Abstract: Intellectualization of the flotation process is a key approach to improving the process efficiency of coking coal preparation plants. This paper first conducts an in-depth analysis of the factors influencing coal slime flotation performance, identifies reagent dosage as the core control variable for flotation intellectualization in most coal preparation plants, and then systematically reviews the research status and technical progress of intelligent reagent dosing models and ash content prediction models for coal slime flotation at home and abroad: in the field of intelligent reagent dosing, it elaborates on the modeling proce. Compares a series of models such as linear regression (including Ridge regression and Lasso regression), support vector regression (SVR), BP neural network, long short-term memory (LSTM) network, and gated recurrent unit (GRU), analyzes the prediction accuracy of each model under different application scenarios as well as the adaptability of each model to operating conditions such as feedstock stability and data volume, while in the field of intelligent ash content prediction, it summarizes the modeling logic, model characteristics of the feature engineering modeling method based on machine vision and the transfer learning method based on convolutional neural network (CNN), along with their adaptability to coal preparation plants in terms of data volume, computing power, and feedstock stability. The current mainstream flotation reagent dosing control system adopts a “feedforward + feedback” regulation architecture: in the feedforward link, the reagent addition model constructed based on machine learning or neural networks can make up for the shortcomings of traditional manual reagent dosing, and in the feedback link, the ash content prediction model based on machine vision-based feature engineering and CNN-based transfer learning overcomes the defects of lag and poor accuracy of traditionline ash measurement methods, providing reliable quality feedback for closed-loop control. On this basis, the paper further prospects the intelligent control system for coal slime flotation: supported by the closed-loop mechanism of “feedforward intelligent reagent dosing - feedback ash content prediction”, this system will establish a full-process intelligent regulation system featuring “data-driven - model collaboration - closed-loop optimization”, promote the transformation of the flotation process from experience-driven to data-driven, provide references for the intelligent upgrading of flotation in coal preparation plants, and contribute to the high-quality development of the coal industry.