Yanfang Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Yanfang Liu has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
Naiming Qi
7 shared publications- Naiming Qi
Harbin Institute of Technology
6 shared publications - Desong Du
School of Astronautics, Harbin Institute of Technology, Harbin 150001, China
3 shared publications - Jiayu She
Department of Aerospace Engineering, Harbin Institute of Technology, Harbin, China
3 shared publications - Chenrui Ni
Department of Aerospace Engineering, Harbin Institute of Technology, Harbin, China
2 shared publications Haiyan Duan
School of Astronautics, Harbin Institute of Technology, Harbin, China
2 shared publications
Latest Publications
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (7) , 8728-8735
Oscillation-Based Locomotion of a Spherical Robot Driven by Flywheels With Low Control Torque*
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (7) , 8728-8735
Oscillation-Based Locomotion of a Spherical Robot Driven by Flywheels With Low Control Torque*
- 7 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 171
Joint multi-target detection and single-target tracking framework for vehicles tracking based on UAV
- 7 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 171
Joint multi-target detection and single-target tracking framework for vehicles tracking based on UAV
- Open Access
- 5 November 2025
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 39 (5)
A unified framework for small object analysis in UAV imagery: Tracking single target from multiple detected targets
- 5 November 2025
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 39 (5)
Open AccessA unified framework for small object analysis in UAV imagery: Tracking single target from multiple detected targets
- 6 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (8) , 8522-8532
Adaptive Neural Indirect Inverse Control of Fractional-Order Rate-Dependent Hysteretic Nonlinear Systems and Its Application
Focusing on the high-precision motion control of nonlinear time-delay systems actuated by smart-material based actuators and motivated by flourishing development of fractional-order control theory, an adaptive fractional-order backstepping indirect inverse motion control (AFBIIMC) scheme is proposed to mitigate the rate-dependent hysteresis with the following features: 1) a new asymmetric rate-dependent hysteresis model, which can accurately depict the hysteresis in different smart-material based actuators such as magnetostrictive and dielectric elastomer actuators (DEA), is developed with the help of classical Prandtl–Ishlinskii hysteric model. 2) A novel hysteresis indirect inverse control algorithms of fractional-order hysteric nonlinear systems is designed, where the very difficult or even impossible work of construction of rate-dependent hysteresis direct inverse model are not required any more. Then, the rate-dependent hysteresis nonlinearity is effectively mitigated and the high-precision motion control performance is achieved. 3) The DEA motion platform is constructed, then, a new fractional-order model of the motion platform is proposed and motion control experiments are implemented to show the effectiveness of the proposed control scheme.
- 6 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (8) , 8522-8532
Adaptive Neural Indirect Inverse Control of Fractional-Order Rate-Dependent Hysteretic Nonlinear Systems and Its Application
Focusing on the high-precision motion control of nonlinear time-delay systems actuated by smart-material based actuators and motivated by flourishing development of fractional-order control theory, an adaptive fractional-order backstepping indirect inverse motion control (AFBIIMC) scheme is proposed to mitigate the rate-dependent hysteresis with the following features: 1) a new asymmetric rate-dependent hysteresis model, which can accurately depict the hysteresis in different smart-material based actuators such as magnetostrictive and dielectric elastomer actuators (DEA), is developed with the help of classical Prandtl–Ishlinskii hysteric model. 2) A novel hysteresis indirect inverse control algorithms of fractional-order hysteric nonlinear systems is designed, where the very difficult or even impossible work of construction of rate-dependent hysteresis direct inverse model are not required any more. Then, the rate-dependent hysteresis nonlinearity is effectively mitigated and the high-precision motion control performance is achieved. 3) The DEA motion platform is constructed, then, a new fractional-order model of the motion platform is proposed and motion control experiments are implemented to show the effectiveness of the proposed control scheme.
- 4 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (5) , 8010-8024
General Hamiltonian Neural Networks for Dynamic Modeling: Handling Sophisticated Constraints Automatically and Achieving Coordinates Free
Embedding the Hamiltonian formalisms into neural networks (NNs) enhances the reliability and precision of data-driven models, in which substantial research has been conducted. However, these approaches require the system to be represented in canonical coordinates, i.e., observed states should be generalized position-momentum pairs, which are typically unknown. This poses limitations when the method is applied to real-world data. Existing methods tackle this challenge through coordinate transformation or designing complex NNs to learn the symplectic phase flow of the state evolution. However, these approaches lack generality and are often difficult to train. This article proposes a versatile framework called general Hamiltonian NN (GHNN), which achieves coordinates free and handles sophisticated constraints automatically with concise form. GHNN employs two NNs, namely, an HNet to predict the Hamiltonian quantity and a JNet to predict the interconnection matrix. The gradients of the Hamiltonian quantity with respect to the input coordinates are calculated using automatic differentiation and are then multiplied by the interconnection matrix to obtain state differentials. Subsequently, ordinary differential equations (ODEs) are solved by numerical integration to provide state predictions. The accuracy and versatility of the GHNN are demonstrated through several challenging tasks, including the nonlinear simple and double pendulum, coupled pendulum, and real 3-D crane dynamic system.
- 4 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (5) , 8010-8024
General Hamiltonian Neural Networks for Dynamic Modeling: Handling Sophisticated Constraints Automatically and Achieving Coordinates Free
Embedding the Hamiltonian formalisms into neural networks (NNs) enhances the reliability and precision of data-driven models, in which substantial research has been conducted. However, these approaches require the system to be represented in canonical coordinates, i.e., observed states should be generalized position-momentum pairs, which are typically unknown. This poses limitations when the method is applied to real-world data. Existing methods tackle this challenge through coordinate transformation or designing complex NNs to learn the symplectic phase flow of the state evolution. However, these approaches lack generality and are often difficult to train. This article proposes a versatile framework called general Hamiltonian NN (GHNN), which achieves coordinates free and handles sophisticated constraints automatically with concise form. GHNN employs two NNs, namely, an HNet to predict the Hamiltonian quantity and a JNet to predict the interconnection matrix. The gradients of the Hamiltonian quantity with respect to the input coordinates are calculated using automatic differentiation and are then multiplied by the interconnection matrix to obtain state differentials. Subsequently, ordinary differential equations (ODEs) are solved by numerical integration to provide state predictions. The accuracy and versatility of the GHNN are demonstrated through several challenging tasks, including the nonlinear simple and double pendulum, coupled pendulum, and real 3-D crane dynamic system.
- Open Access
- 20 March 2024
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 37 (8) , 329-341
Feature-aided pose estimation approach based on variational auto-encoder structure for spacecrafts
- 20 March 2024
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 37 (8) , 329-341
Open AccessFeature-aided pose estimation approach based on variational auto-encoder structure for spacecrafts
- 31 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 27 (2) , 725-732
Learning Piezoelectric Actuator Dynamics Using a Hybrid Model Based on Maxwell-Slip and Gaussian Processes
Piezoelectric actuators (PEAs) are increasingly applied to the field of ultraprecision positioning; however, unwanted nonlinearities, such as hysteresis, degrade their performance. This article integrates a Gaussian process (GP) with a Maxwell-slip (MS) model to capture nonlinearities in PEAs. The GP predicts the output of a PEA or the error of the MS model by using the current and historical applied input voltages, as well as the measured historical output displacements, as input. The measured output is replaced by the output of the MS model in the case that a displacement sensor is unavailable. The experimental results show that by properly selecting system order, the proposed approach achieves a precise result and provides the boundary of the prediction error in terms of variance. If the output displacement is available and the GP predicts the output of a PEA directly, the normalized root-mean-square error (NRMSE) is as low as 0.29%. The NRMSE is reduced further to 0.14% if the dataset used for prediction is updated online.
- 31 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 27 (2) , 725-732
Learning Piezoelectric Actuator Dynamics Using a Hybrid Model Based on Maxwell-Slip and Gaussian Processes
Piezoelectric actuators (PEAs) are increasingly applied to the field of ultraprecision positioning; however, unwanted nonlinearities, such as hysteresis, degrade their performance. This article integrates a Gaussian process (GP) with a Maxwell-slip (MS) model to capture nonlinearities in PEAs. The GP predicts the output of a PEA or the error of the MS model by using the current and historical applied input voltages, as well as the measured historical output displacements, as input. The measured output is replaced by the output of the MS model in the case that a displacement sensor is unavailable. The experimental results show that by properly selecting system order, the proposed approach achieves a precise result and provides the boundary of the prediction error in terms of variance. If the output displacement is available and the GP predicts the output of a PEA directly, the normalized root-mean-square error (NRMSE) is as low as 0.29%. The NRMSE is reduced further to 0.14% if the dataset used for prediction is updated online.
- 13 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 68 (10) , 10050-10057
Hybrid Model Based on Maxwell-Slip Model and Relevance Vector Machine
This article proposes a hybrid model based on Maxwell-slip (MS) model and relevance vector machine (RVM) for piezoelectric actuators (PEAs). The MS model maps the input voltage into the internal state space and provides an estimation of the output displacement. The output function of the MS model provides a linear relationship between the deformation vector and the output displacement. Replacing the regression function of the RVM by the output function of the MS model enables the RVM to update parameters of the MS model. The RVM is chosen since it is also able to provide the distribution of the estimated output displacement. Integration of the MS model with the RVM also reduces the dimension of the RVM from the number of support data points to the number of MS elements, which will reduce the computational time. Results demonstrate that the hybrid model is capable of describing rate-dependent and asymmetric hysteresis, is robust to excitations, enhances the accuracy of the MS model, and requires much shorter computational time than the standalone RVM. The modeling error under random excitation is 0.54% and for asymmetric hysteresis in a reluctance actuator is 0.31%. Compensation for hysteresis is carried out by connecting the inverse model in serial with a PEA and achieves tracking error of less than 0.81%.
- 13 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 68 (10) , 10050-10057
Hybrid Model Based on Maxwell-Slip Model and Relevance Vector Machine
This article proposes a hybrid model based on Maxwell-slip (MS) model and relevance vector machine (RVM) for piezoelectric actuators (PEAs). The MS model maps the input voltage into the internal state space and provides an estimation of the output displacement. The output function of the MS model provides a linear relationship between the deformation vector and the output displacement. Replacing the regression function of the RVM by the output function of the MS model enables the RVM to update parameters of the MS model. The RVM is chosen since it is also able to provide the distribution of the estimated output displacement. Integration of the MS model with the RVM also reduces the dimension of the RVM from the number of support data points to the number of MS elements, which will reduce the computational time. Results demonstrate that the hybrid model is capable of describing rate-dependent and asymmetric hysteresis, is robust to excitations, enhances the accuracy of the MS model, and requires much shorter computational time than the standalone RVM. The modeling error under random excitation is 0.54% and for asymmetric hysteresis in a reluctance actuator is 0.31%. Compensation for hysteresis is carried out by connecting the inverse model in serial with a PEA and achieves tracking error of less than 0.81%.
- 1 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 36-41
Hybrid Model Based on the Maxwell-Slip Model and a Support Vector Machine for Hysteresis in Piezoelectric Actuators
Hysteresis in piezoelectric actuators (PEAs) degrades the positioning accuracy and requires suppression. This paper models the hysteresis via a hybrid approach, which combines the Maxwell-slip (MS) model and a least-squares support vector machine (LS-SVM). The MS model maps the input voltage into internal state space and provides an estimation of the output displacement. The estimated displacement then serves as an exogenous input, and a nonlinear regression model is formulated via the LS-SVM approach to predict the current output displacement. Once parameters are obtained, the above hybrid model does not require a sensor to measure the output displacement. Hysteresis compensation based on the inverse hybrid model is also presented. The experimental results show that the hybrid approach enhances the performance and is robust to excitations. The modeling error is less than 0.66%, and the tracking error of the compensated system is less than 1.2%.
- 1 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 36-41
Hybrid Model Based on the Maxwell-Slip Model and a Support Vector Machine for Hysteresis in Piezoelectric Actuators
Hysteresis in piezoelectric actuators (PEAs) degrades the positioning accuracy and requires suppression. This paper models the hysteresis via a hybrid approach, which combines the Maxwell-slip (MS) model and a least-squares support vector machine (LS-SVM). The MS model maps the input voltage into internal state space and provides an estimation of the output displacement. The estimated displacement then serves as an exogenous input, and a nonlinear regression model is formulated via the LS-SVM approach to predict the current output displacement. Once parameters are obtained, the above hybrid model does not require a sensor to measure the output displacement. Hysteresis compensation based on the inverse hybrid model is also presented. The experimental results show that the hybrid approach enhances the performance and is robust to excitations. The modeling error is less than 0.66%, and the tracking error of the compensated system is less than 1.2%.
- Open Access
- 27 February 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 40901-40908
Modeling and Identification of Temperature-Dependent Hysteresis in Piezoelectric Materials Considering Parameter Sensitivity
Hysteresis is an important nonlinearity effect in piezoelectric materials and is sensitive to temperature. This paper extends the saturated capacitor model to capture temperature-dependent hysteresis by replacing the constant parameters with functions of temperature. To obtain these functions, several sets of constant parameters are identified at different temperatures, and then each term of the parameters is fitted as a function with respect to temperature. In the curve fitting process, the parameter sensitivity is employed as a weight factor. The effectiveness of the proposed modeling and parameter determination procedure is verified with experimental data. The overall normalized root mean square error is approximately 5%, including 3% error caused by the identified parameters.
- 27 February 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 40901-40908
Open AccessModeling and Identification of Temperature-Dependent Hysteresis in Piezoelectric Materials Considering Parameter Sensitivity
Hysteresis is an important nonlinearity effect in piezoelectric materials and is sensitive to temperature. This paper extends the saturated capacitor model to capture temperature-dependent hysteresis by replacing the constant parameters with functions of temperature. To obtain these functions, several sets of constant parameters are identified at different temperatures, and then each term of the parameters is fitted as a function with respect to temperature. In the curve fitting process, the parameter sensitivity is employed as a weight factor. The effectiveness of the proposed modeling and parameter determination procedure is verified with experimental data. The overall normalized root mean square error is approximately 5%, including 3% error caused by the identified parameters.