The objective of this paper is to develop a manual measurement system (MMS) for the Lachman test using stereo markers. A novel calculation method that is fit for stereo markers is proposed to analyze knee joint motion in real-time based on the extraction of markers attached on the femur and tibia. In our experiments, knee extension movement and tibial translation are performed with imitation bones to evaluate the accuracy of the system. Further, a simulation of the Lachman test is performed in vivo measurement. The mean error of the knee extension movement in ten cases (range 0° to 90°) was 0.41° with a standard deviation of 0.44°. The mean error of the tibial translation was approximately 0.3 ± 0.9 mm. Experimental results confirmed the acceptable performance of the proposed measurement system, which can be considered for application in clinical manual tests.
In this paper, we propose a novel approach toward the development of a perceptual color space, FHSI, which stands for “Fuzzy HSI,” because it is based on the fuzzification of the well-known HSI color space. FHSI represents a set of fuzzy colors obtained by partitioning the gamut of feasible colors in the HSI model corresponding to standardized linguistic tags. In fact, color categorization was performed on the basis of personal judgments of humans collected by way of an online survey. This approach helps to significantly enhance color matching and similarity searches by producing more intuitive and human-consistent output for users. The introduced method has potential for use in various color image applications involving query processing, for example, in the coordination of online apparel shopping.
Protein structural class prediction is beneficial to study protein function, regulation and interactions. However, protein structural class prediction for low-similarity sequences (i.e., below 40% in pairwise sequence similarity) remains a challenging problem at present. In this study, a novel computational method is proposed to accurately predict protein structural class for low-similarity sequences. This method is based on support vector machine in conjunction with integrated features from evolutionary information generated with position specific iterative basic local alignment search tool (PSI-BLAST) and predicted secondary structure. Various prediction accuracies evaluated by the jackknife tests are reported on two widely-used low-similarity benchmark datasets (25PDB and 1189), reaching overall accuracies 89.3% and 87.9%, which are significantly higher than those achieved by state-of-the-art in protein structural class prediction. The experimental results suggest that our method could serve as an effective alternative to existing methods in protein structural classification, especially for low-similarity sequences.
Accurate time-to-go estimation with large heading angle error is difficult for homing guidance laws, especially for the impact time control. Considering this, a new cooperative guidance law which requires no time-to-go estimation is investigated. First, the impact time control problem is transformed to the look angle command tracking problem. The look angle command guarantees that the range-to-go error converges to zero asymptotically. Then the proposed guidance law considering the seeker’s field-of-view constraint is derived using sliding mode control to track the desired look angle signal. Numerical simulations are performed to verify the effectiveness of the proposed guidance law for one-to-one and many-to-one engagement scenarios.
Isolated power systems (IPSs) worldwide are traditionally powered by diesel generators that are very expensive to run and produce harmful emissions. In order to mitigate these problems, wind turbines are being introduced into existing IPSs. Although this integration has been reasonably effective at reducing running costs and emissions, high levels of wind penetration cause large system frequency variations, resulting in a prolonged synchronization process for newly dispatched diesel generators. Long synchronization can compromise the stability of a small IPS. This paper examines the diesel synchronization problem using a real IPS as a case study and offers a solution by introducing the concept of predictive synchronization based on adaptive neuro-fuzzy systems. Simulation results demonstrate a significant reduction in diesel generator synchronization times.
This study proposes a new real-coded genetic algorithm (RCGA) taking account of extrapolation, which we call adaptive extrapolation RCGA (AEGA). Real-world problems are often formulated as black-box function optimization problems and sometimes have ridge structures and implicit active constraints. mAREX/JGG is one of the most powerful RCGAs that performs well against these problems. However, mAREX/JGG has a problem of search inefficiency. To overcome this problem, we propose AEGA that generates offspring outside the current population in a more stable manner than mAREX/JGG. Moreover, AEGA adapts the width of the offspring distribution automatically to improve its search efficiency. We evaluate the performance of AEGA using benchmark problems and show that AEGA finds the optimum with fewer evaluations than mAREX/JGG with a maximum reduction ratio of 45%. Furthermore, we apply AEGA to a lens design problem that is known as a difficult real-world problem and show that AEGA reaches the known best solution with approximately 25% fewer evaluations than mAREX/JGG.
A stand-alone doubly fed induction generator (DFIG)-based wind power generation system using a third-harmonic injection indirect matrix converter (THIIMC) is proposed. The THIIMC has the same performance of a back-to-back pulse width modulation converter, but does not require the bulky direct current (dc)-link capacitor. Because of both its compact construction and high reliability, it is very suitable for embedding into DFIG-based wind generators. It also overcomes the drawbacks of indirect matrix converters and improves the reactive power output capability. The THIIMC consists of a rectifier-side converter, an inverter-side converter (ISC), and an active third-harmonic current injection circuit. A direct stator voltage vector control scheme for the ISC provides the desired stator voltage to the loads. The control scheme is designed to compensate the reactive power of the loads based on the THIIMC working principle. Maximum power point tracking control is performed by a battery energy storage system, which is placed in the dc-link of the THIIMC to smooth out the power fluctuations caused by load or wind speed variations. Simulation results demonstrate the performance and feasibility of the proposed topology and control scheme.
Arthroscopic surgery is a minimally invasive surgical procedure that is widely used on joints. However, conventional endoscope-based arthroscopic surgery does not provide stereoscopic information to the surgeon. To overcome this limitation, we have developed a modified endoscopic system that uses an optical fiber (1 mm diameter) for three-dimensional imaging of knee joints. Our endoscopic system is able to operate underwater in real time. It consists of a laser beam that is projected on the surface of the object to be imaged via an optical fiber. A prism is used to decrease the length and diameter of baseline and endoscope tube, respectively. The small diameter (8 mm) of our endoscope makes it extremely convenient for application in arthroscopic surgery. The feasibility of the proposed system has been demonstrated via experiments on synthetic knee joints.
One promising approach to pixel-wise semantic segmentation is based on conditional random fields (CRFs). CRF-based semantic segmentation requires ground-truth annotations to supervisedly train the classifier that generates unary potentials. However, the number of (public) annotation data for training is limitedly small. We observe that the Internet can provide relevant images for any given keywords. Our idea is to convert keyword-related images to pixel-wise annotated images, then use them as training data. In particular, we rely on saliency filters to identify the salient object (foreground) of a retrieved image, which mostly agrees with the given keyword. We utilize saliency information for back-and-foreground CRF-based semantic segmentation to further obtain pixel-wise ground-truth annotations. Experiment results show that training data from Google images improves both the learning performance and the accuracy of semantic segmentation. This suggests that our proposed method is promising for harvesting substantial training data from the Internet for training the classifier in CRF-based semantic segmentation.
Stroke is one of the leading causes worldwide of motor disability in adults. Motor imagery is a rehabilitation technique for potentially treating the results of stroke. Based on bimanual movement coordination, we designed hand motor imagery experiments. Transcranial magnetic stimulation (TMS) was applied to the left motor cortex to produce motorevoked potentials (MEP) in the first dorsal interosseous (FDI) of the right hand. Ten subjects were required to perform three different motor imagery tasks involving the twisting of a bottle cap. The results showed that contralateral hand imagery evoked the largest MEP, meaning that the brain’s motor area was activated the most. This work may prove to be significant as a reference in designing motor imagery therapy protocols for stroke patients.
Traditional new event detection is first proposed by Topic Detection and Tracking and it is actually first event detection. However, one topic usually consists of many events. The automatic instant detection of each event in one topic, not only the first event but also the second, the third and so on, is very useful for users to correctly understand the main development trend of the topic. In this paper, we address the problem of new event detection in one single topic and propose a novel topic model to detect new events along with the topic evolution. Our topic model treats new event detection as novel semantic aspect identification in one topic, rather than measuring the analog degrees between content items by lexical congruence. Besides, it can automatically determine the appropriate number of aspects needed and can naturally adapt dynamic change in the vocabulary along with the topic evolution. We use a sequential Gibbs sampling algorithm for posterior inference, which well realizes the online new event detection. Experiments are presented to show the performance of our proposed technique. It is found that our proposed technique outperforms the comparable techniques in previous work.
Powered exoskeletons have been proposed and developed in various works with the aim of compensating for motor paralysis or reducing weight, workload, or metabolic energy consumption. However, development of the power-assist system depends on the development and evaluation of real powered exoskeletons, and few studies have evaluated the performance of the power-assist system by means of computer simulation. In this paper, we propose an evaluation framework based on computer simulation for the development of an effective power-assist system and demonstrate an analysis of a power-assisted upper-arm reaching movement. We employed the optimality principle to obtain the adapted movements of humans for power-assist systems and compared the performances of power- and non-power-assisted movements in terms of the evaluation index of the power-assist system.
In this paper, Robust Genetic Network Programming (R-GNP) for generating trading rules for stocks is described. R-GNP is a new evolutionary algorithm, where solutions are represented using graph structures. It has been clarified that R-GNP works well especially in dynamic environments. In the proposed hybrid model, R-GNP is applied to generating stock trading rules with variance of fitness values. The unique point is that the generalization ability of R-GNP is improved by using the robust fitness function, which consists of the fitness functions with the original data and a good number of correlated data. Generally speaking, the hybrid intelligent system consists of three steps: priority selection by the portfolio β, optimization by the Genetic Relation Algorithm (GRA), and stock trading by R-GNP. In the simulations, the trading model is trained using the stock prices of 10 brands on the Tokyo Stock Exchange, and then the generalization ability is tested. From the simulation results, it is clarified that the trading rules created by the proposed R-GNP model obtain much higher profits than the traditional methods even in the world-wide financial crisis of 2007. Hence, its effectiveness has been confirmed.