Abstract: In the face of dynamic demand fluctuations in the commodity market,the existing static mining production plan is difficult to match the changing market demand,resulting in low resource allocation efficiency and high operational risk.A monthly plan adjustment and optimization method based on the "prediction-generation-evaluation" integrated framework is pro posed.Firstly,a hidden Markov model (HMM) is constructed to mine the time series characteristics of futures trading data and identify the hidden state (high/medium/low) of market demand.Secondly,a plan generation model coupled with dynamic de mand and risk preference is established.Different levels of Gaussian disturbance are introduced to simulate the risk preference of decision makers (aggressive/neutral/conservative),and a differentiated alternative plan set is generated in combination with mine process constraints.Finally,the DEA-AHP comprehensive evaluation system is constructed to quantify the objective pro duction efficiency and integrate the management strategy weights under different market conditions to achieve the optimal solu tion screening.The simulation test based on the actual mining production data of a mining enterprise shows that the comprehen sive utility score of the optimization plan generated by this method is better than the original plan in all 9 test scenarios,with an average increase of about 0.94%.It effectively solves the problem of disconnection between production decision and dynamic demand,and provides theoretical support for scientific decision-making in dynamic market environment.