JPS6394006A - Starting device for power generation plant - Google Patents

Abstract

PURPOSE:To make it possible shorten the required time for starting the captioned plant by optimizing the starting schedule through the repetitive calculation method under the estimation of the starting characteristic obtained in a dynamic characteristic model, thus permitting the operation restriction condition to be satisfied. CONSTITUTION:A starting schedule preparing means 1000 quantitatively calculates the starting characteristic in a dynamic characteristic model 100 according to the standard schedule in the starting mode selected in response to the stopping time of a power generation plant 2000. In order to increase the estimation precision, an initial value estimating means 200 estimates the initial value based on the difference between the standard cooling characteristic and the actual cooling state through the application of the fuzzy estimation 500. A schedule optimizing means 400 optimizes the starting schedule through the repetitive calculation method, evaluating the starting characteristic, in order to permit the starting in the shortest time and satisfy the operation restriction condition by using the initial schedule of a fundamental schedule preparing means 300. The starting control 3000 is thus performed. Therefore, the plant can be safely started in the shortest time.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a startup device for a power generation plant, and particularly to an initial value prediction method suitable for predicting startup characteristics with high accuracy, and to improving convergence of startup schedule optimization calculations. This invention relates to a method for creating a suitable initial schedule.

[Conventional technology]

The conventional method for starting up a thermal power plant is based on the initial amount of fuel input to the boiler, the time function of main steam temperature and pressure rise, the turbine speed increase and A method has been adopted in which a time function of the load increase is determined as a startup schedule, and this startup schedule is executed by a control system installed in each system of the plant.

The second most typical method is Electrical Wo
rld.

Vofl・, 165. &6 paper “Thermal 5
tressInfluence Starting,
Loading of BoilersTurbine
This method determines the startup schedule uniquely based on the initial conditions of a limited portion of the plant.In other words, the startup schedule is determined based on the initial conditions of a limited portion of the plant. This method determines the steam turbine speed increase rate, initial load amount, speed maintenance, and load maintenance time and load change rate of the steam turbine according to the initial values. Since the temperature rise characteristics of the boiler-generated steam, which is important for managing the thermal stress of the turbine, cannot be predicted before startup, this uncertainty is absorbed by allowing some leeway in the startup schedule.

Therefore, the created startup schedule tends to be longer than necessary.

Another conventional method is US P3,446.
No. 224 and US P 4,228,359 are known. These methods attempt to quickly start up a steam turbine while monitoring the thermal stress generated in the steam turbine online in real time, but, like the conventional methods described above, there is no mention of a method for starting the boiler.

As a conventional method aimed at shortening the start-up time of a boiler, Japanese Patent Application Laid-Open No. 157402/1984 is known. This method aims to rapidly raise the temperature of the steam generated by the boiler while monitoring the thermal stress generated in the boiler online in real time. However, this method makes no mention of starting the steam turbine.

The startup time for the entire plant can be shortened by coordinating the boiler and steam turbine, but the conventional methods described above are all rapid startup methods that focus on only one side of the boiler or steam turbine. Even if individual methods are combined, there is no guarantee that the start-up time of the entire plant will be the shortest.

This is because the boiler and the steam turbine have extremely strong mutual interference, and individual optimization does not necessarily result in overall optimization.

Furthermore, in the conventional method described above, the startup schedule is created based on the actual measured value of the initial temperature layer of the plant immediately before the boiler ignition, so the startup schedule is created at an arbitrary time before the boiler ignition, and the central supply/power supply It was not possible to create a startup schedule that would allow the startup to be completed accurately at the startup completion time specified by the command center (hereinafter referred to as "Nakagyo") (hereinafter referred to as "scheduled startup").

In this way, the conventional method tends to take a long time to start up, and as a result, energy loss during startup (hereinafter referred to as startup loss) increases, which is an economical problem in plant operation. There was a safety issue in executing the schedule.

[Problem that the invention seeks to solve]

An object of the present invention is to solve the problem of predicting the initial state of the plant at startup in the conventional method by utilizing the knowledge of the cooling characteristics of the plant that the operator has experientially and predicting the initial state. In order to improve the accuracy and obtain good convergence of start-up schedule optimization through iterative calculation, operators' qualitative knowledge about the initial plant state and appropriate start-up schedule is used from experience. An object of the present invention is to provide a power generation plant startup device that enables optimization of an initial schedule.

[Means for solving problems]

The above objectives are to develop a dynamic characteristic model for quantitatively calculating the start-up characteristics of a plant, an initial value prediction means for predicting the initial temperature state of the plant when the boiler temperature is high, and a means for predicting the initial temperature in the shortest time. A schedule optimization means for optimizing the startup schedule using an iterative calculation method while evaluating the startup characteristics obtained by the above dynamic characteristic model in order to achieve startup, and a basic method for determining the initial schedule for the above-mentioned iterative calculation. This can be achieved by using a schedule creation means.

The above-mentioned initial value prediction means and basic schedule creation means apply fuzzy inference and use a method similar to the thinking method of a skilled operator, thereby achieving high-precision initial value prediction and repetition necessary for high-precision starting characteristic prediction, respectively. It becomes impossible to create an appropriate initial schedule that provides good convergence through calculations.

[Effect]

Therefore, the startup characteristics when the plant is started can be quantitatively calculated. Therefore, it is possible to know the relationship between the startup schedule and startup characteristics before actually starting the plant, and it is also possible to confirm in advance whether the plant status values satisfy the operation limit conditions, which improves safety. You can create high startup schedules.

The initial value prediction method applies fuzzy reasoning to predict the initial value in a way similar to the thinking method of a skilled operator, and uses qualitative differences between the standard cooling characteristics of the plant and the actual cooling state. A method is used to predict the initial value. This enables highly accurate initial value prediction, and as a result,
It is possible to predict startup characteristics with high accuracy using a dynamic characteristics model.The schedule optimization means applies fuzzy inference to optimize the schedule using a method similar to the thinking method of a skilled operator. Using an iterative calculation method, we qualitatively evaluate the startup characteristics obtained by the model, modify the schedule, calculate the startup characteristics again using the dynamic characteristics model, qualitatively evaluate the results, and modify the schedule. Optimization is being performed. The way a skilled operator thinks is similar to the way a skilled operator thinks, in that the amount of schedule modification is determined based on the qualitative cause-and-effect relationship between the schedule and startup characteristics. As a result, the convergence for finding the optimal schedule is extremely good, so it is possible to use a large-scale dynamic characteristic model using detailed calculation formulas, and it is possible to predict startup characteristics with high accuracy.

The basic schedule creation means is for creating an initial schedule to be used in the schedule optimization according to the initial value obtained by the initial value prediction means, and is used to qualitatively compare the predicted initial value and the standard initial value. An appropriate initial schedule is created by modifying the standard schedule due to the differences. As a result, it is possible to imitate the thinking method of a skilled operator regarding schedule creation, and the convergence of the above-mentioned schedule optimization can be improved.

By means of the means described above, it is possible to create an optimal startup schedule that satisfies the plant operation restriction conditions, minimizes the time required for startup, and completes startup at the time designated by the intermediate supply.

(Example〕 An embodiment of the present invention will be described below.

No. 1111 shows the overall configuration of a power plant starting device to which the present invention is applied. This device is used in a plant 2000 consisting of a boiler, a steam turbine, and a generator.
Startup schedule creation function 1ooo for determining equipment operation timing and control targets during the entire startup process from the boiler point high until reaching the target load (generally the load level commanded from the intermediate supply). A startup control function 3000 automatically starts the plant according to a set startup schedule, and displays necessary information regarding plant startup on a display device 115000 in response to a request from a plant operator 6000, and creates a schedule. The information in the function 1000 is converted into a dynamic characteristic model 100. Initial value prediction function 200. Basic schedule creation function 300. Schedule optimization function 400. Fuzzy inference function 500.
It consists of 600 knowledge bases. SOmbe-x6oo further includes an initial value prediction rule 610. Basic schedule creation rules 640. It consists of schedule optimization rules 670.

Before explaining the startup schedule creation function 1000 in detail, the purpose of each function will be explained.

The dynamic characteristic model 100 starts the plant according to the startup schedule given by the basic schedule creation function 300 or the schedule optimization function 400, with the plant state at the time of boiler ignition predicted by the initial value prediction function 200 as an initial value 210. This is for quantitatively calculating the starting characteristics 140 for the case.

The initial value prediction function 200 predicts the plant state at the time of boiler ignition at an arbitrary time before startup, and generates a dynamic characteristic model 100 and a basic schedule 'creation function 300.
This is for setting.

The basic schedule creation function 300 is for determining a basic schedule 310 as an initial schedule and setting it in the dynamic characteristic model 100 in order to obtain good convergence in the optimum value calculation in the schedule optimization function 400.

The schedule optimization function 400 uses the dynamic characteristic model 100
The optimal schedule 420 is determined by an iterative calculation method in which the startup characteristics are predicted using the Dynamic Characteristics Model 100, the schedule is corrected according to the results, the startup schedule 410 is set in the dynamic characteristics model 100 again, and the startup characteristics are predicted. be.

The fuzzy inference function 500 includes the initial value prediction function 20 described above.
0. It acts on the basic schedule creation function 300 and the schedule optimization function 400, and in each process, by simulating the thinking method of a skilled operator, it aims to improve the prediction accuracy of startup characteristics and speed up the solution of the optimal schedule. Basic schedule creation rules 640. A schedule optimization rule 670 is prepared and used as a knowledge base.

Here, the basic concept of startup schedule optimization by the schedule optimization function 400 will be explained using FIG. 2.

In FIG. 2, the broken lines indicate the turbine stress, startup pattern, and startup time before schedule optimization, that is, when the plant is started according to the basic schedule. In addition, the solid lines indicate the results after schedule optimization. In this figure, the connection time (the time to connect the generator to the power grid) is specified by the central supply, but the start-up is completed. Even if the time is specified, the present invention can basically optimize the schedule using the same method.If the plant is started according to the startup schedule before optimization, as shown in the figure,
Turbine stress has a large margin with respect to the limit value in the first half of startup, and in the second half, the margin becomes smaller and stress exceeding the limit value is generated in some parts.

When such startup characteristics are predicted by the dynamic characteristics model 100, the schedule optimization function 400 operates the fuzzy inference function 500 using the schedule optimization rules 670, and in the first half of startup, the ignition time is delayed or the turbine The startup time is reduced by shortening the speed holding time and load holding time. Additionally, in the latter half of startup, the load holding time is extended to alleviate turbine stress. As described above, by using the starting device to which the present invention is applied, it is possible to suppress the turbine stress, which is a factor limiting operation, to a limit value or less, and to start the turbine at the designated time from intermediate supply in the shortest possible time.

Below, 100 startup schedule creation functions outlined above
0 will be explained in detail for each constituent function.

(1) Initial value prediction function 200 In order to predict startup characteristics with high accuracy using the dynamic characteristics model 100, the plant initial values (values at the time of boiler ignition, mainly temperature condition fi) used in the dynamic characteristics model are It is necessary to predict with high accuracy. However, when analyzing the cooling characteristics based on the current calculated values, ′, nno, ′) l! -Determining this is difficult because it is affected by the operation details at the time of shutdown and the current environment of the plant.

However, operators with extensive operating experience can fairly accurately predict future deviations based on how much the current temperature condition deviates from the standard condition. This method uses the operator's knowledge to create rules for how much the values predicted by the standard cooling characteristics should be corrected in response to the current deviation. The present initial value prediction function 200 will be specifically explained below.

The difference between the previous disconnection time (the time when the generator was disconnected from the power system) tpr and the specified time tpt for parallel connection from the intermediate supply is defined as the stop time Δtri. A corresponding startup mode is determined from standard schedules prepared for each of four startup modes defined in advance according to this stop time, and a startup schedule is selected. The startup schedule here is defined by the following parameters that define the startup pattern from boiler ignition to completion of startup.

(a) Boiler startup time (from ignition to ventilation) (b)
1st speed holding time (at 1000 rpm+m) (C
) 2nd speed holding time (at 2800 rpm) (d
) 3rd speed holding time (at 3600 rpm) (a
) 1st load holding time (at initial load level) (f)
Second load holding time (at 20% load level) (g) Third load holding time (at 40% load level) Also, turbine speed increase rate (rpm/min) and load change rate (
%/min) is also uniquely determined in accordance with the startup mode.

Once the above schedule parameters are determined, the boiler ignition time tra is determined by back calculation from the joining time. Select the appropriate initial value depending on the startup mode. The initial values used here relate to the following plant conditions.

(a) Drum temperature (b) Superheater outlet steam temperature (C) Reheater outlet steam temperature (d) Main steam pipe metal temperature (θ) Reheat steam pipe metal temperature (f) Ice wall inlet internal fluid temperature (g) Economizer outlet internal fluid temperature (h) Economizer inlet internal fluid temperature (i) High pressure turbine (HPT) 1st stage post metal temperature (j
) Intermediate pressure turbine (IPT) first stage after metal temperature (k)
Drum Pressure On the other hand, standard values for the above-mentioned state quantities at present, that is, at the time of creation of the start-up schedule, are determined using the standard cooling characteristics of the turbine and boiler while they are stopped. This will be called the current standard value estimation function. Note that the above standard cooling characteristic is defined by the following equation.

- Mutual skin T TC T engineering (TPFT-T^) θ +T^ ・・・
(1) Here, TPFT Standard temperature when nibrant is stopped (℃) T^: Atmospheric temperature (''C) Tlsf!: Elapsed time after line separation (minutes) Chome Tc: Cooling time constant (minutes) Regarding the drum pressure mentioned above. The same is true.

In addition, the initial state at the time of boiler ignition is similarly predicted.

In order to perform this prediction with high accuracy, the actual measured value of the current state 2
The predicted value is corrected by fuzzy inference taking into consideration the difference between the standard value obtained by Equation (1) and the standard value obtained by Equation (1).

This correction method will be explained below.

Now, the current value deviation E(1), E(2),...E
(11) is the drum temperature X1 (toL superheater outlet steam temperature X5 (to)) actually measured at the present moment (to), respectively.
,...Drum pressure Xxx(to) and its standard value Xzs(tode Xts(t) obtained from equation (1)
(1), +++++ This is the difference from Xtns(t o) normalized by a standard value, and is defined by the following formula.

FIG. 3 shows constants that define the shape in Figure 0, which is a membership function for qualitatively evaluating the magnitude of the current value deviation, PB, PM, and PS.

ZO, NS, NM, and NB are the names given to the membership functions to qualitatively evaluate the magnitude of the deviation E,
Each has the following meaning.

PB: Po5itive BigP M: P
o5itive MediumPS: Po5it
ive SmallZ O: Zar.

NS: Nagatlve Small NM:
Negative Mediu@N B: Neg
active Big Also, the vertical axis of the figure is the membership value. Using this membership function, the current value deviation E(i) (i=1 to 11) regarding the 11 state quantities is calculated.
Evaluate qualitatively.

Figure 4 qualitatively organizes the degree of influence that the current value deviation has on the initial stage of boiler ignition. Predicted value of superheater outlet steam temperature at ignition (
For example, rule Na3 is for qualitatively determining the correction amount DtIa(3) of
In the case of 2, I F (E(2) is NS and
E(3) is NM)THEN(DTIQ(2)i
s NS).

Figure 6 shows the membership function for converting the qualitatively determined initial value correction amount into a quantitative value.・PB is a constant that defines the shape.

PM, PS, zO, NS, NM, NBjt These are the names given to the membership functions to qualitatively represent the magnitude of the correction amount, and correspond to the names used in FIG. Moreover, the vertical axis of the figure is the membership value. The applied rule qualitatively determines which membership function the initial value modification amount belongs to. The actual quantitative correction amount 51 is calculated based on the corrected valuable average value for the initial value.
0 (see FIG. 1), thereby predicting each initial value.

(2) Basic schedule creation function 300 The basic schedule is the initial schedule for the convergence calculation of schedule optimization, and to obtain good convergence, it is desirable to set it as close to the optimal value as possible. The 0-line basic schedule creation function 300, which allows the operator to fairly accurately determine the startup schedule according to the initial state at the time of boiler ignition, focuses on this point.
It utilizes the knowledge of operators who have established rules for how much to modify the parameters of a standard schedule prepared in advance in response to deviations from standard values in the initial state. The basic schedule creation function 300 will be specifically explained below.

The startup schedule selected in accordance with the startup mode is one selected from standard schedules, and is not necessarily the one that matches the current startup conditions. Therefore, the standard initial value and the above Power,
'4. .

This function corrects the schedule by taking into account the difference from the initial value predicted by the first method. This amount of correction is determined by fuzzy inference according to the difference between the predicted values.

The method for determining the amount of correction will be explained below.

Now, the predicted deviation at the time of ignition Ep(1), Ep(2),...
... Ep (11) is the initial value prediction function 20, respectively.
The drum temperature Xt (txoL) at the time of boiler ignition predicted at
Superheater outlet steam temperature Xz (tto) *...Drum pressure Xzt (two) and its standard value Xt (ttaL
X5 (tXo).

. . . The difference from Xzi(tto) is normalized by a standard value, and is defined by the following formula.

FIG. 7 shows the membership function Epxn(i)(
i=1 to 7) are constants that define the form of membership functions, PR, PM, and PS.

ZO, NS, NM, and NB are names given to membership functions in order to qualitatively evaluate the magnitude of the deviation Ep, and their meanings are the same as those in the initial value prediction function 200. Moreover, the vertical axis of the figure is the membership value.

Figure 8 summarizes the knowledge needed to create an appropriate basic schedule according to the predicted initial value at the time of boiler ignition, and which schedule parameters are effective to modify depending on the predicted deviation at the time of ignition. This is an organized list of the targets. An example of the basic schedule creation rule created in accordance with this is shown in Figure 9. This figure shows how the drum temperature at boiler ignition is deviated from the deviation Hp (1), the boiler start time required correction amount Dp (1), and the first speed maintained. Time correction amount Dp(2),・
...0 for qualitatively determining the third load holding time correction amount DP(7) For example, in the case of rule &6.

I F(Ep(1)ls PM) THI N(Dp(1) is PM and Dp(
2) is P 5and Dp(3) 1s P S
and Dp(4) is P 5and Dp(5
) is P S and Dp(6) is P 5
and Dp(7) is P S).

Figure 10 shows the membership function for converting the qualitatively determined schedule parameter modification amount into a quantitative value. It is a constant that defines the shape. P.B., P.
M, PS, zO, NS.

NM and NB are names given to membership functions to qualitatively represent the magnitude of the correction amount, and correspond to the names used in FIG. Moreover, the vertical axis of the figure is the membership value. The applied rule qualitatively determines which membership function the modification amount of the schedule parameter belongs to. If the modification value for the schedule parameter consisting of 1 is defined by multiple memberships according to multiple rules, the weighted average value corresponding to each membership value is set as the actual quantitative modification amount 520 (see Figure 1). This means that a basic schedule has been created.

(3) Schedule optimization function 400 Based on the basic schedule created by the above-mentioned basic schedule creation function 300, the optimal schedule is one that satisfies the operation restriction conditions in the entire startup process from boiler ignition to completion of startup, and the required startup time. This schedule optimization function 400 determines the schedule parameter that minimizes . FIG. 11 shows the overall processing procedure of this function.

This function uses the plant dynamic characteristic model 100 in order to evaluate the starting characteristics of the turbine thermal stress, which is an operation limiting condition. There is a strong correlation between the anaphoric cover turn at startup and the schedule parameters, and the smaller the margin (hereinafter referred to as margin) for the thermal stress limit value, the shorter the startup time. However, when using conventional constrained nonlinear optimization algorithms such as the complex method.

In order to obtain the optimal solution (optimal schedule), if seven variables are used as parameters as in this example, it is necessary to repeat calculations at least 100 times (calculation of startup characteristics using a dynamic characteristics model).
is required, which is not realistic.

Therefore, we aim to significantly improve convergence by using an optimization algorithm that applies fuzzy inference.

Plant operators know from experience which parameters can be shortened by how much, depending on the margin, given predicted values of thermal stress characteristics before startup. This empirical and qualitative knowledge is utilized to determine the amount of schedule modification for optimization.

Specifically, as shown in FIG. 11, first, the turbine thermal stress characteristics are controlled when the plant is started according to the basic schedule determined by the schedule parameters pt (i=1 to 7) given from the basic schedule creation function 300. Prediction is made using the characteristic model 100. Here, as shown in FIG. 12, the dynamic characteristic model 100 includes a schedule calculation function 110 for calculating target values of a boiler ignition command, turbine speed, and load when schedule parameters are given, and a boiler starting characteristic. A boiler model 120 for calculating , and a turbine model 1 for calculating thermal stress of the turbine in response to steam conditions generated from the boiler.
It consists of 30.

The turbine thermal stress calculated here is from four areas: rotor surface stress and bore stress of the high-pressure turbine, and rotor surface stress and bore stress of the intermediate-pressure turbine, all of which are important operational limiting factors that should be noted when starting the turbine. It is. The starting characteristic evaluation function 140 shown in FIG. 11 divides the starting process into seven sections, and calculates the minimum stress margin ma (j=
1 to 7) In this figure, m4 is the minimum margin Ms(j) between high pressure rotor surface stress and medium pressure rotor surface stress in section j, and the minimum margin Ma(j) between high pressure rotor bore stress and medium pressure rotor bore stress. The zero starting characteristic evaluation function 140 shown in both senses is for extracting the characteristics of the hot melt cover turn in the subsequent stress margin evaluation function 150.

The stress margin evaluation function 150 uses the membership function shown in FIG.
MMB (i) (i, = 1 to 6) in Figure 1 is a form of membership function. NM and NB are the names given to the membership functions in order to qualitatively evaluate the magnitudes of the stress margins Ms(j) and MB(j). Moreover, the vertical axis of the figure is the membership value.

The schedule optimization rule 670 is piecemeal knowledge such as ``What schedule parameters should be modified and how much should the hot melt cover pattern be?''. 14th
The figure summarizes the knowledge for modifying the schedule according to the turbine hot melt cover turn predicted using the dynamic characteristic model 100. This is a summary of whether it is effective to modify schedule parameters. Here, Ms(I L Ms(2), . . . M
s(7) and Ma(1), Ma(2), -・”MB(7
) are the first. Second. . . . The minimum stress margin on the rotor surface and the minimum stress margin on the rotor bore in the seventh section.

6 Figure 15 shows the scale 1 degree created according to the above idea.
? 2) shows an example of the LtR optimization rule.

This figure shows the 5th section minimum stress margin Ms(5) and the 6th section minimum stress margin Ms(6) regarding the rotor surface stress.
), calculate the third speed holding time correction amount as the schedule parameter correction amount DPT(4)1 the first load holding time correction amount DPT(5) and the second load holding time correction amount DPT
(6) is for qualitatively determining 0. For example,
In the case of rule NQ54, I F (Ms(5) is
P B and Ms(6) is PM) THEN
(DPT(4) is NM and Dpt(5)
isNM and DPT(6)isNS.

FIG. 16 shows DPTMB (i) (i = 1 to 7) in FIG.
) is a constant that defines the form of the membership function. P
B, PM, PS, 20. N.S., N.M.

NB is a name given to the membership function to qualitatively represent the magnitude of the correction amount. Moreover, the vertical axis of the figure is the membership value.

The modified parameter selection function 160 shown in FIG. 11 selects modified parameters by comparing the characteristics of the response pattern extracted by the stress margin evaluation function 150 with the schedule optimization rules 670, and selects the modified parameters by comparing the characteristics of the response pattern extracted by the stress margin evaluation function 150 with the schedule optimization rules 670. It is qualitatively determined to which membership function in FIG. 16 the modification amount belongs.

If the amount of modification to a schedule parameter that consists of is specified by multiple memberships in multiple rules,
The weighted average value corresponding to each membership value is used as the actual quantitative modification amount 530 (see FIG. 1).The schedule modification amount determination function 17 shown in FIG. 11 performs this.
It is 0.

The amount of modification of the schedule parameters is determined above, and the dynamic characteristic model 1 is used again to predict the startup characteristics when the plant is started according to the modified schedule.
The optimal activation schedule 420 (see FIG. 1) can be obtained by repeating 0 or more operations in which 0 is activated.

Note that the convergence determination function 180 in FIG. 11 is for determining the optimality of the startup schedule created by adding the above-described iterative calculations. In addition, the criterion for this is that among the startup schedules in which the turbine thermal stress is below the limit value during the entire startup process, the time reduction rate compared to the previous startup schedule is below a predetermined value; The startup schedule that requires the shortest startup time is defined as the optimal schedule 420 (see FIG. 1).

The optimal schedule 420 determined here is displayed on the display device 115000 via the user interface 4000 and set in the activation control function 3000, as shown in FIG. The optimal schedule 420 set in the startup control function 3000 is executed upon receiving an execution command 430 from the operator 6000. Startup control function 3
000 is.

To do this, it receives process input 3010 from plant 2000 and provides control output 3010 to the plant.

Operator 6000 adds knowledge base 600.

If changes or deletions are necessary, please contact User Interface No. 17.
The figure shows the schedule optimization process in the schedule optimization function 400 that applies fuzzy inference. The numbers in the figure indicate the startup schedules created at the 1st, 3rd, 5th, and 200th iterations of repeated calculations, and the schedule optimization process at that time. Showing the startup characteristics, here, the first time corresponds to the basic schedule,
Corresponds to the 200th optimal schedule. As can be seen from this figure, a schedule close to the optimal value was obtained in the fifth time, and the convergence of the single algorithm is extremely good. In addition, the created startup schedule is displayed at the specified merging time (in this figure, it is shown as the elapsed time from the time of train separation).
80 minutes (8 hours)). The schedule has been revised to shorten the startup time by omitting speed holding and load holding in the first half of startup, when the thermal stress margin of the turbine is large, and by extending load holding in the second half of startup, when the thermal stress margin is negative. This is aimed at alleviating thermal stress. It is possible to start up in this way.

The schedule error monitoring function 700 will be described next. The schedule error monitoring function 700 monitors the occurrence of schedule errors using the following two functions.

One is an error detection function that monitors whether a schedule error has actually occurred, and the other is an error prediction function that predicts the occurrence of an error before it actually occurs.

The error detection function monitors whether a startup pattern defined by startup schedule parameters created before startup is accurately executed during actual startup. Therefore, this deviation in the activation pattern is monitored.

On the other hand, the error prediction function monitors the behavior of the plant state quantity that causes the deviation before it appears as a deviation in the starting pattern, and predicts the behavior of the plant state quantity using the dynamic characteristic model 100 when creating the starting schedule.

If schedule errors are monitored using the above two functions and the occurrence of an error is detected or predicted, a message to that effect 710 is sent to the user interface 40 along with related plant state quantities.
00 through 1 display bag [Display in 5000. This allows the operator 600o to take prompt action.

(Modified example of the invention) In the embodiment of the present invention described above, the schedule parameters are focused on the boiler startup time, the turbine speed holding time, and the load holding time, but it is not necessarily limited to these, and the turbine speed change It is clear that the present invention can be implemented without changing the basic principle as long as the parameters are representative of the plant start-up pattern, such as rate, load change rate, boiler temperature increase rate, pressure increase rate, etc.

Further, in the embodiments of the present invention, attention is focused on turbine thermal stress as an operation limiting condition, but it is not necessarily limited to this. The state quantity that can be estimated visually or the turbine expansion difference,
Important limiting factors for operation, such as boiler steam pressure increase rate and boiler combustion gas temperature, vary depending on the plant, so they can be taken into consideration as necessary, and it is clear that the present invention can be practiced without changing its essence.

In addition, in the embodiment of the present invention, a method for creating a startup schedule that accurately adheres to the joining time specified by the central power dispatch center was explained as an example, but in implementing the present invention,
It is clear that the essence of the present invention does not change even if the joining time does not necessarily have to be designated, and the boiler ignition time, turbine ventilation time, target load arrival time, etc. can be specified depending on the plant.

In addition, the conditions focused on as the initial plant values in this example should be selected appropriately depending on the equipment configuration and measurement position of the plant in which the present invention is implemented, and it is not necessarily necessary to use the same values as in this example. Although it is expressed as a time function from time, even if it is expressed as a time function based on the time when the plant cools down, such as the boiler extinguishing time, the essence of the present invention does not change.

In addition, it is not necessary to necessarily adopt the convergence determination method in the schedule optimization function of this embodiment, but it is necessary to perform repeated calculations a predetermined number of times and satisfy the operation restriction conditions,
A method that considers the startup schedule that minimizes the startup time as the optimal value, or performs repeated calculations until a predetermined number of startup schedule candidates that satisfy the operation restriction conditions are obtained, and then calculates the startup schedule that takes the minimum startup time among them. Of course, the essence of the present invention does not change even if a method is adopted in which the optimum value is regarded as the optimum value.

Furthermore, although all membership functions used in this example are triangular, they do not necessarily have to be of this shape; quadratic or exponential curves may be adopted depending on the characteristics of the plant and the knowledge of the operator. However, the essence of the present invention does not change even if not only the form of the men/ship functions but also the number of them is arbitrarily set, the essence of the present invention does not change.

〔Effect of the invention〕

According to the present invention, the time required to start up a plant can be reduced by about 30% compared to conventional systems, thereby stabilizing power systems that require frequent starting and stopping of power plants due to fluctuations in load demand. Economical operation becomes possible. This also significantly reduces the burden on the operator. Also,
As the start-up time is reduced, energy loss during start-up can be reduced by about 15%, resulting in a significant reduction in power plant operating costs.

In addition, according to the present invention, it is possible to predict startup characteristics with high accuracy, and startup can be performed while faithfully observing operating limit conditions and specified times, improving safety in plant operation and providing accurate power to the power system. supply becomes possible.

Further, according to the present invention, the calculation time required for the startup schedule optimization calculation can be shortened, so there is no need to use an expensive computer with high computational performance, and an inexpensive startup device can be provided.

[Brief explanation of the drawing]

Figure 1 shows the overall configuration of the power plant starting device;
Figure 3 shows the basic concept of startup schedule optimization, Figure 3 shows the membership function for evaluating the current value deviation, Figure 4 shows the degree of influence of the current value deviation on the initial value at the time of boiler ignition, and Figure 5. The figure shows an example of the initial prediction rule. Figure 6 shows the initial value correction electric conversion membership function.
The figure shows the membership function for evaluating the predicted deviation at the time of ignition. Fig. 8 shows the relationship between the predicted ignition deviation and the schedule parameters to be corrected, Fig. 9 shows an example of rules for creating a basic schedule, Fig. 10 shows membership functions for correcting schedule parameters, and Fig. 11 12 shows the dynamic characteristic model; FIG. 13 shows the membership function for stress margin evaluation; FIG. 14 shows the relationship between the stress margin and schedule parameters to be modified. FIG. 15 shows an example of a schedule optimization rule. Figure 16 shows member dormitory 30 Tachibana 5's ηDial*-rv 2E(Z>, ε(3) ff) for modifying schedule parameters.
raDTIC (Zle & 1jis3ru - JVUF (a(Z)
Is NS −ε(j)isNM) TMEN(
DT + 4 (Z] Is NS) 寥i 90th rule Q 吏aj, (ppo, ab*sl) @fugnA
optD ts ps) No. 10 Kuchifu 13

Claims (1)

[Claims] 1. A boiler for generating steam using heat generated by combustion of fuel, a steam turbine for converting the thermal energy of the generated steam into mechanical energy, and a steam turbine for converting the thermal energy of the generated steam into mechanical energy. A power generation plant comprising a generator for converting into electrical energy, and a power generation plant comprising a startup schedule creation means for determining equipment operation time and control target values according to an initial temperature state before startup of the power generation plant. , in order to predict the initial temperature state, a standard cooling characteristic function that expresses the plant cooling characteristics after the plant is stopped as a time function, and the time (current time) at which the startup schedule is created.
current state measuring means for actually measuring the plant temperature state at the current time; current state estimating means for estimating the plant temperature state at the current time using the standard cooling characteristic function; and start-up starting using the standard cooling characteristic function. The difference between the estimated current value obtained by the first initial value prediction means for predicting the plant temperature state at the time and the current state estimation means and the actual measured current value obtained by the current state actual measurement means. A current value estimation error calculation means for determining a certain current value estimation error, and a second initial value prediction for correcting the predicted value obtained by the first initial value prediction means according to the current value estimation error. a startup mode selection means for selecting a corresponding startup mode according to the actual plant shutdown time from among a plurality of startup modes predefined according to the length of the plant shutdown time in order to create the startup schedule; a standard schedule selection means for selecting a standard schedule corresponding to the selected startup mode; and a temperature for determining a difference between a standard initial temperature corresponding to the selected standard schedule and an actual initial temperature state. A power generation plant starting device characterized by having a deviation calculation means and a basic schedule creation means for modifying the selected standard schedule according to the temperature deviation and creating a new schedule. 2. The second initial value prediction means according to claim 1, further comprising a current value estimation error evaluation means for qualitatively evaluating the magnitude of the current value estimation error; a predicted value correction means for determining a correction amount of the predicted value obtained by the first initial value prediction means by applying a correction rule prepared in advance in relation to the current value estimation error; In order to create the basic schedule, a temperature deviation evaluation means for qualitatively evaluating the magnitude of the temperature deviation and a correction rule prepared in advance in relation to the qualitatively evaluated temperature deviation are applied. 1. A power generation plant starting device comprising standard schedule modification means for determining the amount of modification of the selected standard schedule.