CN104898416B - One sterilization medium energy-saving method of steam injection based modeling armax - Google Patents

One sterilization medium energy-saving method of steam injection based modeling armax Download PDF

Info

Publication number
CN104898416B
CN104898416B CN201510158900.3A CN201510158900A CN104898416B CN 104898416 B CN104898416 B CN 104898416B CN 201510158900 A CN201510158900 A CN 201510158900A CN 104898416 B CN104898416 B CN 104898416B
Authority
CN
China
Prior art keywords
control
steam
time
system
sterilization
Prior art date
Application number
CN201510158900.3A
Other languages
Chinese (zh)
Other versions
CN104898416A (en
Inventor
曹晖
张士良
张彦斌
贾立新
司刚全
Original Assignee
西安交通大学
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 西安交通大学 filed Critical 西安交通大学
Priority to CN201510158900.3A priority Critical patent/CN104898416B/en
Publication of CN104898416A publication Critical patent/CN104898416A/en
Application granted granted Critical
Publication of CN104898416B publication Critical patent/CN104898416B/en

Links

Abstract

本发明公开了一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法,该方法以生物发酵用培养基蒸汽直接加热混合灭菌控制过程为对象,利用生物制药用培养基蒸汽灭菌过程数据建立外部输入自回归滑动(ARMAX)模型,在该模型基础上推导灭菌过程涉及的过程参数对应的传递函数,建立动态矩阵并实现动态矩阵控制(DMC)。 The present invention discloses a medium steam injection sterilization ARMAX model based on energy-saving control method of biological fermentation medium directly mixing steam sterilization heating control process for the object, using the biopharmaceutical medium steam sterilization process data Setup autoregressive moving the external input (ARMAX) model derivation process parameters of the sterilization process according to a transfer function corresponding to the model is established based on the dynamic matrix and dynamic matrix control (DMC). 为节约控制过程中蒸汽使用量,在达到精确控制的目标的同时,本发明中对控制器的设计上加入了节能指标,使控制过程中的蒸汽使用效率更高,降低生产过程中的蒸汽使用成本;本发明控制方法简单,控制策略和控制器参数设计基于实际工业运行过程数据,对灭菌过程涉及到的工艺环节和设备分析的依赖性小,可实现性强,易于应用。 To save the steam amount control process, the target to achieve precise control, while the present invention is designed to control the energy index is added to make more efficient use of steam to control the process, decrease of the steam used in the production process cost; simple control method of the present invention, the control strategy based on the actual design parameters of the controller and the industrial process data operation, small sterilization process involves apparatus and process link dependency analysis can be achieved and strong, easy to apply.

Description

一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法 One sterilization medium energy-saving method of steam injection based ARMAX model

技术领域 FIELD

[0001] 本发明属于生物控制技术领域,具体涉及一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法。 [0001] The present invention belongs to the technical field of biological control, particularly relates to steam injection sterilization medium ARMAX model based on energy-saving control method.

背景技术 Background technique

[0002] 生物制药是利用微生物的代谢作用产生药物成分,对代谢后的微生物和培养基混合物进行提纯后获取有效成分的过程。 [0002] Biopharmaceutical using microbial metabolism produce a pharmaceutical composition, of a mixture of medium and microbial metabolism of the active ingredient, the process acquired after purification. 在生物制药流程中,对培养基的消毒灭菌是不可缺少的一个环节。 Biopharmaceutical process, the sterilization of the culture medium is an indispensable part. 杀灭培养基中的杂菌,可使目标细菌在发酵中处在无其他微生物干扰的环境中,避免对培养基营养物质的竞争和对目标微生物代谢过程的干扰,因而消毒灭菌是发酵成功的基础。 Kill the bacteria in the culture medium, it can make no other target bacteria in a microbial interference environment in the fermentation medium to avoid competition for nutrients and microbial metabolic process on the target interference, and thus sterilization is fermented success The basics. 常见的培养基灭菌方法是用蒸汽直接加热灭菌,又分为实罐灭菌和连续灭菌。 Common sterilizing the medium is directly heated with steam sterilization, sterilization tank is divided into real and continuous sterilization. 实罐灭菌是指将蒸汽直接通入到装有培养基的料罐中进行灭菌,连续灭菌是指使用蒸汽喷射器,在喷射器中完成蒸汽喷射培养基加热灭菌的方式。 Real sterilization tank means to direct the steam into the feed tank containing the medium was sterilized, sterilization is the use of continuous steam ejector, steam injection is completed embodiment of the heating and sterilizing the medium in the ejector. 相比于实罐灭菌,连续灭菌对培养基加热灭菌效果更彻底,效率更高,对生产过程的连续性更为支持,因而成为灭菌方法的发展趋势。 Compared to the real tank sterilization, the sterilization of the medium continuous heat sterilization effect more complete, more efficient, more support for continuous production process, thus become the trend sterilization methods. 但连续灭菌,即使用喷射器对培养基进行蒸汽喷射加热灭菌的方式控制系统较为复杂,该过程控制需要有效的控制策略以达到控制效果。 However, continuous sterilization, i.e. using an ejector for ejecting steam heating and sterilizing medium is controlled systems are more complex, the need for effective control of the process control strategy to achieve the control effect. 灭菌过程控制以培养基被蒸汽喷射加热后在喷射器出口处的温度控制精度为技术指标,同时需要处理在灭菌过程中多个系统扰动量的影响。 After the sterilization process to control steam injection medium was heated at a temperature at the outlet of the injector control accuracy specifications, also need during sterilization treatment on a plurality of system disturbances. 在蒸汽喷射灭菌控制系统中,控制量输入为喷射器蒸汽管路上的阀门开度,输出为培养基在加热灭菌后在喷射器出口处的温度,扰动量有培养基在设备管路中的流速,培养基在喷射器入口处的温度及蒸汽温度。 Sterilization steam injection control system, the control input for the valve opening degree of the steam ejector pipe line, the output medium is sterilized by heating at a temperature at the outlet of the ejector, there is disturbance in the medium line device flow rate, medium temperature and the injector inlet of the steam temperature. 在实际生产中,扰动量的干扰作用十分明显,而多个环节亦容易使扰动量发生变化,如生产用蒸汽在温度和压力上存在的扰动, 多个车间和生产环节对蒸汽管路的共用对蒸汽质量的干扰,不同批次培养基在不同物料罐中的成分配比和温度的不同,灭菌设备和管路对培养基流速的影响等等,使得灭菌控制过程面临难题,而多个扰动的存在,也使建立被控对象的精确模型进而进行系统分析的方法难以实现。 In actual production, the amount of interference disturbance is obvious, but also easy to make a plurality of links changes in the amount of disturbance, such as process steam temperature and pressure in the presence of perturbations, and the workshop production processes common to a plurality of steam pipes interference steam quality, medium components in different batches of different materials in the tank and the temperature difference ratio, the influence of the sterilization apparatus and the medium flow conduit or the like, such that the sterilization process control problems faced, and multi- the presence of disturbance, also precisely controlled object model further systematic analysis methods is difficult to achieve.

[0003] 此外,培养基灭菌过程大量使用蒸汽,蒸汽成本成为发酵过程中除电力和物料之外的重要部分,常规方法为达到灭菌目的,往往开大蒸汽管路阀门,加大蒸汽使用量以确保灭菌的彻底,甚至有蒸汽使用过多而加热过度,造成培养基营养成分被破环的现象。 [0003] In addition, the medium widely used sterilization process steam, costs become an important part of the fermentation process and materials other than electricity, a conventional method to achieve the purpose of sterilization, tend to turn up the steam line valves, used to increase the steam amount to ensure thorough sterilization, and even excessive use of steam and excessive heating, resulting in nutrient media phenomenon is broken ring. 因此, 如何设计灭菌控制策略,如何使灭菌目标与节能目标达成配合,在达到灭菌目的,提高灭菌控制精度的同时,节约蒸汽使用量,降低生产成本,成为生物制药实际生产中面临的问题。 Therefore, how to design control strategies sterilization, sterilization target and how to make energy-saving target to achieve cooperation, reaching the purpose of sterilization, sterilization at the same time improve the control precision, saving steam usage, lower production costs, to be confronted with the actual production of biopharmaceuticals The problem.

[0004] 在过程控制领域中,预测控制以其良好的控制效果在多种过程控制应用中得以实现。 [0004] In the field of process control, predictive control with good control performance can be achieved in a variety of process control applications. 该控制方法自上世纪七十年代被提出,至今仍在工业过程控制中发挥重要作用。 The control method since the early seventies been proposed, industrial process control still play an important role. 以动态矩阵为特征的动态矩阵控制是预测控制的一种,该控制方法基于工业过程采样数据建立控制对象逼近模型,并根据所建立模型获取过程参数的阶跃响应,形成动态矩阵并完成控制过程。 It characterized the dynamic matrix control is a dynamic matrix predictive control, the control method based on the sample data to establish an industrial process control object model approximation, step response model and the parameters of the acquisition process in accordance with the established form and complete the dynamic matrix control process . 该方法可避免工业过程的精确建模困难,同时以其滚动优化的方式进行计算,并可在控制量的计算中加入约束,可改善控制效果,提高过程控制精度。 This method avoids the difficulties of accurate modeling of an industrial process, simultaneously calculating its rolling optimization manner, constraint can be added to the calculated control amount, the control effect can be improved to improve the accuracy of process control. 在本发明中,通过分析控制过程对蒸汽使用量的影响,在控制策略中加入对蒸汽使用量的优化约束,使控制过程具备优化蒸汽使用量的作用,配合控制精度目标,使得灭菌控制过程更具节能效果,可降低生产成本,提高经济效益。 In the present invention, by analyzing the influence of the steam amount control process, the optimization constraints added amount of steam used in the control strategy of the control process with optimized use of steam action, with the control accuracy of target, such that the sterilization process control more energy-saving effect, can reduce production costs, improve economic efficiency.

发明内容 SUMMARY

[0005] 为了解决上述现有技术存在的问题,本发明的目的在于提供一种基于ARMAX (外部输入自回归滑动模型)建模的培养基蒸汽喷射灭菌节能控制方法,其控制指标是培养基即物料和蒸汽混合后在喷射器出口出的温度精度,以蒸汽流量,在本发明中体现为蒸汽管路上的阀门开度为控制量,并加入对系统扰动量的分析,实现对培养基蒸汽喷射灭菌温度的精确控制。 [0005] In order to solve the prior art problems, an object of the present invention is based is to provide a ARMAX (Auto Regressive Moving external input model) steam injection sterilization medium energy saving control method of modeling, which is a control indicator medium i.e., after mixing the material temperature accuracy and a steam ejector outlet to the steam flow, as embodied in the present invention, the steam valve opening control amount of the pipeline, and the added amount of the disturbance analysis system, to achieve the medium steam precise control of the injection of the sterilizing temperature.

[0006] 为了达到上述发明目的,本发明采取的技术方案是: [0006] In order to achieve the above object, the present invention takes the following technical solution:

[0007] 一种基于ARMX建模的培养基蒸汽喷射灭菌节能控制方法,包括如下步骤: [0007] A steam injection sterilization medium based modeling ARMX saving control method comprising the steps of:

[0008] 步骤1:利用培养基蒸汽灭菌过程数据,建立灭菌过程ARMAX模型 [0008] Step 1: Using the data medium steam sterilization process, the sterilization process ARMAX model establishment

[0009] 首先收集工业运行现场培养基蒸汽灭菌过程数据,通过对过程数据进行分析,找到灭菌过程所涉及到的过程参数,并确定该控制系统的输入、输出及扰动量,由此确定灭菌控制系统的模型结构;经分析,系统输入为用于加热培养基的蒸汽的单位流量,在实际运行中表现为喷射器(即蒸汽和培养基完成混合过程的设备)蒸汽入口管路的阀门开度,系统输出为喷射器出口温度,即蒸汽和培养基混合后的温度,系统扰动量有:培养基在喷射器入口处温度,培养基流动速度,喷射器入口处蒸汽温度;本发明中采用建模的模型结构为ARMAX 模型,结合灭菌过程的参数及个数,其形式如下: [0009] First the operating site collecting industrial data medium steam sterilization process, by analyzing the process data, to find the process parameters of the sterilization process involved, and determine the input, output and disturbance of the control system, thereby determining sterilization control system model structure; analyzed, the system input unit flow rate of steam for heating medium, and in the actual operation of the injector (i.e., steam and mixing equipment to complete the process medium) steam inlet line valve opening, an output system is an ejector outlet temperature, i.e. the temperature of the steam and the mixed medium, the system disturbance are: temperature of the medium in the inlet of the ejector, the medium flow rate, the inlet temperature of the steam injector; present invention the model structure is modeled using the ARMAX model, and the number of binding parameters of the sterilization process, the following form:

[0010] A (z) y ⑴=B (z) u ⑴ +Di (z) ei ⑴ +D2 (z) e2 ⑴ +D3 (z) e3 ⑴ [0010] A (z) y ⑴ = B (z) u ⑴ + Di (z) ei ⑴ + D2 (z) e2 ⑴ + D3 (z) e3 ⑴

[0011] A (z) =ai+a2Z—i+H.anaZ—na+1 [0011] A (z) = ai + a2Z-i + H.anaZ-na + 1

[0012] B(z) =bi+b2z—^"bnbZ-nb+1 [0012] B (z) = bi + b2z - ^ "bnbZ-nb + 1

Figure CN104898416BD00071

[0014] 其中,y⑴为系统输出,u⑴为系统输入量,A (z)为y⑴对应的延迟因子所构成的多项式,B (z)为控制量u⑴对应的延迟因子所构成的多项式,D1 (z)为第1个扰动量的⑴对应的延迟因子构成的多项式,D2 (Z)为第2个扰动量Θ2 (t)对应的延迟因子构成的多项式,D3 (Z)为第3个扰动量e3⑴对应的延迟因子构成的多项式,D1 (Z)为第i个扰动量对应的延迟因子所构成的多项式,ai,a2,......ana,bi,b2,......bnb,dii,di2,......山吨为待辨识系数;ei⑴, e2 (t),e3 (t)分别为t时刻培养基在喷射器入口处温度,培养基流动速度和喷射器入口处蒸汽温度; [0014] wherein, y⑴ system output, u⑴ system input, A (z) polynomial y⑴ corresponding delay factor constituted, B (z) is the control amount u⑴ polynomial corresponding delay factor constituted, D1 ( polynomial z) is the first one of perturbation ⑴ corresponding delay factor constituted, D2 (Z) of the second perturbation Θ2 (t) polynomial corresponding delay factor constituted, D3 (Z) for the first three disturbances polynomials corresponding delay factor e3⑴ configuration, D1 (Z) is the i th perturbation corresponding delay factor polynomials constituted, ai, a2, ...... ana, bi, b2, ...... bnb, dii, di2, ...... tonnes Hill coefficient to be identified; ei⑴, e2 (t), e3 (t) at time t, respectively, at the inlet temperature of the spray medium, a medium flow velocity and the ejector steam temperature at the inlet;

[0015] 步骤2:离散传递函数的推导和控制矩阵的生成 [0015] Step 2: Derivation and generating a control transfer function matrix of discrete

[0016] 利用工业运行数据辨识ARMAX模型中的参数,获得参数模型后直接推导对应过程参数的离散传递函数: [0016] With the industrial operation data identification parameter ARMAX model, directly after derivation of the corresponding process parameters of the model parameters discrete transfer function:

Figure CN104898416BD00072

[0018] 故控制量和三个扰动量ei,e2,e3对应的离散传递函数F^F1,F2,F3依次为: [0018] Therefore, three control amount and the disturbance ei, e2, e3 corresponding to the discrete transfer function F ^ F1, F2, F3 as follows:

Figure CN104898416BD00073

[0023] 为计算各参数对应的动态矩阵,先求各参数在对应离散传递函数下的阶跃响应, 其形式如下: [0023] To calculate the dynamic matrix of each parameter, each parameter to find the corresponding step responses at discrete transfer function, the following form:

Figure CN104898416BD00081

[0024] 对于控制量u,采样时刻1,2,……,nu对应的系统输出分别为: [0024] For controlling the amount of u, sampling time 1,2, ......, nu output corresponding system are:

Figure CN104898416BD00082

[0026] 对于扰动量ei,采样时刻1,2,……,!^以寸应的系统输出分别为: [0026] For perturbation ei, sampling time 1,2, ......, ^ inch corresponding to the output system are!:

Figure CN104898416BD00083

[0028] 对于扰动量e2,采样时刻1,2,……,应的系统输出分别为: [0028] For perturbation e2, sampling time 1,2, ......, corresponding system output are:

Figure CN104898416BD00084

[0030] 对于扰动量e3,采样时刻1,2,……,n:e3对应的系统输出分别为: [0030] For perturbation e3, sampling time 1,2, ......, n: e3 corresponding to the output of the system are:

Figure CN104898416BD00085

[0032] 由此,控制量u,扰动量ei,e2,e3对应的动态矩阵Gu,Gi,G2,G3分别为: [0032] Accordingly, the control amount u, disturbance ei, e2, e3 corresponding dynamic matrix Gu, Gi, G2, G3 are:

Figure CN104898416BD00086

Figure CN104898416BD00091

[0037] 其中,m为控制步长,N为预测步长; [0037] where, m is the control step, N being a prediction step;

[0038] 步骤3:控制量的计算 [0038] Step 3: calculate the amount of control

[0039] 基于步骤2所得的各参数动态矩阵,系统预测输出值y'可表达为: [0039] Based on the parameters obtained in step 2 dynamic matrix, the system predicts the output values ​​y 'may be expressed as:

Figure CN104898416BD00092

[0041]上式中,u为控制量按时间序列构成的向量,Gi为扰动量ei的增量对应的动态矩阵,Ei 为扰动量&的增量按时间序列构成的向量;fu为控制量中影响预测值的只与过去时刻数值有关的部分所构成的列向量,fu= [fu(t+l),fu(t+2)……fu(t+N)]T,fEi为扰动e冲影响预测值的只与过去时刻数值有关的部分所构成的列向量, [0041] In the above formula, u is the control amount vector composed of time series, the dynamic matrix of the perturbation Gi corresponding increment ei, Ei is disturbances & amp; delta time series vectors consisting of; FU control Effect of the predicted value of the amount of only the past time value column vector composed of the relevant portion, fu = [fu (t + l), fu (t + 2) ...... fu (t + N)] T, fEi of disturbance e DRAWING only the column vector of predicted values ​​associated with the past portion constituting the time value,

Figure CN104898416BD00093

其中: among them:

Figure CN104898416BD00094

[0044] 其中fu(t+k)为fu中的第k个元素, [0044] wherein fu (t + k) is the k th fu elements,

Figure CN104898416BD00095

为fEj中的第k个元素,N为预测步长, Δ u (ti)为ti时刻的控制量增量,△ ei (tj)为扰动量ei在tj时刻的增量; FEj to the k-th element, N is the prediction steps, Δ u (ti) to control the increment of time ti, △ ei (tj) is the disturbance ei increase in time tj;

[0045] 为简便起见,将系统预测输出值y'记为: [0045] For simplicity, system predictive output value y 'referred to as:

[0046] y7 =Guu+f [0046] y7 = Guu + f

[0047] 其中 [0047] in which

Figure CN104898416BD00096

,因Ei在未来时刻的值不能预测,故设置为0,系统建模和动态矩阵获得后,即可确定目标函数进而计算控制量;常规目标函数J形式为: , Because the value Ei at the time of the next can not be predicted, it is set to 0, the matrix obtained and dynamic system modeling, and then calculate the objective function to determine the control amount; conventional objective function J of the form:

Figure CN104898416BD00097

[0049] 其中λ为平衡目标函数两部分(即预测误差和控制耗能)权重的系数,m为控制步长,Δ u (t+j-Ι)为t+j-Ι时刻控制量的变化量,y' (t+j 11)为在t时刻预测的t+j时刻的系统输出值,w (t+j)为t+j时刻的输出参考轨迹,一般设置为如下形式: [0049] wherein λ is the equilibrium target function in two parts (i.e., the prediction error and control energy) the weight coefficient, m is the control step size, Δ u (t + j-Ι) for the j-Ι timing control amount t of change + amount, y '(t + j 11) the system output values ​​predicted at time t time t + j, w (t + j) is the output reference trajectory time t + j, is generally provided in the following form:

[0050] w (t+j) =aw (t+j-1) + (1-a) r (t+j) [0050] w (t + j) = aw (t + j-1) + (1-a) r (t + j)

[0051] w ⑴=y(t) [0051] w ⑴ = y (t)

[0052] 其中r (t+j)为系统输出在t+j时刻的设定值,a为〇到I之间的常数; [0052] wherein R & lt (t + j) is the system output j at time t + a set value, a is a constant between the square I;

[0053] 求目标函数的最小值,即可获得未来控制步长内最优的控制量。 [0053] The minimization of the objective function, an optimum control amount can be obtained in the next control step. 但考虑到蒸汽使用成本,为节能降耗起见,目标函数中应当加入对蒸汽使用量的目标约束; But considering the cost of steam for energy saving purposes, the objective function should be added to certain constraints on the amount of steam;

[0054] 蒸汽使用量Q可表示为如下形式: [0054] Use of steam amount Q can be expressed as follows:

[0055] Q = T*fs [0055] Q = T * fs

[0056] 式中,T为时间间隔,fs为蒸汽流速,而fs正比与阀门开度,在本发明中,控制步长范围内的各个扫描间隔内的蒸汽使用量Q(t+1),Q(t+2),......Q(t+m)为: [0056] where, T is the time interval, fs is the steam flow rate, and fs proportional to the valve opening, in the present invention, the control amount of the steam in the respective scanning interval step length range of Q (t + 1), Q (t + 2), ...... Q (t + m) is:

[0057] Q (t+1) =K* [u ⑴ + Δ u (t+1) ] * Δ T [0057] Q (t + 1) = K * [u ⑴ + Δ u (t + 1)] * Δ T

[0058] Q (t+2) =K* [u ⑴ + Δ u (t+1) + Δ u (t+2) ] * Δ T [0058] Q (t + 2) = K * [u ⑴ + Δ u (t + 1) + Δ u (t + 2)] * Δ T

[0059] · [0059] ·

[0060] · [0060] ·

[0061] · [0061] ·

[0062] Q (t+m) =K* [u ⑴ + Δ u (t+1) + Δ u (t+2)......Δ u (t+m) ] * Δ T [0062] Q (t + m) = K * [u ⑴ + Δ u (t + 1) + Δ u (t + 2) ...... Δ u (t + m)] * Δ T

[0063] 式中,K为常系数,△ T为控制间隔时间,即扫描时间间隔,u (t)为当前时刻控制量输入的数值,m为控制步长;在控制步长内,蒸汽总耗量E为: [0063] where, K is a constant coefficient, △ T is the control interval of time, i.e., the scan time interval, u (t) is the current time value of the control input, m is a control step; in the control step, the steam total E consumption as follows:

Figure CN104898416BD00101

[0065] 上式可表达为如下形式: [0065] The above equation may be expressed as follows:

[0066] E=KXmXu (t)+KXMtU [0066] E = KXmXu (t) + KXMtU

[0067] 其中M为蒸汽总消耗量计算式推导后产生的矩阵,u为未来时刻控制量u (t+1),u (T +2),……u (t+m)按时间序列构成的向量: [0067] where the matrix M is generated after the total steam consumption calculation formula is derived, u is the next timing control amount u (t + 1), u (T +2), ...... u (t + m) constituting the time series vector:

[0068] U= [u (t+1) ,u (T+2) ,......u (t+m) ] T [0068] U = [u (t + 1), u (T + 2), ...... u (t + m)] T

[0069] M= [m,ml,......1] T [0069] M = [m, ml, ...... 1] T

[0070] 如此,加入了对节能指标的考虑后,目标函数J确定为: After [0070] Thus, the added consideration of energy conservation target, the objective function J is determined to be:

Figure CN104898416BD00102

[0072] 其中 [0072] in which

Figure CN104898416BD00103

,求目标函数的最小值,即可得到未来时刻控制量输入;将目标函数转换成矩阵运算形式后求导,即可得到: , Find the minimum value of the objective function, to obtain a control input future time; after converting the derivative into the objective function in the form of matrix operations, can be obtained:

Figure CN104898416BD00104

[0074] 其中w为输出轨迹按时间顺序构成的序列,T为转置符号,I为单位对角矩阵,上标_ 1为矩阵逆运算符号,以上即为动态矩阵控制方法得到的未来时刻控制量输入,按照滚动优化的方式,每次计算u都只需要计算第一行第一列的数值,然后送至系统执行器即可。 [0074] where w is the output track sequence composed of chronologically, T is the transpose symbol, I is a unit diagonal matrix, the superscript _ an inverse matrix notation, the above is the dynamic matrix control method to obtain the future time control input according rolling optimized manner, each calculation value u only necessary to calculate a first row and first column, and then sent to the system actuators.

[0075] 本发明和现有技术相比,具有如下优点: [0075] The present invention as compared to the prior art, has the following advantages:

[0076] 本发明公开了一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法,以ARMAX 模型结构建立培养基蒸汽喷射灭菌过程数学模型,根据所建立的数学模型推导灭菌控制过程的离散传递函数,使用数学工具,建立基于所得传递函数的阶跃响应,并由此产生动态控制矩阵,用于系统控制;该控制方法采用ARMAX数学建模手段,从灭菌过程数据中提取有用信息进行建模,实现对工业运行数据的有效利用;本发明考虑到生产厂家对节能的需求,在控制策略构建过程中加入对蒸汽使用量的目标约束,使得该控制方法在克服控制精度问题同时,实现减少生产过程蒸汽的使用量,以降低生产成本,提高经济效益。 [0076] The present invention discloses a steam injection sterilization of the medium based on energy-saving control method ARMAX model, ARMAX model structure to establish a medium steam injection sterilization process mathematical model, the mathematical derivation of the sterilization procedure according to the control model established the discrete transfer function, the use of mathematical tools to create a step response based on the resulting transfer function, and thereby generating a dynamic control matrix for controlling the system; ARMAX useful in the control method using mathematical modeling means, extracting data from the sterilization process information modeling, effective utilization of industrial operation data; the present invention contemplates the manufacturer's demand for energy efficient, the amount of steam added to certain constraints on the control strategy used in the build process, such that the control method while overcoming the problem of control accuracy , achieve a reduction of the amount of process steam, to reduce production costs, increase economic efficiency.

附图说明 BRIEF DESCRIPTION

[0077] 附图为本发明基于ARMX建模的培养基蒸汽喷射灭菌节能控制系统结构图。 [0077] The steam injection sterilization medium energy control system structure based on the drawings of the present invention ARMX modeled.

[0078] 图中:u⑴为培养基蒸汽喷射灭菌系统控制量输入,即喷射器蒸汽阀门开度;y (t) 为培养基蒸汽喷射灭菌控制系统输出,即喷射器物料出口温度。 [0078] FIG: u⑴ medium is steam injection sterilization system control input, i.e., the steam injector valve opening; y (t) for the steam injection sterilization media control system output, i.e., the ejector outlet temperature of the material.

具体实施方式 Detailed ways

[0079] 以下结合附图及具体实施例,对本发明作进一步的详细描述。 [0079] The following embodiments in conjunction with accompanying drawings and specific embodiments, the present invention will be further described in detail.

[0080] 以某生物制药厂为例,给出本发明的一个具体应用。 [0080] In an example of a bio-pharmaceutical, given a particular application of the invention. 该制药厂灭菌工作流程为:首先在配料罐完成培养基的准备过程,将相关培养基成分加水混合后搅拌均匀,以供灭菌之用。 The pharmaceutical and sterilization procedures: First, complete the preparation of the culture medium mixing tank, the stir is mixed with water-related media components, for sterilization purposes. 在灭菌设备经过蒸汽消毒灭菌后开始培养基灭菌过程:由电机带动打料栗进入喷射器中与蒸汽完成灭菌过程。 In the sterilization apparatus after sterilizing the medium steam sterilization process begins: the motor is driven by a knock Li into the ejector with steam sterilization process is completed.

[0081] 该培养基蒸汽喷射灭菌节能控制方法的实施包括如下三个部分: [0081] The medium is steam sterilized injection saving control method embodiment comprises the following three parts:

[0082] (1)利用培养基蒸汽灭菌过程数据,建立灭菌过程ARMAX模型 [0082] (1) using a data medium steam sterilization process, the sterilization process ARMAX model establishment

[0083] 首先收集工业运行现场培养基蒸汽灭菌过程数据,通过对过程数据进行分析,找到灭菌过程所涉及到的过程参数,并确定该控制系统的输入、输出及扰动量,由此确定灭菌控制系统的模型结构。 [0083] First the operating site collecting industrial data medium steam sterilization process, by analyzing the process data, to find the process parameters of the sterilization process involved, and determine the input, output and disturbance of the control system, thereby determining sterilization control system model structure. 经分析,系统输入为用于加热培养基的蒸汽的单位流量,在实际运行中表现为喷射器(即蒸汽和培养基完成混合过程的设备)蒸汽入口管路的阀门开度,系统输出为喷射器出口温度,即蒸汽和培养基混合后的温度,系统扰动量有:培养基在喷射器入口处温度,培养基流动速度,喷射器入口处蒸汽温度。 Units of flow of steam for heating medium, by analyzing the performance of the system in actual operation of the input injectors (i.e., steam and mixing equipment to complete the process medium) the valve opening of the steam inlet pipe, for the injection system output outlet temperature, i.e. the temperature of the steam and the mixed medium, the system disturbance are: temperature of the medium in the inlet of the ejector, the medium flow rate, the inlet temperature of the steam injector. 本发明中采用建模的模型结构为ARMAX 模型,结合灭菌过程的参数及个数,其形式如下: Model structure employed in the present invention, ARMAX model is modeled, and the number of binding parameters of the sterilization process, the following form:

[0084] A (z) y ⑴=B (z) u ⑴ +Di (z) ei ⑴ +D2 (z) e2 ⑴ +D3 (z) e3 ⑴ [0084] A (z) y ⑴ = B (z) u ⑴ + Di (z) ei ⑴ + D2 (z) e2 ⑴ + D3 (z) e3 ⑴

[0085] A (z) =ai+a2Z—i+H.anaZ—na+1 [0085] A (z) = ai + a2Z-i + H.anaZ-na + 1

[0086] B (z) =bi+b2z—^..bnbz—nb+1 [0086] B (z) = bi + b2z - ^ .. bnbz-nb + 1

Figure CN104898416BD00111

[0088] 其中,y⑴为t时刻系统输出,u⑴为t时刻系统输入量,A (z)为y⑴对应的延迟因子所构成的多项式,B (z)为控制量u⑴对应的延迟因子所构成的多项式,D1 (z)为第1个扰动量ei (t)对应的延迟因子构成的多项式,D2 (z)为第2个扰动量e2 (t)对应的延迟因子构成的多项式,D3 (z)为第3个扰动量e3 (t)对应的延迟因子构成的多项式,D1 (z)为第i个扰动量对应的延迟因子所构成的多项式,ai,a2,......ana,bi,b2,......bnb,dii,di2,......山叫为待辨识系数;ex⑴,e2⑴,e3⑴分别为t时刻培养基在喷射器入口处温度,培养基流动速度和喷射器入口处蒸汽温度。 [0088] wherein, y⑴ output time t system, u⑴ for the time t system input, A (z) is y⑴ corresponding delay factor constituted polynomial, B (z) is the control amount u⑴ corresponding delay factor consisting of polynomials, D1 (z) for the 1st disturbances ei (t) polynomial corresponding delay factor constituted, D2 (z) of the second perturbation e2 (t) polynomial corresponding delay factor constituted, D3 (z) 3 for the first disturbance e3 (t) corresponding to the delay factor polynomial configuration, D1 (z) is the polynomial corresponding to the i-th delay factor of disturbance constituted, ai, a2, ...... ana, bi , b2, ...... bnb, dii, di2, ...... called Hill coefficient is to be identified; ex⑴, e2⑴, e3⑴ time t respectively at the inlet temperature of the spray medium, a medium flow rate steam temperature and the injector inlet.

[0089] 经辨识,ARMAX的模型及参数如下所示: [0089] are recognized, ARMAX model and parameters are as follows:

[0090] y (t) = 1 · 8347y (t-1) -0 · 9132y (t-2) +0 · 0368y (t-3) +0 · 0196y (t-4) [0090] y (t) = 1 · 8347y (t-1) -0 · 9132y (t-2) +0 · 0368y (t-3) +0 · 0196y (t-4)

[0091] -0 · 0077u (t-1) +0 · 0993u (t-2) -0 · 0502u (t-3) [0091] -0 · 0077u (t-1) +0 · 0993u (t-2) -0 · 0502u (t-3)

[0092] -0 · 0310u (t-4) +0 · 2407ei (t-1) -0 · 1771ei (t-2) [0092] -0 · 0310u (t-4) +0 · 2407ei (t-1) -0 · 1771ei (t-2)

[0093] -0 · 2257ei (t-3) +0 · 1705ei (t-4) -0 · 1772e2 (t-1) [0093] -0 · 2257ei (t-3) +0 · 1705ei (t-4) -0 · 1772e2 (t-1)

[0094] +0 · l〇81e2 (t-2) +0 · 0115e2 (t-3) +0 · 0395e2 (t-4) [0094] l〇81e2 +0 · (t-2) +0 · 0115e2 (t-3) +0 · 0395e2 (t-4)

[0095] -0 · 〇966e3 (t-1) -0 · 0049e3 (t-2) +0 · 3456e3 (t-3) [0095] -0.3 〇966e3 (t-1) -0 · 0049e3 (t-2) +0 · 3456e3 (t-3)

[0096] -0.238Ie3 (t-4) [0096] -0.238Ie3 (t-4)

[0097] ⑵离散传递函数的推导和控制矩阵的生成 [0097] The derivation and generation of control matrix ⑵ discrete transfer function

[0098] 利用工业运行数据辨识ARMAX模型中的参数,获得参数模型后直接推导对应过程参数的离散传递函数: [0098] With the industrial operation data identification parameter ARMAX model, directly after derivation of the corresponding process parameters of the model parameters discrete transfer function:

Figure CN104898416BD00121

[0100] 故控制量和三个扰动量ei,e2,e3对应的离散传递函数Fu,Fi,F2,F3依次为: [0100] Therefore, three control amount and the disturbance ei, e2, e3 corresponding to the discrete transfer function Fu, Fi, F2, F3 as follows:

Figure CN104898416BD00122

[0105] 为计算各参数对应的动态矩阵,先求各参数在对应离散传递函数下的阶跃响应, 对于控制量u,采样时刻1,2,……,20对应的系统输出分别为: [0105] To calculate the dynamic matrix of each parameter, each parameter to find the corresponding step responses at discrete transfer function, for controlling the amount of u, sampling time 1,2, ......, 20 corresponds to the output of the system are:

Figure CN104898416BD00123

[0107] 对于扰动量ei,采样时刻1,2,……,20对应的系统输出分别为: [0107] For perturbation ei, sampling time 1,2, ......, 20 corresponds to the output of the system are:

Figure CN104898416BD00124

[0109] 对于扰动量e2,采样时刻1,2,……,20对应的系统输出分别为: [0109] For perturbation e2, sampling time, 2, ......, 20 corresponds to the output of the system are:

Figure CN104898416BD00125

Figure CN104898416BD00131

[0111] 对于扰动量e3,采样时刻I,2,……,20对应的系统输出分别为: [0111] For perturbation e3, sampling time I, 2, ......, 20 respectively corresponding system output:

Figure CN104898416BD00132

[0113] 控制量u,扰动量ei,e2,e3对应的动态矩阵611,61,62,〇3分别为: [0113] control amount u, disturbance ei, e2, e3 corresponding dynamic matrix 611,61,62, 〇3 were:

Figure CN104898416BD00133

[0118] 其中,m为控制步长,在本例中设置为6,N为预测步长,在本例中设置为10。 [0118] where, m is the control step, is set to 6, N is the prediction steps in the present embodiment, is set to 10 in this embodiment. 将对应参数代入各矩阵中,可得: Each of corresponding parameters into the matrix, can be obtained:

Figure CN104898416BD00141

[0123] (3)控制量的计算 [0123] (3) calculation of the control amount

[0124] 基于步骤2所得的各参数动态矩阵,系统预测输出值y'可表达为: [0124] Based on the parameters obtained in step 2 dynamic matrix, the system predicts the output values ​​y 'may be expressed as:

Figure CN104898416BD00151

[0126]上式中,u为控制量按时间序列构成的向量,Ei为扰动量增量△ ei按时间序列构成的向量,Gi为扰动量ei对应的动态矩阵。 [0126] In the above formula, u is the control amount vector composed of time series, Ei is a time sequence of vector ei disturbance increment △, Gi is a dynamic matrix ei corresponding to the disturbance. fu为控制量中影响预测值的只与过去时刻数值有关的部分所构成的列向量,fu= [fu (t+1),fu (t+2)……fu (t+N) ] T,fEi为扰动&中影响预测值的只与过去时刻数值有关的部分所构成的列向量, fu column vector prediction value of the control amount affecting only the time value associated with the last portion constituted, fu = [fu (t + 1), fu (t + 2) ...... fu (t + N)] T, fEi disturbance is & amp; affecting only the predicted value and the value of the last time the relevant portion of the column vector composed of,

Figure CN104898416BD00152

其中fu (t+k)为fu中的第k个元素, Wherein fu (t + k) is the k th fu elements,

Figure CN104898416BD00153

为%中的第k个元素,k= 1,2,……10: % Of the k-th element, k = 1,2, ...... 10:

Figure CN104898416BD00154

[0128] 其中Δ u (ti)为ti时刻的控制量。 [0128] wherein Δ u (ti) to control the amount of time ti.

Figure CN104898416BD00155

[0130] 其中Δ ei (ti)为扰动量61在卜1时刻的增量,fEl为扰动針中影响预测值的只与过去时刻数值有关的部分所构成的列向量, [0130] wherein Δ ei (ti) is a disturbance at one time increments of 61 BU, FEL only affect the column vector value associated with the last time constituting part of the predicted value of disturbance needle,

Figure CN104898416BD00156

为Ie1中的第k个元素,k = 1,2,...... 10。 Ie1 is the k-th element, k = 1,2, ...... 10.

Figure CN104898416BD00157

[0132] 其中Λ e2 (ti)为扰动量62在卜1时刻的增量,fE2为扰动e2*影响预测值的只与过去时刻数值有关的部分所构成的列向量 [0132] wherein Λ e2 (ti) is a disturbance in a time increment of 62 BU, fE2 influence of disturbance e2 * column vector with only the last time value associated portion constituting the predicted value

Figure CN104898416BD00161

为中的第k个元素,k = 1,2,...... 10。 Is the k-th element, k = 1,2, ...... 10.

Figure CN104898416BD00162

[0135] 其中Ae3(ti)为扰动量e3在ti时刻的增量,fEs为扰动e3中影响预测值的只与过去时刻数值有关的部分所构成的列向量 [0135] wherein Ae3 (ti) is the disturbance e3 increase in the time ti, fEs influence of the disturbance predicted value e3 only past time value column vector composed of the relevant portion

Figure CN104898416BD00163

为中的第k个元素,k=l,2,…… 10。 Is the k-th element, k = l, 2, ...... 10.

[0136] 以上一系列式子中,等式的右侧由已测的数值构成,且E1S未来时刻扰动量构成的向量,在当前时刻取值为0。 [0136] In the above series of formulas, the right side of the equation is made of the measured values, and disturbance E1S future time vector composed of the values ​​at the current time is 0. 由此可计算控制量和三个扰动量中影响预测值的只与过去时刻数值有关的部分所构成的列向量f为: Thereby affecting the calculated predicted value of the control amount and the disturbance three column vector f and only the last time values ​​related portion is configured:

Figure CN104898416BD00164

[0138] 为简便起见,将系统预测输出值y'记为: [0138] For simplicity, system predictive output value y 'referred to as:

[0139] y7 =Guu+f [0139] y7 = Guu + f

[0140] 系统建模和动态矩阵获得后,即可确定目标函数进而计算控制量。 [0140] After obtaining matrix and dynamic system modeling, and then calculate the objective function to determine the control amount. 常规目标函数J 形式为: General objective function J in the form of:

Figure CN104898416BD00165

[0142] 其中λ为平衡目标函数两部分(即预测误差和控制耗能)权重的系数,此处取〇.l,m 为控制步长,A u (t+j-1)为t+j-1时刻控制量的变化量,y' (t+j 11)为在t时刻预测的t+j时刻的系统输出值,w (t+j)为t+j时刻的输出参考轨迹,本例中设置为: [0142] wherein λ is a function of two balancing target portion (i.e., the prediction error and control energy) the weight coefficient, here take 〇.l, m is control step, A u (t + j-1) to t + j timing control amount change amount of -1, y '(t + j 11) the system output values ​​predicted at time t time t + j, w (t + j) is the output reference trajectory t + j time, the present embodiment set as follows:

[0143] w (t+j) =0.5w (t+j-1) + (1-0.5) r (t+j) [0143] w (t + j) = 0.5w (t + j-1) + (1-0.5) r (t + j)

[0144] w (t) =y (t) [0144] w (t) = y (t)

[0145] 其中r (t+j)为系统输出在t+j时刻的设定值。 [0145] where r (t + j) is the system output in time t + j set value.

[0146] 求目标函数的最小值,即可获得未来控制步长内最优的控制量。 [0146] minimization of the objective function, an optimum control amount can be obtained in the next control step. 但考虑到蒸汽使用成本,为节能降耗起见,目标函数中应当加入对蒸汽使用量的目标约束。 But considering the cost of steam for energy saving purposes, the objective function should be added to certain constraints on the amount of steam used.

[0147] 蒸汽使用量Q可表示为如下形式: [0147] Use of steam amount Q can be expressed as follows:

[0148] Q = Ws [0148] Q = Ws

[0149] 式中,T为时间间隔,fs为蒸汽流速,而fs正比与阀门开度,在本发明中,控制步长范围内的各个扫描间隔内的蒸汽使用量Q(t+1),Q(t+2),......Q(t+m)为: [0149] where, T is the time interval, fs is the steam flow rate, and fs proportional to the valve opening, in the present invention, the control amount of the steam in the respective scanning interval step length range of Q (t + 1), Q (t + 2), ...... Q (t + m) is:

[0150] Q (t+1) =K* [u ⑴ + Δ u (t+1) ] * Δ T [0150] Q (t + 1) = K * [u ⑴ + Δ u (t + 1)] * Δ T

[0151] Q (t+2) =K* [u ⑴ + Δ u (t+1) + Δ u (t+2) ] * Δ T [0151] Q (t + 2) = K * [u ⑴ + Δ u (t + 1) + Δ u (t + 2)] * Δ T

[0152] · [0152] ·

[0153] · [0153] ·

[0154] · [0154] ·

[0155] Q (t+m) =K* [u ⑴ + Δ u (t+1) + Δ u (t+2)......Δ u (t+m) ] * Δ T [0155] Q (t + m) = K * [u ⑴ + Δ u (t + 1) + Δ u (t + 2) ...... Δ u (t + m)] * Δ T

[0156] 式中,K为常系数,此处取0.05,Δ T为控制间隔时间,即扫描时间间隔,此处取0.5ms,u⑴为当前时刻控制量输入的数值,m为控制步长,此例中数值为6。 [0156] where, K is a constant coefficient, here taken 0.05, Δ T is the control interval of time, i.e., scan interval, here taking 0.5ms, u⑴ value of the current time point as a control input, m is a control step, value is 6 in this embodiment. 在控制步长内, 蒸汽总耗量E为: In the control step, a total steam consumption of E:

Figure CN104898416BD00171

[0158] 上式可表达为如下形式: [0158] the above formula can be expressed as follows:

[0159] E=KXmXu (t)+MTu [0159] E = KXmXu (t) + MTu

[0160] 其中M为总蒸汽使用量经推导后产生的矩阵,u为未来时刻控制量u (t+1),u(t+ 2),……u (t+m)按时间序列构成的向量: [0160] wherein the matrix generated by M is the total amount of steam used was derived, u vector for the future time control amount u (t + 1), u (t + 2), ...... u (t + m) time series consisting of :

[0161] U= [u (t+1) ,u (t+2) ,......u(t+m)]T [0161] U = [u (t + 1), u (t + 2), ...... u (t + m)] T

[0162] M= [6,5,4,3,2,1]T [0162] M = [6,5,4,3,2,1] T

[0163] 如此,加入了对节能指标的考虑后,目标函数J确定为: After [0163] Thus, the added consideration of energy conservation target, the objective function J is determined to be:

Figure CN104898416BD00172

[0165] 求目标函数的最小值,即可得到未来时刻控制量输入。 [0165] minimization of the objective function, to obtain the next timing control input. 将目标函数转换成矩阵运算形式后求导,即可得到: After the objective function is converted into the form of a matrix derivation operation can be obtained:

[0166] J = uT (GuGuT+XI) u_2uTGuT (w_f) +K [mu ⑴ +uTM] [0166] J = uT (GuGuT + XI) u_2uTGuT (w_f) + K [mu ⑴ + uTM]

Figure CN104898416BD00173

[0168] 其中w为目标轨迹按时间序列构成的列向量, [0168] where w is the target track in a time series of a column vector composed,

Figure CN104898416BD00174

,T为转置符号,I 为单位对角矩阵,上标-1为矩阵逆运算符号。 , T is the transpose symbol, I is a unit diagonal matrix, the superscript -1 is an inverse matrix operation symbol. 以上即为动态矩阵控制方法得到的未来时刻控制量输入,按照滚动优化的方式,每次计算u都只需要计算第一行第一列的数值,式中(οΛυ+λΐΓ1为6 X 6矩阵, Above is the dynamic matrix control method to obtain the future time control input, in accordance with the rolling optimized manner, each calculation value u only necessary to calculate a first row and first column, in the formula (οΛυ + λΐΓ1 to 6 X 6 matrix,

Figure CN104898416BD00175

为6 X 1矩阵,故只要取(GjGu+Air1第一行与 To 6 X 1 matrix, as long as it takes (GjGu + and the first row Air1

Figure CN104898416BD00176

相乘即可,λ已取值为1,(G/Gu+Xir1第一行H为: By multiplying, λ has a value of 1, (G / Gu + Xir1 first line H is:

[0169] H= [0.692 ,-0.248 ,-0.187 ,-0.131 ,-0.086 ,-0.051] [0169] H = [0.692, -0.248, -0.187, -0.131, -0.086, -0.051]

[0170] 故下一时刻控制量u (t+1)的值为: [0170] Therefore, the next time the control amount u (t + 1) values:

Figure CN104898416BD00181

[0172] 计算得出数值后将其送至系统执行器即可,如此滚动计算,每一个时刻计算下一个时刻的控制量。 [0172] After calculating the value derived thereof to the actuator system can, thus rolling calculation, the control amount at a time for each time point is calculated.

Claims (2)

  1. I. 一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法,其特征在于:包括如下步骤: 步骤1:利用培养基蒸汽灭菌过程数据,建立灭菌过程ARMAX模型首先收集工业运行现场培养基蒸汽灭菌过程数据,通过对过程数据进行分析,找到灭菌过程所涉及到的过程参数,并确定灭菌控制系统的输入、输出及扰动量,由此确定灭菌控制系统的模型结构;经分析,系统输入为用于加热培养基的蒸汽的单位流量,在实际运行中表现为喷射器蒸汽入口管路的阀门开度,系统输出为喷射器出口温度,即蒸汽和培养基混合后的温度,系统扰动量有:培养基在喷射器入口处温度,培养基流动速度,喷射器入口处蒸汽温度;采用建模的模型结构为ARMX模型,结合灭菌过程的参数及个数,其形式如下: I. one kind of steam injection sterilization medium ARMAX model-based energy-saving control method, characterized by: comprising the following steps: Step 1: using a data medium steam sterilization process, the sterilization process ARMAX model established first collects industrial operation field medium steam sterilization process data, by analysis of the process data, to find the process parameters of the sterilization process involved, and determine the input, output and disturbance sterilization control system, thereby determining a control system model structure sterilization ; after analysis, the system input unit flow rate of steam for heating medium, and in the actual operation of the steam ejector inlet conduit valve opening, the ejector system output is the outlet temperature, i.e., after mixing the steam and the medium temperature, the system disturbance are: injector temperature at the inlet of the medium, the medium flow rate, the inlet temperature of the steam injector; model structure is modeled using ARMX model, and the number of binding parameters of the sterilization process, which form as follows:
    Figure CN104898416BC00021
    其中,y⑴为系统输出,u⑴为系统输入量,A (z)为y (t)对应的延迟因子所构成的多项式,B (z)为控制量u⑴对应的延迟因子所构成的多项式,D1 (z)为第1个扰动量ei⑴对应的延迟因子构成的多项式,D2 (Z)为第2个扰动量Θ2⑴对应的延迟因子构成的多项式,D3 (Z)为第3个扰动量e3 (t)对应的延迟因子构成的多项式,D1 (Z)为第i个扰动量对应的延迟因子所构成的多项式,ai,a2,......ana,bi,b2,......bnb,dii,di2,......(^化山为待辨识系数;ei (t),e2 (t),e3 (t)分别为t时刻培养基在喷射器入口处温度,培养基流动速度和喷射器入口处蒸汽温度; 步骤2:离散传递函数的推导和控制矩阵的生成利用工业运行数据辨识ARMAX模型中的参数,获得参数模型后直接推导对应过程参数的离散传递函数: Wherein, y⑴ system output, u⑴ system input, the polynomial A (z) to y (t) corresponding to the delay factor constituted, B (z) is the control amount u⑴ corresponding delay factor constituted polynomial, Dl ( z) is of a disturbance polynomial ei⑴ corresponding delay factor constituted, D2 (Z) is a polynomial of second perturbation Θ2⑴ corresponding delay factor constituted, D3 (Z) of the third disturbance e3 (t) polynomials corresponding to the configuration of the delay factor, D1 (Z) is the i th perturbation corresponding delay factor polynomials constituted, ai, a2, ...... ana, bi, b2, ...... bnb , dii, di2, ...... (^ coefficient of the mountain to be identified; ei (t), e2 (t), e3 (t) at time t, respectively, at the inlet temperature of the spray medium, a medium flow speed and temperature of the steam inlet of the injector; step 2: industrial generated using a discrete derivation and a transfer function control matrix operation data identification parameters ARMAX model, directly after derivation of the corresponding process parameters of the model parameters discrete transfer function:
    Figure CN104898416BC00022
    故控制量和三个扰动量ei,e2,e3对应的离散传递函数Fu,F1,F2,F3依次为: So that the control amount and the disturbance three ei, e2, e3 corresponding to the discrete transfer function Fu, F1, F2, F3 as follows:
    Figure CN104898416BC00023
    为计算各参数对应的动态矩阵,先求各参数在对应离散传递函数下的阶跃响应,其形式如下: 对于控制量u,采样时刻1,2,……,nu对应的系统输出分别为: gi,g2,......,Snu 对于扰动量ex,采样时刻I,2,……,应的系统输出分别为: fll,fl2,......,打%! 对于扰动量e2,采样时刻1,2,……,!^2对应的系统输出分别为: f2i,f2z,……,f2ne2 对于扰动量e3,采样时刻1,2,……,应的系统输出分别为: f3i,f32,……,βΠθ3 由此,控制量U,扰动量ΘΙ,Θ2,Θ3对应的动态矩阵Gu,Gi,G2,G3分别为: To calculate the dynamic matrix of each parameter, each parameter to find the corresponding step responses at discrete transfer function, the following form: For the control amount u, sampling time 1,2, ......, nu output corresponding system are: gi, g2, ......, Snu perturbation for ex, sampling time I, 2, ......, corresponding system output are:! fll, fl2, ......, for playing perturbation% e2, sampling time 1,2, ......, ^ 2 corresponding to the output of the system are:! f2i, f2z, ......, f2ne2 for perturbation e3, sampling time 1,2, ......, corresponding system output are: f3i, f32, ......, βΠθ3 thereby controlling the amount of U, disturbance ΘΙ, Θ2, Θ3 corresponding dynamic matrix Gu, Gi, G2, G3 are:
    Figure CN104898416BC00031
    其中,m为控制步长,N为预测步长; 步骤3:控制量的计算基于步骤2所得的各参数动态矩阵,系统预测输出值y'表达为: Wherein, m is the control step, N is the prediction steps; Step 3: the control parameters based on the calculated amount of dynamic matrix obtained in step 2, the system predicts the output values ​​y 'are expressed as:
    Figure CN104898416BC00032
    AT,u 73 e市|」M ί女H、」丨叩亍少P乂tfJ I η」M,β i为扰动量ei的增量按时间序列构成的向量,Gi为扰动量ei对应的动态矩阵;fu为控制量中影响预测值的只与过去时刻数值有关的部分所构成的列向量,fu= [fu(t+l),fu(t+2)……fu(t+N)]T,fEi为扰动出中影响预测值的只与过去时刻数值有关的部分所构成的列向i AT, u 73 e City | "M ί F H," right foot tapping less Shu qe tfJ I η P "M, β i is the disturbance vector ei increment time series configuration, Gi is the corresponding dynamic disturbance ei matrix; fu column vector over time and only partially affected values ​​related to the predicted value of the control amount constituted, fu = [fu (t + l), fu (t + 2) ...... fu (t + N)] column T, fEi disturbance to the time value associated with the last portion of affecting only the prediction value of the i constituted
    Figure CN104898416BC00041
    其中: among them:
    Figure CN104898416BC00042
    其中fu (t+k)为fu中的第k个元素,iEi (t + k)为.¾中的第k个元素,N为预测步长,△ u (t- i)为ti时刻的控制量增量,△ ei (tj)为扰动量&在卜」时刻的增量,gl为控制量在i时刻的阶跃响应值,fij为扰动量ei在j时刻的阶跃响应值; 为简便起见,将系统预测输出值y7记为: y7 = Guu+f 其中 Wherein fu (t + k) is the k th fu elements, iEi (t + k) for the k-th .¾ the element, N is the prediction steps, △ u (ti) to control the timing ti increment, △ ei (tj) is the disturbance & amp; increment Bu "time, gl step response of the control amount value at the time i, fij disturbance step response value ei j at moment; is simplicity, the system is referred to as a predictive output value y7: y7 = Guu + f wherein
    Figure CN104898416BC00043
    :i,系统建模和动态矩阵获得后,即可确定目标函数进而计算控制量;常规目标函数J形式为: : I, the system model and the dynamic matrix is ​​obtained, to determine the objective function further calculates a control amount; conventional objective function J of the form:
    Figure CN104898416BC00044
    其中λ为平衡目标函数两部分即预测误差和控制耗能权重的系数,N为预测步长,m为控制步长,Δ u (t+j-Ι)为t+j-Ι时刻控制量的变化量,y' (t+j 11)为在t时刻预测的t+j时刻的系统输出值,w (t+j)为t+j时刻的输出参考轨迹,设置为如下形式: Where λ is the balancing target function of two parts i.e., the prediction error and control weight energy weighting coefficient, N is the prediction steps, m is control step size, Δ u (t + j-Ι) of t + j-Ι timing control amount the amount of change, y '(t + j 11) the system output values ​​predicted at time t time t + j, w (t + j) is the output time t + j reference trajectory is provided as follows:
    Figure CN104898416BC00045
    其中r (t+j)为系统输出在t+j时刻的设定值,y⑴为t时刻系统输出量,α为〇到1之间的常数; 求目标函数的最小值,即获得未来控制步长内最优的控制量;但考虑到蒸汽使用成本, 为节能降耗起见,目标函数中应当加入对蒸汽使用量的目标约束; 蒸汽使用量Q表示为如下形式: Q = T*fs 式中,T为时间间隔,fs为蒸汽流速,而fs正比与阀门开度,则控制步长范围内的各个扫描间隔内的蒸汽使用量Q(t+1),Q(t+2),……Q(t+m)为: Wherein r (t + j) is the system output set value in the time t + j, y⑴ output of the system time t, [alpha] is a constant between the square to 1; minimization of the objective function, i.e., to obtain the next control step the optimal length of the control amount; but considering the cost of steam for energy saving purposes, the objective function should be added to certain constraints on the amount of steam used; the use of steam amount Q is expressed as follows: Q = T * fs formula , T is the time interval, as the steam flow rate fs, fs proportional to the valve opening, the control amount of the steam used for each scanning interval step length range of Q (t + 1), Q (t + 2), ...... Q (t + m) is:
    Figure CN104898416BC00046
    式中,K为常系数,AT为控制间隔时间,即扫描时间间隔,u⑴为当前时刻控制量输入的数值,m为控制步长;在控制步长内,蒸汽总耗量E为: Where, K is a constant coefficient, the AT control interval, i.e., the scan time interval, u⑴ value of the current time point as a control input, m is a control step; in the control step, a total steam consumption of E:
    Figure CN104898416BC00051
    上式表达为如下形式: The formula is expressed as follows:
    Figure CN104898416BC00052
    其中m为控制步长,M为从蒸汽总消耗量的计算过程中推导出的矩阵,u为未来时刻控制量u(t+l),u(t+2),……u (t+m)按时间序列构成的向量,T为转置符号: Wherein m is the control step size, M being deduced from the calculation of the total consumption of steam in the matrix, u is the next timing control amount u (t + l), u (t + 2), ...... u (t + m ) according to the time sequence of the vector, T is the transpose symbol:
    Figure CN104898416BC00053
    如此,加入了对节能指标的考虑后,目标函数J确定为: After that, it takes into account of energy-saving targets, the target function J is determined as:
    Figure CN104898416BC00054
    求目标函数的最小值,即得到未来时刻控制量输入;将目标函数转换成矩阵运算形式后求导,即可得到: Seeking the minimum of the objective function, i.e., to obtain the next time the control input; after converting the derivative into the objective function in the form of matrix operations, can be obtained:
    Figure CN104898416BC00055
    其中 among them
    Figure CN104898416BC00056
    _ f为输出轨迹按时间顺序构成的序列,T为转置符号,I为单位对角矩阵,上标-1为矩阵逆运算符号; 以上即为动态矩阵控制方法得到的未来时刻控制量输入,按照滚动优化的方式,每次计算U都只需要计算第一行第一列的数值,然后送至系统执行器即可。 F _ a sequence of outputs chronologically track configuration, T is the transpose symbol, I is a unit diagonal matrix, the superscript -1 is an inverse matrix operation symbol; a control above is the future time input dynamic matrix control method obtained, rolling optimization according to the embodiment, each calculation value U only necessary to calculate a first row and first column, and then sent to the system actuators.
  2. 2.权利要求1所述的一种基于ARMAX建模的培养基蒸汽喷射灭菌节能控制方法,其特征在于:应用于生物制药中以蒸汽喷射器对各种发酵用培养基进行喷射加热灭菌的控制系统。 One of the claim 1, the medium steam injection sterilization method of the power saving control based on ARMAX model, wherein: biopharmaceutical applied to various fermented steam injector is injected to heat sterilization medium control system.
CN201510158900.3A 2015-04-03 2015-04-03 One sterilization medium energy-saving method of steam injection based modeling armax CN104898416B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510158900.3A CN104898416B (en) 2015-04-03 2015-04-03 One sterilization medium energy-saving method of steam injection based modeling armax

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510158900.3A CN104898416B (en) 2015-04-03 2015-04-03 One sterilization medium energy-saving method of steam injection based modeling armax

Publications (2)

Publication Number Publication Date
CN104898416A CN104898416A (en) 2015-09-09
CN104898416B true CN104898416B (en) 2017-06-23

Family

ID=54031135

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510158900.3A CN104898416B (en) 2015-04-03 2015-04-03 One sterilization medium energy-saving method of steam injection based modeling armax

Country Status (1)

Country Link
CN (1) CN104898416B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5568378A (en) * 1994-10-24 1996-10-22 Fisher-Rosemount Systems, Inc. Variable horizon predictor for controlling dead time dominant processes, multivariable interactive processes, and processes with time variant dynamics
CN101609480A (en) * 2009-07-13 2009-12-23 清华大学 Inter-node phase relation identification method of electric system based on wide area measurement noise signal
CN101807083A (en) * 2010-03-30 2010-08-18 杭州电子科技大学 Method for optimally controlling humidity in dyeing process
CN103269176A (en) * 2013-05-17 2013-08-28 南京信息工程大学 Inverter control method based on fractional order PI forecasting function

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9447963B2 (en) * 2010-08-16 2016-09-20 Emerson Process Management Power & Water Solutions, Inc. Dynamic tuning of dynamic matrix control of steam temperature

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5568378A (en) * 1994-10-24 1996-10-22 Fisher-Rosemount Systems, Inc. Variable horizon predictor for controlling dead time dominant processes, multivariable interactive processes, and processes with time variant dynamics
CN101609480A (en) * 2009-07-13 2009-12-23 清华大学 Inter-node phase relation identification method of electric system based on wide area measurement noise signal
CN101807083A (en) * 2010-03-30 2010-08-18 杭州电子科技大学 Method for optimally controlling humidity in dyeing process
CN103269176A (en) * 2013-05-17 2013-08-28 南京信息工程大学 Inverter control method based on fractional order PI forecasting function

Also Published As

Publication number Publication date
CN104898416A (en) 2015-09-09

Similar Documents

Publication Publication Date Title
Dochain et al. Modelling and adaptive control of nonlinear distributed parameter bioreactors via orthogonal collocation
Wang et al. Hybrid differential evolution for problems of kinetic parameter estimation and dynamic optimization of an ethanol fermentation process
Lee et al. Enhanced process monitoring of fed-batch penicillin cultivation using time-varying and multivariate statistical analysis
Chen et al. Modelling and optimization of fed-batch fermentation processes using dynamic neural networks and genetic algorithms
US6088630A (en) Automatic control system for unit operation
Agachi et al. Model based control: case studies in process engineering
Karakuzu et al. Modelling, on-line state estimation and fuzzy control of production scale fed-batch baker's yeast fermentation
Aly PID parameters optimization using genetic algorithm technique for electrohydraulic servo control system
Yu et al. Implementation of neural network predictive control to a multivariable chemical reactor
Gernaey et al. Application of mechanistic models to fermentation and biocatalysis for next-generation processes
Vasičkaninová et al. Neural network predictive control of a heat exchanger
Singh et al. Model-based computer-aided framework for design of process monitoring and analysis systems
Selişteanu et al. Sliding mode and adaptive sliding‐mode control of a class of nonlinear bioprocesses
Royce A discussion of recent developments in fermentation monitoring and control from a practical perspective
US20100082312A1 (en) Optimizing product drying through parallel lines of centrifuges and dryer process units
Mahadevan et al. Differential flatness based nonlinear predictive control of fed-batch bioreactors
US9046882B2 (en) Nonlinear model predictive control of a batch reaction system
CA2509558A1 (en) Steam temperature control system, method of controlling steam temperature and power plant using the same
Simeonov et al. Linearizing control of the anaerobic digestion with addition of acetate (control of the anaerobic digestion)
Farinas et al. Modeling the effects of solid state fermentation operating conditions on endoglucanase production using an instrumented bioreactor
Ye et al. Approximating necessary conditions of optimality as controlled variables
Williams COMPUTERS AUTOMATION AND PROCESS CONTROL
Arbaoui et al. Counter-current tubular heat exchanger: Modeling and adaptive predictive functional control
Oberle et al. Numerical computation of optimal feed rates for a fed-batch fermentation model
Celikovsky et al. Singular perturbation based solution to optimal microalgal growth problem and its infinite time horizon analysis

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
GR01