JP2016044881A - Defect event occurrence prediction device, defect event occurrence prediction method and defect event occurrence prediction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a defect event occurrence prediction device capable of predicting a defect event occurrence time without pulling out a furnace wall pipe in a plant including a boiler.SOLUTION: A defect event occurrence prediction device comprises a factor extraction part, a weighting calculating part, an equation derivation part and a defect event occurrence prediction part. The factor extraction part extracts a plurality of factors that are related to occurrence of defect events of a plant in reference to the collected plant data through a statistical analysis. The weighting calculation part calculates a contribution rate of each factor to the occurrence of defect events on the basis of each of the factors extracted by the factor extraction part and presence or non-presence of occurrence of defect events in the plant. The equation derivation part derives the prediction equation for obtaining presence or non-presence of defect events on the basis of each of the factors extracted by the factor extraction part and each of the contribution rates calculated by the weighting calculation part. The defect event occurrence prediction part predicts occurrence of defect events on the basis of the prediction equation calculated by the equation derivation part.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a failure event occurrence prediction device, a failure event occurrence prediction method, and a program.

In a once-through boiler provided in a thermal power plant or the like, the boiler water evaporates in the furnace wall tube, and iron oxide (hematite; Fe ₂ O ₃ ) having a particle size of several microns in the boiler water is deposited on the inner surface of the furnace wall tube. Are known. The deposited iron oxide particles are generally called a hematite scale or a powder-like (powdered) hematite scale (hereinafter referred to as “hematite scale”). The hematite scale has a high thermal resistance, so when it accumulates on the inner surface of the furnace wall tube, the temperature rises and there is a possibility of causing a malfunction such as damage to the furnace wall tube due to overheating.
Patent Document 1 describes, as a related technique, a technique for detecting an abnormal temperature increase in a furnace wall tube and preventing occurrence of a malfunction event such as damage to the furnace wall tube.
Patent Document 2 describes a technique for predicting the thickness of hematite scale deposition in a furnace wall tube as a related technique.

JP-A-2005-221170 JP 2011-214764 A

By the way, in a plant equipped with the boiler as described above, a malfunction event such as damage to the furnace wall tube caused by hematite scale deposition is likely to occur at a portion of the boiler having the highest thermal load. However, the phenomenon of hematite scale deposition has not been elucidated, and failure events may occur outside the region of the boiler where the heat load is highest. Therefore, it is difficult to predict the occurrence location of the malfunction event, and when using the technique described in Patent Document 1, it is difficult to determine the installation location of the thermometer.
In addition, even when using the technique described in Patent Document 2, it is necessary to specify the furnace wall tube part to be predicted for the thickness of the hematite scale deposition, and to measure the temperature, pressure, and iron concentration. It is difficult to predict and measure where an event occurs.

The time of occurrence of a malfunction event depends on plant operating conditions such as the magnitude of load and water quality. Therefore, it is difficult to predict the occurrence time of a malfunction event from only the total operation time. In general, the pre-measures for the occurrence of a failure event due to hematite scale deposition in the furnace wall tube are periodically taken out of the furnace wall tube and observed in cross-section to measure the thickness of the hematite scale deposition. This is a measure to remove the hematite scale by performing chemical cleaning when the thickness exceeds the thickness.
Therefore, in a plant equipped with a boiler, there has been a demand for a technique that can predict the occurrence time of a malfunction event without extracting the furnace wall tube.

  Therefore, an object of the present invention is to provide a failure event occurrence prediction device, a failure event occurrence prediction method, and a program that can solve the above-described problems.

  In the first aspect, a factor extracting unit that extracts a plurality of factors correlated with the occurrence of a plant malfunction event from the collected plant data by statistical analysis, the factor extracted by the factor extraction unit, and a malfunction event in the plant Based on the presence or absence of occurrence, a weight calculation unit that calculates the contribution rate of each factor to the occurrence of a failure event, each factor extracted by the factor extraction unit, and each contribution rate calculated by the weight calculation unit A defect derivation unit that derives a prediction formula for determining whether or not a malfunction event has occurred based on the formula, and a malfunction event occurrence prediction unit that predicts the occurrence of a malfunction event based on the prediction formula derived by the formula derivation unit It is an event occurrence prediction device.

  Moreover, a 2nd aspect is a malfunction event generation | occurrence | production prediction apparatus which is a malfunction event resulting from hematite scale depositing on the furnace wall pipe inner surface of a boiler in the 1st aspect.

  Further, the third mode is the second mode, wherein the equation deriving unit is based on the economizer entrance iron concentration and the contribution rate of the economizer entrance iron concentration to the occurrence of the malfunction event. It is a malfunction event occurrence prediction device for deriving a prediction formula.

  According to a fourth aspect, in the second or third aspect, the equation deriving unit derives the prediction equation based on an operation time and a contribution rate of the operation time to the occurrence of the malfunction event. It is a malfunction event occurrence prediction device.

  Further, according to a fifth aspect, in any one of the second to fourth aspects, the formula deriving unit is based on an average mass flow rate and a contribution rate of the average mass flow rate to the occurrence of the malfunction event. It is a malfunction event occurrence prediction device for deriving a prediction formula.

  Further, according to a sixth aspect, in any one of the second to fifth aspects, the formula deriving unit determines the prediction formula based on an oxygen concentration and a contribution rate of the oxygen concentration to the occurrence of the malfunction event. Is a failure event occurrence prediction device for deriving.

  Further, according to a seventh aspect, in any one of the second to sixth aspects, the formula deriving unit is based on a main steam iron concentration and a contribution rate of the main steam iron concentration to the occurrence of the malfunction event. A failure event occurrence prediction device for deriving the prediction formula.

  Further, according to an eighth aspect, in any one of the second to seventh aspects, the formula deriving unit derives the prediction formula based on pH and a contribution rate of the pH to the occurrence of the malfunction event. This is a malfunction event occurrence prediction device.

  Further, according to a ninth aspect, in any one of the second to eighth aspects, the formula deriving unit is configured to calculate the prediction formula based on the number of activations and a contribution rate of the activation number to the occurrence of the malfunction event. Is a failure event occurrence prediction device for deriving.

  Further, a tenth aspect is any one of the first to ninth aspects, wherein the malfunction event occurrence prediction unit predicts a malfunction event occurrence time based on the prediction formula derived by the formula deriving unit, This is a failure event occurrence prediction device that promotes precautions against occurrence of a failure event.

  The eleventh aspect is any one of the first to tenth aspects, wherein the malfunction event occurrence prediction unit predicts a malfunction event occurrence time based on the prediction formula derived by the formula deriving unit, A malfunction event occurrence prediction device that changes a value of an operation factor indicating an operation condition so as to delay occurrence of a malfunction event.

  In a twelfth aspect according to any one of the first to eleventh aspects, the equation deriving unit derives the prediction equation for each part of the plant, and the failure event occurrence predicting unit derives the equation. This is a malfunction event occurrence prediction device that predicts a site where a malfunction event has occurred based on the prediction formula derived by the unit, and urges advance measures against the occurrence of the malfunction event.

  A thirteenth aspect is a malfunction event occurrence prediction device including a notifying unit for notifying a user of information on occurrence of a malfunction event predicted by the malfunction event occurrence prediction unit in any of the first to twelfth aspects. .

  Further, the fourteenth aspect is the step of extracting a plurality of factors correlated with the occurrence of a plant malfunction event from the collected plant data by statistical analysis, the factor extracted by the step of extracting the plurality of factors, Based on the presence or absence of a failure event occurrence in the plant, based on the step of calculating the contribution rate of each of the factors to the failure event occurrence, the step of extracting the plurality of factors and the step of calculating the contribution rate, A plant failure event occurrence prediction method comprising: a step of deriving a prediction expression for determining whether or not a failure event has occurred; and a step of predicting a failure event occurrence based on the prediction expression derived by the step of deriving the prediction expression. is there.

  In the fifteenth aspect, the computer is configured to extract, from a collected plant data, a plurality of factors correlated with the occurrence of a malfunction event of the plant by statistical analysis, and the factors extracted by the factor extracting unit, Based on the presence / absence of occurrence of a failure event in the plant, the weight calculation means for calculating the contribution rate of each factor to the failure event occurrence, the factor extracted by the factor extraction means and the weight calculation means A formula for deriving a prediction formula for determining whether or not a fault event has occurred based on each contribution rate, and a fault event occurrence prediction for predicting a fault event based on the prediction formula derived by the formula derivation means It is a program that functions as a means.

  According to the malfunction event occurrence predicting apparatus according to the embodiment of the present invention, the malfunction event occurrence timing can be predicted in a plant including a boiler without extruding the furnace wall pipe.

It is a figure which shows an example of the systematic diagram of the boiler. It is a figure which shows an example of a structure of the malfunction event occurrence prediction apparatus 2 by 1st embodiment. It is a figure which shows an example of the plant data of the thermal power plant which has the boiler. It is a figure which shows an example of the correlation of the factors used when narrowing down the factor with a strong correlation. It is a figure which shows an example of the processing flow of the malfunction event occurrence prediction apparatus 2 by 1st embodiment. It is a figure which shows an example of the malfunction event occurrence time estimated from the prediction formula. It is a figure which shows an example of the generation | occurrence | production risk of the malfunction event with respect to actual total operation time. It is a figure which shows an example of the processing flow of the malfunction event occurrence prediction apparatus 2 by 2nd embodiment. It is a figure which shows an example of the malfunction event generation | occurrence | production time which the malfunction event occurrence prediction apparatus 2 estimated.

  Hereinafter, embodiments will be described in detail with reference to the drawings. Below, the example which applies a malfunction event occurrence prediction apparatus to the boiler 1 is demonstrated.

<First embodiment>
First, the boiler 1 will be described.
FIG. 1 is a diagram illustrating an example of a system diagram of the boiler 1.
The boiler 1 includes a superheater 10, a high pressure turbine 20, a reheater 30, an intermediate pressure turbine 40, a low pressure turbine 50, a condenser 60, a condensate pump 70, and a condensate demineralizer 80. The condensate boost pump 90, the ground steam condenser 100, the low pressure heater 110, the deaerator 120, the boiler feed pump 130, the high pressure heater 140, the economizer 150, the furnace wall pipe 160, A steam-water separator 170.

The superheater 10 heats the saturated steam in the steam separator 170, thereby increasing only the temperature while keeping the pressure constant, and generates superheated steam having higher thermal energy.
The high-pressure turbine 20 takes out the thermal energy of the superheated steam and converts it into mechanical energy such as rotation.
The reheater 30 returns the steam used as energy for rotating the high-pressure turbine 20 to the boiler and reheats it. The reheater 30 supplies the reheated steam to the intermediate pressure turbine 40. Thereby, thermal efficiency can be improved.
The intermediate pressure turbine 40 takes out the thermal energy of the steam heated by the reheater 30 and converts it into mechanical energy such as rotation.
The low-pressure turbine 50 takes out the thermal energy of the steam discharged from the intermediate-pressure turbine 40 and converts it into mechanical energy such as rotation.
The condenser 60 cools the steam discharged from the low-pressure turbine 50 with seawater or the like and returns it to water.
The condensate pump 70 sucks the condensate accumulated in the condenser 60. The condensate pump 70 sends the sucked condensate to the condensate demineralizer 80.
The condensate demineralizer 80 removes foreign matter in the condensate sent from the condensate pump 70.
The condensate booster pump 90 raises the pressure of the water from which the condensate demineralizer 80 has removed foreign matter. The condensate booster pump 90 sends the water whose pressure has been increased to the ground steam condenser 100.
The ground steam condenser 100 condenses the leaked steam from the ground (packing presser) and returns it to the water.
The low-pressure heater 110 is a heat exchanger in the water supply system on the condenser side of the boiler water supply pump 130. The low-pressure heater 110 heats the boiler feed water by extracting air from the turbine.
The deaerator 120 directly heats the feed water with steam and physically separates and removes dissolved gas in the feed water.
The boiler feed water pump 130 is a pump that raises the pressure of feed water and pushes it into the boiler.
The high-pressure heater 140 is a heat exchanger in the water supply system on the boiler side with respect to the boiler water supply pump 130. The high-pressure heater 140 heats the boiler feed water by extracting air from the turbine.
The economizer 150 uses the heat of the boiler exhaust gas to heat the boiler feed water. Thereby, boiler efficiency is improved.
The furnace wall tube 160 is a tube through which hot water and steam pass. Hematite scale is deposited on the inner surface of the furnace wall tube 160.
The steam separator 170 forcibly separates the drain that is generated and mixed in the steam pipe at the time of start-up, stop, and low load.

Next, the failure event occurrence prediction device 2 according to the first embodiment will be described.
FIG. 2 is a diagram illustrating an example of the configuration of the failure event occurrence prediction device 2 according to the first embodiment.
The failure event occurrence prediction device 2 includes a factor extraction unit 201, a weight calculation unit 202, an expression derivation unit 203, a failure event occurrence prediction unit 204, and a notification unit 205.
In FIG. 2, reference numeral 3 denotes a storage unit. The memory | storage part 3 has memorize | stored the plant data containing the operation data which the user collected about the some thermal power plant which has the boiler 1, design data, and the presence or absence of malfunction event generation | occurrence | production.
The factor extraction unit 201 reads out plant data collected from the thermal power plant having the boiler 1 from the storage unit 3. The factor extraction unit 201 extracts a plurality of factors correlated with the occurrence of a malfunction event in the thermal power plant from the plant data read from the storage unit 3 by statistical analysis. For example, the occurrence of a malfunction event in a thermal power plant is that hematite scale accumulates on the inner surface of the furnace wall tube of the boiler 1 of the thermal power plant.
The weight calculation unit 202 calculates the contribution rate of each factor to the occurrence of a malfunction event based on the factor extracted from the plant data by the factor extraction unit 201 and the presence or absence of the malfunction event in the thermal power plant.
The formula deriving unit 203 derives a prediction formula for determining whether or not a malfunction event has occurred based on each factor extracted from the plant data by the factor extracting unit 201 and each contribution rate calculated by the weight calculating unit 202.
The failure event occurrence prediction unit 204 predicts the occurrence of a failure event based on the prediction formula derived by the equation deriving unit 203.
The notification unit 205 notifies the user of information on the occurrence of the failure event predicted by the failure event occurrence prediction unit 204.

FIG. 3 is a diagram illustrating an example of plant data of a thermal power plant having the boiler 1.
Plant data includes operating factors that indicate operating conditions that can be changed during the operation of the thermal power plant and the values of the factors (operating data), design factors that indicate design conditions that cannot be changed during the operation determined for the design, and factors Value (design data) and information on the presence / absence of a failure event. For example, the operating factors are operating time (cumulative operating time after CWT conversion, cumulative operating time after chemical cleaning, operating time at the time of failure event), average mass flow rate, pH, oxygen concentration, average iron concentration at economizer, rating These include the maximum value of iron concentration at the entrance of time saving coal, the average concentration of main steam iron, and the number of start / stop times (start-up times). The design factors are boiler structure (spiral type, vertical pipe type, etc.), pipe outer diameter, number of pipes, burner structure (swirl combustion, counter combustion, etc.), furnace width, and the like.
The failure event occurrence prediction device has a significant difference with respect to the presence or absence of a failure event based on the collected plant data indicated by each factor and the presence or absence of occurrence of the failure event in the thermal power plants A to D as shown in FIG. Extract a factor. For example, in FIG. 3, the boiler structure that is one of the factors can be classified into two groups, a spiral type and a vertical tube type. For factors that can be classified into two groups, the occurrence of a malfunction event using a method called t-test that examines whether there is a statistically significant difference between the average values obtained from the samples of the two groups Whether or not there is an association between the factors Moreover, in FIG. 3, pH, a condensate condenser, etc. which are one of the factors take three or more values and can be classified into three or more groups. For factors that can be classified into three or more groups, we use a method called analysis of variance to examine whether there is a statistically significant difference in the mean values obtained from the samples of the three groups. Determine if there is an association between the event occurrence and the factor.

Further, the malfunction event occurrence prediction device narrows down factors having a strong correlation based on a simple correlation between factors.
FIG. 4 is a diagram illustrating an example of correlation between factors used when narrowing down a factor having a strong correlation.
In FIG. 4, factors are arranged in the vertical direction and the horizontal direction, respectively. The numerical value at the intersection of the vertical factor and the horizontal factor indicates the strength of the correlation between the factors. The numerical value is 1 or less, and the closer to 1, the stronger the correlation between factors. When it is determined that the correlation between factors is strong, the failure event occurrence prediction device assigns the same identifier to each of the factors, and performs statistical analysis by regarding those factors as one factor. In FIG. 4, the numerical value indicated by hatching is 0.8 or more, indicating a strong correlation. For the numerical values shown in FIG. 4, for example, the numerical value indicating the correlation between the factor 1 and the factor 9 is 0.9222, the numerical value indicating the correlation between the factor 1 and the factor 10 is 0.8154, the factor 9 and the factor 10 are The numerical value indicating the correlation is strong as 0.8938, and the factor 1, factor 9, and factor 10 can be regarded as one factor.

FIG. 5 is a diagram illustrating an example of a processing flow of the failure event occurrence prediction device 2 according to the first embodiment.
Next, processing of the failure event occurrence prediction device 2 that performs the failure event occurrence prediction in the thermal power plant including the boiler 1 according to the first embodiment will be described.
The factor extraction unit 201 included in the failure event occurrence prediction device 2 according to the first embodiment reads the plant data collected for the thermal power plant having the boiler 1 from the storage unit 3 (step S1). The factor extraction unit 201 extracts, from the plant data read from the storage unit 3, a plurality of factors that are correlated with the occurrence of a malfunction event in the thermal power plant by statistical analysis (step S2). For example, the factor extraction unit 201 determines that the factor 1 has a correlation with the occurrence of the malfunction event of the thermal power plant when the malfunction event occurs when the value of the factor 1 is large and the malfunction event does not occur when the value of the factor 1 is small. To do. In addition, the factor extraction unit 201 does not change the result that the failure event does not occur regardless of the value of the factor 2 when the failure event does not occur when the value of the factor 2 is large or small. 2 determines that there is no correlation with the occurrence of a malfunction event in the thermal power plant. Further, the factor extraction unit 201 does not change the result that the failure event occurs regardless of the value of the factor 3 when the failure event occurs when the value of the factor 3 is large or small. 3 determines that there is no correlation with the occurrence of a malfunction event in the thermal power plant. However, in practice, the factor extraction unit 201 extracts a factor that correlates with the occurrence of a malfunction event in the thermal power plant for a condition in which a plurality of factors are combined in a complex manner. Therefore, the factor extraction unit 201 extracts correlated factors using a statistical analysis method such as t-test or analysis of variance.
The factor extraction unit 201 narrows down factors having a strong correlation based on a single correlation between a plurality of factors extracted from plant data by statistical analysis (step S3). For example, the factor extraction unit 201 obtains the correlation between the factors illustrated in FIG. 4 by cluster analysis or the like. Then, the factor extraction unit 201 assigns the same identifier to each of the factors that showed a strong correlation, and performs statistical analysis by regarding those factors as one factor.
The factor extraction unit 201 narrows down a plurality of factors that are correlated with the occurrence of a malfunction event to about four. Whether or not the factor narrowed down by the factor extraction unit 201 is appropriate is determined by a test for verifying whether or not a prediction formula generated later is appropriate.
The factor extraction unit 201 calculates a plurality of factors extracted and narrowed down from plant data when a failure event occurs by statistical analysis, and a plurality of factors extracted and narrowed down from plant data when a failure event does not occur To the unit 202 and the expression deriving unit 203.

When a plurality of factors narrowed down from the factor extraction unit 201 and the presence / absence of occurrence of a malfunction event in the thermal power plant are input, the weight calculation unit 202 is based on the input multiple factors and the occurrence / absence of malfunction event in the thermal power plant. Then, the contribution ratio of each factor to the occurrence of a malfunction event is calculated. For example, the weight calculation unit 202 assumes a linear relationship between a plurality of independent variables and dependent variables, and uses the multiple regression analysis that predicts the value of the dependent variable from the plurality of independent variables to obtain the results indicated by the plant data. Calculate the contribution ratio of each applicable factor to the occurrence of a malfunction event. More specifically, the weight calculation unit 202 assumes 1 = a × factor 1 + b × factor 2 + c × factor 3+... As a prediction formula for the occurrence of a malfunction event. Further, the weight calculation unit 202 assumes 0 = a × factor 1 + b × factor 2 + c × factor 3+... As a prediction formula for a case where no malfunction event occurs. Then, the weight calculation unit 202 uses multiple regression analysis on the assumed prediction formula 1 = a × factor 1 + b × factor 2 + c × factor 3+. Each of the coefficients a, b, c,... Is calculated. Further, the weight calculation unit 202 uses a multiple regression analysis on the prediction formula represented by 0 = a × factor 1 + b × factor 2 + c × factor 3+. , B, c,... Are calculated.
The weight calculation unit 202 outputs the calculated coefficient of each factor to the formula deriving unit 203.

The expression deriving unit 203 inputs a plurality of factors narrowed down from the factor extracting unit 201 and the presence / absence of occurrence of a malfunction event, and inputs coefficients of a plurality of factors narrowed down from the weight calculation unit 202. A prediction formula for predicting the occurrence of the malfunction event is derived based on the presence / absence of the malfunction event and the coefficient of each of the plurality of factors (step S4). When deriving an expression for calculating the driving time until the occurrence of a malfunction event, the weighting calculation unit 202 uses a prediction formula in which one of a plurality of factors is the driving time, that is, 1 = a × factor 1 + b × factor. 2 + c × operation time +... And 0 = a × factor 1 + b × factor 2 + c × operation time +. Then, the weight calculation unit 202 calculates the coefficient of each factor using multiple regression analysis on the assumed prediction formula. The weight calculation unit 202 outputs the calculated coefficient of each factor to the formula deriving unit 203. The formula deriving unit 203 calculates the operation time until the occurrence of the malfunction event based on the plurality of factors input from the factor extraction unit 201, the presence / absence of occurrence of the malfunction event, and the coefficients of the plurality of factors input from the weight calculation unit 202. A prediction formula to be calculated is derived. That is, the operation time until the occurrence of the malfunction event can be expressed as operation time = (a1 × factor 1 + b1 × factor 2+... −1) ÷ c1.
If the failure event in the thermal power plant is a failure event due to the hematite scale, factors that may have a large impact on the occurrence of the failure event are the concentration of iron in the economizer, operating time, average mass flow rate, oxygen concentration , Main steam iron concentration, pH, number of start-ups, etc.

The formula deriving unit 203 determines whether or not the derived prediction formula is valid using a multiple regression analysis test (step S5). For example, the formula deriving unit 203 substitutes plant data for the derived prediction formula, and determines whether or not the derived prediction formula is valid based on whether or not a failure event occurs for the plant data. To do.
When the formula deriving unit 203 determines that the derived prediction formula is not valid (NO in step S5), the formula deriving unit 203 returns to the process of step S1 and derives the prediction formula again. For example, the formula deriving unit 203 derives a prediction formula again using a plurality of factors regarded as one factor as separate factors. Further, the formula deriving unit 203 derives the prediction formula again by adding the factors removed at the time of narrowing down the factors.
When the formula deriving unit 203 determines that the derived prediction formula is valid (step S5, YES), the formula deriving unit 203 determines whether new malfunction event occurrence information has been added (step S6). That is, the formula deriving unit 203 determines whether or not a new malfunction event is added and a prediction formula for predicting the occurrence of the malfunction event is derived.
When the formula deriving unit 203 determines that new malfunction event occurrence information has been added (step S6, YES), the formula deriving unit 203 returns to the process of step S1 and derives a prediction formula for the new malfunction event.
If the formula deriving unit 203 determines that new malfunction event occurrence information has not been added (NO in step S6), the formula deriving unit 203 outputs the prediction formula derived in the process of step S5 to the malfunction event occurrence prediction unit 204.

  When the prediction event is input from the equation deriving unit 203, the failure event occurrence prediction unit 204 predicts the occurrence of the failure event based on the input prediction equation (step S7). For example, the failure event occurrence predicting unit 204 predicts the failure event occurrence time based on the operation time indicating the operation time until the occurrence of the failure event = (a1 × factor 1 + b1 × factor 2+... −1) ÷ c1.

FIG. 6 is a diagram illustrating an example of the failure event occurrence time predicted from the prediction formula.
In FIG. 6, the vertical axis indicates the failure event occurrence time predicted from the prediction formula. The horizontal axis indicates the actual operation time of the thermal power plant. Further, in FIG. 6, a failure event occurrence line in which the failure event occurrence prediction time and the actual total operation time coincide with each other, and a chemical cleaning recommendation line indicating a time when chemical cleaning should be performed as a countermeasure before the failure event occurs. Is shown.
In the example shown in FIG. 6, the total operation time of the thermal power plant A is ta1, and it is presumed that long-term sound operation is possible until the failure event occurrence prediction time indicated by ta2. In addition, the total operation time of the thermal power plant C is tc1, and it is presumed that some countermeasures such as chemical cleaning will be necessary soon.
Therefore, when the failure event occurrence prediction unit 204 predicts the occurrence of a failure event in the thermal power plant C as shown in FIG. 6, the notification unit 205 takes some measures for the user before the failure event occurs. It is necessary to encourage them. For example, the notification unit 205 may notify the user of the recommended time for performing the chemical cleaning shown in FIG.
Note that the recommended timing of implementation of the countermeasure notified by the notification unit 205 is, for example, the period of the periodic inspection before the time when the occurrence of the malfunction event is predicted. In addition, the recommended timing of implementation of countermeasures notified by the notification unit 205 is, for example, a function indicating the risk of occurrence of a failure event with respect to the actual total operation time as shown in FIG. The event occurrence risk may be regarded as 1, and any measure may be a time indicating the actual total operation time when the failure event occurrence risk is 0.8. In addition, the recommended timing for implementing measures may be a time set based on the past results of measures.

When the failure event occurrence prediction unit 204 performs the process of step S <b> 7, the failure event occurrence prediction unit 204 outputs the predicted failure event occurrence time to the notification unit 205.
When the failure event occurrence time is input from the failure event occurrence prediction unit 204, the notification unit 205 notifies the user of the input failure event occurrence time (step S8).
For example, the notification unit 205 includes any one of a display unit, a speaker, and a vibration unit. When the input failure event occurrence time is one year later, the notification unit 205 displays “There is a possibility that a failure event will occur within one year” on the display unit. Further, the notification unit 205 may output a synthesized voice “There is a possibility that a malfunction event may occur within one year” from the speaker. Further, the notification unit 205 may notify the failure event occurrence time by the vibration of the vibration unit according to the failure event occurrence time (for example, the vibration of the vibration unit becomes longer and stronger as the failure event occurrence time approaches). .
A user who recognizes the time of occurrence of a malfunction event may take appropriate measures before the malfunction event occurs. For example, when the malfunction event of the thermal power plant is a malfunction event caused by hematite scale depositing on the inner wall of the furnace wall tube of the boiler, chemical cleaning is performed as a countermeasure. As a countermeasure, the load may be reduced. Further, as a countermeasure, a filter for filtering the hematite scale may be installed to prevent it from being brought into the boiler.
By doing so, the user can recognize the time of occurrence of the trouble event, and can take appropriate measures before the trouble event occurs.

The trouble event occurrence prediction device 2 according to the first embodiment of the present invention has been described above. In the malfunction event occurrence prediction device 2 described above, the factor extraction unit 201 extracts, from the collected plant data, a plurality of factors correlated with the occurrence of the malfunction event of the thermal power plant by statistical analysis. The weight calculation unit 202 calculates the contribution rate of each factor to the occurrence of a malfunction event based on the factor extracted from the plant data by the factor extraction unit 201 and the presence or absence of the malfunction event in the thermal power plant. The formula deriving unit 203 derives a prediction formula for determining whether or not a malfunction event has occurred based on each factor extracted from the plant data by the factor extracting unit 201 and each contribution rate calculated by the weight calculating unit 202. The failure event occurrence prediction unit 204 predicts the occurrence of a failure event based on the prediction formula derived by the equation deriving unit 203.
By carrying out like this, in a thermal power plant provided with the boiler 1, a malfunction event generation | occurrence | production time can be estimated, without extracting a furnace wall pipe.
In the failure event occurrence prediction device 2, the notification unit 205 notifies the user of the information on the occurrence of the failure event predicted by the failure event occurrence prediction unit 204.
By doing so, the user can take an appropriate measure at an appropriate time so as not to cause a malfunction event in the thermal power plant. As a result, the occurrence of a malfunction event in the thermal power plant can be prevented.

<Second Embodiment>
Next, the process of the failure event occurrence prediction device 2 that performs failure event occurrence prediction in a thermal power plant including the boiler 1 according to the second embodiment will be described.
The failure event occurrence prediction device 2 according to the second embodiment has the same configuration as the failure event occurrence prediction device 2 according to the first embodiment shown in FIG.

FIG. 8 is a diagram illustrating an example of a processing flow of the failure event occurrence prediction device 2 according to the second embodiment.
The processing flow of the failure event occurrence prediction device 2 according to the second embodiment is that when the user changes the operation condition in order to delay the occurrence time of the failure event of the thermal power plant, the failure event occurrence prediction is performed to determine the operation factor. It is a processing flow when the apparatus 2 is used.
Steps S1 to S8 of the processing flow of the malfunction event occurrence prediction device 2 according to the second embodiment are the same as the processes of Step 1 to Step S8 in the processing flow of the malfunction event occurrence prediction device 2 according to the first embodiment. .
Here, the failure event occurrence prediction device 2 according to the second embodiment predicts the failure event occurrence timing obtained by the process of step S7 and the failure event occurrence prediction device 2 performs the prediction result in step S9. The process of step S10 will be described.

First, the prediction result of the failure event occurrence time obtained by the process of step S7 performed by the failure event occurrence prediction apparatus 2 according to the second embodiment will be described.
FIG. 9 is a diagram illustrating an example of a failure event occurrence time predicted by the failure event occurrence prediction device 2.
In FIG. 9, the vertical axis represents the average mass flow rate [10 ³ t / hm ² ]. The horizontal axis represents the economizer inlet iron concentration [ppb (parts per bill)]. In FIG. 9, the relationship between the cumulative operation time after the oxygen treatment is applied to the six thermal power plants A to F and the malfunction event occurrence timing based on the initial operation start timing is indicated by circles.
In FIG. 9, the vertical axis represents the average mass flow rate, which is one of the design conditions. Moreover, the horizontal axis has shown the economizer entrance iron concentration which is one of the operating conditions.
Moreover, in FIG. 9, the actual total operation time and malfunction event occurrence timing are shown for each of the thermal power plants A to F. The time shown in the upper row is the actual total operation time. Moreover, the time shown in the lower part for each thermal power plant is the time of occurrence of the malfunction event.

Thermal power plants C to F are known to be thermal power plants in which a malfunction event has already occurred from actual operation results. It can be seen that in the thermal power plants C to F, the actual total operation time and the failure event occurrence time coincide with each other within an error range of about 1 percent.
Thermal power plants A and B are known to be thermal power plants in which no malfunction event has occurred from actual operation results. In FIG. 9, paying attention to the thermal power plant A, it can be seen that when the current operating conditions are continued, the malfunction event occurs at approximately the middle of 10 years and 20 years, approximately 15 years. Here, for example, if the thermal power plant A has been operating for 14 years, a malfunction event will occur after about one year. In such a case, it is necessary for the thermal power plant A to perform inspection and countermeasures so that a malfunction event does not occur within one year. However, inspection and countermeasures within one year may be difficult due to constraints such as budget and operation plan. In this case, a request for changing the operating condition and delaying the occurrence time of the malfunction event by, for example, one year occurs. At this time, in order to delay the malfunction event occurrence time by one year, that is, to operate the thermal power plant for one year longer, it is possible to specify which operating factor should be set to which value and actually operate the thermal power plant. It is necessary to consider whether or not.
In such a case, it is only necessary to specify which driving factor should be set to which value by the failure event occurrence prediction device 2 performing the processes of step S9 and step S10. Moreover, the user should just examine whether the driving | operation is actually possible based on the setting which the malfunction event occurrence prediction apparatus 2 specified.

Next, the process of step S9 and step S10 performed by the malfunction event occurrence prediction device 2 will be described.
It is assumed that the failure event occurrence prediction unit 204 provided in the failure event occurrence prediction device 2 predicts that a failure event will occur within a period shorter than the operation period desired by the user in the process of step S7.
The failure event occurrence prediction unit 204 outputs the predicted occurrence time of the failure event to the notification unit 205.
When the failure event occurrence time is input from the failure event occurrence prediction unit 204, the notification unit 205 notifies the user of the input failure event occurrence time in the process of step S8.

  The failure event occurrence prediction unit 204 determines whether or not the operation period desired by the user is satisfied based on the predicted occurrence time of the failure event (step S9). For example, the user records a desired driving period in the storage unit 3 in advance. The failure event occurrence prediction unit 204 reads out the operation period desired by the user from the storage unit 3, and based on whether or not the predicted failure event occurrence time is later than the operation period desired by the user, It is determined whether or not the desired operation period is satisfied.

  If it is determined that the predicted failure event occurrence time is later than the operation period desired by the user, the failure event occurrence prediction unit 204 determines that the operation period desired by the user is satisfied (step S9, YES). . Then, the malfunction event occurrence prediction device 2 ends the series of processes.

If the failure event occurrence prediction unit 204 determines that the predicted failure event occurrence time is within the operation period desired by the user or before the operation period desired by the user, the operation period desired by the user is not satisfied. (Step S9, NO). In this case, the malfunction event occurrence predicting unit 204 identifies the driving factor so as to satisfy the driving period desired by the user, and changes the value of the driving factor indicating the driving condition (step S10). For example, the user records the priority order and the value range of the driving factor for a plurality of driving factors. If the failure event occurrence prediction unit 204 determines that the operation period desired by the user is not satisfied in the process of step S <b> 9, the failure event occurrence prediction unit 204 reads the priority order of the operation factors and the range of the operation factor values from the storage unit 3. Then, the failure event occurrence prediction unit 204 changes the driving factor within the range of values in descending order of priority. For example, the malfunction event occurrence prediction unit 204 performs a change to reduce the concentration of iron at the economizer.
Then, the malfunction event occurrence prediction unit 204 returns to the process of step S7 and predicts the malfunction event occurrence time with respect to the changed value of the driving factor.
The processing of step S7 to step S10 in the malfunction event occurrence prediction device 2 is repeated until the operation period desired by the user in step S9 is satisfied.

The trouble event occurrence prediction device 2 according to the second embodiment of the present invention has been described above. In the malfunction event occurrence prediction device 2 described above, the factor extraction unit 201 extracts, from the collected plant data, a plurality of factors correlated with the occurrence of the malfunction event of the thermal power plant by statistical analysis. The weight calculation unit 202 calculates the contribution rate of each factor to the occurrence of a malfunction event based on the factor extracted from the plant data by the factor extraction unit 201 and the presence or absence of the malfunction event in the thermal power plant. The formula deriving unit 203 derives a prediction formula for determining whether or not a malfunction event has occurred based on each factor extracted from the plant data by the factor extracting unit 201 and each contribution rate calculated by the weight calculating unit 202. The failure event occurrence prediction unit 204 predicts the occurrence of the failure event based on the prediction formula derived by the equation deriving unit 203.
By carrying out like this, in a thermal power plant provided with a boiler, a malfunction event generation | occurrence | production time can be estimated, without extracting a furnace wall pipe.
Further, in the above-described failure event occurrence prediction device 2, the failure event occurrence prediction unit 204 determines whether or not the operation period desired by the user is satisfied based on the predicted occurrence time of the failure event. In addition, when it is determined that the operation period desired by the user is not satisfied, the failure event occurrence prediction unit 204 changes the value of the operation factor so as to satisfy the operation period desired by the user.
By doing so, it is possible to set the value of the operating factor for changing the malfunction event occurrence time. As a result, the occurrence of a malfunction event in the thermal power plant can be prevented.

<Third embodiment>
Next, processing of the failure event occurrence prediction device 2 that performs failure event occurrence prediction in a thermal power plant including the boiler 1 according to the third embodiment will be described.
The failure event occurrence prediction device 2 according to the third embodiment has the same configuration as the failure event occurrence prediction device 2 according to the first embodiment shown in FIG.
The processing flow of the malfunction event occurrence prediction device 2 according to the third embodiment is the same as the process flow of the malfunction event occurrence prediction device 2 according to the first embodiment shown in FIG.
However, the prediction formula derived in step S4 by the failure event occurrence prediction device 2 according to the third embodiment is a prediction equation derived for a factor different from the operation time until the failure event occurs. In 1st embodiment, the memory | storage part 3 memorize | stored the plant data of the presence or absence of malfunction event generation | occurrence | production with respect to the factor value for every plant as shown in FIG. On the other hand, in 3rd embodiment, the memory | storage part 3 memorize | stores design data and operation data, such as a boiler level, for example instead of the design data and operation data of plant data. The factors in the design data and operation data such as the boiler level are different from the factors in the design data and operation data of the plant data.
The formula deriving unit 203 provided in the failure event occurrence prediction device 2 reads design data such as boiler level and operation data from the storage unit 3, and uses the design data to derive a prediction formula for each part where the failure event actually occurs. To do.
By doing this, it is possible to specify in more detail the factors that affect the occurrence of the malfunction event. And the site | part which a malfunction event tends to generate | occur | produce can be estimated. As a result, when designing a thermal power plant, it is possible to improve the design in consideration of a site where a trouble event is likely to occur. In addition, the occurrence of a malfunction event can be prevented by changing the part where the malfunction event is likely to occur after the design to a member made of a material that is less likely to cause the malfunction event. In addition, when the furnace wall tube is extracted and the cross section is observed, the tube can be extracted at a more appropriate position than in the past. Furthermore, as in the techniques described in Patent Document 1 and Patent Document 2, when the occurrence of a malfunction event is prevented based on actual measurement values, a measuring instrument such as a thermometer can be installed at a more appropriate location.

The malfunction event occurrence prediction device 2 according to the third embodiment of the present invention has been described above. In the malfunction event occurrence prediction device 2 described above, the factor extraction unit 201 extracts, from the collected plant data, a plurality of factors correlated with the occurrence of the malfunction event of the thermal power plant by statistical analysis. The weighting calculation unit 202 calculates the contribution rate of each factor to the occurrence of a malfunction event based on the factor extracted from the plant data by the factor extraction unit 201 and the presence or absence of the malfunction event in the thermal power plant. The formula deriving unit 203 derives a prediction formula for determining whether or not a malfunction event has occurred based on each factor extracted from the plant data by the factor extracting unit 201 and each contribution rate calculated by the weight calculating unit 202. The failure event occurrence prediction unit 204 predicts the occurrence of the failure event based on the prediction formula derived by the equation deriving unit 203.
By carrying out like this, in a thermal power plant provided with a boiler, a malfunction event generation | occurrence | production time can be estimated, without extracting a furnace wall pipe.
In the above-described malfunction event occurrence prediction device 2, the formula deriving unit 203 reads design data such as the boiler level from the storage unit 3, and uses the design data to predict each part where the malfunction event actually occurred. Derive an expression.
By doing this, it is possible to specify in more detail the factors that affect the occurrence of the malfunction event. And the site | part which a malfunction event tends to generate | occur | produce can be estimated. As a result, when designing a thermal power plant, it is possible to improve the design in consideration of a site where a trouble event is likely to occur. In addition, the occurrence of a malfunction event can be prevented by changing the part where the malfunction event is likely to occur after the design to a member made of a material that is less likely to cause the malfunction event. In addition, when the furnace wall tube is extracted and the cross section is observed, the tube can be extracted at a more appropriate position than in the past. Furthermore, as in the techniques described in Patent Document 1 and Patent Document 2, when the occurrence of a malfunction event is prevented based on actual measurement values, a measuring instrument such as a thermometer can be installed at a more appropriate location.

  Note that the storage unit 3 in the present invention may be provided anywhere within a range in which appropriate information is transmitted and received. Further, the storage unit 3 may exist in a plurality of ranges within a range where appropriate information is transmitted and received, and may store data in a distributed manner.

  In the processing flow according to the embodiment of the present invention, the order of processing may be changed within a range where appropriate processing is performed.

  In addition, although embodiment of this invention was described, the above-mentioned malfunction event generation | occurrence | production prediction apparatus 2 has a computer system inside. The process described above is stored in a computer-readable recording medium in the form of a program, and the above process is performed by the computer reading and executing this program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. Various omissions, replacements, and changes can be made without departing from the scope of the invention.

DESCRIPTION OF SYMBOLS 1 ... Boiler 2 ... Failure event generation | occurrence | production prediction apparatus 3 ... Memory | storage part 10 ... Superheater 20 ... High pressure turbine 30 ... Reheater 40 ... Medium pressure turbine 50 ... Low-pressure turbine 60 ... condenser 70 ... condensate pump 80 ... condensate demineralizer 90 ... condensate booster pump 100 ... ground steam condenser 110 ... low-pressure heater 120 Degasser 130 ... Boiler feed pump 140 ... High pressure heater 150 ... Carrier 160 ... Furnace wall tube 170 ... Steam separator 201 ... Factor extraction unit 202 ... Weight calculation unit 203... Formula derivation unit 204... Failure event occurrence prediction unit 205.

Claims (15)

A factor extraction unit that extracts a plurality of factors correlated with the occurrence of a failure event of the plant from the collected plant data by statistical analysis;
Based on the factor extracted by the factor extraction unit and the presence or absence of occurrence of a failure event in the plant, a weighting calculation unit that calculates a contribution rate of each factor to the occurrence of a failure event;
Based on each factor extracted by the factor extraction unit and each contribution rate calculated by the weight calculation unit, an equation derivation unit for deriving a prediction equation for determining whether or not a malfunction event has occurred,
Based on the prediction formula derived by the equation deriving unit, a failure event occurrence prediction unit that predicts the occurrence of a failure event, and
A failure event occurrence prediction device comprising: The malfunction event occurrence prediction device according to claim 1, wherein the malfunction event of the plant is a malfunction event caused by hematite scale accumulating on an inner surface of a furnace wall tube of a boiler. The malfunction event occurrence according to claim 2, wherein the formula deriving unit derives the prediction formula based on the economizer entrance iron concentration and the contribution rate of the economizer entrance iron concentration to the malfunction event occurrence. Prediction device. The malfunction event occurrence prediction device according to claim 2, wherein the formula deriving unit derives the prediction formula based on an operation time and a contribution rate of the operation time to the occurrence of the malfunction event. The formula deriving unit derives the prediction formula based on an average mass flow rate and a contribution rate of the average mass flow rate to the occurrence of the malfunction event. Failure event occurrence prediction device. The malfunction event according to any one of claims 2 to 5, wherein the formula deriving unit derives the prediction formula based on an oxygen concentration and a contribution rate of the oxygen concentration to the malfunction event occurrence. Occurrence prediction device. The formula deriving unit derives the prediction formula based on the main steam iron concentration and the contribution rate of the main steam iron concentration to the occurrence of the malfunction event. The described failure event occurrence prediction device. The failure event occurrence prediction according to any one of claims 2 to 7, wherein the equation deriving unit derives the prediction equation based on a pH and a contribution rate of the pH to the failure event occurrence. apparatus. The malfunction event according to any one of claims 2 to 8, wherein the formula deriving unit derives the prediction formula based on the number of activations and a contribution rate of the activation frequency to the occurrence of the malfunction event. Occurrence prediction device. The failure event occurrence predicting unit predicts a failure event occurrence time based on the prediction formula derived by the equation deriving unit, and prompts a countermeasure in advance for the occurrence of the failure event. The malfunction event occurrence prediction device according to item. The failure event occurrence prediction unit predicts a failure event occurrence time based on the prediction formula derived by the equation deriving unit, and changes a factor value so as to delay the occurrence of the failure event. The malfunction event occurrence prediction device according to any one of the above. The formula deriving unit derives the prediction formula for each part of the plant,
The failure event occurrence prediction unit predicts a failure event occurrence part based on the prediction formula derived by the equation deriving unit, and prompts a countermeasure in advance of the occurrence of the failure event. The malfunction event occurrence prediction device according to one item. The failure event occurrence prediction device according to any one of claims 1 to 12, further comprising a notification unit that notifies a user of occurrence information of a failure event predicted by the failure event occurrence prediction unit. Extracting, from the collected plant data, a plurality of factors correlated with the occurrence of a plant malfunction event by statistical analysis;
Based on the factor extracted by the step of extracting the plurality of factors and the presence or absence of occurrence of a failure event in the plant, calculating a contribution rate of each factor to the occurrence of the failure event;
Deriving a prediction formula for determining the presence or absence of a malfunction event based on the step of extracting the plurality of factors and the step of calculating the contribution rate;
Predicting the occurrence of a malfunction event based on the prediction formula derived by the step of deriving the prediction formula;
A failure event occurrence prediction method for a plant comprising: Computer
Factor extraction means for extracting a plurality of factors correlated with the occurrence of a plant malfunction event from the collected plant data by statistical analysis;
Based on the factor extracted by the factor extracting means and the presence or absence of occurrence of a failure event in the plant, weight calculation means for calculating a contribution rate of each factor to the occurrence of the failure event;
Formula deriving means for deriving a prediction formula for determining whether or not a malfunction event has occurred, based on each factor extracted by the factor extracting means and each contribution rate calculated by the weight calculating means,
A failure event occurrence predicting means for predicting the occurrence of a failure event based on the prediction formula derived by the equation deriving means;
Program to function as.

Images