A Braille Reading System Based on Electrotactile Display With Flexible Electrode Array

Dear editor,Due to visual impairment and lack of social attention,the reading problem of the blind has not been well resolved.Many existing Braille reading devices still have many problems and are not accepted by the public.This paper developed a portable Braille reading system based on electrotactile display technology.The proposed system is composed of a six-channel electrotactile stimulator,a flexible electrode array for Braille display and a graphical user interface(GUI)for monitoring and control.Based on the flexibility of the system,two Braille reading strategies have been designed:spatial mode and sequential mode.The system was preliminary tested by six sighted subjects.The results showed that subjects were able to recognize Braille characters with more than 80%accuracy within less than 10 seconds for the initial use.Compared to existing Braille reading systems,this system is portable,wearable and flexible in control.

A Semi-Discretizing Method Based Efficient Model for Fluidelastic Instability Threshold Prediction of Tube Bundles

Fluidelastic instability is destructive in tube bundles subjected to cross flow.Flow channel model proposed by Leaver and Weaver is well used for modeling this problem.However,as the tube motion is supposed to be harmonic,it may not simulate the general dynamic behaviors of tubes.To improve this,a model with arbitrary tube motion is proposed by Hassan and Hayder.While,due to involving in the time delay term,the stability problem cannot be solved by the eigenvalue scheme,and time domain responses of the tube have to be obtained to assess the instability threshold.To overcome this weakness,a new approach based on semi-discretizing method(SDM)is proposed in this study to make the instability threshold be predicted by eigenvalues directly.The motion equation of tube is built with considering the arbitrary tube motion and the time delay between fluid flow and tube vibration.A time delay integral term is derived and the SDM is employed to construct a transfer matrix,which transforms the infinite dimensional eigenvalue problem into a finite one.Hence the stability problem become solvable accordingly.With the proposed method,the instability threshold of a typical square tube array model is predicted,and the influences of system parameters on stability are also discussed.With comparing with prior works,it shows significant efficiency improvement in prediction of the instability threshold of tube bundles.

Precise in-situ characterization and cross-validation of the electromagnetic properties of a switched reluctance motor

The electrical-magnetic characteristics of a Switched Reluctance Motor(SRM)exhibit highly nonlinear relationshipwith respect to the rotor position and excitation current,which poses challenges for both precise static measurements and exact calculation of these properties in real-time control.To guarantee that an in-lab test result can be used in the application,firstly a measurementmethod is proposed to characterize the SRM's electromagnetic properties such as the flux linkage,magnetic co-energy,phase inductance and electromagnetic torque on the basis of an installed SRM control circuitry and half-bridge power converter.By this means the characterization process is equivalent to the online observation in its results.Secondly,a theoreticalmodel is built to discriminate the physical meaning between the incremental inductance and the phase inductance,which is the origin of other relevant parameters.This helps to guide the correct utilization of the characterization result.Thirdly an in-situ cross-validation experimentation according to the magnetizing and demagnetizing status measurement verifies the feasibilities and accuracy of the proposed inductance measuring method,which avoid a dubious FEM-based comparison between the numerical calculation and experimental results.Cross-validation experiment shows that the proposed in-situ characterization scheme obtains an accurate full-range electromagnetic properties.The proposed methodology breaks the barrier between the in-lab measurement and on-line utilization of the SRM parameters,highlighting the merits that it completely includes the in-situ factors and replicates the operational scenario without the need of specifically designed instrumentation,which is especially suitable for rapid field characterization for high power motors.

Continual learning fault diagnosis:A dual-branch adaptive aggregation residual network for fault diagnosis with machine increments

As a data-driven approach, Deep Learning(DL)-based fault diagnosis methods need to collect the relatively comprehensive data on machine fault types to achieve satisfactory performance. A mechanical system may include multiple submachines in the real-world. During condition monitoring of a mechanical system, fault data are distributed in a continuous flow of constantly generated information and new faults will inevitably occur in unconsidered submachines, which are also called machine increments. Therefore, adequately collecting fault data in advance is difficult. Limited by the characteristics of DL, training existing models directly with new fault data of new submachines leads to catastrophic forgetting of old tasks, while the cost of collecting all known data to retrain the models is excessively high. DL-based fault diagnosis methods cannot learn continually and adaptively in dynamic environments. A new Continual Learning Fault Diagnosis method(CLFD) is proposed in this paper to solve a series of fault diagnosis tasks with machine increments. The stability–plasticity dilemma is an intrinsic issue in continual learning. The core of CLFD is the proposed Dual-branch Adaptive Aggregation Residual Network(DAARN).Two types of residual blocks are created in each block layer of DAARN: steady and dynamic blocks. The stability–plasticity dilemma is solved by assigning them with adaptive aggregation weights to balance stability and plasticity, and a bi-level optimization program is used to optimize adaptive aggregation weights and model parameters. In addition, a feature-level knowledge distillation loss function is proposed to further overcome catastrophic forgetting. CLFD is then applied to the fault diagnosis case with machine increments. Results demonstrate that CLFD outperforms other continual learning methods and has satisfactory robustness.