CN116224795A - Thermoelectric production equipment control method based on machine learning model - Google Patents
Thermoelectric production equipment control method based on machine learning model Download PDFInfo
- Publication number
- CN116224795A CN116224795A CN202310206142.2A CN202310206142A CN116224795A CN 116224795 A CN116224795 A CN 116224795A CN 202310206142 A CN202310206142 A CN 202310206142A CN 116224795 A CN116224795 A CN 116224795A
- Authority
- CN
- China
- Prior art keywords
- data set
- machine learning
- value
- learning model
- variable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000010801 machine learning Methods 0.000 title claims abstract description 95
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000006870 function Effects 0.000 claims abstract description 52
- 238000005457 optimization Methods 0.000 claims abstract description 16
- 238000010276 construction Methods 0.000 claims abstract description 10
- 230000003247 decreasing effect Effects 0.000 claims description 4
- 238000009826 distribution Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 abstract description 3
- 239000003245 coal Substances 0.000 description 4
- 238000004590 computer program Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000003860 storage Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to a thermoelectric production equipment control method based on a machine learning model, which comprises the following steps: s1, acquiring current operation data of thermoelectric production equipment in operation of the thermoelectric production equipment; the operation data of the thermoelectric production equipment comprises controlled variables, controlled variable adjustment target targets and/or pre-designated auxiliary variables; s2, inputting the current operation data of the thermoelectric production equipment into a pre-established machine learning model, and outputting an adjustment value of an operation variable; the pre-established machine learning model is obtained by utilizing a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function; the final modeling data set is established in advance for the pre-collected thermoelectric production device operating data according to the pre-set construction rules. The method solves the technical problems that the existing control method has higher dependence on manual experience, the control model training process is complex, and the requirements on data quantity and data quality are higher.
Description
Technical Field
The invention relates to the technical field of thermoelectric production, in particular to a thermoelectric production equipment control method based on a machine learning model.
Background
Along with the continuous development of industrialization process and scientific technology, more and more control methods are applied to the thermoelectric production process, and in the prior control technology, the production process is monitored by a temperature, pressure and other detection instruments, and the automatic control of thermoelectric production is realized by using a DCS control system, or a model is built by adopting a system identification or neural network fuzzy logic method, and the control of the production process is realized by combining a model predictive control method.
However, in the existing control method, the DCS control requires engineers to adjust parameters on site according to actual scenes, and related parameters are required to be continuously adjusted along with the time lapse, the change of production conditions and the like so as to ensure that the expression capacity of the model is not attenuated. The control method using the model has complex implementation process and higher requirements on data quantity and data quality, wherein the system identification has no ability to a complex system and easily falls into a local optimal solution, and the method combining neural network fuzzy logic and the like solves the control effectiveness and sacrifices the interpretability of the system.
Therefore, the existing control method has the problems of higher dependence on artificial experience, complex control model training process, higher requirements on data quantity and data quality and lower effectiveness on control of thermoelectric production equipment.
Disclosure of Invention
First, the technical problem to be solved
In view of the above-mentioned shortcomings and disadvantages of the prior art, the present invention provides a control method for a thermoelectric production device based on a machine learning model, which solves the technical problems of high dependence on manual experience, complex training process of the control model, high requirements on data quantity and data quality, and low control effectiveness for the thermoelectric production device in the existing control method.
(II) technical scheme
In order to achieve the above purpose, the main technical scheme adopted by the invention comprises the following steps:
the embodiment of the invention provides a thermoelectric production equipment control method based on a machine learning model, which comprises the following steps:
s1, acquiring current operation data of thermoelectric production equipment in operation of the thermoelectric production equipment;
the operation data of the thermoelectric production equipment comprises controlled variables, controlled variable adjustment target targets and/or pre-designated auxiliary variables;
s2, inputting the current operation data of the thermoelectric production equipment into a pre-established machine learning model, and outputting an adjustment value of an operation variable by the pre-established machine learning model;
the pre-established machine learning model is obtained through a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function;
the final modeling data set is established in advance for the pre-collected thermoelectric production device operation data according to a pre-set construction rule.
Preferably, before S1, the method further includes:
s0, establishing a final modeling data set according to a preset construction rule aiming at the operation data of the thermoelectric production equipment, and acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
Preferably, the S0 specifically includes:
s01, collecting operation data of N thermoelectric production devices at different moments, forming a device operation data set, and preprocessing the device operation data set to obtain an initial data set;
the device operation data set comprises N data groups which are respectively in one-to-one correspondence with N different moments, wherein each data group comprises: operating variables, controlled variables and/or pre-specified auxiliary variables;
wherein the pre-specified auxiliary variable comprises at least one specified variable in operation of the thermoelectric production device;
wherein N is greater than or equal to 1000;
s02, establishing a final modeling data set by adopting a preset construction rule aiming at the initial data set;
s03, acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
Preferably, the S01 includes:
removing the data groups which are completely the same in the equipment operation data set until only one data group remains, and deleting the incomplete data groups in the equipment operation data set to obtain an initial data set;
the data sets which are completely the same are the same in the values of the operation variables, the same in the values of the controlled variables and the same in the values of each specified variable in different data sets;
the incomplete data set is the values of all specified variables of the manipulated, controlled and auxiliary variables, one of which is a null value.
Preferably, the S02 includes:
s02-1, acquiring the number H of data groups in an initial data set, and acquiring an operation variable adjustment value delta x of an operation variable in the data group i relative to a currently set j value by adopting a formula (1) aiming at any one data group i from the T to the H-T in the initial data set ij And obtaining a controlled variable adjustment value deltay of the controlled variable in the data set i with respect to the currently set j value using formula (2) ij ；
Wherein, the formula (1) is:
Δx ij ＝x i -x i+j ；
wherein, the formula (2) is:
Δy ij ＝y i -y i+j ；
x i operating variable values in the ith data group in the initial dataset;
y i the method comprises the steps of setting controlled variable values in an ith data group in an initial data set;
j is an integer with an absolute value less than or equal to a constant T, and T is more than or equal to 20;
wherein, T is more than or equal to i and less than or equal to H-T;
s02-2, will be anyAn operation variable adjustment value Deltax corresponding to the data group i ij Controlled variable adjustment value deltay ij The controlled variable value and/or the auxiliary variable value form a characteristic group;
the auxiliary variable values include each specified variable value;
s02-3, updating the value of j, and repeating the steps S02-1 to S02-3 until the value of j is traversed in a first range, so as to obtain a first characteristic data set;
the first feature data set includes: feature sets at each j value in a first range to which each of the T to H-T data sets of the initial data set corresponds, respectively;
the first range is greater than or equal to-T and less than or equal to T;
s02-4, screening the characteristic groups in the first characteristic data set according to monotonicity of the controlled variable and the operation variable in the operation of the thermoelectric production equipment to obtain a second characteristic data set;
s02-5, extracting the feature group according to a preset extraction mode based on the second feature data set to form a final modeling data set.
Preferably, the step S02-4 specifically comprises:
if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonicity increasing relation, eliminating the characteristic group meeting the first setting condition in the first characteristic data set;
the first setting condition is: the product of the operation variable adjustment value and the controlled variable adjustment value in the characteristic group is smaller than 0;
if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonically decreasing relation, eliminating the characteristic group meeting the second setting condition in the second characteristic data set;
the second setting condition is: the product of the manipulated variable adjustment value and the controlled variable adjustment value in the feature set is greater than 0.
Preferably, the step S02-5 specifically comprises:
s02-5-1, dividing the range of all operation variable adjustment values in the second characteristic data set into k equal-length numerical intervals; wherein k is greater than or equal to 5;
s02-5-2, acquiring feature groups corresponding to each numerical value interval respectively for each numerical value interval, and judging whether the number of the feature groups corresponding to each numerical value interval is more than or equal to M;
wherein M is greater than or equal to 100;
the characteristic group corresponding to each numerical value interval is the characteristic group of the operation variable adjusting value in the numerical value interval;
if the number is greater than or equal to M, randomly extracting M characteristic groups from the characteristic groups corresponding to the numerical value interval to form a sub-characteristic set corresponding to the numerical value interval;
if the number of the feature groups is less than M, all the feature groups corresponding to the numerical value interval are combined into a sub-feature set corresponding to the numerical value interval;
s02-5-3, merging the sub-feature sets respectively corresponding to all the numerical intervals to form a final modeling data set.
Preferably, the method comprises the steps of,
the preset fitness function is as follows:
wherein: alpha is a first model parameter to be solved;
beta is a second model parameter to be solved;
gamma is a third model parameter to be solved;
w is the total number of feature sets in the final modeling dataset;
ε w a random item which is subject to standard normal distribution and corresponds to the w feature group in the final modeling data set;
g () is the model hyper-parameter to be solved;
Y w the values of the controlled variables in the w feature group in the final modeling data set are obtained;
ΔY w adjusting values for controlled variables in a w-th feature group in the final modeling data set;
ΔX w adjusting values for the operating variables in the w-th feature set in the final modeling data set;
when the data group of the equipment operation data set comprises a pre-designated auxiliary variable, the value of d is equal to 1; when the data group of the equipment operation data set does not comprise a pre-designated auxiliary variable, the value of d is equal to 0;
Z w the total value of all appointed variables in the auxiliary variables in the w characteristic group in the final modeling data set is obtained;
in the sine and cosine optimization algorithm;
initializing a population size [100, 500];
the optimizing space of the first model parameter alpha to be solved is a real number which is not 0;
the optimizing space of the second model parameter beta to be solved and the optimizing space of the third model parameter gamma to be solved are all real numbers;
the optimizing space of the model hyper-parameters g () to be solved is: basic elementary functions such as power functions, exponential functions, logarithmic functions, trigonometric functions, inverse trigonometric functions and the like.
Preferably, the step S03 includes:
s03-1, according to the final modeling data set, solving a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super parameter g ();
s03-2, obtaining an initial machine learning model based on a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super-parameter g (), which are obtained when the preset fitness function value is minimum;
wherein the initial machine learning model is:
ΔY＝f(ΔX)＝α 0 *g 0 (ΔX)+β 0 *Y+d×γ 0 *Z；
α 0 a specific value of a first model parameter alpha to be solved when the preset fitness function value is the smallest;
β 0 the specific value of the second model parameter beta to be solved is the specific value of the second model parameter beta to be solved when the preset fitness function value is the minimum;
γ 0 the specific value of the third model parameter gamma to be solved is the specific value of the third model parameter gamma to be solved when the preset fitness function value is the minimum;
g 0 () Model hyper-parameters g ()'s for minimizing the preset fitness function value;
y is a controlled variable value which needs to be input into an initial machine learning model;
z is the total value of all specified variables in auxiliary variables required to be input into an initial machine learning model;
Δx is the manipulated variable adjustment value that needs to be input into the initial machine learning model;
Δy is a controlled variable adjustment value output by the initial machine learning model after Y, Z and Δx are input into the initial machine learning model;
s03-3, aiming at an initial machine learning model, acquiring the machine learning model;
wherein the machine learning model is:
ΔX＝f -1 (ΔY)＝F(ΔY，Y，Z)。
preferably, the S2 specifically includes:
taking the result of subtracting the current controlled variable from the controlled variable adjustment target as delta Y, taking the current controlled variable as Y, taking the current auxiliary variable as Z, and further substituting delta Y, Y and/or Z into the machine learning model, wherein the preset machine learning model outputs the adjustment value of the operation variable.
(III) beneficial effects
The beneficial effects of the invention are as follows: according to the thermoelectric production equipment control method based on the machine learning model, the current thermoelectric production equipment operation data are input into a pre-established machine learning model, and the pre-set machine learning model outputs an adjustment value of an operation variable; the machine learning model is obtained by utilizing a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function, so that the machine learning model is prevented from sinking into a local optimal solution, meanwhile, the machine learning model structure formed by the preset fitness function is simple in implementation process and low in dependence on data, meanwhile, the problem of interpretability of the machine learning model is solved, and the control effectiveness of thermoelectric production equipment is improved.
Drawings
FIG. 1 is a flow chart of a method for controlling a thermoelectric production device based on a machine learning model according to the present invention;
fig. 2 is a schematic diagram showing a comparison of a control process of a thermoelectric production device control method based on a machine learning model according to the present invention and a control process of a conventional PID control method.
Detailed Description
The invention will be better explained by the following detailed description of the embodiments with reference to the drawings.
In order that the above-described aspects may be better understood, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
Referring to fig. 1, the present embodiment provides a thermoelectric production device control method based on a machine learning model, the method comprising:
s1, acquiring current operation data of the thermoelectric production equipment in the operation of the thermoelectric production equipment.
The thermoelectric production device operating data includes a controlled variable, a controlled variable adjustment target, and/or a pre-specified auxiliary variable.
S2, inputting the current operation data of the thermoelectric production equipment into a pre-established machine learning model, and outputting an adjustment value of an operation variable by the pre-established machine learning model.
In practice, operators of current thermoelectric production facilities adjust the operating variables according to the adjustment values.
The pre-established machine learning model is obtained through a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function.
The final modeling data set is established in advance for the pre-collected thermoelectric production device operation data according to a pre-set construction rule.
In a practical application of the present embodiment, the thermoelectric production device control method based on the machine learning model in the present embodiment further includes, before the step S1:
s0, establishing a final modeling data set according to a preset construction rule aiming at the operation data of the thermoelectric production equipment, and acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
Specifically, the S0 specifically includes:
s01, collecting operation data of the thermoelectric production equipment at N different moments, forming an equipment operation data set, and preprocessing the equipment operation data set to obtain an initial data set.
The device operation data set comprises N data groups which are respectively in one-to-one correspondence with N different moments, wherein each data group comprises: an operating variable, a controlled variable and/or a pre-specified auxiliary variable.
In the embodiment, if the operation variable is the coal feeding amount, the controlled variable is the boiler load, and only one specified variable in the auxiliary variables is the total air quantity; the partial data set in the device operational dataset as in table 1:
wherein the pre-specified auxiliary variable comprises at least one specified variable in operation of the thermoelectric production device.
Wherein N is greater than or equal to 1000.
The step S01 specifically comprises the following steps:
and eliminating the data groups which are completely the same in the equipment operation data set until only one data group remains, and deleting the incomplete data groups in the equipment operation data set to obtain an initial data set.
The data sets which are completely the same are the same in the values of the operation variables in different data sets, the values of the controlled variables y are the same and the values of each specified variable are the same; for example, if there are an operating variable, a controlled variable, and two specified variables (for example, specified variable a and specified variable B) included in one data set, if there are the same values of the operating variable in the other data set and the operating variable in the previous data set, the controlled variable and the controlled variable in the previous data set are the same, the specified variable a and the specified variable a in the previous data set are the same, and the specified variable B in the previous data set are the same, then these two data sets are referred to as identical data sets, for example, if there are 5 data sets that are identical, then in this embodiment, these 5 identical data sets are deleted by 4 but only 1 is retained.
The incomplete data set is the values of all specified variables of the manipulated, controlled and auxiliary variables, one of which is a null value.
In this embodiment, if one data set includes an operation variable, a controlled variable, and two specified variables (for example, a specified variable a and a specified variable B), the data set is an incomplete data set if the operation variable is null, that is, if no specific value of the upper operation variable is acquired.
S02, establishing a final modeling data set by adopting a preset construction rule aiming at the initial data set.
The S02 includes:
s02-1, acquiring the number H of data groups in an initial data set, and acquiring an operation variable adjustment value delta x of an operation variable in the data group i relative to a currently set j value by adopting a formula (1) aiming at any one data group i from the T to the H-T in the initial data set ij And obtaining a controlled variable adjustment value deltay of the controlled variable in the data set i with respect to the currently set j value using formula (2) ij 。
Wherein, the formula (1) is:
Δx ij ＝x i -x i+j 。
wherein, the formula (2) is:
Δy ij ＝y i -y i+j 。
x i is the value of the operational variable in the i-th data set in the initial dataset.
y i Is the controlled variable value in the i-th data set in the initial dataset.
j is an integer with an absolute value less than or equal to a constant T, and T is greater than or equal to 20.
Wherein, T is more than or equal to i and less than or equal to H-T.
S02-2, adjusting the operation variable corresponding to any one of the data sets i to be delta y ij Controlled variable adjustment value deltay ij The controlled variable value and/or the auxiliary variable value form a characteristic group; the set of features constructed as in table 2 is the set of partial data in the device operational dataset of table 1:
the auxiliary variable values include each of the specified variable values.
S02-3, updating the value of j, and repeating the steps S02-1 to S02-3 until the value of j is traversed within a first range, so as to obtain a first characteristic data set.
The first feature data set includes: and (3) characteristic groups at each j value in the first range corresponding to each data group from the T to the H-T of the initial data set.
The first range is-T or more and T or less.
S02-4, screening the characteristic group in the first characteristic data set according to the monotonicity of the controlled variable and the operating variable in the operation of the thermoelectric production equipment to obtain a second characteristic data set.
The S02-4 specifically comprises:
and if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonicity increasing relation, eliminating the characteristic group which satisfies the first setting condition in the first characteristic data set.
The first setting condition is: the product of the manipulated variable adjustment value and the controlled variable adjustment value in the feature set is less than 0.
For example, if the coal feed is an operating variable, the boiler load is a controlled variable, and the total air quantity is an auxiliary variable, then there is actually an objective positive increasing relationship between the coal feed and the boiler load, that is, if the coal feed is objectively increased, the boiler load is increased, and this relationship is called as the increasing relationship between the operating variable and the controlled variable in the operation of the thermoelectric production device.
And if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonically decreasing relation, eliminating the characteristic group which satisfies the second setting condition in the second characteristic data set.
The second setting condition is: the product of the manipulated variable adjustment value and the controlled variable adjustment value in the feature set is greater than 0.
For example, if the manipulated variable increases objectively, then the controlled variable decreases, and this relationship is called decreasing the manipulated variable and the controlled variable during operation of the thermoelectric production device.
S02-5, extracting the feature group according to a preset extraction mode based on the second feature data set to form a final modeling data set.
The S02-5 specifically comprises the following steps:
s02-5-1, dividing the range of all operation variable adjustment values in the second characteristic data set into k equal-length numerical intervals; wherein k is 5 or more.
S02-5-2, for each numerical value interval, acquiring the feature groups corresponding to each numerical value interval respectively, and judging whether the number of the feature groups corresponding to each numerical value interval is more than or equal to M.
Wherein M is 100 or more.
The feature group corresponding to each numerical value interval is the feature group of the operation variable adjusting value in the numerical value interval.
If the number is greater than or equal to M, randomly extracting M characteristic groups from the characteristic groups corresponding to the numerical value interval to form a sub-characteristic set corresponding to the numerical value interval.
If the number is less than M, all the feature groups corresponding to the numerical value interval are combined into a sub-feature set corresponding to the numerical value interval.
S02-5-3, merging the sub-feature sets respectively corresponding to all the numerical intervals to form a final modeling data set.
S03, acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
In this embodiment, the predetermined fitness function is:
wherein: alpha is a first model parameter to be solved.
Beta is a second model parameter to be solved.
And gamma is a third model parameter to be solved.
W is the total number of feature sets in the final modeled dataset.
ε w Is a random term subject to standard normal distribution corresponding to the w-th feature group in the final modeling data set.
g () is the model hyper-parameter to be solved.
Y w The values of the controlled variables in the w-th feature set in the final modeling data set.
ΔY w And adjusting the values for the controlled variables in the w-th feature group in the final modeling data set.
ΔX w The values are adjusted for the operating variables in the w-th feature set in the final modeling dataset.
When the data group of the equipment operation data set comprises a pre-designated auxiliary variable, the value of d is equal to 1; when the data set of the device operation data set does not comprise a pre-specified auxiliary variable, the value of d is equal to 0.
Z w The total value of all specified variables in the auxiliary variables in the w-th feature set in the final modeling data set.
The sine and cosine optimization algorithm.
The population size is initialized [100, 500].
The optimizing space of the first model parameter alpha to be solved is a real number which is not 0.
The optimizing space of the second model parameter beta to be solved and the optimizing space of the third model parameter gamma to be solved are all real numbers.
The optimizing space of the model hyper-parameters g () to be solved is: basic elementary functions such as power functions, exponential functions, logarithmic functions, trigonometric functions, inverse trigonometric functions and the like.
In practical application of the embodiment, if for a certain value c, the optimizing space of the model super parameter g (c) to be solved is:
specifically, the step S03 includes:
s03-1, according to the final modeling data set, solving a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super parameter g () to be solved when the preset fitness function value is minimum by utilizing a sine and cosine optimization algorithm.
S03-2, obtaining an initial machine learning model based on a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super-parameter g (), which are obtained when the preset fitness function value is minimum.
Wherein the initial machine learning model is:
ΔY＝f(ΔX)＝α 0 *g 0 (ΔX)+β 0 *Y+d×γ 0 *Z。
α 0 and the specific value of the first model parameter alpha to be solved is the specific value of the first model parameter alpha to be solved when the preset fitness function value is the minimum.
β 0 And the specific value of the second model parameter beta to be solved is the specific value of the second model parameter beta to be solved when the preset fitness function value is the minimum.
γ 0 And the specific value of the third model parameter gamma to be solved is the specific value when the preset fitness function value is the minimum.
g 0 () And (3) a model super-parameter g () for minimizing the preset fitness function value.
Y is the controlled variable value that needs to be entered into the initial machine learning model.
Z is the total value of all specified variables in the auxiliary variables that need to be entered into the initial machine learning model.
Δx is the manipulated variable adjustment value that needs to be input into the initial machine learning model.
Δy is the controlled variable adjustment value output by the initial machine learning model after Y, Z and Δx are input into the initial machine learning model.
S03-3, aiming at the initial machine learning model, acquiring the machine learning model.
Wherein the machine learning model is:
ΔX＝f -1 (ΔY)＝F(ΔY，Y，Z)。
in practice the machine learning model in this embodiment is an inverse function of the initial machine learning model.
In this embodiment, the S2 specifically includes:
taking the result of subtracting the current controlled variable from the controlled variable adjustment target as delta Y, taking the current controlled variable as Y, taking the current auxiliary variable as Z, and further substituting delta Y, Y and/or Z into the machine learning model, wherein the preset machine learning model outputs the adjustment value of the operation variable.
In this embodiment, the preset adjustment value of the machine learning model output operation variable may enable the device controlled variable to reach or stabilize at the target set value target from the current position.
The embodiment provides a control method of a thermoelectric production device based on a machine learning model, which uses a control process of the machine learning model on a boiler load and a conventional PID control pair, as shown in fig. 2, and has better model interpretability, and compared with the conventional control method, the control process of the thermoelectric production device based on the machine learning model in the embodiment reaches a control target faster and has no overshoot phenomenon.
In the thermoelectric production equipment control method based on the machine learning model, the current thermoelectric production equipment operation data is input into a pre-established machine learning model, and the pre-set machine learning model outputs an adjustment value of an operation variable; the machine learning model is obtained by utilizing a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function, so that the machine learning model is prevented from sinking into a local optimal solution, meanwhile, the machine learning model structure formed by the preset fitness function is simple in implementation process and low in dependence on data, meanwhile, the problem of interpretability of the machine learning model is solved, and the control effectiveness of thermoelectric production equipment is improved.
The present embodiment also provides a thermoelectric production device control system based on a machine learning model, including: at least one processor; and at least one memory communicatively coupled to the processor, wherein the memory stores program instructions executable by the processor, the processor invoking the program instructions capable of executing a machine learning model based thermoelectric production device control method as in the embodiments.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions.
It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the terms first, second, third, etc. are for convenience of description only and do not denote any order. These terms may be understood as part of the component name.
Furthermore, it should be noted that in the description of the present specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with the embodiment or example being included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art upon learning the basic inventive concepts. Therefore, the appended claims should be construed to include preferred embodiments and all such variations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, the present invention should also include such modifications and variations provided that they come within the scope of the following claims and their equivalents.
Claims (10)
1. A method of controlling a thermoelectric production facility based on a machine learning model, the method comprising:
s1, acquiring current operation data of thermoelectric production equipment in operation of the thermoelectric production equipment;
the operation data of the thermoelectric production equipment comprises controlled variables, controlled variable adjustment target targets and/or pre-designated auxiliary variables;
s2, inputting the current operation data of the thermoelectric production equipment into a pre-established machine learning model, and outputting an adjustment value of an operation variable by the pre-established machine learning model;
the pre-established machine learning model is obtained through a sine and cosine optimization algorithm according to a final modeling data set and a preset fitness function;
the final modeling data set is established in advance for the pre-collected thermoelectric production device operation data according to a pre-set construction rule.
2. The machine learning model-based thermoelectric production facility control method of claim 1, further comprising, prior to S1:
s0, establishing a final modeling data set according to a preset construction rule aiming at the operation data of the thermoelectric production equipment, and acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
3. The machine learning model-based thermoelectric production facility control method of claim 2, wherein S0 specifically comprises:
s01, collecting operation data of N thermoelectric production devices at different moments, forming a device operation data set, and preprocessing the device operation data set to obtain an initial data set;
the device operation data set comprises N data groups which are respectively in one-to-one correspondence with N different moments, wherein each data group comprises: operating variables, controlled variables and/or pre-specified auxiliary variables;
wherein the pre-specified auxiliary variable comprises at least one specified variable in operation of the thermoelectric production device;
wherein N is greater than or equal to 1000;
s02, establishing a final modeling data set by adopting a preset construction rule aiming at the initial data set;
s03, acquiring a machine learning model by utilizing a sine and cosine optimization algorithm according to the final modeling data set and a preset fitness function.
4. The machine learning model-based thermoelectric production facility control method of claim 3, wherein S01 comprises:
removing the data groups which are completely the same in the equipment operation data set until only one data group remains, and deleting the incomplete data groups in the equipment operation data set to obtain an initial data set;
the data sets which are completely the same are the same in the values of the operation variables, the same in the values of the controlled variables and the same in the values of each specified variable in different data sets;
the incomplete data set is the values of all specified variables of the manipulated, controlled and auxiliary variables, one of which is a null value.
5. The machine learning model-based thermoelectric production facility control method of claim 4, wherein S02 comprises:
s02-1, acquiring the number H of data groups in an initial data set, and acquiring an operation variable adjustment value delta x of an operation variable in the data group i relative to a currently set j value by adopting a formula (1) aiming at any one data group i from the T to the H-T in the initial data set ij And obtaining a controlled variable adjustment value deltay of the controlled variable in the data group i with respect to the currently set j value using the formula (2) ij ；
Wherein, the formula (1) is:
Δx ij ＝x i -x i+j ；
wherein, the formula (2) is:
Δy ij ＝y i -y i+j ；
x i operating variable values in the ith data group in the initial dataset;
y i the method comprises the steps of setting controlled variable values in an ith data group in an initial data set;
j is an integer with an absolute value less than or equal to a constant T, and T is more than or equal to 20;
wherein, T is more than or equal to i and less than or equal to H-T;
s02-2, adjusting the operation variable corresponding to any one of the data sets i to be delta x ij Controlled variable adjustment value deltay ij The controlled variable value and/or the auxiliary variable value form a characteristic group;
the auxiliary variable values include each specified variable value;
s02-3, updating the value of j, and repeating the steps S02-1 to S02-3 until the value of j is traversed in a first range, so as to obtain a first characteristic data set;
the first feature data set includes: feature sets at each j value in a first range to which each of the T to H-T data sets of the initial data set corresponds, respectively;
the first range is greater than or equal to-T and less than or equal to T;
s02-4, screening the characteristic groups in the first characteristic data set according to monotonicity of the controlled variable and the operation variable in the operation of the thermoelectric production equipment to obtain a second characteristic data set;
s02-5, extracting the feature group according to a preset extraction mode based on the second feature data set to form a final modeling data set.
6. The machine learning model-based thermoelectric production facility control method of claim 5, wherein S02-4 specifically comprises:
if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonicity increasing relation, eliminating the characteristic group meeting the first setting condition in the first characteristic data set;
the first setting condition is: the product of the operation variable adjustment value and the controlled variable adjustment value in the characteristic group is smaller than 0;
if the operation variable and the controlled variable in the operation of the thermoelectric production equipment are in monotonically decreasing relation, eliminating the characteristic group meeting the second setting condition in the second characteristic data set;
the second setting condition is: the product of the manipulated variable adjustment value and the controlled variable adjustment value in the feature set is greater than 0.
7. The machine learning model-based thermoelectric production facility control method of claim 6, wherein S02-5 specifically comprises:
s02-5-1, dividing the range of all operation variable adjustment values in the second characteristic data set into k equal-length numerical intervals; wherein k is greater than or equal to 5;
s02-5-2, acquiring feature groups corresponding to each numerical value interval respectively for each numerical value interval, and judging whether the number of the feature groups corresponding to each numerical value interval is more than or equal to M;
wherein M is greater than or equal to 100;
the characteristic group corresponding to each numerical value interval is the characteristic group of the operation variable adjusting value in the numerical value interval;
if the number is greater than or equal to M, randomly extracting M characteristic groups from the characteristic groups corresponding to the numerical value interval to form a sub-characteristic set corresponding to the numerical value interval;
if the number of the feature groups is less than M, all the feature groups corresponding to the numerical value interval are combined into a sub-feature set corresponding to the numerical value interval;
s02-5-3, merging the sub-feature sets respectively corresponding to all the numerical intervals to form a final modeling data set.
8. The method for controlling a thermoelectric production facility based on a machine learning model as claimed in claim 7,
the preset fitness function is as follows:
wherein: alpha is a first model parameter to be solved;
beta is a second model parameter to be solved;
gamma is a third model parameter to be solved;
w is the total number of feature sets in the final modeling dataset;
ε w a random item which is subject to standard normal distribution and corresponds to the w feature group in the final modeling data set;
g () is the model hyper-parameter to be solved;
Y w the values of the controlled variables in the w feature group in the final modeling data set are obtained;
ΔY w adjusting values for controlled variables in a w-th feature group in the final modeling data set;
ΔX w adjusting values for the operating variables in the w-th feature set in the final modeling data set;
when the data group of the equipment operation data set comprises a pre-designated auxiliary variable, the value of d is equal to 1; when the data group of the equipment operation data set does not comprise a pre-designated auxiliary variable, the value of d is equal to 0;
Z w the total value of all appointed variables in the auxiliary variables in the w characteristic group in the final modeling data set is obtained;
in the sine and cosine optimization algorithm;
initializing a population size [100, 500];
the optimizing space of the first model parameter alpha to be solved is a real number which is not 0;
the optimizing space of the second model parameter beta to be solved and the optimizing space of the third model parameter gamma to be solved are all real numbers;
the optimizing space of the model hyper-parameters g () to be solved is: basic elementary functions such as power functions, exponential functions, logarithmic functions, trigonometric functions, inverse trigonometric functions and the like.
9. The machine learning model-based thermoelectric production facility control method of claim 8, wherein S03 comprises:
s03-1, according to the final modeling data set, solving a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super parameter g ();
s03-2, obtaining an initial machine learning model based on a first model parameter alpha to be solved, a second model parameter beta to be solved, a third model parameter gamma to be solved and a model super-parameter g (), which are obtained when the preset fitness function value is minimum;
wherein the initial machine learning model is:
△Y＝f(△X)＝α 0 *g 0 (△X)+β 0 *Y+d×γ 0 *Z；
α 0 to make the pre-formThe specific value of a first model parameter alpha to be solved when the preset fitness function value is minimum;
β 0 the specific value of the second model parameter beta to be solved is the specific value of the second model parameter beta to be solved when the preset fitness function value is the minimum;
γ 0 the specific value of the third model parameter gamma to be solved is the specific value of the third model parameter gamma to be solved when the preset fitness function value is the minimum;
g 0 () Model hyper-parameters g ()'s for minimizing the preset fitness function value;
y is a controlled variable value which needs to be input into an initial machine learning model;
z is the total value of all specified variables in auxiliary variables required to be input into an initial machine learning model;
DeltaX is an operation variable adjustment value required to be input into the initial machine learning model;
delta Y is a controlled variable adjustment value output by the initial machine learning model after Y, Z and delta X are input into the initial machine learning model;
s03-3, aiming at an initial machine learning model, acquiring the machine learning model;
wherein the machine learning model is:
△X＝f -1 (△Y)＝F(△Y，Y，Z)。
10. the machine learning model-based thermoelectric production facility control method of claim 9, wherein S2 specifically comprises:
the result of subtracting the current controlled variable from the controlled variable adjustment target is taken as delta Y, the current controlled variable is taken as Y, the current auxiliary variable is taken as Z, and delta Y, Y and/or Z are further substituted into the machine learning model, and the preset machine learning model outputs the adjustment value of the operation variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310206142.2A CN116224795B (en) | 2023-03-06 | 2023-03-06 | Thermoelectric production equipment control method based on machine learning model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310206142.2A CN116224795B (en) | 2023-03-06 | 2023-03-06 | Thermoelectric production equipment control method based on machine learning model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116224795A true CN116224795A (en) | 2023-06-06 |
| CN116224795B CN116224795B (en) | 2023-11-17 |
Family
ID=86587061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310206142.2A Active CN116224795B (en) | 2023-03-06 | 2023-03-06 | Thermoelectric production equipment control method based on machine learning model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116224795B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110032780A (en) * | 2019-02-01 | 2019-07-19 | 浙江中控软件技术有限公司 | Commercial plant energy consumption benchmark value calculating method and system based on machine learning |
| CN112130453A (en) * | 2020-07-30 | 2020-12-25 | 浙江中控技术股份有限公司 | Control method and system for improving MCS production stability based on machine learning |
| JP2021188568A (en) * | 2020-05-29 | 2021-12-13 | 株式会社Jij | Device for constructing machine learning model, method and program therefor |
| US20220101198A1 (en) * | 2020-09-30 | 2022-03-31 | OnScale, Inc. | Automated generation of a machine learning model from computational simulation data |
| CN114721263A (en) * | 2022-03-16 | 2022-07-08 | 中国中材国际工程股份有限公司 | Intelligent regulation and control method for cement decomposing furnace based on machine learning and intelligent optimization algorithm |
| CN115045854A (en) * | 2022-07-06 | 2022-09-13 | 北京全应科技有限公司 | Parallel fan balance control method based on machine learning |
| CN115309037A (en) * | 2022-08-09 | 2022-11-08 | 西安热工研究院有限公司 | Denitration control intelligent PID self-adaptive correction method based on particle swarm optimization |
-
2023
- 2023-03-06 CN CN202310206142.2A patent/CN116224795B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110032780A (en) * | 2019-02-01 | 2019-07-19 | 浙江中控软件技术有限公司 | Commercial plant energy consumption benchmark value calculating method and system based on machine learning |
| JP2021188568A (en) * | 2020-05-29 | 2021-12-13 | 株式会社Jij | Device for constructing machine learning model, method and program therefor |
| CN112130453A (en) * | 2020-07-30 | 2020-12-25 | 浙江中控技术股份有限公司 | Control method and system for improving MCS production stability based on machine learning |
| US20220101198A1 (en) * | 2020-09-30 | 2022-03-31 | OnScale, Inc. | Automated generation of a machine learning model from computational simulation data |
| CN114721263A (en) * | 2022-03-16 | 2022-07-08 | 中国中材国际工程股份有限公司 | Intelligent regulation and control method for cement decomposing furnace based on machine learning and intelligent optimization algorithm |
| CN115045854A (en) * | 2022-07-06 | 2022-09-13 | 北京全应科技有限公司 | Parallel fan balance control method based on machine learning |
| CN115309037A (en) * | 2022-08-09 | 2022-11-08 | 西安热工研究院有限公司 | Denitration control intelligent PID self-adaptive correction method based on particle swarm optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116224795B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| De Veaux et al. | A comparison of two nonparametric estimation schemes: MARS and neural networks | |
| Valarmathi et al. | Real-coded genetic algorithm for system identification and controller tuning | |
| DE102019126310A1 (en) | System and method for optimizing the combustion of a boiler | |
| CN106779068A (en) | The method and apparatus for adjusting artificial neural network | |
| CN110991568B (en) | Target identification method, device, equipment and storage medium | |
| CA2941352A1 (en) | Neural network and method of neural network training | |
| WO2015043806A1 (en) | Method for the computer-aided control and/or regulation of a technical system | |
| EP2880499A1 (en) | Method for controlling and/or regulating a technical system in a computer-assisted manner | |
| Ignatyev et al. | System for automatic adjustment of intelligent controller parameters | |
| CN116224795B (en) | Thermoelectric production equipment control method based on machine learning model | |
| Gladkov et al. | Electronic computing equipment schemes elements placement based on hybrid intelligence approach | |
| CN114612450A (en) | Image detection segmentation method and system based on data augmentation machine vision and electronic equipment | |
| DE102019214625A1 (en) | Method, device and computer program for creating an artificial neural network | |
| Shahdi et al. | Supervised active learning method as an intelligent linguistic controller and its hardware implementation | |
| Abdulghafor et al. | Nonlinear convergence algorithm: structural properties with doubly stochastic quadratic operators for multi-agent systems | |
| Nasir et al. | Fractional-order pid controller design using pso and ga | |
| Muravyova et al. | Development of the intellectual complex for parallel work of steam boilers | |
| Dymova | Application of Characterization Analysis Methods to Investigation of Logical Networks Structures | |
| DE102019126293B4 (en) | System and method for regulating the operation of a boiler | |
| DE202019105304U1 (en) | Device for creating an artificial neural network | |
| EP3736749A1 (en) | Method and device for controlling a device using a dataset | |
| Zhou et al. | A membership function selection method for fuzzy neural networks | |
| Alfarraj et al. | Optimized automatic generation of fuzzy rules for nonlinear system based on subtractive clustering algorithm for medical image segmentation | |
| EP3650964A2 (en) | Method for controlling or regulating a technical system | |
| CN111279276A (en) | Randomized reinforcement learning for controlling complex systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |