Wenguo Xiang

Nonlinear identification of alstom gasifier based on wiener model

In this work a nonlinear identification approach has been developed and implemented on Alstom gasifier with Wiener model. The linear element of the Wiener model is identified by a combined subspace state space method, which integrates MOESP (Multivariable Output-Error State Space) and N4SID (Numerical algorithms for subspace state space system identification) method in the estimation of system matrices. A single layer neural network is chosen as the nonlinearity of the model. The quadruple system matrices are identified firstly according to the given input-output sample data. Then an initial approximation of the static nonlinear part is determined with the output sequence of linear part. At last, all parameters of the wiener model are optimized by Levenberg-Marquardt algorithm, using the model parameters obtained formerly as the initial estimates. A nonlinear model of the plant at 0% load is adopted as a base model for estimation because it is the most difficult case to control among three operating conditions. The proposed model identification method was used to model Alstom gasifier with strong nonlinearity and multivariable couples, compared to a combined linear subspace identification method. The results demonstrate that the nonlinear identification proposed, which may be applied to nonlinear predictive control, behave better approximation than the linear method.

Nonlinear system identification with modified differential evolution and RBF networks

In this paper, a new control parameter adaptation scheme is introduced into the classical differential evolution (DE) algorithm. Then, a method for nonlinear system identification is proposed. The method combines modified differential evolution (MDE) and radial basis function (RBF) neural networks, which can auto-configure the structure of RBF networks and obtain the model parameters. The RBF network structure and parameters could be determined simultaneously based on input-output data without a priori knowledge. Finally, an example of nonlinear function identification is given to illustrate the effectiveness of the proposed approach.

Dynamic modeling and control of the air separation unit in an IGCC power plant

The technology of Integrated Gasification Combined Cycle (IGCC) power plant is one of the most promising energy producing and coal utilization methods. It offers the opportunity to use fossil energy sources as well as satisfy the strict environmental constraints. A typical IGCC power plant mainly consists of an air separation unit, the gasification system, the gas clean-up process, and the combined-cycle unit. This paper closely investigates the dynamic characteristics and control strategy of the air separation unit. The rigorous first principles mathematic model of the cryogenic rectification column, which is the core of the air separation, is firstly developed. Moreover, with several step changes introduced, dynamic simulations are carried out to illustrate the transient behavior of the column. Considering it a multi-input, multi-out and energy intensive system, the control scheme according to the character of the column is presented, and the performance of the control strategy is simulated.

Gaisfier Following-Based Coordinated Control for the IGCC Power Plant

This paper proposes a gasifier following-based coordinated control strategy for the IGCC power plant. The gasifier and the combined-cycle model are integrated to form the IGCC power plant model and a load and pressure controller is designed to reduce pressure variations with load changes. Simulation results show the good load-tracking property and satisfactory control performance of the proposed controller.

On-line identification of thermal process using a modified ts-type neuro-fuzzy system

In this paper, a modified TS-type neuro-fuzzy system (MTSNFS) for on-line identification is proposed, which possesses six layers of neurons to perform the fuzzy inference. A modified self-organizing competitive learning algorithm with capabilities of dynamical rules recruitment and cancellation is proposed for structure identification. A hybrid learning algorithm combining recursive least squares (RLS) estimation and ordered derivative learning is used for parameter estimation. Both the structure and parameters could be automatically determined online without a priori knowledge. Comparisons with other related works are made via identification of Box-Jenkins furnance. Identification of bed temperature of a circulating fluidized bed boiler using the MTSNFS is also presented in this paper. The results demonstrate that the proposed identification approach is of high accuracy and compactness, and suitable for on-line modeling and prediction.

Online coal quality analyzer-based decentralized PID control for the ALSTOM gasifier

This paper focuses on the design of decentralized PID control scheme for the ALSTOM gasifier based on online coal quality analyzing technique. Taking into account characteristics changing with coal quality variations during gasifier operation, corresponding control strategies were added into the decentralized PID control scheme. Simulation results show good control performance for the ALSTOM gasifier benchmark test with fine coal quality signal. The proposed control scheme adopted the traditional PID control algorithm which could be easily implemented. Thus, provides good guidelines for practical control system design for gasifiers that are subject to coal quality variations.

Development of a simplified method for the determination of ampere-hour capacity of lead–acid battery

Selection of the optimum level of current for charging and discharging operations is an important factor for the performance of lead–acid batteries in PV application. Realizing this situation, an experimental study was carried out to determine the performance of battery under different charging and discharging current. The battery was charged at an input current of 6, 12, and 18 A, whereas under these input charging currents the battery was discharged at constant loads of 5.7, 11.4, and 17.1 A. Then algebraic equations for the determination of battery ampere-hour capacity, in relation with state of charge, were formulated with the help of MATLAB software. The proposed model provides the battery output directly without going through the calculation of constant unknowns and battery ampere-hour capacity. It is different from previous models, which only evaluate the battery ampere-hour capacities with the help of already calculated battery ampere-hour capacities and other battery parameters from the same battery. During the study, it was found that the rate of charge and discharge affected the duration of charge and discharge as well as battery ampere-hour capacity. When the rate is increased, the number of ampere hours was decreased along with the battery operational time duration. The developed model equations were validated with ours as well as other researcher’s measured values; the corresponding values were much closer when the values of state of charge were at the range of 100–90%. This study may be useful to understand the energy storage within a PV system and to select the optimum level of current, which consequently lengthens the life of the battery and improves the overall performance of a Photovoltaic panel (PV) system.

Reduction Behavior of Iron Oxide for Chemical-Looping Hydrogen Generation in a Compact Fluidized Fuel Reactor

Chemical-looping hydrogen generation (CLHG) integrates chemical-looping combustion (CLC) and the steam-iron process. It is a process for hydrogen production with inherent CO2 separation. CLHG includes three reactors: a fuel reactor, a steam reactor and an air reactor, with iron oxide as oxygen carrier. This paper presents a compact fluidized fuel reactor for CLHG to produce reductive FeO with CO2 sequestration. An iron ore as oxygen carrier was tested, and CO and syngas were used as fuels. The results showed that through this compact fuel reactor, reductive FeO for further hydrogen generation was obtained and a high concentration CO2 was separated at the outlet of the fuel reactor. The influence of riser temperature, bubble fluidized bed temperature and Fe2O3/CO ratio on the bed performance was investigated. It revealed that the bubble fluidized bed temperature and Fe2O3/CO ratio had a significant impact on the fuel conversion while the effect of riser temperature was marginal. The iron ore exhibited good reactivity and no agglomeration was found in the experiment.Copyright

Hydrodynamic Analysis of a Three-Fluidized Bed Reactor Cold Flow Model for Chemical Looping Hydrogen Generation: Pressure Characteristics

Chemical looping hydrogen generation (CLHG) can produce pure hydrogen with inherent separation of CO2 from fossils fuel. The process involves a metal oxide, as an oxygen carrier, such as iron oxide. The CLHG system consists of three reactors: a fuel reactor (FR), a steam reactor (SR) and an air reactor (AR). In the FR, the fuel gases react with iron oxides (hematite Fe2O3, magnetite Fe3O4, wustite FeO), generating reduced iron oxides (FeO or even Fe), and with full conversion of gaseous fuels, pure CO2 can be obtained after cooling the flue gas from the fuel reactor; in the SR, FeO and Fe reacts with steam to generate magnetite (Fe3O4) and H2, the latter representing the final target product of the process; in the AR, the magnetite is oxidized back to hematite which is used in another cycle.

A new neuro-fuzzy approach for nonlinear system identification based on differential evolution

In this paper, a new neuro-fuzzy approach for complex dynamical systems identification is proposed. The approach combines the merits of fuzzy logic theory, radial basis function neural networks, and differential evolution algorithm. The structure of the proposed algorithm model is a four-layer radial basis function fuzzy neural network (RBFFNN). The differential evolution algorithm is used for network optimization. A parameter called contribution factor is introduced to find out unimportant rules, and delete them. Both the fuzzy network structure and parameter learning can be performed automatically from input-output samples without a priori knowledge. Finally, examples of thermal processes identification are given to illustrate the effectiveness of the proposed approach.