In response to the problems such as scattered cost sources, complex process coupling, and insufficient dynamic response of traditional accounting methods during the operation of smart mines, this paper proposes a full-cost analysis framework and risk assessment method under the drive of smart mines. The research integrates multi-source heterogeneous data including production scheduling, equipment monitoring, energy measurement, material circulation, safety and environmental protection, and financial accounting, and establishes a mapping relationship of “resource consumption – operation activities – cost objects” to achieve the calculation of costs in all stages of mining, transportation, beneficiation, maintenance, and management. On this basis, by combining long short-term memory networks and random forest algorithms, a cost risk assessment model that takes into account both temporal fluctuation characteristics and static structural characteristics is established, and a low, medium, and high-level risk output mechanism is formed. Experimental results show that the accuracy rate on the test set reaches 88.3%, the recall rate is 84.9%, and the F1 value is 0.854, which is overall superior to a single model. The research shows that the computer simulation technology and multi-source information fusion technology can more accurately identify the cause path and diffusion mode of mining expenditure anomaly, which provides a certain theoretical support for the realization of fine management of intelligent mine.
Povzetek: Študija razvija celovito ogrodje za analizo polnih stroškov in oceno tveganj v pametnih rudnikih. Združuje večvirove podatke o proizvodnji, opremi, energiji, materialih in financah ter uporablja LSTM in random forest modele. Rezultati kažejo izboljšano zaznavanje anomalij stroškov in napovedovanje tveganj, pri čemer ostajajo izzivi pri prenosu med scenariji, kakovosti podatkov in integraciji v realnem času.
