The general control problem can be stated as follows: Given a system with unknown dynamics, in order to construct a suitable controller for the system a model is often required. A model is any device that can imitate the behavior of the system. The process of constructing a model when only the relationship between the inputs of the system and the outputs from the system are available is known as identification. The model itself is called an identifier  . Once a model is available, an inverse model can be constructed to serve as a controller of the real system or plant. This type of problem accrues in many control settings including robotic systems, drive motors for various systems, automatic weld control, truck backer upper and so on. Examples of two such systems are described below. The nation of using an adaptive ANN as a model is shown in Fig. 1, where the weights of the ANN, which is receiving the same input signal as the unknown system, are adaptively modified until its output closely matches the output of the system.
such method is not suitable for real time control. Thus, this paper aims to investigate the possible use of a neuralnetwork approach for lateral controlof such railway wheel sets. The suitability of application to this problem is motivated by promising theoretical studies of higher-order neural units (HONUs), especially the quadratic neural unit for engineering problems for -. These studies are focused on the use of supervised learning based approaches for polynomial structured neural units, also known as a class of HONUs, for adaptive identification and controlof real engineering systems. Further motivation arises from the successful implementationof a quadratic neural unit controller (Neuro-controller), for controlof a bathyscaphe system located in the automatic control laboratories of CTU . Where, here, such controller adhered more closely to the desired behaviour of the system than the conventionally used PID controller. An extension on this result can be recalled in the work . Where, further study was made via introduction of a new software for adaptive identification and control, along with further testing on both a theoretical system and the previously mentioned bathyscaphe system. Given this, in this paper we aim to investigate use ofneuralnetwork (NN) approaches in the following manner. We will begin by explaining in more depth the problem behind the recently employed state feedback and cascade PID controlof the linearization of the CTU roller rig. The proceeding section will then describe the principles and control schemes behind the various experimented methods, for adaptive identification and control. Following this, an experimental analysis of the various approaches of NN, focusing firstly on adaptive identification of the CTU roller rig system and then to test the various NN methods for control. The final component of this paper will then be, to analyse and compare the produced experimental results, with a conclusion to be drawn at the end.
The neuralnetwork controller is created directly based on the neuralnetwork identifier. Its design is fully incorporated the learning strategy into the trained identifier. The weights of the neuralnetwork identifier are constantly verified against the actual plant output. This ensures that the weights allow the neuralnetwork identifier to properly predict actual plant output. The neuralnetwork identifier is used as means to back propagate the performance error to get the equivalent error at the output of the neuralnetwork controller. The accuracy of the plant model is critical in ensuring that the controller is accurate as well. The error between the plant output and the identifier output is also checked for the accuracy level of the identifier. This error is used to back propagate adjust the weights of the identifier to provide the most accurate representation of the plant. The neuralnetwork for controller is also designed as a three-layer neuralnetwork. It has an input layer, a hidden layer and an output layer.
The coolant has also an important effect as a lubricant. Direct lubrication with grinding fluids becomes mainly important in the creep-feed grinding (Malkin, 1989). Some analysis of heat transference indicate that the use of a faster work speed, keeping the same removal rate would lower the temperature and reduce the thermal damage, but this is not always true in practice. The fundamental difficulty of controlling damages caused in the grinding process is the lack of a reliable method in supplying feedback in real time during the process. Acoustic emission and electric power signals together combined have successfully been used in determining indicative parameters when burn takes place. Once these signals are processed and combined properly, they allow the on-line implementationof a burn monitoring system, optimizing the grinding process. This would be of great benefit for the dependent companies on this process, once the quality demand and international competitiveness grow more and more with the world globalization (Aguiar, 1997).
Abstract: The speed of ultrasonic motor of piezo-electric type is usually measured using mechanical sensors such as pulse encoders. However, these sensors are costly and bulky. In this paper, a numerical speed estimation approach of a piezo-electric motor (PEM) is implemented using multi- layer perception neuralnetwork (MLP-NN). The proposed model evaluates rotational speed and load torque based on the amplitude and driving frequency of the terminal voltage, considering the temperature variation. The estimated speed is employed to enhance the performance of the adaptive- fuzzy based speed control system. The model is validated and examined to achieve a minimized relative error in speed estimation approaches.
Over the last two decades, ANN has gained a lot of appreciation from many engineering fields. The electric power industry is no exception. With the advent of powerful and cheap computers, digital control is being used in a growing number of power system applications. Many new effective algorithms have been developed and implemented in real time. Artificial NeuralNetwork controller (ANNC) is a new control approach with a great potential for real-time applications.
Several different methods have been proposed for tuning PID regulators, starting from the work of Ziegler-Nichols . For instance, Åstrom and Hägglund  proposed the use of relay feedback, in conjunction with different design methods , to automate the tuning process of PID regulators. This is a robust technique; however some reservations on its use have been expressed by commissioning engineers, mainly because the method is so different from the standard procedures used for manual tuning. In practice, the commissioning engineer applies an input step to either the plant or the closed-loop system, and, based on the system response and on his experience on the plant, iteratively tunes the PID controller. In this respect, there are similarities to the method proposed in this thesis. The main difference, however, lies in the replacement of the human operator by one type of artificial neuralnetwork, multilayer perceptrons . These gain their experience in tuning by being trained off-line. The skill level of the engineer corresponds to the criterion used to derive the PID parameters used in the training phase. By using criteria that produce good responses, the neural PID tuner mimics an experienced plant operator, with the advantage that several iterations are not needed for tuning.
Multilevel inverter topology has developed recently as a very important alternative in the area of high power medium voltage energy control. In multilevel inverter, thethree basic types of topologies used are diode clamped inverter (neutral point clamped), capacitorclamped (flying capacitor) and cascaded multi cell with separate dc sources. Multilevel inverters are used in medium voltage and high power applications with less harmonic contents. This paper proposes a software implementationof neutral point clamped (NPC) three level voltage source inverter using space vector pulse width modulation (SVPWM) techniques. The inverter feeds an electrical system which is controlled by field oriented control (FOC).The improvement of the control technique is achieved using intelligence techniques. The operation of the electrical system is verified in steady state and transient state responses. This software implementation is performed by using matlab/Simulink software. This paper gives comparison between SVPWM three phase three level with neutral point clamped and without neutral point clamped. Finally, the comparative study of different techniques was implemented.
modules to generate all the power they are capable of producing . MPPT is an electronic tracking system, whether digital or analog. There are many approaches to find and track the maximum power point for PV cells. For instance, neuralnetwork, open circuit voltage, short circuit current, fuzzy logic control, perturb and observe, and incremental conductance are the most popular methods to track the maximum power point tracking. As a matter of fact, many systems have combined methods, for example, using the open circuit voltage (OV) to find the starting point for the iterative methods such as Perturbation and Observation or Incremental. The levels of irradiance play an important part in changing from one scheme to another. For example, at low levels of irradiance, methods like open circuit voltage and short circuit current could be more suitable as they can be more noise immune. When the cells are connected in a series, the iterative methods can be a preferable solution. When a portion of the string is partial shade, search algorithms are needed .In general, the accurate method is better than the fast because fast methods tend to bounce around the maximum power point due to noise present in the power conversion system. Of course, an accurate and fast method would be preferred, but the cost ofimplementation needs to be considered.
When using WT an important decision has to be made: the selection of a suitable mother wavelet, in the particular STLF case, the choice of Daubechies family wavelets is almost consensual, with special emphasis on Daubechies of order 2 and 4 (db2 and db4) , , , . However, larger temporal data sets, specifically a load time series, can present considerable variability over time. Thus, the best mother (base) wavelet for a certain time interval may not be the most appropriate for others, for example in  the authors found that different orders Daubechies were best suited for each season of the year. The same reason led the authors in  to implement a wavelet-based ensemble scheme composed by eight mother wavelets, sectioning the load data series and finding the mother wavelet that best fits each section. Yet when selecting the eight best candidates, the authors also relied on the consensus about Daubechies in selecting half the base wavelets, while the remaining four where picked from the Coiflets family, based on experimental results.
Vibration analyses of structural systems have been performed with the aid of different methods [6-15]. However, the complex shaped structures may be analyzed with soft computing techniques more easily. Soft Computing is a general term for a collection of computing techniques . These well- known techniques constitute artificial neural networks (ANN), fuzzy logic, evolutionary computation, machine learning and probabilistic reasoning. Soft computing methods differ from classical computing methods in that, unlike classical computing methods it is tolerant of imprecision, uncertainty, partial truth to achieve tractability, approximation, robustness, lows solution cost and better rapport with reality .
RESUMO – Contexto – A deglutição é um processo motor com muitas discordâncias e de difícil estudo quanto a sua neurofisiologia. Talvez por essa razão sejam tão raros os artigos sobre esse tema. Objetivo – Descrever o controle neural da mastigação e a qualificação do bolo que se obtém durante a fase oral. Revisar os nervos cranianos envolvidos com a deglutição e suas relações com o tronco cerebral, cerebelo, núcleos de base e córtex. Métodos – Re- visão da literatura com inclusão de trabalhos pessoais e novas observações buscando dar consistência a necessária revisão dos conceitos, muitas vezes conflitantes. Resultados e Conclusão – Em relação a fase oral da deglutição consideramos o controle neural em cinco distintas possibilidades. Fase oral nutricional voluntária, fase oral cortical voluntária primaria, fase oral semiautomática, fase oral em goles subsequentes e fase oral espontânea. Em relação ao controle neural da fase faríngea da deglutição, pode-se observar que o estímulo que dispara a fase faríngea não é o toque produzido pela passagem do bolo, mas sim a distensão pressórica, tenha ou não conteúdo em passagem. Na deglutição nutricional, alimento e pressão são transferidos, mas na fase oral da deglutição primária cortical somente pressão é transferida e temos resposta faríngea similar a nutricional. A fase faríngea incorpora como parte de sua dinâmica as atividades orais já em curso. A fase faríngea se inicia por ação do plexo faríngeo composto pelos nervos glossofaríngeo (IX), vago (X), e acessório (XI), com envolvimento do trigêmeo (V), do facial (VII), glossofaríngeo (IX) e hipoglosso (XII). O plexo cervical (C1, C2), e o nervo hipoglosso, a cada lado, formam a alça cervical de onde, com origem cervical, um ramo segue para o músculo gênio-hioide, um músculo que atua na dinâmica de elevação do complexo hiolaríngeo. Foi também considerado o controle neural da fase esofágica da deglutição. Além de outras hipóteses foi considerado que é possível que a camadas musculares consideradas como longitudinal e circular para o esôfago sejam a longitudinal composta por fibras espirais de passo longo e a circular por fibras espirais de passo curto. Essa morfologia associada ao conceito de preservação de energia, nos permite admitir que a contração da camada longitudinal por seu arranjo espiral seja capaz de alargar o esôfago diminuindo sua resistência ao fluxo e provavelmente e também abrindo a transição esofagogástrica. Desse modo a camada circular, espiral de passo curto, pode propelir o bolo por constrição sequencial de cranial para caudal.
In the majority of proposed adaptive controller a complementary term, called supervisory term, is added to the output of the fuzzy inference system or of the neuralnetwork system controller in order to guarantee global stability using Lyapunov theory . When the system is operating within a prescribed range, the supervisory controller is turned off. The difference between the various methods lies in the structure of the model and in the form of the supervisory term, the approximated control law is based on the certainty equivalence approach while the supervisory term can take different forms, but most often, it is based on the sliding mode technique. Sanner and Slotine first proposed Radial Basis Function (RBF) NNs for controlof affine systems and provided rigorous analysis of stability. They introduced a sliding mode component in the control law for keeping the evolving dynamics within the predetermined compact set of interest .
We showed through analytical derivations and numerical simulation that the optimal threshold for selecting overt actions is a declining function of time. Such a collapsing decision bound has previously been obtained for decision making under a deadline [11,29]. It has also been proposed as an ad-hoc mechanism in drift diffusion models [28,30,43] for explaining finite response time at zero percent coherence. Our results demonstrate that a collapsing bound emerges naturally as a consequence of reward maximiza- tion. Additionally, the POMDP model readily generalizes to the case of decision making with arbitrary numbers of states and actions, as well as time-varying state.
III. SELECTION OF PROCESSOR The selection of right device on which to base our design was a challenging task; typically in the scenario where there is a need to have a right balance of price, performance and power consumption . The x86(intel) and PowerPC processors are proven architecture which are currently being used. The ARM offers goodness of 32 bit processing and peripherals on a single chip (SoC) which makes the hardware design extremely simple. ARM eliminates the monopoly of a single silicon vendor as it is an IP core company with a huge silicon vendor base. Hence ARM was an obvious choice for this architecture.
Abstract: Problem statement: Voltage disturbances are the most common power quality problem due to the increased use of a large numbers of sophisticated electronic equipment in industrial distribution system. The voltage disturbances such as voltage sags, swells, harmonics, unbalance and flickers. High quality in the power supply is needed, since failures due to such disturbances usually have a high impact on production cost. There are many different solutions to compensate voltage disturbances but the use of a DVR is considered to be the most cost effective method. The objective of this study is to propose a new topology of a DVR in order to mitigate voltage swells using a powerful power custom device namely the Dynamic Voltage Restorer (DVR). Approach: New configuration of a DVR with an improvement of a controller based on direct-quadrature-zero method has been introduced to compensate voltage swells in the network. Results: The effectiveness of the DVR with its controller were verify using Matlab/Simulink’s SimPower Toolbox and then implemented using 5KVA DVR experimental setup. Simulations and experimental results demonstrate the effective dynamic performance of the proposed configuration. Conclusion: The implimentation of the proposed DVR validate the capabilities in mitigating ofvoltage swells effectiveness.During voltage swells, the DVR injects an appropriate voltage to maintain the load voltage at its nominal value.
Generally transistors are used as the switches of the DC-DC converters. In traditional PWM converters, power transistors such as MOSFET, IGBT are adopted as the switching devices. MOSFET is preferable for low voltage and high current applications where as IGBT is preferable for high voltageapplications. The duty cycle of the power switch is controlled to achieve buck/boost topology. During the turn on instant current increases linearly results in switching losses. Although the traditional PWM converter is efficient than conventionally adopted linear power converter, it has some drawbacks. The operating frequency of conventional PWM converter must be raised to high value to minimize the volume and weight of the converter. But increase in switching frequency leads to increases the switching losses and it also increases the severity of switching stress and EMI. So there is a frequency limitation on traditional PWM converter (hard switching converter).
An Elman Recurrent NeuralNetwork is a network which in principle is set up as a regular feed forward network. This means that all neurons in one layer are connected with all neurons in the next layer. An exception is the so-called context layer which is a special case of a hidden layer. The neurons in the context layer (context neurons) hold a copy of the output of the hidden neurons. The output of each hidden neuron is copied into a specific neuron in the context layer. The value of the context neuron is used as an extra input signal for all the neurons in the hidden layer one time step later. Therefore, the Elman network has an explicit memory of one time lag (Elman, 1990). Similar to a regular feed forward neuralnetwork, the strength of all connections between neurons are indicated with a weight. Initially, all weight values are chosen randomly and are optimized during the stage of training.
The new achievements in mobile device technologies opened the way for new applications designed to run on mobile devices. In the beginning, mobile devices offered very limited functionality due to small memory, computing power and difficult interaction. Nowadays, mobile devices become more and more popular, available memory grew considerably, being comparable with some desktop computers, mobile processors have improved performance, interaction is becoming more user friendly. These characteristics allow the development of complex applications that make use of available hardware capabilities. Business Intelligence applications help managers to make decisions based on quantitative methods applied to available business data. Mobile business intelligence applications extend such functionality on devices used by the decision makers . Such applications take several forms:
We select the simulation software TrueTime 1.5 . The WNCS is composed by the wireless network, actuator/sensor, controller, interference nodes and the controlled plants. Wireless network is the IEEE 802.11b/g (WLAN), the data rate is 800,000 bits/s, minimum frame size is 272 bits, transmit power is 20 dbm, receiver signal threshold is − 48.00 dbm, path loss exponent is 3.5, act timeouts is 0.00004 s, retry limit is one, error coding threshold is 0.03, the distance between nodes is 20.0 m, maximum signal reach is 86.67 m, and sampling period of the sensor is 0.01s. Reference signal r adopts square wave and its amplitude is from −1 to 1.There are some data dropouts in the closed loop, and the network delays are allowed to be random, time-variant and uncertain, possibly large compared to one, even tens sampling periods.