Design of Intelligent Predictive Controller for Electrically Heated Micro Heat Exchanger

Notice of Violation of IEEE Publication Principles<br><br>"Internal model control based on locally linear model tree (LOLIMOT) model with application to a PH neutral process"<br>by M. Jalili-Kharaajoo, A. Rahmati and F. Rashidi,<br>in proceedings of the 2003 IEEE Internationa Conference on Systems, Man, and Cybernetics, vol 4, pp. 3051-3-55.<br><br>After careful and considered review of the content and authorship of this paper by a duly constituted expert committee, this paper has been found to be in violation of IEEE's Publication Principles.<br><br>This paper contains portions of original text from the paper cited below. The original text was copied without attribution (including appropriate references to the original author(s) and/or paper title) and without permission.<br><br>"Nonlinear Internal Model Control for MISO Systems Based on Local Linear Neuro-Fuzzy Models"<br> by A. Fink, O. Nelles and R. Isermann,<br>in 15th IFAC World Congress, Barcelona, Spain, 2002.<br><br> <br/> The internal model control (IMC) scheme has been widely applied in the field of process control. So far, IMC has been mainly applied to linear processes. This paper discusses the extension of the IMC scheme to nonlinear processes based on local linear models where the properties of linear design procedures can be exploited. The IMC scheme results in controllers that are comparable to conventional multi layer perceptron (MLP) networks. In practice, the tuning of conventional MLP based controllers can be very time-consuming whereas the IMC design procedure is very simple and reliable. The design effort of the IMC based on locally linear model tree (LOLIMOT) algorithm are discussed and the control results are compared by application to nonlinear control of an industrial-scale PH neutralization process. Simulation studies of a PH neutralization process confirm the excellent nonlinear modeling properties of the proposed locally linear network and illustrate the potential for set point tracking and disturbance rejection within an IMC framework.

Acquiring new customers in any business is much more expensive than trying to keep the existing ones. Thus many prediction models are presented to detect churning customers. The objective of this paper was to improve the predictive accuracy and interpretability of churn detection. For this purpose, the application of the locally linear model tree (LOLIMOT) algorithm, which integrates the advantage of neural networks, tree model and fuzzy modeling, was experimented. Applied to the data of a major telecommunication company, the method is found to improve prediction accuracy significantly compared to other algorithms, such as artificial neural networks, decision trees, and logistic regression. The results also indicate that LOLIMOT can have accurate outcome in extremely unbalanced datasets.

This chapter deals with local linear neuro-fuzzy models, also referred to as Takagi-Sugeno fuzzy models, and appropriate algorithms for their identification from data. The local linear modeling approach is based on a divide-andconquer strategy. A complex modeling problem is divided into a number of smaller and thus simpler subproblems, which are solved (almost) independently by identifying simple, e.g., linear, models. The most important factor for the success of such an approach is the division strategy for the original complex problem. Therefore, the properties of local linear neuro-fuzzy models crucially depend on the applied construction algorithm that implements a certain division strategy. This chapter focuses on the local linear model tree (LOLIMOT) algorithm proposed by Nelles [267, 271, 286].

We applied neuro-fuzzy model to predict reservoir properties from seismic attributes. We used local linear neuro-fuzzy model tree (LOLIMOT) algorithm to train our model. This model uses well log data and seismic attributes in a well location for training.

A nonlinear adaptive output feedback control approach based on LOLIMOT is proposed for a class of nonaffine system in the non-strict feedback form with unknown nonlinearities in this paper. In order to overcome the disadvantages of output feedback control, LOLIMOT which can identify the nonlinear system accurately is applied. During the control process, backstepping approach is utilized to design the adaptive output feedback controller and the internal pseudo-controller is used as one-step predictor. The effectiveness of proposed scheme is shown by simulation on a proper nonlinear system.

One of the most challenging topics in analyzing multi-dimensional geo-spatial data such as geophysical data-sets is detecting outlier data. The issue mainly originates from the difficulty in describing “normality” or “abnormality” due to the complexity of the relationships between the data elements. Considerable number of methods have been proposed and applied for detecting outliers whether they are assumed to be noise, anomalies within the data-set or simply isolated events. A new outlier detection method reached from automatic training of Local Linear Model Tree (LOLIMOT) network, and based on the data selected by K-Nearest Neighborhood (KNN) search is proposed in this research. The procedure of selecting data pairs is through decile analysis using distances calculated during KNN data grouping. Experiment on a synthetic 12 cluster 3D data-set is indicative of the method’s robust performance where calculated Cumulative Error Percentage (CEP) is 13% for the method whereas the nearest value for the KNN is 19%. Also, by applying the method on a micro-gravimetric data and an earthquake catalogue related to the north Zagros- west Alborz, and based on the output of the analyses performed, the superiority of the method in outlier detection was confirmed.

Heat exchangers are expedients used to transfer of heat energy from one medium to other while preventing the two from mixing. Heat exchangers are hired in a selection of domestic, commercial, and industrial applications, such as air conditioning, power production, refrigeration, process industry, and manufacturing industry. A larger variation in the temperature affects property of non-Newtonian fluids, thus affects the uniform heat transfer process. This paper is primarily to deal the response analysis of different controllers that is regulating the outlet temperature of fluid inside the PHE is proposed. The cold fluid flow is highly influenced on the hot water temperature, the research work mainly focuses on controlling cold fluid flow. In order to attain the precise model of the system, a mathematical model by first principle methods and process history based methods were used. A PI, PID and Model Predictive Controller (MPC) are known for its better control action of the system. The controller’s performance analysed such as PI, PID and MPC controller to adjust the temperature of hot fluid outlet to a required setpoint. Performance criteria analysis value for various controllers are PI, PID and MPC. It is understood that MPC controller results lower value of performance measures compared to other controllers, this implies that it could track the setpoint by rejecting the external disturbances with less settling time and moderate oscillation. The transient performance and the error criteria of the controllers are investigated and the best controller to regulate the temperature is found out. It is found out that the MPC controller outperforms PI and PID controller from the results of simulation.

The present paper is an extension of the work presented by authors at SDEWES2015 conference, where changes in the dynamic behaviour of shell-and-tube heat exchangers induced by fouling build-up were studied. A mathematical model was proposed of transient heat exchange with the influence of thermal resistance of fouling taken into account. Next a mathematical model of the entire heat exchanger was implemented and simulations were carried out using Simulink in MATLAB environment. The model was validated using data records acquired during 3 years of operation of a heat exchanger network coupled with a Crude Distillation Unit (CDU). Using the validated dynamic model, one can simulate the behaviour of a heat exchanger together with its PID control loop and investigate the influence of fouling build-up on specific indexes of quality of operation. A simulation study was carried out on selected PID-controlled exchangers operated in a heat exchanger network coupled with a CDU rated 220 kg/s crude oil. For different periods of uninterrupted exchanger operation, that is, at different values of the thermal resistance of fouling, tuning parameters were determined using the Ziegler-Nichols method. It was found that the optimal values of tuning parameters changed when the thermal resistance of fouling was changed. However, oscillations remained in the characteristics of some controlled exchangers indicating that the new values of tuning parameters determined by the simple Ziegler–Nichols method were ineffective. In the present work, to improve the quality indices of the exchangers control under fouling conditions, a more advanced method of the optimization of controller tuning is presented using Signal Constraint block in MATLAB/SIMULINK.

To improve the unit load following capacity of a large-capacity coal-fired supercritical (SC) power unit under automatic generation control (AGC), an intelligent model predictive optimal control (MPOC) scheme is proposed. It employs a nonlinear autoregressive moving average (NARMA) neural network model as the prediction model of the energy conversion unit and a more efficient simplified particle swarm optimization (PSO) algorithm as the optimal solution search approach. The effectiveness of the proposed MPOC scheme is evaluated by extensive control simulation tests on a commercial-grade simulator of a 600MW SC power unit. It is shown that the proposed approach can greatly improve the load following speed of the supercritical energy conversion unit, and at the same time keeps main steam pressure within safety limits.

Nowadays, simulation has become a cost-effective option for the evaluation of infrastructure improvements, on-road traffic management systems, and in vehicle driver support systems due to the fast evolution of computational modeling techniques. This paper presents a Locally Linear Neuro-Fuzzy (LLNF) model to simulate and predict the future behavior of a Driver-Vehicle-Unit (DVU). Local Linear Model Tree (LOLIMOT) learning algorithm is applied to train the model using real traffic data. This model was developed based on a new idea for estimating the instantaneous reaction of DVU, as an input of LLNF model. The model?s performance was evaluated based on real observed traffic data and also through comparisons with the results of LLNF models based on constant reaction delay. The results showed that LLNF model based on instantaneous reaction delay input outperformed the other car following models.

Modern Diesel engines with exhaust gas recirculation have achieved a significant progress in intake system, fuel consumption and emissions. So the process became more complex. Therefore, fault detection and diagnosis is difficult to be done and need to be improved. This contribution shows a system of fault detection and diagnosis methods for diesel engines based on physical model and data-driven model. By applying physical dynamic process models, identification with local linear model tree (LOLIMOT), data-driven models and residuals are generated by parity equations. Measured data in fault-free operation is used to build data-driven models. Detectable deflections of these residuals lead to symptoms which are the basis for the detection of faults. In final applications look-up tables can be generated using data-driven models. Experiments with a diesel engine intake system on MATLAB have demonstrated the detection and diagnosis of faults is suitability for application with reasonable calculation effort.

The walking beam furnace (WBF) is one of the most prominent process plants often met in an alloy steel production factory and characterized by high non-linearity, strong coupling, time delay, large time-constant and time variation in its parameter set and structure. From another viewpoint, the WBF is a distributed-parameter process in which the distribution of temperature is not uniform. Hence, this process plant has complicated non-linear dynamic equations that have not worked out yet. In this paper, we propose one-step non-linear predictive model for a real WBF using non-linear black-box sub-system identification based on locally linear neuro-fuzzy (LLNF) model. Furthermore, a multi-step predictive model with a precise long prediction horizon (i.e., ninety seconds ahead), developed with application of the sequential one-step predictive models, is also presented for the first time. The locally linear model tree (LOLIMOT) which is a progressive tree-based algorithm trains these models. Comparing the performance of the one-step LLNF predictive models with their associated models obtained through least squares error (LSE) solution proves that all operating zones of the WBF are of non-linear sub-systems. The recorded data from Iran Alloy Steel factory is utilized for identification and evaluation of the proposed neuro-fuzzy predictive models of the WBF process.

A resistive gas sensor (RGS) under temperature modulation regime is considered as a system for gas detection. Five target gases including Methanol, Ethanol, 2-Propanol, 1-Butanol, and Hydrogen each at 11 concentration levels, were selected for diagnosis using a single commercial gas sensor. For modulating the sensor, a staircase containing five voltage steps each with 20s plateau is applied to micro-heater of the sensor. This, in turn, alters both the temperature and the resistance profiles of the sensing layer which are considered as the input and the output of the defined system, respectively. In this way, five systems corresponding to five steps of the system input can be distinguished. Next, each system under the influence of the examined target gases is modeled with neuro-fuzzy network. Local linear model tree (LOLIMOT) used as learning algorithm of the systems and weights of the trained networks utilized as the features of the sensor in presence of target gas. Mapping these feature vectors using linear discriminant analysis showed successful classification of all target gases.

This study proposes an optimized hybrid visual servoing approach to overcome the imperfections of classical two-dimensional, three-dimensional and hybrid visual servoing methods. These imperfections are mostly convergence issues, non-optimized trajectories, expensive calculations and singularities. The proposed method provides more efficient optimized trajectories with shorter camera path for the robot than image-based and classical hybrid visual servoing methods. Moreover, it is less likely to lose the object from the camera field of view, and it is more robust to camera calibration than the classical position-based and hybrid visual servoing methods. The drawbacks in two-dimensional visual servoing are mostly related to the camera retreat and rotational motions. To tackle these drawbacks, rotations and translations in Z-axis have been separately controlled from three-dimensional estimation of the visual features. The pseudo-inverse of the proposed interaction matrix is approximated by a neuro-fuzzy neural network called local linear model tree. Using local linear model tree, the controller avoids the singularities and ill-conditioning of the proposed interaction matrix and makes it robust to image noises and camera parameters. The proposed method has been compared with classical image-based, position-based and hybrid visual servoing methods, both in simulation and in the real world using a 7-degree-of-freedom arm robot.

In this paper, modelling the torque of a variable valve timing (VVT) spark ignition (SI) engine is studied using multilayer Perceptron (MLP), radial basis function (RBF) and local linear model tree (LOLIMOT) networks. Real data collection, model order selection, model test are discussed. The obtained model is finally tested in real time on-line mode by an engine test bench. The model output is compared with the process output and compared among different network models. Selected model performs well and can be used in the model-based engine control.