Online adaptive least squares support vector machine and its application in utility boiler combustion optimization systems

Boiler combustion optimization is a key measure to improve the energy efficiency and reduce pollutants emissions of power units. However, time-variability of boiler combustion systems and lack of adaptive regression models pose great challenges for the application of the boiler combustion optimization technique. A recent approach to address these issues is to use the least squares support vector machine (LS-SVM), a computationally attractive machine learning technique with rather legible training processes and topologic structures, to model boiler combustion systems. In this paper, we propose an adaptive algorithm for the LS-SVM model, namely adaptive least squares support vector machine (ALS-SVM), with the aim of developing an adaptive boiler combustion model. The fundamental mechanism of the proposed algorithm is firstly introduced, followed by a detailed discussion on key functional components of the algorithm, including online updating of model parameters. A case study using a time-varying nonlinear function is then provided for model validation purposes, where model results illustrate that adaptive LS-SVM models can fit variable characteristics accurately after being updated with the ALS-SVM method. Based on the introduction to the proposed algorithm and the case study, a discussion is then delivered on the potential of applying the proposed ALS-SVM method in a boiler combustion optimization system, and a real-life fossil fuel power plant is taken as an instance to demonstrate its feasibility. Results show that the proposed adaptive model with the ALS-SVM method is able to track the time-varying characteristics of a boiler combustion system.     

Optimization and control of a stand-alone hybrid solid oxide fuel cells/gas turbine system coupled with dry reforming of methane

This paper presents the hybrid solid oxide fuel cells (SOFC)/gas turbine (GT) system coupled with dry reforming of methane (DRM). The DRM is a syngas producer by consuming greenhouse gas. The stand-alone (off-the-grid) power system is developed by using a combination of a post-burner, recuperators and pressurized recycles in place of external energy supplies. To address the stand-alone operation and meet the complete combustion condition for the burner, the optimal operating conditions are initially determined by solving a constrained optimization algorithm for maximizing the hybrid power efficiency, and the dynamic control loops are implemented in a plantwide environment. In the proposed plantwide control strategy, the inventory control framework is added to regulate the plant component inventory, an air/fuel cross-limiting combustion control is added to ensure complete combustion and reduce heat loss, and the power and CO₂ emission control configuration is added to achieve the quality control performance. Finally, the simulation shows that the IMC-based multi-loop control scheme can efficiently regulate the total system power and control CO₂ emissions per kWh of electricity as well.     

一种时空协同的图卷积长短期记忆网络及其工业软测量应用

软测量技术的发展有效解决了工业过程中对于难以直接测量的质量变量的感知困难,为过程的控制与优化提供了有力保障.通常在含有多个质量变量的过程中,样本间的时序关系和多个质量变量间相互影响的空间关系能够反映过程本身的特性,这种时空特性的挖掘有益于软测量模型性能的提升,而传统软测量方法往往局限于对时序关系的学习而并未考虑对质量变量间的空间关系进行有效利用.对此,提出一种时空协同的图卷积长短期记忆网络(graph convolution long short-term memory networks, GC-LSTM),并应用于工业软测量场景.采用多通道网络结构将图卷积网络的空间关系挖掘能力与长短期记忆网络的时序关系学习能力相结合,对过程进行时空协同学习以实现软测量应用.具体而言,每条通道用于对每种质量变量进行独立学习;对于过程的时序特性,利用各通道内的长短期记忆网络提取针对不同质量变量的时序特征;对于过程的空间特性,构建质量变量间空间关系的图结构,采用跨通道的图卷积运算将不同通道内不同质量变量的时序特征基于空间关系进行融合,得到兼具过程时空特性的特征,从而在软测量建模中实现过程时空协同学习与融合...     

Explicit stochastic predictive control of combustion plants based on Gaussian process models

Energy production is one of the largest sources of air pollution. A feasible method to reduce the harmful flue gases emissions and to increase the efficiency is to improve the control strategies of the existing thermoelectric power plants. This makes the Nonlinear Model Predictive Control (NMPC) method very suitable for achieving an efficient combustion control. Recently, an explicit approximate approach for stochastic NMPC based on a Gaussian process model was proposed. The benefits of an explicit solution, in addition to the efficient on-line computations, include also verifiability of the implementation, which is an essential issue in safety-critical applications. This paper considers the application of an explicit approximate approach for stochastic NMPC to the design of an explicit reference tracking NMPC controller for a combustion plant based on its Gaussian process model. The controller brings the air factor (respectively the concentration of oxygen in the flue gases) on its optimal value with every change of the load factor and thus an optimal operation of the combustion plant is achieved.     

On-line soft sensor for polyethylene process with multiple production grades

Since online measurement of the melt index (MI) of polyethylene is difficult, a virtual sensor model is desirable. However, a polyethylene process usually produces products with multiple grades. The relation between process and quality variables is highly nonlinear. Besides, a virtual sensor model in real plant process with many inputs has to deal with collinearity and time-varying issues. A new recursive algorithm, which models a multivariable, time-varying and nonlinear system, is presented. Principal component analysis (PCA) is used to eliminate the collinearity. Fuzzy c-means (FCM) and fuzzy Takagi–Sugeno (FTS) modeling are used to decompose the nonlinear system into several linear subsystems. Effectiveness of the model is demonstrated using real plant data from a polyethylene process.     

PEMFC发电系统的自适应模糊控制与动态分析

监测和控制燃料电池的过程中,需要获得各种实时数据.质子交换膜燃料电池(PEMFC)发电系统中的参数强耦合、高度非线性特性增加了对其控制的难度,传统的PI控制虽然对模型精确的系统有较好的控制效果,但对于参数波动的系统则无法获得较高的控制性能.针对以上情况,基于PEMFC发电系统的动态仿真模型,根据重整器在燃料电池发电系统中的作用,设计了自适应模糊控制器.利用模糊控制规则在线控制氢气摩尔流,从而控制PEMFC发电系统的输出功率.仿真结果表明,该动态模型能够预测输出电压.响应曲线显示出自适应模糊控制算法能够较好控制燃料电池有功和无功功率的输出.模型具有良好的负载跟踪特性.     

Robust data reconciliation of combustion variables in multi-fuel fired industrial boilers

This paper introduces an application of simultaneous nonlinear data reconciliation and gross error detection for power plants utilizing a complex but computationally light first principle combustion model. Element and energy balances and robust techniques introduce nonlinearity and the consequent optimization problem is solved using nonlinear optimization. Data reconciliation improves estimation of process variables and enables improved sensor quality control and identification of process anomalies. The approach was applied to an industrial 200 MW_th fluidized bed boiler combusting wood, peat, bark, and slurry. The results indicate that the approach is valid and is able to perform in various process conditions. As the combustion model is generic, the method is applicable in any boiler environment.     

Combined quasi-static backward modeling and look-ahead fuzzy control of vehicles

Vehicle modeling can play an important role in vehicle power train design, control and energy management investigation. This paper presents a method for vehicle power train modeling. The key feature of the method is its presentation of the dynamic of vehicle based on the road information. This ability makes the method suitable for look-ahead energy management and fuel economy optimal control problems. With the aid of a road slope database, road geometry ahead of the vehicle is extracted. A fuzzy controller is developed that receives this information and controls the velocity of the vehicle with respect to its fuel consumption. In order to maintain the operation of the combustion engine near its efficient region, the fuzzy controller commands a continuously variable transmission. Simulations are carried out using real road data. The results are presented and discussed.     

300MW燃煤锅炉低氮燃烧器改造研究

由于新环保标准《火电厂大气污染物排放标准》（GB13223-2011）的公布,国家对环保要求越来越严格,为更好地达到环保要求以及进一步降低脱硝运行成本,经过调研及针对电厂目前燃煤情况,对某电厂300MW燃煤锅炉进行了低氮燃烧器改造的必要性与可行性的研究分析,决定采用某锅炉厂新研发的多维深度分级燃烧系统,并实施了低氮燃烧器技术改造。 为了考察新型低氮燃烧系统改造后的运行状态,以及为电厂日常运行寻找最佳工况,依据国家以及行业相关标准进行多维深度分级燃烧系统（多维深度分级）的调试,进行制粉系统试验和燃烧优化调整试验,通过一系列的磨煤机出力特性试验、磨煤机分离器挡板特性试验、煤粉细度调整试验,还有习惯工况下变氧量试验、变高位燃尽风量试验、变高位燃尽风摆角试验、变氧试验、磨煤机投运组合试验、变低位燃尽风量试验、变夹心风量试验、二次风配风方式试验、负荷特性试验、空预器漏风测试等一系列试验,优化低氮燃烧器的燃烧调整,寻找较优运行工况,优化锅炉燃烧状况,更好地实现脱硝反应器前入口烟气中NOx含量低于300mg/Nm3的排放要求,延长昂贵的SCR反应器中金属催化剂的寿命,进一步降低SCR脱硝装置的运行成本,实现节能降耗,令锅炉燃烧更加地稳定、可靠与经济,防止出现水冷壁严重结焦和大屏严重挂焦问题,提高锅炉效率,达到了改造目标。     

基于多目标进化算法的锅炉热损失优化研究

针对锅炉热损失模型的特点,提出基于Pareto最优概念的多目标进化算法实现运行工况寻优,然后根据模糊集理论在Pareto解集中求得满意解,获得最佳的锅炉燃烧调整方式.通过某600MW锅炉热损失的优化研究,并与基于神经网络的寻优结果比较,数值计算表明支持向量机模型寻优结果在Pareto前沿具有更好的多样性,结果更优,可指导运行人员进行参数优化调整,提高燃烧经济性.     

Optimal sensor location in the presence of nonstationary noise

Observability conditions for distributed parameter systems with nonstationary state excitation noise are derived. Because these conditions are violated in most cases of interest in process control an optimal projection approach and a sensor location criterion are developed to minimize the error caused by the lack of observability. The value of these concepts was demonstrated in real time experiments on a system with two spatial dimensions, the heating of a cylindrical ingot. It was found that through the nonstationary noise model and the optimal sensor location significantly improved state estimates can be obtained when persistent or slowly varying unmeasured disturbances affect the system. The real-time requirements for the filter were very modest and an application in the process industries where these disturbances always present a serious problem, seems entirely feasible.     

Soft sensor based on stacked auto-encoder deep neural network for air preheater rotor deformation prediction

Soft sensors have been widely used in industrial processes over the past two decades because they use easy-to-measure process variables to predict difficult-to-measure ones. Some success has been achieved by the dominant traditional methods of modeling soft sensors based on statistics, such as principal components analysis (PCA) and partial least square (PLS), but such sensors usually become inaccurate and inefficient when processing strong nonlinear data. In this paper, a new soft sensor modeling approach is proposed based on a deep learning network. First, stacked auto-encoders (SAEs) are employed to extract high-level feature representations of the input data. In the process of training each layer of a SAE, the Limited-memory Broyden-Fletcher-Goldfarb-Shanno algorithm (L-BFGS) is adopted to optimize the weights parameters. Then, a support vector regression (SVR) is added to predict the target value on the basis of the features obtained from the SAE. To improve the model performance, Genetic Algorithm (GA) is used to obtain the optimal parameters of the SVR. To evaluate the proposed method, a soft sensor model for estimating the rotor deformation of air preheaters in a thermal power plant boiler is studied. The experimental results demonstrate that the soft sensor based on the SAE-SVR algorithm is more effective than the existing methods are.     

Optimal heat release shaping in a reactivity controlled compression ignition (RCCI) engine

The present paper addresses the optimal heat release (HR) law in a single cylinder engine operated under reactivity controlled compression ignition (RCCI) combustion mode to minimise the indicated specific fuel consumption (ISFC) subject to different constraints including pressure related limits (maximum cylinder pressure and maximum cylinder pressure gradient). With this aim, a 0-dimensional (0D) engine combustion model has been identified with experimental data. Then, the optimal control problem of minimising the ISFC of the engine at different operating conditions of the engine operating map has been stated and analytically solved. To evaluate the method viability a data-driven model is developed to obtain the control actions (gasoline fraction) leading to the calculated optimal HR, more precisely to the optimal ratio between premixed and diffusive combustion. The experimental results obtained with such controls and the differences with the optimal HR are finally explained and discussed.     

Decentralized estimation of regression coefficients in sensor networks

Consider a wireless sensor network with a fusion center deployed to estimate a common non-random parameter vector. Each sensor obtains a noisy observation vector of the non-random parameter vector according to a linear regression model. The observation noise is correlated across the sensors. Due to power, bandwidth and complexity limitations, each sensor linearly compresses its data. The compressed data from the sensors are transmitted to the fusion center, which linearly estimates the non-random parameter vector. The goal is to design the compression matrices at the sensors and the linear unbiased estimator at the fusion center such that the total variance of the estimation error is minimized. In this paper, we provide necessary and sufficient conditions for achieving the performance of the centralized best linear unbiased estimator. We also provide the optimal compression matrices and the optimal linear unbiased estimator when these conditions are satisfied. When these conditions are not satisfied, we propose a sub-optimal algorithm to determine the compression matrices and the linear unbiased estimator. Simulation results are provided to illustrate the effectiveness of the proposed algorithm.     

基于模型的虚拟温度传感器设计

针对航空发动机控制系统的半物理仿真过程中实时温度场难以模拟、接口电路的温度通道不能检验等问题,提出了一种基于模型的虚拟温度传感器设计思路。在Matlab/Simulink平台上建立传感器的变时间常数算法模型,使用RTW-EC的自动生成代码工具基于数字信号处理器(DSP)C2000微控制器设计了虚拟温度传感器,进一步结合CCS集成开发环境(IDE)对所生成的温度传感器代码进行了处理器在环(PIL)验证。仿真结果表明:所构建的虚拟传感器能够高精度模拟真实温度传感器的特性,可以满足航空发动机控制系统半物理仿真试验的要求。     

Extremum seeking-based optimal EGR set-point design for combustion engines in lean-burn mode

In lean combustion mode, exhaust gas ratio (EGR) is a significant factor that affects fuel economy and combustion stability. A proper EGR level is beneficial for the fuel economy; however, the combustion stability (coefficient of variation (COV) in indicated mean effective pressure (IMEP)) deteriorated monotonously with increasing EGR. The aim of this study is to achieve a trade-off between the fuel economy and combustion stability by optimizing the EGR set-point. A cost function (J) is designed to represent the trade-off and reduce the calibration burden for optimal EGR at different engine operating conditions. An extremum-seeking (ES) algorithm is adopted to search for the extreme value of J and obtain the optimal EGR at an operating point. Finally, a map of optimal EGR set-value is designed and experimentally validated on a real driving cycle.     

具有传感器增益退化、随机时延和丢包的分布式融合估计器

研究了具有传感器增益退化、模型不确定性、数据传输时延和丢包的网络化多传感器分布式融合估计问题,模型的不确定性描述为系统矩阵受到随机扰动,传感器增益退化现象通过统计特性已知的随机变量来描述,随机时延和丢包现象存在于局部最优状态估计向融合中心传输的过程中.首先,设计了一种局部最优无偏估计器,然后将传输时延描述为随机过程,并在融合中心端建立符合存储规则的时延-丢包模型,利用最优线性无偏估计方法,导出最小方差意义下的分布式融合估计器.最后,通过算例仿真证明所设计融合估计器的有效性.     

Real Time Relative and Absolute Dynamic Localization of Air–Ground Wireless Sensor Networks

A novel approach for relative and absolute localization of wireless sensor nodes using a potential field method is presented. The main idea of our work is to develop relative and absolute localization algorithms for the position estimate of stationary unattended ground sensor (UGS) nodes using a potential field method. A dynamical model is derived for each sensor node to estimate the relative and absolute position estimates under the influence of a certain fictitious virtual force. In the algorithm the sensor nodes do not move physically, but a virtual motion is carried out to generate optimal position estimates. The convergence of the estimator system to a least squares solution is guaranteed using Lyapunov theory. Separate control algorithms for relative and absolute localization are developed which guarantee the convergence of the position estimates. The relative localization algorithm assumes that distance (i.e. range) measurements between UGS nodes are available and for absolute localization algorithm, uninhabited aerial vehicles (UAV) are available with on board GPS such that they have absolute position information together with range measurement information. In the relative localization algorithm the UGS nodes are localized with respect to an internal co-ordinate frame. In absolute localization the UGS nodes are localized with respect to the known absolute position of UAV in the air–ground network. The effectiveness of the control algorithm is highlighted by the real time implementation results.     

Fuel blend fraction estimation for fuel-flexible combustion control: Uncertainty analysis

“Fuel-flexible combustion control” strategies require accurate fuel composition information. A physics-based fuel blend fraction estimator for lean-burn combustion is used to assess the impact of uncertain variables on the estimator accuracy. For biodiesel blends in a diesel engine, the strategy is shown to be minimally affected by biodiesel feedstock variations, and expected fuel flow and oxygen sensor errors. However, observed airflow estimate errors are expected to lead to large estimator errors. When applied to ethanol blends, the estimator error is lower due to a higher stoichiometric mixture fraction for ethanol, and inherent differences in typical lean-burn gasoline engine operating conditions.     

Numerical simulation of turbulent combustion of subsonic gas jet flows

A physical and mathematical model of turbulent combustion of subsonic gas fuel jet flows flowing into an air space is proposed. The processes are described by averaged equations of the boundary layer with a turbulent viscosity model and a combustion diffusion model. As turbulent viscosity models, the well-known two-parameter k-? standard and k-?? models are taken. The results of the averaged and pulsating flow characteristics?? comparison of numerical calculations with the experimental data are presented.