Model development fir evaluation of process inefficiency and optimization

我们研究了社会系统和生产系统过程无效率的评价和优化方法。各种情况下DEA方法的发展(1)给出了效率分解和乘数界的性质。(2)建立了基于上下文的DEA模型，引入了基于松弛度的效率度量方法，提供了合适的效率评价和改进目标。(3)建立了评价供应链绩效效率的DEA模型。(4)结果表明，DEA可以有效地应用于多目标调度的Pareto评价。(1)将无效率的概念引入统计方法，考虑用无效率识别模型来定量评价无效率。(2)建立了低效率约束下的回归模型。针对多输入因素的多元回归模型，提出了一种高效的无效率缺口识别算法。(3)作为对其他统计方法的扩展，我们采用随机前沿分析建立了无效率的析因实验模型。本文还研究了应用DEA方法寻找技术和绩效进步的前沿转移指标、竞争环境下的绩效指标、采用正交布局的变量选择方案等。这些结果将在使用实际数据验证其有效性后发表。

We have studied the evaluation and optimization methods to identify the process inefficiency in social systems and production systems.1. Development of DEA method for various circumstances(1) We have shown the properties on the decomposition of inefficiency and multiplier bounds.(2) We have developed the context-dependent DEA and introduced the slack based measure of efficiency, which provide the appropriate efficiency evaluation and goal for improvement.(3) We have developed the DEA model to evaluate the efficiency of supply chain performance.(4) We have shown that DEA can be effectively applied to Pareto evaluation for multiobjective scheduling.2. Development of inefficiency identification model for statistical method(1) We have considered the inefficiency identification model to evaluate the quantitative inefficiency by introducing the concept of inefficiency into statistical method.(2) We have developed the regression model subject to inefficiency. The efficient algorithm to identify the inefficiency gap for multi regression model to deal with multiple input factors.(3) As an extension to the other statistical method, we have developed the factorial experiment model subject to inefficiency by adopting stochastic frontier analysis.Other works on the application of DEA method have been studied, such as the index for frontier shift to find the progress of technology as well as performance, the performance index under competitive environment, variable selection scheme by using orthogonal layout. These results will be published after the verification of their effectiveness by using practical data.