The differential evolution (DE) and harmony search (HS) are two well-known nature-inspired computing techniques. Both of them can be applied to effectively cope with nonlinear optimization problems. In this paper, we propose and study a new DE method, DE–HS, by utilizing the fresh individual generation mechanism of the HS. The HS-based approach can enhance the local search capability of the original DE. Optimization of some unconstrained and constrained benchmark problems and a real-world wind generator demonstrate that our DE–HS has an improved convergence property.

This paper describes three implementation alternatives of a multiplicative general parameter finite impulse response filter (MGP-FIR) used as a precise current reference generator for shunt-type active power filters. Compared to the traditional techniques that use fixed digital or analogue filters, the MGP-FIR is an adaptive and predictive algorithm for tracking typical frequency variations and thus avoiding the harmful phase shift of the fundamental sinusoid. This characteristic is essential for maximizing the harmonics reduction ability of the active power filter. The fixed subfilters of the MGP-FIR are pre-optimized by the use of a genetic algorithm. The harmonics reduction ability was verified by extensive simulations. Moreover, a laboratory prototype was developed and experimentally tested under typical non-linear loads. Compared to the previously presented articles on MGP-FIR-based current reference generators, this is the first time when real-valued or fixed-point coefficients are used with the MGP-FIR. In addition, we also improved the originally proposed amplitude detector. The results are compared with those presented earlier.

The harmony search (HS) method is an emerging meta-heuristic optimization algorithm. In this paper, we propose two modified HS methods to deal with the uni-modal and multi-modal optimization problems. The first modified HS method is based on the fusion of the HS and differential evolution (DE) technique, namely, HS-DE. The DE is employed here to optimize the members of the HS memory. The second modified HS method utilizes a novel HS memory management approach, and it targets at handling the multi-modal problems. Several nonlinear functions are used to demonstrate and verify the effectiveness of our two new HS methods.

In this paper, based on the fusion of the clonal selection algorithm (CSA) and differential evolution (DE) method, we propose a novel optimization scheme: CSA-DE. The DE is employed here to increase the affinities of the clones of the antibodies (Abs) in the CSA. Several nonlinear functions are used to verify and demonstrate the effectiveness of this hybrid optimization approach. It is further applied for the construction of the cascade-correlation (C-C) neural network, in which the optimal hidden nodes can be obtained.

This paper presents a hybrid optimization method based on the fusion of the clonal selection algorithm (CSA) and harmony search (HS) technique. The CSA is employed to improve the members of the harmony memory in the HS method. The hybrid optimization algorithm is further used to optimize a fuzzy classification system for the Fisher Iris data classification. Computer simulations results demonstrate the effectiveness of our new approach.

This paper proposes a particle swarm optimization (PSO)-based detector optimization scheme in the negative selection algorithm (NSA). The NSA is a natural immune response inspired pattern discrimination method. In the new scheme, the NSA detectors are optimized by the PSO to collectively occupy the maximal coverage of the nonself space so that they can achieve the best anomaly detection performance. Two numerical examples including the discriminant analysis of Fisher's iris data are demonstrated to verify the effectiveness of our approach.

The performance of evolutionary algorithms in optimization is tightly coupled to the computational effort required by the evaluation of the objective function. If the objective function is too expensive to evaluate, then, the elaboration of the procedures of the search algorithm alone may not result in the required improvement in algorithm’s performance. However, if there is a way to speed up or decrease the number of objective function evaluations, even a basic algorithms can potentially achieve better results due to the increased number of generation run in given time. This paper considers a probabilistic objective function evaluation scheme in which the candidate solutions are evaluated and evolved based on their objective function value.

This paper proposes a hybrid optimization method based on the ant colony and clonal selection algorithms, in which the cloning and mutation operations are embedded in the ant colony to enhance its search capability. The novel algorithm is employed to deal with a few benchmark optimization problems under both static and dynamic environments. Simulation results demonstrate the remarkable advantages of our approach in diverse optimal solutions, closely tracking varying optimum, as well as improved convergence speed.

In this paper, we employ the clonal optimization method to optimize the detectors in the negative selection algorithm (NSA). Taking advantage of the clonal optimization strategy, the NSA detectors can be optimized for anomaly detection. A new motor fault detection scheme using our NSA is also discussed. We demonstrate the efficiency of the proposed approach with an example of bearings fault detection.

This paper proposes a hybrid particle swarm optimization (PSO) method, which is based on the fusion of the PSO, clonal selection algorithm (CSA), and mind evolutionary computation (MEC). The clone function borrowed from the CSA and MEC-characterized similartaxis and dissimilation operations are embedded in the original PSO. Simulations of nonlinear function optimization are made to compare this hybrid PSO with the regular PSO. It has been demonstrated that our hybrid algorithm can achieve a better convergence performance, and provide diverse solutions to multi-model optimization problems.

In this paper, we first propose a linguistic information feedforward-based dynamical fuzzy system (LIFFDFS) in which the past fuzzy inference output represented by a membership function is fed forward locally with trainable parameters. Our LIFFDFS can overcome the common static mapping drawback of conventional fuzzy systems. We also give a detailed description of its underlying principle and general structure. Next, based on the gradient descent method, an adaptive learning algorithm for the feedforward parameters is derived. The proposed LIFFDFS is further employed in the prediction of time series. The well-known Box-Jenkins gas furnace data are used here as an evaluation example. Simulation results demonstrate that this new dynamical fuzzy system has the advantage of inherent dynamics and is, therefore, well suited for handling temporal problems, such as process modeling and control.

Soft computing (SC) and hard computing (HC) methodologies are fused together successfully in numerous industrial applications. The principal aim is to develop computationally intelligent hybrid systems that are straightforward to analyze, with highly predictable behavior and stability, and with computational burden that is no more than moderate. All these goals are particularly important in embedded real-time applications. This paper is intended to clarify the present vagueness related to the fusion of SC and HC methodologies. We classify the different fusion schemes to 12 core categories and six supplementary categories, and discuss the characteristic features of SC and HC constituents in practical fusion implementations. The emerging fusion approach offers a natural evolution path from pure hard computing toward dominating soft computing.

The design of control systems for large-scale and complex industrial plants involves numerous trade-off problems, such as costs, quality, environmental impact, safety, reliability, accuracy, and robustness. Some of these parameters are even conflicting. Thus, the use of a multidiscipline approach is suggested to satisfy these requirements in an acceptable and well-balanced manner, and a fusion of soft computing and hard computing appears to be a natural and practical choice. Although the state-of-the-art soft computing technology has distinguished features, the use of soft computing technology would be ineffective, if it is improperly fused with conventional hard computing technology and control processes. Proper fusion is key to success, and a general model of fusion is worth examining. In this paper, through a survey of published literature, a general fusion model and fusion topologies are shown at the system level as well as at the algorithm level.

We introduce a genetic algorithm (GA)-based method for structural optimization of multiplicative general parameter (MGP) finite impulse response (FIR) filters. These computationally efficient reduced-rank adaptive filters are robust, suitable for predictive configurations, and they have numerous applications in 50/60 Hz power systems instrumentation. The design process of such filters has three independent stages: Lagrange multipliers-based optimization of the sinusoid-predictive basis filter, genetic algorithm-based search of optimal FIR tap cross-connections and, finally, the online MGP-adaptation phase guided by variations in signal statistics. Thus, our multi-stage design procedure is a complementary fusion of hard computing (HC) and soft computing (SC) methodologies. Such advantageous fusion (or symbiosis) thinking is emerging among researchers and practicing engineers, and it can potentially lead to competitive combinations of individual HC and SC methods.

Soft computing (SC) and hard computing (HC) methodologies are fused together in numerous industrial and other applications. The aim is typically to develop intelligent systems that would be straightforward to analyze, their behavior and stability could be predictable, and the computational burden would be no more than moderate. All these goals are particularly important in embedded real-time applications. This paper clarifies the vagueness related to the fusion of SC and HC methodologies that slows down the acceptance of soft computing outside the computational intelligence community. We categorize the different fusion topologies to 12 core classes and 6 supplementary classes, and discuss the characteristic features of the SC and HC constituents in practical hybrid implementations.

We introduce an evolutionary programming based method for designing robust and computationally efficient adaptive bandpass filters. These predictive filters are optimized for generating current references in active power filters (APFs). The accuracy (phase/amplitude) of the reference current is crucial in current injection type systems, because it directly affects the harmonics reduction ability of the APF. Our digital filtering approach has the following advantages: selective bandpass response, efficient attenuation of specific harmonic components, capability to handle typical frequency alteration, small number of multiplications, and structural simplicity. In an illustrative example, the total harmonic distortion of an artificial current waveform was reduced from 36.7% to less than 3.7% within the line frequency range 49-51 Hz. The proposed scheme is a combination of the hard computing (HC) type multiplicative general parameter method and evolutionary programming that is a constituent of soft computing (SC). Such fusion thinking is emerging among researchers and engineers, and it can potentially lead to efficient combinations of HC and SC methodologies - both on the algorithm level and on the system level.

We explore the feasibility of using both feedforward and Elman neural networks to detect assembly faults in ink jet printers. The method is an extension of the motor fault detection scheme proposed by Gao and Ovaska (2002). Two types of cartridge faults are studied here: encoder belt misalignment and encoder strip error. These two faults are detected from the characteristics variants in the neural networks-based prediction of cartridge velocity signals. Results of experiments with real-world data demonstrate that neural networks can be trained to effectively detect the inherent encoder faults. Some discussions on the selection of appropriate fault detection criteria are also given.

Prediction of sinusoidal signals with time. varying frequencies has been an important research topic in power electronics systems. To solve this problem, we propose a new fuzzy predictive filtering scheme, which is based on a Finite Impulse Response (FIR) filter bank. Fuzzy logic is introduced here to provide appropriate interpolation of individual filter outputs. Therefore, instead of regular 'hard' switching, our method has the advantageous 'soft' switching among different filters. Simulation comparisons between the fuzzy predictive filtering and conventional filter bank-based approach are made to demonstrate that the new scheme can achieve an enhanced prediction performance for slowly changing sinusoidal input signals.

Soft computing (SC) is an emerging collection of methodologies which aims to exploit tolerance for imprecision, uncertainty, and partial truth to achieve robustness, tractability, and low total cost. It differs from conventional hard computing (HC) in the sense that, unlike hard computing, it is strongly based on intuition or subjectivity. Therefore, soft computing provides an attractive opportunity to represent the ambiguity in human thinking with real life uncertainty. Fuzzy logic (FL), neural networks (NN), and genetic algorithms (GA) are the core methodologies of soft computing. However, FL, NN, and GA should not be viewed as competing with each other, but synergistic and complementary instead. Considering the number of available journal and conference papers on various combinations of these three methods, it is easy to conclude that the fusion of individual soft computing methodologies has already been advantageous in numerous applications. On the other hand, hard computing solutions are usually more straightforward to analyze; their behavior and stability are more predictable; and, the computational burden of algorithms is typically either low or moderate. These characteristics. are particularly important in real-time applications. Thus, it is natural to see SC and HC as potentially complementary methodologies. Novel combinations of different methods are needed when developing high-performance, cost-effective, and safe products for the demanding global market. We present an overview of applications in which the fusion of soft computing and hard computing has provided innovative solutions for challenging real-world problems. A carefully selected list of references is considered with evaluative discussions and conclusions.

A class of computationally efficient adaptive algorithms for transversal filters is discussed. The algorithms, which are based on the so-called general parameter method, use typically one or a few dynamically adjusted parameters, each to be added to a block of coefficients of a fixed basis FIR filter. Thus the overall filter is adapted so that the output error is minimized. The adaptive extension can be constructed as an 'add-on' element to be used in parallel with fixed-coefficient filters. An efficient implementation structure is proposed, and the stability and convergence properties of the multiple-parameter algorithm are analyzed.

This paper compares the performance of nonlinear radial basis function network-based (RBFN) and linear autoregressive (AR) model-based general parameter (GP) methods in a fault detection application. We use the efficient GP approach for initializing the weights of the RBFN model in the beginning of the off-line system identification phase, as well as for fine-tuning the modeling accuracy of RBFN and AR models online. Our fault detection scheme is based on monitoring the expectation value of the scalar general parameter. This provides improved robustness and detection sensitivity over such methods where the online prediction error is used directly in the decision-making process. In order to illustrate the performance of the efficient nonlinear and linear schemes, they are applied to fault detection of automobile transmission gears. As the acoustic sound level time-series, providing the necessary basis information for fault detection, is slightly nonlinear, the GP-RBFN outperformed the linear methods: the GP-AR method and conventional AR inverse filtering. Both of the GP-based methods provide competitive alternatives for real-world fault detection and diagnosis.

Soft computing (SC) is an evolving collection of methodologies, which aims to exploit tolerance for imprecision, uncertainty, and partial truth to achieve robustness, tractability, and low total cost. It differs from conventional hard computing in that it is strongly based on intuition or subjectivity. Therefore, soft computing provides an attractive opportunity to represent the ambiguity in human thinking with real life uncertainty. Fuzzy logic (FL), neural networks (NN), and evolutionary computation (EC) are the core methodologies of soft computing. However, FL, NN, and EC should not be viewed as competing with each other, but synergistic and complementary instead. In this paper, we discuss the role of soft computing in developing innovative features for home appliances. Such development work is particularly active in Japan and South Korea. In those Asian countries even ordinary consumers do appreciate the capabilities of SC in developing user-friendly products with high machine IQ. Although home appliance technology is not a popular academic research area, it is an important application area for soft computing methods.

In this paper, we propose a fuzzy logic-based power command enhancement scheme for the mobile station in the cellular communications systems. By defining necessary linguistic terms of the power commands and corresponding fuzzy inference rules, an Embedded Fuzzy Unit (EFU) is constructed. The role of our EFU is to produce appropriate digital power commands instead of the one-bit discrete commands directly obtained from the base station. Simulations show that the EFU-based mobile station can generate considerably smoother received power at the base station. The transient overshoot and steady channel tracking error are small. Moreover, our method has the advantage of simplicity in both structure and algorithm, and is thus easy for hardware implementation.

Fuzzy logic, neural networks, and evolutionary computation are the core methodologies of soft computing (SC). SC is causing a paradigm shift in engineering and science fields since it can solve problems that have not been able to be solved by traditional analytic methods. In addition, SC yields rich knowledge representation, flexible knowledge acquisition, and flexible knowledge processing, which enable intelligent systems to be constructed at low cost. This paper reviews applications of SC in several industrial fields to show the various innovations by TR, HMIQ, and low cost in industries that have been made possible by the use of SC. Our paper intends to remove the gap between theory and practice and attempts to learn how to apply soft computing practically to industrial systems from examples/analogy, reviewing many application papers.

By taking advantage of fuzzy systems and neural networks, a fuzzy-neural network with a general parameter (GP) learning algorithm and heuristic model structure determination is proposed in this paper. Our network model is based on the Gaussian radial basis function network (RBFN). We use the flexible GP approach both for initializing the off-line training algorithm and fine-tuning the nonlinear model efficiently in online operation. A modification of the robust unbiasedness criterion using distorter (UCD) is utilized for selecting the structural parameters of this adaptive model. The UCD approach provides the desired modeling accuracy and avoids the risk of over-fitting. In order to illustrate the operation of the proposed modeling scheme, it is experimentally applied to a fault detection application.

Prospective authors are requested to submit new, unpublished manuscripts for inclusion in the upcoming event described in this call for papers.

During recent years, fuzzy neural networks have found extensive applications in numerous engineering areas. It is known that the fusion of neural networks and fuzzy logic can overcome their individual drawbacks and benefit from each other's merits. However, current fuzzy neural networks often have complex structures and training algorithms. In addition, some of them cannot deal with fuzzy knowledge directly. Inspired by the alpha-level cut representation of fuzzy numbers, we propose a simple neural network-based approach for processing fuzzy information. By numerical simulations, our scheme is illustrated to be capable of coping with fuzzy input and output without a need for new network topology or learning algorithm

Fault detection methods are crucial in acquiring safe and reliable operation in motor drive systems. Remarkable maintenance costs can also be saved by applying advanced detection techniques to find potential failures. However, conventional motor fault detection approaches often have to work with explicit motor models. In addition, most of them are deterministic or non-adaptive, and therefore cannot be used in time-varying cases. We propose an Elman neural network-based motor fault detection scheme to overcome these difficulties. The Elman neural network has the unique time series prediction capability because of its memory nodes as well as local recurrent connections. Motor faults are detected from changes in the expectation of the feature signal prediction error. A genetic algorithm (GA)-aided training strategy for the Elman neural network is further introduced to improve the approximation accuracy and achieve better detection performance. Computer simulations of a practical automobile transmission gear with an artificial fault are carried out to verify the effectiveness of our method. Encouraging fault detection results have been obtained without any prior information of the gear model.

Recently, researches on group robot systems (multi-agent intelligent robots: MAIR) have been reported, where a number of robots behave in a group like birds or ants. It is generally known that each robot has a limited intellectual power, but the robots can behave more intellectually in a group because they can interact each other. It has been reported that the multi-agent robot systems which can do many kinds of tasks efficiently by training the rules between environments and actions using reinforcement learning. Intelligent agent systems are also applied to decision making, human interface, perception and others even in a robot system (intelligent function agent robot: IFAR). This paper first introduces extended soft computing (ESC) which is the fusion/combination of fuzzy, neuro, genetic and chaotic computings and immune network theory in order to explain, what they call, complex systems and cognitive and reactive AIs. Then, contemporary intelligent system concept is discussed while the ESC is promising to realize it. Finally, a decision making robot with multi-agents (antigens and antibodies of immune networks), fuzzy inference and reinforcement learning (genetic computing) is described, as an example. It is confirmed that the ESC plays an important role in constructing intelligent robots.

We give a comparison between the conventional power control scheme and a soft computing-based approach in a mobile communications application. At the base station, the 'bang-bang' control strategy and a neural network-based prediction control method are employed respectively. In addition, full power command, single-bit command transmission mode, and fuzzy logic-based power command enhancement unit, are considered. Based on numerical simulation experiments, we quantitatively evaluate the performance of various combinations of these control methods and command transmission modes. Finally, conclusions on the optimal configuration are given.

Dealing with impulsive noise remains one of the great engineering challenges due to the difficulties in modeling and filtering impulsive noise. Since impulsive noise is, in general, a non-stationary stochastic process, neural networks and fuzzy systems have been applied to deal with it. In many cases, however, the underlying primary signal is smoothly changing. Therefore, a more simple approach, namely polynomial predictive filtering, can successfully be applied instead of soft computing methods. In this paper, we present a filtering system which consists of a cascade of a median filter and a polynomial predictive filter. The proposed system is capable of removing the strong impulses that are present in the polynomial-like input signal without introducing a disturbing delay. The simulation results are compared with the corresponding results obtained by applying fuzzy filtering. Finally, we consider an interesting concept called smoothness of a signal; we will show that the choice of feasible filtering approach and its parameters is dependent on the degree of smoothness of the signal to be filtered.

Although neural networks have already found many successful applications in speech signal processing, it is difficult to compare their performance with that of the traditional methods objectively. We address this problem by using the predictive minimum description length (PMDL) principle to select the optimal network size in modeling speech signals.

We investigate the characteristics of some one-step-ahead nonlinear predictors based on a two-layer feed-forward neural network (2LFNN). The behavior of neural networks (NN) is investigated in the frequency domain using two frequency response estimation techniques, and in the time domain, by analyzing the unit step and triangular pulse responses. Some of the estimated frequency responses of these NNs resemble those of corresponding linear polynomial predictors, revealing the nearly polynomial nature of the applied training signals. The similarity of the two frequency response estimates is an indication of good generalization properties.

Presents a novel neural-network-based predictor for received power level prediction in direct sequence code division multiple access (DS/CDMA) systems. The predictor consists of an adaptive linear element (Adaline) followed by a multilayer perceptron (MLP). An important but difficult problem in designing such a cascade predictor is to determine the complexity of the networks. We solve this problem by using the predictive minimum description length (PMDL) principle to select the optimal numbers of input and hidden nodes. This approach results in a predictor with both good noise attenuation and excellent generalization capability. The optimized neural networks are used for predictive filtering of very noisy Rayleigh fading signals with 1.8 GHz carrier frequency. Our results show that the optimal neural predictor can provide smoothed in-phase and quadrature signals with signal-to-noise ratio (SNR) gains of about 12 and 7 dB at the urban mobile speeds of 5 and 50 km/h, respectively. The corresponding power signal SNR gains are about 11 and 5 dB. Therefore, the neural predictor is well suitable for power control applications where ldquodelaylessrdquo noise attenuation and efficient reduction of fast fading are required.

Delta operator realizations of IIR filters have received a lot of interest due to their good finite-wordlength properties. We present a detailed analysis of the computationally efficient transposed direct form II delta structure, which is also known to have excellent roundoff noise performance under fast sampling. The focus is on the roundoff noise minimization in fixed-point implementations. The free /spl Delta/ parameter in the delta operator is optimized with respect to the output roundoff noise within the scaling constraints. Moreover, the required internal wordlength to obtain a prespecified noise performance is derived.