RU1781670C - System of adaptive control over production process - Google Patents

Abstract

The invention relates to process control systems and can be used for adaptive control of technological processes in the presence of unmeasured uncontrolled inputs of an object that significantly affect the control criterion. The aim of the invention is to improve the quality of control. This goal is achieved by the fact that in the system containing the control object, sequentially connected a block for generating corrective actions, a block for generating control actions and a block for actuators, as well as blocks for measuring quality indicators and signals of controlled and uncontrolled inputs, a block for classifying observations is introduced, the inputs of which are connected to blocks for measuring quality indicators, signals of controlled and uncontrolled inputs and with a class master, and a block for determining optimal controls constituent influences the inputs of which are connected to the block classification observations measuring unit and measuring the quality metrics signals actuated inputs and mode setting device, the output unit for determining the optimum steering effects connected with the inputs forming unit and corrective action forming unit actuating vozdeystviy.3 yl. (L C

Description

The invention relates to process control systems and can be used for adaptive control of technological processes in the presence of unmeasured uncontrolled inputs of an object that significantly affect the quality of control. In this case, the control of individual process variables is carried out by local control devices.

An example of such objects can be, for example, metallurgical units: rolling mills, steelmaking and blast furnaces. Such objects often have unmeasured uncontrolled inputs i.e. inputs whose signals are not available for

directly controlled and not directly measurable. For example, for an object - a continuous broadband hot strip mill - an uncontrolled unmeasured input, among others, is the slab temperature and its volume distribution. This temperature is determined by the design and mode of operation of the heating devices, physicochemical properties and the dimensions of the slab and can only be estimated by crude indirect methods, since they should be covered with a thick layer of scale and heated unevenly in volume. At the same time, the slab temperature has a significant effect on the rolling process and its results - achievement of vvj

00

-

ON VI

ABOUT

given sizes of rolled metal and its physical properties. The number of classes of an unmeasured uncontrolled input can be determined, for example, by the number of heating devices with different configurations or modes, the number of operator workers of these devices with different personal manner of controlling the heating process, and the like.

Known adaptive control system containing sequentially connected block forming

corrective actions, the block forming the control actions and the control object. The disadvantage of this system is the low quality of control, because When determining the optimal control actions, neither the values of the signals of the uncontrolled inputs of the control object nor the influence of the unmeasured inputs of the control object were taken into account.

A device for adaptive process control, which we have adopted as a prototype, comprising series-connected corrective actions generating unit, control actions generating unit, actuators and control objects, as well as controlled inputs signals measuring unit, uncontrolled signals measuring unit inputs and a unit for measuring quality indicators, moreover, the inputs of the unit for measuring signals of controlled inputs and the unit for measuring signals of unmanaged inputs in are connected to the corresponding inputs of the control object, and the inputs of the unit for measuring quality indicators are connected to the outputs of the control object.

The disadvantage of this device is the low quality of control, since the choice of control actions is carried out without taking into account the influence of unmeasured inputs of the control object, which significantly affect the optimal values of the controlled variables.

The aim of the invention is to improve the quality of process control.

The goal is achieved due to the fact that in the adaptive process control system comprising sequentially connected corrective actions generating unit, control actions generating unit and actuator unit, the outputs of which are connected to the inputs of the measurement unit of the signals of the controlled inputs and to the controlled inputs of the control object , unmanaged inputs of the control object are connected to the inputs of the measuring unit of signals of uncontrolled inputs, and the outputs of the control object are connected to the inputs block for measuring quality indicators, a block for classifying observations was introduced, the first

a group of information inputs of which is connected to the outputs of the measurement unit of the signals of the controlled inputs, a second group of information inputs of the measurement block of the observations of the inputs is connected to the outputs of the measurement unit of the signals of uncontrolled inputs, a third group of information inputs of the block of measurement of the observations of the inputs is connected to the outputs of the block of measurement of quality indicators, and a determination unit 5 optimal control actions, the information input of which is connected to the output of the observation classification block, the first group of information s inputs - outputs of the measurement unit with

0 quality indicators, the second group of information inputs - with the outputs of the unit for measuring the signals of the controlled inputs, the output of the unit for determining the optimal control actions is connected to the inputs of the block for generating corrective actions and to the second inputs of the block for forming control actions, as well as the class selector and mode dial the outputs of which are connected respectively

0 with the inputs of the task of the classification block for observations and the block for determining the optimal control actions, while the control object has unmeasured inputs of p classes,

5 At the same time, the classification block of observations contains nodes consisting of the first adder and chains of the second adder and the first multiplier, the two inputs of which are connected to

0 by the course of the second adder, the outputs of the first multiplier elements are the outputs of the chains and are connected to the inputs of the first adder, the number of chains in each node is equal to the sum of the number of controlled inputs, uncontrolled inputs and outputs of the block for measuring quality indicators, the number of nodes is equal to the number p of classes of non-dimensions inputs of the control object, the first inputs of the corresponding second adders of each node

0 are combined and form, respectively, the first, second and third group of information inputs of the observation classification block, the second inputs of the second adders of all nodes are inputs

5 of the task of the classification block of observations, the output of each first adder is connected to the first inputs of the first comparison elements, the number of which for each adder is one less than the number of nodes, and to the second inputs of the other first comparison elements, the number of which is also one less than the number of nodes, the first the inputs of each of these other first elements of comparison are connected respectively to the outputs of all the other first adders, the total number of first elements of comparison is equal to twice the number of combinations of the number of nodes in two, outputs ervyh comparing elements, first inputs of which are connected to the same first adder connected to inputs of AND gates, the number of AND gates equal to the number of nodes, elements, and outputs connected to inputs of an encoder, whose output is the output of the classification of observations.

In addition, the block for determining the optimal control actions contains second multiplier elements, the number of which is equal to the number of outputs of the block for measuring quality indicators, the first inputs of the second multiplier elements form the first group of information inputs of the block for determining the optimal control actions, the second inputs of the second multiplier elements are job inputs unit for determining the optimal control actions, the outputs of the second multiplying elements are connected to the input of the third adder, and also the omitting device, the address input of which is the information input of the optimal control action determining unit, the output of the third adder is connected to the first input of the second comparison element and to the first information input of the storage device, the first information output of the storage device is connected to the second input of the second comparison element. the output of the second comparison element is connected to the input of the element NOT and to the control inputs of the first group of elements BAN, the information inputs of which are cosiness is the second group of information inputs of the block for determining the optimal control actions, the output of the element is NOT connected to the control input of the storage device and to the control inputs of the second group of BAN elements, the information inputs of which are connected to the second information outputs of the storage device, the outputs of the first group of BAN elements are connected to the second information inputs of the storage device and with the first input of the group of OR elements, the second inputs of the group of OR elements are connected to the outputs the second group of elements is FORBID, the outputs of the group of elements OR are outputs of the block

to determine the optimal control actions, the number of the BAN elements of the first group is equal to the number of BAN elements of the second group, and is equal to the number of OR elements and equal to the number of controlled inputs of the control object.

Introduction to the system of the classification block for observations and the block for determining the optimal control actions, and

Also, the class setter and mode setter can improve the quality of control by separately determining the optimal control actions from the classes of observed quantities, moreover

the class combines observations corresponding to close values of the set of parameters of unmeasured inputs of the control object; the proximity of parameters is estimated by the classification block of observations.

The proposed control system is invariant to the system of equations describing the control object, since the functioning of the control system

is described not by the system of equations describing the control object, but by the division of observations into classes and the control criterion selected during the design of the system. The system provides control

object in accordance with the selected control criterion (objective function of the quality indicators of the control object) so that the control actions on the next time interval, the choice of which

carried out in the block 6 for determining the optimal control actions, they always provide the minimum value of the objective function among a finite number of admissible control actions in the current time interval. The direction of change of control actions is an improvement in the control criteria. The control system using the proposed technical solution has its distinguishing features: control in it is carried out taking into account not only uncontrollable influences acting on the object, but also unobservable factors, while not losing the properties of dynamism, adaptability and stability.

The stability of the system, meaning by this term the limited increment of control actions with a limited change in the input data (see / 10 /, p. 12), is ensured by the fact that when the input variables are limited to one class, the control actions are guaranteed to be selected from the same class under action

the same control criteria and, therefore, are limited by the size of the class.

The dynamism of control follows from the consideration of temporary changes in the vectors of uncontrollable variables, controlled variables, quality indicators, and membership of the set of n-eremennial classes.

Adaptability of control is provided by changing the class membership vector at a pace with changes in all observed and unobserved input quantities. In FIG. 1 shows a block diagram of a system; in FIG. 2 is a block diagram of a case classification unit; in FIG. 3 is a block diagram of a block for determining optimal control actions.

The system (Fig. 1) contains a control object 1cm controlled inputs, n uncontrolled inputs, unmeasured inputs of p classes, 2 measurements of signals of uncontrolled inputs, block 3 measurements of signals of controlled inputs, block 4 of measurement of quality indicators with k outputs , block 5 for classifying observations, block 6 for determining optimal control actions, block 7 for generating corrective actions, block 8 for forming control actions, block 9 for actuators, master of classes 10 and master of modes 11.

Uncontrolled n inputs of control object 1 are connected to n inputs of the block. 2 measurements of signals of unmanaged inputs, outputs of control object 1 are connected to the inputs of block 4 for measuring quality indicators, m outputs of block 3 for measuring signals of controlled inputs are connected to the first group of information inputs of block 5 of the classification of observations, the second group of information inputs of block 5 of the observation classification is connected to the outputs of the block 2 of measuring signals of uncontrolled inputs, and the third group of information inputs of the block 5 of the classification of observations is connected to the k outputs of the block 4 of measuring indicators of quality va, wherein p «(m + n + k) input specifying the classification unit 5 connected to the outputs of observations classes setpoint 10, and an output unit 5 connected to the labeling of observation data input unit 6 for determining the optimum steering effects. The mode dial 11 is connected by its k outputs to the corresponding k inputs of the reference unit 6 for determining the optimal control actions, the first group of information inputs of which are connected to the k outputs of the measurement unit 4

quality indicators, and a third group of information inputs (t inputs) with t outputs of the unit 3 for measuring signals of the controlled inputs and outputs of the unit 6 for determining the optimal control actions are connected to m inputs of the unit 7 for generating corrective actions and to m second inputs of the unit-8 for forming control impacts, while

0 m first inputs of the control action generating unit 8 are connected to m outputs of the corrective action generating unit 7, am outputs of the control action generating unit 8 are connected to the corresponding m inputs of the actuator unit 9, m outputs of the actuator unit 9 are connected to m controlled inputs of control object 1 and to m inputs

0 block 3 measurement of signals of controlled inputs,

The control object is, for example, a continuous broadband hot-rolling mill for steel strips. The number of classes p depends on the unobservable factors acting on the control object and is equal to the number of different combinations of all possible qualitative values of these factors. The number of classes is determined based on the characteristics of a particular control object at the stage of preliminary development of the system.

For the control object — a continuous broadband hot rolling mill — the following simplified set of input and output variables can be given:

controlled inputs (two): vertical and horizontal compression of the workpiece (or

0 deviations of these reductions from any base);

uncontrolled input (one): the thickness of the workpiece (or its deviation from the set, from any base, etc.);

5 outputs (two): thickness and width of the rolled product (or their deviation from the rolling task);

quality indicators (two): standard deviation of the strip thickness from

0 reference (from the nominal value) along the entire length of the strip, standard deviation of the strip width from the reference along the entire length of the strip;

unmeasured inputs: various modes of heating the workpiece (slab), characterized by the design features of the heating devices (for example, different furnaces on the mill, from which they are handed out for rolling arbitrarily), different teams of operators with individual features of the heating process and etc. - only three types of modes, i.e. - Three classes of unmeasured inputs.

For such a simplified set of variables characterizing the control system, a target function of the form can be used:

C K1 Z1 + K2 Z2, where C is the value of the objective function;

K1, K2 — coefficients of the variables of the objective function;

Z1 is a first quality indicator, for example, the standard deviation of the strip thickness from the reference over the entire length of the strip;

Z2 is a second quality indicator, for example, the standard deviation of the strip width from the reference over the entire length of the strip.

The aim of the control is to produce rolled products with a minimum standard deviation in thickness and width from the reference, taking into account the influence of unmeasured inputs of the control object, i.e. it is necessary to issue control actions on the actuators with adaptation to the current class of unmeasured inputs (depending on how the workpiece was heated before rolling),

Block 5 classification of observations (Fig. 2) contains p nodes (for example, -3) consisting of m + n + k (for example, 2 + 1 + 2) chains (p is the number of classes of unmeasured inputs, m is the number of controlled inputs, n is the number of uncontrolled inputs, k is the number of outputs of the unit for measuring quality indicators) and the first adder

14. Each of the chains contains a second adder 12 and a first multiplier element 13. The block of classification of observations 5 also includes the first elements of comparison 15, the number of which for each of p nodes is p-1, elements And 16, the number of which is equal to the number of nodes p and the encoder 17. Thus, for m 2, n 1, k 2 and p 3, the block of classification of observations 5 contains (see Fig. 3): second adders 12 - p (m + n + k) 3 (2 + 1 + 2) 15, the first multiple elements 13 - p (m + n + k) 3 (2 + 1 + 2)

15, the first adders 14 to p 3, the first comparison elements 15 to p (p-1) 6, the AND elements 16 to p 3, the encoder 17-1.

In each of the chains containing the second adder 12 and the first multiplier element 13, both inputs of the first multiplier element 13 are connected to the output of the second adder 12 ,. second adders 12 are provided with direct first and inverse second inputs.

The first inputs of the second summation 1§ are combined over all p nodes so that with the bbt corresponding m, n, k of the first inputs of the second adders 12 form, respectively, 5 the first, second and third groups of information inputs of the observation classification block 5.

The second inputs of the second adders 12 of all nodes are the input inputs of the block

0 5 classification of observations, the signals at these inputs correspond to the reference values, respectively, m signals of controlled inputs, n signals of uncontrolled inputs and k quality indicators for all p

5 classes. The outputs of the first multiplier elements 13 of each node are outputs of the chains and are connected to the inputs of the first adders 14 of this node.

The outputs of the first adders 14 are 0 with the outputs of the nodes and are connected to the first inputs of the first elements of comparison 15, the number of which for each of the first adders is one less than the number of nodes (for example, 2, Fig. 2), and with the second inputs

5 other first elements of comparison, the number of which is also one less than the number of nodes (for example, 2), moreover, the first inputs of each of these other first elements of comparison are connected, respectively, to

0 outputs of all other first adders, the total number of first comparison elements is equal to twice the number of combinations from the number of nodes by 2 (for example, for three nodes the number of combinations from 3 to 2 is 3, the number of first

5 elements of comparison is 6). The outputs of the first elements of comparison, the first inputs of which are connected to the same first adder, are connected to the inputs of the elements 16 AND, the number of elements AND is equal to the number

0 nodes (e.g. 3), the outputs of the AND elements are connected to the inputs of the encoder 17, the output of which is the code of the class to which this observation belongs. Observation is a set of values of the variables of the control system - the integral values of the measured values of the signals of the controlled inputs of the control object, uncontrolled inputs of the control object and indicators

0 quality control system corresponding to the same time interval (for the control object of a broadband hot rolling mill - the interval is the rental time of one

5 stripes). The class code to which this observation refers is the output of the case classification block.

Block 6 determining the optimal control actions (Fig. 3) contains a storage device 18, a group of k

2 second multiplying elements 19, the number of which is equal to the number of outputs of the quality indicator measuring unit 4, the third adder 20, the second comparison element 21, the element NOT 22, the elements BAN 23 of the first group, elements BAN 24 of the second group, and the group of elements 25 OR, the number of elements 23 PROHIBITION of the first group is equal to the number of elements 24 PROHIBITION of the second group, as well as equal to the number m of controlled inputs of the control object.

The first inputs of the second multiplying elements 19 form the first group of information inputs of the optimal control actions determination unit 6, the second inputs of the second multiplying elements 19 are the input inputs of the optimal control actions determining unit 6, the outputs of the second multiplying elements 19 are connected to the input of the third adder 20,

The output of the third adder 20 is connected to the first input of the second comparison element 21 and to the first information input of the storage device 18.

The address input of the storage device 18 is the information input of the optimal control actions determining unit 6, the first information input of the storage device 18 is connected to the second input of the second comparison element 21, the output of the second comparison element 21 is connected to the input of the element NOT and to the control inputs of the elements 23 the first group.

The information inputs of the elements 23 FORBIDDING of the first group form the second group of information inputs of the block 6 for determining the optimal control actions. The output of the element HE 22 is connected to the control input of the storage device 18 and to the control inputs of the elements BAN 24 of the first group. The information inputs of the BAN elements of the second group are connected to the second information outputs of the storage device and to the first inputs of the group of 25 OR elements, The second inputs of the group of elements 25 OR are connected to the outputs of the elements 24 BAN of the second group, the output 1 of the group of 25 OR elements are outputs of block 6 determining optimal control actions.

The system operates as follows.

The control object 1 (continuous broadband hot rolling mill) carries out a technological process, for example, metal rolling. Heated billets are fed to the mill (slu). After crimping the slabs between the rolls of the mill stands

the slab extends into a strip. The uncontrolled inputs signal measuring unit 2 measures the uncontrolled variables acting on the control object, The controlled inputs signal measuring unit 2 measures the controlled variables affecting the control object (output variables of the actuator unit 9). Quality Measurement Unit 4

0 measures the output variables of control object 1 and converts them into quality indicators of the control system.

The measurement of signals by blocks 2, 3, 4 is carried out, as is customary in measuring technology, by converting the measured value (both an electric signal and non-electric) into a standard electric signal and averaging it, which is integral (over the entire band) characterizes the rolling process, Block 4 measuring quality indicators additionally converts the output variables of the control object 1 into the quality indicators of the process control system.

5 The output signals of the unit 2 for measuring the signals of unmanaged inputs, the unit 3 for measuring signals of the controlled inputs and the unit 4 for measuring quality indicators are received respectively in the second, first

0 and the third group of information inputs of the observation classification block 5, In addition, the output signals of the control inputs signal measurement unit 3 and the quality indicators measurement unit 4 are received,

5, respectively, on the second and first groups of information inputs of the block 6 for determining the optimal control actions. The inputs of the reference unit 5 of the classification of observations and the unit 6 for determining the optimal control actions receive signals, respectively, from the master of classes 10 and the master of modes 11. Signals of the task of the master of classes 10 are 5 class reference values of all measured inputs and quality indicators of the object management. The reference signals of the mode master 11 are the target coefficients. functions for

0 indicators of the quality of the control object, Block classification of observations 5 according to the state of all measured inputs and quality indicators of the control object of the system (signals from blocks 2, 3 and 4) for the current

5 observations and the class-wise reference values of all measured inputs and quality indicators (signals from class adjuster 10) determines the class number to which this observation belongs. Class number is selected by calculation

the distance in the space of measured variables of the system from the point of current observation to points with the reference values of variables and quality indicators for all classes, and the choice of the class number to which the current observation belongs, according to the minimum of these distances,

Block 6 determining the optimal control actions by the class number of the current observation (signals from block 5 classification of observations), the values of controlled variables and quality indicators (signal from block 3 measuring signals of controlled inputs and block 4 measuring quality indicators) and the coefficients of the objective function at quality indicators (signals from the mode dial 11), it generates the optimal control action for the control object for the next time interval, the length of which is equal to the rental time of one field sy. The objective function is a function of quality indicators, which is a numerical expression for the quality of control.The determination of the optimal control effect is carried out according to the current value of the optimality criterion (for current observation) and the best-remembered best value of the optimality criterion for a given class - or by choosing a known ( previously stored) controls, or maintaining current controls. The control generated in the block 6 for determining the optimal control actions is transmitted to the inputs of the block 7 for generating corrective actions and for the block 8 for forming the control actions.

The corrective actions generating unit 7 from the values of the optimal control actions coming from the output of the optimal control actions determining unit 6 generates the values of the corrective actions that are output from its output to the input of the control actions generating unit 8. These values are coefficients by which it is necessary to multiply the values of the optimal control actions in order to obtain the real allowable values of the control actions. The indicated coefficients have the meaning of limitations, for example, according to technological modes, according to the capabilities of actuators, for reasons of safety, etc. The control action generating unit 8 for each controlled variable multiplies the optimal control

actions (output signals of the block 6 for generating optimal control actions) on the coefficients (output signals of the block 7 for forming corrective actions). The control actions from the output of the control action generating unit 8 go to the input of the actuator unit 9 and, through its actuators, affect the technological process carried out by the control object 1. Thus, after rolling the next strip at the outputs of blocks 2, 3, and 4 new system variables and corresponding quality indicators are being formed. For these values of variables and quality indicators, block 5 for classifying observations according to the reference values of variables and quality indicators for each class (coming from the outputs of class adjuster 10) determines the class number to which this strip belongs. This number is transmitted to the information input of the optimal control unit 6

5 impacts. Block 6 by the class number, signals of controlled inputs, quality indicators and coefficients of the objective function for the next strip (coming from the outputs of the task of modes 11) determines the settings for the optimal control actions for rolling the next strip. The final formation of the settings is carried out by the unit 8 of the formation of control actions by multiplication

5 corresponding output signals of blocks 6 and 7. The settings thus formed are fed to the inputs of block 9 of the actuators and the operating modes of the actuators are determined

0 of the control object 1 during the rolling of the next lane. After rolling the next lane, the process is repeated, and for each lane control is selected that maintains or improves the value of the objective function. This provides adaptive process control.

Block 5 classification of observations 0 (Fig. 2) works as follows.

The second adders 12, the first multiplier elements 13, and the first adders 14 calculate the distances in the space of the variable systems from the point 5 of the current observation to the points of the reference values by classes, for example, by the formula

d - S (J - f j) 2 -I (UCH - f m + q) 2 + Ј (z, 2

T m + n + lj ,,

where di is the distance between the current observation point and the class I point with the reference values of the measured variables;

xi, j 1 ... m are the values of the signals of the controlled inputs for the current observation;

f i, J 1 ... m are the reference values of the controlled inputs for the 1st class;

yq, q 1 ... n are the values of the signals of the uncontrolled inputs for the current observation;

f m + q, q 1 ... l - reference values of the signals of uncontrolled inputs for the 1st class;

2Ј. 1 1 ... k are the values of the output signals of the quality indicator measuring unit for the current observation;

f m + n + i, 1 - 1 ... k - reference values of quality indicators for the 1st class,

The values of the input variables and quality indicators of the control object 1 for the current observation are fed to the direct inputs of the first adders 12 for all p nodes of the observation classification block. The inverse inputs of adders 12 from the output of the class 10 master receive reference values of all m + n + k measured system variables for all p classes. Reference values are determined at the stage of system design by processing experimental data using known algorithms (8). The output of each of adders 12 is the difference between the measured and the reference values of each variable. This signal is applied to both inputs of the corresponding (the first multiplying element 13, which squares its input signal. The output signals of the first multiplying elements 13 of each node are supplied to the inputs of the first adder 14 of the same node. Thus, the output signal of the first adder 14 of the node related to the i-th class, it makes sense the sum of the squares of the distances in the space of system variables from the current observation point to the first-class point with reference values of the measured variables of the first class,

The output signals of the first adders 14 of all p nodes in combinations of 2 are fed to the inputs of the first elements of comparison 15, which perform pairwise comparisons of the outputs of the first adders 14. The outputs of the first elements of comparison 15, the first inputs of which are connected to the same first adder 14 , served on the inputs of the elements And 16. Each of the first elements of comparison 15 is triggered (generates a signal 1 at the output), if the signal at its second input is greater

than the signal at its first input, i.e., if the square of the distance in the space of the measured variables of the system from the current observation point to the point with the reference

the values of the class (and its corresponding node) corresponding to the first input of the first comparison element 15 are less than the same value for the other class (and its corresponding node) corresponding to the second input of this first comparison element 15. Thus, if the first input any of the first element of comparison 15 receives a signal from the output of the first adder of the 1st node, and to the second

5, the input of the same adder receives a signal from the output of the first adder of the j-ro node, then this first comparison element produces an output signal 1 in the case when the observation is closer to the reference set

0 values of i-ro class variables than to a set of j-ro class, i.e. if the signal at its first input is less than at the second.

The signals from the outputs of the first elements of comparison 15, the first inputs of which are connected to the output of the same first adder 14 of the 1st node, go to the inputs of one of the elements And 16. If all the input signals of the element And 16 are 1, then its output signal 1 also appears. This

0 means that the current observation point in the space of the measured variables of the system is located closer to the point with the reference values of class I than to all other points with the reference values of 5 other classes. In this case, the signal O will be generated at the outputs of other elements And 16. The output signals of all elements And 16 are fed to the inputs of the encoder 17, at the output of which a class code is generated

0 i with the smallest distance in the space of measured system variables from the current observation point to the point with the reference values of class i variables. In case the distance between the point

5 of the current observation and reference points for two classes are equal and less distances between the current observation point and reference points for all other classes, at the outputs of all elements And 16 will be

0 signal O. Encryptor 17 is made on the basis of a table entered in a programmable logic matrix or in a programmable ROM. The case of equal minimum distances applies when designing a system to one of the classes. The specific table view of the encoder 17 is determined during the system design phase.

The unit 6 for determining the optimal control actions (see Fig. 3) of the operating inputs of the unit for determining the optimal control actions, from the outputs of the unit for measuring quality indicators 4 (see Fig. 1), the measured system quality indicators are received.

The second inputs of the second multiplying elements 19, which are the inputs of the unit 6 for determining the optimal control actions, output the coefficients of the objective function from the outputs of the mode sensor 11 (see Fig. 1). These coefficients depend on the specifics of the control object and are selected at the design stage of the control system for different operating modes

The second multiplying elements 19 (see Fig. 3) multiply the quality indicators of the system by the corresponding coefficients of the objective function, the values of the products of the indicated values. From the outputs of the second multiplying elements 19 go to the inputs of the third adder 20, which summarizes the values of these products. Thus, the second multiplier elements 19 and the third adder 20 carry out the computational values of the objective function for a particular set of system quality indicators (for current monitoring). This value is applied to the first input of the second comparison element 21 and to the first information input of the storage device 18.

At the address input of the storage device 18, which is the information input of the block for determining the optimal control actions, the class code to which the current observation belongs is output from the output of the observation classification block 5. Regarding this . with the code from the memory device 18, the best value of the objective function of this class achieved during the previous observation period is read and delivered to its first information output; this value is fed to the second input of the second element of comparison 21. The initial matrix of optimal control actions by the classes of variables and the corresponding vector of values of the target functions are built during the design phase of a particular control system and stored in a storage device. As the system operates, the matrix is updated by writing to the best 18 in the sense of the established criterion of optimality of control actions. At the same time, in accordance with the code of the current observation class, from the storage device 18, the values of the controlled variables of the system corresponding to this class of the current observation and the value of the objective function at the first input of the storage device 18 are read and output to its second information outputs. variables of the system are fed to the information inputs of the second group of elements BAN24 The second element of comparison 21 is 0 — it compares the value of the objective function for the current observation nor (at the first input) and the best previously achieved value of the objective function of the class to which the current observation belongs (at the second input 5 de). If the value of the objective function of the current observation is better than the best of the previously achieved values of the objective function of the class to which the current observation belongs, then the second element, compared to 21, generates a signal 1 at its output. For example, if the target function is built to minimize the deviation of the quality indicators from the set standard for the process then the best

5 out of two values is the minimum value of the objective function.

The output signal of the second element of comparison 21 is supplied to the control inputs of the first group of elements

0 and through element 22, inversely, to the control inputs of the second group of elements BAN 24 and memory 18.

The information inputs of the first group of elements are PROHIBITED 23, which form the second group of information inputs of the unit for determining the optimal control actions, from the outputs of the unit for measuring signals of the controlled inputs, the measured signal values

0 controlled inputs of the control object 1. Elements FORBID 23 of the first group and elements FORBID of the second group can transmit to their outputs the values of the signals of optimal control actions

5 If signal 1 is generated at the output of the second comparison element 21, then the BAN 23 elements are opened and the BAN 24 elements are closed. If a signal O is generated at the output of the second element 21, then the elements FORBID 24 are opened and the elements FORBID 23 are closed. The outputs of the elements PROHIBIT 23 and PROHIBIT 24 are fed to the inputs of the group of elements OR 25. Since elements of the first 23 and second 24 groups ZA5 PRET cannot be opened at the same time, the outputs of the OR elements 25, which are the outputs of the unit for determining the optimal control actions, will display either the signals of the controlled inputs of the control object 1 from the output of block 3

measuring the signals of the controlled inputs (if the target function of the current observation is better than the best one previously achieved for this class), or the previously stored set of signals of the controlled inputs of the control object 1, which is read from the memory device 18 and corresponds to the best previously stored value of the target function.

From the outputs of the elements of the PROHIBITION 23 of the first group, signals are sent both to the inputs of the elements OR 25 and to the second information inputs of the memory device 18. Information from these information inputs is stored in the memory device 18 by the signal at the control input of the memory device 18 the transition of this signal from state 1 to state O. At the same time, a signal is written to memory device 18 from its first information input - the new best value of the objective function for this class.

The new values of the objective function and the corresponding new values of the best for this class of control actions are recorded instead of the outdated information at the previous address - in accordance with the class number of the current observation and the code at the information input of the optimal control actions determination unit 6

Thus, in the process of functioning of the system in the storage device 18 of the unit for determining the optimal control actions, there is a constant accumulation of information about the control object 1 and the system as a whole, there is a constant improvement in the quality of control, thereby achieving the aim of the invention is to improve the quality of control technological process

Improving the quality of management is achieved by:

-by taking into account the influence on the process of unmeasured inputs of an object, for example, the parameters of raw materials determined by its technological prehistory (technology modes on the previous units in the technological chain);

- due to the constant correction of optimal control actions, which is carried out in the block for determining the optimal control actions;

- due to the unconditional attainability of the optimal point inherent in the structure of the system itself, formed in the block for determining the optimal control actions for the current state of the control object, and for the class defined in the block for classifying observations;

- by eliminating the possibility of control actions going beyond technological limits by selecting control actions in the control actions generating unit from the set of permissible control actions.

The proposed system can be used for adaptive control of various technological processes and for various control objects, for example, in metallurgy on a continuous broadband hot-rolling mill for steel strips, in oil refining at a catalytic cracking unit, etc.

The proposed system can be implemented on known in automation sensors, actuators, nodes and elements.

Block 2 (see Fig. 1) for measuring signals of uncontrolled inputs, block 3 for measuring signals for controlled inputs, block A for measuring quality indicators are implemented on conventional sensors of technological monitoring and control systems that convert electrical and non-electrical input signals to standard electrical signals. For the control object — a continuous broadband hot rolling mill, such sensors are, for example, IUM series force meters, X-ray meters 1 of strip thickness, and VNIIMETMASH rolled width meters. The sensors contain signal processing means, or may be supplemented with such standard means. For example, modern x-ray thickness gauges have built-in computers with programs for static calculations.

The actuator unit 9 may be implemented on standard actuators in accordance with the type of control object. For example, for a continuous broadband hot rolling mill with two controlled inputs - vertical and horizontal compression of the workpiece - conventional electromechanical or hydraulic press devices of a rolling mill, for example, a pistonless hydraulic press, can be used as actuators

Block 5 classification of observations, block 6 determining the optimal control actions, block 7 forming corrective actions, block 8 forming control actions, sensor classes 10 and mode dial 11 can be implemented on a serial electronic computer technology, for example, computers Electronics 60 - 15 VUMS-28 (5), or - on separate serial electronic elements.

The corrective actions generating unit 7 can be implemented, for example, in the form of a coefficient table entered in a programmable logic matrix or in a permanent storage device, for example, on the KR573RF2, KR556RT5 microcircuits,

In this case, the address inputs of the microcircuits are the inputs of the block 7 of the formation of corrective actions, the outputs of the microcircuits are the outputs of this block.

The control action generating unit 8 can be implemented as a set of multiplying elements, the number of which is equal to the number of controlled inputs of the control object, for example, on the multipliers series КР1802.

The class switch 120 can be implemented, for example, on software switches xP6-11T (ОУЗ 602.160ТУ), or on memory devices КР573РФ2, КР556РТ5. Such a storage device must store a set of reference signals for controlled inputs, uncontrolled inputs and quality indicators of the control object for all classes of non-measured inputs. The definition of these standards is carried out at the stage of designing an optimal control system, for example, according to one of the well-known cluster analysis algorithms, for example, ISODATA, King's algorithm, etc. Another way to build reference values is to build classes from the results of preliminary studies of raw materials and objects Management, Or according to experts, and calculating arithmetic mean values in each constructed class,

The mode switch 11 can be implemented on program switches, for example, PP6-11T (ОУ 602.160 ТУ) or on memory devices КР556РТ5, КР573РФ2, which store a set of coefficients with variable optimality criteria. The determination of these coefficients is carried out at the design stages or trial operation of the system in accordance with the requirements for product quality. The regimes and the corresponding coefficients are changed when the assortment or size of the products is changed

The encoder 17 may be constructed, for example, on the basis of a table entered in a programmable logic matrix or in programmable read-only memory, for example, on K573RF2, KR556RT4 microcircuits.

The storage device 18 can be implemented, for example, on the memory chips, such as K155RU2, KR565RU1A.

The first 4, second 12 and third 20 adders can be implemented, for example, on adder microchips, such as K155IM2, K155IMZ. Inverted inputs of the adders can be implemented, for example, inverters K155LN1.

The first 15 and second 21 comparison elements can be implemented, for example, on adders with direct Vi inverse inputs, performing the code subtraction operation, the sign output of the adder is the output of the comparison element. Such adders are implemented on conventional adder circuits (K155IM2, K155IMZ) with inverse inputs on K155LN1 circuits.

The first 13 and second 19 multiplier elements can be implemented, for example, on the multiplier circuits of the KR1802 series.

Element And 16 can be implemented, for example, on K155LI5 chips. Element HE 22 can be implemented, for example, on K155LN1 microcircuits.

BAN elements first 23 and second

24 groups can be implemented, for example, on K155LI1 microcircuits. Elements OR

25 can be implemented, for example, on K155LL1 microcircuits.

The economic effect of the implementation of the proposed system is formed by improving the quality of management by increasing the volume of products, improving the quality of products and reducing the cost of their production

For example, for a continuous broadband hot-rolling mill of steel strips with the introduction of this system, it is possible to achieve an annual economic effect by increasing the shipment by theoretical weight:

E TV (K2 - K1) C - (KA + KR) SI - - 2,000,000 (0.0513 - 0.05) 100000 - (0.12 + 0.05) 100000 - 0.15 300000 200000 rub .;

where TV 2000000 - the volume of shipment by theoretical weight, tons;

K1 0.05, K2 0.0513 — shipping factors by theoretical weight before and after the implementation of the system;

Ts 100 - the price of a given metal, rub per ton:

KA 0.12, KR 0.05 - ratios of depreciation and current repair;

KN 0.15 - regulatory efficiency coefficient;

SR 300000, SI 100000 - the cost of developing and manufacturing the system, rubles.

The effect on the proposed system will be 25%, or 50,000 rubles per year.

Claims (1)

The claims An adaptive process control adaptive control system comprising serially connected corrective action generation unit, control action generation unit and actuator block, the outputs of which are connected to the inputs of the measurement unit of the signals of the controlled inputs and to the controlled inputs of the control object, uncontrolled inputs of the control object connected to the inputs of the unit for measuring signals of uncontrolled inputs, the outputs of the control object are connected to the inputs of the unit for measuring quality indicators characterized in that, in order to improve the quality of control of objects with non-measurable inputs, a class selector, a mode dial, an optimal control actions determination unit and an observation classification unit are introduced, the first group of information inputs of which are connected to the outputs of the controlled signal measurement unit of inputs, the second group of information inputs of the classification block of observations is connected to the outputs of the measurement block of signals of uncontrolled inputs, the third group of information inputs of the classification block of observations is connected to the outputs of the block for measuring quality indicators, the information input of the block for determining optimal control actions is connected to the output of the classification block 1 for observations, the first group of information inputs for the outputs of the block for measuring quality indicators, the second group of information inputs for the outputs of the block measuring signals of controlled inputs, the output of the block for determining the optimal control actions is connected to the inputs of the block for generating corrective actions and to the group of inputs of the block the worlds of control actions, the outputs of the class master and mode master are connected respectively with the inputs of the flea task of the classification of observations and the block for determining the optimal control actions. 1 FIG. 2 FIG. 5