An optimization method based on an improved least squares support vector regression machine（LS-SVR） and the technique for order preference by similarity to an ideal solution（TOPSIS） was proposed for the parameter optimization of fused deposition molding（FDM） multiresponse that simultaneously reduced warpage and shortened printing time. Firstly,the print speed was considered as a functional parameter and modeled using a Bezier curve. The curve definition points together with nozzle temperature,hotbed temperature,and fill rate were the design variables.The experimental design was carried out by Latin hypercube sampling. Secondly,the LS-SVR based on the kernel function improved by the Fréchet distance was used to model the action relationship between parameters and each response,so that the functional parameters were embedded in the overall model in a functional form. Finally,the non-dominated sorting genetic algorithm Ⅱ and TOPSIS were used to find and determine the optimal combination of parameters. The optimization results show that the proposed method is better than the LS-SVR method based on Euclidean distance and the LS-SVR based on scalar parameters,which illustrates the effectiveness of the method.
As a critical activity of urban waterlog response,reasonable allocation of pumping equipment can not only decrease economic costs,but also decrease a series of risks,such as the loss of residents’ property and casualties. This paper presented a two-stage stochastic nonlinear programming model with objectives of minimizing operational and transportation costs of pumping equipment,and penalty cost（which is a function of pumping time）. The quantity of pumping equipment,pumping time and transportation time were constructed as a nonlinear term,and piecewise linear approximation was designed. A progressive hedging-based solution methodology was designed for large-scale problems. Pudong District of Shanghai was employed as the research region. Rainfall intensity and scope were considered to design a serial of scenarios for numerical studies. The results show that piecewise intervals have great impact on error of total costs and computational time. Moreover,the proposed algorithm is efficient for large-scale problems.
The objective in this study was to investigate the effects of multitasking type and information modality on task performance and users’ subjective perceptions in multitasking with intelligent assistant. An experiment with a two-factor within-subject design was used to test task performance,perceived ease of use,perceived usefulness,intention to use and cognitive load. Thirty participants were recruited to complete parallel multitasking and sequential multitasking tasks in three information presentation scenarios（visual-auditory bimodality, visual unimodality, and auditory unimodality）. Obtained data was analyzed by repeated measures ANOVA. The results show that parallel multitasking takes significantly less time to complete than sequential multitasking. Visualauditory bimodality is significantly better than visual unimodality and auditory unimodality in terms of task completion time and subjective perceptions. The interaction of multitasking type and information modality is not significant. The experiment provides a theoretical basis and practical guidance for the design of information presentation modes and task organization modes for intelligent assistant.
In an uncertain environment,the project planning and controlling are very important for the smooth implementation of the project. Based on the simulation analysis,how to determine the best combination of project input parameters for robust scheduling optimization under random activity duration was studied. Firstly,the research problem was defined. Then,a robust project scheduling optimization model was constructed. The simulation analysis process was designed. The corresponding relation between project input parameters and simulation indicators was analyzed. Next,the tabu search heuristic algorithm was developed to solve the robust scheduling optimization model.The simulation implementation process was described. Finally,an real case was used to illustrate the research. The results show that the robustness of schedule increases with the increase of project schedule duration and resource availability and remains stable after reaching a certain value. When different simulation indicators are selected,the optimal combination of the input parameters is different. This allows the manager to select the corresponding indices according to their actual needs.Guidance for robust project scheduling under uncertain conditions is provided. Improvement of robustness of project scheduling is made through validation.
Drones have been used to improve the efficiency of urban logistics and distribution "last mile" because they are not restricted by terrain. However,they were limited by the maximum flight time and maximum load. Combining with the characteristics of truck and drone delivery,a class of truck distribution route optimization problem with drone assistance was studied. Considering the factors such as the maximum flight time,maximum load and flight speed of drone,a mixed integer programming model was established to minimize the delivery completion time. A hybrid variable neighborhood search algorithm combined with adaptive K-means clustering search was designed. A calculation example was constructed based on Solomon Benchmark C101,R101,and RC101. The analysis results show that the hybrid variable neighborhood search algorithm embedded with a simple heuristic algorithm can solve the proposed routing optimization problem and improve the timeliness of logistics and distribution services. The less the flight speed is affected by the load and the longer the flight duration,the more conducive to shortening the delivery time.
Most of the existing internal knowledge management systems of manufacturing enterprises select the document labels manually,which is inefficient. Extracting keywords automatically to generate document labels using natural language processing technology contributes to the intelligentization of the knowledge management system. For the keyword extraction of automobile research and development documents, this paper proposed the BERT-BiLSTM-TFIDF keyword extraction model. This proposed model added sentence weights and external corpus to improve TFIDF method. The sentence weights were calculated with a designed BERT-Bi LSTM model. The proposed keyword extraction method has improved the shortcomings that the existing keyword extraction methods could not make use of the semantic information and context of the word. The proposed BERT-Bi LSTM-TFIDF method achieves a good result through experimental verification.
Aiming at the problems of low yield,unreasonable workshop layout,and high machine failure rate of JH’s automotive air-conditioning aluminum flat tube production line,a closed-loop optimization process of "discovering problems（VSM and ISM）,scheme design,and simulation verification" was proposed. Firstly,the current value stream map of the production line was drawn to discover the problems. Secondly, a production line problem interpretive structure model was established. The production line problems were divided into deep-level problems,middle-level problems and surface problems,in order to find out the root causes and direct causes of the low efficiency of the production line. Finally,considering the time and cost factors,a multi-objective optimization model was established for workshop layout,and solved by PSO algorithm,while lean thinking was put forward to improve the program to other issues,and Flexsim was used for simulation.This result shows that the combination of VSM and ISM can be more efficient in the process of optimizing the production line.
Shelf space allocation has always been the retail market’s focus because of the shortage of shelf resources. First,a mixed-integer non-linear programming model was established for the twodimensional shelf space allocation problem with the influence of the product spatial adjacency relationship（SAR）. Secondly,an improved random key genetic algorithm（RKGA） was designed to solve the model. Finally,two-scale instances were simulated to verify it. Results show that the improved RKGA has superior performance over the benchmark genetic algorithm and Lingo. In smallscale instances,the improved RAGA has significantly improved solving accuracy and efficiency compared to Lingo. In large-scale instances,it also performs better when the convergence time is similar to benchmark GA. The profit value of all instances has increased by adding SAR effect. As the number of products increases,the profit and its increment appear to increase first and then decrease.
In order to minimize the operating cost and gaseous pollutant emission of the microgrid,which includes renewable energies,a 24 hours day-ahead multi-objective interval optimization model was built. Simultaneously,the possibility degree was introduced to handle uncertain variables and the model was converted for satisfying the decision makers’ risk preference. In addition,in view of the model characteristics,the metaheuristic strategies about initialization and repair of solution were designed. Meanwhile, a multi-objective evolutionary algorithm based on hybrid decomposition（MOEA/HD） was constructed,which used the fuzzy membership degree and Chebyshev function in parallel to decompose the multi-objective optimization problem,to solve the model. Finally,the simulation results proved that the MOEA/HD was more efficient,which could get a set of nondominated solutions with higher quality,wider range and more uniform distribution compared with other algorithms.
The flexible assignment of multi-skilled workers is critical to realize the flexibility of production in Seru production system. Studies always focus on improving productivity,which easily lead to unfair workload allocation and affect the acceptance of workers for production plan. From the perspective of realizing equitable allocation of workload among multi-skilled workers, a Min-Max type optimization indicator was used to describe the workload equity among workers and an equity-oriented mixed integer nonlinear programming model with the objective of minimizing the maximum workload of workers was proposed to solve the multi-skilled worker-batch-task assignment problem in Seru production system. To verify the effectiveness of the equity-oriented model, an efficiency-oriented model with the aim of minimizing the total labor time was compared. In contrast with the efficiencyoriented model, the exact solutions of the computational instances based on both multi-skilled workers and full-skilled workers show that the equity-oriented model can significantly realize the fair assignment of workload among multi-skilled workers within an appropriate production efficiency level.
The typical GERT model can only handle exclusive-OR type nodes,which greatly limits its practical application. Therefore,based on the Phase-Type（PH） distribution,a flexible GERT model that could handle exclusive-OR,OR and AND nodes was proposed. First of all,according to the closeness of the PH distribution operation,a PH distribution based generalized activity network（GAN） was transformed into a PH distribution-based GERT network. The problem of moment generating function when the GAN converted into the GERT was solved. Secondly,the transfer function was constructed based on the moment generating function of the PH distribution. The flexible GERT network could be solved by the Mason formula. It was proved. Furthermore,considering the denseness of the PH distribution,the PH distribution could fit all the distributions on the positive part of the axis. The HyperStar tool based on the principle of clustering fitting was used to fit a large amount of activity time data collected by the sensor to obtain the corresponding PH distribution parameter. Then,with the help of the GERT analytical method and Mathematica12.0software,the parameters of expected completion time,variance and probability of completion of the flexible GERT network were obtained. Finally,an engine overhaul example was used to verify the practicability of this method.
With regard to the "supplier-led" supply chain consisting of multiple suppliers and multiple retailers,three Stackelberg game models were constructed by not considering the constraints（CN）,only considering the countervailing power（MN）,and jointly considering the countervailing power and capacity constraints（MY）. The influences of the number of retailers and the fluctuation of market demand on equilibrium decisions and profit were investigated. Furthermore,the differences in decisions of supply chain with the capacity constraints or not under countervailing power were discovered,where the numerical analysis was used to demonstrate the optimal decision of supply chains. It is found that from the retailer countervailing power side,the increase in the number of retailers will weaken the buyer power,and the retailers will willingly endorse the extra expenses for their orders as well,thereby resulting in higher wholesale prices. In this case,under the consideration of capacity constraints,when the relevant suppliers’ production capacity is lower than the equilibrium production without capacity constraints,the effects on wholesale keep positive and the supplier profits are not always decreased.
In order to improve the user interaction efficiency,accuracy and user comfort of the natural hand interactive interface in virtual reality（VR） scenes,an empirically comparative evaluation on the natural hand interaction performance and user perceived fatigue level was conducted in performing different arm postures at diverse target positions. The evaluation was completed through a target selection experiment,in which both the subjective and objective results were measured,including the task completion time,target selection accuracy and the user perceived arm fatigue. The results were also compared in different conditions of the target position and the arm postures. The results show that natural hand interaction in spaces at the same side of the hand is more efficient and accurate than in spaces at the converse side. The former is also found to be more comfortable with fewer perceived arm fatigue. Hand interaction at the height lower than the shoulders is also found to be more efficient,accurate and labor-saving than that at higher positions.
溯源系统的引入不仅为回收提供便利,还可进一步提高消费者的信任度,从而提高回收量.但不同回收模式下的回收量因受供应链各级成员溯源水平的影响会出现差异,这使得回收模式选择及供应链决策变得更加复杂和困难.因此,开展以制造商为领导者的制造商回收、零售商回收、制造商和零售商混合回收三种模式下的溯源水平、定价策略和回收量的最优决策研究具有重要的应用和学术价值.研究发现:溯源水平与溯源成本边际系数、回收成本节省和回收模式竞争程度有关.溯源水平的提高会提高批发价和销售价,降低回收价.制造商应根据消费者溯源偏好、溯源对回收的影响程度和成本弹性系数设置溯源水平.当竞争程度较低时,制造商会选择混合回收模式,否则选择自行回收;当消费者偏好高于一定阈值时,混合回收模式下回收量的提高更加明显,对环境也更加有利.
In the operation process of the intelligent drug distribution robot in the hospital,the error of the operator may lead to the failure of the operation task. In order to study the human factor reliability of the operation process,the cognitive reliability and error analysis method were used to predict the human error rate. The general performance condition was evaluated quantitatively by the analytic hierarchy process. The operational task was analyzed by the hierarchical task analysis method. Finally,the overall human error rate was obtained by formula. The results demonstrate that the improved method can more accurately predict the human error rate and identify operations with a high error rate during the operation of personnel,thereby effectively avoiding the occurrence of human error. It also makes a certain reference value for the analysis and prediction of human factors in other industries.
The environmental protection policy of garbage classification has been implemented in major cities across the country one by one. However,the survey shows that there are still problems of citizens’ insufficient knowledge of waste separation and perfunctory approach to the separation process.Therefore, in order to study how the government can motivate residents to make efforts in participation,a pure moral hazard model in which neither principal nor agent knew the sorting ability ex ante was firstly established. Then,a mixed model in which adverse selection and moral hazard existed at the same time was established from a realistic perspective,in order to motivate residents to make efforts while considering whether to screen residents from the perspective of government benefits in response to adverse selection. It finds that,the presence of adverse selection makes the incentive cost of dual information asymmetry higher than that of a single moral hazard in general.However,when the probability of an environmentalist’s effort to obtain a reward meets certain conditions,the presence of adverse selection instead makes the government more favorable. When the probability of receiving a reward is high and the quality of the population is generally high,the government should choose the optimal reward scheme under a uniform incentive policy that ignores information screening. When the probability of receiving a reward is moderate and the ability of the population varies,the benefits of using two incentive strategies for information screening are higher than the benefits of not screening the type of population. It also finds when the probability of environmentalists receiving rewards is relatively moderate and low,their positive effect is less. The government has to increase the rewards to ensure residents’ participation. The incentives for nonenvironmentalists will be wasted. Therefore,the government should only consider a uniform incentive policy that encourages residents to make efforts to sort.
Effective scheduling of multi-skilled human resources is essential to the success of software projects. In software development,employees’ personality traits often affect the performance of their work. The software project scheduling problem with multi-skilled employees considering employees’ personality was studied. The proposed problem aimed at assigning employees with different skills and personality traits to suitable tasks,and determining the start time of each task under the constraints of skills and precedence relations,such that the total cost of the project was minimized.A mixed integer linear programming model was formulated and a double priority rules based heuristic algorithm was proposed. Based on a benchmark dataset that was generated using full factorial design of experiment,computational experiments were conducted to analyze the performance of the proposed algorithm. The algorithm was compared with the commercial solver CPLEX and genetic algorithm.The results show that the algorithm is competitive in terms of efficiency and effectiveness.
In order to reduce the cost of port operation as much as possible under the premise of ensuring the ships to complete the loading and unloading operation as planned,an optimization model of berth and quay crane cooperative scheduling based on quay crane sharing strategy was constructed with a goal of the minimum total cost of the port side,and decision variables were berthing position of ships,numbers of opened quayside and calculated numbers of quay cranes. This model took advantages of the feature that the container quayside bridge could move between adjacent ships.According to the characteristics of the model,an algorithm based on the genetic algorithm was designed,in which the position optimization module and quay sharing optimization module were nested. Finally,the method was used to optimize the scheduling of container terminal operations in Dalian port,which verified the effectiveness of the model and algorithm. The results show that the method can reduce the number of quay crane start-up and save the port operation cost. Through the algorithm comparison,the designed algorithm has better optimization results and stability.
Individual environmental awareness and behavior states play an important role in the process of green products diffusion. Environmental awareness increases the preference of individuals to choose the green products,and individual behavior states change the interaction patterns of individuals.The influence of individual environmental awareness and behavior states on the diffusion of green products was studied,and a multi-layer network spreading model based on complex network theory was proposed. The model was composed of environmental awareness spread as the upper layer and green products diffusion as the lower layer. By the mean field method,the dynamic evolution process of the system was analyzed and the critical thresholds of uncorrelated heterogeneous networks were calculated. Finally,the numerical simulation of the scale-free network verified the theoretical analysis and studied the final size of green product diffusion in different situations. The results show that the diffusion scale of green products is positively correlated with environmental awareness and individual activity.
The "winner takes all" feature requires that the pricing strategy of the internet platform gives priority to obtain consumer scale. In view of this,a duopoly competition model of internet platform was constructed to investigate the equilibrium of competition under different pricing strategies and explore the maintenance effect on the scale of consumer users from partial zero pricing strategy.Then the motivation and effect of the partial zero pricing strategy of the internet platform were examined. The research finding is that the internet platform uses price discrimination to divide consumers into two groups under the partial zero pricing strategy:the consumer with low willingness to pay and the consumer with high willingness to pay. Then the internet platform sets a uniform price for consumers with high willingness to pay,and sets discriminatory prices（i. e.,below the uniform price and zero price） for consumers with low willingness to pay. Thus,the competitive results fill the gap of consumer scale caused by setting the uniform price,which covertly increases platform profits and gradually cannibalizes consumer surplus.
