基于工业生产期望指标的人工智能模型学习方法及系统

By introducing comprehensive expectation indicators for industrial production applications and a closed-loop self-learning system into the artificial intelligence model, the problem of inconsistency between model training objectives and business expectations was solved, realizing autonomous optimization and efficient adaptability of the model, and improving the overall efficiency of industrial production.

CN122412968APending Publication Date: 2026-07-17SHANGHAI BAOSIGHT SOFTWARE CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHANGHAI BAOSIGHT SOFTWARE CO LTD
Filing Date
2026-06-22
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

In existing technologies, the training objectives of artificial intelligence models in industrial production are inconsistent with the overall business expectations, resulting in poor model application performance. Furthermore, after deployment, the models cannot be autonomously optimized based on real-time production results, making it difficult to adapt to changes in working conditions.

Method used

By introducing comprehensive expected indicators for industrial production applications, a closed-loop self-learning system is constructed, including modules for data management, indicator evaluation, model training, self-learning, and inference control. This system monitors production data in real time and optimizes the model to meet the needs of comprehensive production benefits.

Benefits of technology

It improves the applicability of artificial intelligence models, enabling them to self-optimize based on actual production performance, maintain a high level of operation, lower the application threshold, and enhance the interpretability and credibility of the models.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122412968A_ABST
    Figure CN122412968A_ABST
Patent Text Reader

Abstract

本发明提供了一种基于工业生产期望指标的人工智能模型学习方法及系统，旨在解决人工智能模型训练目标与工业生产综合业务期望不一致，以及模型部署后无法适应动态工况变化的问题。该方法包括：建立一套工业生产应用期望指标集；根据该指标集构建一个工业生产应用综合期望指标；将一个基于该综合期望指标构建的正则项加入到人工智能模型的损失函数中，以引导模型训练方向与生产综合业务目标对齐。该方法还包括建立闭环自学习机制，实时计算综合期望指标的当前值，当该当前值满足预设触发条件时，触发对人工智能模型的自学习。本申请解决了模型目标与业务目标脱节的问题，实现了模型的闭环自主优化，增强了模型在工业场景下的适用性和可信度。
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Deep reinforcement learning training acceleration method based on expected value function

    CN111339690A

  • An Unbiased Learning Ranking Method Oriented Towards Increased Returns

    CN112612951B

  • Cold rolling performance simulation and dynamic optimization method based on historical data analysis

    CN118332772A