KR20190024032A - Energy optimization system through dyeing process control of machine learning base in dyeing processing factory - Google Patents

Abstract

The present invention relates to a system to optimize energy through dyeing process control based on machine learning in a dyeing processing factory and, more specifically, relates to a system to optimize energy through dyeing process control based on machine learning in a dyeing processing factory, capable of calculating dyeing process control information for reducing energy consumed for a dyeing process. The system includes: a plurality of sensors sensing environment information; a DTC controller controlling the operation of a dyeing machine; a site management server transmitting process control information of the dyeing machine of the DTC controller; and an analysis server conducting analysis based on machine learning.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an energy optimization system through a dyeing process control of a machine learning base in a salt processing plant,

The present invention relates to an energy optimization system through a machine learning based dyeing process control in a salt processing factory, and more particularly, to an energy optimization system using a machine learning based on sensing data in a dyeing process, The present invention relates to an energy optimization system using a machine learning based dyeing process control in a salt processing plant, which calculates dyeing process control information that can reduce input energy in a dyeing process by performing a learning process.

Korea is an energy-free, non-industrial industry with an energy consumption level that is three times higher than that of the world's advanced economies in terms of 10 years. In particular, in the second energy plan, As a goal.

The target for greenhouse gas reduction in Korea's textile sector is 6.3% of the 2020 emission estimate (BAU) target of 600,000 tons, with the salt processing industry accounting for 10.5% of the textile industry, with a target of about 60,000 tons (Growth Power Industry Research Center, 2011)

Among them, the dyeing industry is a representative energy-consuming industry, consuming 77% of the total textile industry fuel consumption and 54% of the electricity consumption, and the proportion of SMEs is 98%, which is much higher than other industries.

Since the dyeing industry accounts for 30-50% of the energy cost in the production cost, it is very advantageous to reduce the proportion of the input cost in order to secure industrial competitiveness.

In Korea, 'Ansan Smart Hub National Industrial Complex Cloud FEMS Establishment Project' (Ministry of Commerce, Industry and Energy, '14) conducted an energy saving experiment by analyzing factories in other industries rather than similar process analyzes. Since it is specialized in the unit business site, there are restrictions on securing and utilizing various data to be applied to the business sites of the dyeing industry.

In addition, the EMS system installed at POSCO Gwangyang Works' Oxygen Plant has limitations that can be applied to a single large-scale business site as a system for optimizing the system for energy saving while preventing problems in operation of oxygen production facilities.

As a result, proposals for an energy optimization system that is very specific to the dyeing industry, which has a high proportion of energy consumption and a high proportion of SMEs, is required to minimize production costs and to minimize environmental pollution caused by greenhouse gases.

SUMMARY OF THE INVENTION The present invention has been conceived to solve the above problems, and it is an object of the present invention to provide a machine for calculating process control information for continuously applying sensing data detected from a plurality of sensors measuring various environmental parameters in a dyeing process in real time, Based on learning-based learning, it is possible not only to control the process through the existing energy monitoring, but also to connect the energy and production variables, so that the process control can be more accurate and sensitive to changes in the production environment. The present invention has been made in view of the above problems.

The present invention has the following features in order to achieve the above object.

The present invention relates to a dyeing process through a dyeing machine, comprising: a plurality of sensors for sensing environmental information including electric power, water supply, and steam amount; A DTC controller receiving sensing data from the plurality of sensors and performing drive control of the dyeing machine; A field management server connected to the DTC controller so as to be capable of data communication and transmitting process control information of the dyeing machine of the DTC controller; And sensing data of a plurality of sensors sensed at the time of performing a process according to the process control information of the dyeing machine and the process control information of the dyeing machine from the DTC controller and receiving the fiber property information and production information specified in the dyeing process and dyeing the sensing data And an analysis server for performing a machine learning based analysis for calculating process control information that can optimize energy input to the process, wherein the analysis server analyzes previous process control information and sensing data And transmits the information to the field management server, so that the entire process of controlling the process of the dyeing machine in accordance with the update process control information in the field management server continues .

The plurality of sensors include a temperature sensor for measuring temperature, a smart power sensor for measuring power supplied, a water meter for measuring cold water or hot water, a water level sensor for measuring water level, a calorimeter for measuring steam amount, , A pressure sensor which is applied to a rapid dyeing machine and which measures nozzle pressure or body pressure, and an encoder or inverter which is applied to a jigger dyeing machine and which measures bubble speed, rolling number or tension.

In addition, the fiber property information, which is specified in the dyeing process in the analysis server, includes fabric type, weight, yarn diameter, yarn count, weaving type, and production information includes quantity, color, Wherein the field management server transmits process control information including temperature control, water supply, drainage, or door input to the DTC controller to perform control decision making, and generates production information including operation details of the dyeing machine, The DTC controller transmits the sensing data received from the plurality of sensors to the middleware unit and the middleware unit transmits the sensing data to the analysis server .

In addition, the analysis server outputs the process control information for optimizing the energy input to the dyeing process by learning the machine learning based analysis on the neural network based on the sensing data transmitted in real time, The input variables of the artificial neural network are composed of yard, weight per yard, dye number, fabric name, number of stairs, water saving, weaving, and fabric weight. Output variables are nozzle pressure, main pump speed, steam amount, dye amount, The temperature rise time, the temperature fall time, the temperature fall time, the maximum temperature, and the maximum temperature holding time, and the intermediate layer is composed of two places so that learning is performed.

According to the present invention, more accurate energy optimization can be realized as machine learning based process control information is calculated through interconnection between energy and production variables rather than process control method through analysis of existing energy monitoring.

In addition, since the analysis server receives the continuous sensor data, the instant update process control information is calculated and reflected in the control of the real time dyeing machine, it becomes possible to optimize the energy sensitive to the production environment change.

1 is a diagram showing a schematic configuration of an energy optimization system according to an embodiment of the present invention.
2 is a flow chart illustrating a process control flow of an energy optimization system according to an embodiment of the present invention.
3 is a diagram illustrating a learning process of an artificial neural network according to an embodiment of the present invention.
FIG. 4 is a diagram showing a schematic configuration of an energy optimization system according to another embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

First, the terminology used in the present application is used only to describe a specific embodiment, and is not intended to limit the present invention, and the singular expressions may include plural expressions unless the context clearly indicates otherwise. Also, in this application, the terms "comprise", "having", and the like are intended to specify that there are stated features, integers, steps, operations, elements, parts or combinations thereof, But do not preclude the presence or addition of features, numbers, steps, operations, components, parts, or combinations thereof.

In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

2 is a flowchart showing a process control flow of an energy optimization system according to an embodiment of the present invention. FIG. 3 is a flow chart of a process flow of the energy optimization system according to an embodiment of the present invention. 1 is a diagram illustrating a learning process of an artificial neural network according to an embodiment of the present invention.

Referring to the drawings, an energy optimization system 1000 according to an exemplary embodiment of the present invention includes a plurality of sensors 100 for sensing environmental information including power, water supply, and steam in a dyeing process through a dyeing machine, A DTC controller 200 receiving sensing data from the sensors 100 and performing driving control of the dyeing machine and sensing data from the DTC controller 200 connected to the DTC controller 200 so as to be able to communicate with the DTC controller 200 A field management server 300 for transmitting process control information of a dyeing machine of the DTC controller 200 and a plurality of sensors sensed at the time of performing a process according to the process control information of the dyeing machine and the process control information from the DTC controller 200 The sensing data of the sensor 100 and the fiber property information and the production information specified in the dyeing process and the sensing data are used as the energy of the energy Analysis comprises a server 400 that performs a machine learning based on the analysis for the calculation of the process control information in an optimization can be made.

Here, the plurality of sensors 100 are provided to detect various environmental variables measured in the dyeing process and utilize them for energy optimization analysis. The plurality of sensors 100 may be different depending on the characteristics of the dyeing machine .

For example, a temperature sensor for measuring temperature, a smart power sensor for measuring power supplied, a water meter for measuring cold water or hot water, a water level sensor for measuring water level, a calorimeter for measuring steam amount, Pressure sensors for measuring nozzle pressure or body pressure, and encoders or inverters for bubble velocity, rolling number or tensile force, which are applied to a jigger dyeing machine.

The sensors 100 may be classified into a sensor that is collected on a digital contact basis and a sensor that is collected on an analog change basis. In the former case, a water meter, a smart power sensor, or a calorimeter may be used. In the latter case, , A level sensor or a pressure sensor may be applied.

On the other hand, the DTC controller 200 is provided at one side of the dyeing machine to perform driving control of the dyeing machine. The DTC controller 200 is equipped with a display unit such as an LCD to monitor the operator, (Water supply, drainage, and various valves) as well as various controls such as dye, preparation metering and transfer input, and can be configured to perform water supply, temperature control, water washing, drainage, Process command information to control the entire dyeing process such as reel speed control.

For example, it has process command information according to various working patterns according to salt method such as completion salt method, basic salt method or combination salt method, multiplexes analog channel into 8 channels of input and 8 channels of output to control temperature, auxiliary tank temperature, , A dosing output, a quantity value input, and the like.

The DTC controller 200 receives sensing data from the plurality of sensors 100 and transmits the sensed data to the middleware unit 310. The analysis server 400 performs analysis through learning using sensing data Respectively.

Meanwhile, the field management server 300 is connected to the DTC controller 200 and an analysis server 400 to be described later so as to be able to communicate data, and transmits the process control information received from the analysis server 400 to the DTC controller 200, Of the process control information.

The field management server 300 transmits process control information including temperature control, water supply, drainage, or door input to the DTC controller 200 to perform control decision making, and controls the operation history of the dyeing machine, Production monitoring information including status information, schedule management information, input / output management information, and material management information are calculated and stored.

In addition, the field management server 300 receives sensing data of the plurality of sensors 100 received by the DTC controller 200 through the DTC controller 200 or the middleware unit 310. Here, the middleware unit 310 Receives the sensing data from the DTC controller 200, and transmits the sensed data to the analysis server 400.

Of course, the field management server 300 may be configured to receive the sensing data through the analysis server 400 without receiving the sensing data from the DTC controller 200 or the middleware unit 310.

In addition, the field management server 300 can be connected to an external server or a terminal through a wired / wireless communication network. The field management server 300 can generate production monitoring information, schedule management information, input / output management information And material management information, and can be provided and monitored.

Meanwhile, the analysis server 400 receives the sensing data of the plurality of sensors 100, which are sensed at the time of performing the process according to the process control information of the dyeing machine and the process control information, from the DTC controller 200, Based analysis of the fiber property information and production information to be specified and the process control information for optimizing the energy input to the dyeing process of the sensing data.

The analysis server 400 analyzes the previous process control information and sensing data of the sensors 100 according to the previous process control information to calculate update process control information that can reduce the energy applied to the dyeing process, The field management server 300 repeatedly performs the entire process of performing the process control of the dyeing machine in accordance with the update process control information.

That is, the process control information condition of the accumulated big data type and the change of the sensing data and the information of the currently specified order (fiber property information and production information) are inputted into the dyeing process through the learning process based on the annealing process And to calculate process control information capable of optimizing energy.

The order information is specified before the dyeing process and corresponds to the fiber property information and the production information. The fiber property information may include the kind of fabrics, the weight, the diameter of the yarn, the number of occupants, the weaving pattern, Quantity, color, reclamation, and job classification.

In addition, the analysis server 400 may perform various analyzes based on the real-time sensing data received from the plurality of sensors 100 and the real-time machine-specific process control information of the DTC controller 200. For example, It is possible to automatically calculate the color classification method such as color, color, and color. It can automatically convert the toe value through the amount of electricity and steam, and can generate the curve of the dwelling curve pattern and the graph of the energy pattern.

In addition, it is possible to generate the analysis value of the nucleation curve pattern information by the regression analysis method.

Also, the analysis server 400 can output the process control information for optimizing the energy input to the dyeing process by learning the analysis of the machine learning based on the sensing data transmitted in real time to the neural network. The input variables of the artificial neural network are composed of yard, weight per yard, dye number, fabric name, number of stairs, water saving, weaving, and fabric weight, and output variables are nozzle pressure, main pump speed, The dye amount, the temperature rising speed, the temperature rising time, the temperature falling speed, the temperature falling time, the maximum temperature, and the maximum temperature holding time, and the intermediate layer is composed of two places.

A generation process of input variables and output variables specified in the artificial neural network will be described.

First, F-test was conducted to confirm the significant variables of energy consumption among the order data collected during the ordering and the field data collected through the sensor during the field work. As a result, 17 variables are considered as shown in [Table 1] , 8 variables were acquired at the time of ordering and 9 variables were found to be variables that can be collected during field work.

Analysis of variance on energy use Variable name Degree of freedom Mean squared F-Value Probability of significance (> F) Remarks Fabric weight One 265574 5.876e + 02 <2e-16 *** Fabric name One 5591425 1.237e + 04 <2e-16 *** Waterproof One 26173297 5.791e + 04 <2e-16 *** Weft number One 11490901 2.542e + 04 <2e-16 *** Weight per yard One 7360919 1.629e + 04 <2e-16 *** yard One 151239 3.346e + 02 <2e-16 *** Weaving One 638201 1.412e + 03 <2e-16 *** Dye number One 15890845 3.516e + 04 <2e-16 *** Dye amount One 8909421 1.971e + 04 <2e-16 *** Reiteration One 921199 2.038e + 03 <2e-16 *** Temperature rise speed One 50510310 1.118e + 05 <2e-16 *** Temperature rise time One 3023445 6.690e + 03 <2e-16 *** Maximum temperature One 12300115 2.721e + 04 <2e-16 *** Maximum temperature holding time One 24818315 5.497e + 04 <2e-16 *** Temperature falling speed One 69236950 1.532e + 05 <2e-16 *** Temperature fall time One 68941897 1.525e + 05 <2e-16 *** working time One 213 4.170e-01 0.4976 Water volume (cold water) One 0 1.000e-03 0.9793 Water volume (hot water) One 47 1.040e-01 0.7490 Steam amount One 2333 5.162e + 00 0.0302 * Main pump speed One 113 2.500e-01 0.6206 Reel speed (Right) One 218 4.830e-01 0.4926 Reel speed (left) One 1187 2.626e + 00 0.1152 Nozzle pressure One 1448 3.203e + 00 0.0833 Residual 31 452

As a result of F-test, it was confirmed that there are 7 data related to the re-emission of the field data in total. Of the six parameters, the maximum temperature is 130C at the highest temperature, as in the case of polyester dyeing. Therefore, unlike the energy consumption, it can not affect the re-crystallization.

Analysis of variance for reiteration Variable name Degree of freedom Mean squared F-Value Probability of significance (> F) Remarks Dye amount One 0.03619 6.905 <2e-16 * Temperature rise speed One 0.31025 59.2 2.09e-09 *** Temperature rise time One 0.03014 5.752 0.021228 * Maximum temperature One 0.00026 0.050 0.824113 Maximum temperature holding time One 0.09947 18.980 8.95e-05 *** Temperature falling speed One 0.07362 14.048 0.000563 *** Temperature fall time One 0.00755 1.441 0.237088 working time One 0.00721 1.375 0.247885 Water volume (cold water) One 0.00007 0.013 0.908670 Water volume (hot water) One 0.00062 0.119 0.731924 Steam amount One 0.01327 2.532 0.119405 Main pump speed One 0.10091 19.254 8.12e-05 *** Reel speed (Right) One 0.00618 1.179 0.284098 Reel speed (left) One 0.00006 0.012 0.913388 Nozzle pressure One 0.08671 16.545 0.000217 *** Residual 40 0.00524

Therefore, we derive the components of the model in order to specify the parameters based on the machine learning model using field data. The input factors are variables such as yard and yard weight, salt number, , And the output element is set as a significant variable among the parameters required for the process control specification when the nozzle pressure, the main pump speed, the steam amount, Respectively.

Accordingly, we derived the components of the machine learning model so as to indicate the process control information for the new order by inputting 8 variables which are generated during the new order and output 10 variables.

Deriving Components of the Machine Learning Model No Input variable Output variable One yard Nozzle pressure 2 Weight per yard Main pump speed 3 Dye number Steam amount 4 Fabric name Dye amount 5 Weft number Temperature rise speed 6 Waterproof Temperature rise time 7 Weaving (pattern) Temperature falling speed 8 Fabric weight Temperature fall time 9 Maximum temperature 10 Maximum temperature holding time

According to one embodiment of the present invention, the input variables and the output variables for learning are composed of yard, weight per yard, dye number, fabric name, number of yarns, water saving, weaving, The parameters were pressure, main pump speed, steam amount, dye amount, temperature rise speed, temperature rise time, temperature fall speed, temperature fall time, peak temperature, And the output value is converted into the actual value by post-processing.

At this time, the data for learning was selected by learning only data that did not cause re-salting. The threshold value was set to 0.005, the hidden layer to 2 places, and the neuron to learn (8,8). FIG. 3 shows a model of the artificial neural network in which learning has been completed.

Therefore, based on the weight and bias of the model, the process control information is calculated based on the input variable of the artificial neural network.

As shown in FIG. 2, the process control flow of the optimization system 1000 according to an exemplary embodiment of the present invention will be described. First, sensing data is detected by a plurality of sensors 100 during a dyeing process using a dyeing machine (S10 The corresponding sensing data is transmitted from the DTC controller 200 to the analysis server 400 via the middleware unit 310 (S20).

Accordingly, the analysis server 400 calculates the optimization process control information through the sensing data and the order information, and transmits the optimization process control information to the site management server 300 (S30), and the site management server 300 transmits the process control information to the DTC controller 200 (S40). The DTC controller 200 performs individual control of the dyeing machine through the process control information received from the field management server 300 (S50).

FIG. 4 is a diagram showing a schematic configuration of an energy optimization system according to another embodiment of the present invention.

Referring to the drawings, an energy optimization system 1000 according to an embodiment of the present invention includes a plurality of DTC controllers 200 attached to a plurality of site management servers 300 and respective site management servers 300 in one analysis server 400, And a plurality of sensors 100.

In the case where the DTC controller 200 and the plurality of sensors 100 are provided by providing a plurality of dyeing machines for each of the dyeing factories, Calculates corresponding process control information for each of a plurality of dyeing factories, and transmits the process control information to the field management server 300.

As described above, the analysis server 400 according to the present invention generates the process control information for optimizing the energy inputted through the sensing data and the order information delivered in real time through a learning process based on the machine learning, The process control is performed in real time, and the plurality of sensors 100 are sensed again so that the sensing data can be applied when the process control information needs to be updated through the analysis server 400, It becomes possible to respond sensitively to environmental changes at the work site.

If necessary, the field management server 300 directly transmits the process control information transmitted from the analysis server 400 to the manager of the site management server 300 as the process control information received in consideration of the previous process control information and the field conditions. It may be configured to determine whether to apply the current process control information to the current process control information or to apply the current process control information to the current process control information,

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the scope of the present invention .

Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

100: a plurality of sensors 200: DTC controller
300: field management server 310: middleware department
400: Analysis Server 1000: Energy Optimization System

Claims (6)

A plurality of sensors 100 for sensing environmental information including power, water supply, and steam in a dyeing process through a dyeing machine;
A DTC controller 200 receiving sensing data from the plurality of sensors 100 and performing driving control of the dyeing machine;
A field management server (300) connected to the DTC controller (200) for data communication and transmitting process control information of the dyeing machine of the DTC controller (200); And
The DTC controller 200 receives the sensing data of the plurality of sensors 100 sensed at the time of performing the process according to the process control information of the dyeing machine and the process control information of the dyeing machine and receives the fiber property information and the production information And an analysis server (400) for performing a machine learning based analysis for calculating process control information capable of optimizing the energy input to the dyeing process of the sensing data,
The analysis server 400 analyzes the previous process control information and sensing data of the sensors 100 according to the previous process control information to calculate update process control information that can reduce the energy applied to the dyeing process, And the process control of the dyeing machine is continuously repeated in accordance with the update process control information in the field management server 300. In the dyeing process control based on the machine learning in the salt processing factory, Energy optimization system.
The method according to claim 1,
The plurality of sensors (100)
A smart power sensor for measuring the supplied power, a water meter for measuring cold water or hot water, a water level sensor for measuring the water level, a calorimeter for measuring the steam amount, and a rapid dyeing machine A pressure sensor for measuring a nozzle pressure or a body pressure, and an encoder or inverter which is applied to a Ziegler stainer and measures bubble velocity, rolling number or tension, and a machine learning based dyeing process control Energy optimization system.
3. The method of claim 2,
The fiber property information specified in the dyeing process in the analysis server 400 includes fabric type, weight, yarn size, number of yarns, weaving type, and the production information includes quantity, color, Based on dyeing process control in a salt processing plant.
The method according to claim 1,
The field management server (300)
The process control information including temperature control, water supply, drainage, or door input is transmitted to the DTC controller 200 to perform control decision, and production monitoring information including the operation history of the dyeing machine, The DTC controller 200 transmits the sensing data received from the plurality of sensors 100 to the middleware unit 310 and transmits the sensing data to the middleware unit 310. [ 310) is configured to deliver it to an analysis server (400). An energy optimization system through a machine learning based dyeing process control in a salt processing plant.
5. The method of claim 4,
The analysis server 400
A machine learning based analysis in a salt processing factory, characterized by outputting process control information for energy injection into a dyeing process by learning the analysis of the machine learning based on the sensed data transmitted in real time to the neural network Energy optimization system through process control. 6. The method of claim 5,
The input parameters of the artificial neural network in learning through the neural network are composed of yard, weight per yard, dye number, fabric name, number of yarns, water saving, weaving, , A steam amount, a dye amount, a temperature rising speed, a temperature rising time, a temperature falling speed, a temperature falling time, a maximum temperature and a maximum temperature holding time, Energy Optimization System through Machine Learning Based Dyeing Process Control in Processing Plant.

Images