Skip to main content

Table 5 Energy disaggregation performance in terms of estimation accuracy for the two-stage NILM methodology

From: Robust energy disaggregation using appliance-specific temporal contextual information

Dataset

Baseline NILM

TCI (stage 1)

w = 1

w = 3

w = 5

w = 7

w = 9

w = 11

w = 13

wopt

ECO-1

70.0

72.8

70.6

72.7

73.2

73.7

67.6

67.1

67.5

76.1

ECO-2

75.0

77.3

75.1

76.6

76.5

76.4

79.9

76.1

76.9

84.1

ECO-4

79.7

81.1

82.3

83.1

82.9

83.4

83.2

82.9

82.1

86.4

ECO-5

84.5

86.8

87.3

87.3

87.4

87.3

89.8

87.4

87.6

89.9

ECO-6

80.8

81.5

81.1

82.0

80.7

80.5

80.5

80.4

80.3

82.8

REDD-1

69.2

71.2

70.0

72.9

73.6

69.2

70.1

69.4

67.2

73.9

REDD-2

73.8

76.9

74.3

74.4

77.9

76.4

75.3

71.9

73.0

80.1

REDD-3

62.5

63.7

66.6

68.7

68.9

67.1

63.2

62.8

60.8

69.7

REDD-4

70.7

73.8

73.9

73.2

74.3

74.7

73.0

73.1

72.9

76.3

REDD-6

77.9

79.1

78.5

79.2

79.1

77.4

77.5

77.7

76.7

81.3

iAWE

63.1

68.9

63.9

64.0

66.3

64.7

66.9

71.4

64.1

73.1

  1. The best performing length of the temporal contextual window w for each of the evaluated datasets is indicated in italics