CN101038277B - Soft measurement method for coal power fineness in powdering producer - Google Patents

Soft measurement method for coal power fineness in powdering producer Download PDF

Info

Publication number
CN101038277B
CN101038277B CN 200710011021 CN200710011021A CN101038277B CN 101038277 B CN101038277 B CN 101038277B CN 200710011021 CN200710011021 CN 200710011021 CN 200710011021 A CN200710011021 A CN 200710011021A CN 101038277 B CN101038277 B CN 101038277B
Authority
CN
China
Prior art keywords
gt
lt
model
mi
soft
Prior art date
Application number
CN 200710011021
Other languages
Chinese (zh)
Other versions
CN101038277A (en
Inventor
岳恒
张君
张立岩
柴天佑
Original Assignee
东北大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 东北大学 filed Critical 东北大学
Priority to CN 200710011021 priority Critical patent/CN101038277B/en
Publication of CN101038277A publication Critical patent/CN101038277A/en
Application granted granted Critical
Publication of CN101038277B publication Critical patent/CN101038277B/en

Links

Abstract

A pulverized coal fineness soft-sensing method in a powder-making process based on a least squares support vector machine depends on a hardware platform, measuring instruments and a computer system used for performing a software calculation, wherein said software performs a soft-sensing through real-time process data obtained by a controlling computer of a data collector. Said method comprises the steps of selecting assistant variables, determining the structure of sample data, determining sample components, building a soft-sensing model based on the least squares support vector machine and performing the model training, parameter selection as well as the model validation and the model parameter emendation. The present invention is applied simply and easily, has a lower cost, has more accurate soft-sensing results, and helps to achieve an optimal control and an optimal operation of a pulverizing process.

Description

基于最小二乘_支持向量机的制粉过程煤粉细度软测量方 _ Based on least squares SVM during milling of soft coal fineness measurement side

law

技术领域 FIELD

[0001] 本发明属于自动化测量技术领域,特别涉及对制粉过程中用于煤粉研磨的由钢球磨煤机、粗粉分离器和布袋收集器组成的中储式钢球磨煤机制粉系统的最终产品一煤粉的细度指标进行软测量的方法。 [0001] The present invention belongs to the field of automation technology measurements, in particular, to the reservoir of formula steel - ball mill for grinding coal to a ball mill pulverizing process, coarse separator and collector bag composed of final product the method of a coal fineness index measuring soft.

[0002] 背景技术 [0002] BACKGROUND OF THE INVENTION

[0003] 制粉生产中,由钢球磨煤机、粗粉分离器和布袋收集器组成的中储式钢球磨煤机广泛用于将煤粉研磨至工艺要求的细度范围内。 [0003] Production of milling, ball mill in the reservoir by a ball-type mill, coarse separator and collector bag composed of widely used to grind coal to a fineness within the range of process requirements. 煤粉的细度是考核制粉过程的重要工艺指标,是提高产品质量和经济效益的重要因素,对于采用燃煤方式的发电、烧结等过程控制占有十分重要的地位。 Fineness of pulverized coal is an important process indicators for assessing milling process, is an important factor to improve product quality and economic efficiency, the use of coal for power generation mode, sintering and other process control plays a very important role. 目前,煤粉细度在线检测的工业应用中,由于实际条件的限制,没有进行工业性试验,没有实际的可操作性,没有成熟的检测装置,在实际生产过程中主要依靠采样、化验的方式离线获得。 Currently, industrial applications coal fineness line detection, due to the limitation of the actual condition, there is no industrial test, no practical operability, no sophisticated detection means, rely mainly on the sample in the actual production process, the way the assay offline obtained. 该方法的不足在于:1.人工操作时人为因素影响大,测量结果的客观性差;2.测量的时间间隔长,测量结果反馈的时间也长,因此得到的信息对操作人员缺乏指导意义;3.对煤粉取样时不可避免出现煤粉泄露现象,造成环境污染。 Disadvantage of this method is that: 1 large when manual human factors, the measurement results objectivity difference; 2 measured time interval is long, the measurement result feedback time is long, the information obtained by the lack of guidance for the operator; 3. inevitably pulverized coal leakage phenomenon occurs, causing environmental pollution when coal dust sampling. 由于样本数据数量有限和较大的干扰误差对训练的影响,采用常见的神经网络方法,模型的收敛速度慢,并且容易出现过拟合现象。 Due to the limited number of sample data and a greater impact of interference error with the training, the use of common neural network convergence velocity model is slow, and prone to over-fitting phenomenon.

[0004] 发明内容 [0004] SUMMARY OF THE INVENTION

[0005] 为了解决现有的制粉过程煤粉细度测量方法之不足,本发明提供一种基于最小二乘_支持向量机的煤粉细度软测量方法,通过常规在线测量仪表提供的辅助变量的测量参数,给出当前的煤粉细度的估计值,为制粉过程的优化操作和优化运行提供关键工艺指标。 [0005] In order to solve the problem of the method of measuring coal fineness conventional milling process, the present invention provides a method of least squares coal fineness _ soft measurement based on support vector machines, the auxiliary line provided by a conventional measuring device measuring the parameter variables, provides an estimate of the current pulverized coal fineness, key process indicators provide optimal operation and optimize the operation of the milling process.

[0006] 本发明方法由硬件平台及测量软件组成,其中硬件平台核心由钢球磨煤机、粗粉分离器、布袋收集器以及相关设备组成,同时配备了测量仪表以及进行软件计算的计算机系统。 Method [0006] The present invention is measured by the hardware platform and software components, wherein the hardware platform of the core ball mill, coarse separator, bag collector composition and associated equipment, measuring instruments, and equipped with a computer system calculated by the software. 其硬件的联接是钢球磨煤机的输入端与给煤机、热风和冷风管道相接,钢球磨煤机输出端与粗粉分离器连接,粗粉分离器的回粉管道与磨煤机输出端连接,出口与布袋除尘器相接,煤粉经布袋收集器实现风粉分离,煤粉由输送装置送入煤粉仓。 The hardware is an input terminal coupled to the ball mill coal feeder, hot air and the cold air duct contact, Ball Mill Classifier output terminal is connected to return conduit coarse dust separator and coal pulverizer output end connected to the bag filter outlet contact, realized through the coal dust bag collector air separation, coal fed from the coal storage conveying means. 详细结构如图1所示。 Detailed structure shown in Figure 1. 以钢球磨煤机、粗粉分离器、布袋收集器组成的制粉系统,其测量仪表包括: Ball milling system to mill, coarse separator, bag collector composition, which measuring device comprises:

[0007] 在热风管道上安装一个热电偶,用于在线测量热风温度Tkfto ; [0007] A thermocouple is mounted on the hot air conduit for online measurement of hot air temperature Tkfto;

[0008] 在磨煤机出口管道上安装一个热电偶,用于在线测量磨煤机出口煤粉温度Tckto ; [0008] A thermocouple mounted on the mill outlet conduit for online measurement of coal mill outlet temperature Tckto;

[0009] 在磨煤机和布袋收集器的给料管道、出料管道上分别安装四个压力计,用于在线测量磨煤机进出口压力PMm、PMi和布袋收集器进出口压力psm、Psm用于计算磨煤机和布袋除尘器差压Pmjcy、Psjqcy ; [0009] in the mill and the feed pipe collector bag, four discharge pipes are attached to a pressure gauge for measuring the mill outlet pressure line PMm, PMi and bag collector outlet pressure psm, Psm bag filter used to calculate the differential pressure and mill Pmjcy, Psjqcy;

[0010] 给煤机转速的反馈信号,用于估计给煤量Wauzs;如果有给煤量的测量信号,直接使用给煤量的检测值。 [0010] coal feeder speed to a feedback signal for estimating the amount of coal Wauzs; if a coal feed rate signal measured directly using the detected value to the amount of coal. 本说明书中用给煤机转速Wauzs估计给煤量。 This specification by the coal feeder speed to a coal Wauzs estimate.

[0011] 该磨矿系统同时配置了控制计算机(分布式计算机控制系统(DCS)、或可编程逻辑控制器(PLC)、或工业控制计算机(IPC))或数据采集器、用以采集测量仪表的信号。 [0011] The grinding configure the system control computer (distributed computer control system (the DCS), or a programmable logic controller (PLC), or industrial control computer (the IPC)) or a data collector for collecting measuring instrument signal of.

[0012] 本发明的软测量软件既可以运行在可编程逻辑控制器上、计算机控制系统的监控计算机上,也可以运行于独立的计算机上,该软件通过与控制计算机或数据采集器进行通讯,获得实时的过程数据,并给出煤粉细度的估计结果。 [0012] Soft measurement software of the present invention may be run on a programmable logic controller, a computer control system monitoring computer, may run on a separate computer, the software for communication control by a computer or data acquisition, real-time process data, and the results give an estimate of the fineness of pulverized coal.

[0013] 本发明方法包括以下步骤:(1)辅助变量的选择,(2)样本数据的结构,(3)样本的组成,⑷基于最小二乘-支持向量机软测量模型,(5)模型参数的校正。 [0013] The method of the present invention comprises the steps of: (1) selection of auxiliary variable structure (2) sample data, the composition (3) of the sample, based on least squares ⑷ - SVM model soft sensor, (5) Model correction parameters.

[0014] (1)辅助变量的选择,从影响煤粉细度的经验和理论分析煤粉细度的相关因素分析,本发明选择辅助变量都是通过仪表在线检测得到的,包括以下辅助变量: [0014] (1) selecting an auxiliary variable, from empirical and theoretical impact of coal fineness related factors fineness coal analysis, the present invention selects, comprising the auxiliary variable auxiliary variables are obtained by detecting the meter:

[0015] 热风温度Tkfto ; [0015] The hot air temperature Tkfto;

[0016] 磨煤机出口温度Tckto ; [0016] mill outlet temperature Tckto;

[0017] 磨煤机差压Psuct :磨煤机入口压力PmepL与磨煤机出口压力PMCTl的差值; [0017] Mill differential Psuct: inlet pressure difference between the mill and the mill outlet pressure PmepL of PMCTl;

[0018] 收集器工作周期的差压最大值Pmaxstqct :布袋收集器入口压力Ps■与布袋收集器出口力PSCYL工作周期的最大差值; [0018] The maximum differential collector Pmaxstqct duty cycle: maximum difference PSCYL duty cycle with inlet pressure Ps ■ power outlet collector bag collector bag;

[0019] 给煤机转速Wauzs,或给煤量。 [0019] The coal feeder speed Wauzs, or the amount of coal.

[0020] (2)样本数据的结构: [0020] (2) the structure of the sample data:

[0021] 煤粉磨制过程按照如下的结构组成样本,并收集样本数据。 [0021] Coal grinding process according to a structure comprising the sample and collect the sample data. 样本表达为{xk,yk}, 其中Xk为样本的输入,既选取的辅助变量——热风温度Tkfto、磨煤机出口温度Tckto、磨煤机差压Pwy、收集器工作周期的差压最大值Pmaxstqct、给煤机转速Wauzs。 Sample expressed as {xk, yk}, where Xk is the input sample, both the selected auxiliary variables - the hot air temperature Tkfto, mill outlet temperature Tckto, differential Pwy mill, the maximum value of the duty cycle of the differential collector Pmaxstqct, coal feeder speed Wauzs. 样本的输出为待估计的主导变量——煤粉细度Lmfxd。 Leading variable output sample is to be estimated - coal fineness Lmfxd. 样本采集记录结构如表1,时间为样本获取的时间,为减少化验误差,根据同一样本煤粉细度采用三次化验结果进行样本取舍,最后取平均值: Sample acquisition record structure as shown in Table 1, the sample acquisition time is the time, in order to reduce error tests, laboratory results for the three samples using a choice based on the same sample coal fineness, and finally averaged:

[0022] 表1样本数据结构 [0022] Table 1 sample data structure

[0023] [0023]

[0024] 考虑到样本数据应该具有代表性,并且尽可能覆盖范围较宽,至少应该包括工业对象正常工作范围,通过手动调节风阀开度、给煤机频率,在生产允许的范围内尽可能改变系统的工作点,每次操作条件改变系统运行平稳后取样化验。 [0024] Considering the sample data to be representative, and covers a wide range as much as possible, it should include at least the normal operating range of industrial objects, by manually adjusting the opening degree of the damper, the stoker frequency, maximum extent permitted production changing the operating point of the system, each operating conditions change sampling was conducted after the system is stable.

[0025] (3)样本的组成: [0025] (3) composition of the sample:

[0026] 本发明用于软测量模型的样本数据由三组组成:模型训练样本、误差训练样本和验证样本组成。 Measuring a soft sample data model for the invention [0026] This consists of three groups: model training samples, the training and validation samples error components. 考虑到训练样本数据应该具有代表性,并且尽可能覆盖范围较宽,至少应该包括工业对象正常工作范围,通过手动调节排风阀开度、给煤机频率,在生产允许的范围内改变系统的工作点,每次操作条件改变后,当系统运行平稳后取样化验。 Considering the training data to be representative, and covers a wide range as much as possible, it should include at least the normal operating range of industrial objects, by manually adjusting the degree of opening of the exhaust valve, the stoker frequency, the range of changes in the system to allow the production of operating point, after each operation conditions change, the system is stable when the assay sample. 在实际生产过程中,对煤粉细度抽查化验时按照表一纪录,获得一定数量的样本数据,将该部分数据分为两组,一组用于模型误差最小的训练来选择模型的参数,另一组用于模型的验证。 In the actual production process, when testing by the fineness of the coal in accordance with a table record, to obtain a certain amount of sample data, the partial data is divided into two groups, one of the smallest error for selecting a training model parameters of the model, another group used to validate the model.

[0027] (4)基于最小二乘支持向量机软测量模型 [0027] (4) Test Method of Soft LSSVM

[0028] 本发明的基于最小二乘支持向量机软测量模型,其模型训练的基本流程框图如图2所示。 [0028] Based on the measurement LSSVM soft model, the basic model training process which block diagram shown in Figure 2 of the present invention. 包括如下详细步骤: Detailed comprising the steps of:

[0029] (A)初始化:进行所有变量的初始化。 [0029] (A) Initialization: initializes all variables.

[0030] (B)是否进行软测量? [0030] (B) whether the soft sensor? 如果是,则转至(C),根据输入数据和软测量模型预测煤粉细度;如果否,则转至(G),进行模型训练、验证与参数修正的过程。 If yes, then go to (C), based on the input data and the predictive coal fineness soft measurement model; if not, go to (G), the model training, validation and parameter modification process.

[0031] 步骤(C)至(F)为软测量模型应用流程,根据输入数据和软测量模型预测煤粉细度。 [0031] Step (C) to (F) of soft measurement model application process, the fineness of pulverized coal measured in accordance with the input data and the soft model predictions.

[0032] (C)读取当前工况数据; [0032] (C) read the current condition data;

[0033] 也就是在线读取需要进行细度软测量的当前工况辅助变量的描述参数。 [0033] The line read is required fineness soft measurement parameters describing the current operating conditions of the auxiliary variables.

[0034] (D)输入数据标准化: [0034] (D) normalized input data:

[0035] 将输入变量参数数据进行标准化处理。 [0035] The variable input parameter data is normalized.

[0036] (E)软测量模型计算 [0036] (E) soft sensor model calculation

[0037] 把经过处理的输入数据代入到经过训练、验证后的如下模型计算煤粉细度的软测 [0037] After processing the input data into the model through the following training, validation calculate the soft coal fineness measured

量结果。 The amount of results.

[0038] [0038]

[0039] 式中xk—样本的输入,既选取的辅助变量——热风温度Tkfto、磨煤机出口温度TCKTO、磨煤机差压Pwct、收集器工作周期的差压最大值Pmaxstqct、给煤机转速Wauzs ; [0039] xk- input samples wherein both the selected auxiliary variables - the hot air temperature Tkfto, mill outlet temperature TCKTO, Pwct mill differential pressure, the duty cycle of the maximum differential collector Pmaxstqct, stoker speed Wauzs;

[0040] y(x)——待估计的主导变量——煤粉细度L· [0040] y (x) - Leading variable to be estimated - L · coal fineness

[0041] α k、b—训练后的模型系数; [0041] α k, model coefficients after training b-;

[0042] k = 1,2, ... ,η ; [0042] k = 1,2, ..., η;

[0043] (F)显示并保存细度软测量结果 [0043] (F) the measurement result is displayed and saved soft fineness

[0044] 将最小二乘_支持向量机模型计算的煤粉细度的软测量值显示在人机界面上,并与当前其他参数描述一起保存到历史数据库中,以供实际化验比较及其它操作使用。 [0044] The least squares _ coal fineness support vector machine Soft calculated measurement value displayed on the display unit, and stored with other parameters described in the current history database together, for comparing the actual testing, and other operations use.

[0045] 为了充分发挥本发明的细度软测量模型的功能,软测量程序应能实时自动读取当前工况描述,具体时限长短根据具体工艺情况确定,然后自动对读取的当前工况描述进行细度软测量计算,把软测量结果显示在操作者或者技术工人容易看到的画面上(可以是监控画面的“优化管理”画面),并把结果及其它数据保存到相关数据库中。 [0045] In order to function sufficiently soft fineness measurement model of the present invention, a soft real-time measurement procedure should be able to automatically read the description of the current operating conditions, particularly the length of time determined depending on the process circumstances, and then automatically reads the current operating conditions described in calculating soft fineness measurement, the measurement result soft displayed on a screen or operator skilled workers easily seen (the monitor screen may be "optimal management" screen), and to save the results and other data into a relational database.

[0046] 步骤(G)至⑴为模型训练、验证与参数修正的过程。 [0046] Step (G) to ⑴ model training, validation and parameter modification process.

[0047] (G)选择模型误差惩罚参数集和核参数集 [0047] (G) to select the model and error penalty parameter set parameter set nuclear

[0048] 根据具体情况确定对模型训练使用的误差惩罚参数和核参数的区间范围,该区间应当包含模型的最优参数,以便从中选择最佳的模型参数。 [0048] determined depending on the circumstances of the error model used in the training interval range parameter and the penalty kernel parameter, the optimal interval should contain the parameters of the model, in order to select the optimal model parameters.

[0049] (H)所有样本数据标准化 [0049] (H) data are normalized for all samples

[0050] 将训练样本、误差评价样本和模型验证样本进行标准化处理,用于模型训练、误差评价和模型验证的输入。 [0050] The training sample, the sample error evaluation and validation samples are normalized for model training, evaluation and validation error input.

[0051] (I)读入训练样本数据 [0051] (I) reads the training data

[0052] 读入准备用于模型训练的样本数据 [0052] reads the model ready for training sample data

[0053] (J)调整模型误差惩罚参数和核参数 [0053] (J) to adjust the model parameters and error penalty kernel parameter

[0054] 从误差惩罚参数和核参数的范围设置的下限开始,每次每个参数循环增加一个步长,作为调整后的参数,用于建立相应的模型对该组参数进行误差评价。 [0054] The lower limit of the range from the start setting parameters and error penalty nuclear parameters, each parameter per cycle increased by one step, as adjusted parameters, the model for establishing a respective set of parameters for error evaluation. [0055] (K)建立模型 [0055] (K) model

[0056] 基于最小二乘_支持向量机的模型建立过程描述如下: [0056] _ based on least squares SVM model is described as follows:

[0057] 本发明以采用径向基核函数(RBF)为例建立软测量模型,该核函数形式为: [0057] In the present invention uses radial basis function (RBF) model soft sensor as an example, the core function of the form:

[0058] [0058]

[0059] 式中:δ——核参数; [0059] wherein: δ-- kernel parameter;

[0060] 该软测量模型根据以下推导过程有:对于给定训练集为{xk,yJk=^,其中Xk e Rdyk e R,φ (.),d为辅助变量个数。 [0060] The soft sensor model according to the following derivation process are: For a given training set as {xk, yJk = ^, where Xk e Rdyk e R, φ, d is the number of auxiliary variables (.). 支持向量机的基本建模思想是通过非线性映射Φ (·),把输入样本从输入空间Rd映射到特征空间Φ (χ) = (Φ^χ), Φ2(χ),... Φη(χ)),从而将非线性估计问题转化为高维特征空间的线性函数估计问题。 The basic modeling idea is SVM nonlinear mapping Φ (·), the input samples Rd mapping from the input space to the feature space Φ (χ) = (Φ ^ χ), Φ2 (χ), ... Φη ( χ)), so that the nonlinear estimation problem is transformed into a linear function of the high dimensional feature space estimation. 即采用如下形式的函数对模型函数进行估计: That function takes the form of the model estimation function:

[0061] [0061]

[0062] 式中:WT——权向量 [0062] wherein: WT-- weight vector

[0063] b——偏置量 [0063] b-- offset

[0064] 根据问题求解目标和结构化风险最小为原则,上式需要同时满足下式: [0064] According to the objectives and problem solving structural risk minimization principle, the formula needs to satisfy the following formulas:

[0065] [0065]

[0066] 采用最小二乘_支持向量机方法,即定义误差损失函数为误差的二次项ek2,上述问题可以表述为在权空间W内,min·/(妒,e) =去扩识+去;^>42满足约束条件:Yk = ffT f [0066] _ least squares support vector machine, i.e., the error is defined as the quadratic loss function ek2 error, the above problem can be described as the right within the space W, min · / (jealous, e) = + despreading identification go; ^> 42 satisfy the constraints: Yk = ffT f

(xk)+b+ek (k= l,2,...,n)。 (Xk) + b + ek (k = l, 2, ..., n). 其中γ为误差惩罚因子。 Wherein γ is a penalty factor error. 用拉格朗日方法求解上述优化问 Asked to solve the above optimization method using the Lagrange

题,定义拉格朗日函数如下: Title, Lagrange function is defined as follows:

拉格朗日乘子。 Lagrange multipliers. 上述问题归结为二次规划问题,令 The above-mentioned problem is reduced to quadratic programming problems, so

[0068] [0068]

[0069]消除W,e,有 [0069] Elimination of W, e, there

,其中: ,among them:

[0071] 通过上式确定系数b和α,得到软测量模型为: [0071] is determined by the coefficient b, and α, measured to obtain soft model:

[0073] (L)读取误差评价样本数据 [0073] (L) Evaluation of the sample data read error

[0074] 读入准备用于误差评价的一组样本数据 [0074] The read ready for error evaluation of a sample data set

[0075] (M)记录误差评价结果和参数 [0075] (M) and the evaluation results of the recording error parameters

[0076] 取误差训练样本数据集S,包含样本数目为1,定义误差函数: [0076] The training data set taken error S, comprising a number of samples, the error function is defined:

[0079] 其中i = 1,2,. . .,1。 [0079] where i = 1,2 ,..., 1. 选择最终误差评价函数为: Select the final error evaluation function:

[0080] e( γ , δ) = min(el+ ne2) [0080] e (γ, δ) = min (el + ne2)

[0081] 式中:γ—误差惩罚参数; [0081] wherein: γ- error penalty parameter;

[0082] δ——核参数; [0082] δ-- kernel parameters;

[0083] η——权重参数; [0083] η-- weight parameter;

[0084] 根据经验选择均方差和最大方差的权重,一般可以选择η = 1。 [0084] The variance weights are selected empirically and maximum variance weight, generally choose η = 1. 在给定的参数区间内利用(K)建立的软测量模型得到误差评价指标,并记录对应的参数。 Using parameters within a given interval (K) to give the soft sensor model error evaluation, and record the corresponding parameters.

[0085] (N)参数调整是否已经到上限 If [0085] (N) parameters have been adjusted to an upper limit

[0086] 若Y+l> yup(y为误差惩罚参数,1为调整步长,Yup为惩罚参数范围的上限) 和S+l> Yup (δ为核参数、1为调整步长,Sup为核参数范围的上限)同时满足,则说明所有参数组合的误差评价工作完成;否则重复(J)〜(N)的工作。 [0086] When Y + l> yup (y is the error penalty parameter 1 for the adjustment step, Yup upper limit of the penalty parameter range) and S + l> Yup (δ nuclear parameter 1 is adjustment step, Sup is the upper limit of the nuclear parameters) meet, then error evaluation is completed for all combinations of parameters; otherwise repeat (J) ~ (N) work.

[0087] (0)选择误差评价最好的模型参数 [0087] (0) to select the best model parameter error evaluation

[0088] 将(M)中记录的误差评价指标,寻找其中的极小值,为误差评价最好的模型参数, 选择对应的参数为模型参数。 [0088] The error evaluation (M) recorded, wherein the minimum value seeking for the best evaluation error of model parameters, the corresponding parameters for the selected model parameters.

[0089] (P)确定模型 [0089] (P) is determined model

[0090] 根据(0)中选择的模型参数,确定模型训练结果,确定软测量模型。 [0090] The model parameters (0) is selected, determining model training results, the soft sensor model.

[0091] (Q)读入验证样本数据 [0091] (Q) sample data read verification

[0092] 读入准备用于模型验证的一组样本数据 [0092] The model validation is ready for reading a set of sample data

[0093] (R)模型验证 [0093] (R) model verification

[0094] 采用⑵中的模型按照(M)中的el指标和e2指标。 [0094] The model in accordance ⑵ el and e2 index indicator (M) is.

[0095] (S)验证精度是否满意 If [0095] (S) verify the accuracy of satisfaction

[0096] 用验证样本数据对软测量模型进行验证,比较软测量模型的输出和实际化验值。 [0096] Soft sensing model to verify the data with the validation sample, relatively soft measurement model output and the actual laboratory values. 如果模型输出值与实际化验值的比较误差在误差允许范围内,该模型可以用于实际煤粉细度在线估计;如果该模型输出值与实际化验值的比较误差较大,分析训练样本数据,继续实验,适当增加训练样本数据,重复(B)〜(S)的工作。 If the model output value and compares the error in the actual laboratory values ​​within tolerance limits, the model can be used to estimate the actual coal fineness line; If the model output value and the actual value comparison error larger assay, analyzing training data, continue the experiment, a suitable training data increases, repeating (B) ~ (S) of the work.

[0097] (T)重新构造训练样本 [0097] (T) to re-structure the training samples

[0098] 验证精度不能满足软测量的需要,需要增加试验次数,重新构造训练样本。 [0098] Soft verify the accuracy of measurement can not meet the needs, it is necessary to increase the number of tests, reconstructed training samples.

[0099] (U)结束 [0099] (U) end

[0100] (5)模型参数的校正: [0100] (5) correction of model parameters:

[0101] 随着时间的推移,化验得到的样本数据不断增加,为进一步修正模型参数,提高模型精度提供了可能。 [0101] Over time, the sample data obtained by testing increasing, provides the possibility of further correction model parameters to improve the accuracy of the model. 选择典型的样本添加到训练样本,如果重新训练的软测量模型能够明显提高模型的精度,则用新的模型代替原有模型,否则不改变原模型。 Select representative sample added to the sample training, soft measuring model if re-training can significantly improve the accuracy of the model, with the new model to replace the original model, or not to change the original model. 如果新的样本数据重复出现样本输入条件相同或相近,而样本输出相差较大的情况,用新样本代替原样本数据, 重新训练模型,替换原有模型。 If the new samples of the input sample data is repeated the same or similar conditions, the difference between the larger of the sample output, the sample instead of the original data with the new sample, retrain the model, to replace the original model.

[0102] 本发明的优点在于:利用计算机系统和常规的检测仪表提供的在线过程数据,仅仅通过少量的人工采样,实现了煤粉细度的基于最小二乘-支持向量机的软测量。 [0102] advantages of the present invention is that: the computer system using the online process data and conventional instrumentation is provided, only a small amount of manual sampling, to achieve the fineness of pulverized coal based on Least Square - Soft Sensor Support Vector Machine. 解决了没有煤粉细度在线检测仪表难以在线检测的难题;与人工取样化验相比,减少了操作人员的工作量,降低了人为操作引入的测量的不确定性,提高了测量的时效性,减少了取样带来的煤粉泄漏的问题。 Solve the problem of no-line coal fineness instrumentation line is difficult to detect; compared with manual sampling tests, reduced operator workload, reducing measurement uncertainty introduced by manual operation, improving the timeliness of measurement, It reduces the problems caused by coal dust sampling leak. 与传统的模糊和神经网络软测量方法相比本发明中软测量方法克服了输入维数高、样本数量少和由于系统干扰因素多样本有时引入较大的误差造成的模型泛化和过拟合的问题。 Compared with the traditional method of soft and fuzzy neural network of the present invention is measured in soft measurement overcomes the high input dimensions, a small number of samples and due to the system due to multiple sample introduced into the model may cause a larger error generalization and overfitting problem. 另外,由于本发明全面考虑了影响煤粉细度的因素,大量采用现有的常规检测信号来实现煤粉细度的在线估计,应用简单、容易,成本较低,软测量结果也较精确。 Further, since the present invention fully considers the factors affecting coal fineness, the extensive use of existing conventional line estimate the detection signal to achieve the fineness of pulverized coal, the application is simple, easy, low cost, flexible measurement results are more accurate. 该方法有助于实现制粉过程的优化控制和优化运行。 This approach helps to optimize the milling process control and optimization operation.

附图说明 BRIEF DESCRIPTION

[0103] 图1制粉过程的流程、测量仪表及计算机配置图; [0103] milling process flow of FIG. 1, FIG measuring instruments and computer configuration;

[0104] 图2基于最小二乘_支持向量机的制粉过程煤粉细度软测量模型建立流程图 [0104] FIG. 2 _ based on least squares SVM during milling coal fineness flowchart soft sensor model

[0105] 图1中:1钢球磨煤机,2布袋收集器,3粗粉分离器,4给煤机,5布袋收集器入口压力计,6布袋收集器入出口压力计,7给煤机变频转速或者给煤量称重,8热电偶,9磨煤机入口压力计,10磨煤机出口压力计,11热电阻,12PLC、控制计算机或专用计算机 In [0105] FIG 1: a ball mill, a collector bag 2, 3 Classifier, coal feeder 4, the bag collector inlet pressure gauge 5, 6 into the collector bag outlet pressure gauge 7 stoker frequency weighing speed or the amount of coal, thermocouple 8, 9 mill inlet pressure gauge, 10 gauge mill outlet, 11 heat resistance, 12PLC, control or special purpose computer

[0106] 图1中所用标记符号如下: [0106] in FIG. 1 by the reference character as follows:

[0107] 热风温度——Tefwd [0107] Hot air temperature --Tefwd

[0108] 磨煤机出口温度——Tckwd [0108] mill outlet temperature --Tckwd

[0109] 磨煤机入口压力——Pmepl [0109] mill inlet pressure --Pmepl

[0110] 磨煤机出口压力——Pmcyl [0110] mill outlet pressure --Pmcyl

[0111] 布袋收集器入口压力——Pseyl [0111] bag collector inlet pressure --Pseyl

[0112] 布袋收集器出口压力——Pscyl [0112] bag collector outlet pressure --Pscyl

[0113] 给煤机转速或给煤量——Wgsuzs [0113] or a coal feeder speed to a coal --Wgsuzs

[0114] 在图1中,实线箭头表示物流(煤粉,热风和冷风),虚线表示信号流。 [0114] In Figure 1, solid arrows indicate stream (coal, hot and cold), a broken line indicates the signal flow.

[0115] 具体实施方式 [0115] DETAILED DESCRIPTION

[0116] 以氧化铝烧成回转窑所用的钢球磨煤机制粉系统为例,如图1所示煤粉磨制过程由给煤机、磨煤机、粗粉分离器和布袋除尘器、煤粉仓等设备组成。 [0116] In an alumina calcination kiln used in steel - ball mill, for example, by the process shown in FIG grinding coal to coal, mill, coarse separator and a bag dust, coal FIG 1 bunker and other equipment. 原煤和热风在钢球磨煤机内磨制和干燥,煤粉在排风机的作用下进入粗粉分离器,回粉从段仓返回到磨煤机继续研磨,合格煤粉经布袋除尘器实现风粉分离,煤粉由输送装置送入煤粉仓。 Coal and hot air and grinding in a ball mill and dried, pulverized coal into the coarse separator under the effect of the exhaust fan, to return back from the powder cartridge section to continue the grinding mill, the pulverized coal passing through the bag filter to achieve air powder separation, coal fed by the coal storage conveying means.

[0117] 磨煤机规格为Φ3. 2X9. 5m,设计生产能力28t/h,磨煤机分为烘干仓、球仓、段仓三仓;主电机型号YR80-8/1180,额定功率800KW。 [0117] Mill specifications Φ3 2X9 5m, design production capacity of 28t / h, into drying ovens mill, ball positions, three positions para positions;.. Part main motor YR80-8 / 1180, rated power 800KW .

[0118] 该磨煤机系统具有本说明书的要求安装如下的测量仪表,包括: [0118] The requirements of this specification mill system has installed the following measuring device, comprising:

[0119] 热风温度; [0119] hot air temperature;

[0120] 磨煤机出口温度; [0120] mill outlet temperature;

[0121] 磨煤机进出口压力; [0121] mill outlet pressure;

[0122] 收集器进出口压力; [0122] collector outlet pressure;

[0123] 给煤机转速反馈。 [0123] coal feeder speed feedback.

[0124] 以可编程控制器(PLC)实现基础控制回路的自动控制,并根据模型需要滤波、计算得到:[0125] 热风温度Tkfto ; [0124] In the programmable controller (PLC) to achieve automatic control of the control circuit based, and model requires filtering is calculated according to: [0125] hot air temperature Tkfto;

[0126] 磨煤机出口温度Tckto ; [0126] mill outlet temperature Tckto;

[0127] 磨煤机差压Pmjcy; [0127] Pressure mill Pmjcy;

[0128] 收集器工作周期的差压最大值Pmaxsjqct ; [0128] Pressure maximum duty cycle collector Pmaxsjqct;

[0129] 给煤机转速Wgmjzs。 [0129] coal feeder speed Wgmjzs.

[0130] 读取上述的过程数据。 [0130] The above-described read process data. 在监控计算机中以西门子的WINCC软件实现监控人机界 In the computer monitor to Siemens WINCC software monitoring human machine interface

[0131] 该制粉系统的煤粉细度(175#筛上)正常工作范围,在供应两台窑时指标设定为15%〜20%。 [0131] (175 # mesh) pulverized coal fineness of the normal operating range of the milling system, when supplying two indicators kiln was set to 15% ~ 20%.

[0132] 软测量模型用西门子PLC的st印7的模块化编程实现,数据存储在DB块中。 [0132] Soft measured with a Siemens PLC model of modular programming st plate 7 implementation, the data stored in the DB. 在WINCC监控系统提供模型数据的修改接口,用于离线分析更改模型参数。 Providing a modification interface model data WINCC monitoring system, change the model parameters for offline analysis. 软测量软件在PLC 上运行,有效的保证了模型输出的时效性,方便了系统的过程监控。 Soft sensor software running on the PLC, effectively guarantee the timeliness of the model output for easy system monitoring.

[0133] 按照本说明书所述的实现方法选择辅助变量,获取实验样本,建立软测量模型,并通过误差最小训练,获得模型参数,得到Y =0.5和δ =0.8。 [0133] The choice of auxiliary variables implemented method according to the present specification, the test sample acquiring soft sensor model and minimal training by the error to obtain the model parameters, to obtain Y = 0.5 and δ = 0.8. 在该参数下,计算a,b的值分别为: In this parameter, calculated values ​​a, b are respectively:

[0134] b= 19. 4587; [0134] b = 19. 4587;

[0135] a = [-3. 16 —1.11 —1.86 —2.6 0. 77 2.6 1. 78 17.35 -3. 04 5.41 -3. 97 -1. 95-9. 04 0. 68 -2. 54 1. 61 -0. 32 -0. 97 0. 36] [0135] a = [-3. 16 -1.11 -1.86 -2.6 0. 77 2.6 1. 78 17.35 -3. 04 5.41 -3. 97 -1. 95-9. 04 0. 68 -2. 54 1. 61-0. 32-0. 97 0.36]

[0136] 本实例由于收集器的工作周期为336s,每隔336s得到一次收集器的最大差压值, 从而运行一次软测量程序,进行一次该时间工况下的细度软测量,并通过监控计算机把软测量结果输出到系统管理画面,为操作者或相关技术工人进行参数优化和监督操作提供一个很好的的参考、指导作用,同时把软测量结果保存到系统实时数据库中。 [0136] Since the present example, the duty cycle of the collector is 336s, 336s maximum differential pressure value of the primary collector to run a software program measured, once the soft fineness measurement conditions of time, and by monitoring every computer software output measurement results to the system management screen, the parameters for the operator or related skilled workers to optimize operations and supervision to provide a good reference, guidance, and save measurement results to the soft real-time database system.

[0137] 本发明即基于最小二乘-支持向量机的煤粉细度软测量模型在制粉系统正常运行期间,能够根据过程的实时数据估计煤粉粒度,相对误差不超过2%,成为一个具有很高实用价值的、低成本的细度计量手段。 [0137] The present invention is based on Least Square - soft coal fineness measurement model SVM milling system during normal operation, coal particle size can be estimated according to the real-time data process, the relative error is less than 2%, as a It has a high practical value, low-cost fineness measurement instruments.

Claims (5)

  1. 一种基于最小二乘‑支持向量机的制粉过程煤粉细度指标软测量方法,该方法依赖硬件平台及测量仪表和计算机系统,其特征在于通过控制计算机或数据采集器获得实时的过程数据,进行软测量,该方法包括以下步骤:(1)辅助变量的选择,选择的辅助变量包括:热风温度TRFWD;磨煤机出口温度TCKWD;磨煤机差压PMJCY:磨煤机入口压力PMRPL与磨煤机出口压力PMCYL的差值;收集器工作周期的差压最大值PMAXSJQCY:布袋收集器入口压力PSRYL与布袋收集器出口压力PSCYL工作周期的最大差值;给煤机转速WGMJZS,或给煤量;(2)确定样本数据的结构,如下表:其中xk为样本的输入,既选取的辅助变量——热风温度TRFWD、磨煤机出口温度TCKWD、磨煤机差压PMJCY、收集器工作周期的差压最大值PMAXSJQCY、给煤机转速WGMJZS,样本的输出yk为待估计的主导变量——煤粉细度LMFXD;(3)样 Based on least squares - supports soft coal fineness index measuring method during milling vector machine, the method relies on the hardware platform and measurement instruments and a computer system, wherein the data obtained by real-time process control computer or data logger , soft sensor, the method comprising the steps of: (1) selecting an auxiliary variable and select the secondary variables comprising: a hot air temperature TRFWD; mill outlet temperature TCKWD; mill pressure PMJCY: mill inlet pressure and PMRPL PMCYL the pressure difference between the mill outlet; PMAXSJQCY maximum differential collector duty cycle: the maximum pressure difference between the outlet collector bag PSCYL work cycle with inlet pressure PSRYL bag collector; coal feeder speed WGMJZS, or coal amount; (2) determining the structure of the sample data in the following table: where xk is the input sample, both the selected auxiliary variables - the hot air temperature TRFWD, mill outlet temperature TCKWD, mill differential pressure PMJCY, the duty cycle collector the maximum differential pressure PMAXSJQCY, coal feeder speed WGMJZS, output sample yk dominant variables to be estimated - coal fineness LMFXD; (3) sample 的组成用于软测量模型的样本数据由三组组成:模型训练样本、误差训练样本和验证样本组成;模型训练样本至少应该包括工业对象正常工作范围,通过手动调节排风阀开度、给煤机频率,在生产允许的范围内改变系统的工作点,每次操作条件改变后,当系统运行平稳后取样化验的样本数据;对煤粉细度抽查化验时按照样本数据结构纪录,获得一定数量的样本数据,将该部分数据分为两组,一组用于模型误差最小的训练来选择模型的参数,另一组用于模型的验证;(4)建立基于最小二乘‑支持向量机软测量模型,如下式: <mrow> <mi>y</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>&alpha;</mi> <mi>k</mi> </msub> <mi>k</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <msub> <mi>x</mi> <mi>k</mi> </ Composition for soft sample data measured by the three models: the model training samples, the training and validation samples errors composition; model training sample should include at least the normal operating range of industrial objects, by manually adjusting the degree of opening of the exhaust valve, coal local frequency, changes the system operating point within the range of allowable production, after each operation condition is changed, the system is stable when the assay sample data sampled; sample data structure according to the record, is obtained when a certain number of testing by the coal fineness sample data, the authentication data into two portions, a minimum set of parameters for selecting training model error model, a model for the other groups; (4) the establishment of least squares based on - SVM soft measurement model, the following formula: <mrow> <mi> y </ mi> <mrow> <mo> (</ mo> <mi> x </ mi> <mo>) </ mo> </ mrow> <mo > = </ mo> <munderover> <mi> & Sigma; </ mi> <mrow> <mi> k </ mi> <mo> = </ mo> <mn> 1 </ mn> </ mrow> < mi> k </ mi> </ munderover> <msub> <mi> & alpha; </ mi> <mi> k </ mi> </ msub> <mi> k </ mi> <mrow> <mo> ( </ mo> <mi> x </ mi> <mo>, </ mo> <msub> <mi> x </ mi> <mi> k </ mi> </ msub> <mo>)</mo> </mrow> <mo>+</mo> <mi>b</mi> </mrow>αk、b——训练后的模型系数;k=1,2,...,n;然后进行模型的训练、参数选择及模型验证;(5)模型参数的校正;选择典型的样本添加到训练样本,如果重新训练的软测量模型能够明显提高模型的精度,则用新的模型代替原有模型,否则不改变原模型;如果新的样本数据重复出现样本输入条件相同或相近,而样本输出相差较大的情况,用新样本代替原样本数据,重新训练模型,替换原有模型。 msub> <mo>) </ mo> </ mrow> <mo> + </ mo> <mi> b </ mi> </ mrow> αk, b-- model coefficients after training; k = 1,2 , ..., n-; and training, parameter selection and model validation of the model; correcting (5) the model parameters; selecting representative sample added to the training sample, if soft measurement model retraining can significantly improve the accuracy of the model, it is replaced with a new model of the original model, or not changing the original model; if new samples of the input sample data is repeated the same or similar conditions, the difference between the larger of the sample output, the sample instead of the original data with the new sample, the model retrained , replace the original model.
  2. 2.根据权利要求1所述的一种基于最小二乘-支持向量机的制粉过程煤粉细度指标软测量方法,其特征在于步骤(4)所述的模型建立、训练、参数选择及模型验证按以下步骤进行:(A)初始化:进行所有变量的初始化;(B)是否进行软测量? According to one of the claims 1 based on least squares - coal fineness index measuring method during milling soft support vector machine, wherein the step of modeling (4), the training parameters and the model validation perform the following steps: (a) initialization: initializes all variables; (B) whether soft sensor? 如果是,则转至(C),根据输入数据和软测量模型预测煤粉细度;如果否,则转至(G),进行模型训练与参数修正的过程;(C)读取当前工况数据;⑶输入数据标准化处理;(E)软测量模型计算;(F)显示并保存细度软测量结果,执行(U) ; (G)选择模型误差惩罚参数集和核参数集;(H)所有样本数据标准化;⑴读入训练样本数据;(J)调整模型误差惩罚参数和核参数;⑷建立模型;(L) 读取误差评价样本数据;(M)记录误差评价结果和参数;(N)参数调整是否已经到上限;若参数调整到上限,选择参数;若不到上限,根据调整步长调整新参数重复误差评价过程,返回(J) ; (0)选择误差评价最好的模型参数;(P)确定模型;(Q)读入验证样本数据;(R)模型验证;(S)验证精度是否满意,如果满意,模型参数选择完毕;如果不满意,增加化验数据, 重新构造训练样本,执行 If yes, then go to (C), based on the input data and the predictive coal fineness soft measurement model; if not, go to (G), and training the model parameters of the correction process; (C) read the current operating conditions data; ⑶ input data normalized; (E) soft measurement model calculation; (F.) is displayed and saved soft measurements fineness is performed (U); (G) to select the model error penalty parameter set and the nuclear parameter sets; (H) All sample data normalized; ⑴ reads the training data; (J) to adjust the model parameters and error penalty kernel parameter; ⑷ model; (L) evaluation of the sample data read error; (M), and evaluation results of the recording error parameter; (N ) whether the parameter adjustment to an upper limit; parameter adjustment if the upper limit selection parameter; if not to limit, according to the new adjustment parameter adjustment step is repeated error evaluation process, return (J); (0) to select the best model parameter error evaluation ; (P) determining the model; (Q) sample data read verification; (R & lt) model verification; if (S) verify the accuracy satisfied, if satisfied, the model parameter selection is completed; if not, increase the assay data, reconstructed samples training ,carried out (T) ; (T)重新构造训练样本;(U)结束。 (T); (T) to reconstruct the training samples; (U) end.
  3. 3.根据权利要求2所述的一种基于最小二乘-支持向量机的制粉过程煤粉细度指标软测量方法,其特征在于步骤(M)所述的记录误差评价结果和参数时使用以下函数: According to one of the claim 2 based on least squares - measuring methods used to support soft coal fineness index vector during milling machine, wherein said step of recording error (M) and the evaluation results of the parameter the following functions:
  4. 4.根据权利要求2所述的一种基于最小二乘-支持向量机的制粉过程煤粉细度指标软测量方法,其特征在于步骤(N)所述的判断参数调整是否已经到上限时应符合以下条件: Y+1> Yup,式中:Y为误差惩罚参数、ι为调整步长,Yup为惩罚参数范围的上限;和S+1 > Yup,式中:S为核参数、1为调整步长,Sup为核参数范围的上限,同时满足。 According to one of the claim 2 based on least squares - coal fineness index measuring method during milling soft support vector machine, wherein determining whether the parameter adjustment step (N) has reached the upper limit should meet the following conditions: Y + 1> Yup, wherein: Y is error penalty parameter iota for the adjustment step, Yup is a penalty parameter range limit; and S + 1> Yup, the formula where: S is the kernel parameter, 1 It is a step size, the upper limit Sup core parameters while satisfying.
  5. 5.根据权利要求1所述的一种基于最小二乘-支持向量机的制粉过程煤粉细度指标软测量方法,其特征在于其依赖的硬件平台核心由钢球磨煤机、粗粉分离器、布袋收集器以及相关设备组成,同时配备了测量仪表以及进行软件计算的计算机系统;磨煤机的输入端与给煤机、热风和冷风管道相接,磨煤机出口、粗粉分离器及布袋收集器依次相连,在热风和磨煤机出口管道上分别装有热电偶和热电阻测温原件,磨煤机和布袋收集器的进出口管道上分别安装压力计,给煤机调速变频器输出频率反馈信号,系统配置控制计算机或数据采集器,采集测量仪表信号;测量软件运行在计算机控制系统的监控计算机上或独立的计算机上。 5 based on least squares according to one of the claims 1 - coal fineness index measuring method during milling soft support vector machine, characterized in that the core of the hardware platform dependent ball mill, coarse separation , a bag collector composition and associated equipment, and a measuring device equipped with a computer system calculated by the software; and input to a coal pulverizer, in contact with hot air and the cold air duct, the mill outlet, Classifier and in turn connected to the collector bag, and the hot air in the mill outlet conduit are respectively provided with the original temperature thermocouple and thermal resistance, are mounted pressure gauge and inlet and outlet piping bag collector mill, coal feeder speed output frequency feedback signal, the control computer system configuration or the data acquisition, acquire measurement instrument signal; measuring software running on a computer control system monitoring computer or a separate computer.
CN 200710011021 2007-04-19 2007-04-19 Soft measurement method for coal power fineness in powdering producer CN101038277B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200710011021 CN101038277B (en) 2007-04-19 2007-04-19 Soft measurement method for coal power fineness in powdering producer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200710011021 CN101038277B (en) 2007-04-19 2007-04-19 Soft measurement method for coal power fineness in powdering producer

Publications (2)

Publication Number Publication Date
CN101038277A CN101038277A (en) 2007-09-19
CN101038277B true CN101038277B (en) 2010-12-08

Family

ID=38889320

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200710011021 CN101038277B (en) 2007-04-19 2007-04-19 Soft measurement method for coal power fineness in powdering producer

Country Status (1)

Country Link
CN (1) CN101038277B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105930666A (en) * 2016-04-27 2016-09-07 厦门大学 Online soft measurement method for temperature of hearth of Shell entrained-bed gasifier

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101334666B (en) 2008-07-15 2010-10-27 西安艾贝尔科技发展有限公司 Double-inlet double-outlet steel ball coal mill straight blowing type milling system optimized control method
CN102778538B (en) * 2012-07-06 2014-12-10 广东电网公司电力科学研究院 Soft measuring method based on improved SVM (Support Vector Machine) for measuring boiler unburned carbon content in fly ash
CN103336107B (en) * 2013-05-30 2014-05-07 中国科学院沈阳自动化研究所 Soft measurement method for f-CaO content of cement clinker
CN103472867B (en) * 2013-09-22 2015-11-04 浙江大学 The optimizing temperature of pesticide production waste liquid incinerator system and method for support vector machine
CN103472728B (en) * 2013-09-22 2016-08-24 浙江大学 The pesticide incinerator hazardous emission controls up to par system and method for machine learning
CN103699782B (en) * 2013-12-09 2017-02-01 国家电网公司 Coal feeding amount soft measuring method of middle-speed powder milling and preparation system
CN104374675B (en) * 2014-10-15 2016-10-26 国家电网公司 A kind of coal pulverizer fineness of pulverized coal on-line monitoring method
CN104634706A (en) * 2015-01-23 2015-05-20 国家电网公司 Neural network-based soft measurement method for pulverized coal fineness
CN104991037B (en) * 2015-05-15 2016-06-22 广西大学 A kind of data measuring method for cane sugar boiling and crystal and device
CN106529008B (en) * 2016-11-01 2019-11-26 天津工业大学 A kind of double integrated offset minimum binary modeling methods based on Monte Carlo and LASSO
CN106546704A (en) * 2016-12-09 2017-03-29 华北电力大学(保定) A kind of nitrogen oxides hard measurement device of compressed sensing and its measuring method
CN107016176A (en) * 2017-03-24 2017-08-04 杭州电子科技大学 A kind of hybrid intelligent overall boiler burning optimization method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1153906A (en) 1996-01-03 1997-07-09 山东省电力学校 Fast automatic duff fineness detector
US20040089202A1 (en) 2002-11-07 2004-05-13 Vladimir Ronin Method for the treatment of fly ash
CN2752755Y (en) 2004-10-15 2006-01-18 南京大陆中电科技股份有限公司 Powder coal fineness on-line detector

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1153906A (en) 1996-01-03 1997-07-09 山东省电力学校 Fast automatic duff fineness detector
US20040089202A1 (en) 2002-11-07 2004-05-13 Vladimir Ronin Method for the treatment of fly ash
CN2752755Y (en) 2004-10-15 2006-01-18 南京大陆中电科技股份有限公司 Powder coal fineness on-line detector

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
赵晴川.燃煤锅炉风煤比和煤粉细度在线测量技术.热力发电 5.2005,(5),57,58.

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105930666A (en) * 2016-04-27 2016-09-07 厦门大学 Online soft measurement method for temperature of hearth of Shell entrained-bed gasifier
CN105930666B (en) * 2016-04-27 2019-01-22 厦门大学 A kind of online soft sensor method of entrained flow gasifying furnace bore temperature

Also Published As

Publication number Publication date
CN101038277A (en) 2007-09-19

Similar Documents

Publication Publication Date Title
US4390058A (en) Method of monitoring condenser performance and system therefor
CN101334666B (en) Double-inlet double-outlet steel ball coal mill straight blowing type milling system optimized control method
CN102227691B (en) Detection of anomalies in aircraft engine
CN1224720C (en) Blast furnace smelt controlling method with intelligent control system
Wei et al. Development of a multisegment coal mill model using an evolutionary computation technique
CN102607857A (en) Turbocharger performance qualification method and apparatus
CN101902195B (en) Method for automatically calibration of excitation system modelling and PSS optimization
ES2293511T3 (en) Procedure for monitoring the performance of an industrial equipment.
CN103028480A (en) Intelligent control system for vertical mill based on fuzzy PID (proportion integration differentiation) algorithm
CN1749727A (en) Method for low-cost estimation of steam turbine performance
CN102213708A (en) Method for testing air leak rate of air preheater
CN2826422Y (en) Rapid monitoring device for strap coal sampling
CN100524131C (en) System and method for detecting an abnormal situation associated with a heater
CN1207399C (en) Intelligent blast furnace smelt controlling system
CN1694109B (en) Material data correction method in chemical and oil refinement process
JP4562108B2 (en) Method for online measurement of fuel thermal function of fuel in a combustion turbine unit
CN103615881A (en) On-line measuring and control method and system of grain circulation drying moisture based on total-weight detection
CN202628482U (en) Automatic blower fan performance testing system
CN1221787C (en) Separated metering method and system for central air conditioner
Vasquez et al. System identification of steam pressure in a fire-tube boiler
CN100538316C (en) Thermal power plant ball mill load soft-sensing method
CN103975150B (en) Method and apparatus for being monitored to gas sensor
CN101105520B (en) Generating set primary frequency regulation test method and emulating instrument for test
CN102175282A (en) Method for diagnosing fault of centrifugal air compressor based on information fusion
CN102680235B (en) Method and system for the analysis of turbine

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model
CF01