A recurrent self-evolving fuzzy neural network predictive control for microwave drying process

A recurrent self-evolving fuzzy neural network (RSEFNN) predictive control scheme is developed for microwave drying process in this paper. During microwave drying process, the temperature, power absorption efficiency, and moisture variation characteristic in the drying material cannot be exactly known for the complex application environment. So a RSEFNN is constructed to predict the microwave drying process. Based on the RSEFNN, to achieve a highly efficient and safe microwave drying process, a multiple objectives predictive control algorithm is constructed to get a suitable input power over a prediction horizon. To identify the feasibility of the proposed recurrent self-evolving fuzzy neural network predictive control (RSEFNNPC) algorithm, a simulation of Red Maple and an actual application of lignite drying were analyzed in this paper. In the Red Maple drying process, temperature and moisture content are chosen as control objectives. As the simulation results show, the RSEFNNPC could achieve multiple objectives optimization. In the actual lignite drying process, the difference between lignite temperature and presupposed temperature was below 2?K. The difference between RSEFNN prediction and actual sampling temperature was below 1?K.     

HYBRID NEURAL APPROACHES FOR MODELLING DRYING PROCESSES FOR PARTICULATE SOLIDS

ABSTRACT

This work presents methods for synthesizing drying process models for particulate solids that combine prior knowledge with artificial neural networks. The inclusion of prior knowledge is investigated by developing two applications with the data from two indirect rotary steam dryers. The first application consisted in the modelling of the drying process of soya meal in a batch indirect rotary dryer, The external and internal mass transfer resistances were associated in the hidden layer of the network to linear and sigmoidal nodes, respectively. The second application consisted in the modelling of the drying process of soya meal in a continuos indirect rotary dryer. The model was constructed using the Semi-parametric Design Approach. The model predicts the evolution of solid moisture content and temperature as a function of the solid position in the dryer. The results show that the hybrid model performs better than the pure “ black box” neural network and default models. They also shows that prior knowledge enhances the extrapolation capabilities of a neural network model,     

HYBRID NEURAL APPROACHES FOR MODELLING DRYING PROCESSES FOR PARTICULATE SOLIDS

This work presents methods for synthesizing drying process models for particulate solids that combine prior knowledge with artificial neural networks. The inclusion of prior knowledge is investigated by developing two applications with the data from two indirect rotary steam dryers. The first application consisted in the modelling of the drying process of soya meal in a batch indirect rotary dryer, The external and internal mass transfer resistances were associated in the hidden layer of the network to linear and sigmoidal nodes, respectively. The second application consisted in the modelling of the drying process of soya meal in a continuos indirect rotary dryer. The model was constructed using the Semi-parametric Design Approach. The model predicts the evolution of solid moisture content and temperature as a function of the solid position in the dryer. The results show that the hybrid model performs better than the pure “ black box” neural network and default models. They also shows that prior knowledge enhances the extrapolation capabilities of a neural network model,     

Differential recurrent neural network based predictive control

In this paper an efficient algorithm to train general differential recurrent neural network (DRNN) is developed. The trained network can be directly used in the nonlinear model predictive control (NMPC) context. The neural network is represented in a general nonlinear state-space form and used to predict the future dynamic behavior of the nonlinear process in real time. In the new training algorithms, the ODEs of the model and the dynamic sensitivity are solved simultaneously using Taylor series expansion and automatic differentiation (AD) techniques. The same approach is also used to solve the online optimization problem in the predictive controller. The efficiency and effectiveness of the DRNN training algorithm and the NMPC approach are demonstrated through a two-CSTR case study. A good model fitting for the nonlinear plant at different sampling rates is obtained using the new method. A comparison with other approaches shows that the new algorithm can considerably reduce network training time and improve solution accuracy. The DRNN based NMPC approach results in good control performance under different operating conditions.     

NMPC of an industrial crystallization process using model-based observers

This paper illustrates the benefits of a nonlinear model-based predictive control (NMPC) approach applied to an industrial crystallization process. This relevant approach proposes a setpoint tracking of the crystal mass. The controlled variable, unavailable, is obtained using an extended Luenberger observer. A neural network model is used as internal model to predict process outputs. An optimization problem is solved to compute future control actions taking into account real-time control objectives. The performances of this strategy are demonstrated via simulation in cases of setpoint tracking and disturbance rejection. The results reveal a significant improvement in terms of robustness and energy efficiency.     

基于LSTM神经网络的流化床干燥器内生物质颗粒湿度预测

生物质作为一种储量丰富、环境友好且易于获取的可再生能源,日渐成为能源研究利用领域的热点。生物质湿度是影响生物质利用效率的关键因素,因此干燥是生物质利用之前的必要步骤。流化床由于其良好的传热传质特性,在干燥过程中得到了广泛的应用。为了实时监测生物质颗粒的干燥过程,利用弧形静电传感器阵列,结合用于时间序列建模的长短期记忆（LSTM）神经网络,实现了流化床干燥器内生物质颗粒湿度的预测。在实验室规模的流化床干燥器上进行了多工况实验获取训练和测试数据,通过模型参数优化确定了LSTM模型。通过与标准循环神经网络（RNN）模型的预测结果的对比表明,LSTM神经网络模型的平均相对误差较小,能够较为准确地预测流化床干燥器内生物质颗粒的湿度。     

Dynamics And Assessment of Some Control Strategies of a Simulated Industrial Rotary Dryer

Rotary dryers are widely used for the continuous drying of minerals and chemicals on a large scale. Hot gases are passed parallel to the flowing solid to achieve the desired product moisture content. Because these dryers are energy intensive, it is mandatory to operate them as efficiently as possible to respond to economic pressures. Using a dynamic rotary dryer simulator for mineral concentrate, five control strategies are evaluated and compared. Two control strategies are based on PI controllers and the others use neural network models. Results clearly show that a feedforward action, in conjunction with a PI controller or incorporated within the structure of a neural network model, led to the best performances provided an accurate measurement of the feed moisture content is available.     

On-Farm Drying and Storage Systems

ABSTRACT

Rotary dryers are widely used for the continuous drying of minerals and chemicals on a large scale. Hot gases are passed parallel to the flowing solid to achieve the desired product moisture content. Because these dryers are energy intensive, it is mandatory to operate them as efficiently as possible to respond to economic pressures. Using a dynamic rotary dryer simulator for mineral concentrate, five control strategies are evaluated and compared. Two control strategies are based on PI controllers and the others use neural network models. Results clearly show that a feedforward action, in conjunction with a PI controller or incorporated within the structure of a neural network model, led to the best performances provided an accurate measurement of the feed moisture content is available.     

Off-Line Determination of the Optimal Command Policies for a Sludge Drying Application

The main objective of this work is to develop, using a predictive control method, an off-line determination of the operating parameters for a sludge drying stage. At each time step, two operating parameters are identified by simultaneously minimizing three objective functions over a finite horizon. A laboratory dryer is briefly presented and used, in order to evaluate the suitability of the direct model employed to simulate sludge drying. Surface temperature, drying kinetics, and evaporated mass flux obtained from experimental measurements are compared to numerical simulations. Afterward, the optimization procedure is carried out and the results are discussed.     

Non‐Linear Model Predictive Control: A Personal Retrospective

An overview of non‐linear model predictive control (NMPC) is presented, with an extreme bias towards the author's experiences and published results. Challenges include multiple solutions (from non‐convex optimization problems), and divergence of the model and plant outputs when the constant additive output disturbance (the approach of dynamic matrix control, DMC) is used. Experiences with the use of fundamental models, multiple linear models (MMPC), and neural networks are reviewed. Ongoing work in unmeasured disturbance estimation, prediction and rejection is also discussed.     

NEURAL NETWORK-BASED PREDICTIVE CONTROL FOR MULTIVARIABLE PROCESSES

In this paper, the systematic derivations of setting up a nonlinear model predictive control based on the neural network are presented. This extends our previous work (Chen, 1998) into a multivariable system to explore the characteristics of the design. There are two stages for the development of nonlinear neural network predictive control: a neural network model and a control design. In the neural network model design, a parallel multiple-input, single-output neural network autoregressive with a model of exogenous inputs (NNARX) is proposed for multistep ahead predictions. In control design, the controller with extended control horizon is developed. The Levenberg-Marquardt algorithm is applied to training the NNARX model. Also, the sequential quadratic programming is used to search for the optimal manipulated inputs. The gradient of the objective function and constraints that require computation of Jacobian matrices are completely derived for optimization calculation. To demonstrate the control ability of MIMO cases, the proposed method is applied through two nonlinear simulation problems.     

Neural network-based predictive control for multivariable processes

In this paper, the systematic derivations of setting up a nonlinear model predictive control based on the neural network are presented. This extends our previous work (Chen, 1998) into a multivariable system to explore the characteristics of the design. There are two stages for the development of nonlinear neural network predictive control: a neural network model and a control design. In the neural network model design, a parallel multiple-input, single-output neural network autoregressive with a model of exogenous inputs (NNARX) is proposed for multistep ahead predictions. In control design, the controller with extended control horizon is developed. The Levenberg-Marquardt algorithm is applied to training the NNARX model. Also, the sequential quadratic programming is used to search for the optimal manipulated inputs. The gradient of the objective function and constraints that require computation of Jacobian matrices are completely derived for optimization calculation. To demonstrate the control ability of MIMO cases, the proposed method is applied through two nonlinear simulation problems.     

Optimization-based strategies for the operation of low-density polyethylene tubular reactors: nonlinear model predictive control

In this work, we present a general nonlinear model predictive control (NMPC) framework for low-density polyethylene (LDPE) tubular reactors. The framework is based on a first-principles dynamic model able to capture complex phenomena arising in these units. We first demonstrate the potential of using NMPC to simultaneously regulate and optimize the process economics in the presence of persistent disturbances such as fouling. We then couple the NMPC controller with a compatible moving horizon estimator (MHE) to provide output feedback. Finally, we discuss computational limitations arising in this framework and make use of recently proposed advanced-step MHE and NMPC strategies to provide nearly instantaneous feedback.     

An Investigation of Drying Process of Shelled Pistachios in a Newly Designed Fixed Bed Dryer System by Using Artificial Neural Network

In this paper, the drying of Siirt pistachios (SSPs) in a newly designed fixed bed dryer system and the prediction of drying characteristics using artificial neural network (ANN) are presented. Drying characteristics of SSPs with initial moisture content (MC) of 42.3% dry basis (db) were studied at different air temperatures (60, 80, and 100 °C) and air velocities (0.065, 0.1, and 0.13 m/s) in a newly designed fixed bed dryer system. Obtained results of experiments were used for ANN modeling and compared with experimental data. Falling rate period was observed during all the drying experiments; constant rate period was not observed. Furthermore, in the presented study, the application of ANN for predicting the drying time (DT) for a good quality product (output parameter for ANN modeling) was investigated. In order to train the ANN, experimental measurements were used as training data and test data. The back propagation learning algorithm with two different variants, so-called Levenberg–Marguardt (LM) and scaled conjugate gradient (SCG), and tangent sigmoid transfer function were used in the network so that the best approach can be determined. The most suitable algorithm and neuron number in the hidden layer are found out as LM with 15 neurons. For this number level, after the training, it is found that Root-mean squared (RMS) value is 0.3692, and absolute fraction of variance (R²) value is 99.99%. It is concluded that ANNs can be used for prediction of drying SSPs as an accurate method in similar systems.     

Prediction of Residence Time Distribution of Solid Wastes in a Rotary Dryer

The residence time distribution (RTD) of heterogeneous citrus waste particles was determined in a semi-pilot-scale rotary dryer with concurrent flow under several operational conditions. The experimental methodology was based on the stimulus-response technique, which consisted of injecting pulse-like tracers in the dryer feed stream. Measurements of RTD were performed to build up experimental curves that were numerically integrated to provide the mean residence time. A perfectly-stirred-tank in series model and a neural network model were derived. In addition, empirical and semi-empirical correlations from the literature were used to estimate residence time and the influence of operating conditions on this variable was investigated.     

Considerations on nonlinear model predictive control techniques

The nonlinear model predictive control (NMPC) is an on-line application based on nonlinear convolution models. It is an appealing control methodology, but it is difficult to implement and its solution is not so performing since it unavoidably means to solve a usually large-scale, constrained, and multidimensional optimization. To increase the difficulty, this optimization problem is subject to computationally heavy differential and algebraic constraints constituting the same convolution model and the least squares nature of the objective function easily leads to narrow valleys and multimodality issues.Beyond a short review of the state-of-the-art, the paper is aimed at highlighting the possibility to exploit at best the intrinsic features of the specific system one is going to control using the NMPC. The idea is to give the NMPC the possibility to automatically select the best combination of algorithms (differential solvers and optimizers) in accordance with the specific problem to be solved. From this perspective, the NMPC could be easily extended to many scientific fields traditionally far from process systems and computer-aided process engineering and the user has not to worry about which specific differential solvers and optimizers are needed to solve his/her problem.     

On‐Line PEMFC Control Using Parameterized Nonlinear Model‐Based Predictive Control

In this work, a fast nonlinear model‐based predictive control (NMPC) strategy is designed and experimentally validated on‐line on a real fuel cell. Regarding NMPC strategies, the most challenging part remains to achieve on‐line implementation, especially when dealing with fast dynamic systems. As previously demonstrated in a recent work, the proposed control strategy is ideally suited to address this problem. Indeed, it is 30 times faster than classical NMPC controllers. This strategy relies on a specific parameterization of the control actions to reduce the computational time and achieve on‐line implementation. Due to its short computational time compared to mechanistic models, an artificial neural network model is designed and experimentally validated. This model is employed as internal model in the NMPC controller to predict the system behavior. To confirm the applicability and the relevance of the proposed NMPC controller varying control scenarios are investigated on a test bench. The built‐in controller is overridden and the NMPC controller is implemented externally and executed on‐line. Experimental results exhibit the outstanding tracking capability and robustness against model‐process mismatch of the proposed strategy. The parameterized NMPC controller turns out to be an excellent candidate for on‐line applications.     

Mathematical Modeling the Drying of Poplar Wood Particles in a Closed-Loop Triple Pass Rotary Dryer

Closed-loop drying systems are an attractive alternative to conventional drying systems because they provide a wide range of potential advantages. Consequently, type of drying process is attracting increased interest. Rotary drying of wood particles can be assumed as an incorporated process involving fluid–solid interactions and simultaneous heat and mass transfer within and between the particles. Understanding these mechanisms during rotary drying processes may result in determination of the optimum drying parameters and improved dryer design. In this study, due to the complexity and nonlinearity of the momentum, heat, and mass transfer equations, a computerized mathematical model of a closed-loop triple-pass concurrent rotary dryer was developed to simulate the drying behavior of poplar wood particles within the dryer drums. Wood particle moisture content and temperature, drying air temperature, and drying air humidity ratio along the drums lengths can be simulated using this model. The model presented in this work has been shown to successfully predict the steady-state behavior of a concurrent rotary dryer and can be used to analyze the effects of various drying process parameters on the performance of the closed-loop triple-pass rotary dryer to determine the optimum drying parameters. The model was also used to simulate the performance of industrial closed-loop rotary dryers under various operating conditions.     

COMPARISON AMONG PI,DMC, QDMC,AND GPC ALGORITHMS IN THE CONTROL OF THE SPOUTED BED DRYER

This article presents comparative analysis between the classical PI (proportional-integral control) and MPC (model predictive control) techniques for a drying process on spouted beds. The on-line experimental setups were carried out in a laboratory-scale plant of a spouted bed dryer. The main objective was to optimize the plant operation by searching for the best control structure to be used in future scale enlargement. The major drawbacks encountered in this kind of system were high interactivity among variables, a malfunction as a result of calculated variables out of the operational window, and modeling mismatch. Despite the robustness of the operational PI, the control actions of this strategy did not overcome the variable interactions. The DMC (dynamic matrix control) and the QDMC (quadratic dynamic matrix control) algorithms performed satisfactorily over the major drawbacks. Special attention was given to the latter algorithm due to its ability to hold the variables under constrained oscillations. However, the best results were found for the adaptive GPC (generalized predictive control) algorithm whose actions prevailed over the modeling mismatch due to the strong nonlinear behavior intrinsic to the process. The main goal of the present work is to describe a procedure that can be standardized for other types of dryers and different scales. This is especially the case for the adaptive GPC, whose control structure is independent of the dryer nature and scale and whose implementation does not require previous identification procedures (self-tuning) and/or structural changes.     

COMPARISON AMONG PI, DMC, QDMC, AND GPC ALGORITHMS IN THE CONTROL OF THE SPOUTED BED DRYER

This article presents comparative analysis between the classical PI (proportional-integral control) and MPC (model predictive control) techniques for a drying process on spouted beds. The on-line experimental setups were carried out in a laboratory-scale plant of a spouted bed dryer. The main objective was to optimize the plant operation by searching for the best control structure to be used in future scale enlargement. The major drawbacks encountered in this kind of system were high interactivity among variables, a malfunction as a result of calculated variables out of the operational window, and modeling mismatch. Despite the robustness of the operational PI, the control actions of this strategy did not overcome the variable interactions. The DMC (dynamic matrix control) and the QDMC (quadratic dynamic matrix control) algorithms performed satisfactorily over the major drawbacks. Special attention was given to the latter algorithm due to its ability to hold the variables under constrained oscillations. However, the best results were found for the adaptive GPC (generalized predictive control) algorithm whose actions prevailed over the modeling mismatch due to the strong nonlinear behavior intrinsic to the process. The main goal of the present work is to describe a procedure that can be standardized for other types of dryers and different scales. This is especially the case for the adaptive GPC, whose control structure is independent of the dryer nature and scale and whose implementation does not require previous identification procedures (self-tuning) and/or structural changes.