Xiaoyu Song

New particle swarm algorithm for job shop scheduling problems

A new particle swarm algorithm is proposed to make up for the drawback of conventional one on solving job shop scheduling problem, and improve the performance of searching solutions. In the new algorithm, the particle swarm algorithm is adopted to search in the solution space globally. According to the feature of job shop problem solutions, a critical operation-based selection method is proposed, and the TS algorithm based on it is adopted as the local algorithm to strengthen the capability of local search. The new particle swarm algorithm has been tested with 13 hard benchmark problems. The result demonstrates that the obtained best solution and the average value of ten times result are better than parallel genetic algorithm and taboo search algorithm. So the validity of the proposed new particle swarm algorithm is validated.

A Hybrid Algorithm of PSO and SA for Solving JSP

A hybrid algorithm of particle swarm optimization (PSO) and simulated annealing (SA) algorithm (HPSOSA) is proposed, which is used to overcome the deficiency of resolving job shop problem (JSP) and improve the quality of searching solutions. According to the characteristics of random and large-scale search of PSO, we adopt PSO to construct the parallel initial solutions of SA. At the same time, we increase shifting bottleneck technology and memory device in local SA. By this way, the search efficiency of SA is improved. HPSOSA algorithm has been tested with the 13 hard benchmark problems. The result shows that the average relative error percentage of the average value in ten time experiments is 2.46% and 0.08% which are respectively smaller than parallel genetic algorithm (PGA) and taboo search algorithm with back jump tracking (TSAB). So it can be concluded that the proposed hybrid particle swarm optimization algorithm is effective.

CBR based emergency case generating system under engineering accidents

With the flourishing development of construction and real estate industry, engineering accidents happen more frequently. The purpose of this paper is to generate emergency case as quick as possible so as to improve the quick emergency reaction capability when engineering accident happens. The flow of CBR based emergency case generating system (CBR/ECGS) under engineering accidents is studied. Both the static and dynamic case presentation technologies for CBR/ECGS are designed, and the case retrieval, case adaptation and case maintenance technologies for CBR/ECGS are proposed. It can be seen from running result of CBR/ECGS that, comparing with the result under the mode in which artificial adjustment is allowed, the overcast degree of emergency cases generated by CBR/ECGS under the mode in which only automatic system adjustment is allowed is almost the same. The validity of CBR/ECGS proposed in this paper is validated, and CBR is suitable for solving complex, non-structure problems such as emergency case generating under engineering accidents.

GM(1,1) based grey programming model for emergency goods scheduling

To schedule emergency goods scientifically and effectively so as to meet the demand of emergency goods within limited time under uncertain circumstance, grey theory is introduced to establish a multi-objective grey programming model in this paper. GM(1,1) is adapted and employed to predict emergency goods demand quantities. Both reducing emergency goods consumption and transportation cost to the greatest extent are considered during the course of emergency goods scheduling. In the objective function, the earliest emergency activity starttime, the fewest participated emergency goods supply nodes and the highest preference degree of emergency constraints is presented. The genetic arithmetic (GA) for above model is designed, and an instance abstracted from practical emergency scheduling work is solved. By adjusting three key control parameters of GA, test and improve the performance of the algorithm. The experimentation results by GA is compared with that by the time limit maximum(TLM) method. It shows that under the condition that the total emergency goods quantities provided by all supplied nodes is const, the optimizing capability of genetic algorithm is stronger than the TLM method when the emergency goods demand quantities are not too high. GA is suitable for solving the multi-objective gray programming problem.

Base state amendments spatio-temporal data model with dynamic selection of base state

The traditional Base State with Amendments Spatio-temporal data model often select the operation of base state on sequential. The method is relatively easy to implement, but there are obvious deficiencies. This paper introduces the concepts of operation of the base state and Operand, and proposes a base state amendments spatio-temporal data model with dynamic selection of base state. The retrieval time choose the operation of base state which having smaller operand in the adjacent two base states. The problem of selecting a distant base state causing a large number of redundant operations because of using sequential retrieval method can be solved by this approach, thereby enhancing the retrieval efficiency.

Base state with amendments spatio-temporal data model on time-slot retrieval based on temporary base-state

The method of traditional base state with amendments spatio-temporal data model processing slot-time retrieval is simply extended from time-point retrieval, neglected the characteristics of the adjacent of each point-time. This paper proposed the concept of temporary base-state, which set the current retrieved time as temporary base-state and the next point-time select the operation of base-state, having less operand between the temporary base-state and fixed base-state.

Design and implementation of villages and towns housing construction products intelligent service system

To improve the adaptation of the housing construction product in the villages and towns housing, meet the requirements of the villages and towns housing for the energy-saving building products and fittings. Microsoft.Net framework and the MVC design pattern are adopted, the other techniques are applied like Ajax, which design and construct the intelligent service system that apples for the villages and towns housing construction product. It realizes the generalized processing and the intelligent query optimization of the villages and towns housing construction product, affords an intelligent service platform for the popularization of the construction product and management. The implement of the system affords an efficient service platform to the management in villages and towns residential building construction product in our country, which has a high practical value and operability.

Complex network characters in object-oriented software desigh

In order to picture the complexity of software architecture, class collaboration network is defined to abstract the architecture of object-oriented software on class level. Eclipse is selected as example of object-oriented software, and its class collaboration network is constructed and the degree distribution of the network is analyzed by statistical physics with the conclusion that degree-rank, degree-pdf, degree-ccdf are all power law, which accords with the basic character of complex network. So based on complex network theory, degree-degree correlation, cluster-degree correlation and right-degree correlation in Eclipse class collaboration are all studied, and results show that in the class collaboration network, the degrees of the linked two nodes are negative correlative, and the out-degree of origin node and the in-degree of end node are negative correlative too; The cluster degree of Eclipse is big, and the cluster coefficient of the node and its degree are power law correlative; The right of the node in Eclipse network and its degree are linear correlative.

Algorithm of arcs recognition based on bar tracking

Recognition for arcs is an important and difficult problem in the study on engineering drawings vectorization. This paper presents an algorithm for recognizing arcs using bar tracking, which acquires the points chains of arcs and combined lines using bar tracking, then the combined lines are segmented, at last the arcs are fitted and validated. In experiments, we test the algorithm using benchmark data and real drawings. The experiments indicate that the algorithm is fast and effective.

Data mining-based engineering project grading technique

The purpose of this paper is to improve the quality of engineering project grading, the basic processes of data mining technique are introduced. Taking the engineering project grading as background, the implement cycles such as business understanding, data understanding, data preparation, modeling, evaluation and deployment are studied in detail. During modeling, the decision tree is adopted as analyzing modeling, and the conventional C4.5 algorithm is adapted. The adapted algorithm is applied to the engineering project grading, and its result is compared with that of conventional C4.5 algorithm. The comparing result demonstrates that for the complex system such as engineering project grading, it can improve in a certain extent on precision and obtained structure of decision tree, it can improve the quality of engineering project grading.