KR102311857B1 - Prediction System for preheating time of gas turbine - Google Patents

Abstract

The present invention relates to a gas turbine preheating time prediction system capable of efficiently managing a gas turbine by predicting the preheating time of the gas turbine.
In one example, a server for predicting the preheating time of the gas turbine; and an operator terminal connected to the server through a wired communication network and receiving the preheating time predicted by the server, wherein the server includes a data collection unit configured to collect pipe temperature data and outdoor temperature data of a gas turbine; Disclosed is a gas turbine preheating time prediction system, characterized in that it comprises a data analysis unit for predicting the preheating time by analyzing the data collected by the data collection unit.

Description

Prediction System for preheating time of gas turbine

The present invention relates to a gas turbine preheating time prediction system.

A combined cycle power plant operates a gas turbine repeatedly with one cycle of starting and stopping. In such a combined cycle power plant, the start time of the gas turbine is planned in advance, and preheating is started so that the gas turbine is normally started at the planned time. On the other hand, in the case of a combined cycle power plant in operation, there is no system for predicting the time consumed for preheating the gas turbine, and the driver prepares for starting by predicting the preheating time empirically. In this way, the time for starting preheating before starting the gas turbine starts before the maximum time consumed for preheating and continues to maintain preheating until the planned start time. Accordingly, since preheating is continued from the time when the preheating is completed to the actual start time, unnecessary fuel resources are consumed. Accordingly, there is a need for a system capable of efficiently managing the preheating time of a gas turbine in a combined cycle power plant.

The present invention provides a gas turbine preheating time prediction system capable of efficiently managing a gas turbine by predicting the preheating time of the gas turbine.

A gas turbine preheating time prediction system according to the present invention includes: a server for predicting a gas turbine preheating time; and an operator terminal connected to the server through a wired communication network and receiving the preheating time predicted by the server, wherein the server includes a data collection unit configured to collect pipe temperature data and outdoor temperature data of a gas turbine; It may include a data analysis unit for predicting the preheating time by analyzing the data collected by the data collection unit.

The data analyzer may predict the preheating time of the gas turbine by using a coefficient of variation including the outside air temperature, humidity, and temperature of each pipe of the gas turbine.

The data analyzer may predict the preheating time of the gas turbine by analyzing the correlation between the coefficients of variation using a Projection to Latent Structures, Partial Least Square (PLS) technique.

The data analyzer may classify and analyze a cold start or a warm start according to a time from when the start of the gas turbine ends to the start of preheating again.

The data analyzer may measure an error by comparing the predicted preheating time with the actual preheating time, and analyze the error.

The operator terminal may further include a monitor for displaying the predicted time transmitted from the server on the screen to the operator.

The server may provide the operator with information for each section of the preheating time predicted in real time, the manual prediction time, and the preheating time performed in real time through the monitor.

The gas turbine preheating time prediction system according to an embodiment of the present invention can predict the preheating time of a gas turbine by using big data on the pipe temperature data and meteorological data of the gas turbine. can be improved

In addition, the gas turbine preheating time prediction system according to an embodiment of the present invention can improve the reliability of the prediction system by analyzing the error between the predicted time and the actual required time, and finding the cause of the error and solving it.

1 is a schematic diagram illustrating a gas turbine preheating time prediction system according to an embodiment of the present invention.
2 is a block diagram illustrating a gas turbine preheating time prediction system according to an embodiment of the present invention.
3 is a diagram illustrating a data pre-processing method during predictive analysis in the gas turbine preheating time prediction system according to the present invention.
4 is a graph for explaining the verification of the prediction technique used in the gas turbine preheating time prediction system according to the present invention.
5 is a diagram illustrating a screen capture displayed on a monitor of an operator terminal.
6 is a graph comparing the time predicted by the server and the time actually spent.
7 is a chart showing the contribution to the error between the predicted time and the actual required time.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Examples of the present invention are provided to more completely explain the present invention to those of ordinary skill in the art, and the following examples may be modified in various other forms, and the scope of the present invention is as follows It is not limited to an Example. Rather, these examples are provided so that this disclosure will be more thorough and complete, and will fully convey the spirit of the invention to those skilled in the art. Also, like reference numerals in the drawings refer to like elements. As used herein, the term “and/or” includes any one and all combinations of one or more of those listed items.

1 is a schematic diagram illustrating a gas turbine preheating time prediction system according to an embodiment of the present invention. 2 is a block diagram illustrating a gas turbine preheating time prediction system according to an embodiment of the present invention.

1 and 2 , the gas turbine preheating time prediction system 100 according to an embodiment of the present invention connects the server 110 , the operator terminal 130 , and the server 110 and the operator terminal 130 . and a wired communication network 120 for exchanging information. The present invention is a prediction system for predicting the preheating time of the gas turbine (10) among power generation facilities. In particular, the gas turbine preheating time prediction system 100 according to an embodiment of the present invention performs multivariate analysis related to the gas turbine preheating time by utilizing big data on the pipe temperature data and weather data of the gas turbine 10 . can

The server 110 may transmit information on the predicted preheating completion time, the previously consumed preheating time, and the preheating time required for each step to the operator terminal 130 through the wired communication network 120 .

The wired communication network 120 may be implemented as a network communication network, and data communication may be performed according to an Internet protocol such as Transmission Control Protocol/Internet Protocol (TCP/IP). The wired communication network 120 may transmit data between the server 110 and the operator terminal 130 .

The operator terminal 130 is a terminal capable of arithmetic, and may be configured in plurality by being assigned to each operator. The operator terminal 130 may include a first terminal to an n-th terminal. For example, although the operator terminal 130 is illustrated as a desktop PC in the drawing, the operator terminal 130 may include a notebook computer. The operator terminal 130 may receive information on the predicted preheating completion time, the previously consumed preheating time, and the preheating time required for each step from the server 110 in real time and display it on the monitor 131 . Accordingly, the operator can efficiently operate the gas turbine 10 by checking the information displayed on the operator terminal 130 . Also, the operator terminal 130 may further include an input unit 132 for receiving an operator's input. For example, the input unit 132 may be formed of a keyboard, a mouse, or the like. The operator may input information for prediction through the input unit 132 . Specifically, the operator may input the temperature of the gas turbine pipe and the outside air temperature expected at the predetermined preheating start time through the input unit 132 . Also, information input through the input unit 132 may be transmitted to the server 110 through the wired communication network 120 .

The server 110 is a database having a memory such as a hard disk drive (HDD), a solid state drive (SSD), and the like. The server 110 collects information on the preheating time of the gas turbine 10 , analyzes it, and transmits the analysis result to the operator terminal 130 . In terms of hardware, the server 110 has the same configuration as a typical web server, and in terms of software, it is implemented through various types of languages such as C, C++, Java, Visual Basic, Visual C, etc. to perform various functions. It may contain program modules. The server 110 includes a data collection unit 111 and a data analysis unit 112 .

The data collection unit 111 receives and stores information of the gas turbine 10 from a plurality of sensors. For example, the data collection unit 111 may receive information measured by the sensor from a Distributed Control System (DCS) through an OLE for Process Control (OPC) communication method and store various data. In addition, the data collection unit 111 may receive data such as a temperature of a gas turbine pipe, an outdoor temperature, and a humidity from a plurality of sensors.

The data analysis unit 112 performs predictive analysis of predicting the preheating completion time of the gas turbine 10 by using the data stored in the data collection unit 111 . Specifically, the data analysis unit 112 performs predictive analysis to predict the time it takes to preheat the gas turbine 10 by using the coefficient of variation including the temperature, outside air temperature, humidity, etc. of each gas turbine pipe. can

For example, the data analysis unit 112 may predict a preheating completion time required to preheat the gas turbine 10 through a multivariate analysis technique. The multivariate analysis technique used in the present invention uses a Projection to Latent Structures, Partial Least Square (PLS) technique. The basic idea of the PLS technique is to analyze the correlation of a large number of variables by dimensional reduction. Dimensional reduction is an algorithm that reduces the number of variables by linearly combining variables in a form that minimizes information loss and facilitates analysis by analysts. Reducing the number of variables here is not removing the variables, but deriving a new variable, that is, a new axis through linear combination. This analysis concept, dimensionality reduction, is similar to PCA (Principal Component Analysis) technique. Both PLS and PCA techniques analyze correlations between variables and reduce dimensions, but the PLS technique differs from the PCA technique in that it specifies an objective variable and predicts the value of the objective variable.

In the PLS technique, correlation analysis is an analysis of the linear correlation between each variable. The linear correlation is a correlation in which the change in the value of one variable is linearly drawn according to the change in the value of one variable when a graph is drawn with each variable as an individual coordinate axis. For example, if the value of one variable tends to increase when the value of the other variable also increases, it is said to have a positive correlation, and vice versa, it is said to have a negative correlation. Variables used in PLS modeling are first analyzed for correlation between variables through cross-correlation analysis. The analysis value has a value between -1 and 1, and the closer the absolute value of the value is to 1, the stronger the correlation. In the present invention, the variables to be used in the analysis of the PLS technique are modeled as variables having a correlation of 0.4 or more.

The PLS technique is similar to regression, which is a representative technique for predicting the value of the objective variable in terms of the modeling purpose. Regression analysis is the most commonly used statistical analysis technique to predict the value of an objective variable. However, there is a problem that occurs when a predictive model is made with data from field facilities, and one of them is multicollinearity. In the case of regression analysis, an important prerequisite of the analysis is to assume that there is little or no correlation between the predictors (or response variables). However, in most cases, there is a high correlation between data from field facilities. Because of this, multicollinearity problems arise in the case of regression analysis, which increases the variance of the regression coefficients, making the model unstable and difficult to interpret. On the other hand, the PLS method performs modeling on the premise that there is a correlation between predictor variables. Therefore, since the PLS method does not have problems such as multicollinearity that occurs in regression analysis, it is more suitable than regression analysis for data modeling of field facilities in terms of correlation analysis between each data and predicting the value of the objective variable through this correlation. do.

Accordingly, the data analysis unit 112 performs modeling through correlation analysis based on data of a section normally operated according to the PLS technique. Here, in order for the data analysis unit 112 to perform modeling, it is very important to collect data of a normally operated section. This is because, even in the same normal section of the collected data, if each driving condition is different, the prediction error can be reduced by dividing the data into different models. In the current combined cycle power plant, when preheating a gas turbine, the external conditions that have a large influence other than the pipe temperature and the outside air temperature are the time from the end of the previous power generation start to the start of preheating again. Therefore, in the present invention, the data analysis unit 112 divided into cold start/warm start based on the time difference from the end of the previous generation to the start time of preheating, and each model was constructed. This is a classification in the modeling construction phase, and when the data analysis unit 112 provides prediction information to the monitor 131 of the operator terminal 130, the time predicted through the condition model is automatically calculated according to the input condition. to provide.

3 is a diagram illustrating a data pre-processing method during predictive analysis in the gas turbine preheating time prediction system according to the present invention.

Data used by the data analysis unit 112 for modeling in predictive analysis is pre-processed by unit variance scaling as shown in FIG. 3 . This is that the data analysis unit 112 pre-processes so that the mean is 0 and the standard deviation becomes 1 in order to minimize the error according to the range of values between each predictive variable, and the variation for each variable is converted to the same standard. . The data analysis unit 112 may minimize the prediction error of modeling that occurs due to the difference in the data range through this pre-processing process.

4 is a graph for explaining the verification of the prediction technique used in the gas turbine preheating time prediction system according to the present invention.

In addition, the verification of the predictive power of the model of the prediction technique used by the data analysis unit 112 is confirmed by the R-squared value as shown in FIG. 4 . It can be said that the closer the R-squared value is to 1, the higher the predictive power for the data used in the current model. However, the higher the R-squared value, the less the model has the optimal predictive power. If the R-squared value is too high, an overfitting problem may occur, and the predictive power of the current data is high, but the predictive power of the future data is rather poor. Therefore, in the verification of the predictive power, the data analysis unit 112 determines that the R-squared value has a minimum reference value, and if the R-squared value is greater than or equal to the R-squared value, the predictive power of the model is meaningful.

5 is a diagram illustrating a screen capture displayed on a monitor of an operator terminal.

Referring to FIG. 5 , the result predicted by the server 110 is transmitted to the operator terminal 130 and displayed on the monitor 131 . Specifically, the preheating time predicted by the server 110 is displayed on the upper part of the monitor 131 in FIG. 5 . Here, the prediction information includes a real-time prediction time at which the server 110 predicts the pre-heating time required by the server 110 in real time when the preheating of the gas turbine 10 starts, and the operator inputs information through the input unit 132 of the operator terminal 130 . It may include a manual prediction time to predict. Here, the operator can check the manual prediction time by inputting information such as the pipe temperature of the gas turbine 10 and the outside air temperature to the input unit 132 . In addition, after the operator inputs information through the input unit 132 , the predicted past history can be inquired through the monitor 131 , and if there is wrongly inputted information, it can be corrected immediately. On the lower part of the monitor 131, the preheating time actually required up to the last 4 times is displayed in detail for each step. Through this, the operator can easily find out which step in which round the problem occurred.

6 is a graph comparing the time predicted by the server with the time actually spent.

As shown in FIG. 6 , the data analysis unit 112 may compare the predicted time with the actual time and display it on the monitor 131 of the operator terminal 130 . Also, referring to FIG. 6 , it can be seen that the error of the prediction time of the gas turbine preheating time prediction system 100 according to the present invention satisfies the reliability of the 2 sigma level. If the error between the predicted time of the gas turbine preheating time prediction system 100 and the actual required time increases, the data analysis unit 112 may perform additional analysis.

7 is a chart showing the contribution to the error between the predicted time and the actual required time.

As shown in FIG. 7 , the data analysis unit 112 may perform error analysis and display it on the monitor 131 of the operator terminal 130 . The data analysis unit 112 may show the operator in the order of greatest contribution to the error generation through error analysis. In other words, the data analysis unit 112 may provide the difference to the operator through the monitor 131 by displaying the difference in the form of a bar graph in the order of showing the most different tendencies compared to the normal driving state. This allows the operator to identify suspected problem areas and take action. For example, if the temperature of pipe A records a much lower temperature than usual through the error analysis as described above, the cause of the problem may be identified by performing an inspection on pipe A. Accordingly, the gas turbine preheating time prediction system 100 according to the present invention can easily find and solve problems occurring during preheating through the detailed analysis of the data analysis unit 112 .

What has been described above is only one embodiment for implementing the gas turbine preheating time prediction system according to the present invention, and the present invention is not limited to the above embodiment, and as claimed in the claims below, the present invention Without departing from the scope of the present invention, it will be said that the technical spirit of the present invention exists to the extent that various modifications can be made by anyone with ordinary knowledge in the field to which the invention pertains.

100: gas turbine warm-up time prediction system 10: server
111: data collection unit 112: data analysis unit
120: wired communication network 130: operator terminal
131: monitor 132: input unit

Claims (7)

a server for estimating the preheating time of the gas turbine; and
and an operator terminal connected to the server through a wired communication network and receiving the predicted warm-up time from the server;
The server is characterized in that it comprises a data collection unit for collecting pipe temperature data and outdoor temperature data of the gas turbine, and a data analysis unit for predicting a preheating time by analyzing the data collected by the data collection unit,
The data analysis unit is characterized in that it divides and analyzes a cold start or a warm start according to the time from the start of the gas turbine to the start of preheating again,
The data analysis unit is a gas turbine preheating time prediction system for preprocessing the data used for modeling in the preheating time prediction analysis by unit variance scaling. The method of claim 1,
The data analysis unit predicts the preheating time of the gas turbine by using the coefficient of variation including the outside temperature, humidity, and temperature of each pipe of the gas turbine. 3. The method of claim 2,
The data analysis unit predicts the preheating time of the gas turbine by analyzing the correlation between the coefficients of variation using a Projection to Latent Structures, Partial Least Square (PLS) technique. delete The method of claim 1,
The data analyzer compares the predicted preheating time with the actual preheating time, measures an error, and analyzes the error. The method of claim 1,
The operator terminal is a gas turbine preheating time prediction system, characterized in that it further comprises a monitor for displaying the predicted time transmitted from the server on the screen to the operator. 7. The method of claim 6,
The server is a gas turbine preheating time prediction system, characterized in that for providing information for each section of the preheating time, manual prediction time, and preheating time predicted in real time through the monitor.

Images