Deep learning-based electroacupuncture constant dose control method
By combining an impedance digital twin model and a tubular robust model predictive controller with a depth predistortion compensation network, the problem of constant dose control in electroacupuncture devices when facing impedance differences and dynamic changes is solved, achieving dose consistency and stability of treatment effect in multi-channel output.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANJING HOSPITAL OF TCM
- Filing Date
- 2026-06-12
- Publication Date
- 2026-07-17
AI Technical Summary
Existing electroacupuncture devices struggle to maintain constant current, cumulative charge, or effective energy when faced with differences in impedance between different patients and acupoints, as well as dynamic changes in impedance during treatment. Furthermore, multi-channel outputs are prone to dose inconsistencies and distortions, and the controller lacks explicit handling of model uncertainties and safety constraints.
Impedance digital twin model is used to predict time-varying impedance parameters and polarization state parameters. Combined with tubular robust model predictive controller and deep predistortion compensation network, a closed-loop control system is constructed. Through real-time acquisition and rolling optimization, pre-compensation drive signal is generated to maintain target dose consistency and suppress waveform distortion.
It achieves safety constraint protection during rapid impedance fluctuations or model mismatch, maintains dose consistency and treatment comfort of multi-channel output, reduces manual frequency adjustment, and reduces stimulation intensity drift and waveform distortion.
Smart Images
Figure CN122398620A_ABST