---

In this study, several data-driven techniques including system identification, adaptive neuro-fuzzy inference system (ANFIS), artificial neural network (ANN) and wavelet-artificial neural network (Wavelet-ANN) models were applied to model rainfall-runoff (RR) relationship. For this purpose, the daily stream flow time series of hydrometric station of Hajighoshan on Gorgan River and the daily rainfall time series belonging to five meteorological stations (Houtan, Maravehtapeh, Tamar, Cheshmehkhan and Tangrah climatologic stations) were used for period of 1983-2007. Root mean square error (RMSE) and correlation coefficient (r) statistics were employed to evaluate the performance of the ANN, ANFIS, ARX and ARMAX models for rainfall-runoff modeling. The results showed that ANFIS models outperformed the system identification, ANN and Wavelet-ANN models. ANFIS model in which preprocessed data using fuzzy interface system was used as input for ANN which could cope with non-linear nature of time series and performed better than others.

---

One of the challenges in non-destructive testing schemes using the ultrasound pulse-echo technique is to identify those defects whose sizes are less than or equal to the detection resolution that is dependent on the width of the ultrasound pulse. Existing methods also require a reference specimen of the same defective material, which may not be available in some cases. We present a new method for detecting and identifying such defects. In the proposed approach, each layer of the test specimen is modeled as a linear time invariant (LTI) filter, and therefore, each defect is characterized by its corresponding impulse response. We use a combination of time gating and system identification techniques to detect and identify the defects, and as such, do not require a reference specimen. To demonstrate the performance of the proposed approach, we tested metallic blocks in which specific defects were impregnated. Results show that the proposed method can detect such defects, does not depend on the test set-up (including the ultrasound transducer), does not require a reference specimen, and is capable of detecting several defects situated in different depth on top of each other

---

In this paper, we describe our implementation of an interior point algorithm for large scale systems. First we identify system with small and medium methods convex optimization, then we use interior point method for identification. Finally we offer an interior point method that uses nonlinear cost function and see that we achieve a good trade-off between error and CPU time. Actually, in this paper, we are looking for a method that can identify large scale systems with low model order, error and CPU time of solution of simulation. Previous articles didn’t check the order of the computed model, and the relationship between the error and CPU time. We assume that the model of our simulation is ARMA. We are going to identify a large scale system and compute the error and CPU time and compare the relationships. Examined data in this paper is related to cutaneous potential recordings of a pregnant woman. These data are pendulous and have a large standard deviation; therefore, it can’t be fitted with ordinary curve fittings, so we use the smoothing spline for computing the order of the model. Finally, we checked the influence of the number of data on error and CPU time and order of model

---

Hilbert-Huang transform (HHT) consists of two main parts: (1) empirical mode decomposition (EMD) to extract intrinsic mode functions (IMFs) and (2) Hilbert spectral analysis to obtain time-frequency characteristics of the IMFs through the Hilbert transform. Recently, a new enhanced HHT is proposed by the authors in which, two mathematical limitations that restrict the application of the Hilbert transform are circumvented and also an additional smoothing parameter is applied to decrease noise effects on the results. In this paper based on the HHT approach, a simple method for output-only identification of natural frequencies of linear structures is proposed in which HHT or enhanced HHT can be employed. In the proposed method, ambient response data measured at all degrees of freedom of the structure are used to obtain an averaged marginal spectrum. The averaged marginal spectrum is used for identifying the natural frequencies of the structure. In order to validate the effectiveness of the proposed identification method, ambient response data of an arch bridge and a 15-story building are examined. In the first case, the first six natural frequencies of the bridge in vertical direction are extracted. And in the second case, the first three natural frequencies of the building in East-West, North-South and torsional directions are identified. From the results, first, it is found that the enhanced HHT by employing the smoothing parameter is more efficient than the HHT in increasing the readability of the time-frequency-amplitude spectrum and also is capable to provide more accurate amplitude-frequency distribution; second, by comparing results of the proposed method with those obtained from other valid methods, it is concluded that the proposed identification method by using the enhanced HHT is accurately able to estimate the natural frequencies of structures. Regarding to simplicity of the proposed method, it can be applied as an efficient tool for identification of structures or employed to extract changes in frequencies due to occurrences of damages during strong ground motions.

---

Fuel control unit (FCU) is one of the most essential parts in a gas turbine engine; therefore it is necessary to be studied as an important part of the fuel control system. This paper report the use of Nonlinear Auto Regressive with eXogenous input (NARX) neural network model for modeling of the jet engine FCU. Therefore, To measure and recording data from the FCU inputs and output, the test bench including hydraulic system, data acquisition system and induction motor control system are designed and constructed. This setup is a mechatronic collection which includes mechanical design, discharge and pressure sensors, tachometer, control unit and piping systems. The process of modeling is carried out in MATLAB software. The identified model is evaluated with validation data and its response is compared with the real system response. Results demonstrate the effectiveness of the NARX neural network model and show that the real system is estimated by the NARX neural network model accurately.

---

Jet engine Fuel Control Unit (FCU) has been manufactured recently in the SSAC laboratory of Iran University of science and technology, and is being tested now using HIL testing. Regarding there is no possibility utilizing a jet engine in this test set; an induction motor was used as an actuator to transit rotation acquired using jet engine real time simulation to the FCU pump. In this paper, the induction motor’s velocity control is presented considering condition for the test set. To do so, system identification method is used to model the components employed in induction motor velocity control system. Then, the model is evaluated using experimental test results. Afterward, it is used to design ANFIS controller, and the controller parameters is adjusted employing IWO optimization algorithm as a strong tool for seeking in vast disorder spaces. Subsequently, the controller designed is implemented on a real system. Results gained using simulation and the designed controller implementation show that ANFIS controller designed using IWO algorithm works appropriately.

---

Aerospace Launch Vehicles (ALVs), used for launching artificial satellites and space stations to Earth orbits, usually encounter with failure in navigation systems . In these cases, survival of an ALV during accurate payloads injection in orbits is one of the most critical issues for Guidance and Control systems.An important challenge for safety of Aerospace Launch Vehicle (ALV) is their reliability against all types of faults. There is a requirement for on-board fault detection without deteriorating the performance of ALV. In this paper, a new software sensor is proposed for fault detection and compensation based on symmetrical behavior of the yaw and pitch channels of an ALV. For this purpose, using identification techniques on the yaw channel, a new software sensor is developed as an online rigid dynamic predictor for the pitch channel. The proposed software sensor is employed to generate the residual of estimation error as an indicator of predefined faults. The main novelty of this software sensor is online tuning of the virtual sensor against unforeseen variations in the parameters of the vehicle. Robustness of the new control system in the presence of asymmetric behavior is investigated. The efficiency of the proposed fault tolerant method is illustrated through simulations.

---

Reaction wheels are angular momentum exchange devices used to stabilize the position of the satellite and maneuvering. This actuator can change the momentum of the satellite to change the attitude of the system. During the process of operation, noise and disturbances arisen from the unbalancing of the wheels lead to inconvenient performance of the reaction wheels. Several works have been presented for active noise cancelation in these devices. But, the practical tools of signal processing such as filter banks and wavelets which used for offline de-noising are samples of very useful noise cancellation methods. If these toolboxes are employed for online de-noising these signal processing approaches are applicable for noisy systems such as reaction wheels. The main challenge of this strategy is delay arisen from the signal processing and this is inevitable. In this paper, a strategy of online wavelet de-noising is designed and proposed for noise cancellation in a reaction wheel. In this regards, for considering the delay compensation the method of Smith predictor is used to lead the delay of the process out of the closed loop control system. The accuracy of this algorithm requires an estimate of the system dynamics and the understanding of the delay system. According to the use of the FIR filter delay can be fully calculated. The recursive least squares used for identification reaction wheel as an estimate of the system.

---

Hybrid systems are a group of dynamical system which their behavior described by the interaction of discrete and continuous dynamical system behaviors. One of the subsets of hybrid systems, is piecewise affine system. Piecewise affine system identification, consists of estimating the parameters of each subsystem and the coefficients of the state-input boundary hyperplanes. In order to clustering the state-input space and estimating the feature matrixes simultaneously, bounded error algorithm and adaline neural network are used. It should be said that in this method, there is no need to know the number of linear subsystems of the piecewise affine system. Moreover, it should be noted that the identification method is extended based on on-line data acquisition from system. In continuation, this method is used to identify a benchmark mathematical piecewise affine system. By comparing the results with the reference paper, it is proven that this method has a good performance in clustering the state-input space and estimating the feature matrixes. In the end, by using the proposed method, an active water tank which its equations are described by the form of a piecewise affine system is identified.

---

In this paper, a novel test rig for a quarter car suspension system of Samand with McPherson mechanism is fabricated and its elasto-damping elements are dynamically identified. The inputs of test rig are road roughness and its acceleration and the outputs are sprung mass acceleration, un-sprung mass acceleration, suspension deflection, and tire deflection which are recorded by sensors. The test rig of suspension system includes McPherson mechanism with nonlinear spring and damper. This system is categorized as a multi-input-multi-output (MIMO) identified system. The nonlinear least squares iterative method, as a gray-box identification method, is used for finding the elasto-damping coefficients of tire and suspension elements. In this method, a nonlinear mathematical model is considered for the system and its parameters are calculated using the test rig data. The Levenberg–Marquardt algorithm (LMA) is used to solve the non-linear least squares problem. The outputs of the identified nonlinear model are compared with the measured experimental data. As a result, the test rig outputs are followed by the outputs of the identified model with acceptable errors. The compared results indicate a good performance of the proposed model to estimate the behavior of the nonlinear suspension elements.

---

In this paper, designing optimal linear controller for non-holonomic wheeled mobile robots based on Linear Quadratic Gaussian (LQG) controller is considered. Parameters of the governing kinematics equation of motion are derived based on system identification techniques by using real experimental data. The autoregressive moving average-exogenous input (ARMAX) models are taken into account. The least square (LS) algorithm is utilized to estimate the parameters of the model. Thereafter, optimal model order and the performance of the model are determined using several statistical analyses. Also, the recursive LS (RLS) with forgetting factor is employed to demonstrate the convergence of the model parameters. Verification of discrete linear model implies the possibility of using the linear controllers. Therefore, the optimal LQG controller for wheeled mobile robots is designed to track the reference trajectory. The Kalman observer is used to estimate un-measurable states of the robot. Furthermore, the optimal linear control together with system identification techniques yields simpler controller than nonlinear controllers. Designed controller and verified model are simulated using the MATLAB-Simulink software. Results show the effectiveness of the controller in tracking the desired reference trajectory.

---

In this paper the effect of artificial noise on the performance of nonlinear system identification method in reconstructing the response of a cantilever beam model having a local nonlinearity is investigated. For this purpose, the weak form equation governing the transverse vibration of a linear beam having a strongly nonlinear spring at the end is discretized by using Rayleigh-Ritz approach. Then, the derived equations are solved via Rung-Kutta method and the simulated response of the beam to impulse force is obtained. By contaminating the simulated response to artificial measurement noise, nonparametric nonlinear system identification is applied to reconstruct the response. Accordingly, intrinsic mode functions of the response are obtained by using advanced empirical mode decomposition, and nonlinear interaction model including intrinsic modal oscillators is constructed. Primary results show that the presence of noise in the response highly affects the sifting process which results in extraction of spurious intrinsic mode functions. In order to eradicate the effect of noise on this process, noise signals are used as masking signals in the advanced empirical mode decomposition method and intrinsic mode functions corresponding to the noise are extracted. Based on this approach, the dynamic of the noise in the response is identified and noise reduced signals are reconstructed by the intrinsic modal oscillators with appropriate accuracy.

---

In order to detect damage in a large-scale and complicated structure, there is need to exact nonlinear numerical modeling that its results have been analyzed using a method of system identification. In this way, the extended finite element model (XFEM) based on cohesive crack model (XFEM Based Cohesive Crack Segments) for concrete material as a reliable model is used for investigating real responses of Karun 3 concrete dam against applied loads and damages. In this model, whole of the structure is potentially under damage risk, while there is no initial crack. The dam is numerically modeled and analyzed using the finite element method (FEM) and XFEM Based Cohesive Crack Segments respectively, and the dam is analyzed under the seismic excitation. Then, for specification of crack effects and nonlinear behavior, the structural modal parameters and their variation should be investigated based on structure response for obtaining damage initiation time and its location by using system identification based on continuous Wavelet (CWT) transform. Results show that the dam natural frequencies decrease after the crack is formed, where decrease in longitudinal and vertical responses are more than the transversal response decrease. Moreover, crack width and its exact location are specified precisely from comparing the intact and damaged crest and central cantilever vibration modes. Therefore, the combination of XFEM Based Cohesive Crack Segments and CWT is useful procedure for structural health monitoring of concrete arch dams.

---

In this paper a controller has been presented based on the predictive control to drive and control the bipedal Nao robot. One of the challenges in the practical applying of these types of controllers is their high computational loading and the time-consuming control operations in each time step, in which it is suggested to use Laguerre Functions to reduce the computational loading of the predictive controller. In this study, at first using the conventional methods for the identification, and via the real data obtained from the Nao robot in Mechatronics research center of Qazvin Azad University, a proper model is proposed for walking the Nao robot which is considered as a two-dimensional motion in the plane. Then a controller will be designed to control the robot motion using the model based predictive controller. The purpose of this control approach in the first place is to stabilize the walking of the robot and then to guide and keep it on the desired trajectory, so that this trajectory tracking can be performed well as much as possible. Moreover, in order to evaluate the efficiency of the proposed controller, this controller has been compared with a proportional-integral-derivative controller and will be studied. The simulation results show the effectiveness of the proposed controller performance in the robot trajectory tracking, which finally comparing the obtained results from both of the control approaches, indicates the efficiency and different capabilities of the proposed method in this study.

---

Boring operations due to the large length to diameter ratio and the high flexibility of tool are prone to self-excited (chatter) vibration. This vibration may cause poor surface quality, low dimensional accuracy and tool breakage. In practice, chatter is the main limitation on production rate. The main reason of chatter phenomenon is the dynamic interaction between cutting process and structure of machine tool. By increasing the length of the cutting tool, the vibration tendency in the tool’s structure increases. Improving dynamic stiffness of the tool is the most effective solution for decreasing vibration and increasing chatter stability. For increasing the stability of the tool in long overhang boring operations, passive and active vibration control has been proposed and implemented. In active control methods, vibrations can be effectively damped over a various cutting conditions. The aim of this research is to enhance chatter stability of an industrial boring bar by increasing the dynamic stiffness. A VCA actuator is used for active vibration control. The designed setup can effectively suppress undesirable vibrations in the radial direction. First, modal parameters of the boring bar are determined by experimental modal analysis. Then, the transfer function of the actuator-tool setup is identified with the sweep frequency excitation. In the following, the direct velocity feedback is successfully implemented in the vibration control loop. The results of cutting tests indicate that the actuator has a great performance in suppressing vibrations and increasing the dynamic stiffness. Hence, the developed method can significantly increase chatter stability of boring operations.

---

Nowadays the exoskeleton, known as a useful device in robotic rehabilitation and elderly assistance, has been attracted the attention of many researches. One of the most important feathers of the exoskeleton robots are the compliant interaction with patient. The Series Elastic Actuators (SEA) not only interact with human compliantly but also provide several advantaged such as torque measurement and torque control. The pervious researches have used an inner position loop and an outer force loop. In this paper, the motor and power transmission model is also integrated in the controller design. In this paper, the parameters of the SEA, motor and links are identified firstly. Then, two model-based torque control is designed and introduced based on the velocity and current commands. In contrast to previous researched, the controller is proposed for the locked and free condition and the Lyapunov stability analysis is presented. Finally, the experimental validation test on the Sharif lower limb exoskeleton is presented for these controller. The experimental results of the controller show that the accuracy of torque control based on the current and velocity is 1.2 and 0.2 N.m, respectively.

---

System identification with the development of dynamic testing of structures has become one of the useful methods for structural health monitoring and damage detection and also finite element model updating. Identification of structural dynamic parameters is done by using excitation-responses data and includes physical-dynamical parameters such as mass, stiffness and damping matrices and/or modal parameters such as natural frequencies, damping ratios and modal shapes. Block pulse functions (BPFs) are a set of orthogonal functions that are used to approximate the variety of functions. These functions have explicit definition and provide simple formulation of complex problems. In this research, structural dynamic equations have been converted to state space equations and based on input BP coefficients and BP coefficients of displacement responses, a transfer function is extracted for each degree of freedom. Transfer functions include important information such as the eigenvalues of plant matrix. The equalization of transfer functions with ARX model led to estimate the eigenvalues of plant matrix and identification of dynamical parameters of structure is done based on these eigenvalues.To prove the validity and feasibility of proposed method, numerical simulation of the three-story shear frame with determined responses at all degrees of freedom and excited on base level is presented. Also, the accuracy of the identification process by applying noise at different levels to the structure response is investigated. The results reveal the proposed method can be beneficial in structural identification with less computational expenses and high accuracy.

---

In this paper, a linear dynamic model for a reaction wheel is identified using experimental analysis. To do this, online input-output data of reaction wheel is sent and received by CAN protocol working with the frequency of one mega bit per second. The experimental hardware consists of reaction wheel, processing board, CAN protocol, and LabVIEW monitoring. Modeling assumes the reaction wheel and its inner control circuit as a black box and takes into account the practical considerations. Initially, behavior of the reaction wheel is examined using test signals for velocity and acceleration as inputs. After that, the test signals are replaced by Chirp and PRBS signals and the output results are saved. According the results obtained in the tests, ARMAX and ARX linear dynamic models are assigned to the motor and different orders of these models are compared with each other to reach the appropriate order of the models. Furthermore, a delay is also incorporated in the model and its proper order is determined by the simulations. Finally, to validate the proposed model, the outputs of the model and plant are compared followed by exerting a new test signal. The results indicate a good agreement between the proposed model and the practical behavior.

---

---

In structural monitoring, modal parameters extracted from vibration data are commonly used to gain some information about the condition of bridges. However, even small amount of uncertainty in extracted modal parameters has a considerable erroneous impact on different processes of structural monitoring, including structural model updating and damage detection. Accordingly, in this research effects of different data processing methods and types of vibration tests such as ambient vibration and free vibration, on extracted modal parameters, have been studied. In this regard, four methods including Covariance based Stochastic Subspace Identification (Cov-SSI), Eigensystem Realization Algorithm (ERA), Frequency Domain Decomposition (FDD), and Analytical Mode Decomposition - Hilbert (AMD-Hilbert) have been used to estimate modal parameters. SSI and ERA are parametric methods in time domain in which mathematical bases are similar. FDD and AMD-Hilbert are non-parametric methods which work in frequency and time-frequency domain, respectively. SSI and FDD methods were used for ambient vibration test data and ERA was used for free vibration test records, while AMD-Hilbert method was applied for both free and ambient vibration data. In this article, vibration data of six points were measured from a girder of Gisha Bridge using three Molecular-electronic seismometer sensors, roved in three different setups. One sensor was chosen as reference and its position was fixed among different setups. Data of this sensor were later used for merging different setups results. Therefore, to extract modal parameters multi-setup merging approaches were inevitably used. The measurements were done in vertical direction which leads to identifying vertical bending modes. Ambient vibration responses were measured while the bridge was excited by wind and traffic under the bridge. Free vibration responses were measured after making an impact on the girder.  Two approaches were considered for merging. In the first approach setups were analyzed separately and their final results were combined together and in the second one, merging was done before the process of system identification which eliminates any need to analyze multiple times. A numerical model was also simulated to compare with the field results. Filtering of the recorded data was done before beginning of the system identification process to remove the drift and sudden changes in the signals. Data processing on ambient vibration responses resulted in the first three vertical bending modes which are compatible among all methods, to some extent. In addition, the first two vertical bending modes were identified from free vibration data. Similarity of the mode shapes between different methods were assessed using MAC criterion. Compatible results between these two types of test and numerical model, verifies the results. It is seen that FDD and SSI methods obtained more stable and reliable modal parameters among different setups. Results indicate more modes were identified for ambient vibration data compared to free vibration data. Since, in free response of the structure the first modes are more dominant, lower number of modes could be identified. Considering the non-stationary condition of the conducted vibration tests, the results indicate that the post-processing multi-setup merging approach works better than the pre-processing multi-setup merging approach.