CN115045854A - Parallel fan balance control method based on machine learning - Google Patents
Parallel fan balance control method based on machine learning Download PDFInfo
- Publication number
- CN115045854A CN115045854A CN202210799091.4A CN202210799091A CN115045854A CN 115045854 A CN115045854 A CN 115045854A CN 202210799091 A CN202210799091 A CN 202210799091A CN 115045854 A CN115045854 A CN 115045854A
- Authority
- CN
- China
- Prior art keywords
- fan
- machine learning
- parallel
- representing
- learning model
- 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.)
- Pending
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04D—NON-POSITIVE-DISPLACEMENT PUMPS
- F04D27/00—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04D—NON-POSITIVE-DISPLACEMENT PUMPS
- F04D27/00—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids
- F04D27/004—Control, e.g. regulation, of pumps, pumping installations or pumping systems specially adapted for elastic fluids by varying driving speed
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Control Of Positive-Displacement Air Blowers (AREA)
Abstract
The invention relates to a parallel fan balance control method based on machine learning, which comprises the following steps: step 1, acquiring historical data of parallel fans; step 2, constructing a machine learning model, training the machine learning model through historical data, and obtaining parameters of the machine learning model; step 3, correcting the observation data of the air duct pressure through the parameters of the machine learning model; step 4, predicting the target frequency of the parallel fans by using the parameters of the machine learning model, the corrected air duct pressure and the current operation data; and 5, adjusting the parallel fans according to the target frequency of the parallel fans until the target frequency with equivalent output is reached. The invention can continuously and automatically carry out balance control on the parallel fans; the labor is saved, and the efficiency in operation is improved; meanwhile, the parallel fans can be quantitatively and accurately controlled in a balanced manner, and the efficiency of the fans in operation is improved.
Description
Technical Field
The invention relates to the technical field of fan control, in particular to a parallel fan balance control method based on machine learning.
Background
In engineering, the hardware of the parallel fans can adopt products of the same type of the same manufacturer, and the parameters of the pipelines and the equipment of the fans are completely the same or close to each other. The existing control method of the parallel fans manually controls the startability of the two fans when starting so as to avoid wind robbing and even reverse rotation of the fans. In the running process of the parallel fans, if wind robbing occurs, the movable blades of the normal fans are manually opened to be large, the other fan is manually closed to be minimum, and when the pressure of the main air pipe returns to be normal, all parameters are stable, the parallel operation is restarted. After the parallel fans are started, the movable blade baffles are required to be opened to a certain degree, and downdraft is prevented. The wind pressure of the parallel fans in operation is kept stable, manual small-scale operation is carried out, the slow change of the wind pressure is guaranteed, and the level and experience of operators are relied on.
In the prior art, patent application number CN201811600062.0 discloses an automatic parallel method and system for primary fans of thermal power generating units, which provides an automatic parallel method when the primary fans are started, and has the disadvantages that whether parallel is successful is judged only by current, and balance control in operation is not performed. The patent application No. CN201410126530.0 control method for eliminating surge of axial flow fan has the defects of narrow scene applicability and multiple manual operations of operators matched with programs during operation. The prior art can guarantee the safety of the parallel fans at the bottom of a pocket, prevent large accidents and hidden dangers, smoothly control the parallel fans to start and close, and keep the air pressure basically stable during operation. However, the control scenes of the technologies are limited, only the scenes of parallel operation and surge during starting are covered, and the scenes of control during running are lacked; meanwhile, manual operation causes low control efficiency, the granularity of the control fan is thick, and continuous automatic control cannot be realized. Meanwhile, the automatic parallel method only has an air blower, lacks a parallel control method for an induced draft fan, and lacks a quantitative method for balance control of parallel fans, so that the parallel efficiency of the parallel fans is difficult to keep optimal.
Disclosure of Invention
The invention provides a general and efficient parallel fan balance control method based on machine learning, so that the output of parallel fans can be continuously and effectively and automatically balanced in the running process after the fans are started.
The invention provides a parallel fan balance control method based on machine learning, which comprises the following steps:
step 1, acquiring historical data of parallel fans;
step 2, constructing a machine learning model, training the machine learning model through historical data, and obtaining parameters of the machine learning model;
step 3, correcting the observation data of the air duct pressure through the parameters of the machine learning model;
step 4, predicting the target frequency of the parallel fans by using the parameters of the machine learning model, the corrected air duct pressure and the current operation data;
and 5, adjusting the parallel fans according to the target frequency of the parallel fans until the target frequency with equivalent output is reached.
Further, in step 1, the parallel fans include two identical pipelines, and the volume flow rates of the pipelines are as follows:
wherein Q is A Denotes the volume flow of the pipe A, Q B The volume flow of the pipeline B is shown, and Q represents the total air quantity in the air duct; α ═ u (u) A S) 2 ，β＝(u B S) 2 ；F A Represents the actual duct pressure, F, of the duct A B Represents the actual duct pressure for duct B; u. u A Denotes the flow coefficient, u, of the conduit A B Represents the flow coefficient, u, of the pipe B A ≈u B (ii) a S represents the cross-sectional area of the conduit.
Further, in step 2, the machine learning model includes first to third machine learning models when the pipe actual current can be obtained, wherein the first machine learning model is:
wherein, I A Represents the actual fan current of the pipeline A, I B Representing the actual fan current of the pipeline B;
the second machine learning model and the third machine learning model are:
wherein λ is A1 A first parameter, λ, representing the pipeline A A2 A second parameter, λ, representing the conduit A A3 A third parameter, λ, representing the pipeline A B1 A first parameter, λ, representing the pipe B B2 A second parameter, λ, representing the pipe B B3 A third parameter representing the pipe B; f. of A Representing the actual fan frequency of the pipeline A; f. of B Representing the actual fan frequency for duct B.
Further, in step 2, when the machine learning model cannot obtain the actual current of the pipeline, a fourth machine learning model is constructed by using a method of least square method:
wherein λ is A4 A fourth parameter, λ, representing the pipeline A A5 Denotes a fifth parameter, λ, of the conduit A B4 A fourth parameter, λ, representing the pipe B B5 Represents a fifth parameter of the pipeline B; f. of A Representing the actual fan frequency, f, of the duct A B Representing the actual fan frequency for duct B, min T represents minimizing the expression.
Further, in step 3, when the machine learning model can obtain the actual current of the pipeline, the machine learning model corrects the observed data of the duct pressure by the actual current:
wherein the content of the first and second substances,
observed data representing the pressure of the conduit in conduit a,
observed data representing the pressure of the ducts in duct B.
Further, in step 3, when the actual current of the pipeline cannot be obtained, the machine learning model corrects the observed number of the duct pressure by the parameters α and β:
wherein the content of the first and second substances,
observed data representing the pressure of the conduit in conduit a,
observed data representing the pressure of the ducts in duct B.
Further, in step 4, the machine learning model predicts the target frequency of the fan when the actual current of the pipeline can be obtained:
wherein, f' A Denotes the target Fan frequency, f 'of duct A' B Representing the target fan frequency for duct B.
Further, in step 4, the target frequency of the fan is predicted:
wherein Q is A ＝Q B ＝Q/2，f’ A Indicating tubeTarget fan frequency of road A, f' B Representing the target fan frequency for duct B.
Furthermore, in step 5, the target frequency of the parallel fan is predicted, the difference is made between the target frequency and the current actual frequency of the parallel fan, the fan with the larger absolute value of the difference starts to be adjusted to be close to the target frequency, then the other fan is adjusted accordingly, and the two fans are gradually adjusted in multiple steps according to the configuration parameters to gradually reach the target frequency with the equivalent output.
The invention achieves the following beneficial effects:
the method is suitable for all scenes when the parallel fans operate, including starting and stable operation.
The parallel fan control system can continuously and automatically carry out balance control on the parallel fans, and has the advantages of saving labor and improving the efficiency in operation and control; meanwhile, balance control can be conducted on the parallel fans quantitatively and accurately, and the positive effect is that the efficiency of the fans in operation is improved.
Drawings
Fig. 1 is a schematic flow chart of a parallel fan balance control method based on machine learning according to an embodiment of the present invention;
fig. 2 is a logic diagram of a parallel fan in the parallel fan balance control method based on machine learning according to the embodiment of the present invention.
Detailed Description
The technical solution of the present invention will be described in more detail with reference to the accompanying drawings, and the present invention includes, but is not limited to, the following embodiments.
As shown in fig. 1, the invention provides a parallel fan balance control method based on machine learning, which comprises the following steps:
step 1, acquiring historical data of parallel fans;
step 2, constructing a machine learning model, training the machine learning model through historical data, and obtaining parameters of the machine learning model;
step 3, correcting the observation data of the air duct pressure through the parameters of the machine learning model;
step 4, predicting the target frequency of the parallel fans by using the parameters of the machine learning model, the corrected air duct pressure and the current operation data;
and 5, adjusting the parallel fans according to the predicted parallel fan frequency until the target frequency with equivalent output is reached.
In the following embodiments, as shown in fig. 2, the logic of the parallel fans is illustrated, and the problem is abstracted into an upper pipeline a and a lower pipeline B, and a communication pipeline C can be arranged in the middle, and the same liquid flows in the pipelines. One direction is from the pipe into the mother pipe D and the other direction is from the mother pipe D into the two pipes. In engineering, the equipment on the pipeline A and the pipeline B is the same type of product manufactured by the same manufacturer, the equipment parameters are completely the same or very close, and the special condition of the fault of the pipeline A or the pipeline B is not considered. Knowing the pressure F of the two pipes A And F B Frequency f of fan unit on main pipe total flow Q, A, B pipeline A 、f B Or opening K A 、K B The two pipelines are equal in output force, the frequency or the opening degree is actually used for controlling the flow in the pipelines, and the output force is equal to the flow Q in the pipelines according to the definition of engineering A And Q B The same value, and the frequency of the fan device is adjusted to achieve the aim.
The invention is further illustrated by the following description in conjunction with the figures and the specific embodiments.
For embodiments in which the parallel fan is a blower (e.g., a secondary fan) or an induced fan, the details are as follows.
The first embodiment is as follows:
in this embodiment, the historical data of the last week is used, updated once a week, modeled again, and predicted by a new model.
Step 1, acquiring historical data of a week and actual frequencies f of two fan devices of the following parameters A And f B Actual duct pressure F of two fan units A And F B Total air quantity Q in air duct, actual current I of two fans A And I B 。
The known fan air volume model is as follows:
for line A, B, there is the following relationship according to "line flow-pressure relationship":
wherein Q is A Denotes the volume flow of the pipe A, Q B Represents the volumetric flow rate of the conduit B; u represents a flow coefficient, related to the valve, the fluid substance and the shape of the pipe, which coefficient is determined when these are determined; s represents the cross-sectional area of the pipeline; f AB Indicating the pressure before the parent D.
According to conservation of mass: q ═ Q A +Q B So, the system of equations can be obtained:
in general, since the parameters at A, B are identical or very close to each other, the flow coefficient u of A, B pipe is considered to be the flow coefficient u in a special case without considering the A or B fault A 、u B Very close, i.e. approximately equal: u. of A ≈u B Therefore:
to obtain Q A 、Q B Then let alpha be (u) respectively A S) 2 And β ═ u (u) B S) 2 Obtaining:
normally without Q A 、Q B The measured points α, β cannot be directly found out, and there are two different ways to solve this problem. In engineering, which method is used is determined according to whether the actual current of the parallel fans exists or not.
Step 2, obtaining the actual current I A And I B Three machine learning models are constructed by using a linear regression method.
According to the fan air volume model, because the fan efficiency, the mechanical transmission efficiency, the rated power and the rated current of the parallel fans are the same, the uncertain factors can be counteracted by calculating the ratio of the air volumes.
Recombined Q ═ Q A +Q B The following can be calculated:
according to the formulas (5) and (7),
joint subtraction to obtain a first machine learning model:
and fitting by using a linear regression method without intercept to obtain alpha and beta.
Substituting alpha and beta according to (6) and (8), fitting parameters in the second machine learning model and the third machine learning model by using two linear regression models, and obtaining a first parameter lambda of the pipeline A A1 Second parameter lambda of the pipeline A A2 A third parameter lambda of the pipeline A A3 First parameter lambda of the pipeline B B1 A second parameter lambda of the pipe B B2 A third parameter lambda of the pipeline B B3 。
The independent variable in the second machine learning model is
And F A Dependent variable is f A (ii) a The third machine learning model argument is
And F B The dependent variable is f B 。
If there is no actual current, a fourth machine learning model is constructed using a method of least squares.
According to the fan air volume model, because the fan efficiency, the mechanical transmission efficiency, the rated power and the rated current of the parallel fans are the same, the actual current is in direct proportion to the actual air volume and the actual pressure:
because the actual current of the fan is equivalent to the actual frequency, the fan has the advantages of low cost and high efficiency
Conversion is to the equation:
the following equation is minimized:
min T＝(λ A4 (Q A ×F A )+λ A5 -f A ) 2 +(λ B4 (Q B ×F B )+λ B5 -f B ) 2 (14)
substituting equation (5) into equation (14), i.e.:
the fourth parameter lambda of the pipeline A can be determined by means of a least-squares minimization (15) A4 Fifth parameter λ of the pipeline A A5 A fourth parameter λ of the pipe B B4 Fifth parameter λ of the pipe B B5 And α, β.
Step 3, correcting the observation data of the air duct pressure in order to make the prediction result more accurate;
and if the actual current exists, correcting the observation data of the air duct pressure through the actual current to adjust the observation data into the actual air duct pressure data:
if no actual current exists, the observation data of the air duct pressure is corrected through the parameters alpha and beta, so that the observation data is adjusted to be actual air duct pressure data:
and 4, predicting the target frequency of the two parallel fans according to the parameters of the machine learning model, the corrected air duct pressure and the current operation data to obtain the target frequency of the fans.
For the case of actual current, the following formula is used to predict the current,
for the case of no actual current, here Q A ＝Q B Q/2, using the following formula,
wherein, f' A Denotes the target Fan frequency, f 'of duct A' B Representing the target fan frequency for duct B.
And 5, predicting the target frequency of the parallel fans, making a difference with the current actual parallel fan frequency, starting adjustment of the fan with a larger absolute value of the difference, adjusting the fan to be close to the target frequency, then adjusting the other fan to follow, and gradually adjusting the two fans in multiple steps according to configuration parameters to gradually reach the target frequency with equivalent output.
Therefore, the complete control method of the parallel fan balance is completed, and the 5 steps comprise three major stages of acquiring data, establishing different models according to conditions, predicting and adjusting the frequency of the parallel fans, and efficiently and quantitatively balancing the air supply of the parallel fans.
The present invention is not limited to the above embodiments, and those skilled in the art can implement the present invention in other various embodiments according to the disclosure of the embodiments and the drawings, and therefore, all designs that can be easily changed or modified by using the design structure and idea of the present invention fall within the protection scope of the present invention.
Claims (9)
1. A parallel fan balance control method based on machine learning is characterized by comprising the following steps:
step 1, acquiring historical data of parallel fans;
step 2, constructing a machine learning model, training the machine learning model through historical data, and obtaining parameters of the machine learning model;
step 3, correcting the observation data of the air duct pressure through the parameters of the machine learning model;
step 4, predicting the target frequency of the parallel fans by using the parameters of the machine learning model, the corrected air duct pressure and the current operation data;
and 5, adjusting the parallel fans according to the target frequency of the parallel fans until the target frequency with equivalent output is reached.
2. The parallel fan balance control method according to claim 1, wherein in step 1, the parallel fan comprises two identical pipes, and the volume flow rates of the pipes are as follows:
wherein Q is A Denotes the volume flow of the pipe A, Q B The volume flow of the pipeline B is shown, and Q represents the total air quantity in the air duct; α ═ u (u) A S) 2 ，β＝(u B S) 2 ；F A Represents the actual duct pressure, F, of the duct A B Represents the actual duct pressure for duct B; u. of A Denotes the flow coefficient, u, of the conduit A B Represents the flow coefficient, u, of the pipe B A ≈u B (ii) a S represents the cross-sectional area of the conduit.
3. The parallel fan balance control method according to claim 2, wherein in step 2, the machine learning model includes first to third machine learning models when the actual current of the pipeline can be obtained, wherein the first machine learning model is:
wherein, I A Represents the actual fan current of the pipeline A, I B Representing the actual fan current of the pipeline B;
the second machine learning model and the third machine learning model are:
wherein λ is A1 A first parameter, λ, representing the pipeline A A2 A second parameter, λ, representing the conduit A A3 A third parameter, λ, representing the pipeline A B1 A first parameter, λ, representing the pipe B B2 A second parameter, λ, representing the pipe B B3 A third parameter representing the pipe B; f. of A Representing the actual fan frequency of the pipeline A; f. of B Representing the actual fan frequency for duct B.
4. The parallel fan balance control method according to claim 2, wherein in step 2, when the machine learning model cannot obtain the actual current of the pipeline, a fourth machine learning model is constructed by using a least square method:
wherein λ is A4 A fourth parameter, λ, representing the pipeline A A5 Denotes a fifth parameter, λ, of the conduit A B4 A fourth parameter, λ, representing the pipe B B5 Represents a fifth parameter of the pipeline B; f. of A Representing the actual fan frequency, f, of the pipeline A B Representing the actual fan frequency for duct B and minT to minimize the expression.
5. The parallel fan balance control method according to claim 3, wherein in step 3, the machine learning model corrects the observed data of the duct pressure by the actual current when the actual current of the duct can be obtained:
wherein the content of the first and second substances,
observed data representing the pressure of the conduit in conduit a,
observed data representing the pressure of the ducts in duct B.
6. The parallel fan balance control method of claim 4, wherein in step 3, the machine learning model corrects the observed number of the pair of wind channel pressures by parameters α and β when the actual current of the pipeline cannot be obtained:
wherein the content of the first and second substances,
observed data representing the pressure of the conduit in conduit a,
observed data representing the pressure of the ducts in duct B.
7. The parallel fan balance control method of claim 5, wherein in step 4, the machine learning model predicts a fan target frequency when the actual current of the pipeline can be obtained:
wherein, f' A Denotes the target Fan frequency, f 'of duct A' B Representing the target fan frequency for duct B.
8. The parallel fan balance control method of claim 6, wherein in step 4, a fan target frequency is predicted:
wherein Q is A ＝Q B ＝Q/2，f’ A Denotes the target Fan frequency, f 'of duct A' B Representing the target fan frequency for duct B.
9. The parallel fan balance control method according to claim 1, wherein in step 5, a target frequency of the parallel fan is predicted, a difference is made between the target frequency and a current actual parallel fan frequency, the fan with a larger absolute value of the difference starts to adjust to approach the target frequency, then the other fan follows to adjust, and the two fans are gradually adjusted in multiple steps according to configuration parameters to gradually reach the target frequency with equivalent output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210799091.4A CN115045854A (en) | 2022-07-06 | 2022-07-06 | Parallel fan balance control method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210799091.4A CN115045854A (en) | 2022-07-06 | 2022-07-06 | Parallel fan balance control method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115045854A true CN115045854A (en) | 2022-09-13 |
Family
ID=83166193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210799091.4A Pending CN115045854A (en) | 2022-07-06 | 2022-07-06 | Parallel fan balance control method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115045854A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116224795A (en) * | 2023-03-06 | 2023-06-06 | 北京全应科技有限公司 | Thermoelectric production equipment control method based on machine learning model |
-
2022
- 2022-07-06 CN CN202210799091.4A patent/CN115045854A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116224795A (en) * | 2023-03-06 | 2023-06-06 | 北京全应科技有限公司 | Thermoelectric production equipment control method based on machine learning model |
| CN116224795B (en) * | 2023-03-06 | 2023-11-17 | 北京全应科技有限公司 | Thermoelectric production equipment control method based on machine learning model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110513767B (en) | Heat supply network hydraulic balance regulation and control method and system based on resistance characteristics of heating power station | |
| CN109764243B (en) | Method for controlling fluid flow through a valve | |
| CN104501421B (en) | A kind of control method of frequency conversion two-stage compression heat pump water heater | |
| CN102207320B (en) | Control method of minimum opening of air conditioning expansion valves and air conditioning expansion valves | |
| CN109059217B (en) | Total air volume control method of variable air volume air conditioning system based on operation curve | |
| CN115045854A (en) | Parallel fan balance control method based on machine learning | |
| DK3179173T3 (en) | Method and system for automatic hydraulic balancing of consumers in a heating and / or cooling system | |
| EP2660527B1 (en) | Method for tuning a ventilation system | |
| CN107728685B (en) | Intelligent constant-temperature water control method | |
| CN115823706B (en) | Self-adaptive variable pressure difference energy-saving control system and method for primary pump | |
| CN110332656B (en) | Ventilation air-conditioning system of underground water power station factory building and operation control method thereof | |
| US20170003695A1 (en) | Method and apparatus for the energy-efficient operation of secondary dust removal systems | |
| JP6033674B2 (en) | Heat supply control device, heat supply system, and heat supply control method | |
| CN106086271A (en) | For the method reducing the energy efficiency of the blast furnace blower compressed air amount of leaking informaton | |
| JP6684453B2 (en) | Extraction control method and control device for steam turbine generator | |
| CN113606656A (en) | Model prediction-based heating system hydraulic balance adjusting method and system | |
| CN206207528U (en) | A.T.C, the heating system of moisturizing | |
| JP6938139B2 (en) | Steam turbine equipment | |
| CN113803121B (en) | Automatic control method and system for low-pressure steam turbine and power generation system | |
| CN109253492B (en) | Adjusting method for solving hydraulic imbalance of heat supply system | |
| Mu et al. | A novel air flowrate control method based on terminal damper opening prediction in multi-zone VAV system | |
| CN112424538A (en) | Method and system for controlling energy transfer of a thermal energy exchanger | |
| CN115789731A (en) | Control method, control device, system and storage medium of central smoke machine system | |
| CN105240993A (en) | Fine energy-saving control system of central air conditioner and achieving method of fine energy-saving control system | |
| EP2715213B1 (en) | Gas heating system for gas pressure reducing systems and method for obtaining said heating effect |
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 |