Multi-State Energy Classifier to Evaluate the Performance of the NILM Algorithm

. 2019 Nov 28;19(23):5236. doi: 10.3390/s19235236

Algorithm 2 Appliance state clustering.

Input:

G_{T} = {g_{1}, g_{2}, \dots, g_{t}}

is the ground truth of appliance m
N = Maximum number of states
Output:

C_{S t a t e} = [C_{1}, C_{2}, \dots, C_{K}]

is the clustered operational states of appliance m

1:
$G_{S c a l e} =$ Standardize the values of $G_{T}$ time series
2:
for $K = 1$ toNdo
3:
Compute within groups sum of squares (WSS)
4:
end for
5:
Obtain K using elbow method
6:
Perform K-Means clustering on $G_{T}$ to find K clusters $C l u s_{1}, C l u s_{2}, \dots, C l u s_{K}$
7:
where $C l u s_{1} = \{G_{T_C l u s_{1}}^{(1)}, \dots, G_{T_C l u s_{1}}^{(n_{1})}\}, \dots, C l u s_{K} = \{G_{T_C l u s_{K}}^{(1)}, \dots, G_{T_C l u s_{K}}^{(n_{K})}\}$
8:
for $i = 1$ toKdo
9:
Get $C_{m e a n}$ and $C_{s t d . d e v}$ of cluster $C l u s_{i}$
10:
$C_{T h r e s} = \frac{C_{s t d . d e v}}{C_{m e a n}}$
11:
$C_{R a t e} = \frac{λ C_{s t d . d e v}}{C_{m e a n}}$ where $λ = 3$
12:
Store $[S_{i}, C_{m e a n}, C_{s t d . d e v}, C_{T h r e s}, C_{R a t e}]$ in $C_{i}$
13:
Store $C_{i}$ in $C_{S t a t e}$
14:
end for
15:
return $C_{s t a t e}$