Non-Intrusive Load Identification Method Based on Gramian Angular Difference Field Image Coding

Ming Fu; Bin Duan

doi:10.3788/LOP230716

Journals >Laser & Optoelectronics Progress >Volume 60 >Issue 24 >Page 2410001 > Article

Laser & Optoelectronics Progress
Vol. 60, Issue 24, 2410001 (2023)

Non-Intrusive Load Identification Method Based on Gramian Angular Difference Field Image Coding

Ming Fu^* and Bin Duan

Author Affiliations

School of Automation and Electronic Information, Xiangtan University, Xiangtan 411105, Hunan, China

show less

DOI: 10.3788/LOP230716 Cite this Article Set citation alerts

Ming Fu, Bin Duan. Non-Intrusive Load Identification Method Based on Gramian Angular Difference Field Image Coding[J]. Laser & Optoelectronics Progress, 2023, 60(24): 2410001 Copy Citation Text

show less

Fig. 1. Chart of load signatures construction

Download full size

Fig. 2. Chart of CBAM attention mechanism implementation process

Download full size

Fig. 3. Chart of load identification network structure

Download full size

Fig. 4. Number of samples for each appliance in the dataset. (a) PLAID dataset; (b) WHITED dataset

Download full size

Fig. 5. Training process of load identification model. (a) PLAID dataset; (b) WHITED dataset

Download full size

Fig. 6. Confusion matrix for the final test results. (a) PLAID dataset; (b) WHITED dataset

Download full size

Fig. 7. Experimental results for different

N_{t}

values

Download full size

Model	Load signature	PLAID dataset		WHITED dataset
Model	Load signature	Accuracy /%	F1 score /%	Accuracy /%	F1 score /%
CBAM	V-I image	95.16	95.11	97.55	97.56
CBAM	GADF image	99.45	99.44	99.24	99.23
ResNet-18	V-I image	97.39	97.41	96.92	96.89
ResNet-18	GADF image	97.41	97.42	97.73	97.72
LeNet-5	V-I image	95.53	95.52	96.15	96.12
LeNet-5	GADF image	96.43	96.44	96.24	96.24
SE	V-I image	95.16	95.15	96.15	96.14
SE	GADF image	98.52	98.53	98.48	98.47
ECA	V-I image	96.27	96.26	96.15	96.21
ECA	GADF image	96.86	96.87	97.18	97.17

Table 1. Comparison of experimental results

Appliance	Accuracy /%
Appliance	Ref.［13］	Ref.［14］	Ref.［28］	Ref.［29］	Ref.［8］	Proposed method
Total	91.0	94.0	97.95	97.7	98.2	99.45
Fluorescent lamp	100	98.5	99.43	100	99.4	100
Fridge	70.0	84.0	81.58	90	78.9	100
Hairdryer	100	94.5	96.79	99.4	98.1	100
Air conditioner	75.0	92.2	92.42	86.6	95.5	100
Laptop	100	92.4	96.51	99.4	99.4	97.96
Microwave	100	98.6	97.84	97.9	100	100
Vacuum	100	100	96.79	97.4	100	100
Incandescent light	100	93.2	100	97.4	100	100
Fan	75.0	83.3	100	96.6	100	100
Washing machine	85.0	90.0	96.15	100	88.5	95.92
Heater	95.0	82.8	100	100	100	100

Table 2. Comparison of accuracy with other load recognition method

Ming Fu, Bin Duan. Non-Intrusive Load Identification Method Based on Gramian Angular Difference Field Image Coding[J]. Laser & Optoelectronics Progress, 2023, 60(24): 2410001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information