Giansalvo Cirrincione

Bearing Fault Detection by a Novel Condition-Monitoring Scheme Based on Statistical-Time Features and Neural Networks

Bearing degradation is the most common source of faults in electrical machines. In this context, this work presents a novel monitoring scheme applied to diagnose bearing faults. Apart from detecting local defects, i.e., single-point ball and raceway faults, it takes also into account the detection of distributed defects, such as roughness. The development of diagnosis methodologies considering both kinds of bearing faults is, nowadays, subject of concern in fault diagnosis of electrical machines. First, the method analyzes the most significant statistical-time features calculated from vibration signal. Then, it uses a variant of the curvilinear component analysis, a nonlinear manifold learning technique, for compression and visualization of the feature behavior. It allows interpreting the underlying physical phenomenon. This technique has demonstrated to be a very powerful and promising tool in the diagnosis area. Finally, a hierarchical neural network structure is used to perform the classification stage. The effectiveness of this condition-monitoring scheme has been verified by experimental results obtained from different operating conditions.

The MCA EXIN neuron for the minor component analysis

The minor component analysis (MCA) deals with the recovery of the eigenvector associated to the smallest eigenvalue of the autocorrelation matrix of the input data and is a very important tool for signal processing and data analysis. It is almost exclusively solved by linear neurons. This paper presents a linear neuron endowed with a novel learning law, called MCA EXINn and analyzes its features. The neural literature about MCA is very poor, in the sense that both a little theoretical basis is given (almost always focusing on the ODE asymptotic approximation) and only experiments on toy problems (at most four-dimensional problems) are presented, without any numerical analysis. This work addresses these problems and lays sound theoretical foundations for the neural MCA theory. In particular, it classifies the MCA neurons according to the Riemannian metric and justifies, from the analysis of the degeneracy of the error cost; the different behavior in approaching convergence. The cost landscape is studied and used as a basis for the analysis of the asymptotic behavior. All the phases of the dynamics of the MCA algorithms are investigated in detail and, together with the numerical analysis, lead to the identification of three possible kinds of divergence, here called sudden, dynamic, and numerical. The importance of the choice of low initial conditions is also explained. A lot of importance is given to the experimental part, where simulations on high-dimensional problems are,presented and analyzed. The orthogonal regression or total least squares (TLS) technique is also presented, together with a real-world application on the identification of the parameters of an electrical machine. It can be concluded that MCA EXIN is the best MCA neuron in terms of stability (no finite time divergence), speed, and accuracy.

A new adaptive integration methodology for estimating flux in induction machine drives

This paper proposes a simple adaptive notch filter for the elimination of the dc component in the integration of signals used for the flux estimation in high performance ac drives. This integration method is composed of two identical adaptive noise cancellers using a linear neural network with just one bias weight. Its behavior has been investigated in simulation as applied to electrical drives and compared with other four traditional integration algorithms. A test bench has been then developed for its experimentation in a field oriented controlled induction machine. It has been verified that this integration algorithm outperforms the other algorithms in estimating the rotor flux even at low speeds.

Sensorless Control of Induction Machines by a New Neural Algorithm: The TLS EXIN Neuron

This paper proposes two speed observers for high-performance induction machine drives, both adopting an online adaptation law based on a new total least-squares (TLS) technique: the TLS EXIN neuron. The first is a model reference adaptive system (MRAS) observer with a neural adaptive integrator in the reference model and a neural adaptive model trained online by the TLS EXIN neuron. This observer, presented in a previous article of the authors, has been improved here in two aspects: first, the neural adaptive integrator has been modified to make its learning factor vary according to the reference speed of the drive, second, a neural adaptive model based on the modified Euler integration has been proposed to solve the discretization instability problem in field-weakening. The second observer is a new full-order adaptive one based on the state equations of the induction machine, where the speed is estimated by means of a TLS EXIN adaptation technique. Both these observers have been provided with an inverter nonlinearity compensation algorithm and with techniques for the online estimation of the stator resistance of the machine. Moreover, a thorough theoretical stability analysis has been developed for them both, with particular reference to the field-weakening region behavior for the TLS MRAS observer and to the regenerating mode at low speeds for the TLS adaptive observer. Both speed observers have been verified in numerical simulation and experimentally on a test setup, and have also been compared experimentally with the BPN MRAS observer, the classic adaptive observer and with an open-loop estimator. Results show that both proposed observers outperform all other three observers in every working condition, with the TLS adaptive observer resulting in a better performance than the TLS MRAS observer

Constrained minimization for parameter estimation of induction motors in saturated and unsaturated conditions

This paper presents the analytical solution of the application of the constrained least-squares (LS) minimization to the online parameter estimation of induction machines. This constrained minimization is derived from the classical linear dynamical model of the induction machine, and therefore it is able to estimated the steady-state value of the electrical parameters of the induction motor under different magnetization levels. The methodology has been verified in simulation with a dynamical model which takes into account iron path saturation effects. After a description of the experimental setup and its signal processing systems, the methodology is verified experimentally under saturated and unsaturated working conditions, and the results are discussed and compared to those obtained with a classical unconstrained ordinary LS technique.

Diagnosis of broken-bars fault in induction machines using higher order spectral analysis.

Detection and identification of induction machine faults through the stator current signal using higher order spectra analysis is presented. This technique is known as motor current signature analysis (MCSA). This paper proposes two higher order spectra techniques, namely the power spectrum and the slices of bi-spectrum used for the analysis of induction machine stator current leading to the detection of electrical failures within the rotor cage. The method has been tested by using both healthy and broken rotor bars cases for an 18.5 kW-220 V/380 V-50 Hz-2 pair of poles induction motor under different load conditions. Experimental signals have been analyzed highlighting that bi-spectrum results show their superiority in the accurate detection of rotor broken bars. Even when the induction machine is rotating at a low level of shaft load (no-load condition), the rotor fault detection is efficient. We will also demonstrate through the analysis and experimental verification, that our proposed proposed-method has better detection performance in terms of receiver operation characteristics (ROC) curves and precision-recall graph.

Sensorless Control of Induction Motors by Reduced Order Observer With MCA EXIN + Based Adaptive Speed Estimation

This paper presents a sensorless technique for high-performance induction machine drives based on neural networks. It proposes a reduced order speed observer where the speed is estimated with a new generalized least-squares technique based on the minor component analysis (MCA) EXIN + neuron. With this regard, the main original aspects of this work are the development of two original choices of the gain matrix of the observer, one of which guarantees the poles of the observer to be fixed on one point of the negative real semi-axis in spite of rotor speed, and the adoption of a completely new speed estimation law based on the MCA EXIN + neuron. The methodology has been verified experimentally on a rotor flux oriented vector controlled drive and has proven to work at very low operating speed at no-load and rated load (down to 3 rad/s corresponding to 28.6 rpm), to have good estimation accuracy both in speed transient and in steady-state and to work correctly at zero-speed, at no-load, and at medium loads. A comparison with the classic full-order adaptive observer under the same working conditions has proven that the proposed observer exhibits a better performance in terms of lowest working speed and zero-speed operation

A new TLS-based MRAS speed estimation with adaptive integration for high-performance induction machine drives

This paper presents a new model reference adaptive system (MRAS) speed observer for high-performance field-oriented control induction motor drives which employs the flux error for estimating the rotor speed, but overcomes the pure integration problems by using a novel adaptive integration method based on neural adaptive filtering. A linear neuron (the ADALINE) is employed for the estimation of both the rotor speed and the rotor flux-linkage with a recursive total least-squares (TLS) algorithm (the TLS EXIN neuron) for online training. This neural model is also used as a predictor, that is with no feedback loops between the output of the neural network and its input. The proposed scheme has been implemented in a test setup and compared with an MRAS ordinary least-squares speed estimation with low-pass filter integration, with the well-known Schauders scheme and with the latest Holtzs scheme. The experimental results show that in the high and medium-speed ranges with and without load, the four algorithms give practically the same results, while in low-speed ranges (that is, below 10 rad/s ) the TLS-based algorithm outperforms the other three algorithms. Successful experiments have also been made to verify the robustness of the algorithm to load perturbations and to test its performance at zero-speed operation.

Linear system identification using the TLS EXIN neuron

Abstract The paper presents a neural approach for the parameter estimation of adaptive IIR filters for linear system identification. It is based on a novel neuron, the TLS EXIN neuron, capable of resolving the TLS problem present in this kind of estimation, where noisy errors affect not only the observation vector but also the data matrix. After a survey of other techniques for solving such parameters estimations, the TLS EXIN neuron is compared both theoretically and numerically with the former techniques, resulting in improved performance. Moreover, it is also proved that the TLS EXIN neuron permits some powerful acceleration techniques, unlike the other approaches. These results are also shown numerically.

A new experimental application of least-squares techniques for the estimation of the induction motor parameters

This paper deals with a new experimental approach to the parameter estimation of induction motors with least-squares techniques. In particular, it exploits the robustness of total least-squares (TLS) techniques in noisy environments by using a new neuron, the TLS EXIN, which is easily implemented on-line. After showing that ordinary least-squares (OLS) algorithms, classically employed in literature, are quite unreliable in presence of noisy measurements, which is not the case for TLS, the TLS EXIN neuron is applied numerically and experimentally for retrieving the parameters of the induction motor by means of a test-bench. Additionally, for the case of very noisy data, a refinement of the TLS estimation has been obtained by the application of a constrained optimisation algorithm which explicitly takes into account the relationships among the K-parameters. The strength of this approach and the enhancement obtained is fully demonstrated first numerically and then verified experimentally.