Algorithm 1: MLSTM-iForest
1:	Inputs: $X_{t,i}$—time-series source data, S—sliding window step, H—height limit,
2:	N—number of trees, A—abnormal score threshold
3:	Outputs: $R_{t,i}$ classification result
4:	Standardizing $X_{t,i}$ to get $X_{t,i}^n$
5:	Inputting $X_{t,i}^n$ into MLSTM to get predicted data $T_{i+n,i}^p$
6:	Calculating the deviation index $inde{x}_s$ based on S
7:	Building iForest based on $inde{x}_s$, H, and N
8:	Outputting abnormal $Scor{e}_{t,i}$ from iForest
9:	if $Scor{e}_{t,i}$> A
10:	$R_{t,i}$ = abnormal
11:	else
12:	$R_{t,i}$ = normal
13:	return $R_{t,i}$ $R_{t,i}$