Non-intrusive load monitoring method combining BiLSTM-Transformer and Kolmogorov-Arnold network

doi:10.11772/j.issn.1001-9081.2025060728

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (6): 2026-2033.DOI: 10.11772/j.issn.1001-9081.2025060728

• Frontier and comprehensive applications • Previous Articles

Non-intrusive load monitoring method combining BiLSTM-Transformer and Kolmogorov-Arnold network

Jun QIN¹, Xintao JIAO¹(), Biqing ZENG²

^1.School of Artificial Intelligence，South China Normal University，Foshan Guangdong 528225，China
^2.Aberdeen Institute of Data Science and Artificial Intelligence，South China Normal University，Foshan Guangdong 528225，China

Received:2025-07-01 Revised:2025-09-03 Accepted:2025-09-24 Online:2025-10-09 Published:2026-06-10
Contact: Xintao JIAO
About author:QIN Jun， born in 2000， M. S. candidate. His research interests include non-intrusive load monitoring， artificial intelligence.
ZENG Biqing， born in 1969， Ph. D.， professor. His research interests include natural language processing， artificial intelligence.
First author contact:JIAO Xintao， born in 1979， Ph. D.， lecturer. His research interests include non-intrusive load monitoring， artificial intelligence.
Supported by:
Horizontal Project Commissioned by Foshan Nanhua Instruments Company Limited(2022440002001077)

融合BiLSTM-Transformer与Kolmogorov-Arnold网络的非侵入式负荷监测方法

秦隽¹, 焦新涛¹(), 曾碧卿²

^1.华南师范大学人工智能学院，广东佛山 528225
^2.华南师范大学阿伯丁数据科学与人工智能学院，广东佛山 528225

通讯作者: 焦新涛
作者简介:秦隽（2000—），男，广东广州人，硕士研究生，主要研究方向：非侵入式负荷监测、人工智能
曾碧卿（1969—），男，湖南衡南人，教授，博士，CCF杰出会员，主要研究方向：自然语言处理、人工智能。
第一联系人：焦新涛（1979—），男，江苏南通人，讲师，博士，CCF会员，主要研究方向：非侵入式负荷监测、人工智能
基金资助:
佛山市南华仪器股份有限公司委托横向项目(2022440002001077)

Abstract

Abstract:

To address the shortcomings of the existing deep learning-based Non-Intrusive Load Monitoring （NILM） methods in capturing long-term dependencies and complex nonlinear dynamic features， an NILM method combining BiLSTM-Transformer and Kolmogorov-Arnold Network （KAN） was proposed， and a mix model BT-KAN was constructed. Firstly， the BiLSTM-Transformer module was designed to combine the advantage of the Bidirectional Long Short-Term Memory （BiLSTM） network in modeling bidirectional sequence dependencies with the capability of Transformer in modeling global context， and a multi-head attention mechanism was employed to capture long-term dependencies of power load effectively， thereby improving the disaggregation accuracy long-cycle appliances. Then， the KAN module was used to capture nonlinear dynamic features of power load signals more accurately through a hierarchical nonlinear mapping mechanism based on the Kolmogorov-Arnold representation theorem， thereby improving the disaggregation accuracy for complex load modes. Experimental results on the REDD （Reference Energy Disaggregation Dataset） and UK-DALE （UK Domestic Appliance-Level Electricity） datasets show that compared with four Transformer-based similar models， the proposed model achieves reduction of at least 1.6% and 5.5% in Mean Absolute Error （MAE）， the improvement of at least 8.3% and 0.7% in F1-score. It can be seen that the proposed method captures long-term dependencies and nonlinear dynamic features in power load signals more accurately and improves the disaggregation effect of complex appliance operating modes.

Key words: Non-Intrusive Load Monitoring (NILM), Kolmogorov-Arnold Network (KAN), Transformer, Bidirectional Long Short-Term Memory (BiLSTM) network, multi-head attention mechanism

摘要：

针对现有的基于深度学习的非侵入式负荷监测（NILM）方法在捕捉长期依赖性和复杂非线性动态特征方面存在不足的问题，提出一种融合BiLSTM-Transformer与Kolmogorov-Arnold网络（KAN）的NILM方法，形成混合模型BT-KAN。首先，BiLSTM-Transformer模块结合双向长短时记忆（BiLSTM）网络在双向序列依赖建模上的优势与Transformer在全局上下文建模上的能力，并利用多头注意力机制有效捕捉电力负荷的长期依赖特性，从而提高长周期电器负荷分解的准确率；其次，KAN模块基于Kolmogorov-Arnold表示定理，通过分层非线性映射机制能更精确地捕捉电力负荷信号中的非线性动态特征，从而提高对复杂负载模式的分解准确率。在REDD（Reference Energy Disaggregation Dataset）和UK-DALE（UK Domestic Appliance-Level Electricity）这2个数据集上的实验结果表明，相较于基于Transformer的4个对比模型，所提方法的平均绝对误差（MAE）至少降低了1.6%和5.5%，F1分数至少提升了8.3%和0.7%。可见，所提方法能更准确地捕捉电力负荷信号中的长期依赖与非线性动态特征，并显著提升复杂电器运行模式的分解效果。

关键词: 非侵入式负荷监测, Kolmogorov-Arnold网络, Transformer, 双向长短时记忆网络, 多头注意力机制

CLC Number:

TP391.7

Jun QIN, Xintao JIAO, Biqing ZENG. Non-intrusive load monitoring method combining BiLSTM-Transformer and Kolmogorov-Arnold network[J]. Journal of Computer Applications, 2026, 46(6): 2026-2033.

秦隽, 焦新涛, 曾碧卿. 融合BiLSTM-Transformer与Kolmogorov-Arnold网络的非侵入式负荷监测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(6): 2026-2033.

Figures/Tables 6

References 28

[1]	TANONI G， PRINCIPI E， SQUARTINI S. Non-Intrusive Load Monitoring in industrial settings： a systematic review［J］. Renewable and Sustainable Energy Reviews， 2024， 202： No.114703.
[2]	何耿生，黄宇，曾金灿，等. 基于多通道融合机制的非侵入式负荷监测方法［J］. 计算机应用， 2025， 45（S1）： 317-322.
	HE G S， HUANG Y， ZENG J C， et al. Non-intrusive load monitoring method based on multi-channel fusion mechanism［J］. Journal of Computer Applications， 2025， 45（S1）： 317-322.
[3]	VASWANI A， SHAZEER N， PARMAR N， et al. Attention is all you need［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2017： 6000-6010.
[4]	LIU Z， WANG Y， VAIDYA S， et al. KAN： Kolmogorov-Arnold networks［EB/OL］. ［2025-01-01］..
[5]	AZAD M I， RAJABI R， ESTEBSARI A. Non-Intrusive Load Monitoring （NILM） using deep neural networks： a review［C］// Proceedings of the 2023 IEEE International Conference on Environment and Electrical Engineering. Piscataway： IEEE， 2023： 1-6.
[6]	WANG C， LI Q， YANG P， et al. A switching event detection method for household appliances based on improved sliding window［C］// Proceedings of the 6th International Conference on Energy， Electrical and Power Engineering. Piscataway： IEEE， 2023： 683-689.
[7]	REHMAN A U， LIE T T， VALLÈS B， et al. Event-detection algorithms for low sampling nonintrusive load monitoring systems based on low complexity statistical features［J］. IEEE Transactions on Instrumentation and Measurement， 2020， 69（3）： 751-759.
[8]	LIU B， CHANG F， LUAN W， et al. Improved appliance transient feature extraction via template matching［C］// Proceedings of the 2023 IEEE International Conference on Acoustics， Speech and Signal Processing. Piscataway： IEEE， 2023： 1-5.
[9]	CHEN K L， CHANG H H， CHEN N. A new transient feature extraction method of power signatures for Nonintrusive Load Monitoring Systems［C］// Proceedings of the 2013 IEEE International Workshop on Applied Measurements for Power Systems. Piscataway： IEEE， 2013： 79-84.
[10]	JIANG L， LI J， LUO S， et al. Power load event detection and classification based on edge symbol analysis and support vector machine［J］. Applied Computational Intelligence and Soft Computing， 2012， 2012： No.742461.
[11]	LAUGHMAN C， LEE K， COX R， et al. Power signature analysis［J］. IEEE Power and Energy Magazine， 2003， 1（2）： 56-63.
[12]	ZOHA A， GLUHAK A， NATI M， et al. Low-power appliance monitoring using factorial hidden Markov models［C］// Proceedings of the IEEE 8th International Conference on Intelligent Sensors， Sensor Networks and Information Processing. Piscataway： IEEE， 2013： 527-532.
[13]	KOLTER J Z， JAAKKOLA T. Approximate inference in additive factorial HMMs with application to energy disaggregation［C］// Proceedings of the 15th International Conference on Artificial Intelligence and Statistics. New York： JMLR.org， 2012： 1472-1482.
[14]	KELLY J， KNOTTENBELT W. The UK-DALE dataset， domestic appliance-level electricity demand and whole-house demand from five UK homes［J］. Scientific Data， 2015， 2： No.150007.
[15]	ZHANG C， ZHONG M， WANG Z， et al. Sequence-to-point learning with neural networks for non-intrusive load monitoring［C］// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2018： 2604-2611.
[16]	MAUCH L， YANG B. A new approach for supervised power disaggregation by using a deep recurrent LSTM network［C］// Proceedings of the 2015 IEEE Global Conference on Signal and Information Processing. Piscataway： IEEE， 2015： 63-67.
[17]	PASCANU R， MIKOLOV T， BENGIO Y. On the difficulty of training recurrent neural networks［C］// Proceedings of the 30th International Conference on Machine Learning. New York： JMLR.org， 2013： 1310-1318.
[18]	YUE Z， WITZIG C R， JORDE D， et al. BERT4NILM： a bidirectional Transformer model for non-intrusive load monitoring［C］// Proceedings of the 5th International Workshop on Non-Intrusive Load Monitoring. New York： ACM， 2020： 89-93.
[19]	PICCIALLI V， SUDOSO A M. Improving non-intrusive load disaggregation through an attention-based deep neural network［J］. Energies， 2021， 14（4）： No.847.
[20]	SYKIOTIS S， KASELIMI M， DOULAMIS A， et al. ELECTRIcity： an efficient transformer for non-intrusive load monitoring ［J］. Sensors， 2022， 22（8）： No.2926.
[21]	REN H， SU X， JENSSEN R， et al. Attention-guided temporal convolutional network for non-intrusive load monitoring［C］// Proceedings of the 2022 IEEE International Conference on Communications， Control， and Computing Technologies for Smart Grids. Piscataway： IEEE， 2022： 419-425.
[22]	WANG L， MAO S， NELMS R M. Transformer for nonintrusive load monitoring： complexity reduction and transferability［J］. IEEE Internet of Things Journal， 2022， 9（19）： 18987-18997.
[23]	ANGELIS G F， TIMPLALEXIS C， SALAMANIS A I， et al. Energformer： a new Transformer model for energy disaggregation［J］. IEEE Transactions on Consumer Electronics， 2023， 69（3）： 308-320.
[24]	XUAN Y， PANG C， YU H， et al. An enhanced bidirectional transformer model with temporal-aware self-attention for short-term load forecasting ［J］. IEEE Access， 2024， 12： 75625-75639.
[25]	IRANI AZAD M， RAJABI R， ESTEBSARI A. NonIntrusive Load Monitoring （NILM） using a deep learning model with a Transformer-based attention mechanism and temporal pooling［J］. Electronics， 2024， 13（2）： No.407.
[26]	ZHOU K， ZHANG Z， LU X. Non-intrusive load monitoring based on an efficient deep learning model with local feature extraction［J］. IEEE Transactions on Industrial Informatics， 2024， 20（7）： 9497-9507.
[27]	LIU H， LIU C， TIAN L， et al. Non-intrusive load disaggregation based on deep learning and multi-feature fusion［C］// Proceedings of the 3rd International Conference on Smart Power and Internet Energy Systems. Piscataway： IEEE， 2021： 210-215.
[28]	KOLTER J Z， JOHNSON M J. REDD： a public data set for energy disaggregation research［EB/OL］. ［2025-01-01］..

数据集	设备	阈值/ W	截断值/W	最短开机持续时间/s	最短关机持续时间/s
REDD	冰箱	50	400	10	2
	微波炉	200	1 800	2	5
	洗碗机	10	1 200	300	300
	洗衣机	20	3 500	300	26
UK-DALE	冰箱	50	300	10	2
	微波炉	200	3 000	2	5
	洗碗机	10	2 500	300	300
	洗衣机	20	2 500	300	26

数据集	设备	阈值/ W	截断值/W	最短开机持续时间/s	最短关机持续时间/s
REDD	冰箱	50	400	10	2
	微波炉	200	1 800	2	5
	洗碗机	10	1 200	300	300
	洗衣机	20	3 500	300	26
UK-DALE	冰箱	50	300	10	2
	微波炉	200	3 000	2	5
	洗碗机	10	2 500	300	300
	洗衣机	20	2 500	300	26

数据集	模型	冰箱			微波炉			洗碗机			洗衣机			平均
数据集	模型	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc
REDD	BERT4NILM	32.35	0.756	0.841	17.58	0.476	0.989	20.49	0.523	0.969	34.96	0.559	0.991	26.35	0.579	0.948
	ELECTRIcity				17.31	0.610	0.989	24.06	0.601	0.968	26.07	0.693	0.998	22.48	0.635	0.985
	GTA-BERT	30.69	0.761	0.887	17.61	0.599	0.997	22.41	0.472	0.962	18.02	0.694	0.993	22.18	0.632	0.960
	TransUNet-NILM	31.50	0.847		18.42	0.351		25.69	0.402		35.71	0.304		27.84	0.476
	BT-KAN	29.25	0.873	0.909	17.03	0.639	0.998	20.36	0.568	0.970	20.67	0.672	0.998	21.83	0.688	0.969
UK-DALE	BERT4NILM	25.49	0.766	0.813	6.57	0.014	0.995	16.18	0.667	0.966	6.98	0.325	0.966	13.81	0.443	0.935
	ELECTRIcity	22.61	0.810	0.843	6.28	0.277	0.996	18.96	0.818	0.984	5.56	0.797	0.994	13.35	0.676	0.954
	GTA-BERT	25.32	0.796	0.812	6.27	0.209	0.996	13.32	0.669	0.979	8.83	0.340	0.959	13.44	0.504	0.937
	TransUNet-NILM	21.50	0.822		6.58	0.413		13.06	0.536		6.05	0.762		11.80	0.633
	BT-KAN	19.93	0.839	0.860	5.94	0.463	0.996	12.78	0.679	0.985	5.95	0.744	0.996	11.15	0.681	0.959

数据集	模型	冰箱			微波炉			洗碗机			洗衣机			平均
数据集	模型	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc
REDD	BERT4NILM	32.35	0.756	0.841	17.58	0.476	0.989	20.49	0.523	0.969	34.96	0.559	0.991	26.35	0.579	0.948
	ELECTRIcity				17.31	0.610	0.989	24.06	0.601	0.968	26.07	0.693	0.998	22.48	0.635	0.985
	GTA-BERT	30.69	0.761	0.887	17.61	0.599	0.997	22.41	0.472	0.962	18.02	0.694	0.993	22.18	0.632	0.960
	TransUNet-NILM	31.50	0.847		18.42	0.351		25.69	0.402		35.71	0.304		27.84	0.476
	BT-KAN	29.25	0.873	0.909	17.03	0.639	0.998	20.36	0.568	0.970	20.67	0.672	0.998	21.83	0.688	0.969
UK-DALE	BERT4NILM	25.49	0.766	0.813	6.57	0.014	0.995	16.18	0.667	0.966	6.98	0.325	0.966	13.81	0.443	0.935
	ELECTRIcity	22.61	0.810	0.843	6.28	0.277	0.996	18.96	0.818	0.984	5.56	0.797	0.994	13.35	0.676	0.954
	GTA-BERT	25.32	0.796	0.812	6.27	0.209	0.996	13.32	0.669	0.979	8.83	0.340	0.959	13.44	0.504	0.937
	TransUNet-NILM	21.50	0.822		6.58	0.413		13.06	0.536		6.05	0.762		11.80	0.633
	BT-KAN	19.93	0.839	0.860	5.94	0.463	0.996	12.78	0.679	0.985	5.95	0.744	0.996	11.15	0.681	0.959

数据集	模型	冰箱			微波炉			洗碗机			洗衣机			平均
数据集	模型	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc	MAE/W	F1	Acc
REDD	w/o BiLSTM	33.99	0.835	0.859	18.71	0.555	0.988	25.08	0.422	0.956	26.90	0.562	0.986	26.17	0.594	0.947
	w/o Transformer	35.88	0.723	0.794	19.64	0.482	0.979	26.72	0.397	0.947	28.19	0.519	0.980	27.61	0.530	0.925
	w/o BiLSTM- Transformer	40.33	0.703	0.725	25.67	0.388	0.936	30.17	0.345	0.931	35.89	0.443	0.951	33.02	0.470	0.886
	w/o KAN	36.02	0.758	0.815	19.01	0.521	0.973	25.87	0.403	0.950	28.92	0.493	0.979	27.46	0.544	0.929
	BT-KAN	29.25	0.873	0.909	17.03	0.639	0.998	20.36	0.568	0.970	20.67	0.672	0.998	21.83	0.688	0.969
UK-DALE	w/o BiLSTM	22.20	0.805	0.824	6.11	0.406	0.979	13.07	0.518	0.967	6.57	0.682	0.962	11.99	0.603	0.933
	w/o Transformer	27.67	0.761	0.779	6.89	0.328	0.957	16.68	0.501	0.938	7.79	0.583	0.959	14.76	0.543	0.908
	w/o BiLSTM- Transformer	29.70	0.707	0.725	7.55	0.309	0.933	19.69	0.480	0.922	8.09	0.566	0.930	16.26	0.516	0.878
	w/o KAN	25.95	0.734	0.751	6.68	0.355	0.948	16.40	0.527	0.947	7.26	0.623	0.946	14.07	0.560	0.898
	BT-KAN	19.93	0.839	0.860	5.94	0.463	0.996	12.78	0.679	0.985	5.95	0.744	0.996	11.15	0.681	0.959

Non-intrusive load monitoring method combining BiLSTM-Transformer and Kolmogorov-Arnold network

融合BiLSTM-Transformer与Kolmogorov-Arnold网络的非侵入式负荷监测方法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 6

References 28

Related Articles 15

Recommended Articles

Metrics

[1]	Wenjun FENG, Xinwei SONG, Yuntao YUE. Time series prediction of environmental electric field intensity with generalized correlation entropy loss function-based Transformer model [J]. Journal of Computer Applications, 2026, 46(6): 1872-1880.
[2]	Lili HE, Meng CAO, Lei ZHANG, Hongjun PAN, Yi LIU, Chengxin SUN. Sign language generation model based on Kolmogorov-Arnold network and diffusion Transformer [J]. Journal of Computer Applications, 2026, 46(6): 1801-1810.
[3]	Yuanhao HE, Jun ZHAO. Defect detection algorithm for train bearing rollers based on FHC-DETR [J]. Journal of Computer Applications, 2026, 46(5): 1624-1633.
[4]	Minqi WU, Yuanhua YANG, Hang LI, Yaqin HU, Zhihao TANG, Teng MEI. Lightweight underwater small object detection based on graph Transformer and RT-DETR [J]. Journal of Computer Applications, 2026, 46(5): 1586-1595.
[5]	Xinyi YAN, Linglong ZHU, Yonghong ZHANG. CDC-DETR： multi-scale real-time human-vehicle detection method for complex traffic scenarios [J]. Journal of Computer Applications, 2026, 46(4): 1283-1291.
[6]	Xiang BAI, Juchuan LI, Huimin WANG, Chao JING, Jian NIU, Xingzhong ZHANG, Yongqiang CHENG. Power image retrieval method based on improved Swin Transformer [J]. Journal of Computer Applications, 2026, 46(4): 1334-1343.
[7]	Jie HU, Pengcheng LI, Jun SUN, Jiaao ZHANG. Key phrase extraction model based on multi-perspective information enhancement and hierarchical weighting [J]. Journal of Computer Applications, 2026, 46(4): 1050-1057.
[8]	Zhi JIANG, Xuebin CHEN, Changyin LUO, Ziye ZHEN. Hybrid optimization framework for improving Kolmogorov-Arnold network in federated learning [J]. Journal of Computer Applications, 2026, 46(4): 1023-1033.
[9]	Shengwei XU, Jianbo WANG, Jijie HAN, Yijie BAI. Face forgery detection method based on tri-branch feature extraction [J]. Journal of Computer Applications, 2026, 46(4): 1292-1299.
[10]	Jixin GUO, Ting ZHANG. Transformer image dehazing based on component collaborative optimization pruning [J]. Journal of Computer Applications, 2026, 46(3): 933-939.
[11]	Ping HUANG, Qing LI, Haifeng QIU, Chengsi WANG, Anzi HUANG, Long FAN. Lightweight method for transmission line defect detection [J]. Journal of Computer Applications, 2026, 46(3): 969-979.
[12]	Hanqing LIU, Guoming SANG, Yijia ZHANG. Remote sensing image captioning model combining dense multi-scale feature fusion and feature knowledge-enhanced Transformer [J]. Journal of Computer Applications, 2026, 46(3): 741-749.
[13]	Jian ZHANG, Jianbo YU, Jian TANG. Municipal solid waste incineration state recognition method based on multilayer preprocessing [J]. Journal of Computer Applications, 2026, 46(3): 940-949.
[14]	Jun WU, Chuan ZHAO. Small object detection method based on improved DETR algorithm [J]. Journal of Computer Applications, 2026, 46(2): 564-571.
[15]	Lijin YAO, Di ZHANG, Piyu ZHOU, Zhijian QU, Haipeng WANG. Transformer and gated recurrent unit-based de novo sequencing algorithm for phosphopeptides [J]. Journal of Computer Applications, 2026, 46(1): 297-304.