M. La Cava

Brief paper: Parameter identification for induction motor simulation

The simulation of squirrel-cage induction motors is very useful to design frequency-controlled a.c. drives. For this purpose the parameters of the induction motor must assume suitable values. Parameter obtained by the open and short circuit tests give generally unsatisfactory results. In this paper it is therefore proposed to apply a suitable identification technique. By assuming the two axes equations as the model of the motor, the parameters are worked out by a curve fitting procedure of the steady state torque and current characteristics. Due to the nonlinearities in the model equations an iterative procedure is proposed which attains the absolute minimum. The parameter can be therefore worked out with a satisfactory accuracy as confirmed by the test presented in this paper.

Comparison of different growing radial basis functions algorithms for control systems applications

Supervised growing neural networks (SGNNs) are a class of self-organizing maps without a predefined structure. In fact the structure of the approximators is generated autonomously from a set of training data. New algorithms for SGNNs have been proposed previously with the objective to provide improved performance for on-line sequential learning. In this work two important class of algorithms for SGNNs are compared: resource allocating networks (RAN) and dynamic cell structures (DCS). The main objective is to provide a clear comparative study, which could help to assess the performance among the different algorithms for on-line real time application purposes. The main performance criteria are: the accuracy following the same amount of training-in terms of standard deviation and estimation error trends-and the computational complexity of the algorithm. The comparison has been performed through two different studies. The first study is relative to the learning of a nonlinear 3-D function. The second study is relative to the learning of a 3-D look-up table of a specific aerodynamic parameter of an aircraft.

Parameter identification for induction motor simulation

Summary The simulation of squirrel-cage induction motors is very useful to design frequency-controlled a.c. drives. For this purpose the parameters of the induction motor must assume suitable values. Due to the unsatisfactory results obtained by the open and short circuit tests, it is necessary to develop a new proce dure. The problem is seen in the general framework of theparameter identification techniques based On the knowledge of the torque-speed and stator current-speed characteristics of the induction motor in a suitable range of the slip. To obtain significant results it is necessary to specify the nonlinear constraints on the parameters. The problem of parameter identification is then formulated as a constrained nonlinear minimization problem. A particular computation procedure is developed to solve this problem. A test confirms the accuracy of the parameter values obtained by the newly developed identification method.

Feature Matching Algorithms for Machine Vision Based Autonomous Aerial Refueling

In this paper a machine vision (MV) based system is proposed as distance estimation sensor to be employed by UAVs during the operations of autonomous aerial refueling. For studying this problem it was developed a simulator featuring a 3D virtual reality (VR) interface that generates the image stream of the AAR maneuver. The proposed MV algorithm performs specific tasks as image processing for features extraction, feature matching and pose estimation. The problem of tanker/UAV attitude estimation from images is investigated in two scenarios: with and without artificial markers installed on the tanker. Two feature matching algorithms are proposed and the performance of the optical feedback signal are analyzed and compared in closed loop simulations

Optimal operation of the leavening process for a bread-making industrial plant

In this work a model-based method is proposed for determining the optimal set points for the leavening process in a continuous bread-making industrial plant. The procedure is based on a model of the leavening process, which has been identified using Neural Networks and ARX models. A Genetic Algorithm has been used to derive the optimal production set points, taking into account economic constraints. Two modes of operation of the plant have been investigated and optimized results have been confirmed by experiments on an industrial plant.

Improving trajectory tracking by feedforward evolutionary multivariable constrained optimization

Improvements in positioning accuracy and reduction of trajectory tracking error in robotic systems require advanced control laws; these should take into account also multivariable concurrent specifications and be able to handle inputs and state constraints. In this work these requirements are considered exploiting a feedforward model-based predictive controller in which the control law is planned online on the basis of a multiobjective cost function, the minimization of which is executed by means of an evolutionary algorithm. The proposed scheme has been applied to an existing robust feedback control scheme, to achieve a more accurate trajectory tracking of the tip of a flexible link. Real time execution is possible; moreover, notwithstanding the stochastic inference engine of the evolutionary algorithms, the proposed scheme is sufficiently reliable, since it reveals a high degree of repeatability of the control signals.

An Approach to Fault Diagnosis for Non Linear Dynamic Systems Using Neural Networks

Abstract This paper proposes an approach to fault detection and identification (FDI) for non linear dynamic systems using neural networks. A fault is considered as a variation of physical parameters; therefore the FDI problem can be reduced to the parameter identification of a non linear system and to the association of the set of the estimated parameters with the mode of operation of the plant. Under some assumptions, this can be efficiently accomplished by a combined use of a linear parameter estimator and a bank of neural classifiers. Each neural network is trained to perform the diagnosis in a certain working point of the plant; a supervisor is introduced to allow interpolation between the working points, which the system has been trained with. The FDI scheme has been tested by simulation on a non linear mechanical oscillator.

ICP and CBF Regulation: A new Hypothesis to Explain the “Windkessel” Phenomenon

The brain tamponade represents the final condition of a progressive intracranial pressure (ICP) increase up to values close to arterial blood pressure (BP) producing a reverberating flow pattern in the cerebral arteries with no net flow. This finding implies intracranial volume changes, therefore a full application of the Monro-Kellie doctrine is impossible. To resolve this contradiction, in eight pigs a reversible condition of brain tamponade was produced by infusing saline into a cerebral ventricle. The following parameters were measured: BP in the common carotid artery, ICP by the same needle utilised for the infusion, arterial and venous blood flow velocity (BFV) at, respectively, internal carotid artery (ICA) and sagittal sinus (SS) site by ultrasound technique. When ICP approached carotid BP values, reverberating BFV waves both at ICA and SS site were simultaneously observed. The arterial and venous reverberating waves appeared to be almost exactly superimposable, with a delay of about 40 msec. This synchronism between the pulsatile arterial and venous BFV indicates that the residual pulsation, still occurring at the arterial proximal level, is compensated by a passive compression-distension of the SS with no blood volume (that is net flow) crossing the intracranial vasculature.

Analysis of SCR Circuits Via Augmented State Transition Matrix

Abstract A method for determining the value of the forced response, free motion, mean value and harmonic coefficients of a linear timeinvariant dynamic system with canonical input is proposed. This method consists in the numerical computation of a particular augmented state system. When the input variable is canonical and periodic, the initial state may also be calculated by elementary matrix algebra. This me thod is particularly suited for analyzing electrical systems supplied bystat - ic convertors because during each conducting interval they are described by linear time-invariant differential equations with canonical inputs. An example is developed to illustrate the advantages of the method.

A simple control scheme for Mini Unmanned Aerial Vehicles

In this paper some results of an ongoing research program on mini UAV control are reported. It is proposed a simple control scheme that can be implemented on a low cost-low weight UAV, with limited feedback information. The control scheme is composed of feedforward term, which generates the trim control commands for the particular mission, and a decoupled feedback controller to stabilize the flight around the trajectory. The performance of the proposed scheme has been evaluated by simulation in various flight conditions