Electromechanical actuator (EMA) exhibits advanced performance in industry, but its dynamic servo responses are constrained by parametric perturbations, load torque variations, and measurement noise. A strong disturbance rejection ability is necessary for EMAs. However, this usually makes them more sensitive to the measurement noise, reducing the steady-state precision. In this article, an adaptive linear active disturbance rejection control (LADRC) controller is proposed to achieve strong antidisturbance performance and reduce noise sensitivity for EMAs. A novel parallel structure is proposed to improve dynamic responses, which replaces the traditional cascade structure of position and speed loops. Aiming to improve the antidisturbance performance, a linear full-order-extended state observer is integrated with the parallel controller, called the LADRC controller. To reduce the difficulty of parameter tuning, the number of tuning parameters of LADRC is reduced to two by a pole placement design. And these two parameters of LADRC can be adjusted adaptively by the hyperbolic tangent function. Finally, the simulation and experimental results are provided to verify the effectiveness of the proposed strategy for EMAs.

Adaptive LADRC-Based Disturbance Rejection Method for Electromechanical Servo System

A practical nonlinear adaptive control of hydraulic servomechanisms with periodic-like disturbances

A traditional linear extended state observer (LESO) does not have enough capacity to handle the rapidly-varying back electromotive force (EMF). This key weakness limits the application of LESO in the internal permanent magnet synchronous motor (IPMSM) sensorless control system. In this article, an IPMSM sensorless drive method using an improved LESO (ILESO) is proposed. The ILESO consists of a pure integral and rapidly-varying sinusoidal interference estimator. The rapidly-varying sinusoidal interference estimator is added to the unknown interference estimation loop to estimate rapidly-varying back EMF. And the frequency of the rapidly-varying sinusoidal interference estimator changes adaptively with the operating frequency of the motor to guarantee the desired estimation accuracy of back EMF in a wide-speed range. This design makes it possible to observe the back EMF in a relatively low bandwidth, thereby solving the tradeoff between LESO bandwidth and high-frequency noise filtering. Then, an enhanced phase-locked loop is presented to improve the ability of tracking the rapidly-changing speed. The effectiveness of the proposed method is verified at a 2.0-kW IPMSM sensorless drive.

A Rotor Position and Speed Estimation Method Using an Improved Linear Extended State Observer for IPMSM Sensorless Drives

In the permanent magnet synchronous motor speed regulation system, the dynamic performance of the conventional active disturbance rejection control (ADRC) system will be deteriorated by using a low bandwidth speed filter. To solve this problem, two ADRC controllers considering the speed measurement noise are proposed in this article. One proposed ADRC system is based on the extended state observer (ESO), the other proposed ADRC system is based on phase-locking loop observer (PLLO). In the proposed two ADRC systems, an integrator is employed as the speed filter so that the measured position can be directly used for observing the speed and disturbance without speed calculation. Meanwhile, by using an integrator as the speed filter, the dynamic performance of the proposed ESO-based ADRC system is not affected since it is independent of the speed filter, and the proposed PLLO-based ADRC system has a better rejection ability for the low-frequency disturbance. Experimental results validate the proposed methods.

/pdf/linear-active-disturbance-rejection-controllers-for-pmsm-xbiugduya4.pdf

Linear Active Disturbance Rejection Controllers for PMSM Speed Regulation System Considering the Speed Filter

Adaptive super-twisting trajectory tracking control for an unmanned aerial vehicle under gust winds

In order to improve the disturbance rejection performance of electro-mechanical actuator (EMA), a robust servo system based on full-order extended state observer (ESO) is proposed for EMA employing permanent magnet synchronous motor (PMSM). Firstly, dynamic model of EMA is analyzed. Parameter uncertainties and load variation are treated as total disturbances. Utilizing this model, a parallel composite control of position and speed with command shaping is presented to improve the frequency bandwidth of EMA. Secondly, a full-order ESO is designed to estimate total disturbances. Thirdly, a simple parameter tuning method is given, which is convenient to analyze and implement the controller. The disturbance sensitivity analysis is realized in frequency domain. Finally, the simulation and experimental results verify the effectiveness of proposed strategy for EMA.

Full-order ESO Based Disturbance Rejection Method With a Simple Parameter Tuning for Electro-Mechanical Servo System

Tensor learning is a powerful tool for big data analytics and machine learning, e.g., gene analysis and deep learning. However, tensor learning algorithms are compute-intensive since their time and space complexities grow exponentially with the order of tensors, which hinders their application. In this paper, we exploit the parallelism of tensor learning primitives using GPU tensor cores and develop high-performance tensor learning algorithms. First, we propose novel hardware-oriented optimization strategies for tensor learning primitives on GPU tensor cores. Second, for big data analytics, we employ the optimized tensor learning primitives to accelerate the CP tensor decomposition and then apply it for gene analysis. Third, we optimize the Tucker tensor decomposition and propose a novel Tucker tensor layer to compress deep neural networks. We employ natural gradients to train the neural networks, which only involve a forward pass without backpropagation and thus are suitable for GPU computations. Compared with TensorLab and TensorLy libraries on an A100 GPU, our third-order CP tensor decomposition achieves up to <inline-formula><tex-math notation="LaTeX">$16.32\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>16</mml:mn><mml:mo>.</mml:mo><mml:mn>32</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq1-3222955.gif"/></alternatives></inline-formula> and <inline-formula><tex-math notation="LaTeX">$32.25\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>32</mml:mn><mml:mo>.</mml:mo><mml:mn>25</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq2-3222955.gif"/></alternatives></inline-formula> speedups; and <inline-formula><tex-math notation="LaTeX">$6.09\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>6</mml:mn><mml:mo>.</mml:mo><mml:mn>09</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq3-3222955.gif"/></alternatives></inline-formula> and <inline-formula><tex-math notation="LaTeX">$6.72\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>6</mml:mn><mml:mo>.</mml:mo><mml:mn>72</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq4-3222955.gif"/></alternatives></inline-formula> speedups for our third-order Tucker tensor decomposition. The proposed fourth-order CP and Tucker tensor decompositions achieve up to <inline-formula><tex-math notation="LaTeX">$30.65\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>30</mml:mn><mml:mo>.</mml:mo><mml:mn>65</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq5-3222955.gif"/></alternatives></inline-formula> and <inline-formula><tex-math notation="LaTeX">$5.41\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>5</mml:mn><mml:mo>.</mml:mo><mml:mn>41</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq6-3222955.gif"/></alternatives></inline-formula> speedups over the TensorLab. Our CP tensor decomposition for gene analysis achieves up to <inline-formula><tex-math notation="LaTeX">$5.88\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>5</mml:mn><mml:mo>.</mml:mo><mml:mn>88</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq7-3222955.gif"/></alternatives></inline-formula> speedup over TensorLy. Compared with a conventional fully connected neural network, our Tucker tensor layer neural network achieves an accuracy of <inline-formula><tex-math notation="LaTeX">$97.9\%$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>97</mml:mn><mml:mo>.</mml:mo><mml:mn>9</mml:mn><mml:mo>%</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq8-3222955.gif"/></alternatives></inline-formula>, a speedup of <inline-formula><tex-math notation="LaTeX">$4.47\times$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>4</mml:mn><mml:mo>.</mml:mo><mml:mn>47</mml:mn><mml:mo>×</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq9-3222955.gif"/></alternatives></inline-formula>, and a compression ratio of <inline-formula><tex-math notation="LaTeX">$2.92$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>2</mml:mn><mml:mo>.</mml:mo><mml:mn>92</mml:mn></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq10-3222955.gif"/></alternatives></inline-formula> at the cost of <inline-formula><tex-math notation="LaTeX">$0.4\%$</tex-math><alternatives><mml:math><mml:mrow><mml:mn>0</mml:mn><mml:mo>.</mml:mo><mml:mn>4</mml:mn><mml:mo>%</mml:mo></mml:mrow></mml:math><inline-graphic xlink:href="liu-ieq11-3222955.gif"/></alternatives></inline-formula> drop in accuracy.

High-Performance Tensor Learning Primitives Using GPU Tensor Cores

Noise injection-based method has been shown to be able to improve the robustness of artificial neural networks in previous work. In this work, we propose a novel noise injection-based training scheme for better model robustness. Specifically, we first develop a likelihood ratio method to estimate the gradient with respect to both synaptic weights and noise levels for stochastic gradient descent training. Then, we design an approximation for the vanilla noise injection-based training method to reduce memory and improve computational efficiency. Next, we apply our proposed scheme to spiking neural networks and evaluate the performance of classification accuracy and robustness on MNIST and Fashion-MNIST datasets. Experiment results show that our proposed method achieves a much better performance on adversarial robustness and slightly better performance on original accuracy, compared with the conventional gradient-based training method.

A Novel Noise Injection-based Training Scheme for Better Model Robustness

/pdf/online-calibrated-energy-aware-and-heading-corrected-23nuzogv.pdf

Zeliang Zhang

Papers

Adaptive LADRC-Based Disturbance Rejection Method for Electromechanical Servo System

Full-order ESO Based Disturbance Rejection Method With a Simple Parameter Tuning for Electro-Mechanical Servo System

High-Performance Tensor Learning Primitives Using GPU Tensor Cores

A Novel Noise Injection-based Training Scheme for Better Model Robustness

Online calibrated, energy-aware and heading corrected pedestrian navigation with foot-mounted MARG sensors