Algorithm 3:${\mathcal{A}}_{21}$ Find rule
Require: The voltage of the device $X_1$, The voltage of the device $X_2$, …, The voltage of the device $X_n$.
if ${\mathcal{C}}_{X_1}-{\varphi }_{tol}<\mathcal{C}<{\mathcal{C}}_{X_1}+{\varphi }_{tol}$ then
${\mathsf{\Psi }}_{{\mathcal{A}}_1}=1$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=0$	▹ Normal standard
else if … then
${\mathsf{\Psi }}_{{\mathcal{A}}_1}=1$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=0$	▹ Normal standard
else if ${\mathcal{C}}_{X_n}-{\varphi }_{tol}<\mathcal{C}<{\mathcal{C}}_{X_n}+{\varphi }_{tol}$ then
${\mathsf{\Psi }}_{{\mathcal{A}}_1}=1$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=0$	▹ Normal standard
else
${\mathsf{\Psi }}_{{\mathcal{A}}_1}=0$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=0$	▹ Normal standard
if ${\mathsf{\Psi }}_{{\mathcal{A}}_1}=0$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=0$ Conform to the rules then
Abnormal but does not affect the status. For exceptions, use data to determine the category, and find exception rules among the exceptions. If it does not match, look for a classification method.
${\mathsf{\Psi }}_{{\mathcal{A}}_1}=1$, ${\mathsf{\Psi }}_{{\mathcal{A}}_{21}}=Ø$
else
Classify new anomalies. Use supervised or unsupervised learning, enhanced learning, use deep learning, new abnormal state, and at the same time conform to the abnormal state of historical data. To use classification methods, record them for later use or detection, and record the results.
Set up ${\mathsf{\Psi }}_{{\mathcal{A}}_{22}}=1$, Judgment belongs to ${\mathcal{A}}_{22}$ There are no rules for exceptions, and unsupervised learning is used to detect whether there are new classifications.
end if
end if