基于进化HMM模型的动态Agent系统建模方法

提出一种基于隐式Markov模型（HMM）的进化建模方法．使用进化算法随机搜索HMM的模型空间,自动选择HMM的结构和参数-完成对动态智能体系统行为的建模,学习智能体对周围环境的分割和反映方式．实验结果表明,该方法可以很好地搜索HMM的模型空间,并且避免了人工确定HMM模型结构的困难和手工设计模型所需的多次反复．     

A genetic-based neuro-fuzzy approach for modeling and control ofdynamical systems

Linguistic modeling of complex irregular systems constitutes the heart of many control and decision making systems, and fuzzy logic represents one of the most effective algorithms to build such linguistic models. In this paper, a linguistic (qualitative) modeling approach is proposed. The approach combines the merits of the fuzzy logic theory, neural networks, and genetic algorithms (GAs). The proposed model is presented in a fuzzy-neural network (FNN) form which can handle both quantitative (numerical) and qualitative (linguistic) knowledge. The learning algorithm of a FNN is composed of three phases. The first phase is used to find the initial membership functions of the fuzzy model. In the second phase, a new algorithm is developed and used to extract the linguistic-fuzzy rules. In the third phase, a multiresolutional dynamic genetic algorithm (MRD-GA) is proposed and used for optimized tuning of membership functions of the proposed model. Two well-known benchmarks are used to evaluate the performance of the proposed modeling approach, and compare it with other modeling approaches.     

粗糙集理论的分层递阶约简算法及其信息理论基础

本文模拟人类认知的分层递阶原则,提出一种粗糙集理论的分层递阶约简算法.该算法首先将信息系统或决策系统的知识在由部分属性所构成的多种层次和多种粒度上表示出来,然后分别对各个属性层次进行递阶约简.因此,该算法具有较强的实用性和较好的动态特性,并且能并行运算.同时,本文从信息理论的角度证明了分层递阶约简的理论基础.文章的最后将该算法应用于某水泥窑炉控制决策的获取中,证实了其有效性.     

Parallel inference in dynamic decision support systems of a semiotic type

The organization of parallel inference in dynamic decision support systems (DDSS) of a semiotic type, oriented towards a solving of ill-formed problems in dynamic applied domains, is considered. As a knowledge representation model, there are used production rules reflecting expert knowledge about a problem domain, an environment and decision making processes. The main concepts and assertions defining possibility and impossibility of parallel executing the production rules are given. Several types of parallelism in an inference process are introduced. The corresponding algorithm of parallel inference is described. Thus, the purpose of this paper is to develop and to research parallel inference methods and procedures that provide efficient processing a large amount of production rules for DDSS of a semiotic type.     

基于邻域粒化的混合信息系统动态规则提取

现有的混合信息系统知识发现模型涵盖的数据类型大多为符号型、数值型条件属性及符号型决策属性,且大多数模型的关注点是属性约简或特征选择,针对规则提取的研究相对较少。针对涵盖更多数据类型的混合信息系统构建一个动态规则提取模型。首先修正了现有的属性值距离的计算公式,对错层型属性值的距离给出了一种定义形式,从而定义了一个新的混合距离。其次提出了针对数值型决策属性诱导决策类的3种方法。其后构造了广义邻域粗糙集模型,提出了动态粒度下的上下近似及规则提取算法,构建了基于邻域粒化的动态规则提取模型。该模型可用于具有以下特点的信息系统的规则提取： （1）条件属性集可包括单层符号型、错层符号型、数值型、区间型、集值型、未知型等; （2）决策属性集可包括符号型、数值型。利用UCI数据库中的数据集进行了对比实验,分类精度表明了规则提取算法的有效性。     

VERIFICATION OF QUALITATIVE PROPERTIES OF RULE-BASED EXPERT SYSTEMS

Frequently, expert systems are developed to operate in dynamic environments where they must reason about time-varying information and generate hypotheses, conclusions, and process inputs that can drastically influence the environment within which they operate. For instance, expert systems used for fault diagnosis and fault accommodation in nuclear power plants reason over sensor data and operator inputs, form fault hypotheses, make recommendations pertaining to safe process operation, and in crisis situations, could generate command inputs to the process to help maintain safe operation. Clearly, there is a pressing need to verify and certify that such expert systems are dependable in their operation and can reliably maintain adequate performance levels. In this article we develop a mathematical approach to verifying qualitative properties of rule-based expert systems that operate in dynamic and uncertain environments. First, we provide mathematical models for the expert system (including the knowledge-base and inference engine) and for the mechanisms for interfacing to the user inputs and the dynamic process. Next, using these mathematical models, we show that while the structure and interconnection of information in the knowledge base influence the expert system's ability to react appropriately in a dynamic environment, the qualitative properties of the full knowledge-base/inference engine loop must be considered to fully characterize an expert system's dynamic behavior. To illustrate the verification approach, we show how to model and analyze the qualitative properties of rule-based expert systems that solve a water-jug filling problem and a simple process control problem. Finally, in our concluding remarks, we highlight some limitations of our approach and provide some future directions for research.     

一种基于信息粒度的动态属性约简求解算法

动态属性约简是粗糙集理论的重要研究内容之一.针对动态决策表构造了一种基于信息粒度的动态属性约简模型,详细分析了决策表中出现新属性动态增加时信息粒度的增量式计算方法;在此基础上,以信息粒度作为启发信息,设计了一种动态属性约简求解算法,该算法能有效利用原决策表的属性约简结果和信息粒度来降低算法的计算复杂度,并使得约简结果具有较好传承性;最后通过算例分析和实验比较进一步验证了本算法的可行性和有效性.     

ON KNOWLEDGE REPRESENTATION FOR INFERENCES IN CAST METHOD BANKS

This article explains why aspects of knowledge representation must be considered in the context of computer aided systems theory (CAST). CAST method banks support human experts during the process of problem solving. They should be understood as decision support systems, as assistants of their human expert users. One key to making this approach work is the communication between the expert and the system. The assistant should provide systematical and goal-directive information about the current problem state for the human expert. Another, even more important requirement is the assistant's knowledge about all available methods at a certain problem-solving state and their expected impact on the further problem-solving process. Knowledge representation denotes how the problem domain is represented within the support system and how it is used. We investigate different forms of knowledge representations and summarize criteria for the applicability of different forms of knowledge representations in CAST systems.     

Online updating belief rule based system for pipeline leak detection under expert intervention

A belief rule base inference methodology using the evidential reasoning approach (RIMER) has been developed recently, where a new belief rule base (BRB) is proposed to extend traditional IF-THEN rules and can capture more complicated causal relationships using different types of information with uncertainties, but these models are trained off-line and it is very expensive to train and re-train them. As such, recursive algorithms have been developed to update the BRB systems online and their calculation speed is very high, which is very important, particularly for the systems that have a high level of real-time requirement. The optimization models and recursive algorithms have been used for pipeline leak detection. However, because the proposed algorithms are both locally optimal and there may exist some noise in the real engineering systems, the trained or updated BRB may violate some certain running patterns that the pipeline leak should follow. These patterns can be determined by human experts according to some basic physical principles and the historical information. Therefore, this paper describes under expert intervention, how the recursive algorithm update the BRB system so that the updated BRB cannot only be used for pipeline leak detection but also satisfy the given patterns. Pipeline operations under different conditions are modeled by a BRB using expert knowledge, which is then updated and fine tuned using the proposed recursive algorithm and pipeline operating data, and validated by testing data. All training and testing data are collected from a real pipeline. The study demonstrates that under expert intervention, the BRB expert system is flexible, can be automatically tuned to represent complicated expert systems, and may be applied widely in engineering. It is also demonstrated that compared with other methods such as fuzzy neural networks (FNNs), the RIMER has a special characteristic of allowing direct intervention of human experts in deciding the internal structure and the parameters of a BRB expert system.     

Dynamic probabilistic analysis of accidents in construction projects by combining precursor data and expert judgments

Construction accident occurrences are essentially rare, stochastic, and dynamic. This study proposes a method for accident prediction that fully captures these natures based on historical data and prior knowledge. The method utilizes the relatively high occurrence frequency of precursor events and the dependency between precursors and accidents. The modeling approach consists of three steps: (1) characterize the stochastic occurrences of precursor events over time based on precursor data; (2) estimate the failure rate of the Poisson model which is assumed to be a prior distribution of accident occurrences; and (3) elicit the expert knowledge about the stochastic dependency between near miss occurrences and accident occurrences. A copula-based Markov model is used to develop the time series model of precursors while a copula-based protocol is proposed to aid expert judgment elicitation and quantification. The probability of accident occurrence is then dynamically updated according to the observed historical near miss numbers. The proposed method is applied to a metro construction project. A five-year long near miss data were collected and used as accident precursor data, while three experts were invited to provide relevant information. The developed accident model is used to predict the accident-prone periods, which are consistent with the months that the observed near miss occurrence frequency deviates significantly from normality. Thus, the model can be used to support the planning of necessary safety improvement programs before the accident risk increased.     

基于测度递进的模糊认知图及其应用

模糊认知图(fuzzy cognitive map,FCM)具有简单的推理机制和较强的因果关系表达能力,已得到广泛关注和研究,但FCM对专家经验知识具有较强的依赖性,故而限制了在复杂动态系统建模中的应用.基于此,提出了一种测度递进策略的模糊认知图学习方法.利用线性回归算法,学习得到模糊认知图权重矩阵粗模型;将神经网络的权值调整算法应用于权重矩阵粗模型的细化过程,将该模糊认知图模型应用在股票市场,实现对股票日均值的预测.实验结果表明了该建模方式是有效的.     

动态电源管理的随机切换模型与策略优化

提出一种基于连续时间Markov决策过程的动态电源管理策略优化方法.通过建立动态电源管理系统的随机切换模型,将动态电源管理问题转化为带约束的策略优化问题,并给出一种基于矢量合成的策略梯度优化算法.随机切换模型对动态电源管理系统的描述精确,策略优化算法简便有效,既能离线计算,也适用于在线优化.仿真实验验证了该方法的有效性.     

基于知识图谱的智能决策支持技术及应用研究

知识图谱是把复杂的领域知识通过数据挖掘、信息处理、知识计量和图形绘制而显示出来,解释知识领域的动态发展规律。知识图谱把所有不同种类的信息(heterogeneous information)连接在一起得到一个关系网络并从"关系"的角度去分析问题。知识图谱目前被广泛应用于智能搜索、智能问答等领域。提出了一种基于知识图谱的智能决策支持框架,用于解决传统决策支持系统存在的问题。通过大数据、知识图谱等海量知识分析和模型构建技术,结合决策支持系统,增强对问题的分解与处理、形成具有关系型网络的知识系统。最后结合电信领域中的经典决策案例,搭建基于知识图谱的欺诈电话智能决策支撑平台。和传统的决策支持系统比较,该研究方法的优点在于结合大数据处理方法提升了知识建模的算力和决策支持的效率,使实时处理大规模信息数据成为现实;基于知识图谱的关系型网络,提升了决策模型的准确性和关联相关性。     

ADAPTIVE DECISION SUPPORT SYSTEM FOR AUTOMATIC PHYSICAL EFFORT PLAN GENERATION—DATA-DRIVEN APPROACH

Mathematical models delivered using both expert knowledge and experimental data improve understanding of dynamic properties of the system under consideration. This is useful for different purposes, such as prediction, diagnosis, decision making, and system control. A data-driven approach has been found to be particularly useful in designing adaptive decision support systems. We demonstrate the usefulness of data-driven models in a custom application designed for sport training management. We have developed a system that makes use of expert knowledge together with measurement data (heart rate, electromyography, and acceleration) as well as environmental (Global Positioning System) in order to generate an optimal training plan. The system performs such tasks as modeling of the athlete's cardiovascular system, estimation of the athlete's parameters, and adaptation of the model to the athlete.     

Expert systems and economic policy analysis: An application to OCS auction leasing

In this paper we argue that expert systems can be powerful tools for modelling microeconomic systems, including both individual decision making and the coordination of individual agents in a resource allocation mechanism. Using the fact that expert systems are essentially computerized versions of decision processes, we illustrate how they can be viewed as generalized process models of decision-making. We argue that the expert system approach is beneficial because it allows a policy analyst to explore the implication of policy alternatives without having to incur the generally prohibitive cost of field implementation studies. Further, enables the incorporation and updating of decision strategies and qualitative information, which human experts typically use but which is not amenable to pure mathematical modelling.One particular microeconomic system we suggest could be modelled as an expert system is the OCS offshore oil lease auction process. Moreover, we argue that constructing such an expert system model would require the development of two integrated expert systems: one for the auction process and subsequent resource allocation and the other to model the individual bidding behavior of the auction participants. We set out the structure of the auction expert system in some detail and discuss rules of thumb used by bidders inferred from our empirical research on past OCS auctions.Such an expert system of an auction leasing process could provide benefits to both bidders (e.g., oil companies) and the auctioneer (e.g., the Department of the Interior) as well. Bidders, by trying different strategies against different hypothesized strategies by their opponents could use such an integrated expert system to improve their bidding performances. The auctioneer, on the other hand, could test the efficiency of various proposed auction institutions under different assumptions about bidding behavior. In some circumstances, it might be desirable to even automate the auction process with a network coordinating the expert systems used by the individual firms and a computerized auctioneer.     

A Strategy of Dynamic Reasoning in Knowledge-Based System with Fuzzy Production Rules

Fuzzy production rules have been successfully applied to represent uncertainty in a knowledge-based system. The knowledge organized as a knowledge base is static. On the other hand, a real system such as the stock market is dynamic in nature. Therefore we need a strategy to reflect the dynamic nature of a system when we make reasoning with a knowledge-based system.This paper proposes a strategy of dynamic reasoning that can be used to takes account the dynamic behavior of decision-making with the knowledge-based system consisted of fuzzy rules. A degree of match (DM) between actual input information and antecedent of a rule is represented by a value in interval [0, 1]. Weights of relative importance of attributes in a rule are obtained by the AHP (Analytic Hierarchy Process) method. Then these weights are applied as exponents for the DM, and the DMs in a rule are combined, with the Min operator, into a single DM for the rule. In this way, the importance of attributes of a rule, which can be changed from time to time, can be reflected to reasoning in knowledge-based system with fuzzy rules.With the proposed reasoning procedure, a decision maker can take his judgment on the given decision environment into a static knowledge base with fuzzy rules when he makes decision with the knowledge base. This procedure can be automated as a pre-processing system for fuzzy expert systems. Thereby the quality of decisions could be enhanced.     

Integrating preventive and predictive maintenance policies with system dynamics: A decision table approach

A decision table is a practical tool that helps systems planners to make operational decisions, especially when they are under stress. With the effect of recent trends, such as the use of machine learning, data mining, and reinforcement learning methods, the maintenance decision has been a dynamic issue depending on system conditions. An expert may execute the maintenance or wait for the next periodic maintenance due to lack of maintenance workers, tools or budget, resources, etc., although the intelligent method predicts a failure approaching. Even sometimes, he/she may ignore the current periodic maintenance. Our method allows making some changes in the maintenance plan systematically. It integrates the results of preventive and predictive maintenance policies, and as different from the literature, it allows ignoring some maintenance actions depending on the maintenance resource levels in a decision table. Such a strategy helps to allocate limited resources to maintenance actions reasonably. We conducted an extensive simulation study on a real-life dataset. The preventive maintenance period is determined using classical approaches such as Weibull analysis. A machine learning algorithm is utilized to predict the type of failure. We have analyzed the performance of the proposed decision table approach under a variety of scenarios and with different parameter settings. We also showed the effect of parameter settings and the marginal utility of each maintenance policy. In addition, the approach provides several choices for planners. As a result, the proposed approach improves the system’s sustainability compared to traditional policies.     

基于粗糙集理论的计算机图像识别系统

为解决专家系统较难获取完备知识的瓶颈问题,设计了一种基于粗糙集理论的诊断系统。研究历史数据所形成的决策表,运用粗糙集理论进行约简,构建专家系统知识库模型。通过计算诊断规则粗糙度,确定诊断规则的置信程度。利用推理机,实现对知识库的动态维护。结合诊断的特点,建立基于粗糙集理论的专家诊断系统模型,快速准确地实现诊断的目标。实例表明,该专家诊断系统有效、实用,具有很好的学习能力。     

概率粗糙集三支决策在线快速计算算法研究

随着大数据和物联网技术的不断发展,动态在线计算已经成为了一种常见的计算模式,在动态在线计算中进行不确定问题的推理和求解是一项具有挑战性的新议题。概率粗糙集三支决策理论是一种处理不确定性知识挖掘的有效工具,根据在线计算模式中数据同步增减的动态特点,提出了一种概率粗糙集三支决策的在线计算方法。首先,以内存滑动窗口模式对在线动态计算的数据变化特点进行理论建模;然后,根据上述模型中在线计算的数据变化模式,推导出不同类型数据变化模式下的三支决策条件概率及三支区域的变化规律;最后,提出了一种新型在线快速计算算法,其获取的三支决策规则与经典概率三支决策算法是等效的。通过与经典三支决策计算算法的多组对比实验,验证了提出的在线快速计算算法的高效性与稳定性。     

Adaptive visual tracking using the prioritized Q-learning algorithm: MDP-based parameter learning approach

This paper introduces an adaptive visual tracking method that combines the adaptive appearance model and the optimization capability of the Markov decision process. Most tracking algorithms are limited due to variations in object appearance from changes in illumination, viewing angle, object scale, and object shape. This paper is motivated by the fact that tracking performance degradation is caused not only by changes in object appearance but also by the inflexible controls of tracker parameters. To the best of our knowledge, optimization of tracker parameters has not been thoroughly investigated, even though it critically influences tracking performance. The challenge is to equip an adaptive tracking algorithm with an optimization capability for a more flexible and robust appearance model. In this paper, the Markov decision process, which has been applied successfully in many dynamic systems, is employed to optimize an adaptive appearance model-based tracking algorithm. The adaptive visual tracking is formulated as a Markov decision process based dynamic parameter optimization problem with uncertain and incomplete information. The high computation requirements of the Markov decision process formulation are solved by the proposed prioritized Q-learning approach. We carried out extensive experiments using realistic video sets, and achieved very encouraging and competitive results.