Journal of Modeling and Simulation
Volume:51 Issue: 2, Summer-Autumn 2019

Due to the important role of the boiler-turbine units in industries and electricity generation, it is important to diagnose different types of faults in different parts of boiler-turbine system. Different parts of a boiler-turbine system like the sensor or actuator or plant can be affected by various types of faults. In this paper, the effects of the occurrence of faults on the actuators are investigated and analyzed and fault detection of boiler-turbine actuators is studied. For fault detection purpose, a dynamic neural network with an internal feedback is applied to generate the residual. After generating the residuals, the decision making step, as the most crucial part of the fault detection process, has to be followed. For designing a proper threshold, which is sensitive to different types of faults and insensitive to noise, the robust threshold is designed using the model error modeling method. The robust threshold is designed using a dynamic neural network with an internal feedback. The results for multiple types of faults and various outputs show the effectiveness of this approach for designing the threshold. As a practical case of study the dynamic model of the boiler-turbine unit, which was represented by Bell and Astrom in their paper, is considered.

In this paper an adaptive neuro fuzzy inference system based on interval Gaussian type-2 fuzzy sets in the antecedent part and Gaussian type-1 fuzzy sets as coefficients of linear combination of input variables in the consequent part is presented. The capability of the proposed method (we named ANFIS2) to function approximation and dynamical system identification is shown. The ANFIS2 structure is very similar to ANFIS, but in ANFIS2, a layer has been added for the purpose of type reduction. An adaptive learning rate based backpropagation with convergence guaranteed is used for parameter learning. Finally, the proposed ANFIS2 are used to control of a flexible joint robot arm that can be used in robot arm. Simulation results shows the proposed ANFIS2 with Gaussian type-1 fuzzy set as coefficients of linear combination of input variables in the consequent part has good performance and high accuracy but more training time. In the simulation, ANFIS2 is compared with conventional ANFIS. The results show that, in abrupt changes, the type-2 fuzzy system proof of efficiency and excellence to the type-1 fuzzy system.

In this paper, the min-time optimal control problem is mainly investigated in the linear time invariant (LTI) continuous-time control system with a constrained input. A high order dynamical LTI system is firstly considered for this purpose. Then the Pontryagin principle and some necessary optimality conditions have been simultaneously used to solve the optimal control problem. These optimality conditions would usually lead to some complicated equations while some integral terms may be presented. Then a systematic procedure based on state transition matrix will be addressed to overcome and simplify the mentioned complexities. Therefore the state transition matrix would be used to determine the exact solution of the min-time control problem in a typical LTI system. The min-time problem would be converted to some algebraic nonlinear equations by using of the state transition matrix. These algebraic equations are depended on some definite parameters. Hence the required design parameters as well as switching times and the possible minimum time would be analytically determined in the minimum-time optimal control problem. Thus the min-time control signal would be explicitly determined by computing of the switching times and also some other constants. The proposed control scheme is applied in some typical dynamical examples to show the effectiveness of the suggested control method.

Analysis of nonlinear autonomous systems has been a popular field of study in recent decades. As a noticeable nonlinear behavior, chaotic dynamics has been intensively investigated since Lorenz discovered the first physical evidence of chaos in his famous equations. Although many chaotic systems have been ever reported in the literature, a systematic and qualitative approach for chaos generation is still a challenging issue. Recently, we have developed an analysis tool which provides globally valid results about the qualitative behavior of some nonlinear systems based on their pseudo-linear form of representation. In this paper, it is applied to generate conservative chaos by focusing on the essential qualitative attribute of conservative chaotic behavior. This feature is the continual stretching and folding of system trajectories which never settle down to a periodic regime. Indeed, it is tried to create this quality of behavior through the aforementioned qualitative analysis tool. The proposed approach helps us to generate a new class of chaotic systems with highly remarkable characteristics. The most elegant one is its almost parameter independency for chaos generation; There is no need for a trial-and-error mechanism to find the exact parameters’ values in order to produce chaotic behavior. It is shown that the system exhibits conservative chaotic dynamics for almost all parameters’ values. The chaotic behavior of the derived system is verified through the analysis of Lyapunov exponents and dimension as well. Besides, the frequency power spectrum analysis is also performed in order to put more emphasis on the chaotic behavior of the system.

Gas metal arc welding (GMAW) is a widespread process used for rapid prototyping of metallic parts. In this process, in order to obtain a desired welding geometry, it is very important to predict the weld bead geometry based on the input process parameters, which are voltage, wire feed rate, welding speed and welding nozzle angle. For this purpose, a global model of the welding geometry must be defined based on these parameters. Due to the non-linear and coupled multivariable relationship between the process parameters and the weld bead geometry, it is not possible to define this model in form of an explicit mathematical expression, and therefore application of supervised learning algorithms can be investigated as an efficient alternative in this problem. In this paper, application of the extreme learning machine (ELM) and support vector machine (SVM), as two efficient and powerful machine learning algorithms for predictive modelling of this process has been investigated and error analysis of the proposed models suggest that the output parameters of this process can be predicted by the ELM algorithm with higher precision and generalization capability.

In this paper, a numerical study has been carried out for coupled mass, momentum and heat transfer in the field under effects of natural convection. For this purpose, the unsteady incompressible Navier-Stokes equations with the terms of the Buoyancy forces (due to temperature gradients), energy conservation and concentration (mass) transfer equations have been simultaneously solved using appropriate numerical methods. In order to discretize spatial terms, a combined formulation contains a second-order central difference method and the first-order upwind scheme has been used. Time integration of the governing equation has been performed using the fourth order Runge-Kutta method. The effect of variations of the mass of contaminant has been studied in changing the flow and thermal fields structure. It is concluded from obtained results, an increase in mass flow rate of secondary (mass) injection, alters the structure of the flow and thermal fields. Comparison of the results obtained from the numerical model with appropriate reference data shown that the model has relatively good accuracy.

In this paper, dynamic modeling of a Vestas 660 kW wind turbine and its validation are performed based on operational data extracted from Eoun-Ebn-Ali wind farm in Tabriz, Iran. The operational data show that the turbine under study, with a classical PI controller, encounters high fluctuations when controlling the output power at its rated value. The turbine modeling is performed by deriving the non-linear dynamic equations of different subsystems. Then, the model parameters are identified such that the model response matches the actual response. In order to validate the proposed model, inputs to the actual wind turbine (wind speed, pitch angle and generator torque) are fed to the model in MATLAB as well as FAST tool, and the output powers are compared. In order to improve the control performance and alleviate fluctuations in the full-load region, considering the nonlinear and complex behavior of the system, a neuro-fuzzy controller is designed and simulated to control the pitch angle. In this controller, neural network is used to adjust the membership functions of the fuzzy controller. Simulation results of the designed neuro-fuzzy controller indicate the improved performance of the closed-loop system compared to the actual and simulated results from the classical PI controller.

In this study, a novel control algorithm, based on a P300-based brain-computer interface (BCI) is deployed to control a 2-DoF robotic arm. Eight subjects, including five men and three women, perform a 2-dimensional target tracking in a simulated environment. Their EEG (Electroencephalography) signals from the visual cortex are recorded and P300 components are extracted and evaluated to deliver a real-time BCI-based controller. The volunteer’s intention is recognized and will be decoded as an appropriate command to control the cursor.  The final processed BCI output is used to control a simulated robotic arm in a 2-dimensional space. The results show that the system allows the robot's end-effector to move between arbitrary positions in a point-to-point session with the desired accuracy. This model is tested and compared on the Dataset II of the BCI competition. The best result is obtained with a multi-classifier solution with a recognition rate of 97 percent, without channel selection before the classification.

Turbulent flow over an array of wall-mounted cubic obstacles has been numerically investigated using large-eddy simulation. The simulations have been performed using high- performance computations with local cluster systems. The array of cubes is fully submerged in a simulated deep rough-wall atmospheric boundary-layer with high turbulence intensity characteristics of environmental turbulent flows. Four different approaches have been tested to reproduce the approaching highly turbulent inflow condition. Significant influence of the inlet boundary condition on the predictive streamwise root mean squared velocity (and second-order turbulence statistics if generalized) have been observed. A pro- posed method based on inserting a solid grid at the inlet of the domain with superimposed correlated random fluctuations has been selected as the inlet boundary condition to conduct the simulations. Three different subgrid-scale (SGS) models have been also used to compare their predictive performance in turbulence statistics and temporal energy spectra. It was observed that the choice SGS model does not have considerable effect on the second-order turbulence statistics, however, it was influential on the predicted energy level in the energy spectra. It was also observed that the flow reaches a self-similar states after the second row of obstacles which was different from the reported value in some of the previous studies.

Supply chain is an integrated system of facilities and activities. Gaining the optimum design of demand satisfaction network is one of the most important live issues in the decision making problems category. Most of previous studies considered unreal assumptions such as the lack of capacity constraints to satisfy demand in the network and in hubs. By considering the nature of the case that have been studied in this research, the assumption of unlimited capacity to satisfy the demand is justified. Another common assumption in hub location problems is the lack of direct connection between the nodes. In this research and in real world problems would be seen that the direct link between the nodes can be effective in reducing system costs and increase the efficiency of the network. The other innovations of current research is considering uncertain nature of the demand data, oscillation and changes in costs anticipation and actual hub establishing costs, fuzzy numbers are used to represent these values. Problem modeling is held in a fuzzy state and a hybrid method is represented to solve the problem. At first, defuzzification of the model is taken place. Afterwards, all possible answers are considered with the help of the Genetic Algorithm. At last, the optimum case were chosen by using VIKOR ranking method. Calculation results for the designing of optical fiber network between cities are showing good and acceptable performance of the proposed method in an acceptable solving time.

Today, one of the important issues of power quality (PQ) in power systems is the current harmonics. Increasing expansion of nonlinear loads at different parts of the electric network makes harmonic distortion flow through the network. This causes the network to have background voltage and current harmonic distortion and even affect on the PQ of linear load performance. Therefore, it is important to determine the harmonic contribution between load and network in distribution networks and industrial centers. In this paper, a new procedure for determining the current harmonic contribution around a quiescent point at the PCC is presented using load modeling based on crossed-frequency admittance matrix. The novelty of this paper is that, firstly, instead of using the harmonic Norton model of load, which has often been used in papers related to the harmonic contribution, it uses the crossed-frequency admittance matrix model, which is closed to actual load model; and secondly, considering the fact that a linear load has a diagonal form for its crossed frequency admittance matrix, the separation of harmonic contribution is made.

A new iterative learning controller is proposed for a general unknown discretetime-varying nonlinear non-affine system represented by NARMAX (Nonlinear Autoregressive Moving Average with eXogenous inputs) model. The proposed controller is composed of an iterative learning neural identifier and an iterative learning controller. Iterative learning control and iterative learning identification are integrated in each iteration. A multi-layer neural network is used for identification. Since the system considered in this paper is time-varying, the proposed neural identifier also is time-varying. The weights of the neural identifier are updated at each iteration, so both tracking performance and identification are improved at each iteration simultaneously. The structure of the proposed neural network used for identification system is affine in control input. Then new iterative learning control law based on the neural identifier is proposed and applied to the system. It should be mentioned that the proposed integrated algorithm has a faster, better and more accurate performance when compared with other iterative learning control algorithms proposed for similar systems. Convergence of both the trajectory tracking error and identification error is guaranteed along the iteration domain with repeating the process within a time-limited range. Simulation and comparison results easily approve the effectiveness of the proposed method.

Intrusion detection is one of the main challenges in wireless systems especially in Internet of things (IOT) based networks. There are various attack types such as probe, denial of service, remote to local and user to root. In addition to the known attacks and malicious behaviors, there are various unknown attacks which some of them have similar behaviors with respect to each other or mimic the normal behavior. So, classification of connections in IOT based networks is a hard and challenging task. In this paper, an intrusion detection framework is proposed for classification of various attacks and separation of them from the normal connections. The double discriminant embedding (DDE) method is used to transform the original feature space of data. This transform is implemented in two steps. In the first step, the difference between the features is maximized; and in the second one, the difference between classes is increased. The extracted features not only have less overlapping with respect to each other and contain less redundant information but also they provide more separation between different classes. The extracted features are fed to the support vector machine (SVM) with polynomial kernel for classification. The experiments on NSL-KDD dataset have shown improvement of the SVM classifier when the DDE features are used.

In this paper, an efficient algorithm for the efficiency maximization of the multilevel linear amplification using nonlinear components (M-LINC) systems is proposed regarding the linearity of the system. In this algorithm, we use the generalized memory polynomial (GMP) to provide a behavioral model for the power amplifier (PA) and calculate the power spectral density (PSD) of the output signal of the system instead of using complicated analytical methods or time-consuming circuit level simulations. In order to have a reliable model, a modeling process which validates the static and dynamic behaviors of the obtained model is provided, and the validation is performed through the time domain signals, PSD, and AM-AM characteristics. As an example, we optimize the efficiency of a 6 level LINC system with a 2.4 GHz 25 W Doherty PA and a 15 MHz three-tone signal using the particle swarm optimization (PSO) method where an upper bound on the adjacent channel leakage ratio (ACLR) is considered as the linearity constraint. Our results show that for each given ACLR limit by a communication standard, the efficiency can be maximized with a certain number of levels in M-LINC system. Furthermore, the results unveil the trade-off between linearity and efficiency in M-LINC systems.

The aim of this study is to investigate the effects of non-Newtonian blood rheology models on the wall shear stress (WSS) distribution in a cohort of patients-specific coronary arteries. Twenty patients with diseased left anterior descending (LAD) coronary arteries (with varying degrees of stenosis severity from mild to severe) who underwent angiography and in-vivo pressure measurements were selected to perform computational fluid dynamics (CFD) simulations. Three-dimensional (3D) patient-specific geometries were reconstructed from 3D quantitative coronary angiography. To compare the effects of rheological properties of blood on WSS along the arteries, each artery was divided into 3 segments; proximal (pre-stenosis), stenosis and distal (post-stenosis). Blood was modelled as a Newtonian and non-Newtonian (Carreau-Yasuda, Casson and Power-law) fluid.Our findings showed that the WSS distributions over proximal and stenosis segments were significantly affected by the non-Newtonian properties of blood whereas the effect was negligible over distal segment. On the other hand, the type of non-Newtonian model is important to achieve accurate results over proximal and stenosis regions, but over distal region, it does not matter what model is used. Therefore, to simplify the simulation, the Newtonian model can be acceptable in finding the wall shear stress distribution over the distal region regardless of severity of stenosis.

In this paper, we explore the impacts of utilizing intelligent reflecting surfaces (IRS) in a power-splitting based simultaneous wireless information and power transfer (PS-SWIPT) system and compare its performance with the traditional decode and forward relaying system. To analyze a more practical system, it is also assumed that the receiving nodes are subject to decoding cost, and they are only informed about the imperfect channel state information (CSI). First, we drive the achievable data rate of single IRS-assisted cooperative communications, and to maximize the achievable rate, optimal phase shits for each elements of the IRS node is derived, and finally the optimal power splitting ratio at the destination is obtained. The, the system model is extended to consider two and multiple IRS-assisted system. The respective achievable rates are derived and optimized accordingly. To evaluate the benefits of using the IRS, we have also derived the achievable rate for a two-hop decode and forward relaying scheme, wherein both the relay and the destination not only did they equip with rededicated power but also they can harvest energy from the received signals to provide the required power for the decoding. For this case, optimal power splitting factor at both the relay and the destination are optimized. Finally, the numerical results are presented to examine and compare the performance of the two considered systems. It is shown that by increasing the size of the reflecting surface, IRS-based cooperative transmission outperforms the conventional relaying scheme.

In this paper the generalizations of the Burgers-Korteweg-de Vries model with small parameter derived by Kudryshov et al[ N.A. Kudryashov, D.I. Sinelshchikov. Extended models of non-linear waves in liquid with gas bubbles, International Journal of Non-Linear Mechanics 63 (2014) 31-38] is studied. A comprehensive study on the approximate symmetry analysis of the waves models is presented. First, we obtain approximate symmetry for the equation. Subsequently, in a physical application, using the first-order approximate symmetries, corresponding approximate invariant solutions to the perturbed non-linear models are obtained.