The deterioration common crossing elements increase the need for the maintenance of crossings and significantly reduce the reliability, availability and safety of railway traffic. The present paper introduces the results of experimental investigation of common crossing monitoring while its deterioration during the lifecycle of crossing. The conventional methods of common crossing condition estimation with time-based features like maximal accelerations, etc. are low appropriate as statistically significant condition indicators. The proposed multifractal analysis enables to extract the significant features from acceleration measurements. The extracted features are fused together with information about the train velocities and longitudinal position of wheels impact with the Lasso regularisation and multivariate linear regression. A ranking of feature importance is done and the redundant features are identified. The developed condition indicator has good correlation to the crossing lifetime and is simple for interpretation.