Deep neural networks in deep learning have been widely demonstrated to have higher accuracy and distinct advantages over traditional machine learning methods in extracting data features. While convolutional neural networks (CNNs) have shown great success in feature extraction and audio classification, it is important to note that real-time audios are dependent on previous scenes. Also, the main drawback of deep learning algorithms is that they need a huge number of datasets to indicate their efficient performance. In this paper, a recurrent neural network (RNN) combined with CNN is proposed to address this problem. Moreover, a Deep Convolutional Generative Adversarial Network (DCGAN) is used for high-quality data augmentation. This data augmentation technique is applied to the UrbanSound8K dataset to improve the environmental sound classification. Batch normalization, transfer learning, and three feature representations map are used to improve the model accuracy. The results show that the generated images by DCGAN have similar features to the original training images and has the capability to generate spectrograms and improve the classification accuracy. Experimental results on UrbanSound8K datasets demonstrate that the proposed CNN-RNN architecture achieves better performance than the state-of-the-art classification models.

This paper presents modeling and analysis of a piezoelectric mounted rotary flexible beam that can be used as an energy scavenger for rotary motion applications. The energy harvester system consists of a piezoelectric bimorph cantilever beam with a tip mass mounted on a rotating hub. Assuming Euler–Bernoulli beam equations and considering the effect of a piezoelectric transducer, equations of motion are derived using the Lagrangian approach followed by relationships describing the harvested power. The equations provide a quantitative description of how the hub acceleration and gravity due to the tip mass contribute power to the energy harvester. In particular, expressions describing optimum load resistance and the maximum power that can be harvested using the proposed system are derived. Numerical simulations are performed to show the performance of the harvester by obtaining tip velocities and electrical output voltages for a range of electrical load resistances and rotational speeds. It is shown that by proper sizing and parameter selection, the proposed system can supply enough energy for operating wireless sensors in rotating mechanisms such as tires and turbines.

This paper presents the design and performance analysis and experimental study of a 3-RRR spherical parallel manipulator in the context of hip exoskeleton applications. First, the mechanism’s inverse kinematics analysis and Jacobian matrix development are revisited. Manipulability, dexterity, and rotational sensitivity indices are then evaluated for two different methods of attachment to the human body. The superior attachment method in terms of these performance measures is indicated, and an experimental study based on the selected method is conducted; the experiment involves testing the capability of a 3-RRR manipulator’s end-effector in tracking the motions experienced by a human hip joint during normal gait cycles. Finally, the results of the experimental study indicate that the manipulator represents a feasible hip exoskeleton solution providing total kinematic compliance with the human hip joint’s 3-degree-of-freedom motion capabilities.

Engine mounts are used in the automotive industry to isolate engine and chassis by reducing the noise and vibration imposed from one to the other. This paper describes modelling, simulation and design of a semi-active engine mount that is designed specifically to address the complicated vibration pattern of variable displacement engines (VDE). The ideal isolation for VDE requires the stiffness to be switchable upon cylinder activation/deactivation operating modes. In order to have a modular design, the same hydraulic engine mount components are maintained and a novel auxiliary magneto-rheological (MR) fluid chamber is developed and retrofitted inside the pumping chamber. The new compliance chamber is a controllable pressure regulator, which can effectively alter the dynamic performance of the mount. Switching between different modes happens by turning the electrical current to the MR chamber magnetic coil on and off. A model has been developed for the passive hydraulic mount and then it is extended to include the MR auxiliary chamber as well. A proof-of-concept prototype of the design has been fabricated which validates the mathematical model. The results demonstrate unique capability of the developed semi-active mount to be used for VDE application.

To reduce the occurrence of wheelchair falls and to develop effective protection systems, we aimed to quantify sideways tip and fall dynamics of electric power wheelchairs (EPWs). We hypothesized that driving speed, curb height and angle of approach would affect impact forces and head injury risk for wheelchair riders. We further expected that fall dynamics and head injury risk would be greater for unrestrained riders compared to restrained riders. Sideways wheelchair tip and fall dynamics were reconstructed using a remotely operated rear wheel drive EPW and a Hybrid III test dummy driving at different approach angles (5 to 63°) over an adjustable height curb (0.30 to 0.41 m) at speeds of 0.6–1.5 m/s. Rigid body dynamics models (Madymo, TASS International, Livonia, MI) were developed in parallel with the experiments to systematically study and quantify the impact forces and the sideways tip or fall of an EPW user in different driving conditions. Shallower approach angles (25°) (p < 0.05) and higher curbs (0.4 m) (p < 0.05) were the most significant predictors of tipping for restrained passengers. Unrestrained passengers were most affected by higher curbs (0.4 m) (p < 0.005) and fell forward from the upright wheelchair when the approach angle was 60°. Head impact forces were greater in unrestrained users (6181 ± 2372 N) than restrained users (1336 ± 827 N) (p = 0.00053). Unrestrained users had significantly greater head impact severities than restrained users (HIC = 610 ± 634 vs HIC = 29 ± 38, p = 0.00013) and several tip events resulted in HICs > 1000 (severe head injury) in unrestrained users. Sideways tips and forward falls from wheelchairs were most sensitive to curb height and approach angle but were not affected by driving speed. Sideways tips and falls resulted in impact forces that could result in concussions or traumatic brain injury and require injury prevention strategies. Seat belts eliminated the risk of falling from an upright chair and reduced head impact forces in sideways wheelchair tips in this study; however, their use must be considered within the ethical and legal definitions of restraints.

This paper describes the design of a versatile and fully controllable active engine mount. The proposed active mount is capable of addressing vibration isolation requirements at various driving conditions. This design addresses a better ride quality that has always been demanded by the automotive industry, as well as satisfying sophisticated vibration isolation requirements for the unconventional engines, i.e. variable displacement, and hybrids. The proposed engine mount replaces the decoupler of the original design with a solenoid actuator. The mathematical model of the active mount is obtained. The dynamic characteristics of the mount are shown to be highly controllable over the operating frequency range of excitation in engines. The effectiveness of the developed active engine mount for various working conditions of engine is also evaluated. Several driving conditions are investigated and proper control strategies are utilized to demonstrate the mount's capability to fulfill the isolation requirements for each condition. The promising results, in addition to compactness, low cost, fail safety, and durability are the main advantages of the proposed active engine mount, which makes it viable for automotive applications.