CN103760131A

Movatterモバイル変換

Info

Publication number: CN103760131A
Application number: CN201410021092.1A
Authority: CN
Inventors: 钱锋; 程辉; 杜文莉
Original assignee: East China University of Science and Technology
Current assignee: East China University of Science and Technology
Priority date: 2014-01-17
Filing date: 2014-01-17
Publication date: 2014-04-30

Abstract

本发明涉及一种基于近红外光谱检测汽油油品属性的实时方法，包括采集汽油样品及其近红外光谱，并利用实验室方法测得汽油的所需属性；然后采用基线校正方法对采集到的汽油光谱样本进行基线校正并选择合适的波长范围；最后，建立属性值与近红外光谱之间的数学模型并用于汽油属性的预测分析。该方法采用在线近红外分析仪，汽油样本的近红外光谱可以实时获取；因此可以对汽油油品属性进行实时预测；具有分析速度快、信噪比高等优点，特别适合汽油管道调和过程中汽油油品属性的在线预测分析。

The invention relates to a real-time method for detecting the properties of gasoline oil based on near-infrared spectroscopy, which includes collecting gasoline samples and their near-infrared spectra, and using laboratory methods to measure the required properties of gasoline; The gasoline spectral samples are baseline corrected and the appropriate wavelength range is selected; finally, a mathematical model between attribute values and near-infrared spectra is established and used for predictive analysis of gasoline attributes. This method uses an online near-infrared analyzer, and the near-infrared spectrum of gasoline samples can be obtained in real time; therefore, the properties of gasoline oil can be predicted in real time; it has the advantages of fast analysis speed and high signal-to-noise ratio, and is especially suitable for gasoline pipeline blending process. Online predictive analysis of product attributes.

Description

Translated fromChinese

基于近红外光谱检测的汽油油品属性实时预测方法Real-time prediction method of gasoline oil properties based on near-infrared spectroscopy detection

技术领域technical field

本发明涉及一种基于近红外光谱检测的汽油油品属性实时预测方法，特别适用于在线预测汽油油品的研究法辛烷值与马达法辛烷值属性。The invention relates to a real-time prediction method of gasoline oil product attributes based on near-infrared spectrum detection, and is particularly suitable for online prediction of gasoline oil product research method octane number and motor method octane number attributes.

背景技术Background technique

汽油产品是炼油厂的一个重要经济来源，因为炼油厂中有60%-70%的原油将最终转化为汽油。而汽油调和作为汽油生产过程中的最后一步，也是至关重要的一个环节，汽油调和系统的好坏直接影响到最终生产的成品汽油的质量，从而影响到企业汽油生产的经济效益。如何充分利用现有油品资源，以更低的成本生产出更多的高质量汽油，是汽油调和一直致力于解决的问题。而且随着我国汽车数量的迅速增加，高标号汽油的需求量越来越大，大量生产高标号汽油满足社会的需要已经成为炼油企业的急迫任务。如何充分利用现有的调和组分、油品罐容等资源，在保证产品质量合格的前提下，生产最大量的高标号汽油，已成为目前炼油企业急需解决的问题。Gasoline products are an important source of income for refineries, as 60%-70% of the crude oil in a refinery is ultimately converted to gasoline. Gasoline blending, as the last step in the gasoline production process, is also a crucial link. The quality of the gasoline blending system directly affects the quality of the final product gasoline produced, thus affecting the economic benefits of the gasoline production of the enterprise. How to make full use of existing oil resources to produce more high-quality gasoline at a lower cost is a problem that gasoline blending has been committed to solving. Moreover, with the rapid increase of the number of automobiles in our country, the demand for high-grade gasoline is increasing. It has become an urgent task for oil refiners to mass-produce high-grade gasoline to meet the needs of the society. How to make full use of existing blending components, oil tank capacity and other resources to produce the largest amount of high-grade gasoline under the premise of ensuring product quality has become an urgent problem for oil refining companies.

近年来国家对车用燃油排放指标限制也越来越严格，不仅对辛烷值等常规指标有要求，而且对硫、烯烃、芳烃等都有愈加严格的限制。因此许多炼油企业采用的传统人工试凑的配方计算方式已经难以满足日益苛刻的汽油指标要求。汽油调和过程优化在国内外已经得到了广泛重视，汽油调和的优化控制依赖于近红外在线分析仪表，其分析的准确性与稳定性直接影响到调和优化控制的成败。由于近红外分析仪是二次测量仪表，即近红外分析仪并不能直接测量物质属性，而必须先建立待测物质的属性与近红外光谱之间的数学模型然后根据模型来测量物质属性。因此，建立精度高、鲁棒性好的近红外模型是近红外技术能否有效应用的关键。In recent years, the country has become more and more strict on the emission indicators of vehicle fuel, not only for conventional indicators such as octane number, but also for sulfur, olefins, and aromatics. Therefore, the traditional formula calculation method of manual trial and error adopted by many refineries has been difficult to meet the increasingly stringent gasoline index requirements. The optimization of gasoline blending process has been widely paid attention at home and abroad. The optimization control of gasoline blending depends on the near-infrared online analysis instrument. The accuracy and stability of its analysis directly affect the success or failure of the optimal control of blending. Since the near-infrared analyzer is a secondary measuring instrument, that is, the near-infrared analyzer cannot directly measure the properties of the material, but must first establish a mathematical model between the properties of the material to be measured and the near-infrared spectrum, and then measure the properties of the material according to the model. Therefore, establishing a near-infrared model with high precision and good robustness is the key to the effective application of near-infrared technology.

发明内容Contents of the invention

为了解决上述技术问题，本发明提出一种基于近红外光谱检测的汽油油品属性实时预测方法，特别适用于在线预测汽油油品的研究法辛烷值与马达法辛烷值属性。该方法可以根据调和过程的实时光谱样本预测汽油油品的研究法辛烷值与马达法辛烷值等属性。In order to solve the above-mentioned technical problems, the present invention proposes a real-time prediction method of gasoline oil properties based on near-infrared spectroscopy, which is especially suitable for online prediction of the research octane number and motor octane number attributes of gasoline oil products. The method can predict properties such as research octane number and motor octane number of gasoline oil products based on real-time spectral samples of the blending process.

本发明方法提供的基于近红外光谱检测的汽油油品属性实时预测方法包括以下步骤：The real-time prediction method of gasoline oil properties based on near-infrared spectrum detection provided by the method of the present invention comprises the following steps:

步骤一:收集汽油样品，利用实验室方法测得汽油的属性Step 1: Collect gasoline samples and use laboratory methods to measure the properties of gasoline

步骤二：采集汽油样品的近红外光谱Step 2: Collect the near-infrared spectrum of the gasoline sample

步骤三：对所述的近红外光谱进行预处理，消除干扰因素，选择波长范围，分别建立汽油研究法辛烷值与马达法辛烷值和近红外光谱之间的近红外模型并检验Step 3: Preprocess the near-infrared spectrum, eliminate interference factors, select the wavelength range, respectively establish the near-infrared model between gasoline research method octane number and motor method octane number and near-infrared spectrum and test

步骤四：采用步骤三中建立的近红外模型对各属性值进行预测Step 4: Use the near-infrared model established in Step 3 to predict the value of each attribute

步骤五：用所建的近红外模型预测汽油调和过程中未知汽油属性。Step 5: Use the built near-infrared model to predict unknown gasoline properties in the gasoline blending process.

较佳的，步骤一所述的方法，光谱采集应该在流量相对稳定，油品属性曲线平稳无毛刺的情况下获取汽油近红外光谱。Preferably, in the method described instep 1, the spectral collection should acquire the near-infrared spectrum of gasoline when the flow rate is relatively stable and the oil property curve is stable without burrs.

较佳的，步骤三所述的方法，采集到的光谱采用基线校正方法进行预处理，如峰谷扣减法，即：每一条新的光谱要减去光谱峰谷处的吸光度的平均值。Preferably, in the method described in step 3, the collected spectra are preprocessed by a baseline correction method, such as the peak-to-valley subtraction method, that is, the average value of the absorbance at the peak and valley of the spectrum is subtracted from each new spectrum.

较佳的，步骤三所述的方法，用于建立校正模型的汽油近红外光谱的波长范围采用间隔偏最小二乘（IPLS）的方法选择。Preferably, in the method described in step 3, the wavelength range of the near-infrared spectrum of gasoline used to establish the calibration model is selected using the method of interval partial least squares (IPLS).

较佳的，步骤三所述的方法，校正模型采用恰当的回归方法建立（如采用偏最小二乘），预测属性与光谱之间的关系如式（1）Preferably, in the method described in step 3, the calibration model is established using an appropriate regression method (such as partial least squares), and the relationship between the predicted attribute and the spectrum is as in formula (1)

y=a₀+a₁x₁+a₂x₂+...+a_nx_n （1）y=a₀ +a₁ x₁ +a₂ x₂ +...+a_n x_n (1)

其中：ｙ为预测属性，a_i为模型参数，x_i为光谱第i个波长点的吸光度Where: y is the predicted attribute, a_i is the model parameter, x_i is the absorbance of the ith wavelength point of the spectrum

本发明具有以下有益效果：（1）本发明采用在线近红外分析仪，具有工业现场应用无需样品预处理、分析速度快、信噪比高等优点。本发明可以同时检测汽油的研究法辛烷值、马达法辛烷值、干点、苯含量、氧含量、烯烃、芳烃、蒸汽压；解决了常规分析方法的费时、浪费问题，提高了分析效率，是汽油属性在线检测的新型有效的方法。（2）利用近红外光谱分析技术分析汽油的属性，结合偏最小二乘方法建立汽油属性与近红外光谱的校正模型，通过模型预测未知样本，结果可靠、精度高。因此，可以将该技术进行推广，尤其是在汽油管道调和过程中汽油属性的在线实时预测，可以促进汽油调和工艺以及油品的升级。The present invention has the following beneficial effects: (1) The present invention adopts an online near-infrared analyzer, which has the advantages of no need for sample pretreatment in industrial field application, fast analysis speed, and high signal-to-noise ratio. The invention can simultaneously detect gasoline's research octane number, motor octane number, dry point, benzene content, oxygen content, olefins, aromatics, and vapor pressure; solves the time-consuming and wasteful problems of conventional analysis methods, and improves analysis efficiency , is a new and effective method for online detection of gasoline properties. (2) Using near-infrared spectrum analysis technology to analyze the properties of gasoline, combined with the partial least squares method to establish a correction model for gasoline properties and near-infrared spectra, and predicting unknown samples through the model, the results are reliable and high-precision. Therefore, this technology can be promoted, especially the online real-time prediction of gasoline properties in the process of gasoline pipeline blending, which can promote the upgrading of gasoline blending process and oil products.

附图说明Description of drawings

图1：汽油近红外原始谱图；Figure 1: The original near-infrared spectrum of gasoline;

图2：预处理后的汽油近红外谱图；Figure 2: Near-infrared spectrum of gasoline after pretreatment;

图3：近红外研究法辛烷值回归模型（数据经过归一化处理）；Figure 3: NIR research method octane number regression model (data normalized);

图4；实时预测方法流程图。Figure 4; Flowchart of the real-time prediction method.

具体实施方式Detailed ways

下面结合附图和实施例对本发明做进一步说明。The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

以下辛烷值预测的实施例用来说明本发明，但不用来限制本发明的范围。The following examples of octane prediction serve to illustrate the invention, but are not intended to limit the scope of the invention.

请参阅图4，为实施例1基于近红外光谱检测的汽油油品属性实时预测方法流程图，具体包括以下步骤：Please refer to Figure 4, which is a flow chart of the real-time prediction method for gasoline oil properties based on near-infrared spectrum detection in Example 1, which specifically includes the following steps:

步骤一：收集100个样品（汽油管道调和工业工程获得），随机分为校正集和验证集（本实施例中比例为4：1）。其中，近红外光谱采用在线近红外仪器，扫描4次取平均值；光谱扫描范围：1000-1600nm，分辨率1nm，见图1，为该汽油近红外原始谱图。较佳的，光谱采集应该在流量相对稳定，油品属性曲线平稳无毛刺的情况下获取汽油近红外光谱。Step 1: Collect 100 samples (obtained from gasoline pipeline reconciliation industrial engineering), and randomly divide them into a calibration set and a verification set (the ratio is 4:1 in this example). Among them, the near-infrared spectrum adopts an online near-infrared instrument, and the average value is obtained by scanning 4 times; the spectral scanning range: 1000-1600nm, and the resolution is 1nm. See Figure 1, which is the original near-infrared spectrum of the gasoline. Preferably, the spectral collection should obtain the near-infrared spectrum of gasoline when the flow rate is relatively stable and the oil property curve is smooth and burr-free.

步骤二：采用标准方法测定汽油样品的研究法辛烷值(RON)属性，并对步骤一中的汽油光谱采用峰谷扣减的方法进行基线校正，并选取1181-1200nm范围的波长用于建立模型。较佳的，峰谷扣减的方法为：每一条新的光谱要减去1088-1092nm,1298-1312nm10个波长处吸光度的平均值。请参阅图1和图2,Step 2: Use standard methods to determine the research octane number (RON) properties of gasoline samples, and perform baseline correction on the gasoline spectra instep 1 using the method of peak-to-valley subtraction, and select wavelengths in the range of 1181-1200nm to establish Model. Preferably, the method of peak-to-valley subtraction is as follows: subtract the average value of absorbance at 10 wavelengths of 1088-1092nm and 1298-1312nm from each new spectrum. Please refer to Figure 1 and Figure 2,

其中图1为原始光谱，光谱的基线漂移严重；图2为基线校正后的谱图，基线漂移基本消除。Among them, Figure 1 is the original spectrum, and the baseline drift of the spectrum is serious; Figure 2 is the spectrum after baseline correction, and the baseline drift is basically eliminated.

步骤三：建立汽油辛烷值与近红外光谱的回归模型。校正模型采用恰当的回归方法建立（本实施例采用偏最小二乘），预测属性与光谱之间的关系如式（1）Step 3: Establish a regression model between gasoline octane number and near-infrared spectrum. The calibration model is established by an appropriate regression method (partial least squares is used in this example), and the relationship between the predicted attribute and the spectrum is shown in formula (1)

ｙ为预测属性，a_i为模型参数，x_i为光谱第i个波长点的吸光度y is the predicted attribute, a_i is the model parameter, x_i is the absorbance of the ith wavelength point of the spectrum

步骤四：利用验证集对模型进行验证。决定系数达0.912，交互验证均方误差为0.15Step 4: Validate the model using the validation set. The coefficient of determination is 0.912, and the mean square error of cross-validation is 0.15

步骤五：输出汽油辛烷值与近红外光谱的回归模型系数a_i，用于汽油管道调和过程的监控和优化控制。Step 5: output the regression model coefficient a_i of gasoline octane number and near-infrared spectrum, which is used for monitoring and optimal control of gasoline pipeline blending process.

基于上述方法，近红外法的研究法辛烷值预测值与实际值的比较结果如表1所示,该模型达预测准确、高效。Based on the above method, the comparison results of the octane number predicted value and the actual value of the near-infrared method are shown in Table 1. The model is accurate and efficient.

表1本发明方法的预测值与实际值结果Predicted value and actual value result of table 1 inventive method

本实施例以分析汽油研究法辛烷值（RON）属性为例，但也可以用于汽油其他属性的预测，如：马达法辛烷值，抗爆指数，雷德蒸汽压，硫含量，密度，烯烃含量，芳烃含量，或/和苯含量。This example takes the analysis of gasoline research method octane number (RON) properties as an example, but it can also be used to predict other properties of gasoline, such as: motor method octane number, antiknock index, Reid vapor pressure, sulfur content, density , olefin content, aromatic content, or/and benzene content.

综上所述仅为发明的较佳实施例而已，并非用来限定本发明的实施范围。即凡依本发明申请专利范围的内容所作的等效变化与修饰，都应为本发明的技术范畴。In summary, the above are only preferred embodiments of the invention, and are not intended to limit the implementation scope of the invention. That is, all equivalent changes and modifications made according to the content of the patent scope of the present invention shall be within the technical scope of the present invention.

Claims

1. the oil product of gasoline attribute real-time predicting method detecting based near infrared spectrum, is characterized in that, comprises the following steps:

Step 1: gather gasoline sample, utilize laboratory method to record the required attribute of gasoline;

Step 2: the near infrared spectrum of gasoline sample described in obtaining step one;

Step 3: the described near infrared spectrum obtaining in step 2 is carried out to pre-service, and chosen wavelength range, sets up near-infrared model check between property value and near infrared spectrum;

Step 4: adopt the near-infrared model of setting up in step 3 to predict each property value.

2. the method for claim 1, is characterized in that, the quantity of gasoline sample described in step 1 is more than or equal to 30.

3. the method for claim 1, is characterized in that, adopts on-line nir system to gather the near infrared spectrum of gasoline sample in step 2.

4. method as claimed in claim 3, is characterized in that, the near infrared spectrum scanning scope of on-line nir system described in step 2 is 1000-1600nm, and replication four times, is averaged near infrared spectrum value.

5. the method for claim 1, is characterized in that, preprocess method described in step 3 is baseline correction method.

5. the method for claim 1, is characterized in that, the range of choice of wavelength described in step 3 is any effective wavelength within 1000-1600nm.

6. the method for claim 1, is characterized in that, near-infrared model described in step 3 adopts homing method to set up.

7. method as claimed in claim 6, is characterized in that, described homing method adopts offset minimum binary method.