Under downhole microseismic monitoring conditions, uneven data quality or large velocity-model errors may cause the inversion to misinterpret noise and local waveform-fitting residuals as spurious non-double-couple components such as ISO and CLVD components, thereby affecting the interpretation of microseismic rupture types. To address this problem, on the basis of integrating P-wave and S-wave waveforms, P-wave first-motion polarity, and S/P amplitude ratio information, this paper proposes a more effective microseismic moment tensor inversion method. At the observation level, weights are assigned according to receiver-recording quality and effective signal quality in different frequency bands. At the source-model level, an adaptive DC skeleton constraint is introduced to first identify a relatively stable dominant shear mechanism and then regulate non-double-couple components such as ISO and CLVD. In synthetic-data tests, representative source-type events and a multistage synthetic microseismic event set are separately designed to verify the effectiveness and stability of the proposed method. The results show that the proposed method can effectively suppress the formation of spurious non-double-couple components and can better preserve the DC, ISO, and CLVD component proportions in mixed-type events. Compared with the waveform-matching method and the fixed-weight joint-constrained method, the proposed method maintains physically acceptable solution proportions ranging from 0.77 to 0.86 under noise interference and from 0.66 to 0.86 under velocity perturbation, while also achieving better performance in terms of moment tensor deviation angle and source-type proportion recovery.