3 results about "Security industry" patented technology

The invention discloses an AI native intelligent central operating system based on reinforcement learning, and relates to the technical field of reinforcement learning, and the AI native intelligent central operating system comprises the steps: determining a corresponding regulation action based on a preset business target and state representation; according to the invention, the plurality of sub-regulation items are determined based on the recognition of the regulation action, and the quantitative execution is carried out along each sub-regulation item, so that the change information of the regulated key performance indicator is predicted, and the accuracy of the change information of the regulated key performance indicator is improved. Determining a corresponding reward value according to the identification of the change information of the key performance indicator; determining to-be-optimized content according to reinforcement learning of the reward value; according to the method, each optimized sub-regulation item is determined according to the matching of the to-be-optimized content and each sub-regulation item, and the final operation content of the security industry AI native intelligent center is constructed in combination with the action constraint relationship of the security database, so that the accuracy of the final operation content of the security industry AI native intelligent center is improved.

The application discloses a WeChat group public opinion on electricity rapid screening, marking and closed-loop treatment method and system based on RPA and AI image recognition, and belongs to the field of information technology and industrial automation. The method comprises the following steps: automatically collecting multimedia messages of WeChat groups through RPA robots; performing image character recognition and conversion by using a locally deployed OCR engine; performing rapid screening and false alarm elimination based on a pre-set domain sensitive word library and an associated non-sensitive word library; performing risk level determination and visual marking on the screening result; finally, generating a standardized work order and automatically pushing the work order to a business system to form a closed loop. Correspondingly, the system comprises a message collection module, a local recognition module, an intelligent screening engine, a hierarchical marking module and a work order pushing module. The core innovation of the application lies in that an AI recognition and processing architecture is locally deployed offline, so that data is not externally overflowed in the whole process, and through intelligent screening, structured preprocessing and business coding management, a full-process, automatic and intelligent closed loop from public opinion discovery to work order treatment is realized, and the application is particularly suitable for social group public opinion monitoring and service improvement in high-security industries such as electric power.