturing structure represents the number and shape of cracks in the target well pattern; determining a stress field active utilization coefficient and a fracture network complexity coefficient of the fracturing scheme according to the fracturing structure, wherein the stress field active utilization coefficient represents the stress field active utilization effect in the fracturing simulation process, and the fracture network complexity coefficient represents the proportion of the crack area in the fracturing simulation process; and selecting an optimal fracturing scheme according to the stress field active utilization coefficient and the fracture network complexity coefficient to fracture the target well pattern. Through the above method, the most sufficient fracturing scheme using the space stress field can be determined, and the optimal fracturing scheme can be selected through the product operation to guide the fracturing of the real target well pattern.

A double patterning method of manufacturing select gates and word lines is provided in the present invention, including forming first string patterns composed of word line patterns and select gate patterns on a target layer, forming a conformal spacer layer on first string patterns, wherein the spacer layer forms trenches between first string patterns, forming a fill layer filling up the trenches on the spacer layer, removing fill layer outside of the trenches, so that fill layer in the trenches forms second string patterns, wherein the second string patterns and the first string patterns are spaced apart, removing exposed spacer layer, so that the first string patterns and the second string patterns constitute target patterns spaced apart from each other on the target layer, and performing an etching process using those target patterns as a mask to remove exposed target layer, so as to form word lines and select gates.

This invention provides a method, apparatus, device, and medium for pattern defect detection and repair. The method includes: preprocessing an input image to obtain a preprocessed image; extracting pattern units from the preprocessed image and calculating parameters for characterizing the pattern space and feature patterns; detecting and locating defects based on the feature vector of the standard unit and the average adjacent spacing; generating a repaired image block; and fusing the repaired image block with the original image. The aim is to completely solve the core pain points of traditional methods, such as strong reliance on manual intervention, high defect miss rate, poor repair effect, and weak scene adaptability.

An image encoding / decoding method, a bitstream transmission method, and a computer-readable recording medium storing a bitstream are provided. An image decoding method according to the present disclosure may be an image decoding method performed by an image decoding apparatus, comprising: constructing candidates for a spatial geometric partition mode (spatial geometric partition mode, SGPM), wherein each of the candidates includes a prediction mode for an intra prediction mode of each partition divided by a partition mode of the SGPM; and predicting a current block based on one of the candidates, wherein a prediction mode of a first candidate among the candidates is determined as information indicating an intra block copy prediction mode of a reference position.

The invention provides a rapid graph spacing screening method for chip design, which can rapidly complete distance check between graphs, reduce computing resources and improve layout design efficiency. Comprising the following steps: step 1, obtaining a graph needing to be checked, and screening out a graph of a specified type; step 2, combining repeated graphs which are in the same graph layer and have the same shape; 3, the current layer is divided into a plurality of rectangular areas with the length and the width being specified sizes, every two rectangular areas are arranged in a partial area overlapping mode, and logic and operation are conducted on the graphs screened in the same layer and each rectangular area; and step 4, performing center point distance calculation on all the graphs in the same rectangular area in pairs, reserving two graphs of which the distance is smaller than a set distance, and recording the positions and the distance of the two graphs so as to complete distance inspection between the graphs.

The invention provides a dynamic graph federal learning method for cross-line network urban rail passenger flow prediction, which comprises the following steps: preprocessing local automatic fare collection data of a client to obtain a local signal graph, and constructing and compressing an urban rail passenger flow dynamic space association graph of the client; local passenger flow mode space features of the urban rail passenger flow dynamic space association diagrams of all the clients are extracted and uploaded to a server for aggregation, so that a global urban rail network passenger flow dynamic space-time diagram is constructed, local model parameters of the clients are trained and optimized, and the local model parameters are uploaded to the server for aggregation; and according to an aggregation result of the local model parameters, the client updates the local model parameters and completes training of a local time sequence prediction model, so that the urban rail line network passenger flow is predicted, through a federated learning framework, AFC data leakage risks are avoided, cross-line global information is fully utilized, the passenger flow prediction accuracy is improved, cross-line communication overhead is greatly reduced, and the urban rail line network passenger flow prediction efficiency is improved. The real-time prediction efficiency is improved, and a reliable basis is provided for global urban rail network operation and train shift scheduling.

Systems, methods, and instrumentalities are disclosed for performing extrapolation filter-based intra prediction (EIP) in combination with other prediction modes. For example, EIP may be combined with one or more of combined inter-intra prediction (CIIP), geometric partitioning modes (GPM), spatial geometric partitioning modes (SGPM), or combined intra block copy geometric partitioning modes IBC-GPM. A video encoding / decoding device may obtain a first prediction associated with a current block. The first prediction may be obtained based on an extrapolation filter. The device may obtain a second prediction associated with the current block. The device may generate a blended prediction based on the first prediction and / or the second prediction. The device may encode / decode the current block based on the blended prediction.

The invention provides an artificial intelligence chip and a cache operation method thereof. The artificial intelligence chip comprises a cache, a plurality of computing cores and a notepad distributing-releasing unit. And the cache dynamically distributes the storage space of the cache into a cache mode space and a notepad mode space. At least one of the plurality of computing cores issues a cache mode access request to the cache to perform data access to the cache mode space. The pad allocation-release unit performs data access to the cached pad mode space on the basis of a pad mode access request issued by at least one of the plurality of computing cores. In response to the pad mode space being insufficient to serve the pad mode access request, the pad allocation-release unit generates a pad allocation request to the cache to dynamically increase the pad mode space.

The invention provides a photoetching test pattern generation method and computer equipment. The generation method of the photoetching test pattern comprises the following steps: constructing an initial test pattern; training a de-noising diffusion probability model by using the initial test pattern; combining the trained de-noising diffusion probability model with random noise to obtain a generated test pattern; and screening the generated test pattern to obtain a target test pattern. According to the scheme provided by the invention, simple and multivariate test patterns can be automatically generated, excessive dependence on artificial experience is avoided, the test period is shortened, the consistency and stability of a photoetching process are improved, and the photoetching yield is improved; according to the method, pattern exploration is carried out in a design space specified by a process design suite of a current process node, the boundary of a known pattern space is expanded, features in various design spaces are covered as far as possible, the coverage rate is increased, and generation of a photoetching test pattern is accelerated.

To provide a game machine capable of suitably increasing a degree of attention to a game.SOLUTION: While "opening operation" including "a correction variation 1" pattern executed when a left reel LR is stopped in a position in which a jackpot pattern deviates within a range of a first step to sixtieth step from a normal stop position on an effective line in a variable performance inversely rotates a middle reel CR and a right reel RR at rest in a normal position by one pattern space from an initial stop position while inversely rotating all the reels LR, CR, RR simultaneously when the opening is started, the left reel LR at rest in a first deviation amount position is inversely rotated by the number of steps obtained by adding one deviation amount step to one pattern space (namely, 20 steps) from the first deviation amount position.SELECTED DRAWING: Figure 115

The present invention relates to the technical field of industrial equipment fault monitoring and diagnosis, and specifically to a deep learning-based industrial equipment fault prediction and diagnosis system, comprising a dynamic data alignment module, a multidimensional feature fusion module, a fault mode decoupling module, and an interpretable diagnostic mapping module. The system comprises: a dynamic data alignment module for constructing a multidimensional equipment status data stream under a unified timeline; a multidimensional feature fusion module for performing orthogonal decomposition in feature space to extract a set of independent feature vectors; a fault mode decoupling module for constructing a fault mode space projection template; and an interpretable diagnostic mapping module for performing constraint matching to generate a visual diagnostic report. The present invention achieves accurate fault location and remaining life prediction by dynamically aligning and multidimensional feature fusion of multi-source heterogeneous sensor data, combining fault mode decoupling and interpretable diagnostic mapping, while reducing feature redundancy and noise interference.

The invention discloses an automatic program repairing method based on mode collaboration, which comprises the following steps of: taking defect codes and change contexts thereof as input, establishing a repairing mode space through change feature representation and template semantic induction, and dividing a repairing task into semantic insertion, statement level, expression level, mixing and the like; on the basis of a pre-training code language model, expert adapters with efficient parameters are constructed for different repair modes, a dynamic gating fusion mechanism is introduced, multiple expert capabilities are adaptively selected and weighted and combined according to representation of defect samples, candidate repair patches are output, and an effective repair result is obtained through engineering constraint verification. Compared with a single-model or single-adapter scheme, the problems of non-uniform data distribution and mutual capability interference in a multi-mode repair scene can be relieved, and the patch generation accuracy and stability are improved.

A display device is provided. The display device includes a display area, a peripheral area located outside the display area and including a plurality of sides and a corner peripheral area that is curved and located between two sides of the plurality of sides extending in different directions, a display element located in the display area, a first dam portion in the peripheral area and arranged to surround the display area, and a pattern group in the corner peripheral area such that the first dam portion is between the display area and the pattern group, in which the pattern group includes a first pattern portion including a plurality of patterns spaced apart from each other in a radial direction.

The embodiment of the invention discloses a dynamic spatial filter, and the filter comprises a plurality of masks which have a preset shape and are used for achieving a specific spatial filtering function; the mask is connected to the motion output end of the X-axis motion mechanism; the X-axis movement mechanism is used for driving the mask to move on the X axis; the movement output end of the X-axis movement mechanism is borne by the movement output end of the Y-axis movement mechanism, and the Y-axis movement mechanism is used for driving the driving output end of the X-axis movement mechanism to move on the Y axis; the two-dimensional position of the mask is adjusted through the X-axis displacement and the Y-axis displacement; meanwhile, the embodiment of the invention discloses an optical detection system adopting the dynamic spatial filter. The shielding area of the spatial filter can be dynamically controlled in a two-dimensional mode, and flexible control over the pattern spatial information frequency spectrum of the to-be-detected object with the pattern is achieved.

A display apparatus includes a display area, a peripheral area located outside the display area and including sides and a corner peripheral area that is located between two of the sides that extend in different directions, a display element located in the display area, a first dam portion located in the peripheral area and arranged to surround the display area, and a pattern group located in the corner peripheral area such that the first dam portion is between the display area and the pattern group, wherein the pattern group includes a first pattern portion including a plurality of patterns spaced apart from each other along a radial direction.

A problem of supervised learning is overcome by using patches to discover objects in unlabeled training images. The discovered objects are embedded in a pattern space. An AI machine replaces manual entry steps of training with a machine-centric process including clustering in a pixel space, clustering in latent space and building the pattern space based on different losses derived from pixel space clustering and latent space clustering. A distance structure in the pattern space captures the co-occurrence of patterns due to frequently appearing objects in training image data. Embodiments provide image representation based on local image patch naturally handles the position and scale invariance property that is important to effective object detection. Embodiments successfully identifies frequent objects such as human faces, human bodies, animals, or vehicles from unorganized data images based on a small quantity of training images.

A layout splitting method, system and device, a storage medium and a computer program, the method comprising: obtaining a mask layout to be split, the mask layout to be split having a plurality of graphs, the plurality of graphs comprising first graphs arranged in parallel at intervals in a first direction and second graphs arranged in parallel at intervals in a second direction, the spaced areas of the first pattern in the first direction and the spaced areas of the second pattern in the second direction are trimming areas; splitting the mask layout to be split into a first sub-mask layout and a second sub-mask layout; performing first merging processing on the first pattern of the first sub-mask layout to form a first merged pattern; performing second combination processing on the second pattern of the second sub-mask layout to form a second combined pattern; a third sub-mask layout is provided, the sub-mask layout comprises a plurality of trimming patterns, and the distribution of the trimming patterns in the third sub-mask layout corresponds to the distribution of the trimming areas. An exposure process window is enlarged, and pattern transmission precision is improved.

This invention discloses a method for rapidly reconstructing the electromagnetic characteristics of an unmanned aerial vehicle (UAV) swarm with variable attitude. The method treats each UAV in the swarm as an independently reusable electromagnetic module, constructing its spherical multi-port response model through a single full-wave simulation. It utilizes analytical spherical propagation operators to describe the free-space electromagnetic coupling between UAVs, and analytically characterizes the impact of UAV attitude and position changes on radiation patterns using spherical geometric transformation and translation operators. After assembling the swarm-level linear system in the spherical pattern space, the radiation and scattering electromagnetic characteristics of the UAV swarm under arbitrary attitude and formation conditions can be rapidly reconstructed without re-performing a full-wave simulation.