Self-sensing Estimation Research Articles

A comparative analysis between deep neural network-based 1D-CNN and LSTM models to harness the self-sensing property of the shape memory alloy wire actuator for position estimation

The article presents a performance-based comparative analysis of popular deep neural network (DNN) models such as 1-dimensional convolutional neural network (1D-CNN) and long short-term memory (LSTM) for position estimation of shape memory alloy (SMA)-based wire actuator. These DNN models utilize the self-sensing property (SSP) for position estimation of the SMA actuator. The phase-dependent electrical resistivity of SMA wire acts as SSP, where the electrical resistivity in the form of SMA wire resistance acts as inputs to the proposed models for precise estimation of the current position of the SMA actuator. For effective position control of the SMA actuator, accurate position sensor feedback is required, utilizing SSP results in the elimination of this external sensor. This will improve the overall system in terms of compactness and reduced interface complexity. Coming to DNN models, 1D-CNN has been meagerly explored in the current literature landscape for self-sensing estimation of SMA actuators. These 1D-CNN models are becoming quite popular for time series prediction for various applications and are emerging as an alternative to widely used LSTM models. In this paper, a novel implementation of a 1D-CNN model for SMA actuator position estimation has been done. A comparative analysis between 1D-CNN and LSTM has been done for prediction capability and inference speed based on performance measures such as Mean Square Error (MSE), Mean Absolute Error (MAE), symmetric Mean Absolute Percentage Error (sMAPE), data distribution, and average inference speed. The proposed comparative results show that 1D-CNN has a matching performance with the LSTM model with respect to prediction capability, however, 1D-CNN offers faster inference speed. The analysis of the proposed work can be useful for choosing a suitable DNN model for deployment on low computing platforms such as microcontrollers for SMA actuator-based real-time applications where time latency is a critical parameter.

Image Registration Technique to Improve a Self-Sensing Estimation for Model Predictive Controlled Interior PMSM Drives

Current responses based self-sensing control estimates rotor position by detecting magnetic saliency of a machine. However, the estimation error can be resulted from several nonideal factors, especially magnetic cross-saturation, as effects of these factors show up in current responses. This article applies image registration technique to improve the accuracy of a self-sensing estimation for model predictive control (MPC) interior permanent magnet synchronous machines (IPMSM) drives by minimizing impacts of nonideal factors on measured current responses. In commissioning process, two images are generated. The inverter voltage vectors in MPC excite current responses, represented by a set of vectors that generate a response image. Nonideal factors that result in estimation errors also deform the response image. A primitive image based on resolver output can be considered as a template. An image registration technique is investigated to register deformed images with corresponding primitive images, and transformations contain operations of rotation, scale, translation, and shear can be derived. During the drive, the deformed pixel points are online mapped to new pixel points by applying these operations. The current responses after transformation are less subject to nonideal factors. As a result, impacts of these deformation factors that contribute to estimation errors can be reduced. Finally, the application of image registration technique to the self-sensing estimation is verified on a 3.7-kW IPMSM drive.

Self-Sensing Estimation of Mechanical Stress in Magnetostrictive Actuators

Magnetostrictive actuators show interesting performances related to high-precision actuation with high-energy density, which focused the interest of research in the past years. Nevertheless, the output deformation is related to the current and the applied stress, through a complex multi-input multi-output (MIMO) hysteresis process. As a consequence, a reliable and accurate control task cannot neglect the stress monitoring. Such task, if performed by an ad hoc stress sensor, e.g., a load cell, will face with the sensor’s placement in the device, resulting in an increase in cost, complexity, and bulkiness of the system. This issue can be circumvent if the current and magnetic flux measurements, available in the control chain, are exploited. To this aim, a novel self-sensing approach to deliver real-time estimation of the stress is presented in this paper, involving some recent results of a thermodynamic compatible MIMO model of hysteresis for magnetostrictive materials.