JPH05187265A - Gas turbine combustion daignosing device - Google Patents

Abstract

PURPOSE:To perform the diagnosis of a combustion state and presentation of guidance by estimating quality of combustion through fuzzy inference based on measurement being the operation result of a gas turbine and effecting presen tation of positioning of a combustion mode, preclassified according to a fuzzy rule, to an operator. CONSTITUTION:A power plant comprises a gas turbine 10, a combustor 100, a compressor 20, a generator 30, a combustion control device 1000, and a combustion diagnosing device 2000. In this case, through fuzzy inference based on measurement being the operation result of the gas turbine 10, quality of combustion is decided. Positioning of a combustion mode preclassified according to a fuzzy rule is presented to an operator. Namely, the temperature 127 of a premixture combustion flame, the temperature 128 of a flame stabilizing ring, and the temperature 42 of exhaust gas by a thermometer 41 are respectively measured. Meanwhile, operation guidance 2120 and a combustion mode 2130 are presented to an operator 200 from the combustion diagnosing device 2000 based on each measurement.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for diagnosing an operating state of a gas turbine based on the measurement information of the gas turbine,
In particular, the present invention relates to a gas turbine combustion diagnosis device that diagnoses a combustion state and presents guidance for adjustment or automatically adjusts when a defective combustion occurs in order to maintain a good combustion state.

[0002]

2. Description of the Related Art Recent gas turbines use N in exhaust gas.
In order to reduce the _Ox concentration and increase the thermal efficiency, the ratio of the fuel amount to the air amount supplied to the combustor (hereinafter referred to as the fuel-air ratio) is set within an appropriate range according to the operating conditions of the gas turbine. Need to drive. However, the combustion state may become unstable, misfire, or high temperature may occur due to changes in the fuel calorific value and atmospheric humidity as the above operating conditions, resulting in reduced thermal efficiency, reduced output, thermal fatigue damage to combustor members, and high temperature damage, there is a risk of causing various troubles such as NO _X concentration increases. Conventionally experienced operator and adjustment members watching a measurement of the gas turbine exhaust temperature and NO _X concentration was not operate the fuel-air ratio manually.

[0003]

However, the above-mentioned prior art has the following problems.

The first problem is that there are many operating conditions to be considered, and it is difficult to objectively quantify the causal relationship with the combustion state.

The second problem is that, in the case of directly and continuously measuring the combustion state, not only the measuring equipment is expensive, but also it is exposed to a high temperature state, so that there is a reliability problem and it is practically impossible. is there.

A third problem is the operation result as the exhaust temperature and NO _X concentration state quantity was measured in is information indirectly representing the combustion state, it is difficult to quantitatively grasp the combustion state That is.

The fourth problem is that, due to the above-mentioned various problems, it is difficult for a trained operator to quantitatively determine an appropriate operation amount even if a conventional measurement information value as an operation condition or an operation result is given. It is also extremely difficult for the coordinator and a heavy burden. An object of the present invention is to enable diagnosis of combustion state and presentation of guidance by fuzzy reasoning.

[0008]

The present invention solves the above problems by the following means.

First, as a basic means, operating know-how based on the experience of skilled operators and coordinators, that is, knowledge on a qualitative operating method for improving the combustion state based on measured values as operation results. Is built in the storage device of the computer system as a fuzzy rule base, and the operation amount is automatically determined by referring to the above fuzzy rule base by fuzzy inference based on the measured value during operation,
A fuzzy inference means for presenting operation guidance to operators and coordinators is provided. Further, the fuzzy inference means has a function of estimating the quality of combustion from the conformity of the fuzzy rule and presenting it together with the operation guidance. Next, if the combustion state is successfully improved when operated in accordance with this operation guidance, the neural network is made to learn the correspondence relationship between the operating condition and the operation result before the operation, the quality of the combustion, and the operation amount, and this is repeated. As a result, learning means is provided to establish the relationship between the two so that the operating condition and the degree of combustion when the operating result is input as the internal connection state of the neural network and the appropriate operation guidance are output.

[0010]

The above means operates as follows in order to solve the above-mentioned conventional problems.

First, the fuzzy rule base makes it possible to use the qualitative operating know-how of an operator or a coordinator as objective knowledge in a computer. The result of fuzzy inference using this is verified by comparing it with the actual driving result, and the correct knowledge is fixed in the neural network through the learning means. By repeating the above, it is possible to easily quantify the causal relationship between the operating condition and the combustion state for the first problem, and it is not necessary to directly and directly measure the combustion state for the second problem. Regarding the problem, it is possible to quantitatively grasp the combustion state from the operation result, and regarding the fourth problem, an appropriate operation amount can be automatically determined from the conventional measurement information, so that a skilled operator or adjustment can be performed. Even if it is not a member, it becomes possible to operate very easily.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A combustion diagnosis device for a gas turbine used in a power plant will be described below as an embodiment of the present invention.

FIG. 1 shows a gas turbine 10 used in a power plant, its combustor 100, a compressor 20, and a generator 3.
0, combustion control device 1000 and combustion diagnosis device 20 of the present invention
00 and the positions of measurement points and operating ends corresponding to those input / output signals are shown. The gas turbine of this embodiment has ten combustors, and each combustor has a double flame structure as shown in FIG. That is, the fuel ejected from the fuel nozzle 111 arranged inside (the primary fuel 112
(Referred to as “primary air 113”) and the air supplied from the surroundings (referred to as “primary air 113”) in a diffused state, and from the fuel nozzle 121 disposed outside to the premixing chamber 125. Ejected fuel (this is the secondary fuel 11
2) and air (which is referred to as secondary air 123) are mixed and then burnt, and then burnt. In the latter case, the fuel-air mixed gas leaving the premixing chamber 125 ignites downstream of the flame holding ring 126 for obtaining stable combustion and forms the premixed combustion flame 124.

In the present embodiment, the temperature 127 of the premixed combustion flame 124 (this is the F ₂ gas temperature T _GF2i (i = 1 to 1 corresponding to the number of combustors) is used as a clue for combustion diagnosis.
0)) and the temperature 128 of the flame holding ring 126 (this is called flame holding ring metal temperature T _RMi (i = 1 to 10 corresponding to the number of combustors)). In addition, 13 thermometers 41 distributed around the inner circumference of the exhaust duct of the gas turbine 41
The exhaust gas temperature 42 measured from (exhaust temperature T _EXi (corresponding to the number of thermometers and called (i = 1 to 13)) is measured.

Next, the combustion control device 1000 will be described. FIG. 3 shows a functional block diagram of the combustion control device 1000. This function receives the total fuel amount command 1001 (FDD) to the combustor according to the load request value, performs various calculations shown in FIG. 3, and then executes this calculation on the primary fuel 113 and the secondary fuel 12
2 target values for 2, that is, the primary fuel amount command value 10
11 (FFD ₁ ) and secondary fuel amount command value 1021 (FFD ₂ )
Split into. At the same time, this function receives the FFD value and receives the total air amount (the total flow rate of the air passing through the cooling holes of the primary air 112, the secondary air 123 and the main chamber sleeve 129).
The IFC opening 1031 that defines the flow rate ratio of the secondary air 123 to Here, the IFC opening 1031 corresponds to the position of the slide ring 1032 shown in FIG. In addition, T _CI is the compressor inlet air temperature 53,
LHV is the fuel heating value 51. Here, the IFC bias 2100 and the F ₂ ratio bias 2200 are input to the main combustion control device 1000 as manual operation amounts from the operator 200 as described later.

Next, the combustion diagnosis device 2000 will be described. FIG. 4 shows a basic functional block diagram of the combustion diagnosis device 2000.

First, FIG. 5 shows a processing procedure in the F ₂ fuel-air ratio calculation function 2500 in this function. Here, I
Rotational speed 45, IGV opening 54, compressor inlet air temperature T for basic secondary air amount x ₆ determined by FC opening
_The actual secondary air amount A ₂ is obtained by multiplying the correction value x ₅ by the _CI 53 and the operating time 1041, and the F ₂ fuel-air ratio 2510 is calculated by the ratio (F ₂ / A ₂ ) to the secondary fuel amount F _2. calculate.

Next, the fuzzy reasoning 2600 instructs the operator 200 on the operation guidance 2120 and the combustion mode 21.
The function of presenting 30 will be described. This function is shown in Figure 6.
It is executed according to the fuzzy rule base shown in. The fuzzy rule base consists of 12 fuzzy rules corresponding to the quality of combustion of the premixed combustion flame. In other words, rule No.1 is the misfire mode (F2
Rule No. 2 to 5 are unstable combustion modes in which flame disappears or arrives in a short time (called unstable region), and Rule No. 6 ignites behind flame stabilizer 126. The first stable combustion mode of continuous combustion (referred to as stable region (1)), rules No. 7 to 10 are transition combustion modes (referred to as transition regions) in which the ignition point repeats going in and out of the flame stabilizer 126, rule No. 11 is the second stable combustion mode in which the flame stabilizer 126 is ignited and continuously burned (referred to as stable region (2)), and Rule No. 12 is high temperature combustion in which the flame is fired before the flame stabilizer 126 and continuously burned at high temperature It corresponds to the mode (called the high temperature region). In the stable region (1), the fuel-air ratio is ideal. The fuel-air ratio becomes smaller toward the left side of this figure (air amount is relatively larger), and the fuel-air ratio becomes closer to the right side. Is large (the amount of air is relatively small). The detection amount as the condition part of the fuzzy rule is, as described above, the fluctuation range and fluctuation cycle of the exhaust temperature of the gas turbine, the temperature level of the flame ring metal temperature or the F ₂ gas temperature, the fluctuation range and fluctuation cycle, and the NO _X level. , C
It is O level. The above exhaust temperature and flame holding ring metal temperature or F ₂ gas temperature are handled by OR logic from a phenomenological viewpoint. Further, the above two, NO _X level and CO level are handled by AND logic. The operation guidance 2120 is obtained as the conclusion of the fuzzy inference, that is, the IFC bias correction amount and the F ₂ bias correction amount as the operation amount prepared in the conclusion part of the fuzzy rule, and this is displayed on the CRT screen. As a function of presenting the combustion mode 2130, the degree of suitability of each fuzzy rule is displayed on the CRT screen in association with the combustion mode name as shown in FIG. 7, so that the combustion state can be grasped.

Next, the combustion diagnosis function by the neural network 2700 will be described. Operation result evaluation 262 of the combustion state as the operation result when the IFC bias or the F ₂ bias is corrected according to the operation guidance 2120 obtained by the fuzzy inference 2600.
0 is evaluated, and a successful example (example in which the combustion state is improved) is learned by the learning function 2610 using the backpropagation method in the neural network having the input / output structure shown in FIG. In this case, as the input signal for learning, five operating conditions (F ₂ fuel-air ratio, LHV, absolute humidity,
T _CD , P _CD ) and the operation results at that time 7 pieces (exhaust temperature fluctuation range, exhaust temperature fluctuation cycle, F ₂ gas temperature level, F ₂ gas temperature fluctuation range, F ₂ gas temperature fluctuation cycle, NO _X level, C
O level) is used. Further, as the teacher signal at this time, the IFC bias correction amount and the F ₂ bias correction amount, which are the operation amounts, the pre-operation combustion mode (fuzzy rule conformance 12
Individual). By continuing the learning of the operation case in which the combustion state is improved in this way, it becomes possible to give the neural network 2700 the ability to predict the combustion state and the ability to present appropriate operation guidance. Regarding the back propagation method,
MIT Press, Neurocomputing Foundations of Research, 1988
Year 318-362 (The MIT Press, Neuro
computing Foundations of Research, 1988, pp318-36
2). FIG. 8 shows a display example of the operation guidance and the combustion mode. Operation result evaluation 2
The operation 620 may be performed by the operator 200, but may be easily performed automatically. That is, it can be considered that the combustion state is improved if the combustion mode of the operation result is closer to the stable region (1) than that before the operation. If the combustion state does not improve as a result of the operation, the fuzzy rule is judged to be unsuitable and the membership function in the condition part or conclusion part of the fuzzy rule is modified by the fuzzy rule modification 2630 according to the moving direction and the moving amount of the combustion mode. To do.

Next, the position determination of the defective combustor by the neural network will be described. Here, the defective combustor diagnosis neural network 2140 is used. The neural network 2140 has an input / output structure shown in FIG. That is, corresponding to the exhaust temperature detection points,
When the detection signals of 13 exhaust gas temperature fluctuation widths, 13 exhaust gas temperature fluctuation cycles, 15 loads and rotation speeds are input to the input layer, the position of the defective combustor is signaled to the output layer position corresponding to the No. By outputting, the position of the defective combustor can be determined in real time. The neural network 2140 can be constructed by learning in advance the causal relationship between the input and output signals from the actual machine data when defective combustion occurs.

In the above embodiment of the present invention, the operator 2
00 receives the operation guidance 2120 and manually operates the IFC bias 2100 and the F ₂ ratio bias 22.
Although 00 is input to the combustion control device 1000 as a correction amount, the operation results are accumulated and the fuzzy rule base 2 is used.
After the validity of 600 is verified, it is of course possible to use an online automatic control method.

Further, in the above-mentioned embodiment of the present invention, the diagnosis of the combustion state of the premixed flame was mainly conducted, but for the diffusion flame, the combustor having the same measuring means and fuel-air ratio operating means as in the embodiment. In, the present invention can be easily applied.

Further, in the above-described embodiment of the present invention, both the IFC bias 2100 and the F ₂ ratio bias 2200 are operated, but if either one of the IFC bias 2100 and the F ₂ ratio bias 2200 can be operated, one of them can be operated. It does not depart from the essence of the invention.

Further, in the above embodiment of the present invention, the combustor in which the slide ring 1032 is operated by the IFC opening command to adjust the fuel-air ratio is targeted, but any combustor having a fuel-air ratio operating means may be used. For example, the present invention can be easily applied.

In the above embodiment of the present invention, the flame holding ring metal temperature and the gas temperature in the combustor are used as the measurement information in the combustor, but the combustor measured using an optical fiber or an image fiber is used. According to the internal combustion flame luminance information, the responsiveness is higher and the reliable combustion situation can be grasped. Therefore, in carrying out the present invention, the combustion flame brightness can be substituted for the flame holding ring metal temperature and the gas temperature in the combustor. In this case, it is desirable to install the optical fiber or the image fiber in a position where the flame condition is within the viewing angle, such as the premix chamber and the main chamber.

Further, the above-mentioned embodiment of the present invention is intended for the gas turbine used in the power generation plant, and is of course applicable to the gas turbine in the exhaust heat recovery type combined cycle power generation plant. Further, the invention can be applied to a gas turbine for driving a machine such as a ship or a vehicle or a gas turbine for an aircraft engine without departing from the essence of the present invention.

According to the embodiment of the present invention, the following effects can be obtained.

[0028] (a) is operated defended ensure emissions regulations such as NO _X concentration in recent years in particular stricter environmental protection surface, it is facilitated without a skilled operator and coordinators.

(B) It is possible to use the qualitative operating know-how of the operator and the coordinator as objective knowledge in the computer.

(C) The causal relationship between the operating condition and the combustion state can be easily quantified, it is not necessary to directly measure the combustion state, and the combustion state can be quantitatively grasped from the operation result.

(D) An appropriate manipulated variable can be automatically determined from conventional measurement information.

[0032]

The effect of the present invention is that the combustion state can be prevented from becoming unstable, misfiring, or becoming high temperature even if various operating conditions such as fuel calorific value and atmospheric humidity change, so that the thermal efficiency is reduced. decrease in the output, thermal fatigue damage and high temperature damage to the combustor member, it is to be avoided in advance that fall into various failure state such as rise of the NO _X concentration. As a result, the thermal efficiency and operating rate of the gas turbine can be significantly improved,
Operating costs and repair costs can be significantly reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing a relationship between a gas turbine power plant and a combustion diagnosis device.

FIG. 2 is a diagram showing an internal structure of a combustor.

FIG. 3 is a functional block diagram of a combustion control device.

FIG. 4 is a basic functional block diagram of a combustion diagnosis device.

FIG. 5 is a block diagram of F ₂ fuel-air ratio calculation.

FIG. 6 is a diagram showing a fuzzy rule base used in fuzzy inference.

FIG. 7 is a diagram showing a combustion mode display example.

FIG. 8 is a diagram showing an example of combustion diagnosis by a neural network.

FIG. 9 is a diagram showing an example of determining the position of a defective combustor using a neural network.

[Explanation of symbols]

10 ... Gas turbine, 20 ... Compressor, 30 ... Generator, 100 ... Combustor, 1000 ... Combustion control device, 200
0 ... Combustion diagnostic device, 2500 ... F ₂ fuel-air ratio calculation function, 2
600 ... Fuzzy inference, 2700 ... Neural network, 2120 ... Operation guidance.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Norio Kuroda 3-1-1, Saiwaicho, Hitachi-shi, Ibaraki Hitachi Ltd. Hitachi factory

Claims (15)

[Claims] 1. A method for estimating the quality of combustion by fuzzy inference based on a measured value as an operation result of a gas turbine, and presenting the operator with a position in a combustion mode pre-classified in accordance with a fuzzy rule. Gas turbine combustion diagnostic device. 2. The ratio of the amount of fuel to the amount of air for combustion,
That is, in a gas turbine having a fuel-air ratio adjusting means,
A gas turbine characterized in that the quality of combustion is estimated by fuzzy inference based on the measured value as an operation result, and operation guidance for adjusting the fuel-air ratio is presented to the operator according to the degree of quality of combustion. Combustion diagnostic device. 3. A ratio of the amount of fuel to the amount of air for combustion,
That is, in a gas turbine having a fuel-air ratio adjusting means,
The quality of combustion is estimated by fuzzy reasoning based on the measured value as the operation result, and operation guidance for adjusting the fuel-air ratio is presented to the operator according to the degree of quality of combustion, and the fuel-air ratio is followed according to the guidance. If the combustion state is successfully improved when is operated, the neural network is made to learn the correspondence relationship between the operating condition and the operation result before the operation and the quality of combustion or the operation amount, and the neural network created by repeating this is learned. A gas turbine combustion diagnostic device characterized in that an operating condition and an operating result are input to an input layer of a network to obtain an estimated value of a combustion state or an operation guidance from an output value from the output layer. 4. In a gas turbine having a plurality of combustors, correspondence between a combustor that has experienced defective combustion in the past, exhaust temperature information obtained from exhaust gas temperature measurement points at that time, and the rotational speed or load of the gas turbine. The relationship is learned in the neural network, and the defective combustor is identified by the output signal from the output layer when the exhaust gas temperature information during actual operation and the rotational speed or load of the gas turbine are input to the input layer of the neural network. A gas turbine combustion diagnostic device characterized by: 5. The gas turbine combustion diagnostic device according to claim 1, 2, or 3, wherein the operation result of the gas turbine is:
Exhaust gas temperature information, the flame holding ring metal temperature, combustor gas temperature, combustion flame brightness, NO _X concentration in the exhaust gas, CO in the exhaust
A gas turbine combustion diagnostic device characterized by using one of the concentrations. 6. The gas turbine combustion diagnostic device according to claim 3, wherein the operating condition of the gas turbine is a fuel-air ratio,
A gas turbine combustion diagnostic device characterized by using any one of fuel calorific value, air humidity, compressor outlet temperature, and compressor outlet pressure. 7. The gas turbine combustion diagnostic apparatus according to claim 2, wherein either one of IFC bias adjusting means and premixed fuel ratio adjusting means is used as the fuel-air ratio adjusting means. Gas turbine combustion diagnostic device. 8. The gas turbine combustion diagnostic apparatus according to claim 1, 2, or 3, wherein a degree of good or bad combustion is a misfire mode in which a flame is extinguished for a long time, or a flame disappears or arrives in a short time. Unstable combustion mode, the first stable combustion mode that ignites behind the flame stabilizer and continuously burns, the transition combustion mode in which the ignition point repeats going in and out of the flame stabilizer, and the continuous combustion that ignites before the flame stabilizer 2. A gas turbine combustion diagnostic device characterized by distinguishing between at least two combustion modes among a stable combustion mode and a high temperature combustion mode in which ignition is performed before the flame stabilizer and continuous combustion is performed at a high temperature. 9. The gas turbine combustion diagnostic device according to claim 1, 2, or 3, wherein the degree of good or bad of combustion is adapted for each of the fuzzy rules prepared for the combustion mode or fuzzy inference defined in claim 8. A gas turbine combustion diagnostic device characterized by being presented to the operator depending on the degree. 10. The gas turbine combustion diagnosis device according to claim 5, wherein a fluctuation range or a fluctuation cycle is used as the exhaust gas temperature information. 11. The gas turbine combustion diagnostic device according to claim 6, wherein any one of a temperature level, a fluctuation range and a fluctuation cycle is used as the flame holding ring metal temperature, the gas temperature in the combustor and the combustion flame brightness. A characteristic gas turbine combustion diagnostic device. 12. The gas turbine combustion diagnostic apparatus according to claim 2, further comprising automatic control means for converting an operation amount corresponding to the operation guidance into an operation signal and automatically operating the operation end. Gas turbine combustion diagnostic device characterized by: 13. A gas turbine combustion diagnostic apparatus, wherein the gas turbine combustion diagnostic apparatus according to any one of claims 1, 2, 3 and 4 is used as a subsystem in an operation control system of a gas turbine power plant. 14. A gas turbine combustion diagnosis system using the gas turbine combustion diagnosis system according to claim 1, 2, 3 or 4 as a subsystem in an operation control system of an exhaust heat recovery combined cycle power plant. apparatus. 15. Use of the gas turbine combustion diagnosis device according to claim 1, 2, 3, or 4 as a subsystem in an operation control system of a gas turbine for driving a machine such as a ship or a vehicle or a gas turbine for an aircraft engine. A characteristic gas turbine combustion diagnostic device.