In this paper, a modified bat algorithm (MBA) is proposed for solving the multi-objective flexible job shop scheduling problem. Three different production performance indicators are considered which are the makespan, the total workload of machines and the critical machine workload. Firstly, to make the algorithm adaptive to the problem, the converting approaches are presented to implement the conversion between the continuous position vector and the discrete scheduling code. Secondly, an initialization scheme combining heuristics and random rule is introduced to ensure good quality and diversity of the initial population. Furthermore, five neighborhood structures are designed based on individual positions. Then, a local search algorithm is embedded into the BA to enhance the local searching ability. Finally, simulation results demonstrate the feasibility and effectiveness of our proposed algorithm.

Submitted on July 21, 2017; Revised on September 8, 2017; Accepted on October 10, 2017
References: 27

Recent advancements in materials and manufacturing processes are becoming difficult in day-to-day times to meet the industrial needs. Assembly is one of the processes in manufacturing, which takeover 20% of overall cost in manufacturing [5]. The assembly of the parts still becomes difficult, if the product consists of parts with more intricate shape. Design For Assembly (DFA) has driven product designers towards minimizing the number of parts in a product to reduce the assembly efforts and manufacturing cost. Until now, there is no generalized method to obtain modified topology of the product by DFA concept. Many industries like Toyota, Sony and many more follow their own designed DFA methodology. Generalization of DFA concept to obtain the modified topology of the product involves high skilled user intervention and demands in deep knowledge of DFA principles. In this paper, an attempt is made to generate modified topology by generalizing the DFA concept. To generalize the DFA concept and to obtain the modified topology, the research work is mainly concentrated on four principles. 1) Material properties of the parts, 2) Relative motion between the parts, 3) Contact between the parts, 4) Functionality of the parts. Depending on these principles, a general methodology has been developed to obtain the modified topology of industrial products. The methodology has been successfully implemented on an industrial product to obtain modified topology with reduced part numbers.

Submitted on May 20, 2017; Revised on October 10, 2017; Accepted on October 21, 2017
References: 18

A hybrid meta-heuristic named variable neighborhood migrating birds optimization (VNMBO), which is a combination of variable neighborhood search (VNS) and migrating birds optimization (MBO). The main aim of this paper is to provide a new way for MBO to solve the flexible job shop scheduling problem (FJSP). A two-stage population initialization scheme was first adopted to improve the quality of the initial solutions. An individual leaping mechanism was introduced to the algorithm in order to avoid the premature convergence. To search the solution space effectively, three neighborhood structures were designed and a VNS was developed to enhance the local searching ability. Finally, to assess the performance of the proposed VNMBO, some published algorithms were compared by using two famous benchmark data sets. The comparison results show that the proposed VNMBO is effective for solving the FJSP with the objective of minimizing the makespan.

Submitted on July 24, 2017; Revised on October 13, 2017; Accepted on October 21, 2017
References: 19

With the video display technique developing, High Efficiency Video Coding (HEVC) to optimize coding efficiency for video coder is proposed. However, the latest coding standard HEVC has a good performance. In the meantime, it introduces large computational complexity in mode decision process. In the coding process, the encoding mode of a coding block is selected in numbers candidate. Therefore, once a coding block in the sequences encoded as skip mode, coding time would be saved largely since the simplification of rate-distortion (RD) cost calculation. In this paper, a decision method of coding blocks (CBs) based on texture analysis of entropy is proposed. Even though a CB included in homogeneous regions in natural test video sequences, the entropy value between current and neighbouring reference coding blocks is computed to decide the CB could be encoded as skip mode. Early skip mode detection can omit unnecessary mode decision since it results in huge computational complexity for RD cost calculation in conventional method. After the detection process, the CBs would be encoded as skip mode directly. Extensive experiment concludes that our method has better performance for computational complexity compared to conventional coding method with a negligible loss.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 22

In view of the fact that SaaS service cannot currently guarantee reliability and instantaneity, a dynamic SaaS service selection strategy based on Skyline (SSBS) is proposed. In this method, to select the matching services in the SaaS services library, the Skyline computation is used to reduce redundancy. Then, service selection can be made using mixed integer programming. The experimental results show that the proposed method can accurately select the most suitable could service.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 12

Various types of defects exist in the solar cell surface because of some uncontrollable factors during the process of production. The solar cell surface defects detection is indispensable for the production of solar cell. The automatic defects detection methods based on computer vision have been widely used because of its convenience, real time and low cost. The state-of-the-art methods of solar cell surface defects detection based on computer vision are reviewed in this paper. Firstly, the typical defects of solar cell surface are summarized. Secondly, the state-of-the-art methods are classified into three categories: local scheme, global scheme and local-global scheme based methods, and separately introduced. Thirdly, the qualitative and exact evaluations of state-of-the-art methods are presented. The main contents of this paper and future development trends are summarized in the end.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 41

Aerospace complex software system is the keypoint of aerospace industry informatization. The complexity and scale of aerospace complex software system is growing with the increase of system requirements. Therefore, the possibility of runtime failures is also increasing. The runtime failures may lead to some serious problems of the aerospace software system and may cause great damage. To reduce the loss of software failures and to ensure the normal operation of aerospace complex software system, this paper focuses on runtime fault detection based on runtime verification. Runtime verification aims to monitor a running system and check whether executions of the monitored system satisfies or violates a given correctness property. This paper proposes a method to realize runtime fault detection and solve the runtime failure problem.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 14

Particle swarm optimization (PSO) is a popular intelligent algorithm to solve the task scheduling optimization problem of work-flow system in cloud computing environment. However, this algorithm is easy to fall into the local optimality. It is the reason that the execution time and cost of the scheduling scheme are higher than other methods. Therefore, by improving the calculation method of the single particle success value, the traditional adaptive inertia weight particle group task scheduling algorithm is optimized. Through each particle fitness and local optimal value and global optimal value that divided into four cases to compare, the inertia weight improved can be used to adjust the particle velocity more accurately. It can better equilibrate search capacity of particles between global and local, and avoid the local maximum of the particles. In this paper, we more accurately describe the particle state and improve the inertia weight. We can get a scheduling scheme with lower execution time and lower cost. The analog results show that the improved algorithm is stable. The convergence accuracy is obviously improved. It can effectively avoid prematurely falling into the local optimality.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 11

Cloud computing is a new type of business computing model. It is connected through the network and can obtain various applications, data and IT services. The core of cloud computing is task scheduling, and the application of genetic algorithm (GA) in cloud computing task scheduling is also a hot topic in recent years. In this paper, the "three-stage selection method" and the genetic strategy of "total-division-total" are put forward to improve genetic algorithm. Using simulation experiments in cloud computing simulation software named Cloudsim, the experimental results show that comparing with the simple genetic algorithm (SGA), the improved genetic algorithm (IGA) is better than the simple genetic algorithm on completion time, and it is an effective task scheduling algorithm in cloud computing environment.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 13

Micro-blog is a new social networking service platform and users are very concerned about real-time personalized information. However, the existing micro-blog platform does not fully consider the user's real-time personalized demands. The paper proposes a micro-blog real-time personalized recommendation model. We constructed partial index mechanism to maintain the latest release or update of micro-blog, and inferred the topic distribution of micro-blog and user interest vector based on the LDA model to meet the real-time personalized demands of users. Experimental results on real datasets show that the proposed method is real-time and effective.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 12

superword level parallel (SLP) algorithm cannot effectively handle the large-scale applications which covered few parallel codes, and the codes which can be vectorized may be adverse to the vectorization. A new improved algorithm for SLP is proposed. First of all, attempt to transform the non-isomorphic statements, which can’t be vectorized to isomorphic statements as far as possible. Namely, locate the opportunities of vectorization which SLP has lost, and then build the Max Common Subgraph (MCS) through adding redundant nodes, process some optimization such as redundant deleting to get the supplement diagram of SLP, it can greatly increase the parallelism of program. At last, using the method of cutting, eliminate the codes harmful to the vectorization, and execute them in serial. This vectorizes the revenue codes, improving the efficiency of programs as far as possible. Experimental results show that, compared with the SLP algorithm, its performance in average is better than it 9.1%.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 15

With plan recognition applications continuously expanding, the speed of traditional plan recognition is too slow, which hampers plan recognition applying to some areas. So, it is important to construct simple and efficient plan recognizing algorithm. Based on the study of plan knowledge graph, the concept of fuzzy cognitive plan map is proposed. Then, the recognizing algorithm based on fuzzy cognitive plan map is proposed. In this algorithm, it uses the operation of matrix to recognize agent’s plan, which can overcome the inefficiency of searching graph in plan knowledge graph. Experimental results show that the plan recognizing algorithm has excellent performance in plan recognition and the same recognition result is true for plan knowledge graph and it can reduce the recognition time greatly. The algorithm of plan recognition does well in intelligent game, smart home, network security and other fields, which has an important significance for extending the application area of plan recognition.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 15

Software trustworthiness has become one of the prominent studies in software quality assurance, in which the trustworthiness measurement is the primary topic. Compared with the method to evaluate the software development process, we measure to what extent the entity of software better fits users’ requirement. In this paper, we propose a bottom-up method of software trustworthiness measurement based on the source code. First, for the trustworthiness measurement of attributes, a comprehensive model is proposed. Second, the validity and stability of the model are verified by Monte Carlo simulation. Finally, the proposed method is developed based on the open source static detection tool for Cppcheck, which forms the software trustworthiness static measurement tool for TSMT.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 11

Architecture Led Incremental System Assurance short for ALISA presents a method to check if a system implementation meets its requirements. This method helps find errors in the early phase of system integration. ALISA provides four notations—requirements specifications, architecture models, verification techniques and assurance cases. The verification plan, which is designed by extracting information from requirement specification and architecture model, can be executed on the system and the result is significant metrics to judge the system quality. There are problems when generating a verification plan. As for the hierarchical architecture model with increasing complexity, the system may be divided into several parts and it is difficult to accomplish the assurance manually for each tier of the architecture. The approach also needs to respond to the ever-changing demands rapidly. New faults may be introduced artificially when designing requirement specifications and verification plans. In the paper, we propose an approach which uses ATL, whose full name is ATLAS transformation language, to help automatically generate verification plans. The meta-model of the verification plan and of requirement specification are given. Thus, designing the transformation rules from the verification part to the requirement part is easy. A lightweight template described in ATL is used to generate the verification plan for critical requirement and quality property.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 22

Because manufacturing grid resources have diversity, heterogeneity and dynamic characteristics, the existing resource retrieval and allocation methods do not take into account the load balance and quality of service simultaneously. In this paper, processing time and processing costs are taken as main factors to construct objective function, into which load balance factor is introduced. Also, completion quality and the reliability are taken as constraint conditions. Thus, a model for resources optimal allocation is proposed to satisfy service quality and maintain the balance of resource load. Further, a hybrid algorithm of ant colony and fish swarm is presented to solve the new model, and solution steps are given in detail. Simulation experiments are carried out in combination with the practical application. Resources load balance is improved significantly with the presented model, which shows that the method is efficient.

Submitted on September 29, 2017; Revised on October 9, 2017; Accepted on October 15, 2017
References: 17

The shape prior based level set method is widely used to segment and track objects’ contours in images and video sequences. However, such type of method is very slow and easy to fall into local minimum and obtain a wrong matching result. To overcome these issues, this paper presents a novel dynamic local level set method with shape prior. To speed up the local level set method, a genetic algorithm is used to dynamically choose the local region. Secondly, the genetic algorithm is also used to help the evolution process jump out of the local optimum when embedding shape prior into the level set function. The main the main strategy is using genetic algorithm to pre-choose shape priors and estimate the parameters. The experimental results prove the effectiveness and efficiency of the proposed method.

Submitted on September 15, 2017; Revised on October 20, 2017; Accepted on October 27, 2017
References: 11

In order to realize precise speed control of driving motor, an adaptive fuzzy PID control strategy for motors was established based on the existing proportional–integral–derivative (PID) control theory. The motor speed control model is built by simplifying the parameters of a brushless DC motor using the Sim Power Systems toolbox in MATLAB/Simulink environment, which involves the simulation of motor speed control including low speed, high speed, and road bump situations in city traffic environment. Results show that the time of the adaptive fuzzy PID control is 0.08s at low speed, the adjustment time of the conventional PID control is 0.22s, and the adjustment times are 0.12s and 0.32s at high speed. After encountering road bumps, the adaptive fuzzy PID control can quickly react and return to normal speed, whereas the conventional PID control is evidently affected by the interference.

Submitted on August 31, 2017; Revised on October 5, 2017; Accepted on October 23, 2017
References: 11

In order to achieve the optimal selection and allocation of various mould manufacturing resources under a manufacturing network environment, this study developed a bi-level programming model and proposed a related solution. Based on a thorough analysis of the allocation process of mould manufacturing resources, the calculation formulas of the task’s satisfaction with resource service quality and the load balance rate of resources were derived, and a bi-level programming mathematical model for the optimal allocation of mould manufacturing resources was established. Moreover, a hierarchical discrete particle swarm optimization (HDPSO) algorithm, which was based on the principle of non-dominated sorting in a multi-objective optimization, was designed for solving the model. Finally, a mould manufacturing project was selected for experimentally validating the feasibility of the proposed bi-level programming model for the optimal allocation of resources and the effectiveness of the proposed HDPSO algorithm.

Submitted on August 29, 2017; Revised on September 30, 2017; Accepted on October 18, 2017
References: 20

Text feature selection is one of the key steps in text classification, and thus can affect performance of text classification. In this paper, the feature dispersion degree of between-class documents is first put forward to measure the feature dispersion between categories (the greater its value, the larger the influence of the feature has). The feature concentration degree of within-class documents is then proposed to measure feature concentration in the text of a category (the greater its value, the larger the influence of feature has). Subsequently, a text feature selection method is presented, which uses both of the proposed degrees comprehensively to measure the importance of features. Experimental comparison results show that the proposed feature selection method can often get more representative feature subsets and improve performance of text classification.

Submitted on July 17, 2017; First Revised on October 7, 2017; Second Revised on October 15, 2017; Accepted on October 17, 2017
References: 11

Fault diagnosis for the maintenance of machinery is more difficult since it becomes more precise, automatic and efficient. To tackle this problem, a feature selection and probabilistic neural network-based method is presented in this paper. Firstly, feature parameters are extracted and selected after obtaining the raw signal. Then, the selected feature parameters are preprocessed according to the faulted characteristic frequencies of components. Finally, the diagnosis results are outputted with the decision method of PNN. Experimental data is utilized to demonstrate the effectiveness of this methodology.

Submitted on May 31, 2017; Revised on October 10, 2017; Accepted on October 18, 2017
References: 13