固体充填开采方法在处理煤基固废和控制地表沉降方面具有较大优势，但其充填效率低、接续时间长、劳动强度大等问题制约着绿色充填开采的发展。针对固体充填技术升级的内生动力、行业发展的迫切需求以及矿山智能化建设的必然趋势，提出视频AI算法分析的固体智能充填方法。本文首先通过分析该固体智能充填方法的内涵及难点，构建出视频AI算法分析的固体智能充填开采方法系统构架，阐述视频AI算法的工作原理与实现流程，并给出视频AI算法可以实现的功能。根据不同地质条件分析了关键充填装备在不同工序下的影响因素，通过Creo进行液压支架骨架建模，实现液压支架在不同工况下机构的运动，给出对应的调控判据及路径，并设计关键充填装备在不同工序下的控制算法流程。根据视频算法特征及算法优缺点初步选择了图像识别算法，将构建好的目标数据集经过算法模型的训练及调参最终确定了最佳算法及对应的参数。通过某矿充填面应用效果分析，SVM各评价指标均优于其他算法，表明SVM模型在工况判别时表现出色，具有高度的准确性及可靠性。本研究可实现关键充填装备机构非正常工况的识别及调控、提高充填效率、机构位姿参数识别、充填空间夯实效果展示，可为视频AI算法分析的固体智能充填开采技术研发与应用提供理论指导。

The solid backfilling mining method has great advantages in handling coal-based solid waste and controlling surface subsidence, but its low backfilling efficiency, long succession time and high labour intensity constrain the development of green backfilling mining. Aiming at the endogenous driving force of solid backfilling technology upgrading, the urgent demand of industry development and the inevitable trend of mine intelligent construction, the solid intelligent backfilling method analysed by video AI algorithm is proposed. Initially, this paper examines the essence and challenges of intelligent solid backfilling methods, establishes the system framework for analyzing the intelligent solid backfilling mining method using video AI algorithms, elucidates the operational principles and implementation process of the video AI algorithm, and outlines the capabilities that can be achieved through this algorithm. It analyzes the influencing factors on key backfilling equipment under different geological conditions. The hydraulic support framework modeling was conducted using Creo to simulate its movement in various working conditions; corresponding control criteria and paths were provided, along with a designed control algorithm flow for key backfilling equipment across different processes. Based on the characteristics of video algorithms and their pros and cons, an image recognition algorithm is initially chosen; after training and adjusting the model, optimal algorithms and corresponding parameters are determined. Through an analysis of application effects on an coal mine fill surfaces, it is found that SVM evaluation indices outperform other algorithms, indicating excellent performance in discriminating working conditions with high accuracy and reliability. The research can realize the identification and regulation of abnormal working conditions of key backfilling equipment mechanism, improve backfilling efficiency, identify the positional parameters of the mechanism, and display the effect of backfilling space tamping, which can provide theoretical guidance for the research, development and application of solid intelligent backfilling and mining technology analyzed by video AI algorithm.