10results about How to "Reduce calls" patented technology

The application discloses a general-purpose learning method and system for a physical robot, which comprises a special talent model library, a monitor set, a general talent model and a multi-robot distributed exploration cluster. The special talent model library contains L pre-trained special talent models. The monitor set contains L monitors, each of which is arranged with a special talent model and used for evaluating whether a current input state belongs to the use range of the special talent model. The multi-robot distributed exploration cluster contains multiple physical robots, each of which is arranged in a different task scene. In the application, multiple robots are introduced to perform different tasks in parallel, the monitor is used to dynamically determine when the special talent model needs to be involved, the appropriate special talent model is selected for guidance, the DAGGER type iterative data aggregation mechanism is adopted, the general talent model continuously learns on the state distribution generated by self-exploration in multiple scenes, and finally, the general talent model which can adapt to multiple scenes and surpass a single special talent is obtained.

The invention discloses a method and device for recognizing and reconstructing vector characters of a CAD drawing, and relates to the technical field of data processing. The method comprises the steps that a PDF file is analyzed, and vector elements and attribute information of the vector elements are extracted; character stroke candidate lines are screened out according to attribute grouping; carrying out connectivity analysis and density-based spatial clustering on the lines, and identifying a text region; rendering the vector lines of the text region into an image with high resolution, carrying out single character segmentation, and recognizing characters by combining drawing-level font library template matching with a single character recognition model; combining the single characters into word groups according to the positions and the context information, and performing word order correction by adopting a language model; the recognition result is written back into an editable CAD text object; according to the method, vector geometric information is directly utilized, distortion of bitmap sampling is avoided, the small character recognition rate and the anti-jamming capability are effectively improved, and end-to-end semantic recovery from vector lines to searchable and editable texts is achieved.

The application discloses a lossless compression method and system for image sensor raw data, and relates to the technical field of image sensors.The application fully utilizes the statistical characteristics that the pixel value difference between adjacent frames under the same exposure is extremely small, combines the inherent color arrangement rule of the Bayer format image, separates the color channels first and then performs time domain difference calculation, greatly reduces the data entropy value, and obtains a compression ratio much higher than that of a general lossless compression algorithm on the premise of completely lossless original pixel values; meanwhile, independent compression and packaging are performed with the minimum service unit as the granularity, target data can be directly located and accessed without decompression and large-scale package, the data retrieval and calculation overheads are reduced, the traceability of the original data is retained, the statistical confidence of the test result is not affected, and the batch data storage requirement in the image sensor production test scene is adapted.

The application provides a kind of multi virtual power plant transaction decision method and device for day-ahead electricity market, it is related to virtual power plant optimization operation technical field.The method comprises: according to distribution network operator and multiple virtual power plant systems, distribution network operator-virtual power plant master-slave game double-layer structure is established;Objective function and constraint condition of distribution network operator pricing game model are constructed;Objective function and constraint condition of deterministic virtual power plant robust transaction model are constructed;The uncertainty of wind, photovoltaic and controllable load is considered, and the deterministic virtual power plant robust transaction model is converted into uncertainty virtual power plant robust transaction model, and then a distributed robust scheduling model is obtained;Dynamic kriging element model is used to solve, and local search mechanism is introduced to obtain transaction decision;Wherein, column and constraint generation algorithm is used to solve distributed robust scheduling model.The application can help virtual power plant aggregator to significantly improve the income in day-ahead transaction.

The invention discloses a Yarn application process tree construction method and device and a storage medium, and relates to the technical field of Hadoop YARN resource management and monitoring. A method for constructing a Yarn application process tree comprises the steps that a monitoring program is started, and a monitor marks monitored application processes as different codes according to configuration; executing corresponding monitoring logic according to the code to obtain an application process tree construction result; the monitoring logic comprises a traditional global scanning construction mode, a single-process mode, a sub-process interface construction mode and an incremental updating mode. The method is simple in implementation mode and suitable for process monitoring and resource metering optimization of the node side of the YARN node-level resource manager, self-adaptive selection for different types of containers is achieved by introducing a multi-mode switchable monitoring framework, and resource waste and performance bottleneck caused by traditional fixed full scanning are avoided.

The embodiment of the invention provides a calling method and device for standard part resources in industrial design software, computer equipment and a medium, and the method comprises the steps: receiving a matching condition input by a user on a design software platform, screening out a standard part three-dimensional model meeting the matching condition according to the matching condition, and storing the standard part three-dimensional model in the design software platform; taking the model as a candidate standard component three-dimensional model; selecting an adaptive standard component three-dimensional model from the candidate standard component three-dimensional models according to actual installation requirements as a target standard component three-dimensional model; recommending a matched standard component three-dimensional model according to the target standard component three-dimensional model; and calling the target standard component three-dimensional model and the matched standard component three-dimensional model for installation, and displaying attribute information on the target standard component three-dimensional model and the matched standard component three-dimensional model. According to the scheme, the design process is optimized through standard part full-life-cycle tracing and global consistency management, and the standard part calling and assembling time is shortened.

The application relates to the technical field of e-commerce marketing compliance auditing, and discloses a marketing content compliance auditing method based on a domain large model; the method first acquires marketing materials to be audited, a delivery context, supervision rules and platform auditing rules, splits the rules into auditing requirements, and establishes a mapping relationship between the auditing requirements and violation categories; carrier recognition, content extraction, format normalization and fact fragment association are performed on texts, pictures, videos and landing page links; requirement level matching, exception situation verification and disposition level aggregation are performed according to the deterministic auditing rules, the confidence interval and the domain large model, and an auditing conclusion and a disposition instruction are generated; the application can make the multi-modal marketing content auditing conclusion locatable, traceable and reviewable, and reduce the amount of manual review.

This application relates to a video rendering processing method, apparatus, electronic device, and storage medium. The method includes: requesting a preset number of window handles and storing the preset number of window handles as a handle pool; when rendering video, obtaining an available window handle from the handle pool, creating the necessary rendering resources for the available window handle, and starting the video rendering process; when stopping video rendering, searching the handle pool for the window handle corresponding to the stopped video rendering, and clearing the rendering resources of the found window handle. This application solves the problem in the prior art where, during frequent and continuous multi-channel video rendering without restarting, the terminal rendering software causes a continuous accumulation of window handles, thus affecting system stability.

The application belongs to the field of rapid design and manufacture of production equipment, and particularly relates to a robot gripper structure parameter rapid simulation optimization method, which comprises the following steps: constructing a robot gripper structure parameterization model; simulating the stress and deformation of the robot gripper structure parameterization model under different sizes and different load conditions by using statics finite element simulation software to obtain stress and deformation simulation data; constructing a stress and deformation prediction model; optimizing the size parameters that meet the stress and deformation constraint conditions under the actual load conditions by establishing a particle swarm optimization algorithm, obtaining the optimal value of the size parameters, and substituting the optimal value into the stress and deformation prediction model to obtain the size parameters under the actual conditions and the stress and deformation under the load, so as to realize the optimization of the robot gripper structure parameters. The application utilizes the Kriging surrogate model to predict the stress and deformation, reduces the calling of the finite element model, and improves the optimization efficiency.

The invention provides a resource search method and device, equipment, a storage medium and a program product, and the method comprises the steps: carrying out the intention recognition of a resource search request inputted by a user based on a pre-trained large language model, and obtaining a request resource entity and a request resource type; determining a plurality of target search engines from a plurality of search engines based on the request resource type, and expanding search terms based on the request resource type; calling each target search engine for searching based on the request resource entity and the expanded search term; and verifying a search result returned by the target search engine, and generating a response of the resource search request based on the verified search result. Multiple search engines are used for parallel retrieval, so that the coverage degree and the search success rate are improved; by automatically verifying the search result, the search result filtering is realized, the user operation is simplified, and the convenience of resource downloading is improved.