9results about How to "Avoid ignoring" patented technology

The invention provides an exception reproduction method, system and device and a computer storage medium. The exception reproduction method comprises the steps that an application front end obtains running data of an application program and determines whether the application program has an exception phenomenon or not based on the running data; under the condition that the application program is abnormal, the application front end sends an operation log of the application program to the reproduction rear end; the reproduction rear end simulates and reproduces the abnormal phenomenon for multiple times based on the operation log; and the reproduction rear end determines the cause of the abnormal phenomenon based on the reproduction result of the multiple simulation reproduction. Through the exception reproduction method, the reproduction success rate of the exception phenomenon is improved, and the exception positioning and repairing period in the application program is shortened.

The invention discloses an ancient book and historic site element screening method based on a multi-dimensional scale. According to the method, the limitation of single-dimensional judgment in traditional screening is effectively avoided by constructing a multi-dimensional evaluation system and combining a depth analysis algorithm. According to the method, firstly, the authenticity of data is ensured through historical cross check and expert interviews, then potential associations among all dimensions are mined, and candidate elements which are balanced in multiple dimensions or have unique values are accurately recognized. The method not only can comprehensively reflect the comprehensive value of the ancient books and historic sites, but also can prevent the overall significance from being ignored due to the shortages of individual dimensions, so that the screening result better meets the actual requirements, and a reliable basis is provided for subsequent protection, research or utilization. Meanwhile, a closed-loop optimization mechanism is formed in the whole process, the deep analysis algorithm can timely find out anomalies or unreasonable parts in the data, adjustment and improvement of subsequent steps are promoted, the screening process is more flexible and efficient, and decision making is more scientific and reasonable.

The application discloses a distributed energy storage scheduling method based on a reinforcement learning algorithm, relates to the technical field of distributed energy storage, and comprises the following steps: an environment model of a battery swap station is constructed, and the scheduling management of the battery swap station is abstracted as a Markov decision process; wherein the environment model comprehensively considers battery charging and discharging characteristics, battery degradation and user charging and battery swap demands; an improved DDPG algorithm network architecture is introduced to strengthen the optimal scheduling strategy output by the Markov decision process; an experience replay pool is set, DDPG algorithm network parameters are updated, and the final optimal scheduling strategy is optimized and obtained. The application abstracts the scheduling management of the battery swap station as a Markov decision process, introduces the improved DDPG algorithm network architecture, and sets the experience replay pool, so that a small amount of abnormal sample experience is avoided from being ignored, the final optimal scheduling strategy can be suitable for a complex dynamic power grid environment, the uncertainty problem of the distributed energy storage power grid can be solved, and actual scheduling demands can be met.

This invention discloses a GPU code policy-driven generation method and system based on parallelism inference constraints. It collects cross-iteration memory access address features of variables using GDB, constructs a spatiotemporal program graph, and applies a spatiotemporal graph convolutional network for modeling. This yields prediction results for loop parallelism types and a five-dimensional optimization policy prediction vector. For serial loops, it directly returns the original C code. For parallelizable loops, the prediction results are converted into natural language parallelism labels and optimization instruction sets, which are then concatenated in a fixed order to construct structured hints. This drives a large language model as a syntax constraint executor to generate target CUDA code. This invention explicitly decouples parallelism inference from code generation, eliminating erroneous parallelization of the large language model and misjudgment of optimization policies, significantly improving the accuracy and execution performance of the generated code. It can be widely applied in the GPU migration of high-performance computing programs and heterogeneous computing fields.

The invention belongs to the technical field of medical information processing, and particularly relates to a POCT detection item management system and method. The system comprises an acquisition module used for acquiring a to-be-detected sample; the detection module is used for detecting a sample and sending a detection result to the processing center; the processing center is used for receiving the detection result and issuing a detection report according to the detection result; the processing center can perform information interaction with the hospital information system, and the processing center can perform information interaction with the laboratory information system; the processing center comprises an alarm module and a quality control management module; the alarm module is used for giving an alarm prompt according to a set threshold value; and the quality control management module is used for performing quality control management on the detection module. Closed-loop management is realized, and an abnormal result is prevented from being ignored. In addition, the timeliness of the detection equipment can be greatly improved, and the detection efficiency is greatly improved.

The invention relates to the technical field of safety early warning, and discloses a factory equipment early warning control method based on the Internet of Things, which comprises the following steps: acquiring input power and output load data of equipment, and performing identification and linear interpolation correction on abnormal data to form a corrected data sequence; subsequently, the device usage intensity is calculated and smoothed, and breaking points are identified by analyzing the intensity change rate, thereby dividing the device operation process into a plurality of consecutive segments. And for each section, calculating a safety index of the section, and constructing a normal equipment use structure fingerprint as a reference. And comparing each index of the real-time section with the normal fingerprint section, and calculating a structural damage index and an accumulated damage index so as to quantify the abnormal degree. According to a preset grading rule, the system can judge the starting point and the range of the abnormal chain, and automatically triggers different grades of control instructions from early warning, alarming to equipment operation stopping according to the damage degree.

The application discloses a 5G intelligent terminal equipment management method and system, and particularly relates to the fields of terminal equipment management and Internet of Things worm protection, and comprises the following steps: collecting the message transmission delay of a terminal and adjacent terminals in a continuous communication process, recording and arranging the messages in time sequence, filling the arrangement result into a two-dimensional delay matrix, with the rows of the two-dimensional delay matrix corresponding to continuous time periods and the columns corresponding to adjacent terminals, and outputting the two-dimensional delay matrix; the scheme constructs a two-dimensional delay matrix, time chain data, residual record table and difference curve data, and combines the processing logic of reverse order rearrangement, disturbance correction and frozen state check chain, realizes delay mode consistency judgment across time intervals and terminals, identifies forged or tampered log tracks, avoids fault tolerance logic from being triggered by an Internet of Things worm to trigger an error rollback, and solves the problem that existing systems lose recognition ability when facing cross-time-space log forgery.

The invention discloses an intelligent factory management method and system based on a smart factory, and relates to the technical field of digital management, and the method comprises the steps: collecting management layer operation behavior data from an informatization management system of the smart factory, carrying out the semantic analysis and data mapping, and then carrying out the ascending order arrangement and chain connection operation; and constructing a management behavior chain L, and obtaining a structure evolution characteristic quantity after data processing. And performing management state evaluation by constructing a trigger rhythm dispersity DL, and performing complexity description analysis on the management behavior structure according to the structure evolution characteristic quantity when the management behavior rhythm is unbalanced. And finally, performing management behavior chain situation assessment by constructing a management behavior chain situation index GXL. According to the invention, through conjoint analysis of the management behavior time sequence characteristics and the structure characteristics, systematic description and evaluation of the operation state of the factory management behavior chain are realized, and the method is suitable for management situation analysis and auxiliary decision making in a smart factory environment.

This disclosure provides a retrieval-enhanced diagnostic report generation method, apparatus, and electronic device, comprising: generating a preliminary diagnostic report for medical images of a case to be analyzed using a first preset large model; calculating the similarity between the preliminary diagnostic report and pre-stored diagnostic reports in a database to obtain similar cases; using the images of similar cases and their actual diagnostic reports as reference information, and inputting the medical images of the case to be analyzed and the reference information into a second preset large model to generate a final diagnostic report. This achieves a dual-driven diagnosis combining visual and semantic elements, accurately capturing pathological semantic relationships, avoiding the neglect of pathological diagnostic logic due to relying solely on visual features, and improving generalization ability.