CN115877811B

CN115877811B - Flow process treatment method, device and equipment

Info

Publication number: CN115877811B
Application number: CN202310215775.XA
Authority: CN
Inventors: 张善贵; 樊泳; 刘聪; 雷鸣
Original assignee: Business Intelligence Of Oriental Nations Corp ltd
Current assignee: Business Intelligence Of Oriental Nations Corp ltd
Priority date: 2023-03-08
Filing date: 2023-03-08
Publication date: 2023-05-23
Anticipated expiration: 2043-03-08
Also published as: CN115877811A

Abstract

The application provides a flow process treatment method, a device and equipment, and relates to the technical field of process control, wherein the method comprises the following steps: obtaining an alignment offset corresponding to an input variable of production process control and sampling values of output variables at a plurality of first sampling moments; performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the sampling values of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating the corresponding relation between the input variable and the output variable; and determining the set value of the input variable according to a preset process condition, the target values of the output variables at the first sampling moments and the data fusion matrix. The effect of optimizing and controlling the flow process is improved.

Description

Flow process treatment method, device and equipment

Technical Field

The present disclosure relates to the field of process control technologies, and in particular, to a method, an apparatus, and a device for processing a process.

Background

The process optimization control is a very important technology in the process manufacturing industry, and the process control comprises an input variable and an output variable, wherein the input variable is an operation variable of the process, and the output variable is a controlled variable of the process.

The output variables and the input variables in the process have causal relation, and different output variables are obtained after corresponding process treatment based on different input variables. Therefore, in the flow process, the purpose of optimizing the flow process can be achieved by optimizing the value of the input variable and controlling the value of the output variable.

In the current process optimization control, the value of an input variable is usually input manually according to an empirical value, and the value is too dependent on human experience, so that the effect of the process optimization control is poor.

Disclosure of Invention

The application provides a flow process processing method, a device and equipment, which are used for solving the problem that the effect of optimizing and controlling the flow process is poor because the value of an input variable in the prior art is too dependent on human experience.

In a first aspect, the present application provides a process treatment method, including:

obtaining an alignment offset corresponding to an input variable of production process control and measured values of output variables at a plurality of first sampling moments;

Performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured value of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variable and the output variable;

and determining the set value of the input variable according to a preset process condition, the target values of the output variables at the first sampling moments and the data fusion matrix.

In one possible implementation manner, the performing an alignment fusion process on the input variable and the output variable according to the alignment offset and the measured values of the output variable at the first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the first sampling moments, where the data fusion matrix includes:

determining sampling intervals of the input variable at the plurality of first sampling moments according to the alignment offset and the plurality of first sampling moments;

and obtaining the data fusion matrix according to the sampling intervals of the input variables at the first sampling moments and the measured values of the output variables at the first sampling moments.

In a possible implementation manner, the determining, according to the alignment offset and the plurality of first sampling moments, a sampling interval of the input variable at the plurality of first sampling moments includes:

determining alignment time between the input variable and the output variable at each first sampling time according to the alignment offset and the plurality of first sampling times;

and determining sampling intervals of the input variables at the plurality of first sampling moments according to the alignment moments and the preset time length.

In one possible implementation manner, the obtaining the data fusion matrix according to the sampling intervals of the input variables at the first sampling moments and the measured values of the output variables at the first sampling moments includes:

obtaining sampling values of the input variable at a plurality of first sampling moments according to the sampling intervals of the input variable at the plurality of first sampling moments, wherein the sampling values of the input variable at the first sampling moments comprise a plurality of sampling values of the input variable, the second sampling moments of which are positioned in the sampling intervals, aiming at the sampling intervals of the input variable at any first sampling moment;

And obtaining the data fusion matrix according to the sampling values of the input variables and the measured values of the output variables at the first sampling moments.

In one possible implementation manner, the determining the set value of the input variable according to the preset process condition, the target value of the output variable at the plurality of first sampling moments and the data fusion matrix includes:

determining an optimized control function between the input variable and the output variable according to the data fusion matrix;

and determining a set value of the input variable according to the target value of the output variable, the optimization control function and the preset process condition.

In a possible implementation manner, the determining an optimized control function between the input variable and the output variable according to the data fusion matrix includes:

determining an initial control function between the input variable and the output variable according to the data fusion matrix;

and updating the initial control function in real time according to the real-time value of the input variable and the actual measurement value of the output variable to obtain the optimized control function.

In one possible implementation manner, the updating the initial control function in real time according to the real-time value of the input variable and the real-time value of the output variable to obtain the optimized control function includes:

Performing at least one first operation, any ith first operation including: acquiring a predicted value of the output variable according to the real-time value of the input variable and the ith round of control function; the i is an integer greater than or equal to 1, and the 1 st round of control function is the initial control function;

when the difference value between the predicted value of the output variable and the actual measured value of the output variable is greater than or equal to a preset value, updating the ith round of control function according to the difference value to obtain an (i+1) th round of control function, and executing a (i+1) th first operation according to the (i+1) th round of control function;

and determining the ith round of control function as the optimized control function when the difference value between the predicted value of the output variable and the measured value of the output variable is smaller than the preset value.

In a second aspect, the present application provides a flow process treatment apparatus, comprising:

the acquisition module is used for acquiring alignment offset corresponding to an input variable of production process control and measured values of output variables at a plurality of first sampling moments;

the processing module is used for carrying out alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured value of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variable and the output variable;

And the determining module is used for determining the set value of the input variable according to the preset process condition, the target values of the output variables at the plurality of first sampling moments and the data fusion matrix.

In a possible implementation manner, the processing module is specifically configured to:

In one possible implementation manner, the determining module is specifically configured to:

In a third aspect, the present application provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the flow process processing method according to any one of the first aspects when executing the program.

In a fourth aspect, the present application provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the flow process treatment method according to any of the first aspects.

In a fifth aspect, the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the flow process treatment method according to any one of the first aspects.

The method, the device and the equipment for processing the flow process provided by the embodiment of the application firstly acquire the alignment offset corresponding to the input variable and the sampling values of the output variable at a plurality of first sampling moments; then, according to the alignment offset and sampling values of the output variables at a plurality of first sampling moments, performing alignment fusion processing on the input variables and the output variables to obtain a data fusion matrix of the input variables and the output variables at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variables and the output variables; and finally, determining a set value of the input variable according to the preset process condition, the target values of the output variables at a plurality of first sampling moments and the data fusion matrix. By aligning the offset and a plurality of first sampling moments, alignment fusion between the input variable and the output variable is realized, and a data fusion matrix is obtained, so that the function corresponding relation between the input variable and the output variable can be determined based on the data fusion matrix, the set value of the input variable is determined based on the function corresponding relation, the output variable control target value and the process constraint condition of the input variable, the subsequent process optimization and control are performed, the set value of the input variable is not required to be set manually, and the process optimization control effect is good.

Drawings

For a clearer description of the present application or of the prior art, the drawings that are used in the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description below are some embodiments of the present application, and that other drawings may be obtained from these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic view of an application scenario provided in an embodiment of the present application;

fig. 2 is a schematic flow chart of a process treatment method according to an embodiment of the present application;

fig. 3 is a schematic diagram of a real-time alignment fusion process provided in an embodiment of the present application;

FIG. 4 is a schematic flow chart of determining a set value of an input variable according to an embodiment of the present disclosure;

FIG. 5 is a flowchart of a process provided in an embodiment of the present application;

FIG. 6 is a schematic diagram of an implementation of the PID controller algorithm provided in an embodiment of the present application;

FIG. 7 is a graph showing the comparison between the original test signal and the output signal of the PID controller under different PID control coefficients according to the embodiment of the present application;

FIG. 8 is a schematic diagram of model algorithm control provided in an embodiment of the present application;

FIG. 9 is a basic schematic diagram of a dynamic matrix control algorithm according to an embodiment of the present application;

fig. 10 is a schematic structural diagram of a flow process treatment device according to an embodiment of the present application;

fig. 11 is a schematic entity structure diagram of an electronic device according to an embodiment of the present application.

Detailed Description

For the purposes of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions in the present application will be clearly and completely described below with reference to the drawings in the present application, and it is apparent that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

Process optimization control is a very important technology in the process manufacturing industry. In recent years, with the rapid development of industrial internet technology, it has become more common and necessary to apply technical means such as big data, artificial intelligence, etc. in the process to realize automation, informatization and intelligence of industrial manufacturing process. Aiming at a specific production process flow, mass data of input variables (operation variables) and output variables (controlled variables) in the production process are collected, and then alignment and mapping of the input variables and the output variables in the production process are realized according to a process flow rule based on a process production mechanism model and a statistical technology, so that causal relation mapping between process input control conditions and output quality results is formed.

The application scenario of the present application may be understood, for example, in connection with fig. 1. Fig. 1 is a schematic view of an application scenario provided in the embodiment of the present application, and as shown in fig. 1, the application scenario is a process scenario of cement production, and raw materials required for cement production include, for example, free calcium, limestone, iron, silicon, and the like.

In the process of producing cement, the input variables can include, for example, the proportions of raw materials such as free calcium, limestone, iron, silicon and the like, and can include wind temperature, high-temperature fan rotating speed, kiln head boiler inlet temperature, head coal wind pressure and the like, and the input variables can influence the quality of the produced cement to a certain extent.

After the input variables are set, the finished cement can be finally generated through a cement production line, and after the finished cement is generated, the finished cement needs to be detected to judge whether the specified requirement is met. The output variables of the cement process can be obtained after the finished cement is tested, and can include, for example, free calcium of the cement, vertical weight, 3-day strength of the cement, 28-day strength of the cement, and the like. Based on these output variables, it is possible to obtain whether the finished cement meets the requirements.

When the input variable is changed, the output variable is correspondingly transformed, namely, a certain causal relation exists between the input variable and the output variable. For example, in fig. 1, if less raw material iron is added, the iron oxide content of the finished cement may be affected. Therefore, the purpose of flow process optimization is to control the output variable by controlling the input variable so as to enable the produced finished product to meet the requirement, further, on the premise of enabling the produced finished product to meet the requirement, the power consumption is reduced as much as possible, and the purpose of energy saving is achieved. It should be noted that, in fig. 1, only the process of cement production is described as an example, but the process of cement production in the embodiments of the present application is not limited to the process of cement production, and may be applied to various processes.

The current control system for the flow process is mainly implemented based on Proportional-Integral-Derivative (PID) and model predictive control (Model Predictive Control, MPC) control methods. PID control is one of the earliest control strategies developed in the industrial field, namely, according to the control deviation between a given value and an actual output value, the deviation is formed into a control quantity by linear combination according to proportion, integral and derivative, and a controlled object is controlled.

PID control is one of the control algorithms developed in the industry at the earliest time, and is still the most widely used control method in the current industrial control field due to the characteristics of simple algorithm structure, good robustness and high reliability. But on the other hand, as a linear controller, there are some disadvantages in terms of complicated flow process control. In particular, in the actual industrial production process, some control scenarios often have characteristics of nonlinearity, time-varying uncertainty, strong interference and the like, and it is sometimes difficult to achieve an ideal control effect by applying a conventional PID controller. In some production application scenes with complex processes, due to the complicated parameter setting method, the conventional PID controller often has poor parameter setting and poor effect, and has poor adaptability to the operation conditions.

The traditional PID controller is triggered only by the principle of industrial process control, does not consider the economic factors of energy conservation and consumption reduction in the industrial process control process, and does not accord with the transformation of the future industry to the digital, informationized and intelligent directions. Because of these factors, conventional PID control systems are limited in the industrial application fields with high complexity and strict performance requirements, and improvements and optimization are needed.

MPC is another common industrial control technology, and the basic principle of MPC is: the current control action of the system is obtained by solving a finite time domain open-loop optimal control problem at each sampling instant, the current state of the process is used as the initial state of the optimal control problem, and the solved optimal control strategy only acts on the current control process. The MPC algorithm is essentially a solution to the open loop optimal control problem, which is also the most essential difference of MPC algorithms from other pre-calculated control algorithms.

Compared with other traditional control algorithms, the MPC algorithm has the following main characteristics: (1) The accuracy requirement on the model is low, the modeling is convenient, and the process description can be obtained by a simple test; (2) The discrete convolution and model described by non-minimization is adopted, so that the information redundancy is large, and the robustness and stability of the system are improved; (3) By adopting a rolling optimization strategy instead of global one-time optimization, the optimization calculation is repeatedly performed on line, and rolling implementation can timely make up uncertainty caused by factors such as model adaptation, distortion, interference and the like, so that the dynamic performance is good; (4) The algorithm is easy to popularize to practical application scenes such as constraint, delay, non-minimum phase, nonlinearity and the like, and the problem of multiple variables and constraint can be effectively solved.

On the other hand, as a control algorithm based on an optimization model, the MPC model predictive control has some disadvantages in the practical application process: the method can not describe an unstable system, is not suitable for an unstable object, is difficult to identify a system model on line, starts from a process energy-saving and consumption-reducing target, starts from a process control target only, and does not have a method for achieving a double optimization control target of process control and energy-saving and consumption-reducing in the process industry.

In summary, the above two process control methods all need to manually give the value of the input variable and then perform the actual control process based on the value of the output variable, that is, the above two process control methods all only relate to a specific control process, but do not relate to a process how to perform optimization based on the value of the input variable before the control process, and cannot achieve effective process optimization.

Based on the above, the embodiment of the application provides a flow process treatment scheme, which determines the target value of the input variable on the basis of realizing the alignment and fusion of the input variable and the output variable, and realizes the optimization of the flow process. The main purpose of the embodiment of the application is to solve the defects of prior art PID control, MPC model predictive control and other prior art control systems in the practical application process, and adopt a process optimization control system based on data real-time alignment fusion to realize real-time optimization control of the process. The scheme of the embodiment of the present application will be described with reference to fig. 2.

Fig. 2 is a flow chart of a flow process treatment method provided in an embodiment of the present application, as shown in fig. 2, the method may include:

s21, obtaining alignment offset corresponding to the input variable of the production process control and measured values of the output variable at a plurality of first sampling moments.

The input variables may also be referred to as manipulated variables, for example, in a flow process, the input variables may include raw material ratios, contents, control temperature, humidity, etc. in the flow. The output variable may also be referred to as a controlled variable, i.e. a variable controlled by an operating variable, the output variable being affected by the input variable. In the flow process, the output variables may be various parameters of the finished product to be produced, for judging whether the produced finished product meets the requirements, and the like. The first sampling time is the sampling time of the output variable, and the measured value is the value obtained by sampling the output variable at the first sampling time. In the embodiment of the application, the input variables may include one or more, and the output variables may also include one or more.

S22, performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured values of the output variable at a plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating the functional correspondence between the input variable and the output variable.

In the process, the output variable corresponding to the input variable is not obtained immediately after the input variable is determined, but a certain delay exists. Taking cement production as an example, after an input variable is set, the cement product corresponding to the input variable can be obtained only by carrying out pipeline production according to the input variable, and then the cement product corresponding to the input variable is sampled to obtain a corresponding output variable. For example, when the input variable is 1800 ℃, after the input variable is 1800 ℃, the production line is required to be carried out according to the set process formula, so that the finished cement produced at the temperature of 1800 ℃ can be obtained, and further the output variable, such as the content of ferric oxide in the cement, is obtained.

In the pipeline, cement is continuously produced, and the input variable and the output variable are also produced in real time, but because a certain delay exists between the input variable and the output variable, for example, after the input variable is determined, the output variable corresponding to the input variable can be obtained after five hours are possibly needed, and the delay is unknown, therefore, alignment fusion processing is needed to be carried out on the input variable and the output variable, a data fusion matrix of the input variable and the output variable is determined, and the corresponding relation between each input variable and the output variable is determined based on the data fusion matrix, so that the causal mapping relation between the input variable and the output variable can be determined based on the aligned corresponding relation.

Therefore, in the embodiment of the application, according to the alignment offset and the first sampling moments, the alignment fusion processing is performed on the input variable and the output variable, so that a data fusion matrix of the input variable and the output variable at the first sampling moments is obtained, and the function corresponding relation of each output variable and each input variable is determined through the data fusion matrix.

S23, determining a set value of an input variable according to preset process conditions, target values of the output variable at a plurality of first sampling moments and a data fusion matrix.

The preset process conditions are necessary conditions for producing qualified finished products, and may include one or more conditions of a temperature range, a humidity range, a raw material proportioning range and the like. Taking the temperature range as an example, the temperature range specified by the preset process condition is set to 1500-1800 ℃, and the temperature range higher than 1800 ℃ can cause equipment damage, the temperature range lower than 1500 ℃ can cause the failure of producing qualified finished products, and the like, so that the temperature needs to be set to 1500-1800 ℃.

According to the preset process conditions, target values of the output variables at a plurality of first sampling moments and the data fusion matrix, the set values of the input variables can be determined. Specifically, in S22, since the data fusion matrix determined according to the first sampling time and the alignment offset indicates the output variable corresponding to each input variable, and alignment fusion between the input variable and the output variable is implemented, based on the data fusion matrix, a causal mapping relationship between the input variable and the output variable may be determined, where the causal mapping relationship reflects how the output variable changes correspondingly in the case where the input variable changes. Thus, based on the causal mapping, and the desired power consumption saving objective, or the end product requirement, embodied by the output variable objective, an optimal set point for the input variable can be determined. That is, the preset process conditions, the target values of the output variables at the plurality of first sampling moments, and the data fusion matrix are combined, and the set values of the output variables can be determined. Subsequently, the set value of the input variable can be input into the process operation system to realize the purpose of optimizing the process.

The flow process processing method provided by the embodiment of the application comprises the steps of firstly, obtaining an alignment offset corresponding to an input variable and sampling values of the output variable at a plurality of first sampling moments; then, according to the alignment offset and sampling values of the output variables at a plurality of first sampling moments, performing alignment fusion processing on the input variables and the output variables to obtain a data fusion matrix of the input variables and the output variables at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variables and the output variables; and finally, determining a set value of the input variable according to the preset process condition, the target values of the output variables at a plurality of first sampling moments and the data fusion matrix. By aligning the offset and a plurality of first sampling moments, alignment fusion between the input variable and the output variable is realized, and a data fusion matrix is obtained, so that the function corresponding relation between the input variable and the output variable can be determined based on the data fusion matrix, the set value of the input variable is determined based on the function corresponding relation, the output variable control target value and the process constraint condition of the input variable, the subsequent process optimization and control are performed, the set value of the input variable is not required to be set manually, and the process optimization control effect is good.

Compared with the current process PID control system or MPC model predictive control system, the technology provided by the embodiment of the application has the greatest characteristics that multiple linear regression, a neural network, a random forest, multi-objective optimization and other artificial intelligent algorithms are introduced: for a distributed control system (Distributed Control System, DCS) of the flow process, real-time data (operation variable or input variable) and process quality index data (controlled variable or output variable), real-time alignment and fusion of the flow process DCS real-time data and the process quality index data are realized according to a flow process mechanism model, process experience and sampling data statistical processing technology.

Then, by fusing the generated multidimensional process data matrix, a proper process input variable (operation variable) and an output variable (controlled variable) are selected, and an input/output control model of the flow process is created. Based on the optimization target of flow process control (for example, the process output variable meets the quality specification requirement of the process, and meanwhile, the energy consumption in the production process is the lowest) and the process input/output control model, an optimization process control model of the process is established, the optimal value of the current process optimization control model is calculated in real time, the current optimal value is used as the target value of the operation variable of the process control system, and the PID control algorithm or the MPC model predictive control algorithm is combined to realize the real-time process control of the process operation variable. This process will be described in detail below.

First, an alignment offset corresponding to an input variable is acquired.

Table 1 below illustrates one input variable and corresponding alignment offset, X in Table 1 _k The input variables are the names of the input variables, the variable types are the input variables, also called the operation variables, t _k For the corresponding alignment offset.

TABLE 1

Then, sampling values of the output variable at a plurality of first sampling moments are acquired.

Table 2 below illustrates an output variable and corresponding first sample time, Y in Table 2 _n Is index data, belongs to one of output variables, T _p Is the corresponding first sampling instant. For example, the first sampling instant is T ₁ Output variable Y ₁ Is y ₁₁ Output variable Y ₂ Is y ₁₂ Output variable Y _n Is y _1n And so on.

TABLE 2

Secondly, the real-time alignment fusion of the process DCS real-time data (process control operation variable or process control input variable) and the process quality index data (process controlled variable or process output variable) is carried out.

The above described real-time alignment fusion process is described below in conjunction with fig. 3. Fig. 3 is a schematic diagram of a real-time alignment fusion process provided in an embodiment of the present application, where, as shown in fig. 3, the process includes:

s31, determining sampling intervals of input variables at a plurality of first sampling moments according to the alignment offset and the plurality of first sampling moments.

After the alignment offset and the plurality of first sampling moments are obtained, firstly determining the alignment moment between the input variable and the output variable at each first sampling moment according to the alignment offset and the plurality of first sampling moments.

Specifically, each process operation variable data and the alignment rule between controlled variable data are based on the sampling time stamp of the controlled variable (process quality index) data.

For each input variable X _i (i=1, 2..and k) with the output variable, according to which the input variable X is first of all _i The alignment offset of the input variable and the first sampling time of the output variable are calculated, and the alignment time is the actual input variableThe time stamps are aligned.

For example for input variable X ₁ It can be seen from the above Table 1 that it is related to the output variable Y ₁ 、Y ₂ 、...、Y _n Is aligned by an offset of t ₁ The alignment time of the output variable with the q-th sampling value is (T) _q -t ₁ ) Wherein q is more than or equal to 1 and less than or equal to p. For arbitrary input variable X _i The alignment time of the sampling value (s is more than or equal to 1 and less than or equal to p) of the s-th sampling value of the output variable is (T) _s -t _i ). That is, the alignment time between the input variable and the output variable at each first sampling time is the difference between the first sampling time of the output variable and the alignment offset of the input variable.

After the alignment time between the input variable and the output variable at each first sampling time is obtained, the sampling intervals of the input variable at a plurality of first sampling times can be determined according to the alignment time and the preset time.

The preset duration is set to be M, and M is greater than or equal to 0. Specifically, in order to reduce alignment errors between the input variable and the output variable, an alignment time interval of the input variable is defined as [ -M, M]For example, input variable X _i The alignment time of the sampling value (s is more than or equal to 1 and less than or equal to p) of the s-th sampling value of the output variable is (T) _s -t _i ) Then input variable X _i The sampling interval with the output variable at the first sampling time is [ T ] _s -t _i -M，T _s -t _i +M]。

S32, obtaining a data fusion matrix according to the sampling intervals of the input variables at a plurality of first sampling moments and the measured values of the output variables at the plurality of first sampling moments.

Firstly, according to sampling intervals of the input variable at a plurality of first sampling moments, sampling values of the input variable at the plurality of first sampling moments are obtained.

Specifically, for a sampling interval of an input variable at any first sampling time, a plurality of sampling values of the input variable, of which second sampling time is located in the sampling interval, are acquired first. In this embodiment of the present application, the first sampling time is a sampling time of an output variable, and the second sampling time is a sampling time of an input variable. After obtaining the sampling interval of the input variable at the first sampling time, a plurality of sampling values of the input variable can be determined according to the sampling interval, and a second sampling time of the plurality of sampling values is located in the sampling interval.

And then, obtaining a data fusion matrix according to the sampling values of the input variables at the first sampling moments and the measured values of the output variables at the first sampling moments. The sampling value of the input variable at the first sampling time comprises a plurality of sampling values of the input variable, wherein the second sampling time of the input variable is located in a sampling interval.

Specifically, according to the plurality of sampling values, a fusion value of the input variable at the first sampling moment is obtained. For example, an average value may be obtained for the plurality of sampling values, a weighted average value may be obtained, and a fusion value of the input variables at the first sampling time may be obtained, thereby obtaining a data fusion matrix.

Still with input variable X _i For example, the alignment time of the sampling value (1.ltoreq.s.ltoreq.p) with the s-th sampling value of the output variable is (T) _s -t _i ) Then input variable X _i With the output variable at a first sampling instant T _s The lower sampling interval is [ T ] _s -t _i -M，T _s -t _i +M]Thus, the first sampling instant T _s Fusion value x of lower input variable _is Can be defined as:

x _is =f(T _s -t _i -M，T _s -t _i +M)（1）

f(T _s -t _i -M，T _s -t _i +M) represents the input variable X _i In the sampling interval [ T ] _s -t _i -M，T _s -t _i +M]An average of a plurality of sample values within.

According to the above rule of alignment of input variables with output variables, the input variables X defined in tables 1 and 2 above ₁ 、X ₂ 、...、X _k And output variable Y ₁ 、Y ₂ 、...、Y _n The aligned and fused process data matrix (i.e., data fusion matrix) can be represented as the following table 3:

TABLE 3 Table 3

In which the variable X is input _i Actual aligned sample values of (i=1, 2,.. _s Fusion value x of lower input variable _is ）x _is =f(T _s -t _i -M，T _s -t _i +M)，f(T _s -t _i -M，T _s -t _i +M) represents the input variable X _i In the sampling interval [ T ] _s -t _i -M，T _s -t _i +M]An average of a plurality of sample values within.

And performing model data preprocessing operation according to the process data matrix.

For example, the model data preprocessing operations such as filtering of the modeling sample data, abnormal sampling value processing and the like can be performed by selecting the fused modeling process data matrix with the same working condition for more than half a year.

Then, the set values of the input variables are determined based on the data fusion matrix.

This process is described below in connection with fig. 4. Fig. 4 is a schematic flow chart of determining a set value of an input variable according to an embodiment of the present application, as shown in fig. 4, including:

s41, determining an optimized control function between the input variable and the output variable according to the data fusion matrix.

First, an initial control function between an input variable and an output variable is determined based on a data fusion matrix.

And then, updating the initial control function according to the real-time value of the input variable and the real-time value of the output variable to obtain an optimized control function.

Specifically, at least one first operation is performed, and any ith first operation includes: obtaining a predicted value of an output variable according to a real-time value of the input variable and an ith round of control function; i is an integer greater than or equal to 1, and the 1 st round of control function is an initial control function;

when the difference value between the predicted value of the output variable and the measured value of the output variable is greater than or equal to a preset value, updating the ith round of control function according to the difference value to obtain an (i+1) th round of control function, and executing the (i+1) th first operation according to the (i+1) th round of control function;

and when the difference value between the predicted value of the output variable and the real-time value of the output variable is smaller than a preset value, determining the ith round of control function as a target control function.

Taking cement production flow process as an example, the objective control function can be determined by the following steps (1) to (7).

And (1) performing correlation analysis on model variables based on the process data matrix after data preprocessing, and then selecting input variables (operation variables) and output variables (controlled variables) of an optimization control model based on correlation analysis results of the model variables and a cement production process mechanism model.

And (2) creating a control function equation between the input variable and the output variable of the optimizing control model by adopting artificial intelligent algorithms such as multi-element linear (nonlinear) regression, random forests, neural networks and the like based on the final input variable (operation variable) and the output variable (controlled variable) of the optimizing control model of the cement production process.

And (3) creating a process real-time optimal control model according to the process control optimization targets of the process (the process control optimization targets of the general process are the constraint conditions of the optimal control model, wherein the process control optimization targets of the general process are the constraint conditions of the optimal control model, and the constraint conditions of the optimal control model are the constraint conditions of the optimal control model, and the energy conservation and consumption reduction of the production process and the minimum production cost of unit products are realized as far as possible while the process quality index meets the product specification requirement.

And (4) acquiring the DCS value or the measured value of the input variable of the current control process in real time based on the process optimization control model input-output function equation, and realizing the real-time prediction of the controlled variable (process quality index) of the current process through the input-output function equation.

And (5) solving an optimal control strategy (optimal set value of a process optimization control operation variable) of the process control in real time by using a conjugate gradient method and a Newton method, then sending the optimal control strategy to a process cache tool (Alternative PHP Cache, APC) pilot control system (adopting a PID control algorithm or an MPC model prediction algorithm) through a communication protocol, and taking the optimal control value of the process operation variable as an initial target value of the operation variable in the PID controller or the model prediction control system for real-time control.

And (6) calculating the deviation between the control model predicted value and the actual measured value of the controlled variable (output variable, generally the quality index of the process) of the optimization control model in real time, and when the control deviation exceeds a threshold allowed by the system, automatically performing self-adaptive adjustment on the process input/output control function equation by the system by adopting an exponential weighted moving Average (Exponentially Weighted Moving-Average) method or a secondary resampling modeling method.

And (7) repeating the steps (1) - (6), so that the real-time optimization control of the flow process can be realized, and the optimal value of the operation variable output by the real-time optimization control system is changed in real time and gradually converges to the global optimal control value of the current working condition optimization control because the optimization control input/output control function equation adopts a self-adaptive adjustment method.

S42, determining a set value of the input variable according to the target value of the output variable, the optimization control function and the preset process condition.

The final objective of the flow process optimization control system based on the data real-time alignment fusion is as follows: the production process is controlled to save energy and reduce consumption as much as possible while ensuring the quality of the technical product to meet the specification requirement, namely the cost of qualified products of production units is the lowest.

The flow process processing system architecture provided by the embodiment of the application comprises: the flow process treatment system comprises a process data acquisition unit, a data alignment fusion unit, a process optimization control unit and a process control system unit.

The process Data acquisition unit is used for acquiring process Data, and comprises a process DCS Data acquisition subunit and a process quality Data acquisition subunit, wherein the process DCS Data acquisition subunit can comprise a real-time library, a process control object connection and embedding (OLE for Process Control, OPC) Data Access (DA) protocol and the like, and the process quality Data acquisition subunit can comprise a relational database or a hypertext transfer protocol (Hyper Text Transfer Protocol, HTTP) and the like.

The data alignment fusion unit is used for acquiring the process DCS data in real time, aligning according to the first sampling time, classifying and fusing according to the process data, acquiring the process quality index data, supplementing the missing value, classifying and fusing according to the quality category attribute and the like.

The process optimization control unit is used for filtering the modeling sample data of the process optimization control model and preprocessing abnormal sampling values; the method is used for carrying out correlation analysis on process control variables and determining input and output variables of an optimal control model; the method is used for creating a process optimization control input/output variable control function equation; the method is used for creating a process optimization control model; the method is used for real-time optimization control of the process and real-time prediction of the controlled variable value of the process; the method is used for calculating the control deviation of the controlled variable of the process in real time, adaptively adjusting an input/output control model of the process and the like.

The process control system unit comprises a PID controller, wherein the PID controller is used for taking the optimal value of the operation variable as an initial state value; including OPC DA service software, etc.

Fig. 5 is a flowchart of a process provided in an embodiment of the present application, as shown in fig. 5, including:

s51, aligning and fusing process data in real time.

The process data includes process DCS real-time data (i.e., input variables) and process quality index data (i.e., output variables). After alignment and fusion, fusion values of the input variables at a plurality of first sampling moments can be obtained, and then a process data matrix after alignment and fusion is obtained.

S52, preprocessing a process data matrix.

The preprocessing is mainly to remove some abnormal data in the process data matrix.

S53, process optimization control variable correlation analysis.

S54, determining the input and output variables of the process optimization control model.

S55, inputting and outputting a variable control function equation by the process optimization control model.

S53 to S55 are mainly to determine causal mapping between input variables and output variables, and may be fitted by data in a process data matrix, with the fitted function being the control function equation between the input variables and the output variables.

S56, creating a process optimization control model.

The model is created mainly according to the target to be achieved and preset process conditions, wherein the target to be achieved can be, for example, energy conservation and consumption reduction, namely, power consumption reduction or cost reduction as much as possible under the condition of meeting the requirement of a finished product, and the preset process conditions are conditions which are needed to be achieved when the process is carried out to produce the product.

S57, solving the process optimization control model in real time.

S58, solving the optimal control strategy value of the process operation variable.

Namely, according to the determined input variable, the actual output variable is obtained by inputting the determined input variable into the system, the deviation between the actual output variable and the output variable calculated according to the control function equation is calculated, if the deviation is large, the steps S55-S58 are repeatedly executed, and if the deviation is smaller than the preset value, the control function can be used as the target control function, and then the set value of the input variable is determined.

The system solves the obtained optimal control strategy of the current working condition in real time, namely the optimal set value of the core operation variable (input variable) of the current working condition, and then sends the optimal set value to the process APC advanced control system (PID control system or MPC system) in real time through a communication protocol, and the optimal set value is used as the initial state value of the operation variable of the current prior control system to control the production process in real time. The system also calculates the deviation between the predicted value and the actual measured value of the control model of the controlled variable (output variable, generally the quality index of the process) of the optimization control model in real time, and when the control deviation exceeds the threshold allowed by the system, the system automatically adopts an EWMA index weighted moving average method or a secondary resampling modeling method to carry out self-adaptive adjustment on the input and output control function equation of the process, so that the optimal value of the core operation variable in the production process is always the global optimal value of the working condition.

In general, the final realization goal of the flow process optimization control system based on the data real-time alignment fusion is as follows: the production process is controlled to save energy and reduce consumption as much as possible while ensuring the quality of the technical product to meet the specification requirement, namely the cost of qualified products of production units is the lowest.

The PID controller is a process control system most commonly used in the field of current-stage industry, and the PID is a linear controller according to an actual control deviation e (t) between a given value r (t) and an actual output value c (t):

e(t)=r(t)-c(t)（2）

fig. 6 is a schematic diagram of an implementation of the PID controller algorithm provided in the embodiment of the present application, as shown in fig. 6, after calculating a deviation between an input and an output of a previous round, a control quantity is formed by linearly combining a proportion (P), a derivative (I) and a derivative (D) of an actual control deviation e (t), and a controlled variable is controlled by a process to obtain an output. The core formula of the PID control algorithm is as follows:

（3）

in the method, in the process of the invention,

for outputting variable values +.>

For the actual control deviation +.>

Is a proportionality coefficient->

Is the integral coefficient of the integral,

is a differential coefficient. Each correction link of PID controllerThe function of (2) is as follows:

the proportion links are as follows: real-time proportional response of actual deviation signal of control system

Once the deviation is generated, the controller immediately generates control to reduce the error. When deviation->

When=0, the control action is also equal to 0. Thus, the proportional control is adjusted based on the deviation, i.e., there is a differential adjustment.

And (3) integrating: can memorize control errors, is mainly used for eliminating static difference and improving the no-difference degree of a system, and the intensity of the integral action depends on an integral time constant

The larger the integration time constant, the weaker the integration effect, and vice versa.

The differential link can reflect the variation trend (variation rate) of signal deviation, and can introduce an effective early correction signal into the system before the deviation signal value becomes too large, thereby accelerating the action speed of the system and reducing the adjustment time.

From the time perspective, the proportional action is to control the current error of the system, the integral action is to the history of the system error, and the differential action reflects the change trend of the system error. Fig. 7 is a comparison chart between an original test signal and a signal output by a PID controller under different PID control coefficients provided in the embodiment of the present application, as shown in fig. 7, which illustrates an original reference signal, and a curve of an ideal output variable can be gradually optimized according to a target value of the input variable under the control of the PID controller under different integral coefficients (curves when the integral coefficients are 0.5, 1 and 2 are illustrated in fig. 7).

The MPC model predictive control algorithm is another common industrial process control technology in the industrial field, the model predictive control is a multivariable control method, and a control mechanism model can be described as follows: at each sampling moment, according to the obtained current measurement information, a finite time open-loop optimization problem is solved on line, and the first element of the obtained control sequence acts on the controlled object. At the next sampling moment, repeating the process, taking the new measured value as an initial condition for predicting the future dynamic state of the system at the moment, refreshing the optimization problem and solving again.

The MPC model predictive control algorithm comprises the following three main implementation steps: predictive model, feedback correction, and rolling optimization.

The prediction model is used for creating a control function equation between the operating variable and the controlled variable of the control system by adopting a multiple regression algorithm, a neural network algorithm or a random forest algorithm according to the historical working condition data information of the input variable (operating variable) and the output variable (controlled variable) of the control system, and after the function equation between the operating variable and the controlled variable of the control system is determined, an arbitrary input variable value (vector or matrix) is given, so that the output value of one controlled variable can be predicted.

In the feedback correction, in the prediction control, the prediction model is adopted to estimate the output value of the controlled variable, but the prediction based on the model cannot be accurately matched with the actual because of uncertain factors such as nonlinearity, model adaptation, interference and the like in the actual control process. Therefore, in the prediction control process, the prediction deviation between the predicted value and the actual sampling value of the controlled variable model needs to be calculated in real time, and the prediction model is corrected in real time by using methods such as the prediction deviation, an EWMA (exponentially weighted moving average) and the like.

The process of applying feedback correction to the predictive model is predictive control with strong immunity to disturbance and system uncertainty. The prediction control is based on a model and utilizes the real-time control deviation feedback information of the controlled variable, so that the model prediction control is a closed-loop optimization control algorithm.

Rolling optimization, predictive control is an optimization control algorithm that requires the determination of future control actions by optimization of a performance metric that is also related to future behavior of the process, which is determined by future control decisions based on a predictive model. However, unlike the usual discrete optimization control algorithm, the optimization in the predictive control does not adopt a constant global optimum, but adopts a rolling finite time domain optimization strategy, i.e. the optimization process is not completed offline at one time but repeatedly performed online. At each sampling instant, the optimization performance index is only designed for a limited time in the future from that instant, and by the next sampling instant, this optimization period will be advanced simultaneously. Thus, predictive control does not use an optimized performance index that is globally the same, but rather has a locally optimized performance index at each instant relative to that instant.

There are several tens of algorithms for model predictive control, of which there are three kinds of model algorithmic control (Model Algorithm Control, MAC), dynamic matrix control (Dynamic Matrix Control, DMC) and generalized predictive control (Generalized Predictive Control, GPC) as typical main.

The structure of model algorithm control comprises 4 calculation links, namely an internal model, feedback correction, rolling optimization and a reference track, and the basic idea of the algorithm is as follows: firstly, predicting future output values of controlled variables, then determining control actions at the current moment according to the future output values, and firstly predicting and then controlling. Because of its predictability, it is evident from conventional PID control systems that output-first feedback-last control. FIG. 8 is a schematic diagram of model algorithm control provided in an embodiment of the present application, and FIG. 9 is a basic schematic diagram of dynamic matrix control algorithm provided in an embodiment of the present application, as shown in FIG. 8, y after rolling optimization _r (k) And obtaining e (k) from the feedback corrected result, then performing rolling optimization on the e (k) to obtain u (k), wherein u (k) can obtain y through a prediction model _r (k|k), jointly performing feedback correction by combining the initial rolling optimization result, and outputting y (k); as shown in FIG. 9, the input is r (k), and y is combined with the feedback correction and prediction model output _r And (k+j|k) obtaining e (k), performing rolling optimization on the e (k) to obtain u (k), and respectively inputting the u (k) into a prediction model and acting on a controlled object, and finally outputting y (k). In the practical application process, the specific algorithms controlled by the model algorithm are numerous, and the model algorithm is controlled by a single-step model algorithmThe multi-step model algorithm control, the single-value model algorithm control, and the incremental model algorithm control are not described in detail herein.

The DMC is different from the MPC in that on the internal model, the algorithm is modified by using an engineering easily measured object step response as the model. The generalized predictive control is based on the previous predictive algorithms, introduces the idea of adaptive control, and the general predictive control algorithm mainly compensates the system error through feedback, and the model can timely compensate the influence caused by time variation, interference and the like by adding a rolling optimization technology.

In order to realize real-time optimal control of a process, an optimal control system based on real-time alignment fusion of process data is realized on the basis of a traditional PID controller and MPC model predictive control: for the modeling sample data matrix after process alignment and fusion, a control function equation of process input and output variables is created by using neural network algorithms such as multiple linear regression, nonlinear regression, a neural network and a random forest, then an optimization target of process control (the process output variables meet the quality specification requirement of the process and the energy consumption in the production process is the lowest) and process optimization control constraint conditions (process optimization control operation variable constraint conditions and process optimization control controlled variable constraint conditions) are combined, an optimization control model of the process is established, the optimal value of the current process optimization control model is calculated in real time, and the optimal value of the current operation variable is sent to a process APC (automatic control system) through a communication protocol or a prediction algorithm of the PID (proportion integration differentiation) control algorithm or the MPC model) to realize real-time optimization control of the process operation variable. In the practical application process, the method for creating the process optimization control model input/output variable control function equation can also adopt other methods such as decision tree regression algorithm, support vector machine regression algorithm and the like. On the other hand, the solving algorithm conjugate gradient method and the quasi-Newton method of the multi-objective optimization control model can also adopt a random gradient descent method, a Newton method and the like in the actual application process.

According to the scheme of the embodiment of the application, the real-time alignment and fusion of DCS variable data and quality index data of a process are realized, and a modeling sample data matrix of a process optimization control model is generated; the second aspect applies the modeling sample data matrix after fusion to analyze the relativity among the process control variables, and then confirms the final input variable (operation variable) and output variable (controlled variable) of the process optimization control model by combining the process mechanism model; in the third aspect, a control function equation between input and output variables of an optimization control model is established by applying artificial intelligent algorithms such as polynary linearity, polynary nonlinearity, a neural network and a random forest; the fourth aspect applies a conjugate gradient method, and a quasi-Newton method solves an optimal control strategy of an optimal control model in real time, namely an optimal control value of an operation variable of a process optimal control model; the optimal control value of the operation variable is sent to an APC (automatic control system) of the process by using an OPC DA (automatic control system) communication protocol, and the optimal control value of the operation variable of the process is used as an initial set value of the operation variable in a PID (proportion integration differentiation) controller or a model predictive control system in the APC system for real-time control; in the sixth aspect, the control deviation between the predicted value and the real-time measured value of the controlled variable of the process optimization control model is calculated in real time, and when the control deviation exceeds the threshold allowed by the system, the system automatically adopts an EWMA index weighted moving average method or a secondary resampling modeling method to carry out self-adaptive adjustment on the process input and output control function equation.

In summary, the process processing scheme provided by the embodiment of the application is a process optimization control method and system based on data real-time alignment fusion, and compared with the existing process PID control system or MPC model prediction control system, on one hand, the scheme realizes alignment and fusion of process DCS data and quality index data according to a process mechanism model and a process production flow rule based on the process DCS real-time data and the process product quality index data, thereby forming a causal relation mapping between process operation variables (input variables, DCS control points) and controlled variables (output variables, process quality indexes), and providing a data basis for accurately creating an input/output control function equation of a process optimization control model. On the other hand, the scheme introduces a multi-element linear regression, nonlinear regression, a neural network, a random forest and other artificial intelligent algorithms, creates a real-time optimal control model among process core control variables based on a modeling sample matrix after process alignment fusion and a relatively mature artificial intelligent algorithm, and adopts a conjugate gradient method to solve the optimal control strategy of the current working condition in real time by adopting a quasi-Newton method, namely, under the precondition that the quality index of a process product meets the process specification, the energy conservation and consumption reduction of the production process are realized as much as possible, and finally the production cost of a unit qualified product is lowest.

The flow process treatment device provided by the application is described below, and the flow process treatment device described below and the flow process treatment method described above can be referred to correspondingly.

Fig. 10 is a schematic structural diagram of a flow process treatment apparatus according to an embodiment of the present application, as shown in fig. 10, including:

an obtaining module 101, configured to obtain alignment offset corresponding to an input variable of production process control, and measured values of an output variable at a plurality of first sampling moments;

the processing module 102 is configured to perform alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured values of the output variable at the plurality of first sampling moments, so as to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, where the data fusion matrix is used to indicate a functional correspondence between the input variable and the output variable;

a determining module 103, configured to determine a set value of the input variable according to a preset process condition, a target value of the output variable at the plurality of first sampling moments, and the data fusion matrix.

In one possible implementation, the processing module 102 is specifically configured to:

In one possible implementation, the determining module 103 is specifically configured to:

Fig. 11 illustrates a physical structure diagram of an electronic device, as shown in fig. 11, which may include: processor 1110, communication interface Communications Interface 1120, memory 1130 and communication bus 1140, wherein processor 1110, communication interface 1120 and memory 1130 communicate with each other via communication bus 1140. Processor 1110 may call logic instructions in memory 1130 to perform a flow process method comprising: obtaining an alignment offset corresponding to an input variable of production process control and measured values of output variables at a plurality of first sampling moments; performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured value of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variable and the output variable; and determining the set value of the input variable according to a preset process condition, the target values of the output variables at the first sampling moments and the data fusion matrix.

Further, the logic instructions in the memory 1130 described above may be implemented in the form of software functional units and sold or used as a stand-alone product, stored on a computer-readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

In another aspect, the present application further provides a computer program product, where the computer program product includes a computer program, where the computer program can be stored on a non-transitory computer readable storage medium, where the computer program when executed by a processor can perform the process treatment method provided by the above methods, and the method includes: obtaining an alignment offset corresponding to an input variable of production process control and measured values of output variables at a plurality of first sampling moments; performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured value of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variable and the output variable; and determining the set value of the input variable according to a preset process condition, the target values of the output variables at the first sampling moments and the data fusion matrix.

In yet another aspect, the present application further provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the flow process method provided by the above methods, the method comprising: obtaining an alignment offset corresponding to an input variable of production process control and measured values of output variables at a plurality of first sampling moments; performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured value of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, wherein the data fusion matrix is used for indicating a functional corresponding relation between the input variable and the output variable; and determining the set value of the input variable according to a preset process condition, the target values of the output variables at the first sampling moments and the data fusion matrix.

The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. A flow process treatment method, comprising:

determining a set value of the input variable according to a preset process condition, target values of the output variable at the plurality of first sampling moments and the data fusion matrix;

the performing alignment fusion processing on the input variable and the output variable according to the alignment offset and the measured values of the output variable at the plurality of first sampling moments to obtain a data fusion matrix of the input variable and the output variable at the plurality of first sampling moments, including:

Obtaining the data fusion matrix according to the sampling intervals of the input variables at the first sampling moments and the measured values of the output variables at the first sampling moments;

the determining, according to the alignment offset and the plurality of first sampling moments, a sampling interval of the input variable at the plurality of first sampling moments includes:

determining sampling intervals of the input variables at the plurality of first sampling moments according to the alignment moments and the preset time length;

the obtaining the data fusion matrix according to the sampling intervals of the input variables at the first sampling moments and the measured values of the output variables at the first sampling moments includes:

2. The method of claim 1, wherein the determining the set value of the input variable based on the preset process condition, the target value of the output variable at the plurality of first sampling moments, and the data fusion matrix comprises:

3. The method of claim 2, wherein said determining an optimal control function between said input variables and said output variables based on said data fusion matrix comprises:

4. A method according to claim 3, wherein said updating the initial control function in real time based on the real-time value of the input variable and the real-time value of the output variable to obtain the optimal control function comprises:

5. A flow process treatment apparatus, comprising:

the determining module is used for determining a set value of the input variable according to a preset process condition, target values of the output variable at the plurality of first sampling moments and the data fusion matrix;

the processing module is specifically configured to:

the step of determining, by the processing module, a sampling interval of the input variable at the plurality of first sampling moments according to the alignment offset and the plurality of first sampling moments includes:

the step of obtaining the data fusion matrix by the processing module according to the sampling intervals of the input variables at the plurality of first sampling moments and the measured values of the output variables at the plurality of first sampling moments includes:

6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the flow process method of any of claims 1-4 when the program is executed.

7. A non-transitory computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the flow process method according to any one of claims 1-4.