CN113962081B - Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information - Google Patents

Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information Download PDF

Info

Publication number
CN113962081B
CN113962081B CN202111222841.3A CN202111222841A CN113962081B CN 113962081 B CN113962081 B CN 113962081B CN 202111222841 A CN202111222841 A CN 202111222841A CN 113962081 B CN113962081 B CN 113962081B
Authority
CN
China
Prior art keywords
energy consumption
per ton
variance
consumption per
auxiliary measurement
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202111222841.3A
Other languages
Chinese (zh)
Other versions
CN113962081A (en
Inventor
栾小丽
高爽
赵顺毅
倪雨青
刘飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangnan University
Original Assignee
Jiangnan University
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 Jiangnan University filed Critical Jiangnan University
Priority to CN202111222841.3A priority Critical patent/CN113962081B/en
Publication of CN113962081A publication Critical patent/CN113962081A/en
Application granted granted Critical
Publication of CN113962081B publication Critical patent/CN113962081B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P80/00—Climate change mitigation technologies for sector-wide applications
    • Y02P80/10—Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Development Economics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Game Theory and Decision Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Air Conditioning Control Device (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明公开了一种基于辅助量测信息的精馏塔单吨能耗估计方法及系统，该方法包括以下步骤：S1、构建精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；S2、求解最优的精馏塔单吨能耗预测分布；S3、对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；S4、更新模态概率；S5、将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差。本发明基于基于辅助量测信息的精馏塔单吨能耗估计方法借助于高质量的辅助测量数据，从贝叶斯状态估计的角度，充分利用更有价值的辅助测量分布，大幅提升精馏塔单吨能耗估计的精度。

Figure 202111222841

The invention discloses a method and system for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information. The method includes the following steps: S1. Constructing a state space model of energy consumption per ton of a rectifying tower, and using Bayesian state estimation Calculate the Bayesian posterior distribution of the heat load data and auxiliary measurement data introduced into the reboiler; S2, solve the optimal prediction distribution of the energy consumption per ton of the rectification column; Update the predicted values of the mean and variance; S4, update the modal probability; S5, fuse the updated mean, variance and modal probability to obtain the final Bayesian estimated value, that is, the single ton energy of the distillation column to be estimated. The mean and variance of the cost. The method for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information in the present invention makes full use of more valuable auxiliary measurement distributions from the perspective of Bayesian state estimation by means of high-quality auxiliary measurement data, and greatly improves rectification Accuracy of energy consumption estimates per ton of towers.

Figure 202111222841

Description

一种基于辅助量测信息的精馏塔单吨能耗估计方法及系统A method and system for estimating energy consumption per ton of distillation column based on auxiliary measurement information

技术领域technical field

本发明涉及系统工程信号处理技术领域，特别涉及一种基于辅助量测信息的精馏塔单吨能耗估计方法及系统。The invention relates to the technical field of system engineering signal processing, in particular to a method and system for estimating energy consumption per ton of a distillation column based on auxiliary measurement information.

背景技术Background technique

石化行业中的精馏塔是生产国家战略性物资不可替代的重大耗能设备，实现重大耗能设备的节能是实现制造业节能减排的关键。而精馏塔单吨能耗只能在生产过程结束后获得，难以在线检测，因此如何在现有检测信息的基础上，通过状态估计手段对精馏塔单吨能耗进行估计尤显重要。现有状态估计方法多样，从贝叶斯状态估计，到卡尔曼状态估计，再到有限脉冲状态估计，涌现了出了许多优异的成果，并且在各行各业都发挥出了让人无法忽视的作用。The rectification tower in the petrochemical industry is an irreplaceable major energy-consuming equipment for the production of national strategic materials, and realizing the energy saving of the major energy-consuming equipment is the key to realizing energy saving and emission reduction in the manufacturing industry. The energy consumption per ton of the rectifying tower can only be obtained after the production process, and it is difficult to detect online. Therefore, it is particularly important to estimate the energy consumption per ton of the rectifying tower by means of state estimation on the basis of the existing detection information. There are various existing state estimation methods, from Bayesian state estimation, to Kalman state estimation, to finite impulse state estimation, many excellent results have emerged, and they have played a role in all walks of life. effect.

近年来，为了获得更精确的精馏塔单吨能耗估计值，各种融合策略被相继提出，其中track to track方法利用相关性将不同精馏塔单吨能耗估计器的估计值进行融合，广泛应用于融合领域。后来，各种延伸的融合估计方法被陆续提出。虽然所提的各种融合策略可以借助于额外的信息，提升能耗估计的精度，但其本质是对估计结果的折衷与加权，即额外信息的利用并没有改变单吨能耗估计器的本质结构，从而导致单吨能耗估计的精度较低。In recent years, in order to obtain more accurate estimates of energy consumption per ton of rectifying towers, various fusion strategies have been proposed. The track to track method uses correlation to fuse the estimated values of energy consumption per ton of different rectifying towers. , which is widely used in the field of fusion. Later, various extended fusion estimation methods have been proposed. Although the various fusion strategies proposed can improve the accuracy of energy consumption estimation with the help of additional information, their essence is the compromise and weighting of the estimation results, that is, the use of additional information does not change the essence of the single-ton energy consumption estimator structure, resulting in lower accuracy of energy consumption estimates per ton.

发明内容SUMMARY OF THE INVENTION

本发明要解决的技术问题是提供一种借助辅助测量数据、提高精馏塔单吨能耗估计精度的基于辅助量测信息的精馏塔单吨能耗估计方法及系统。The technical problem to be solved by the present invention is to provide a method and system for estimating energy consumption per ton of rectifying tower based on auxiliary measurement information, which improves the estimation accuracy of energy consumption per ton of rectifying tower by means of auxiliary measurement data.

为了解决上述问题，本发明提供了一种基于辅助量测信息的精馏塔单吨能耗估计方法，其包括以下步骤：In order to solve the above problems, the present invention provides a method for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information, which comprises the following steps:

S1、构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；S1. Construct a single-ton energy consumption state space model of the rectification column, and use Bayesian state estimation to calculate the Bayesian heat load data and auxiliary measurement data introduced into the reboiler based on the constructed single-ton energy consumption state space model of the rectification column. posterior distribution;

S2、利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；S2. Using the similarity measure, calculate the objective similarity between the Bayesian posterior distribution introduced in the reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution, and aim to maximize the objective similarity , to solve the optimal prediction distribution of energy consumption per ton of distillation column;

S3、对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；S3. Perform a mixed interaction on the mean and variance of the energy consumption per ton of the rectifying tower, calculate the predicted value of the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction, correct the predicted value, and perform a correction on the corrected energy consumption. Update the predicted value of the mean and variance of the energy consumption per ton of the distillation column;

S4、利用贝叶斯公式计算模态的后验分布以更新模态概率；S4. Use the Bayesian formula to calculate the posterior distribution of the mode to update the mode probability;

S5、将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差。S5. Integrate the updated mean and variance and the modal probability to obtain the final Bayesian estimate, that is, the mean and variance of the energy consumption per ton of the distillation column to be estimated.

作为本发明的进一步改进，采用交互式多模型方法、n阶广义伪贝叶斯算法或变结构多模型算法求解精馏塔单吨能耗的均值和方差的混合交互。As a further improvement of the present invention, an interactive multi-model method, an n-order generalized pseudo-Bayesian algorithm or a variable structure multi-model algorithm is used to solve the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column.

作为本发明的进一步改进，所述客观相似度为距离测度、相似测度或匹配测度。As a further improvement of the present invention, the objective similarity is a distance measure, a similarity measure or a matching measure.

作为本发明的进一步改进，步骤S1中，构建的精馏塔单吨能耗状态空间模型如下：As a further improvement of the present invention, in step S1, the constructed rectifying tower energy consumption state space model per ton is as follows:

xk＝F(rk)xk-1+G(rk)wk,x k =F(r k )x k-1 +G(r k )w k ,

Figure BDA0003313261950000021
Figure BDA0003313261950000021

Figure BDA0003313261950000022
Figure BDA0003313261950000022

式中，

Figure BDA0003313261950000023
为精馏塔单吨能耗变量，
Figure BDA0003313261950000024
和
Figure BDA0003313261950000025
分别代表再沸器热负荷数据和辅助测量数据，定义
Figure BDA0003313261950000026
和
Figure BDA0003313261950000027
分别表示再沸器热负荷序列集合和辅助测量序列集合，rk表示在有限空间M＝{1,2,…,M}中取值的离散齐次马尔可夫链，对于任意i,j∈M转移概率定义为
Figure BDA0003313261950000028
F(rk),G(rk)和H(rk)表示rk相关的模型矩阵，噪声项wk～N(wk；0,Qk)，
Figure BDA0003313261950000029
和
Figure BDA00033132619500000210
为独立同分布的高斯噪声，Qk为过程噪声方差，
Figure BDA00033132619500000211
和
Figure BDA00033132619500000212
分别分为再沸器热负荷噪声方差和辅助测量方差，假设初始分布为
Figure BDA00033132619500000213
其中
Figure BDA00033132619500000214
表示均值为
Figure BDA00033132619500000215
方差为P的高斯分布，为了符号简化，定义
Figure BDA00033132619500000216
为rk的第j个模态，
Figure BDA00033132619500000217
以及
Figure BDA00033132619500000218
In the formula,
Figure BDA0003313261950000023
is the energy consumption variable per ton of distillation column,
Figure BDA0003313261950000024
and
Figure BDA0003313261950000025
represent the reboiler heat load data and auxiliary measurement data, respectively, define
Figure BDA0003313261950000026
and
Figure BDA0003313261950000027
represent the reboiler heat load sequence set and the auxiliary measurement sequence set, respectively, r k represents the discrete homogeneous Markov chain taking values in the finite space M={1,2,...,M}, for any i,j∈ The M transition probability is defined as
Figure BDA0003313261950000028
F(r k ), G(r k ) and H(r k ) represent the model matrix related to r k , the noise term w k ～N(w k ; 0, Q k ),
Figure BDA0003313261950000029
and
Figure BDA00033132619500000210
is the independent and identically distributed Gaussian noise, Q k is the process noise variance,
Figure BDA00033132619500000211
and
Figure BDA00033132619500000212
are divided into reboiler heat load noise variance and auxiliary measurement variance, respectively, assuming that the initial distribution is
Figure BDA00033132619500000213
in
Figure BDA00033132619500000214
means that the mean is
Figure BDA00033132619500000215
A Gaussian distribution with variance P, for notation simplification, defines
Figure BDA00033132619500000216
is the jth mode of r k ,
Figure BDA00033132619500000217
as well as
Figure BDA00033132619500000218

作为本发明的进一步改进，利用贝叶斯状态估计计算包含再沸器热负荷数据和辅助测量数据的贝叶斯后验分布，如下：As a further improvement of the present invention, the Bayesian posterior distribution including reboiler heat load data and auxiliary measurement data is calculated using Bayesian state estimation, as follows:

Figure BDA0003313261950000031
Figure BDA0003313261950000031

式中

Figure BDA0003313261950000032
表示再沸器热负荷序列和辅助测量序列总集合。in the formula
Figure BDA0003313261950000032
Represents the total set of reboiler heat duty sequences and auxiliary measurement sequences.

作为本发明的进一步改进，步骤S2中，理想的贝叶斯后验分布如下：As a further improvement of the present invention, in step S2, the ideal Bayesian posterior distribution is as follows:

Figure BDA0003313261950000033
Figure BDA0003313261950000033

其中

Figure BDA0003313261950000034
为似然分布，
Figure BDA0003313261950000035
为精馏塔单吨能耗的预测分布；in
Figure BDA0003313261950000034
is the likelihood distribution,
Figure BDA0003313261950000035
is the predicted distribution of energy consumption per ton of distillation column;

引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布如下：The Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data is as follows:

Figure BDA0003313261950000036
Figure BDA0003313261950000036

其中

Figure BDA0003313261950000037
定义为引入辅助数据后的似然分布，
Figure BDA0003313261950000038
为引入辅助数据后需要优化求解的精馏塔单吨能耗的预测分布；in
Figure BDA0003313261950000037
is defined as the likelihood distribution after introducing auxiliary data,
Figure BDA0003313261950000038
Predicted distribution of energy consumption per ton of distillation column that needs to be optimized after the introduction of auxiliary data;

利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，如下：Using the similarity measure, the objective similarity between the Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution is calculated as follows:

Figure BDA0003313261950000039
Figure BDA0003313261950000039

并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布如下：With the objective of maximizing the objective similarity, the optimal prediction distribution of energy consumption per ton of distillation tower is as follows:

Figure BDA00033132619500000310
Figure BDA00033132619500000310

其中exp(·)代表指数函数，Ef(·)[g(·)]表示计算g(·)分布关于f(·)分布的期望。where exp( ) represents the exponential function, and E f( ) [g( )] represents the expectation of computing the g( ) distribution with respect to the f( ) distribution.

作为本发明的进一步改进，步骤S3包括：As a further improvement of the present invention, step S3 includes:

S31、求取精馏塔单吨能耗的均值和方差的混合交互：S31. Obtain the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column:

Figure BDA00033132619500000311
Figure BDA00033132619500000311

Figure BDA00033132619500000312
Figure BDA00033132619500000312

式中符号(…)表示与前一项相同的项，πij代表从k-1时刻模态i到k时刻模态j的转移概率，

Figure BDA0003313261950000041
为k-1时刻第i个模态的概率，
Figure BDA0003313261950000042
为模态j的预测概率，
Figure BDA0003313261950000043
为k-1时刻第i个模态下精馏塔单吨能耗的均值，
Figure BDA0003313261950000044
为混合交互后j模态下的混合精馏塔单吨能耗均值，
Figure BDA0003313261950000045
为k-1时刻第i个模态下精馏塔单吨能耗的方差，
Figure BDA0003313261950000046
为混合交互后j模态下的混合精馏塔单吨能耗方差；where the symbol (…) represents the same term as the previous term, π ij represents the transition probability from mode i at time k-1 to mode j at time k,
Figure BDA0003313261950000041
is the probability of the i-th mode at time k-1,
Figure BDA0003313261950000042
is the predicted probability of mode j,
Figure BDA0003313261950000043
is the average energy consumption per ton of the distillation column in the i-th mode at time k-1,
Figure BDA0003313261950000044
is the average energy consumption per ton of the mixed distillation column in j mode after mixing interaction,
Figure BDA0003313261950000045
is the variance of the energy consumption per ton of the distillation column in the ith mode at time k-1,
Figure BDA0003313261950000046
is the variance of energy consumption per ton of mixed distillation column in j mode after mixing interaction;

S32、根据预测步公式，计算精馏塔单吨能耗的均值和方差的预测值：S32, according to the prediction step formula, calculate the predicted value of the mean value and variance of the energy consumption per ton of the rectifying tower:

Figure BDA0003313261950000047
Figure BDA0003313261950000047

Figure BDA0003313261950000048
Figure BDA0003313261950000048

S33、根据求解的最优精馏塔单吨能耗预测分布，对上述预测值进行修正：S33, according to the predicted distribution of the optimal energy consumption per ton of the rectifying tower, correct the above predicted value:

Figure BDA0003313261950000049
Figure BDA0003313261950000049

Figure BDA00033132619500000410
Figure BDA00033132619500000410

其中

Figure BDA00033132619500000411
为利用辅助测量数据获得的估计值，
Figure BDA00033132619500000412
in
Figure BDA00033132619500000411
For estimates obtained using auxiliary measurement data,
Figure BDA00033132619500000412

S34、利用如下公式对修正后的精馏塔单吨能耗预测值进行更新：S34, use the following formula to update the revised predicted value of energy consumption per ton of the rectifying tower:

Figure BDA00033132619500000413
Figure BDA00033132619500000413

Figure BDA00033132619500000414
Figure BDA00033132619500000414

其中

Figure BDA00033132619500000415
in
Figure BDA00033132619500000415

作为本发明的进一步改进，步骤S4包括：As a further improvement of the present invention, step S4 includes:

利用贝叶斯公式计算模态的贝叶斯后验分布：Use the Bayesian formula to calculate the Bayesian posterior distribution of the modes:

Figure BDA00033132619500000416
Figure BDA00033132619500000416

其中

Figure BDA00033132619500000417
in
Figure BDA00033132619500000417

作为本发明的进一步改进，在步骤S5中，最终估计的精馏塔单吨能耗的均值和方差如下：As a further improvement of the present invention, in step S5, the mean and variance of the final estimated energy consumption per ton of the rectifying tower are as follows:

Figure BDA00033132619500000418
Figure BDA00033132619500000418

Figure BDA0003313261950000051
Figure BDA0003313261950000051

其中

Figure BDA0003313261950000052
为最终估计的精馏塔单吨能耗均值，Pk表示最终估计的精馏塔单吨能耗方差。in
Figure BDA0003313261950000052
is the final estimated average energy consumption per ton of the rectifying tower, and P k represents the variance of the final estimated energy consumption per ton of the rectifying tower.

本发明还提供了一种基于辅助量测信息的精馏塔单吨能耗估计系统，其包括：The present invention also provides a system for estimating energy consumption per ton of distillation column based on auxiliary measurement information, which includes:

模型构建模块，用于构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；The model building module is used to construct a single ton energy consumption state space model of the rectification column. Based on the constructed single ton energy consumption state space model of the rectification column, the reboiler heat load data and auxiliary measurement data are introduced by Bayesian state estimation calculation. The Bayesian posterior distribution of ;

最优精馏塔单吨能耗预测分布求解模块，用于利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；The optimal rectification tower single-ton energy consumption prediction distribution solution module is used to calculate the difference between the Bayesian posterior distribution and the ideal Bayesian posterior distribution of the reboiler heat load data and auxiliary measurement data using the similarity measure. The objective similarity between the two, and the objective is to maximize the objective similarity to solve the optimal prediction distribution of energy consumption per ton of rectifying tower;

精馏塔单吨能耗的均值和方差的预测值更新模块，用于对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；The update module for the prediction value of the mean and variance of the energy consumption per ton of the rectifying tower, which is used to mix and interact with the mean and variance of the energy consumption per ton of the rectifying tower, and calculate the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction. The predicted value of the rectification tower is revised, and the revised predicted value of the average value and variance of the energy consumption per ton of the rectifying tower is updated;

模态概率更新模块，用于利用贝叶斯公式计算模态的后验分布以更新模态概率；The modal probability update module is used to calculate the posterior distribution of the modal using the Bayesian formula to update the modal probability;

精馏塔单吨能耗估计结果输出模块，用于将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差。The output module for the estimation result of energy consumption per ton of rectification tower is used to fuse the updated mean and variance and modal probability to obtain the final Bayesian estimate, that is, the mean value and sum of energy consumption per ton of rectification tower to be estimated variance.

本发明的有益效果：Beneficial effects of the present invention:

本发明基于基于辅助量测信息的精馏塔单吨能耗估计方法及系统借助于高质量的辅助测量数据，从贝叶斯状态估计的角度，充分利用更有价值的辅助测量分布，大幅提升精馏塔单吨能耗估计的精度。The present invention is based on the method and system for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information by means of high-quality auxiliary measurement data, and from the perspective of Bayesian state estimation, makes full use of more valuable auxiliary measurement distribution, and greatly improves the Accuracy of energy consumption estimates per ton of distillation towers.

上述说明仅是本发明技术方案的概述，为了能够更清楚了解本发明的技术手段，而可依照说明书的内容予以实施，并且为了让本发明的上述和其他目的、特征和优点能够更明显易懂，以下特举较佳实施例，并配合附图，详细说明如下。The above description is only an overview of the technical solutions of the present invention, in order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the description, and in order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand , the following specific preferred embodiments, and in conjunction with the accompanying drawings, are described in detail as follows.

附图说明Description of drawings

图1是本发明优选实施例中基于基于辅助量测信息的精馏塔单吨能耗估计方法的流程图；1 is a flow chart of a method for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information in a preferred embodiment of the present invention;

图2是不同估计方法的均方根误差图。Figure 2 is a plot of RMSE for different estimation methods.

具体实施方式Detailed ways

下面结合附图和具体实施例对本发明作进一步说明，以使本领域的技术人员可以更好地理解本发明并能予以实施，但所举实施例不作为对本发明的限定。The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

如图1所示，为本发明优选实施例中基于基于辅助量测信息的精馏塔单吨能耗估计方法，包括以下步骤：As shown in Figure 1, it is a method for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information in a preferred embodiment of the present invention, comprising the following steps:

S1、构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；S1. Construct a single-ton energy consumption state space model of the rectification column, and use Bayesian state estimation to calculate the Bayesian heat load data and auxiliary measurement data introduced into the reboiler based on the constructed single-ton energy consumption state space model of the rectification column. posterior distribution;

可选的，构建的精馏塔单吨能耗状态空间模型可以是线性(非线性)时不变系统，也可以是随机多模态系统。Optionally, the constructed state space model of single-ton energy consumption of the rectification tower may be a linear (non-linear) time-invariant system, or a random multi-modal system.

具体地，再沸器热负荷数据是利用模型的输出获得的再沸器热负荷值，辅助测量数据是根据导热油流量、进出口温差等计算得到的精确的热负荷数据。Specifically, the reboiler heat load data is the reboiler heat load value obtained by using the output of the model, and the auxiliary measurement data is the accurate heat load data calculated according to the heat transfer oil flow rate, inlet and outlet temperature difference, etc.

可选的，构建的精馏塔单吨能耗状态空间模型如下：Optionally, the constructed state space model of single ton energy consumption of the rectifying tower is as follows:

xk＝F(rk)xk-1+G(rk)wk,x k =F(r k )x k-1 +G(r k )w k ,

Figure BDA0003313261950000061
Figure BDA0003313261950000061

Figure BDA0003313261950000062
Figure BDA0003313261950000062

式中，

Figure BDA0003313261950000063
为精馏塔单吨能耗变量，
Figure BDA0003313261950000064
和
Figure BDA0003313261950000065
分别代表再沸器热负荷数据和辅助测量数据，定义
Figure BDA0003313261950000066
和
Figure BDA0003313261950000067
分别表示再沸器热负荷序列集合和辅助测量序列集合，rk表示在有限空间M＝{1,2,…,M}中取值的离散齐次马尔可夫链，对于任意i,j∈M转移概率定义为
Figure BDA0003313261950000068
F(rk),G(rk)和H(rk)表示rk相关的模型矩阵，噪声项wk～N(wk；0,Qk)，
Figure BDA0003313261950000069
和
Figure BDA0003313261950000071
为独立同分布的高斯噪声，Qk为过程噪声方差，
Figure BDA0003313261950000072
和
Figure BDA0003313261950000073
分别分为再沸器热负荷噪声方差和辅助测量方差，假设初始分布为
Figure BDA0003313261950000074
其中
Figure BDA0003313261950000075
表示均值为
Figure BDA0003313261950000076
方差为P的高斯分布，为了符号简化，定义
Figure BDA0003313261950000077
为rk的第j个模态，
Figure BDA0003313261950000078
以及
Figure BDA0003313261950000079
In the formula,
Figure BDA0003313261950000063
is the energy consumption variable per ton of distillation column,
Figure BDA0003313261950000064
and
Figure BDA0003313261950000065
represent the reboiler heat load data and auxiliary measurement data, respectively, define
Figure BDA0003313261950000066
and
Figure BDA0003313261950000067
represent the reboiler heat load sequence set and the auxiliary measurement sequence set, respectively, r k represents the discrete homogeneous Markov chain taking values in the finite space M={1,2,...,M}, for any i,j∈ The M transition probability is defined as
Figure BDA0003313261950000068
F(r k ), G(r k ) and H(r k ) represent the model matrix related to r k , the noise term w k ～N(w k ; 0, Q k ),
Figure BDA0003313261950000069
and
Figure BDA0003313261950000071
is the independent and identically distributed Gaussian noise, Q k is the process noise variance,
Figure BDA0003313261950000072
and
Figure BDA0003313261950000073
are divided into reboiler heat load noise variance and auxiliary measurement variance, respectively, assuming that the initial distribution is
Figure BDA0003313261950000074
in
Figure BDA0003313261950000075
means that the mean is
Figure BDA0003313261950000076
A Gaussian distribution with variance P, for notation simplification, defines
Figure BDA0003313261950000077
is the jth mode of r k ,
Figure BDA0003313261950000078
as well as
Figure BDA0003313261950000079

进一步地，利用贝叶斯状态估计计算包含再沸器热负荷数据和辅助测量数据的贝叶斯后验分布，如下：Further, the Bayesian posterior distribution containing reboiler heat load data and auxiliary measurement data is calculated using Bayesian state estimation, as follows:

Figure BDA00033132619500000710
Figure BDA00033132619500000710

式中

Figure BDA00033132619500000711
表示再沸器热负荷序列和辅助测量序列总集合。in the formula
Figure BDA00033132619500000711
Represents the total set of reboiler heat duty sequences and auxiliary measurement sequences.

S2、利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；S2. Using the similarity measure, calculate the objective similarity between the Bayesian posterior distribution introduced in the reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution, and aim to maximize the objective similarity , to solve the optimal prediction distribution of energy consumption per ton of distillation column;

可选的，所述客观相似度为距离测度、相似测度或匹配测度。Optionally, the objective similarity is a distance measure, a similarity measure or a matching measure.

其中，相似性度量是为了计算理想的贝叶斯后验分布和引入辅助数据的贝叶斯后验分布之间的距离，可以用距离测度，如欧氏距离、绝对值距离、马氏距离、KL散度、JS散度；也可以用相似测度，如角度相似系数、相关系数、指数相似系数、Pearson系数；也可以用匹配测度，如Tanimoto测度，Rao测度，简单匹配系数，Dice系数，Kulzinsky系数。Among them, the similarity measure is to calculate the distance between the ideal Bayesian posterior distribution and the Bayesian posterior distribution introduced with auxiliary data. Distance measures such as Euclidean distance, absolute value distance, Mahalanobis distance, KL divergence, JS divergence; similarity measures can also be used, such as angle similarity coefficient, correlation coefficient, exponential similarity coefficient, Pearson coefficient; matching measures can also be used, such as Tanimoto measure, Rao measure, simple matching coefficient, Dice coefficient, Kulzinsky coefficient.

可选的，理想的贝叶斯后验分布如下：Optionally, the ideal Bayesian posterior distribution is as follows:

Figure BDA00033132619500000712
Figure BDA00033132619500000712

其中

Figure BDA00033132619500000713
为似然分布，
Figure BDA00033132619500000714
为精馏塔单吨能耗的预测分布；in
Figure BDA00033132619500000713
is the likelihood distribution,
Figure BDA00033132619500000714
is the predicted distribution of energy consumption per ton of distillation column;

引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布如下：The Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data is as follows:

Figure BDA00033132619500000715
Figure BDA00033132619500000715

其中

Figure BDA00033132619500000716
定义为引入辅助数据后的似然分布，
Figure BDA00033132619500000717
为引入辅助数据后需要优化求解的精馏塔单吨能耗的预测分布；in
Figure BDA00033132619500000716
is defined as the likelihood distribution after introducing auxiliary data,
Figure BDA00033132619500000717
Predicted distribution of energy consumption per ton of distillation column that needs to be optimized after the introduction of auxiliary data;

利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，如下：Using the similarity measure, the objective similarity between the Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution is calculated as follows:

Figure BDA0003313261950000081
Figure BDA0003313261950000081

并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布如下：With the objective of maximizing the objective similarity, the optimal prediction distribution of energy consumption per ton of distillation tower is as follows:

Figure BDA0003313261950000082
Figure BDA0003313261950000082

其中exp(·)代表指数函数，Ef(·)[g(·)]表示计算g(·)分布关于f(·)分布的期望。where exp( ) represents the exponential function, and E f( ) [g( )] represents the expectation of computing the g( ) distribution with respect to the f( ) distribution.

S3、对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；S3. Perform a mixed interaction on the mean and variance of the energy consumption per ton of the rectifying tower, calculate the predicted value of the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction, correct the predicted value, and perform a correction on the corrected energy consumption. Update the predicted value of the mean and variance of the energy consumption per ton of the distillation column;

可选的，可用交互式多模型方法、n阶广义伪贝叶斯算法或变结构多模型算法等求解精馏塔单吨能耗的均值和方差的混合交互。Optionally, an interactive multi-model method, an n-order generalized pseudo-Bayesian algorithm or a variable structure multi-model algorithm, etc. can be used to solve the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column.

示例地，对均值和方差的预测和更新可用卡尔曼滤波方法、最大似然、最大先验、粒子滤波、分布式卡尔曼滤波、分布式粒子滤波或协方差一致方法。Illustratively, prediction and updating of the mean and variance can be performed using Kalman filtering methods, maximum likelihood, maximum prior, particle filtering, distributed Kalman filtering, distributed particle filtering, or covariance consensus methods.

可选的，步骤S3包括：Optionally, step S3 includes:

S31、求取精馏塔单吨能耗的均值和方差的混合交互：S31. Obtain the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column:

Figure BDA0003313261950000083
Figure BDA0003313261950000083

Figure BDA0003313261950000084
Figure BDA0003313261950000084

式中符号(…)表示与前一项相同的项，πij代表从k-1时刻模态i到k时刻模态j的转移概率，

Figure BDA0003313261950000085
为k-1时刻第i个模态的概率，
Figure BDA0003313261950000086
为模态j的预测概率，
Figure BDA0003313261950000087
为k-1时刻第i个模态下精馏塔单吨能耗的均值，
Figure BDA0003313261950000088
为混合交互后j模态下的混合精馏塔单吨能耗均值，
Figure BDA0003313261950000089
为k-1时刻第i个模态下精馏塔单吨能耗的方差，
Figure BDA00033132619500000810
为混合交互后j模态下的混合精馏塔单吨能耗方差；where the symbol (…) represents the same term as the previous term, π ij represents the transition probability from mode i at time k-1 to mode j at time k,
Figure BDA0003313261950000085
is the probability of the i-th mode at time k-1,
Figure BDA0003313261950000086
is the predicted probability of mode j,
Figure BDA0003313261950000087
is the average energy consumption per ton of the distillation column in the i-th mode at time k-1,
Figure BDA0003313261950000088
is the average energy consumption per ton of the mixed distillation column in j mode after mixing interaction,
Figure BDA0003313261950000089
is the variance of the energy consumption per ton of the distillation column in the ith mode at time k-1,
Figure BDA00033132619500000810
is the variance of energy consumption per ton of mixed distillation column in j mode after mixing interaction;

S32、根据预测步公式，计算精馏塔单吨能耗的均值和方差的预测值：S32, according to the prediction step formula, calculate the predicted value of the mean value and variance of the energy consumption per ton of the rectifying tower:

Figure BDA00033132619500000811
Figure BDA00033132619500000811

Figure BDA0003313261950000091
Figure BDA0003313261950000091

S33、根据求解的最优精馏塔单吨能耗预测分布，对上述预测值进行修正：S33, according to the predicted distribution of the optimal energy consumption per ton of the rectifying tower, correct the above predicted value:

Figure BDA0003313261950000092
Figure BDA0003313261950000092

Figure BDA0003313261950000093
Figure BDA0003313261950000093

其中

Figure BDA0003313261950000094
为利用辅助测量数据获得的估计值，
Figure BDA0003313261950000095
in
Figure BDA0003313261950000094
For estimates obtained using auxiliary measurement data,
Figure BDA0003313261950000095

S34、利用如下公式对修正后的精馏塔单吨能耗预测值进行更新：S34, use the following formula to update the revised predicted value of energy consumption per ton of the rectifying tower:

Figure BDA0003313261950000096
Figure BDA0003313261950000096

Figure BDA0003313261950000097
Figure BDA0003313261950000097

其中

Figure BDA0003313261950000098
in
Figure BDA0003313261950000098

S4、利用贝叶斯公式计算模态的后验分布以更新模态概率；S4. Use the Bayesian formula to calculate the posterior distribution of the mode to update the mode probability;

可选的，步骤S4包括：Optionally, step S4 includes:

利用贝叶斯公式计算模态的贝叶斯后验分布：Use the Bayesian formula to calculate the Bayesian posterior distribution of the modes:

Figure BDA0003313261950000099
Figure BDA0003313261950000099

其中

Figure BDA00033132619500000910
in
Figure BDA00033132619500000910

S5、将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差。S5. Integrate the updated mean and variance and the modal probability to obtain the final Bayesian estimate, that is, the mean and variance of the energy consumption per ton of the distillation column to be estimated.

可选的，最终估计的精馏塔单吨能耗的均值和方差如下：Optionally, the mean and variance of the final estimated energy consumption per ton of the rectifying tower are as follows:

Figure BDA00033132619500000911
Figure BDA00033132619500000911

Figure BDA00033132619500000912
Figure BDA00033132619500000912

其中

Figure BDA00033132619500000913
为最终估计的精馏塔单吨能耗均值，Pk表示最终估计的精馏塔单吨能耗方差。in
Figure BDA00033132619500000913
is the final estimated average energy consumption per ton of the rectifying tower, and P k represents the variance of the final estimated energy consumption per ton of the rectifying tower.

如图2所示，采样时间为1秒，精馏塔单吨能耗变量初始值设定分别为100。传统方法a为不包含辅助量测信息的交互多模型算法，传统方法b为变结构交互多模型算法，传统方法c为交互多模型融合算法。通过对不同估计方法的均方根误差对比，可以发现本发明的基于辅助量测信息的精馏塔单吨能耗估计方法误差更小，精度更高。As shown in Figure 2, the sampling time is 1 second, and the initial value of the energy consumption variable per ton of the rectifying tower is set to 100 respectively. The traditional method a is an interactive multi-model algorithm without auxiliary measurement information, the traditional method b is a variable-structure interactive multi-model algorithm, and the traditional method c is an interactive multi-model fusion algorithm. By comparing the root mean square errors of different estimation methods, it can be found that the method for estimating energy consumption per ton of distillation column based on auxiliary measurement information of the present invention has smaller errors and higher precision.

本发明优选实施例还公开了一种基于辅助量测信息的精馏塔单吨能耗估计系统，其包括：A preferred embodiment of the present invention also discloses a system for estimating energy consumption per ton of distillation column based on auxiliary measurement information, which includes:

模型构建模块，用于构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；The model building module is used to construct a single ton energy consumption state space model of the rectification column. Based on the constructed single ton energy consumption state space model of the rectification column, the reboiler heat load data and auxiliary measurement data are introduced by Bayesian state estimation calculation. The Bayesian posterior distribution of ;

最优精馏塔单吨能耗预测分布求解模块，用于利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；The optimal rectification tower single-ton energy consumption prediction distribution solution module is used to calculate the difference between the Bayesian posterior distribution and the ideal Bayesian posterior distribution of the reboiler heat load data and auxiliary measurement data using the similarity measure. The objective similarity between the two, and the objective is to maximize the objective similarity to solve the optimal prediction distribution of energy consumption per ton of rectifying tower;

精馏塔单吨能耗的均值和方差的预测值更新模块，用于对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；The update module for the prediction value of the mean and variance of the energy consumption per ton of the rectifying tower, which is used to mix and interact with the mean and variance of the energy consumption per ton of the rectifying tower, and calculate the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction. The predicted value of the rectification tower is revised, and the revised predicted value of the average value and variance of the energy consumption per ton of the rectifying tower is updated;

模态概率更新模块，用于利用贝叶斯公式计算模态的后验分布以更新模态概率；The modal probability update module is used to calculate the posterior distribution of the modal using the Bayesian formula to update the modal probability;

精馏塔单吨能耗估计结果输出模块，用于将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差。The output module for the estimation result of energy consumption per ton of rectification tower is used to fuse the updated mean and variance and modal probability to obtain the final Bayesian estimate, that is, the mean value and sum of energy consumption per ton of rectification tower to be estimated variance.

本发明实施例中的基于辅助量测信息的精馏塔单吨能耗估计系统用于实现前述的基于辅助量测信息的精馏塔单吨能耗估计方法，因此该系统的具体实施方式可见前文中的基于辅助量测信息的精馏塔单吨能耗估计方法的实施例部分，所以，其具体实施方式可以参照相应的各个部分实施例的描述，在此不再展开介绍。The system for estimating energy consumption per ton of rectifying column based on auxiliary measurement information in the embodiment of the present invention is used to realize the aforementioned method for estimating energy consumption per ton of rectification column based on auxiliary measurement information. Therefore, the specific implementation of the system can be seen The foregoing examples of the method for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information are described above. Therefore, the specific implementation methods can be referred to the descriptions of the corresponding examples, and will not be described here.

另外，由于本实施例的基于辅助量测信息的精馏塔单吨能耗估计系统用于实现前述的基于辅助量测信息的精馏塔单吨能耗估计方法，因此其作用与上述方法的作用相对应，这里不再赘述。In addition, since the system for estimating energy consumption per ton of rectifying tower based on auxiliary measurement information in this embodiment is used to realize the aforementioned method for estimating energy consumption per ton of rectifying tower based on auxiliary measurement information, its function is the same as that of the above method. The functions are corresponding and will not be repeated here.

本发明基于基于辅助量测信息的精馏塔单吨能耗估计方法及系统借助于高质量的辅助测量数据，从贝叶斯状态估计的角度，充分利用更有价值的辅助测量分布，大幅提升精馏塔单吨能耗估计的精度。The present invention is based on the method and system for estimating energy consumption per ton of a rectifying tower based on auxiliary measurement information by means of high-quality auxiliary measurement data, and from the perspective of Bayesian state estimation, makes full use of more valuable auxiliary measurement distribution, and greatly improves the Accuracy of energy consumption estimates per ton of distillation towers.

以上实施例仅是为充分说明本发明而所举的较佳的实施例，本发明的保护范围不限于此。本技术领域的技术人员在本发明基础上所作的等同替代或变换，均在本发明的保护范围之内。本发明的保护范围以权利要求书为准。The above embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the present invention is subject to the claims.

Claims (8)

1.一种基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，包括以下步骤：1. a rectifying tower single ton energy consumption estimation method based on auxiliary measurement information, is characterized in that, comprises the following steps: S1、构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；S1. Construct a single-ton energy consumption state space model of the rectification column, and use Bayesian state estimation to calculate the Bayesian heat load data and auxiliary measurement data introduced into the reboiler based on the constructed single-ton energy consumption state space model of the rectification column. posterior distribution; S2、利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；S2. Using the similarity measure, calculate the objective similarity between the Bayesian posterior distribution introduced in the reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution, and aim to maximize the objective similarity , to solve the optimal prediction distribution of energy consumption per ton of distillation column; S3、对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；S3. Perform a mixed interaction on the mean and variance of the energy consumption per ton of the rectifying tower, calculate the predicted value of the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction, correct the predicted value, and perform a correction on the corrected energy consumption. Update the predicted value of the mean and variance of the energy consumption per ton of the distillation column; S4、利用贝叶斯公式计算模态的后验分布以更新模态概率；S4. Use the Bayesian formula to calculate the posterior distribution of the mode to update the mode probability; S5、将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差；S5, fuse the updated mean and variance and the modal probability to obtain the final Bayesian estimate, that is, the mean and variance of the energy consumption per ton of the distillation column to be estimated; 构建的精馏塔单吨能耗状态空间模型如下：The constructed state space model of single ton energy consumption of distillation column is as follows: xk＝F(rk)xk-1+G(rk)wk,x k =F(r k )x k-1 +G(r k )w k ,
Figure FDA0003601973230000011
Figure FDA0003601973230000011
Figure FDA0003601973230000012
Figure FDA0003601973230000012
 式中，
Figure FDA0003601973230000013
为精馏塔单吨能耗变量，
Figure FDA0003601973230000014
和
Figure FDA0003601973230000015
分别代表再沸器热负荷数据和辅助测量数据，定义
Figure FDA0003601973230000016
和
Figure FDA0003601973230000017
分别表示再沸器热负荷序列集合和辅助测量序列集合，rk表示在有限空间
Figure FDA0003601973230000018
中取值的离散齐次马尔可夫链，对于任意
Figure FDA0003601973230000019
转移概率定义为
Figure FDA00036019732300000110
F(rk),G(rk)和H(rk)表示rk相关的模型矩阵，噪声项
Figure FDA00036019732300000111
Figure FDA00036019732300000112
和
Figure FDA00036019732300000113
为独立同分布的高斯噪声，Qk为过程噪声方差，
Figure FDA00036019732300000114
和
Figure FDA00036019732300000115
分别分为再沸器热负荷噪声方差和辅助测量方差，假设初始分布为
Figure FDA00036019732300000116
其中
Figure FDA00036019732300000117
表示均值为
Figure FDA00036019732300000118
方差为P的高斯分布，为了符号简化，定义
Figure FDA00036019732300000119
为rk的第j个模态，
Figure FDA00036019732300000120
以及
Figure FDA00036019732300000121
In the formula,
Figure FDA0003601973230000013
is the energy consumption variable per ton of distillation column,
Figure FDA0003601973230000014
and
Figure FDA0003601973230000015
represent the reboiler heat load data and auxiliary measurement data, respectively, define
Figure FDA0003601973230000016
and
Figure FDA0003601973230000017
respectively represent the reboiler heat load sequence set and the auxiliary measurement sequence set, and r k represents the limited space
Figure FDA0003601973230000018
A discrete homogeneous Markov chain valued in , for any
Figure FDA0003601973230000019
The transition probability is defined as
Figure FDA00036019732300000110
F(r k ), G(r k ) and H(r k ) represent the model matrix related to r k , the noise term
Figure FDA00036019732300000111
Figure FDA00036019732300000112
and
Figure FDA00036019732300000113
is the independent and identically distributed Gaussian noise, Q k is the process noise variance,
Figure FDA00036019732300000114
and
Figure FDA00036019732300000115
are divided into reboiler heat load noise variance and auxiliary measurement variance, respectively, assuming that the initial distribution is
Figure FDA00036019732300000116
in
Figure FDA00036019732300000117
means that the mean is
Figure FDA00036019732300000118
A Gaussian distribution with variance P, for notation simplification, defines
Figure FDA00036019732300000119
is the jth mode of r k ,
Figure FDA00036019732300000120
as well as
Figure FDA00036019732300000121
 利用贝叶斯状态估计计算包含再沸器热负荷数据和辅助测量数据的贝叶斯后验分布，如下：The Bayesian posterior distribution containing the reboiler heat load data and auxiliary measurement data is calculated using Bayesian state estimation as follows:
Figure FDA0003601973230000021
Figure FDA0003601973230000021
 式中
Figure FDA0003601973230000022
表示再沸器热负荷序列和辅助测量序列总集合。in the formula
Figure FDA0003601973230000022
Represents the total set of reboiler heat duty sequences and auxiliary measurement sequences. 2.如权利要求1所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，采用交互式多模型方法、n阶广义伪贝叶斯算法或变结构多模型算法求解精馏塔单吨能耗的均值和方差的混合交互。2. the rectifying tower single ton energy consumption estimation method based on auxiliary measurement information as claimed in claim 1, is characterized in that, adopts interactive multi-model method, n-order generalized pseudo-Bayesian algorithm or variable structure multi-model algorithm Solve the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column. 3.如权利要求1所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，所述客观相似度为距离测度、相似测度或匹配测度。3 . The method for estimating energy consumption per ton of distillation column based on auxiliary measurement information according to claim 1 , wherein the objective similarity is a distance measure, a similarity measure or a matching measure. 4 . 4.如权利要求1所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，步骤S2中，理想的贝叶斯后验分布如下：4. the rectifying tower single ton energy consumption estimation method based on auxiliary measurement information as claimed in claim 1, is characterized in that, in step S2, ideal Bayesian posterior distribution is as follows:
Figure FDA0003601973230000023
Figure FDA0003601973230000023
 其中
Figure FDA0003601973230000024
为似然分布，
Figure FDA0003601973230000025
为精馏塔单吨能耗的预测分布；in
Figure FDA0003601973230000024
is the likelihood distribution,
Figure FDA0003601973230000025
is the predicted distribution of energy consumption per ton of distillation column; 引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布如下：The Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data is as follows:
Figure FDA0003601973230000026
Figure FDA0003601973230000026
 其中
Figure FDA0003601973230000027
定义为引入辅助数据后的似然分布，
Figure FDA0003601973230000028
为引入辅助数据后需要优化求解的精馏塔单吨能耗的预测分布；in
Figure FDA0003601973230000027
is defined as the likelihood distribution after introducing auxiliary data,
Figure FDA0003601973230000028
Predicted distribution of energy consumption per ton of distillation column that needs to be optimized after the introduction of auxiliary data; 利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，如下：Using the similarity measure, the objective similarity between the Bayesian posterior distribution of the introduced reboiler heat load data and auxiliary measurement data and the ideal Bayesian posterior distribution is calculated as follows:
Figure FDA0003601973230000029
Figure FDA0003601973230000029
 并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布如下：With the objective of maximizing the objective similarity, the optimal prediction distribution of energy consumption per ton of distillation column is as follows:
Figure FDA00036019732300000210
Figure FDA00036019732300000210
 其中exp(·)代表指数函数，
Figure FDA00036019732300000211
表示计算g(·)分布关于f(·)分布的期望。where exp( ) represents the exponential function,
Figure FDA00036019732300000211
Represents the expectation of computing the g(·) distribution with respect to the f(·) distribution. 5.如权利要求4所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，步骤S3包括：5. the rectifying tower single ton energy consumption estimation method based on auxiliary measurement information as claimed in claim 4, is characterized in that, step S3 comprises: S31、求取精馏塔单吨能耗的均值和方差的混合交互：S31. Obtain the mixed interaction of the mean and variance of the energy consumption per ton of the distillation column:
Figure FDA0003601973230000031
Figure FDA0003601973230000031
Figure FDA0003601973230000032
Figure FDA0003601973230000032
 式中符号()表示与前一项相同的项，πij代表从k-1时刻模态i到k时刻模态j的转移概率，
Figure FDA0003601973230000033
为k-1时刻第i个模态的概率，
Figure FDA0003601973230000034
为模态j的预测概率，
Figure FDA0003601973230000035
为k-1时刻第i个模态下精馏塔单吨能耗的均值，
Figure FDA0003601973230000036
为混合交互后j模态下的混合精馏塔单吨能耗均值，
Figure FDA0003601973230000037
为k-1时刻第i个模态下精馏塔单吨能耗的方差，
Figure FDA0003601973230000038
为混合交互后j模态下的混合精馏塔单吨能耗方差；where the symbol ( ) represents the same item as the previous item, π ij represents the transition probability from mode i at time k-1 to mode j at time k,
Figure FDA0003601973230000033
is the probability of the i-th mode at time k-1,
Figure FDA0003601973230000034
is the predicted probability of mode j,
Figure FDA0003601973230000035
is the average energy consumption per ton of the distillation column in the i-th mode at time k-1,
Figure FDA0003601973230000036
is the average energy consumption per ton of the mixed distillation column in j mode after mixing interaction,
Figure FDA0003601973230000037
is the variance of the energy consumption per ton of the distillation column in the ith mode at time k-1,
Figure FDA0003601973230000038
is the variance of energy consumption per ton of mixed distillation column in j mode after mixing interaction; S32、根据预测步公式，计算精馏塔单吨能耗的均值和方差的预测值：S32, according to the prediction step formula, calculate the predicted value of the mean value and variance of the energy consumption per ton of the rectifying tower:
Figure FDA0003601973230000039
Figure FDA0003601973230000039
Figure FDA00036019732300000310
Figure FDA00036019732300000310
 S33、根据求解的最优精馏塔单吨能耗预测分布，对上述预测值进行修正：S33, according to the predicted distribution of the optimal energy consumption per ton of the rectifying tower, correct the above predicted value:
Figure FDA00036019732300000311
Figure FDA00036019732300000311
Figure FDA00036019732300000312
Figure FDA00036019732300000312
 其中
Figure FDA00036019732300000313
为利用辅助测量数据获得的估计值，
Figure FDA00036019732300000314
in
Figure FDA00036019732300000313
For estimates obtained using auxiliary measurement data,
Figure FDA00036019732300000314
 S34、利用如下公式对修正后的精馏塔单吨能耗预测值进行更新：S34, use the following formula to update the revised predicted value of energy consumption per ton of the rectifying tower:
Figure FDA00036019732300000315
Figure FDA00036019732300000315
Figure FDA00036019732300000316
Figure FDA00036019732300000316
 其中
Figure FDA00036019732300000317
in
Figure FDA00036019732300000317
 6.如权利要求5所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，步骤S4包括：6. The method for estimating energy consumption per ton of rectifying tower based on auxiliary measurement information as claimed in claim 5, wherein step S4 comprises: 利用贝叶斯公式计算模态的贝叶斯后验分布：Use the Bayesian formula to calculate the Bayesian posterior distribution of the modes:
Figure FDA0003601973230000041
Figure FDA0003601973230000041
 其中
Figure FDA0003601973230000042
in
Figure FDA0003601973230000042
 7.如权利要求6所述的基于辅助量测信息的精馏塔单吨能耗估计方法，其特征在于，在步骤S5中，最终估计的精馏塔单吨能耗的均值和方差如下：7. the rectifying tower single ton energy consumption estimation method based on auxiliary measurement information as claimed in claim 6, is characterized in that, in step S5, the mean value and the variance of the rectifying tower single ton energy consumption of final estimation are as follows:
Figure FDA0003601973230000043
Figure FDA0003601973230000043
Figure FDA0003601973230000044
Figure FDA0003601973230000044
 其中
Figure FDA0003601973230000045
为最终估计的精馏塔单吨能耗均值，Pk表示最终估计的精馏塔单吨能耗方差。in
Figure FDA0003601973230000045
is the final estimated average energy consumption per ton of the rectifying tower, and P k represents the variance of the final estimated energy consumption per ton of the rectifying tower. 8.一种基于辅助量测信息的精馏塔单吨能耗估计系统，其特征在于，包括：8. a rectifying tower single ton energy consumption estimation system based on auxiliary measurement information, is characterized in that, comprises: 模型构建模块，用于构建精馏塔单吨能耗状态空间模型，基于构建的精馏塔单吨能耗状态空间模型，利用贝叶斯状态估计计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布；The model building module is used to construct a single ton energy consumption state space model of the rectification column. Based on the constructed single ton energy consumption state space model of the rectification column, the reboiler heat load data and auxiliary measurement data are introduced by Bayesian state estimation calculation. The Bayesian posterior distribution of ; 最优精馏塔单吨能耗预测分布求解模块，用于利用相似性度量，计算引入再沸器热负荷数据和辅助测量数据的贝叶斯后验分布和理想的贝叶斯后验分布之间的客观相似度，并以客观相似度最大为目标，求解最优的精馏塔单吨能耗预测分布；The optimal rectification tower single-ton energy consumption prediction distribution solution module is used to calculate the difference between the Bayesian posterior distribution and the ideal Bayesian posterior distribution of the reboiler heat load data and auxiliary measurement data using the similarity measure. The objective similarity between the two, and the objective is to maximize the objective similarity to solve the optimal prediction distribution of energy consumption per ton of distillation column; 精馏塔单吨能耗的均值和方差的预测值更新模块，用于对精馏塔单吨能耗的均值和方差进行混合交互，计算混合交互后精馏塔单吨能耗的均值和方差的预测值，并对预测值进行修正，并对修正后的精馏塔单吨能耗的均值和方差的预测值进行更新；The update module for the prediction value of the mean and variance of the energy consumption per ton of the rectifying tower, which is used to perform mixed interaction with the mean and variance of the energy consumption per ton of the rectifying tower, and calculate the mean and variance of the energy consumption per ton of the rectifying tower after the mixed interaction. The predicted value of the rectification tower, and the predicted value is revised, and the revised predicted value of the average value and variance of the energy consumption per ton of the rectifying tower is updated; 模态概率更新模块，用于利用贝叶斯公式计算模态的后验分布以更新模态概率；The modal probability update module is used to calculate the posterior distribution of the modal using the Bayesian formula to update the modal probability; 精馏塔单吨能耗估计结果输出模块，用于将更新的均值和方差以及模态概率进行融合，得到最终的贝叶斯估计值，即待估计的精馏塔单吨能耗的均值和方差；The output module for the estimation result of energy consumption per ton of the rectifying tower is used to fuse the updated mean and variance and modal probability to obtain the final Bayesian estimate, that is, the mean value and sum of the energy consumption per ton of the rectifying tower to be estimated. variance; 构建的精馏塔单吨能耗状态空间模型如下：The constructed state space model of single ton energy consumption of distillation column is as follows: xk＝F(rk)xk-1+G(rk)wk,x k =F(r k )x k-1 +G(r k )w k ,
Figure FDA0003601973230000051
Figure FDA0003601973230000051
Figure FDA0003601973230000052
Figure FDA0003601973230000052
 式中，
Figure FDA0003601973230000053
为精馏塔单吨能耗变量，
Figure FDA0003601973230000054
和
Figure FDA0003601973230000055
分别代表再沸器热负荷数据和辅助测量数据，定义
Figure FDA0003601973230000056
和
Figure FDA0003601973230000057
分别表示再沸器热负荷序列集合和辅助测量序列集合，rk表示在有限空间
Figure FDA0003601973230000058
中取值的离散齐次马尔可夫链，对于任意
Figure FDA0003601973230000059
转移概率定义为
Figure FDA00036019732300000510
F(rk),G(rk)和H(rk)表示rk相关的模型矩阵，噪声项
Figure FDA00036019732300000511
Figure FDA00036019732300000512
和
Figure FDA00036019732300000513
为独立同分布的高斯噪声，Qk为过程噪声方差，
Figure FDA00036019732300000514
和
Figure FDA00036019732300000515
分别分为再沸器热负荷噪声方差和辅助测量方差，假设初始分布为
Figure FDA00036019732300000516
其中
Figure FDA00036019732300000517
表示均值为
Figure FDA00036019732300000518
方差为P的高斯分布，为了符号简化，定义
Figure FDA00036019732300000519
为rk的第j个模态，
Figure FDA00036019732300000520
以及
Figure FDA00036019732300000521
In the formula,
Figure FDA0003601973230000053
is the energy consumption variable per ton of distillation column,
Figure FDA0003601973230000054
and
Figure FDA0003601973230000055
represent the reboiler heat load data and auxiliary measurement data, respectively, define
Figure FDA0003601973230000056
and
Figure FDA0003601973230000057
respectively represent the reboiler heat load sequence set and the auxiliary measurement sequence set, and r k represents the limited space
Figure FDA0003601973230000058
A discrete homogeneous Markov chain valued in , for any
Figure FDA0003601973230000059
The transition probability is defined as
Figure FDA00036019732300000510
F(r k ), G(r k ) and H(r k ) represent the model matrix related to r k , the noise term
Figure FDA00036019732300000511
Figure FDA00036019732300000512
and
Figure FDA00036019732300000513
is the independent and identically distributed Gaussian noise, Q k is the process noise variance,
Figure FDA00036019732300000514
and
Figure FDA00036019732300000515
are divided into reboiler heat load noise variance and auxiliary measurement variance, respectively, assuming that the initial distribution is
Figure FDA00036019732300000516
in
Figure FDA00036019732300000517
means that the mean is
Figure FDA00036019732300000518
A Gaussian distribution with variance P, for notation simplification, defines
Figure FDA00036019732300000519
is the jth mode of r k ,
Figure FDA00036019732300000520
as well as
Figure FDA00036019732300000521
 利用贝叶斯状态估计计算包含再沸器热负荷数据和辅助测量数据的贝叶斯后验分布，如下：The Bayesian posterior distribution containing the reboiler heat load data and auxiliary measurement data is calculated using Bayesian state estimation as follows:
Figure FDA00036019732300000522
Figure FDA00036019732300000522
 式中
Figure FDA00036019732300000523
表示再沸器热负荷序列和辅助测量序列总集合。in the formula
Figure FDA00036019732300000523
Represents the total set of reboiler heat duty sequences and auxiliary measurement sequences.
CN202111222841.3A 2021-10-20 2021-10-20 Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information Active CN113962081B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111222841.3A CN113962081B (en) 2021-10-20 2021-10-20 Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111222841.3A CN113962081B (en) 2021-10-20 2021-10-20 Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information

Publications (2)

Publication Number Publication Date
CN113962081A CN113962081A (en) 2022-01-21
CN113962081B true CN113962081B (en) 2022-05-31

Family

ID=79465024

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111222841.3A Active CN113962081B (en) 2021-10-20 2021-10-20 Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information

Country Status (1)

Country Link
CN (1) CN113962081B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115169721B (en) * 2022-07-25 2023-07-04 江南大学 A method and system for predicting energy consumption per ton in distillation process based on migration identification

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108804784A (en) * 2018-05-25 2018-11-13 江南大学 A kind of instant learning soft-measuring modeling method based on Bayes's gauss hybrid models
CN112116138A (en) * 2020-09-09 2020-12-22 山东科技大学 Power system prediction state estimation method and system based on data driving
CN112562797A (en) * 2020-11-30 2021-03-26 中南大学 Method and system for predicting outlet ions in iron precipitation process

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7289906B2 (en) * 2004-04-05 2007-10-30 Oregon Health & Science University Navigation system applications of sigma-point Kalman filters for nonlinear estimation and sensor fusion
US11966840B2 (en) * 2019-08-15 2024-04-23 Noodle Analytics, Inc. Deep probabilistic decision machines

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108804784A (en) * 2018-05-25 2018-11-13 江南大学 A kind of instant learning soft-measuring modeling method based on Bayes's gauss hybrid models
CN112116138A (en) * 2020-09-09 2020-12-22 山东科技大学 Power system prediction state estimation method and system based on data driving
CN112562797A (en) * 2020-11-30 2021-03-26 中南大学 Method and system for predicting outlet ions in iron precipitation process

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Intelligent State Estimation for Continuous Fermenters Using Variational Bayesian Learning;S. Gao, S. Zhao, X. Luan and F. Liu;《IEEE Transactions on Industrial Informatics》;20210205;第8429-8437页 *
一种基于自适应模糊高斯核聚类的软测量建模方法;夏源等;《上海交通大学学报》;20170628(第06期);第722-726页 *
基于贝叶斯学习的关联向量机及其在软测量中的应用;陈佳等;《华东理工大学学报(自然科学版)》;20070130(第01期);第1-5页 *
基于贝叶斯网络的软测量建模方法;李雅芹等;《计算机与应用化学》;20101028(第10期);第1391-1394页 *
复合高斯噪声中知识辅助的贝叶斯Rao检测方法;高永婵等;《西安电子科技大学学报》;20130710(第06期);第46-51、173页 *
聚合过程贝叶斯统计学习质量模式监测;高爽，郑年年，栾小丽，刘飞;《化工进展》;20181231;第37卷(第12期);第4558-4564页 *

Also Published As

Publication number Publication date
CN113962081A (en) 2022-01-21

Similar Documents

Publication Publication Date Title
US20230297642A1 (en) Bearings-only target tracking method based on pseudo-linear maximum correlation entropy kalman filtering
CN109990786B (en) Maneuvering target tracking method and device
CN110503071A (en) Multi-Target Tracking Method Based on Variational Bayesian Label Multi-Bernoulli Stacking Model
CN108804784A (en) A kind of instant learning soft-measuring modeling method based on Bayes's gauss hybrid models
CN113012766B (en) An Adaptive Soft Sensing Modeling Method Based on Online Selective Integration
Gao et al. Improved IMM algorithm for nonlinear maneuvering target tracking
CN108647272A (en) A kind of small sample extending method based on data distribution
CN113962081B (en) Rectifying tower single-ton energy consumption estimation method and system based on auxiliary measurement information
McCulloch Linear regression with stable disturbances
CN108255791A (en) Maneuvering target tracking method based on distributed sensor consistency
Creel et al. Indirect likelihood inference
CN112381145A (en) Gaussian process regression multi-model fusion modeling method based on nearest correlation spectral clustering
CN111859263A (en) Accurate dosing method for tap water treatment
CN110046377A (en) A kind of selective ensemble instant learning soft-measuring modeling method based on isomery similarity
CN109145421B (en) A spatiotemporal fuzzy modeling method applied to distributed parameter systems
CN106547899B (en) Intermittent process time interval division method based on multi-scale time-varying clustering center change
CN109375160B (en) Angle measurement error estimation method in pure-azimuth passive positioning
CN109253727B (en) A Localization Method Based on Improved Iterative Volume Particle Filter Algorithm
CN110794676A (en) CSTR process nonlinear control method based on Hammerstein-Wiener model
Lü et al. Ladle furnace liquid steel temperature prediction model based on optimally pruned bagging
CN113432608A (en) Generalized high-order CKF algorithm based on maximum correlation entropy and suitable for INS/CNS integrated navigation system
CN112197914A (en) Whale MUSIC algorithm-based gas leakage source estimation method
CN112837351B (en) Improved label multiple Bernoulli distributed optimization fusion tracking method
CN109902762A (en) Data preprocessing method based on 1/2 similarity deviation
Saha et al. Missing value estimation in DNA microarrays using linear regression and fuzzy approach

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant