The smart electricity meter，as the terminal device for smart grid marketing，electricity information，and energy distribution，provides strong data support in various aspects such as distribution network operation and maintenance management and customer experience optimization.However，due to the complex and variable operating environment，it is difficult to accurately assess its running state.Therefore，the modified association rules mining （MARM） method is employed，which improves upon two key importance evaluation criteria in the traditional ARM model.This enhancement not only increases the accuracy of state assessment but also improves the ability to identify potential operational risks.Additionally，to reduce uncertainty when dealing with continuous features，the traditional fuzzy inference system （FIS） is enhanced by introducing fuzzy probability （FP） and the tiered fuzzy inference system （TFIS）.Through model integration，a MARM-FP-TFIS （MFT） model that combines strong association rule identification with probability fuzzy inference is developed to solve the health assessment of the running state of smart electricity meters，realizing the evaluation of the running state of the smart energy meter.Finally，case studies verify the feasibility and functionality of the established model in practical applications，thus achieving an accurate assessment of the running state of smart electricity meters under multidime nsional data conditions.