Search

This paper provides a tutorial introduction and overview of design techniques for a system with both parametric uncertainty and unmodeled uncertainty, using mixed μ-synthesis. This paper also includes LFTs representation for modeling and an example of two-cart mass-spring-damper system (MSDs) is used to analyze its robust stability and performance, based on mixed μ-synthesis.

In view of the traditional genetic algorithm easily fall into local optimum in the late iterations, this paper puts forward an improved chaos genetic algorithm coded orthogonal signal design method which combines the chaos theory and genetic algorithm for MIMO radar. In order to prevent and overcome the ‘premature’ phenomenon in the process of optimization, the traversal features of the chaos optimization is introduced to the genetic algorithm, which reduces the autocorrelation peak side lobe and cross-correlation peak. Simulation results show that the proposed algorithm is feasible and effective.

Although the extreme learning machine(ELM)algorithm is applied in training single hidden layer feedforward neural networks without manual intervention and its fast training speed has been recognized nowadays, there are still several problems need to be solved in this algorithm. In this paper, we propose a new improved ELM algorithm–Error Minimized and Ridge Regression ELM(ER-ELM) derived from how to select the number of hidden nodes and the way to handle with the multicollinearity. We compare the ER-ELM algorithm with other relative algorithms such as ELM et al. Simulation results show that the ER-ELM algorithm has better stability and generalization performance than the other algorithms.

For the coupling characteristics about strip shape and gauge integrated system in hot strip mill, the feedforward compensation control method based on invariance principle was proposed, so the coupling problem of control system was effectively solved. The astringency mechanism and the stability mechanism of single neuron simple adaptive control were thorough analysis in this paper, and its application in strip shape and gauge integrated system of hot strip mill was realized. The simulation experiments achieve desirable control and thus prove its validity and feasibility.

In this paper, active control of repetitive impulsive noise is studied. A novel model-free iterative learning control (MFILC) algorithm based on FFT is used for an active noise control (ANC) system with an unknown or time-varying secondary path. Unlike the model-based method, the controller design only depends on the measured input and output data without any prior knowledge of the plant model. Computer simulations have been carried out to validate the effectiveness of the presented algorithm. Simulation results show that the proposed scheme can significantly reduce the impulsive noise and is more robust to secondary path changes.

Many adaptive schemes for controlling the probabilities of crossover and mutation in genetic algorithms with fuzzy logic have been reported in recent years. However, there has not been known work on comparative studies of these algorithms. In this paper, several fuzzy genetic algorithms are briefly summarized first, and they are studied in comparison with each other under the same simulation conditions. The simulation results are analyzed in terms of search speed and search quality.

In some artificial systems, performing the realistic perception for actors, which will handle both the processes to simulate the sensing organs and identify, will spend most computational time. Unfortunately, this matter even ruins the result of decision based on perception. In order to reduce the computation cost from a systemic view and optimize the performance of system, a brand-new coherent framework was proposed. The framework, which is a cycle of internal-state activizing desire, desire guiding perception, decision-making based on perception, and the executing result feed backing to actor, mimics the principle that the behavior of the actor is excitated both by the internal motivation and external stimuli. Firstly, we modeled internal state of artificial actors, which caused by temporary stimuli and accumulation of physical and mental states. Secondly, the actors' desires were described which are generated by internal state and used for guiding perception. Thirdly, the net value of the possible features combinations for a given desire was figured out using decision-theoretic principles to determine whether the process on the features combinations is worthy or not. Then, a multi-objective decision making algorithm was introduced to achieve decision based on perception, thus the connection between the actual desires and behaviors are established. Finally, we validate the approach and algorithms in establishing an autonomous computer animation system by simulating the realistic behavior of fish in bio-system.

We discuss Pulsed Para-Neural Networks (PPNN) defined as graphs consisting of processing nodes and directed edges, called axons, where axons are pure delays, whereas nodes are paraneurons processing spiketrains of constant amplitude. Pusles are received and produced in discrete moments of time. We consider PPNNs with two kinds of paraneurons: MEXOR-based paraneurons (producing a pulse at clock t if it received one and only one pulse at clock t-1) and counter-based paraneuron. We present a “brain” for a mobile robot as an example of practical PPNN.

The task-oriented proactive seamless migration is one of difficult problems to be solved in pervasive computing paradigm. Apparently, this function of seamless mobility is suitable for mobile services, such as mobile Web-based learning. But when seamless migration for computing task of learning is realized among PC, laptop, or PDA, there are several difficult problems to be solved, such as how to supply the proactive/attentive service with uncertainty for aware context. In order to realize E-learning based on proactive seamless migration, we design and improve relative fuzzy-neural approach (of course, besides it, there are other approaches). Generally, the network can be classified into two. One is that fuzzy logic reasoning is completed by fuzzy weight in neural system. The other is that the input data must be fuzzified in the first or second level, but not weight. We discuss and study the second in this paper. For proactive decision, fusion method based on fuzzy-neural can make Web-based learning system keep advantage of fuzzy logic system and remain adaptive optimum in proactive/attentive service. The correctness and validity of our new approach have been tested.