Search

We apply four machine learning methods to cross-sectional return prediction for hedge fund selection. We equip the forecast model with a set of idiosyncratic features, which are derived from historical returns of a hedge fund and capture a variety of fund-specific information. Evaluating the out-of-sample performance, we find that our forecast method significantly outperforms the four styled Hedge Fund Research indices in almost all situations. Among the four machine learning methods, we find that deep neural network appears to be overall most effective. Investigating the source of methodological advantage of our method using a case study, we find that cross-sectional forecast outperforms forecast based on time series regression in most cases. Advanced modeling capabilities of machine learning further enhance these advantages. We find that the return-based features lead to higher returns than the benchmark of a set of macroderivative features, and our forecast method yields best performance when the two sets of features are combined. This paper was accepted by David Simchi‐Levi, finance.

With the advent of mass customisation, solving the assembly sequence planning (ASP) problem not only involves a non-convex optimisation problem that is hard to solve but also requires a high-speed ...

Robot-assisted prostate intervention under magnetic resonance imaging (MRI) guidance is a promising method to improve the clinical performance compared with the manual method. An MR conditional 6-DOF prostate intervention serial robot is developed and a binocular vision system (BVS) is established to evaluate the needle placement accuracy and located the penetration point precisely. The robot is designed by the MR conditional criteria. The serial configuration of the robot provides adequate flexibility and large workspace and excellent friendliness to the physicians. The kinematics are deduced and the needle placement control flow is proposed according to the configuration of the robot. The robot-assisted prostate intervention is divided into two phases including needle placement and needle penetration. A custom-made robust BVS is developed to obtain the needle tip position automatically in the needle placement phase where the needle cannot be detected by the MRI for lack of hydrogen atom. A simple and general algorithm used for needle tip camera coordinate estimation is proposed. Experiments on the BVS validation and robot accuracy evaluation are performed. The experiment results show that the errors of the BVS are under 0.3621 mm and the position error of the proposed robot is 2.815 mm which indicate the adequate accuracy for the prostate intervention. Graphical abstract.

Advanced building controls and energy optimization for new constructions and retrofits rely on accurate weather data. Traditionally, most studies utilize airport weather information as the decision inputs. However, most buildings are in environments that are quite different than those at the airport miles away. Tree cover, adjacent buildings, and micro-climate effects caused by the larger surrounding area can all yield deviations in air temperature, humidity, solar irradiance, and wind that are large enough to influence design and operation decisions. In order to overcome this challenge, there are many prior studies on developing weather forecasting algorithms from micro-to meso-scales. This paper reviews and complies knowledge on common weather data resources, data processing methodologies and forecasting techniques of weather information. Commonly used statistical, machine learning and physical-based models are discussed and presented as two major categories: deterministic forecasting and probabilistic forecasting. Finally, evaluation metrics for forecasting errors are listed and discussed.

Mass customized production brings great uncertainty to the computer-aided process planning (CAPP). Current CAPP methods based on heuristic optimization assume in advance that manufacturing resources are static and make a deterministic plan that cannot cope with the uncertainty of the manufacture environment. As a promising method in solving complex and dynamic decision-making problems, deep reinforcement learning is employed in this paper for process planning, aiming at promoting the response speed by exploiting the reusability and expandability of past decision-making experiences. To simplify the decision procedure, two different types of decisions, operation sequencing and resource selection, are fused into one by integrating environment states and agent behaviors in a matrix manner. Then, a masking algorithm is developed to screen out currently inexecutable machining operations at each decision step and process planning datasets are generated for training and testing according to the actual processing logic. Next, the Monte Carlo method and the deep learning algorithm are utilized to evaluate and improve the process policy, respectively. Finally, the searching capability of the proposed method for both static and dynamic manufacturing resources are tested in case studies, and the results are discussed. It is shown that the proposed approach can solve the planning problem more efficiently compared with current optimization-based approaches.

Setup planning and operation sequencing are two interrelated problems in computer-aided process planning. However, they are addressed by most traditional methods separately, in which setup planning focuses on precision assurance, while operation sequencing is devoted to cost reduction. One drawback of these methods is that they cannot balance the requirements between precision and cost. In this paper, a constrained optimization approach is proposed to address the setup planning and operation sequencing problems in a combined way. With an objective to minimize the manufacturing cost, the optimization model includes various constraints, especially the precision constraints on setup error, tool error, and stacking error. A modified particle swarm optimization algorithm is developed to search for the optimal solution. The case study shows that the proposed method has relatively more flexibility in guaranteeing the precision and decreasing the cost.

Real-time detection of object is a hot topic in remote sensing applications. In this field, deep network has shown large potential in terms of accuracy. However, it is difficult to design a lightweight hardware and software system to satisfy the satellite edge deployment due to the complexity of the structure of deep network, large amount of computation and high demand for computing resources and storage. In this article, we propose a lightweight object detecting system, based on MPSoC FPGA processor, to solve the problems. First, the lightweight algorithm for object detection which is suitable for the deployment of satellite edge devices is studied. Then, to reduce the number of parameters and the overhead calculation and storage, a network parameter compression algorithm based on dynamic mapping is laid out. Finally, a prototype verification system based on MPSoC FPGA processor is constructed to realize aircraft object detection. The experimental results highlight that the accuracy of detecting the aircrafts can reach more than 92%. For single frame image with input size of 416×416, the reasoning time is reduced to 7ms, suggesting that the requirement of real-time processing of remote sensing target detection can be met.

When using continuous frame space target images to perform space target detection, motion parameter estimation, and motion trajectory extraction, one problem that needs to be solved is the problem of entering and leaving the field of view of satellites and stars and other space targets. Detection is the basis of high-precision motion parameter estimation between frames. Use the STK EOIR module to generate space target sequence images. Based on the singular value decomposition theory, calculate the parameters such as the amount of rotation, translation, and scaling factors of the inter-frame image. The analysis of the inter-field parameters of the field of view identifies the points contained in and out of the field of view and detects them. The experimental results show that the algorithm is simple and fast, and can effectively detect space targets in and out of the field of view.

Visibility prediction has important application value in transportation, aviation, military, agriculture and other fields. Traditional visibility monitoring instruments have the disadvantages of complex operation, expensive price and low accuracy. At the same time, the existing visibility prediction methods based on video images also have the disadvantages of low accuracy. In this paper, a visibility prediction method in foggy weather based on deep learning is proposed. The proposed method uses target detection network YOLOv5 (YOU ONLY LOOK ONCE V5) to detect the ground landmarks in the video, then establishes a camera imaging model to calculate distance of the ground landmarks, and finally obtains the visibility value according to the distance of the farthest landmark that can be detected. In the proposed method, CIOU _Loss is selected as the loss function of YOLOv5 to improve the convergence speed and prediction accuracy. The experimental results show that thanks to yolov5's powerful and fast detection capabilities, after the training of 200 epoches, the proposed method can detect landmarks in foggy images with a 100% recall rate, which has the advantages of low cost and high accuracy.

Task allocation for specialized unmanned robotic agents is addressed in this paper. Based on the assumptions that each individual robotic agent possesses specialized capabilities and that targets representing the tasks to be performed in the surrounding environment impose specific requirements, the proposed approach computes task-agent fitting probabilities to efficiently match the available robotic agents with the detected targets. The framework is supported by a deep learning method with an object instance segmentation capability, Mask R-CNN, that is adapted to provide target object recognition and localization estimates from vision sensors mounted on the robotic agents. Experimental validation, for indoor search-and-rescue (SAR) scenarios, is conducted and results demonstrate the reliability and efficiency of the proposed approach.

To test the accuracy and efficacy of the multimodality imaging-compatible insertion robot with a respiratory motion calibration module designed for ablation of liver tumors in phantom and animal models. To evaluate and compare the influences of intervention experience on robot-assisted and ultrasound-controlled ablation procedures.Accuracy tests on rigid body/phantom model with a respiratory movement simulation device and microwave ablation tests on porcine liver tumor/rabbit liver cancer were performed with the robot we designed or with the traditional ultrasound-guidance by physicians with or without intervention experience.In the accuracy tests performed by the physicians without intervention experience, the insertion accuracy and efficiency of robot-assisted group was higher than those of ultrasound-guided group with statistically significant differences. In the microwave ablation tests performed by the physicians without intervention experience, better complete ablation rate was achieved when applying the robot. In the microwave ablation tests performed by the physicians with intervention experience, there was no statistically significant difference of the insertion number and total ablation time between the robot-assisted group and the ultrasound-controlled group. The evaluation by the NASA-TLX suggested that the robot-assisted insertion and microwave ablation process performed by physicians with or without experience were more comfortable.The multimodality imaging-compatible insertion robot with a respiratory motion calibration module designed for ablation of liver tumors could increase the insertion accuracy and ablation efficacy, and minimize the influence of the physicians' experience. The ablation procedure could be more comfortable with less stress with the application of the robot.

Feature interactions may result in many process alternatives in part machining. Traditional process planning methods only identify one of the options, which is usually not optimal in the sense of engineering. This paper presents an optimisation approach to handle the interacting feature recognition problem in mill-turn parts. The approach subdivides the material removal volume into cells first and then it combines the cells into features. Here, a two-level cell combination method is developed. On the lower level, individual features are formed by searching the combinations of cells near a given part face; on the upper level, the feature distributions are explored by rearranging the order of part faces for feature formation. In order to optimise the feature distribution, a novel optimisation model is proposed, which quantitatively distinguishes its options by considering the factors of feature numbers, tool approaching directions, cutting directions and surface roughness. The combinatorial optimisation pro...

Fault diagnosis of rolling bearings has been an important and challenging research issue. The existing conventional algorithms for diagnosing rolling bearings rely on artificial feature extraction requiring a wealth of knowledge of signal handling, expertise and human efforts, lacking adaptability. With the capacity for learning features from original signals automatically, deep learning methods can solve the shortcomings of conventional diagnosis methods. This paper puts forward a deep learning method for rolling bearing fault diagnosis based on two-dimensional convolutional neural network, called 2dCNN-FD. The method has the capability of automatic feature extraction and robust noise tolerance. Experimental results demonstrate that comparing with other traditional diagnosis algorithms, the 2dCNN-FD model achieves high diagnosis accuracy and better robustness under different noise conditions.

In order to predict the reliability of semiconductor lasers, an accelerated degradation test (ADT) was proposed as an element of reliability testing. Temperature-stressed ADT was applied for 8 Semiconductor lasers which used in space missions, and the degradation characteristics of output power of semiconductor lasers were studied. Then, a reliability model based multi-output Gaussian process regression (MOGP) was proposed to evaluate the lifetime and reliability for laser diodes. The advantage of the proposed MOGP based method is that it utilizes the output correlation between multiple degradation traces to make the outputs utilize each other's information and provide more accurate prediction than single modeling. Thereby improving the prediction accuracy. Furthermore, verifying applications and cases studies are discussed to prove the generality and practicability of the proposed reliability prediction model. Results show that the accuracy of the proposed MOGP based method is twice that of the SVM method.

As equipment becomes more and more complex, it is increasingly difficult to manually extract and select fault features manually based on expert experience or signal processing techniques. In addition, the shallow model such as BP neural network and SVM have trouble to deal with the complex mapping relationship with respect to the measured signal and the health condition of the equipment, who faces the problem of dimensional disaster. Combined with the advantages of deep confidence network (DBN) in features extraction and deal with high-dimensional and nonlinear samples, a fault feature extraction and diagnosis method based on deep confidence network for gearbox is investigated in this framework. The method uses the original time domain signal to train the deep confidence network and completes the intelligent diagnosis through deep learning. The preponderance is that it can take out the dependence on a great quantity of signal processing techniques and diagnostic experience, and accomplish the extraction of fault features and the intelligent diagnosis of health status with the characteristic of self-adaption. The method has no periodic requirements for time domain signals, and has strong versatility and adaptability. The experimental results of the fault diagnosis for the planetary gearbox demonstrated the feasibility and superiority of the presented method.

The main factor, which leads to semiconductor lasers degradation in space mission experiments, is the introduction of lattice defects in the detector material produced by radiation. Therefore, a new remaining useful life (RUL) prediction method considering radiation effect is presented. The prediction method is based on failure mechanisms and particle filters, which realizes the combination of a physical model and data-driven method. The particle filters can abundantly employ degradation data and avoid ignoring the failure mechanism. Accelerated degradation testing (ADT) is proposed for effectively assessing the proposed prediction method, in which the proton radiation is selected as the accelerated stress. The prediction results are compared with other methods to verify the accuracy and effectiveness of the proposed prediction method.

The adoption of precast concrete elements (PCEs) are becoming popular in civil infrastructures. Since quality of connections determines the structure property, design of rebar in PCEs is a mandatory stage in constructions. Due to large number of rebar, complicated shapes of PCEs and complicated rules for arrangement, it is labor-intensive and error-prone for designers to avoid all clashes even using computer software. With the aid of BIM, it is desirable to have an automated and clash-free rebar design. Taking this cue, we introduce a framework with generative adversarial network (GAN) and multi-agent reinforcement learning (MARL) for generating design and automatically avoiding clash of rebar in PCES. We use GAN to generate 2D rebar designs. Then, 2D rebar designs are transformed into digital environments for MARL. In addition, layout of rebar is modelled as path planning of agents in MARL. An illustrative example is presented to test the proposed framework.

To date, to improve construction quality and efficiency and reduce environmental pollution, the use of precast concrete elements (PCEs) has become popular in civil engineering. As PCEs are manufactured in a batch manner and possess complicated shapes, traditional manual inspection methods cannot meet today’s requirements in terms of production rate of PCEs. The manual inspection of PCEs needs to be conducted one by one after the production, resulting in the excessive storage of finished PCEs in the storage yards. Therefore, many studies have proposed the use of terrestrial laser scanners (TLSs) for the quality inspection of PCEs. However, all these studies focus on the data of a single PCE or a single surface of PCE, which is acquired from a unique or predefined scanning angle. It is thus still inefficient and impractical in reality, where hundred types of PCEs with different properties may exist. Taking this cue, this study proposes to scan multiple PCEs simultaneously to improve the inspection efficiency by using TLSs. In particular, a segmentation and recognition approach is proposed to automatically extract and identify the different types of PCEs in a large amount of outdoor laser scan data. For the data segmentation, 3D data is first converted into 2D images. Image processing is then combined with radially bounded nearest neighbor graph (RBNN) algorithm to speed up the laser scan data segmentation. For the PCE recognition, based on the as-designed models of PCEs in building information modeling (BIM), the proposed method uses a coarse matching and a fine matching to recognize the type of each PCE data. To the best of our knowledge, no research work has been conducted on the automatic recognition of PCEs from a million or even ten million of the outdoor laser scan points, which contain many different types of PCEs. To verify the feasibility of the proposed method, experimental studies have been conducted on the PCE outdoor laser scan data, considering the shape, type, and amount of PCEs. In total, 22 PCEs including 12 different types are involved in this paper. Experiment results confirm the effectiveness and efficiency of the proposed approach for automatic segmentation and recognition of different PCEs.

This paper investigates the potential of deep learning methods to detect and segment objects from vision sensors mounted on autonomous robots to support task allocation in unmanned systems. An object instance segmentation framework, Mask R-CNN, is experimentally evaluated and compared with previous architecture, Faster R-CNN. The former model adds an object mask prediction branch in parallel with the existing branches for target objects location and class recognition, which represents a significant benefit for autonomous robots navigation. A comparison of performance between the two architectures is carried over scenes of varying complexity. While both networks perform well on recognition and bounding box estimation, experimental results show that Mask R-CNN generally outperforms Faster R-CNN, particularly because of the accurate mask prediction generated by this network. These results support well the requirements imposed by an automated task allocation mechanism for a group of unmanned vehicles.

Image processing technology has many applications, such as medical image processing, face detection and vehicle detection. At present, in the field of civil engineering, there are many difficulties in the intelligent manufacturing process of precast concrete elements (PCEs), such as non-contact measurement of PCEs using 3D laser scanners. In order to solve the data acquisition problem of PCE data in outdoor laser scan data, this paper is the first attempt to map a 3D laser scan data into 2D images, and then use image processing technology to realize the segmentation of outdoor laser scan data which contains the PCE data. We convert the processed objects from laser scan data to images, and computation time is reduced effectively as a result. This segmentation technique includes four steps, data preprocessing, 3D laser scan data mapping into 2D images, image processing for segmentation and laser scan data processing. Finally, the results obtained by experiments prove the effectiveness of this segmentation technique.

In urban underground environment, a short-conduit is commonly used to provide mechanical protection for the cable. The cable section buried in the short-conduit may experience excessive thermal stresses. However, the thermal rating for the cable in the short-conduit is not specialized in the IEC 60287. It has not been sufficiently addressed in the literature either. This paper proposes a quasi-three-dimensional thermal model for the cable installed in the short-conduit. A simplified model for efficiently and effectively solving the proposed model is then developed. Based on the simplified thermal model, an improved method for assessing the real-time ampacity of the cable installed in the short-conduit is implemented. The performance of the improved method with respect to the IEC method on thermal rating for the cable installed in the short-conduit is verified by the Finite Element Analysis (FEA). The results show that the improved method outperforms the IEC method when the conduit length is less than five meters. The method is useful in real-time thermal rating of the short-conduit cables and paves a way to increase the utilization of cables installed in short-conduits while still ensuring their reliability.

This paper presents an adaptive hierarchical optimization approach to optimize the lamination parameter distribution of variable-stiffness laminated plate with truncated hierarchical B-Splines. Traditional methods use a uniform grid to design the distribution though its variation rate could change dramatically from point to point. As a consequence, the dense grid incurs a high computation cost while the rough one loses the design freedom. The main motivation of this research is to resolve this problem by performing a hierarchy of the optimizations respectively on different level grids of the hierarchical B-Spline for a non-uniform representation of lamination parameters. At the end of each level optimization, some elements in the multi-level grid are selected to be refined further according to a refinement rule. The hierarchical mesh consisting of newly generated elements and unrefined ones, is used to represent the solution, and its corresponding hierarchical B-Spline expression is updated with a new optimization step. Based on the sensitivity analysis approach proposed for the adaptive optimization problem, the sequential quadratic programming algorithm is implemented for the optimizations at individual levels. Finally, three examples of compliance minimization are chosen to validate the efficiency and accuracy of the proposed method.

In the US, people spend more than 90% of their time in buildings, which contributes to more than 70% of overall electricity usage in the country. Occupant behavior is becoming a leading factor impacting energy consumption in buildings. Existing occupant-behavior studies are often limited to a single building and individual behavior, such as presence or interactions in confined spaces. Moreover, studies modeling occupant behavior at the building or community level are limited. With the development of the Internet of Things, mobile positioning data are available through social media and location-based service applications. The goal of this study is to analyze the impacts of more representative occupancy profiles, derived from high resolution urban scale mobile position data, on building energy consumption. . A pilot study was conducted on more than 900 buildings in downtown San Antonio, Texas, with billions of mobile positioning data. We then compared these profiles with the existing Department of Energy prototype models and quantified the differences using a statistical method. On average, the differences in occupancy rates between the ones derived from the empirical profile and the ones from the Department of Energy reference ranged from −30% to 70%. The realistic derived profiles are then simulated in the CityBES. The results show that the predicted cooling energy demand is reduced by up to 40% while the heating energy demand is reduced by up to 60%. This study, therefore, advances knowledge of urban planning as well as urban-scale energy modeling and optimization.

The proliferation of urban sensing, IoT, and big data in cities provides unprecedented opportunities for a deeper understanding of occupant behaviour and energy usage patterns at the urban scale. This enables data-driven building and energy models to capture the urban dynamics, specifically the intrinsic occupant and energy use behavioural profiles that are not usually considered in traditional models. Although there are related reviews, none investigated urban data for use in modelling occupant behaviour and energy use at multiple scales, from buildings to neighbourhood to city. This survey paper aims to fill this gap by providing a critical summary and analysis of the works reported in the literature. We present the different sources of occupant-centric urban data that are useful for data-driven modelling and categorise the range of applications and recent data-driven modelling techniques for urban behaviour and energy modelling, along with the traditional stochastic and simulation-based approaches. Finally, we present a set of recommendations for future directions in data-driven modelling of occupant behaviour and energy in buildings at the urban scale.

The design of reinforcing steel bars (rebars) is critical to reinforced concrete (RC) structures. Generally, a good number of rebars are required by a design code, particularly at member connections. As such, rebar clashes (i.e., collisions and congestions) would be inevitable. It would be impractical, labor-intensive, and error-prone to avoid all possible clashes manually or even using standard design software. The building information modeling (BIM) technology has been utilized by the present architecture, engineering, and construction (ACE) industry for clash-free rebar designs. However, most existing BIM-based approaches offer the clash resolution strategy for moving components with an optimization algorithm, and are only applicable to the RC structures with regular shapes. In particular, the optimized path of rebars cannot be adjusted to avoid the obstacles, thus limiting the practical applications. Furthermore, most existing studies lack the learning from design code and constructibility constraints to realize automatic and intelligent arrangement and adjustment of rebars for avoiding the obstacles encountered in complex RC joints and frame structures. Considering these shortcomings, the authors have recently proposed an immediate reward-based multi-agent reinforcement learning (MARL) system with BIM, towards automatic clash-free rebar designs of RC joints without clashes. However, as the immediate reward is required in the MARL system for guiding the learning of a rebar design, it will not succeed in clash-free rebar designs of complex RC structures where immediate reward is often unavailable. In this study, this study further extends the previous work with Q-learning (a model-free reinforcement learning algorithm) for more realistic path planning considering both immediate and delayed rewards in clash-free rebar designs for real-world RC structures. In particular, the rebar design problem is treated as a path-planning problem of multi-agent system, where each rebar is deemed as an intelligence reinforcement learning agent. Next, by employing the Q-learning as the reinforcement learning engine, the particular form of state, action, and immediate and delayed rewards for the reinforcement MARL for automatic rebar designs considering more actual constructible constraints and design codes can be developed. Comprehensive experiments on three typical beam-column joints and a two-story RC building frame were conducted to evaluate the efficiency of the proposed method. The study results of paths of rebar designs, success rates, and average time confirm that the proposed framework with MARL and BIM is effective and efficient.

The betatron tune is an important parameter in a storage ring to enable stable operation. A tune adjustment tool with a small impact on the beam dynamics is useful for user operation and machine studies. Therefore, a tune knob is developed for the Hefei light source-II (HLS-II) storage ring. Owing to the compactness of the storage ring, a global adjustment mechanism is adopted. To reduce the impact on beam injection, only quadrupole families outside the injection section are used by the tune knob, and the $$\beta$$ functions of the injection section remain unchanged. A code is developed based on the accelerator simulation software, MAD-X, to calculate the adjustment of the quadrupole strengths. The accelerator toolbox is used to double check the accuracy of the tune knob. Online measurement of the tune knob is also performed. The result shows that the tune knob works well when the tune is adjusted in a specific range. Betatron coupling measurement is also carried out, showing an application of the tune knob on machine studies. In this paper, the development of the tune knob and its experimental results in the HLS-II storage ring are reported in detail.

To solve the problem of difficulty in extracting fault features from the vibration signal of centrifugal pump, this paper proposes an integrated method based on multi-scale entropy (MSE) and local characteristic-scale decomposition (LCD) for centrifugal pump fault diagnosis. Firstly, LCD is applied to decompose the signal into a series of components. Secondly, MSE is extracted as the signal feature. Thirdly, probabilistic neural network (PNN) is employed for fault recognition. Finally, a case is studied based on the experimental data of a centrifugal pump, and the present method achieves the accuracy of 98.00%. Compared with other fault diagnostic methods, the proposed approach exhibits the best performance. The results validate the effectiveness and superiority of the present method.

With the continuous progress of the space station project, the rotary mechanism is widely used in the space station. Serious failure of rotating mechanism can cause motor burnout and then safety problems. Therefore, it is necessary to monitor and diagnose rotating mechanism. The vibration signal of the rotating mechanism is nonlinear and non-stationary. Therefore, the acceleration sensor vibration signal is decomposed by CEEMD method to get the intrinsic mode function. Next, Fourier transform is used to obtain time-frequency information according to the obvious components of fault characteristics, and the information entropy of time-frequency information is calculated. The degree of dimensionality reduction about fault feature is determined by selecting the number of principal components through adaptive cumulative contribution. Finally, the fault feature is trained by support vector machine and tested with data. Experimental results show that the optimized method is less computational and can accurately extract fault feature information.

We apply four machine learning methods to cross-sectional return prediction for hedge fund selection. We equip the forecast model with a set of idiosyncratic features, which are derived from historical returns of a hedge fund and capture a variety of fund-specific information. Evaluating the out-of-sample performance, we find that our forecast method significantly outperforms the four styled HFR Indices in almost all situations. Among the four machine learning methods, we find that deep neural network appears to be overall most effective. Investigating the source of methodological advantage of our method using a case study, we find that cross-sectional forecast outperforms forecast based on time series regression in most cases. Advanced modeling capabilities of machine learning further enhance these advantages. We find that the return-based features lead to higher returns than the benchmark of a set of macro-derivative features, and our forecast method yields best performance when the two sets of features are combined.

Maintaining the continuity (blending) of stacking sequences is very vital when the stacking sequences of a multi-region panel is being optimized. The traditional attempt simplifies the difficulty of the blending problem by designing the whole panel based on one stacking template. However, this simplification adds an unnecessary constraint that regions with the same thickness should have the same stacking sequence. In order to overcome the drawback, this paper introduces a region block queue based method to obtain the optimal manufacturable stacking sequences of composite laminates blended from multiple stacking templates. The method discussed here groups regions into blocks, forming a block queue, and designs the entire laminate panel with multiple stacking templates, which are respectively applied to different blocks. The continuity between two adjacent blocks is guaranteed by sequentially rebuilding the template of a block from the sequences of adjacent regions within its previous block in the queue. An 18-region benchmark problem is chosen to prove the effectiveness of the proposed method, and the solution obtained is 0.27kg lighter than the traditional template-based method. The reduction of mass reveals that the use of multiple templates does contribute to searching for lighter solution.

Computer-aided process planning (CAPP) plays an important role in integrated manufacturing system and it can serve as a bridge between CAD and CAM. As a crucial part of CAPP, setup planning is a multi-constraint problem, in which the precision takes priority over efficiency. However, instead of precision constraints, traditional optimization methods have paid much more attention to efficiency requirements. This leads to the reduction in the precision of the final parts. This paper develops an optimization approach for solving computer-aided setup planning problem, which takes into account various constraints, especially the precision requirements specified by designers. First, objective function of the optimization model is formulated and a series of constraints, including feature precedence, tool approaching direction (TAD), and precision requirements are systematically created. Next, the model is solved by using a hybrid particle swarm optimization algorithm. In order to overcome the local optimum trap, mutation and exchange operations are adopted from the genetic algorithm. Finally, a part is tested in the case study and the validation of this method is proved.

With the extensive use of information technology in AAL communication systems, a data management model has recently embodied the 3-V characteristics of big data: volume, velocity, and variety. A lot of work has been done on volume and velocity, but not as much has been reported on variety. To handle the variety of data, universal solutions with acceptable performance are usually much more cost effective than customized solutions. To achieve universality, a basic idea is to first define a universal abstraction that covers a wide range of data types, and then build a universal system for universal abstraction. Traditional database management systems commonly use a multidimensional data type, or feature vectors, as a universal abstraction. However, many new data types in AAL systems cannot be abstracted into multidimensional space. To find a more universal data abstraction and build more universal systems, we propose the concept of big data abstraction, with metric space as a universal abstraction for AAL data types. Furthermore, to demonstrate how metricspace data abstraction works, we survey the state of the art in metric space indexing, a fundamental task in data management. Finally, open research issues are discussed.

The purpose of this study was to explore the assistant function of 3-D information for I125 particle implantation of multineedle intervention under the guidance of ultrasound. The assistant function of 3-D information was verified by a simulation experiment system which consists of an ultrasound probe, an abdominal phantom, the preoperative computed tomography image of a patient, the electromagnetic tracking device, and the self-developed 3-D image navigation software with a practical and friendly graphical user interface. The simulation particle implantation experiments were divided into the two groups. The first group of experiments was performed with the aid of 3-D information. Seven days later, the second group of experiments was carried out with the aid of 2-D information. We made the statistical analysis of the experimental results obtained by nine medical students, nine interventional radiologists, and nine attending physicians. With the assistance of 3-D information, the percentage of tumor coverage increased (p

This paper addresses the problem of task allocation in real-time distributed systems with the goal of maximizing the system reliability, which has been shown to be NP-hard. We take account of the deadline constraint to formulate this problem and then propose an algorithm called chaotic adaptive simulated annealing (XASA) to solve the problem. Firstly, XASA begins with chaotic optimization which takes a chaotic walk in the solution space and generates several local minima; secondly XASA improves SA algorithm via several adaptive schemes and continues to search the optimal based on the results of chaotic optimization. The effectiveness of XASA is evaluated by comparing with traditional SA algorithm and improved SA algorithm. The results show that XASA can achieve a satisfactory performance of speedup without loss of solution quality.

Purpose: To develop a fusion imaging system that combines ultrasound and computed tomography for real-time tumor tracking and to validate the accuracy of performing registration via this approach during a specific breathing phase. Materials and Methods: The initial part of the experimental study was performed using iodized oil injection in pig livers and was focused on determining the accuracy of registration. Eight points (A1-4 and B1-4) at different positions and with different target sizes were selected as target points. During respiratory motion, we used our self-designed system to perform the procedure either with (experimental group, E) or without (control group, C) the respiratory monitoring module. The registration errors were then compared between the 2 groups and within group E. The second part of this study was designed as a preliminary clinical study and was performed in 18 patients. Screening was performed to determine the combination of points on the body surface that provided the highest sensitivity to respiratory motion. Registration was performed either with (group E) or without (group C) the respiratory monitoring module. Registration errors were compared between the 2 groups. Results: In part 1 of this study, there were fewer registration errors at each point in group E than at the corresponding points in group C ( P < .01). In group E, there were more registration errors at points A1 and B1 than at the other points ( P < .05). There was no significant difference in registration errors among the remaining points. During part 2 of the study, there was a significant difference in the registration errors between the 2 groups ( P < .01). Conclusions: Real-time fusion registration is feasible and can be accurately performed during respiratory motions when using this system.

To develop an image visualization system based on graphic processing unit (GPU) hardware acceleration for clinical use in hepatocellular carcinoma (HCC) interventional planning.We developed a liver tumor planning tool to assist the physician in providing patient-specific analysis and visualization. We employed a spatial distance computation algorithm to determine the spatial location of tumors and their relation to the main hepatic vessels. GPU hardware acceleration was implemented for rapid calculation of the spatial distance from the tumor surface to the surrounding vascular territories.The algorithm for spatial distance provided an accurate minimum value for the distance from the tumor surface to the surrounding duct system as well as the region of interest (ROI). Analyzing the data (mean CPU time = 43.14 ± 29.34; mean GPU time = 0.41 ± 0.38) using an independent samples t-test, the result showed a remarkable difference (p0.001). Thus, GPU hardware acceleration performed the distance arithmetic at higher rates than conventional CPUs.The visual assistance tool performs as an intuitive and objective module in clinical cases, and is expected to help physicians achieve a more reliable treatment in liver tumor patients. As such, we believe it represents an improvement in image guided preoperative planning.

The development cycle of an image-guided surgery navigation system is too long to meet current clinical needs. This paper presents an integrated system developed by the integration of two open-source software (IGSTK and MITK) to shorten the development cycle of the image-guided surgery navigation system and save human resources simultaneously. An image-guided surgery navigation system was established by connecting the two aforementioned open-source software libraries. It used the Medical Imaging Interaction Toolkit (MITK) as a framework providing image processing tools for the image-guided surgery navigation system of medical imaging software with a high degree of interaction and used the Image-Guided Surgery Toolkit (IGSTK) as a library that provided the basic components of the system for location, tracking, and registration. The electromagnetic tracking device was used to measure the real-time position of surgical tools and fiducials attached to the patient's anatomy. IGSTK was integrated into MITK; at the same time, the compatibility and the stability of this system were emphasized. Experiments showed that an integrated system of the image-guided surgery navigation system could be developed in 2 months. The integration of IGSTK into MITK is feasible. Several techniques for 3D reconstruction, geometric analysis, mesh generation, and surface data analysis for medical image analysis of MITK can connect with the techniques for location, tracking, and registration of IGSTK. This integration of advanced modalities can decrease software development time and emphasize the precision, safety, and robustness of the image-guided surgery navigation system.

In the development of safety-critical system, ModelBased Safety Assessment (MBSA) process can reduce the development complexity and improve assessment efficiency. As an important part of MBSA, system modeling and safety requirements description with formal methods still lack a well-defined process. In this paper, the relationship between traditional safety assessment process and MBSA is discussed, a system modeling and safety property formalization method based on NuSMV is proposed. Finally, a case study is presented to demonstrate and validate the proposed modeling method.

The purpose of this study was to provide a simulation therapy environment for microwave thermal ablation (MWA) under the guidance of ultrasound, and to present an inexpensive and portable simulator built on real patient-based pre-operative computed tomography (CT) data. We established an experimental simulation system for teaching MWA and present the results of a preliminary evaluation of the simulator's realism and utility for training. The system comprises physical elements of an electromagnetic tracking device and an abdominal phantom, and software elements providing three-dimensional (3D) image processing tools, real-time navigation functions and objective evaluation function module. Details of the novel aspects of this system are presented, including a portable electromagnetic tracking device, adoption of real patient-based pre-operative CT data of liver, operation simulation of MWA, and recording and playback of the operation simulation. Patients with liver cancer were selected for evaluation of the clinical application value of the experimental simulation system. A total of 50 consultant interventional radiologists and 20 specialist registrars in radiology rated the simulator's hardware reality and overall ergonomics. Results show that the simulator system we describe can be used as a training tool for MWA. It enables training with real patient cases prior to surgery, and it can provide a realistic simulation of the actual procedure.

On the hardware platform of Distributed Real-Time Simulation, a communication module based on reflective memory network was realized in FPGA. As an infrastructure module, it could be integrated with other modules seamlessly. Under the speed-first rules, DMA engine in PCI Master mode was designed, and 64bits/32bits bus compatibility was realized. In order to solve the competition problems during network and memory access, the Arbiter strategy based on adjustable time slice was proposed. In addition, network command packets of higher priority were introduced to fit practical simulation requirements, including the interrupts issue.

With the widely use of CAN in the industrial control field, means of transportation, and so on, it is significant to improve the performance of the corresponding simulation platform to make them be more useful for simulating the systems using CAN cards. Firstly, this paper introduces the overall structure of the KD-DRT simulation platform and the characteristic of CAN. Second, determines the form of the CAN card model according to demand. Then designs the CAN card model in 3 respects: the inner structure of a CAN node, variables and API functions. Finally, analyses the main workflow of the model.

Metric-space indexing, also known as distance-based indexing, is a universal indexing to support similarity queries. It only requires that the similarity of data be defined by a metric distance function. To achieve the great universalness, metric-space indexing does not take use of the domain information of data, and is thus outperformed by many domain-specific methods. In this paper, to speed up metric-space similarity query, we first assign one thread for each query in the multi-query case to increase the throughput. Then, for a single query, we assign one thread for each search path from the root of the index tree to decrease the responding time. Last but not least, we implement an in-memory buffer to break the bottleneck of the disk access to the index file. Experimental results show that our efforts result in good speed up and parallel efficiency.

KD-DRT simulation platform tries to manage MATLAB/Simulink models, and make them compatible and cooperating with other C/C++ models in distributed real- time simulation. MATLAB/Simulink together with KD-DRT are compared and analyzed in system structure and properties. The shortages of MATLAB/Simulink in distributed real-time simulation are proposed, and KD-DRT is chosen as the support platform instead. KD-DRT chooses S-Function mode to create simulation models, and chooses Target Language Compiler(TLC) to inline S-Function and collect information of relationship, parameters and ports from Simulink models. Correlated RTW file is created and the structure is analyzed to extract these informations through TLC. Tests are carried out in the end to prove that KD-DRT can control parameters of Simulink models online, and transfer signals between multiple models outside the MATLAB environment. KD-DRT proves to be more suitable to serve as the support platform for distributed real-time simulation ,and is able to combine Simulink models with C/C++ ones in simulation system.