Algorithm 1: The proposed NCM-LGDM-FMEA

1: Determine FM = {FMi, i = 1, 2, …, m}  2: for each ${\mathrm{FM}}_i\in FM$ do  3:  for each $\mathrm{RF}\in \{\mathrm{O},\mathrm{S},\mathrm{D}\}$ do  4:   Collect heterogeneous evaluation data D = {dk, k = 1, …, n}.  5:   Filter invalid data by (15), $D\to D^{\prime }$.  6:   for each $d_k\in D^{\prime }$ do  7:    Convert dk into an NCM ${\tilde{d}}_k$ as Section 3.2, $D^{\prime }\to \tilde{D}$.  8:    Bad data processing on ${\tilde{d}}_k$ by (16), $\tilde{D}\to {\tilde{D}}^{\prime }$.  9:    Outlier processing on ${\tilde{d}}_k$ by (17), ${\tilde{D}}^{\prime }\to {\tilde{D}}^{″}$. 10:    Tolerance limit processing on ${\tilde{d}}_k$ by (18), ${\tilde{D}}^{″}\to {\tilde{D}}^{‴}$. 11:   end for 12:   for each ${\tilde{d}}_k\in {\tilde{D}}^{‴}$ do 13:    Calculate the UDk of ${\tilde{d}}_k$ by (19). 14:    Calculate the DDk of ${\tilde{d}}_k$ by (21). 15:    Calculate the data weight $w_k^{\mathrm{data}}$ of ${\tilde{d}}_k$ by (22). 16:    Calculate the identity weight $w_k^{\mathrm{identity}}$ of ek by (23). 17:    Calculate the comprehensive weight wk of ek by (24). 18:    Calculate the LGDM result by (9) to aggregate ${\tilde{d}}_k$ with wk. 19:   end for 20:   Obtain $\tilde{O}$, $\tilde{S}$, and respectively. 21:  end for 22:  Calculate ${\widetilde{RPN}}_i$ of FMi by (25). 23: end for 24: Prioritize {FMi} by (6) based on their RPNs.