Search

The paper presents a new mechanism for an index finger rehabilitation device. The mechanism was designed based on a pre-specified trajectory obtained by performing flexion/extension experimental trials with index finger. These experiments were performed 30 times and mean of the normalized trajectory was extracted. From this trajectory, six precision points were selected and based on the obtained trajectory, a path generating optimization problem was formulated to imitate the flexion/extension. Besides, rectification constraints were posed to avoid defects that typically encounter during synthesis. The formulated nonlinear optimization problem was solved using nature-inspired and metaphor-less algorithms. It is found that the summed error between the desired and generated trajectories was very less in case of nature-inspired algorithm in comparison with metaphor-less algorithm. Then, all rectification constraints were validated. It is found that obtained mechanism is completely defect-free and is able to imitate the desired flexion/extension trajectory. Therefore, design obtained using nature-inspired algorithm may prove to be beneficial for the patients with dysfunction index finger.

In this paper, an intelligent approach of adaptive neuro-fuzzy inference system (ANFIS) and non-dominated sorting genetic algorithm-II (NSGA-II) was delineated to establish model and optimize the wire electrical discharge machining (WEDM) process. The WEDM experiments were designed utilizing Taguchi L18 mixed orthogonal array for machining of Ti–6Al–4V titanium alloy. The ANFIS model was delineated to explain the influence of input machining characteristics, viz. peak current (IP), pulse on-time (Ton), pulse off-time (Toff) and wire feed (WF) on output response of material removal rate (MRR) and wire wear ratio (WWR). The proximity of results with confirmation experimental results revealed the effectiveness of the developed ANFIS model in prediction of output quality characteristics for the chosen input machining factors. The artificial neural network (ANN) and NSGA-II were integrated and applied for multi-objective optimization in determining optimal WEDM machining process conditions. The optimal results obtained with NSGA-II for multi-objective optimization of input control variables are within tolerance limits and realized an improvement in MRR and WWR (maximum absolute percentage errors as 6.784% and 7.589%) with optimal machining characteristics.

The bat algorithm (BA) is one of prominent swarm-based algorithm that has the suitability in solving only small dimension engineering problems and suffers from drawback of getting trapped in local minimum with slow convergence for multi-dimensional problems. In the context of improving its applicability in solving large scale and constrained engineering design problems, this paper presents a novel upgraded bat algorithm with cuckoo search and Sugeno inertia weight (UBCSIW). In the proposed UBCSIW algorithm, first, the bat algorithm with its competence to exploit the optimal solutions in search space is combined with cuckoo search with its ability to explore best solution globally using Levy flight in the search space. Secondly, a new velocity and position search equation is incorporated in which the bat searches around the best candidate solution. This step helps in establishing adequate balance between exploration and exploitation capability and improving the performance effectively by employing greedy selection to choose the best candidate solution. Finally, Sugeno fuzzy inertia weight is introduced in the velocity updation equation, boosting the flexibility and diversity of bat population and results in stability of results. The effectiveness of the proposed UBCSIW algorithm is tested on 16 standard benchmark functions (unimodal and multimodal) with different dimensions, 12 CEC2015 test functions and 7 well-known constrained engineering design problems. The outputs of the proposed UBCSIW algorithm are validated by comparison with classical BA and other swarm-based state-of-the art algorithms. The simulation results show that proposed UBCSIW algorithm achieves highly competitive results in terms of higher optimization accuracy and improved convergence that outperforms basic BA in all twenty-eight test functions while performs better than other competitive algorithms in 24 functions (13 benchmark and 11 CEC2015 functions).

In this paper, a fuzzy logic-based artificial intelligence technique is delineated to optimize surface reconstruction parameters, initiating from raw point cloud data using contactless scanning of transtibial prosthetic socket. The prosthetic socket having a free-form surface is considered for reconstruction of an accurate and complete digital model, which is further crucial for subsequent design, analysis and production. The input factors considered for the socket surface reconstruction includes number of points, filtering, triangle counts and smoothing level. The experimental trials are performed using face-centered central composite design, and further experimental data are used to establish Mamdani fuzzy logic model to predict the surface standard deviation and required space in computer memory as the response parameters. The confirmation experiment results reveal that the fuzzy model shows good agreement of 96.53% with the experimental measured value. Based on the results, it is proved that established fuzzy model can be used for predicting the surface reconstruction parameters to improve the final accuracy of the transtibial prosthetic socket model. The results concluded that the fuzzy logic provides commendable results; in addition, it successfully predicts the effect of different parameters considered for surface reconstruction on surface accuracy and computer file size.