Constraint-based fuzzy logic controllers (CFLCs) have been recognized as a unified framework of control system design. Arbitrary types of constraints are allowed in a CFLC to specify the desired states of a plant. the...
详细信息
ISBN:
(纸本)0780370783
Constraint-based fuzzy logic controllers (CFLCs) have been recognized as a unified framework of control system design. Arbitrary types of constraints are allowed in a CFLC to specify the desired states of a plant. the purpose of this paper is to present a design methodology of CFLCs, transforming the intention of a control system into a network of fuzzy constraints. the task of designing a CFLC could be viewed as a process of incremental constraint-based problem-solving, namely a fuzzy constraint refinement problem. Specifically, a schema could be derived for the configuration of a CFLC;after that, constraints can be gradually added (or removed) into the schema until the performance requirements are met. Based on this framework, we have successfully designed a CFLC for modeling a simplified rotary cement kiln that has both nonlinear and time varying characteristics of a plant.
the proceedings contains 48 papers. Topics discussed include networking, distributed file and storage systems, video on demand, high speed networking, embedded and internetworking technologies, network routing, modeli...
详细信息
the proceedings contains 48 papers. Topics discussed include networking, distributed file and storage systems, video on demand, high speed networking, embedded and internetworking technologies, network routing, modeling algorithms and techniques, distributed systems, large scale distributed storage system, home and multimedia networking, high performance computing and real time systems.
this paper describes the design of a Fuzzy Logic Scheduled Controllers used for side-slip velocity control of a missile autopilot design. A multi-objective evolutionary algorithm is used to determine the membership fu...
详细信息
ISBN:
(纸本)0780370783
this paper describes the design of a Fuzzy Logic Scheduled Controllers used for side-slip velocity control of a missile autopilot design. A multi-objective evolutionary algorithm is used to determine the membership function distribution of the fuzzy trajectory controller within an outer loop control system. Scaling factors of the FLC inputs and outputs for each required demand are obtained by using a polynomial fit for a large range of multiple velocity demands (1g,5g,10g,15g-lateral acceleration equivalent). the design meets mulitple objectives related to closed loop performance such as: steady state error, overshoot, settling and rise time. Multiple solutions are obtained simultaneously by using nondominated sorting for forming the Pareto front, combined with a reference point approach to incorporate preference information into the GA to direct the search towards feasible desirable areas which satisfy specific values of the objectives. Simulationn results are presented showing the fuzzy gain surface and extreme models in the multiple model population.
A demonstration program called the Weapons system Open Architecture (WSOA) enabling innovative techniques for time critical target prosecution was discussed. the WSOA used tactical communication links, collaborative p...
详细信息
A demonstration program called the Weapons system Open Architecture (WSOA) enabling innovative techniques for time critical target prosecution was discussed. the WSOA used tactical communication links, collaborative planning, information mining, quality of service based technology, and link bandwidthoptimization. Common object request broker architecture (CORBA) in WSOA enabled distributed object computing on different operating systems and hardware platforms.
PLNs are neural networks with linear and metric neurons that separate a non-linear input space into several linear regions. the separation is done by LVQ-like metric neurons [1]. Linear output neurons provide a linear...
详细信息
ISBN:
(纸本)0780370783
PLNs are neural networks with linear and metric neurons that separate a non-linear input space into several linear regions. the separation is done by LVQ-like metric neurons [1]. Linear output neurons provide a linear mapping from input to output space. Unlike other neural networks there is a 3-dimensional weight matrix between input, hidden and output layer rather than a 2-dimensional weight matrix between hidden and output layer. For training of PLNs different training strategies like gradient descent or linear regression exist that are combined with a constructive method producing hidden units [2]. the fastest method is an incremental regression with only 2 to 5 cycles of the complete training set. Incremental training means that the network weights are refreshed after each presentation of a pattern. One drawback of this method is the non-smooth approximation of the objective function, especially for those spots corresponding to the first few patterns of a linear region. One bad effect is an insufficient generalization in this region. Pruning solves this problem. Different strategies are known like Optimal Brain Damage or Optimal Brain Surgeon. Both are very time consuming. A novel algorithm demonstrates a good trade-off in respect of generalization quality and computing effort. Results are presented. Applications are seen mainly in approximation tasks and here, especially, in control tasks and system identification [3].
this paper described a heuristic search technique carrying out collision avoidance for Autonomous Underwater Vehicles(AUVs). Fuzzy relational products are used as the mathematical implement for the analysis and synthe...
详细信息
this paper described a heuristic search technique carrying out collision avoidance for Autonomous Underwater Vehicles(AUVs). Fuzzy relational products are used as the mathematical implement for the analysis and synthesis of relations between obstacles that are met in the navigation environment and available candidate nodes. In this paper, we propose a more effective evaluation function that reflects the heuristic information of domain experts on obstacle clearance, and an advanced heuristic search method performing collision avoidance for AUVs. the search technique adopts fuzzy relational products to conduct path-planning of intelligent navigation system. In order to verify the performance with Å search method through simulation in view of the CPU time, the optimization of path and the amount of memory usage.
In the context of object-oriented formalism, stereotypes are presented as a new means for user-defined extensions of the Unified modeling Language (UML). Unfortunately, the stereotypes syntax and semantics vary depend...
详细信息
the explosive spread of the Internet in recent years has made it possible to access a huge number of information sources. these sources include web home pages, relational databases, image databases and so on. To enabl...
详细信息
this paper describes a newly developed digital dynamic simulation software package, which is used to simulate the fatigue & fracture process of offshore structures under random ocean loading. the system is program...
详细信息
this paper describes a newly developed digital dynamic simulation software package, which is used to simulate the fatigue & fracture process of offshore structures under random ocean loading. the system is programmed with C++ and may run on the Windows 98 or Windows NT4.0. the simulation system introduces a new reliability model considering both inspection and repair (I&R) processes, by which both fatigue failure and fracture may be considered together, and improvement in reliability by I&R can be evaluated quantitatively. Several NDT probabilistic models and other models describing the effect on behavior of a structural member by I&R are also integrated in the software system. By means of a digital intelligent simulator integrated in the software system, the dynamic fatigue and fracture risk may be monitored in real-time, and in-service inspection and repair optimization for a crack may be easily realized. the system provides a good easy-to-use tool for offshore structures risk management and reliability improvement.
In this paper two ART based neuro-fuzzy systems, FasArt and its recurrent version RFasArt, are compared in order to test the performance of the recurrency in this type of architecture. Both models has been employed in...
详细信息
In this paper two ART based neuro-fuzzy systems, FasArt and its recurrent version RFasArt, are compared in order to test the performance of the recurrency in this type of architecture. Both models has been employed in several areas such as pattern recognition and modeling/identification of systems. In the first area, the document understanding was involved, here the document components should be classified according to the relationships amongst them. In the second area a waste water treatment plant and an AC electrical motor have been selected. Better successful rates, non ambiguities in the classification process and a reduction of the number of fuzzy rules or the complexity have been some of the main advantages founded.
暂无评论