Silvio Simani

Diagnosis techniques for sensor faults of industrial processes

A model-based procedure exploiting analytical redundancy for the detection and isolation of faults in input-output control sensors of a dynamic system is presented. The diagnosis system is based on state estimators, namely dynamic observers or Kalman filters designed in deterministic and stochastic environments, respectively, and uses residual analysis and statistical tests for fault detection and isolation. The state estimators are obtained from an input-output data process and standard identification techniques based on ARX or errors-in-variables models, depending on signal to noise ratio. In the latter case the Kalman filter parameters, i.e., the model parameters and input-output noise variances, are obtained by processing the noisy data according to the Frisch scheme rules. The proposed fault detection and isolation tool has been tested on a single-shaft industrial gas turbine model. Results from simulation show that minimum detectable faults are perfectly compatible with the industrial target of this application.

Fault diagnosis in power plant using neural networks

Abstract Fault diagnosis and identification (FDI) have been widely developed during recent years. Model-based methods, fault tree approaches and pattern recognition techniques are among the most common methodologies used in such tasks. Neural networks have been used in FDI problems for model approximation and pattern recognition as well. However, because of difficulties to perform Neural Network training on dynamic patterns, the second approach seems more adequate. In this paper, the FDI methodology consists of two stages. In the first stage, the fault is detected on the basis of residuals generated from a bank of Kalman filters, while, in the second stage, fault identification is obtained from pattern recognition techniques implemented by Neural Networks. The proposed fault diagnosis tool has been tested on a model of a power plant and results from simulations are reported and commented in the paper.

High-speed DSP-based implementation of piecewise-affine and piecewise-quadratic fuzzy systems

Abstract The paper tackles the problem of executing high-dimensional fuzzy inferences following zeroth- and first-order Takagi–Sugeno inference models. The choice of a proper AND operator, which is compatible with all the semantic requirements for a conjunctive aggregation, results in an input–output relationship which is piecewise affine or quadratic. The proposed inference procedures exploit this to perform the computation from inputs to output in a time that does not grow exponentially with the number of inputs. Some details of the implementation of the two inference procedures on a TMS320C6201 are given along with some simulation results demonstrating the effectiveness of the piecewise approach.

Parameter identification for piecewise-affine fuzzy models in noisy environment

Abstract In this paper the problem of identifying a fuzzy model from noisy data is addressed. The piecewise-affine fuzzy model structure is used as non-linear prototype for a multi–input, single–output unknown system. The consequents of the fuzzy model are identified from noisy data which are collected from experiments on the real system. The identification procedure is formulated within the Frisch scheme, well established for linear systems, which is extended so that it applies to piecewise-affine, constrained models.

Data Driven Approach for Wind Turbine Actuator and Sensor Fault Detection and Isolation

Abstract In order to improve reliability of wind turbines, it is important to detect and isolate faults as fast as possible, and handle them in an optimal way. An important component in modern wind turbines is the converter, which for a wind turbine control point–of–view normally provides the torque acting on the wind turbine generator, as well as measurement of this torque. In this work, a diagnosis strategy based on fuzzy prototypes is presented, in order to detect these faults in the converter, and isolate them either to be an actuator or a sensor fault. The fuzzy system is used since the model under investigation is nonlinear, whilst the wind speed measurement is highly noisy, due to the turbulence around the rotor plane. The fuzzy system consists of a set of piecewise affine Takagi–Sugeno models, which are identified from the noisy measurements acquired from the simulated wind turbine. The fault detection and isolation strategy is thus designed based on these fuzzy models. The wind turbine simulator is finally used to validate the achieved performances of the suggested fault detection and isolation scheme.

Identification of piecewise affine models in noisy environment

This paper addresses the identification of non-linear systems. A wide class of these systems can be described using non-linear time-invariant regression models, that can be approximated by means of piecewise affine prototypes with an arbitrary degree of accuracy. This work concerns the identification of piecewise affine model parameters through input-output data affected by additive noise. In order to show the effectiveness of the developed technique, the results obtained in the identification of both a simple simulated system and a real dynamic process are reported.

Fault Diagnosis of a Wind Turbine Benchmark via Identified Fuzzy Models

In order to improve the availability of wind turbines and to avoid catastrophic consequences, the detection of faults in their earlier occurrence is fundamental. This paper proposes the development of a fault diagnosis scheme relying on identified fuzzy models. The fuzzy theory is exploited since it allows approximating uncertain models and managing noisy data. These fuzzy models, in the form of Takagi-Sugeno prototypes, represent the residual generators used for fault detection and isolation (FDI). A wind turbine benchmark is used to validate the achieved performances of the designed FDI scheme. Finally, extensive comparisons with different fault diagnosis methods highlight the features of the suggested solution.

Model-based robust fault detection and isolation of an industrial gas turbine prototype using soft computing techniques

This study proposes a model-based robust fault detection and isolation (RFDI) method with hybrid structure. Robust detection and isolation of the realistic faults of an industrial gas turbine in steady-state conditions is mainly considered. For residual generation, a bank of time-delay multilayer perceptron (MLP) models is used, and in fault detection step, a passive approach based on model error modelling is employed to achieve threshold adaptation. To do so, local linear neuro-fuzzy (LLNF) modelling is utilised for constructing error-model to generate uncertainty interval upon the system output in order to make decision whether a fault occurred or not. This model is trained using local linear model tree (LOLIMOT) which is a progressive tree-construction algorithm. Simple thresholding is also used along with adaptive thresholding in fault detection phase for comparative purposes. Besides, another MLP neural network is utilised to isolate the faults. In order to show the effectiveness of proposed RFDI method, it was tested on a single-shaft industrial gas turbine prototype model and has been evaluated based on the gas turbine data. A brief comparative study with the related works done on this gas turbine benchmark is also provided to show the pros and cons of the presented RFDI method.

Hybrid Model–Based Fault Detection of Wind Turbine Sensors

Abstract In order to improve reliability of wind turbines, it is important to detect faults in their very early occurrence, and to handle them in an optimal way. This paper focuses on the pitch sensors of the turbine blade system, as they are mainly used for wind turbine control, in order to maximise the power production, and the efficiency of the whole process. On the other hand, as the input-output behaviour of the system under diagnosis is nonlinear, this work suggests a modelling scheme relying on piecewise affine models, whose parameters are identified through the acquired input–output measurements affected by measurement uncertainty. Therefore, these hybrid prototypes are exploited for generating suitable residual signals, which allow the detection and the isolation of the considered sensor faults. This noise rejection scheme is used since the wind turbine measurements are not very reliable, due to the uncertainty of wind speed acting on the wind turbine, and to the turbulence around the rotor plane. A detailed benchmark model simulating the wind turbine where realistic fault conditions can be considered shows the effectiveness of both the identification and fault diagnosis techniques.

Application of a data-driven fuzzy control design to a wind turbine benchmark model

In general, the modelling of wind turbines is a challenging task, since they are complex dynamic systems, whose aerodynamics are nonlinear and unsteady. Accurate models should contain many degrees of freedom, and their control algorithm design must account for these complexities. However, these algorithms must capture the most important turbine dynamics without being too complex and unwieldy, mainly when they have to be implemented in real-time applications. The first contribution of this work consists of providing an application example of the design and testing through simulations, of a data-driven fuzzy wind turbine control. In particular, the strategy is based on fuzzy modelling and identification approaches to model-based control design. Fuzzy modelling and identification can represent an alternative for developing experimental models of complex systems, directly derived directly from measured input-output data without detailed system assumptions. Regarding the controller design, this paper suggests again a fuzzy control approach for the adjustment of both the wind turbine blade pitch angle and the generator torque. The effectiveness of the proposed strategies is assessed on the data sequences acquired from the considered wind turbine benchmark. Several experiments provide the evidence of the advantages of the proposed regulator with respect to different control methods.