JP5065653B2 - Coal gasifier operation control method, coal gasifier operation control device, and coal gasifier operation control program - Google Patents

Description

  The present invention relates to a coal gasifier operation control method, a coal gasifier operation control device, and a coal gasifier operation control program. More specifically, the present invention relates to a coal gasification furnace operation control method, a coal gasification furnace operation control device, and a coal gasification furnace operation control program suitable for improving the efficiency and safety of the operation of the coal gasification furnace. .

  As one of the next-generation power generation systems, research is underway for the practical application and commercialization of coal gasification combined power generation (IGCC), which generates high-efficiency power by combining steam turbines and gas turbines.

  In Non-Patent Document 1, an optimization calculation is performed based on a performance evaluation function of a gasifier with respect to a combustor temperature, a generated gas heat generation amount, and a recycled char amount, in which the gasifier burden, the air ratio, and the coal supply ratio are operating parameters. An optimal operation support system for coal gasifiers has been proposed that has the function of performing and setting operation control parameters.

Hiroaki Watanabe, Kenichi Tokoro et al. "Development of optimum operation support system for coal gasifier-Examination of optimization method for gasifier operation-" Central Research Institute of Electric Power Research report: M04001,2004

  However, the performance evaluation function of the gasifier in the technique described in Non-Patent Document 1 is constructed based on data such as the combustor temperature measured once after operation under one operating condition of the gasifier. Yes, it was not possible to build and modify the performance evaluation function during operation. In addition, since the data measured after the operation includes data acquired under the conditions that do not satisfy the operation restriction conditions of the gasifier, the operation control parameters obtained by the optimization calculation based on the data are included. The accuracy was low and the gasifier could not be operated efficiently.

  Accordingly, the present invention provides a coal gasification furnace operation control method, a coal gasification furnace operation control device, and a coal gasification furnace operation control program for determining a highly accurate operation control parameter to improve the efficiency of coal gasification furnace operation. The purpose is to provide.

  To achieve this object, the coal gasification furnace operation control method according to claim 1 obtains at least the combustor temperature, the generated gas heat generation amount, and the recycled char amount as online data from the operating coal gasification furnace. A time interval in which a correlation between the online data and the operation control parameter is found in advance for the processing and the coal gasifier, a statistical process for averaging the online data at the time interval, and the averaged online data Assuming that the error between the performance evaluation function estimation process that estimates the performance evaluation function based on the normal distribution and the performance evaluation function and the online data follows a normal distribution, the probability of violating the driving constraint conditions during the preset driving is a risk factor And a risk factor setting process that determines online data settings based on the normal distribution and the risk factor Generating a maximization of the gas heating value for the purpose, and is to perform the operation control parameter determination processing for setting the operation control parameter solve the optimization problem of constraint and operating constraints based on the set value.

  Further, the coal gasifier operation control device according to claim 4 acquires at least the combustor temperature, the generated gas heat generation amount, and the recycled char amount as online data from each measuring device installed in the operating coal gasifier. Online data acquisition means to be stored in the storage device, and the time interval at which the online data has a correlation with the operation control parameter for each coal gasifier is stored in the storage device, and the online data is stored in the time interval. Statistical means for averaging and storing in the storage device, performance evaluation function estimating means for reading out the averaged online data, obtaining a performance evaluation function by regression analysis and storing it in the storage device, and a value of the performance evaluation function Obtain a histogram of errors from the online data, assume that the histogram follows a normal distribution, and set a preset The risk factor setting means to determine the online data setting value based on the normal distribution and the risk factor and store it in the storage device with the probability of violating the operation constraint conditions during the rotation, and to maximize the generated gas calorific value And an operation control parameter determining means for solving an optimization problem with an objective and a constraint equation based on a set value as an operation constraint condition to obtain an operation control parameter.

  The coal gasifier operation control program according to claim 6 obtains at least the combustor temperature, the generated gas heat generation amount, and the recycled char amount as online data from each measuring device installed in the coal gasifier in operation. Online data acquisition processing to be stored in the storage device, and the time interval at which the online data has a correlation with the operation control parameter in advance for each coal gasifier is stored in the storage device, and the online data is stored in the time interval. Statistical processing to be averaged and stored in the storage device, performance evaluation function estimation processing to read the averaged online data, obtain a performance evaluation function by regression analysis and store it in the storage device, and a value of the performance evaluation function Obtain a histogram of errors from online data, assume that the histogram follows a normal distribution, and predict The probability of violating the operation constraint conditions during the set operation is regarded as a risk factor, a risk factor setting process for determining a set value of online data based on the normal distribution and the risk factor and storing it in a storage device, and a generated gas calorific value An operation control parameter determination process for solving an optimization problem for the purpose of maximization and using a constraint equation based on a set value as an operation constraint condition and outputting an operation control parameter is executed by a computer.

  Therefore, data such as combustor temperature measured by each measuring device in the operating gasification furnace is acquired as online data, and the online data obtained for each gasification furnace in advance is the air ratio, coal feed ratio, etc. Online data is averaged at a time interval (hereinafter also referred to as a minimum time interval) that has a correlation with the operation control parameter. Further, by estimating the performance evaluation function from the online data averaged at the minimum time interval, it is possible to estimate a performance evaluation function having high correlation. Furthermore, based on the knowledge that the error distribution of this highly correlated performance evaluation function and the measured value online data approximates the normal distribution, the risk factor is set in advance as a point on the normal distribution. The value can be obtained mathematically. Furthermore, using the constraint formula with the set value as a limit value for the obtained online data as the operating constraint condition, non-linear optimization calculation for the purpose of improving the efficiency of gasifier operation is performed, and acquired during operation On the basis of the online data, a highly accurate operation control parameter is determined while continuing the operation.

  The invention according to claim 2 is the gasifier operation control method according to claim 1, wherein the performance evaluation function is the first performance evaluation function, and the normal distribution based on the first performance evaluation function is the first normal distribution. When the first normal distribution and the second normal distribution based on the second performance evaluation function estimated in the next performance evaluation function estimation process are determined to be significantly different, the first performance evaluation function is The reconstruction process of the performance evaluation function to be changed to the second performance evaluation function is performed.

  The invention described in claim 5 is the gasification furnace operation control device according to claim 4, further comprising a first performance evaluation function as the performance evaluation function and a first normal distribution based on the first performance evaluation function. When it is determined that the first normal distribution and the second normal distribution based on the second performance evaluation function estimated in the next performance evaluation function estimation process are significantly different from each other, A performance evaluation function restructuring means for changing the performance evaluation function to the second performance evaluation function is provided.

  The invention according to claim 7 is the coal gasifier operation control program according to claim 6, further comprising a first performance evaluation function as a performance evaluation function and a normal distribution based on the first performance evaluation function as a first distribution. If the first normal distribution and the second normal distribution based on the second performance evaluation function estimated in the next performance evaluation function estimation process are determined to be significantly different from each other, The performance evaluation function restructuring process for changing the performance evaluation function to the second performance evaluation function is executed by the computer.

  Accordingly, if the two normal distributions are significantly different from each other when the two normal distributions are compared, that is, if the first performance evaluation function is no longer compatible with the online data that is continuously acquired, the first performance evaluation is performed. The operation control parameter is changed by replacing the function with a second performance evaluation function and performing the subsequent processing.

  According to a third aspect of the present invention, in the gasification furnace operation control method according to the first or second aspect, the molten slag flow monitoring process for monitoring the flow state of the molten slag flow is performed, and a sign of slag clogging is obtained. If it is detected, the driving constraint condition is changed.

  Accordingly, when a sign of slag clogging is detected, it is determined that there is a problem with the operation constraint condition, and the operation constraint condition is changed. For example, because the combustor temperature is too low, the operation constraint conditions for the combustor temperature are changed.Specifically, the lower limit set value of the combustor temperature is set higher, the optimization calculation is executed, and the operation control parameter is set. I am trying to change it.

  As described above, according to the coal gasification furnace operation control method, the coal gasification furnace operation control apparatus, and the coal gasification furnace operation control program according to the present invention, online data is obtained from each measuring device during the operation of the gasification furnace. It is possible to estimate the performance evaluation function while acquiring. In addition, by estimating the performance evaluation function from the online data averaged at the minimum time interval and obtaining the optimization calculation constraint based on the performance evaluation function, high-precision operation control during the operation of the gasifier The parameter is determined, and the gasifier can be operated efficiently based on the parameter.

  Further, assuming that the error distribution is a normal distribution, the risk factor can be reflected in the driving constraint condition. Therefore, even if the driving restriction condition is temporarily not satisfied during driving, the driving control parameter can be set to continue driving if the overall risk rate is not exceeded, thereby eliminating the driving range restriction. It is possible to optimize safe operating conditions in consideration of variations in measured values.

  According to the coal gasifier operation control method according to claim 2, the coal gasifier operation control device according to claim 5, and the coal gasifier operation control program according to claim 7, two normal distributions are obtained. When the error becomes large, it becomes possible to autonomously change the operation control parameter. In addition, as the operation continues, it becomes possible to estimate a highly-correlated performance evaluation function, and the longer the operation time, the more efficiently the gasifier can be operated based on highly accurate operation control parameters. Become.

  According to the coal gasifier operation control method according to claim 3, operation control parameters capable of autonomously preventing slag clogging, misfire, etc. can be set, and safe operation of the gasifier is performed. Is possible.

  Hereinafter, the configuration of the present invention will be described in detail based on the embodiment shown in FIGS.

  The coal gasification furnace operation control apparatus 1 of the present embodiment acquires at least the combustor temperature, the generated gas heat generation amount, and the recycled char amount as online data from each measuring device installed in the operating coal gasification furnace, and stores it in the storage device. Online data acquisition means for storing, and for each coal gasification furnace, a minimum time interval at which the online data has a correlation with the operation control parameter is obtained in advance, and the online data is averaged at the minimum time interval and the storage device A statistical means for storing data, a performance evaluation function estimating means for reading out the averaged online data, obtaining a performance evaluation function by regression analysis and storing it in the storage device, and an error between the value of the performance evaluation function and the online data A histogram is obtained, and it is assumed that the histogram follows a normal distribution. As a risk factor setting means for determining a set value of online data and storing it in the storage device based on a risk factor that is a probability of violating the operation constraint condition during operation stored in the storage device, and Operation control parameter determination means for solving the optimization problem for the purpose of maximization and solving the optimization problem using the constraint equation based on the set value as the operation constraint condition is provided.

  As shown in FIG. 1, the gasification furnace 2 of the present embodiment is a pressurized two-stage entrained bed coal gasification furnace composed of two chambers: a lower combustor 3 and an upper reductor 4.

  Moreover, the imaging means 7 for imaging simultaneously the molten slag 6 which flows down from the discharge port 5 of the gasification furnace 2 from a different direction is provided. In the present embodiment, the photographing means 7 is, for example, three CCD cameras 7a, 7b, 7c.

  The combustor 3 is provided with a combustor pulverized coal burner 8 for supplying coal and a combustor burner 9 for supplying a recycled char collected at the furnace outlet, and the reductor 4 is provided with a reductor pulverized coal burner for supplying coal. 10 is installed.

  Further, in order to obtain images simultaneously from these three CCD cameras 7, a synchronous control device 11 for synchronizing the three CCD cameras 7, a three-dimensional shape measuring means 15 for obtaining a three-dimensional shape, and a three-dimensional surface temperature distribution. An image processing device 12 functioning as a temperature measuring means 16 to be obtained, an online data acquiring means 17 for acquiring online data from each measuring device installed in the coal gasification furnace, and statistics for averaging the online data at a minimum time interval Based on the distribution of the error between the performance evaluation function and the online data (normal distribution) and a preset risk factor, based on the average means 18, the performance evaluation function estimation means 19 for estimating the performance evaluation function from the averaged online data, A risk factor setting means 20 for obtaining a set value of online data, an operation control parameter for obtaining an operation control parameter by performing an optimization calculation. Meter determining means 21, performance evaluation function restructuring means 22 for reconstructing the performance evaluation function when the normal distribution based on the performance evaluation function is significantly different, and detecting the slag clogging sign and changing the operation control parameter And an optimization arithmetic unit 13 that functions as the molten slag flow monitoring means 23.

  Further, although not shown, a measuring device such as a thermocouple for measuring the temperature in the combustor 3 (hereinafter, combustor temperature) is provided at the outlet of the combustor 3, and coal generated at the outlet of the gasification furnace in the reductor 4. Measurement equipment such as gas chromatography (GC) and mass spectrometer for measuring the calorific value of gasification gas (hereinafter referred to as product gas) (hereinafter referred to as product gas calorific value), and char supply facilities include recycled char (burning of pulverized coal) (Remaining) A hopper weight change rate and an impact flow meter for recycling char are measured.

  The online data such as the combustor temperature, the generated gas heat generation amount, the recycled char amount, and the like, and the gasification performance data of the gas that is the product of the coal gasification furnace 2 are input to the optimization computing device 13 via the measuring device 14. (Refer to the arrow indicated by symbol a in FIG. 1). Further, the optimization calculation device 13 performs control based on operation control parameters obtained for each operation target device of the gasification furnace 2 such as the reductor pulverized coal burner 8, the combustor pulverized coal burner 10, the combustor char burner 9, and the like. A signal is sent, and these driving operation target devices are controlled (see the arrow indicated by symbol b in FIG. 1).

  The synchronization control device 11, the image processing device 12, and the optimization calculation device 13 can be realized by using existing computer resources (computer system). FIG. 2 shows an example of the configuration of the optimization arithmetic unit 13 as an example. The optimization arithmetic device 13 stores an output device 24 such as a display, an input device 25 such as a keyboard and a mouse, a central processing arithmetic device (CPU) 26 that performs arithmetic processing, and data, parameters, and the like being calculated. A main storage device (memory, RAM) 27, a hard disk as an auxiliary storage device 28 in which various data such as calculation results are recorded, an input / output interface 29 for inputting / outputting data to / from the outside, and the like. Hereinafter, the main storage device 26 and the auxiliary storage device 27 are collectively referred to simply as a storage device. The above hardware resources are electrically connected through a bus 30, for example.

  Moreover, the coal gasifier operation control program of the present embodiment is recorded in, for example, the auxiliary storage device 28, and the computer functions as the optimization computing device 13 by the CPU 26 reading and executing the program. The online data acquisition means 17, statistical means 18, performance evaluation function estimation means 19, risk factor setting means 20, operation control parameter determination means 21, performance evaluation function reconstruction means 22, and molten slag flow monitoring means 23 are The software executed by the CPU 26 can be configured to be executed by a computer. Data necessary for the execution is loaded into the RAM 27.

  Further, the auxiliary storage device 28 is not necessarily an internal device of the computer, and it goes without saying that an external hard disk or an external storage device accessible via a network may be used. In the present embodiment, the operation result database 31 is configured in the auxiliary storage device 28. The configuration of the optimization operation 13 described above is an example and is not limited to this. Also, the hardware configurations of the synchronization control device 11 and the image processing device 12 are configured in the same manner except for the means for executing processing. In addition, the synchronization control device 11, the image processing device 12, and the optimization calculation device 13 do not necessarily have to be configured by separate hardware, and a plurality of devices can be realized by a single hardware.

  Moreover, the outline | summary of the coal gasifier 2 whole used by FIG. 3 at this embodiment is shown. According to the coal gasification furnace 2, raw coal is finely pulverized and dried by a pulverizer 32 to a 200 mesh (74 [mm]) pass 90 [%] or more, and a storage bin 33, a pressure lock hopper 34, a weighing hopper Through 35, the set flow rate is supplied to the combustor 3 and the reductor 4 by air conveyance.

  Further, the secondary air (see the arrow indicated by the symbol d in FIG. 3) is heated to a predetermined temperature and supplied only to the combustor 3. The combustor 3 has a function of burning pulverized coal and char at a high temperature and discharging coal ash as molten slag 6 and a function of supplying heat necessary for a gasification reaction in the reductor 4.

  In the reductor 4, the hot gas rising from the combustor 3 and the pulverized coal blown from the reductor pulverized coal burner 8 are mixed to promote thermal decomposition of the coal. In the reductor, a gasification reaction of char which is an endothermic reaction occurs. The product gas is cooled by the heat exchanger 36, and the char in the product gas is collected by the cyclone 37. The collected char is air-flowed and supplied to the combustor 3 in the same manner as pulverized coal. Reference numeral 38 is a raw coal bunker, 39 is an air compressor, 40 is an air heater, 41 is a pulverized coal distributor, 42 is an ash hopper, 43 is an ash lock hopper, 44 is cooling water, and 45 is a starting heavy oil combustor. , 46 is a level meter, 47 is a water tank, 48 is a char distributor, 49 is a water washing tower, 50 is a pressure reducing valve, 51 is a cooling water pump, 52 is a cooling water pound, 53 is a cooling tower, and 54 is a product gas incinerator. , 55 is a bag filter, 56 is a desulfurizer, and 57 is a chimney.

In this embodiment, various data are measured as described below. Table 1 shows each data, its measurement method, sampling cycle, and sampling position. Hereinafter, unless otherwise specified, measurement is performed based on Table 1. In FIG. 3, the numbers surrounded by circles indicate the sampling positions of the respective data (see the sampling position column in Table 1).

  Next, the coal gasifier operation control method of this embodiment will be described. The coal gasification furnace operation control method of the present embodiment includes an online data acquisition process for acquiring at least the combustor temperature, the generated gas heat generation amount and the recycled char amount as online data from the operating coal gasification furnace, and the coal gasification furnace. A minimum time interval in which a correlation between online data and operation control parameters can be found in advance, statistical processing for averaging the online data at the minimum time interval, and a performance evaluation function based on the averaged online data. Assuming that the performance evaluation function estimation process to be estimated and the error between the performance evaluation function and online data follow a normal distribution, there is a risk that the probability of violating the driving constraint conditions during the operation set as the initial parameter and the normal distribution in advance Risk rate setting process to determine the set value of online data based on the rate, and generated gas For the purpose of maximizing the heat, and is performed a driving control parameter determination processing for setting the operation control parameter solve the optimization problem of constraint and operating constraints based on the set value.

The performance evaluation function of the coal gasifier used for the optimization calculation is expressed as a function of the operation control parameter with respect to the performance evaluation index of the coal gasifier. The performance evaluation index can be measured from the gasifier 2 as online data as described above. In the present embodiment, three values of “combustor temperature”, “generated gas calorific value (coal gasification gas calorific value)”, and “recycled char amount” are used as performance evaluation indexes. The combustor temperature is evaluated in order to completely melt the ash contained in the coal and discharge it as molten slag. The generated gas calorific value is evaluated as the quality of fuel gas for the gas turbine combustor. This is an index for evaluating the reactivity of coal in the gasifier. Here, the performance evaluation index is not limited to these, and in addition to the above performance evaluation index, for example, “in-furnace carbon conversion rate” and “cold gas efficiency” may be used. “In-furnace carbon conversion rate” and “cold gas efficiency” are values proportional to the total coal flow rate to be supplied to the gasification furnace 2, and “in-furnace carbon conversion rate” is inversely proportional to “recycled char amount”. It becomes a relationship. “Cold gas efficiency” indicates how much chemical heat of coal has been transferred to product gas. “In-furnace carbon conversion rate” and “cold gas efficiency” are expressed by Equation 1 and Equation 2, respectively.

  In the present embodiment, an operation control parameter function is constructed for the above three performance evaluation indexes. As the operation control parameters of the coal gasifier, combustor coal flow rate, reductor coal flow rate, coal feed ratio, combustor pulverized coal burner secondary air flow rate, combustor char burner secondary air flow rate, etc. can be used. In the embodiment, “air ratio” and “coal supply ratio” are used as operation control parameters. Each operation control parameter will be described below.

The “air ratio” is the ratio of the air flow rate input to the gasification furnace 2 to the theoretical air flow rate relative to the total coal flow rate input to the gasification furnace, and is expressed by Equation 3. The air flow rate can be measured from a gas flow meter installed in each burner, and the coal flow rate can be measured from the weight change rate of the coal supply hopper and the fluctuation value of the impact flow meter.

The “coal supply amount ratio” refers to the ratio of the amount of coal supplied to the combustor 3 and the reductor 4, and is expressed by Equation 4. In the case where a plurality of burners are installed in the gasifier 2, the ratio of coal input to each burner may be used as the operation control parameter. This controls the combustor temperature.

  In this embodiment, “gasifier load” is set as an initial parameter. “Gasifier load” refers to the ratio between the rated condition of the gasifier 2 and the total coal flow rate actually input. For example, when the rated coal flow rate is input, the gasifier load is 100%. If a coal flow rate of 80% of the rated condition is input, the gasifier load becomes 80%.

  Next, statistical processing in the present embodiment will be described with reference to FIGS. As an example for explaining the processing, the operating conditions are set to target air ratio = 0.525 and target coal supply ratio = 0.55, once per second of the gas flow rate and gas flow rate to the gasifier 2. 4 to 6 show measured values of the online data (hereinafter referred to as instantaneous data). In addition, it is preferable to acquire online data at the minimum sampling period that can be measured by each device in order to improve accuracy.

  As shown in FIG. 4, it can be seen that the coal flow rate is not only disturbed when it is received into the weighing hopper 35 but also fluctuates constantly. Further, as shown in FIG. 5, it can be seen that the gas flow rate periodically fluctuates at a cycle of about 100 seconds.

  In addition, as shown in FIG. 6, it can be seen that the temperature of each part in the gasification furnace 2 also fluctuates with time. This is because the air ratio changes with time because the flow rates of coal and gas fluctuate. In particular, the combustor temperature and the combustor outlet temperature tend to fluctuate greatly, and it can be seen that this tendency decreases with increasing distance from the top of the reductor. This indicates that the region closer to the burner is more susceptible to changes in the combustion state of pulverized coal due to fluctuations in the air ratio.

  FIG. 7 is a graph plotting the relationship between the instantaneous data of the combustor temperature and the corresponding instantaneous data of the air ratio. As shown in FIG. 7, since no correlation can be found between the combustor temperature and the air ratio, a performance evaluation function cannot be constructed.

  The reason why the combustor temperature, which should have a strong correlation with the air ratio, shows such a tendency is that the location where the coal flow rate is measured, the location where the gas flow rate is measured, and the location where the gas temperature is measured are within one gasifier 2. Even so, it is thought that the reason is that they capture different phenomena because they are different. However, it is impossible to measure each measured value at the same measurement point because of the design of the gasification furnace 2. That is, it can be said that there is a time constant in the gasifier 2, and it takes time for the influence to appear in the performance of the gasifier even if the operation control parameter is changed.

  The inventors of the present application have newly found that even if instantaneous data such as shown in FIG. 7 cannot be found at all, a high correlation with the operation control parameter can be found by performing statistical processing. It has been newly found that the performance evaluation function of the combustor temperature using the air ratio as a parameter can be expressed by a linear function. Here, statistical processing refers to finding the time average value at the integration time while integrating instantaneous data, and finding the minimum time interval that can find a high correlation coefficient in the time average value (hereinafter referred to as the minimum time interval). ) And average the instantaneous data at the minimum time interval.

  In the present embodiment, the minimum time interval is obtained based on the above knowledge. The algorithm for performing the statistical processing is not particularly limited. For example, the time average value is a correlation coefficient with respect to the air ratio in the instantaneous data, and then the correlation coefficient with respect to the air ratio in the average data for 2 seconds,. . . The time interval at which the correlation coefficient reaches the peak value can be set as the minimum time interval. In this case, in order to reduce the amount of calculation, for example, the correlation coefficient is first compared using average data in units of 100 seconds, and the correlation coefficient is obtained in units of 1 second in units of 100 seconds in which a peak value is considered to exist. Anyway.

  Here, the minimum time interval at which the correlation can be found differs depending on the gasification furnace 2. This is natural because the distance between the points differs even if the gasification is different in size and shape even if it is at a fixed place (combustor exit or the like). Therefore, in the coal gasifier operation control method of the present embodiment, a minimum time interval for performing statistical processing is determined in advance for each target gasifier 2, and statistical processing is performed for each minimum time interval. Is preferred. The minimum time interval may be obtained in advance for each gasification furnace, or may be obtained by the above-described method after the actual operation is started, and statistical processing may be performed using this.

  For example, in the case of the gasification furnace 2 in the present embodiment, it has been found that the minimum time interval needs to be at least about 600 seconds. Here, the minimum time interval at which data can be acquired depends on restrictions on the apparatus, but even in this case, the minimum time interval can be obtained. For example, in the case of the combustor temperature, it was possible to acquire data every second, but in the case of the generated gas calorific value, in the gasification furnace 2 of the present embodiment, once every 5 minutes (300 seconds). However, there was a limitation on the device that data could only be obtained. In this case, no correlation was found at 300 seconds, and a correlation could be found at 600 seconds. In addition, the correlation could be found even at 900 seconds. From the above, in the case of the generated gas heat generation amount, it is considered that there is a minimum time interval for which a correlation is required at least between 300 seconds and 600 seconds. Hereinafter, in the present embodiment, considering that the minimum time interval of the combustor temperature is required to be at least about 600 seconds, the minimum time interval at which a correlation can be found for both the combustor temperature and the generated gas heating value is 900 seconds. Will be described.

  Similarly, FIG. 8 shows the change over time in the coal supply ratio. If the minimum time interval is set to 900 seconds, the coal supply ratio varies with the flow rate fluctuation of the pulverized coal supplied to the combustor 3 and the reductor 4, respectively. FIG. 9 shows the change over time in the amount of generated gas heat (minimum time interval = 900 seconds), and FIG. 10 shows the change over time in the amount of recycled char (minimum time interval = 900 seconds). As with the coal feed ratio, the generated gas heat generation also fluctuated over time. On the other hand, it can be seen that the amount of recycled char changes stepwise. This is because the char supply is controlled by the rotational speed of the positive displacement feeder.

  In the present embodiment, the performance evaluation function described below can be obtained by regression analysis (linear regression) from an average value of online data (hereinafter also referred to as operation performance data) at the minimum time interval obtained by statistical processing. it can. That is, the performance evaluation function described below varies depending on various operating conditions such as experimental equipment and online data, and is not fixed. In the present embodiment, the least square method is used for regression analysis, but the present invention is not limited to this. For example, a maximum likelihood estimation method or the like can be used. The performance evaluation function to be estimated is not limited to a linear function, and may be estimated as a non-linear function.

First, the performance evaluation function of the combustor temperature will be described. As shown in FIG. 11, since the combustor temperature (T _COM ) increases in proportion to the air ratio (λ), the performance evaluation function of the combustor temperature can be defined as a linear expression using the air ratio as a parameter (formula 5).
<Equation 5>
T _COM = 549.8 · λ + 1350.85

  The coefficient of determination representing the rate of change in combustor temperature that can be explained by the performance evaluation function shown in Formula 5 was 0.4692, and the variance of the difference (error) between the estimation formula and the online data was 180.04. From the above, it can be seen that the combustor temperature has a large correlation with the air ratio.

Further, as shown in FIG. 12, the combustor temperature (T _COM ) is also correlated with the coal supply ratio (rt). Similarly, looking at FIG. 12 showing the distribution of combustor temperature using the air ratio and the coal supply ratio as parameters, it can be seen that the combustor temperature tends to increase as the coal supply ratio decreases even at the same air ratio. Therefore, the combustor temperature can be defined as a binary function of the air ratio and the coal supply ratio (Formula 6).
<Equation 6>
T _COM = 383.2 · λ−297.9 · rt + 1612.45

  The coefficient of determination in the performance evaluation function expressed by Equation 6 is 0.5788, and the error variance is 146.25, which is higher than the case where only the air ratio is used as a parameter. Therefore, in the present embodiment, the evaluation function of the combustor temperature is a function using the air ratio and the coal supply ratio as parameters.

  Furthermore, in order to see the tendency of the error of the obtained performance evaluation function, FIG. 13 shows a histogram summarizing the appearance frequency of the error between the instantaneous data and the performance evaluation function. Here, the distribution of the error can be estimated by a statistical test method. The test method is not particularly limited as long as a known or new test means is used, but in this embodiment, a re-force test (Lilliefors Test), a Shapiro-Wilk test, etc. Is going to be used.

  The inventors of the present application have found that the distribution of errors indicated by this histogram approximates a normal distribution and can be assumed to be a normal component. When this error is plotted on the normal probability graph, it becomes as shown in FIG. 14, and it can be seen that the error distribution is very close to the normal distribution. When the error distribution is assumed to be a normal distribution, the average is 0 and the standard deviation is 11.82K. In the present embodiment, it is possible to assume a normal distribution with a significance level of 95%, for example.

Next, the performance evaluation function of the generated gas heat generation amount will be described. As shown in FIG. 15, the generated gas heating value H _gas tends to decrease in proportion to the air ratio. Therefore, the performance evaluation function of the generated gas calorific value can be defined as a linear expression using the air ratio as a parameter (Formula 7).
<Equation 7>
H _gas = -3.707 · λ + 4.780

At this time, the coefficient of determination was 0.3751 and the variance of error was 2.88, and a high correlation was obtained. FIG. 16 shows a histogram with respect to the appearance frequency of the generated gas heat generation error, and a normal probability graph of the error shown in FIG. Similarly, it can be seen that the error distribution of the generated gas heat generation is extremely close to the normal distribution. When the error distribution is assumed to be a normal distribution, the average is 0, and the standard deviation is 0.11 MJ / m ³ N.

Next, a performance evaluation function for the amount of recycled char will be described. As shown in FIG. 18, the amount of recycled char Q _char tends to decrease in proportion to the air ratio. Therefore, the performance evaluation function of the amount of recycled char can be defined as a linear expression using the air ratio as a parameter (Formula 8).
<Formula 8>
Q _char = -117.3377 ・ λ + 80.5804

  At this time, the coefficient of determination was 0.3930, and the variance of error was 11.19, and a high correlation was obtained. FIG. 19 shows a histogram with respect to the appearance frequency of the error in the amount of recycled char, and FIG. 20 shows a normal probability graph of the error. Similarly, it can be seen that the distribution of errors in the amount of recycled char is very close to the normal distribution. If the error distribution is assumed to be a normal distribution, the average is 0 and the standard deviation is 3.31 kg / h.

  Thus, according to the coal gasifier operation control method of the present embodiment, it is possible to estimate the performance evaluation function for each performance evaluation index. However, as described above, an error occurs between the estimated value based on the performance evaluation function and the actual measured value of the online data. Therefore, it is impossible to estimate each performance evaluation index with 100% accuracy using the above performance evaluation functions.

  For this reason, when optimizing the operation conditions, it is necessary to set the operation constraint conditions in consideration of the error of the estimation formula so as not to exceed the limit value of the operation constraint conditions. However, under conditions that satisfy the operating constraint conditions throughout the entire operating time, the gasifier must be set at a point that is considerably safer than the limit values of the operating constraint conditions. The driving efficiency will deteriorate. Here, even in a very limited state, if it is a very short time, it does not immediately lead to slag clogging or misfire.

  Therefore, in the gasifier operation control method of the present embodiment, the probability of violating the constraint condition during operation is set as a risk factor, and the risk factor is reflected in the operation control parameter, thereby ensuring the safety of the gasifier. Driving efficiency can be improved (risk rate setting process).

  In the present embodiment, the risk factor can be defined as a percentage value of a normal distribution. As described above, the histogram of the difference between the estimated value based on the performance evaluation function and the actual measured value of the online data can be assumed to be a normal distribution. For example, when the risk factor is 10% as shown in FIG. When driving in a state where the constraint conditions are violated for a maximum of 6 minutes, the lower limit value is 10% (= 0.1) of the integral value (= 1.0) of the whole normal distribution. Can be approximated. The approximation method may be a known or new approximation method, and is not particularly limited. For example, Toda's approximation formula or the like can be used.

  Thus, if the risk factor is low, safe driving that satisfies the constraint conditions is possible, but the performance is relatively low. On the other hand, if the risk factor is high, gasification performance is improved and the gasification furnace can be operated efficiently. However, if the risk factor is set too high, the operation time in which the operation constraint condition is broken becomes long, and there is a possibility that slag clogging or misfire may occur. Although the setting of the risk factor depends on the concept of the plant operation by the driver, the operation restriction range can be set by setting the upper limit and the lower limit of the minimum risk factor. According to the knowledge of the inventors of the present application, the risk factor is preferably about 10% or less, for example. In this case, it is because the probability of violating the driving constraint condition continuously at two driving points at intervals of 900 seconds can be suppressed to 1% or less.

  In this embodiment, the optimization problem is formulated using the performance evaluation function and the risk factor described above, and the optimum operation control parameter is obtained by solving the optimization problem.

  In the present embodiment, for example, the following three constraint conditions are used as the operation constraint conditions of the coal gasifier.

(1) Combustor temperature lower limit In order to discharge the ash contained in the coal as molten slag 6 from the slag hole at the bottom of the combustor 3, it is necessary to maintain the temperature in the combustor 3 at or above the melting point of ash. Note that the combustor temperature lower limit is determined based on, for example, the composition of coal ash.
(2) Lower limit value of generated gas calorific value Because the generated gas must have a calorific value that exceeds the required specification as fuel for the gas turbine combustor, it is necessary to maintain stable combustion in the gas turbine combustor. is there. In addition, the generated gas calorific value lower limit is determined based on, for example, the specifications of the gas turbine combustor.
(3) Upper limit of amount of recycled char The char collected at the coal gasifier outlet is re-introduced into the combustor 3, so it is constant in the coal gasifier so that no char exceeding the recycling capacity is generated. This is because it is necessary to maintain the above reaction rate. The upper limit value of the recycled char amount is determined based on, for example, the recycled capacity.

  Note that the operation constraint condition is not limited to the above example, and for example, an in-furnace carbon conversion lower limit value, a cold gas efficiency lower limit value, and the like can be used as the operation constraint condition. For example, since the in-furnace carbon conversion rate is inversely proportional to the amount of recycled char as described above, the lower limit value of the in-furnace carbon conversion rate may be used as the operation constraint condition instead of the upper limit value of the recycled char amount. Further, by adding the cold gas efficiency to the operation constraint conditions, the optimization calculation can be performed under more severe constraint conditions, and the operation control parameters can be further improved in accuracy.

  In the present embodiment, the optimization problem is solved using a trust region method, but the solution of the optimization problem is not particularly limited, and for example, evolution such as a genetic algorithm. Other solution methods such as a dynamic calculation method and a robust optimization method may be used.

  When this optimization problem is solved and the optimal operation control parameters are determined, the calculation results are reported to the operator and fed back to the plant input device such as a burner as a control signal, and the operation continues based on the optimal operation control parameters. Is done. The obtained operation control parameter may be changed by the operator by inputting the parameter, or may be changed automatically.

  However, since it is necessary to follow a change in the operation status of the gasifier, in this embodiment, even after optimization, the instantaneous data of the online data is continuously acquired, and statistically based on the instantaneous data. Processing is also performed continuously. Therefore, the setting of the operation control parameter based on the initially estimated performance evaluation function may not satisfy the risk factor constraint. Therefore, in the coal gasification furnace operation control method of the present embodiment, the performance evaluation function is estimated every time statistical processing is executed, and for example, compared with the performance evaluation function in the previous processing (for example, 900 seconds before). The performance evaluation function is reconstructed.

  The performance evaluation function is reconstructed by performing a mathematical test between the previous performance evaluation function and the current performance evaluation function. In the present embodiment, it is assumed that the two populations (the values of the two performance evaluation functions) both follow a normal distribution, a test whether the averages are equal, and because they are heterogeneous, welch (Welch) T-test is used as a test method. In addition, the test method is not limited to this, and a known or new test method may be used. Further, the performance evaluation function reconstruction process does not necessarily have to be performed every minimum time interval, and each time the statistical process is executed a plurality of times, the performance evaluation function may be compared with the previous performance evaluation function.

  In this embodiment, when the average of two normal distributions is determined to be significantly different by Welch's t-test, the two normal distributions are determined to be different. When the two normal distributions are different, the risk factor and the optimization calculation are performed as described above based on the new performance evaluation function. Thereby, it becomes possible to continue driving | running on efficient driving | running conditions, changing a parameter autonomously. In addition, the more accurate the operation control parameters can be obtained as the operation is continued.

  Further, in the coal gasification furnace operation control method of the present embodiment, the molten slag flow that is sulfided from the furnace outlet in conjunction with the above-described processing is photographed from different directions by a plurality of CCD cameras, and the operating state is obtained from the slag flow image. It is preferable to perform the process of diagnosing in parallel. The process for monitoring the molten slag flow is not particularly limited, and for example, a “molten slag flow monitoring device” described in JP-A-2006-118744 can be used. According to this molten slag flow monitoring device, the three-dimensional shape and surface temperature distribution of the molten slag flow can be obtained.

  Here, the constraint condition of the combustor temperature lower limit value is set by estimating the melting point from the composition of the ash contained in the coal used as described above. However, even if, for example, coal of the same brand name is used, the actual ash content may vary slightly depending on the lot. In such a case, the slag flow may be deteriorated even when the operation control parameter of the gasifier is set to the optimum value by the optimization calculation. Since the deterioration of the slag flow leads to the blockage of the slag hole, it must be treated immediately. Here, the deterioration of the slag flow means that the combustor temperature is low, that is, the setting of the combustor temperature lower limit is too high.

  In this embodiment, when the deterioration of the slag flow is detected in this way, the combustor temperature lower limit value is corrected. Specifically, when a deterioration pattern of the flow state is detected in the slag flow image acquired from the installed CCD camera, the slag melting burner is automatically ignited in order to prevent solidification blockage of the slag hole. The correction of the combustor temperature lower limit and the optimization calculation are performed. The monitoring process of the molten slag flow is performed independently of the other processes, and is performed as an interrupt operation to the process executed in the flowchart shown in FIG.

  Next, the detail of the process which the coal gasifier operation control program of this embodiment performs is demonstrated using the flowchart of FIG.

  In this embodiment, as initial parameters before operation of the coal gasifier, “gasifier load”, “combustor temperature lower limit value”, “generated gas heating value lower limit value”, “recycled char amount upper limit value”, “risk rate” ”Is set, and the processing counter k is initialized (S1).

  Next, online data acquisition processing (S2) is performed. Specifically, on-line data such as combustor temperature, generated gas heat generation amount, recycled char amount, air ratio, and coal supply amount ratio data are stored in the auxiliary storage device 28 of the optimization arithmetic unit 13 or the like.

  In the present embodiment, the acquired online data is statistically processed once every 900 seconds based on the above knowledge. Accordingly, the online data acquisition process acquires instantaneous data once per second (the generated gas heat generation is data once every 5 minutes) and stores it in the storage device. When data for 900 seconds is stored, S3 To move to statistical processing.

  In the statistical process (S3), an average value (operation result data) of 900 seconds is calculated based on the online data, and the data is stored in the operation result database 31 in the auxiliary storage device 28.

  Next, it is determined whether the processing number counter k = 0 (S4). That is, when it is the first process (S4; Yes), the performance evaluation function is estimated (S5), and when it is the second and subsequent processes (S4; No), the performance evaluation function is reconstructed (S11). I do.

  In the estimation of the performance evaluation prediction function (S5), the operation performance data recorded in the operation performance database 31 is read, and the performance evaluation function using the operation control parameters as explanatory variables is estimated from the operation performance data by the least square method.

  Next, error calculation (S6) of the estimated performance evaluation function is performed. Specifically, an error with respect to past operation performance data is calculated for the performance evaluation function of the combustor temperature and the amount of generated char among the estimated performance evaluation functions. Here, the past operation result data refers to, for example, operation result data in the previous loop processing.

The error of the performance evaluation function is calculated by calculating the combustor temperature T _{t at} the past time t, the generated char amount Q _t , the air ratio λ _t , the coal feed ratio rt _t , and the estimated combustor temperature prediction formula T _com (λ, rt) and Q _char (λ) as the prediction formula for the amount of generated char, the combustor temperature error ECom _t at time _t is calculated, for example, by Equation 9, and the char amount error EChar _t is calculated by, for example, Equation 10. The
<Equation 9>
ECOM _t = T _com (λ _t , rt _t ) −T _t
<Equation 10>
EChar _t = Q _char (λ _t ) −Q _t

Next, the error distribution between the combustor temperature prediction function and the char generation amount prediction function is estimated based on the combustor temperature error ECom _t and the char amount error EChar _t (S7). In the present embodiment, the estimation is performed by the re-force test (Lilliefors Test).

  Next, a setting value that makes the risk factor less than or equal to the allowable probability is determined (S8). More specifically, the set temperature T is determined such that the risk factor for which the combustor temperature is set in advance is equal to or less than the allowable probability, assuming that the prediction function follows the estimated normal distribution. Similarly, the setting value V of the char amount prediction formula is obtained in which the probability that the amount of char generated exceeds the risk rate is equal to or less than the allowable probability. In the present embodiment, the set temperature T and the set value V can be obtained from Toda's approximate expression from the error distribution.

The optimization problem shown in Formula 11 is formulated using the obtained combustor temperature setting value T and char amount setting value V (S9).
<Equation 11>
Objective: Maximize generated gas heat generation (maxH _gas (λ, rt))
Constraint: Generated char amount ≦ V (Q _char (λ) ≦ V)
Combustor temperature ≧ T (T _com (λ, rt) ≧ T)

  An optimal operation control parameter is determined by solving the above optimization problem (S10). When the optimum operation control parameter is determined, the calculation result is reported to the operator and fed back to the plant input device such as a burner as a control signal. In the program, the process count counter is incremented by 1 (S12), and if the system termination is not instructed (S13; No), the process returns to the online data acquisition process of S2 to perform the loop process.

  In the case of processing other than the first time (S4; No), the performance evaluation function is reconstructed (S11). The reconstruction of the performance evaluation function is to compare the performance evaluation function in the previous loop process and the population of the performance evaluation function in the current loop process. In this embodiment, when the average of two normal distributions is determined to be significantly different by Welch's t-test, the two normal distributions are determined to be different.

  If it is determined that the two normal distributions are the same, it means that the operation may be continued with the operation control parameter based on the performance evaluation function up to the previous time with little error. That is, there is no need to reconstruct the performance evaluation function (S11; No), the process counter is incremented by 1 (S12), and if the system termination is not instructed (S13; No), the online data acquisition process of S2 is performed. Go back and loop.

  On the other hand, if it is determined that the two normal distributions are different, if the operation is continued with the operation control parameters based on the performance evaluation function up to the previous time, the risk factor constraint may not be satisfied. It is. That is, since it is necessary to reconstruct the performance evaluation function (S11; Yes), the process proceeds to S5 and the performance evaluation function is estimated.

  The loop process described above is continuously executed until the process end is instructed (S13; Yes). The above is description of the process which the coal gasifier operation control program of this embodiment performs.

  Next, an example of the monitoring process of the molten slag flow will be described with reference to the flowchart of FIG.

  First, slag images are acquired from the three CCD cameras 7a, 7b, and 7c that are synchronously controlled by the synchronous control device 11 installed immediately below the slag hole (S21), and molten slag flow state data (three-dimensional) is obtained by image processing. (Shape and surface temperature distribution) are obtained (S22).

  Next, the molten slag flow state data is subjected to a slag flow state by pattern matching using a learning device such as a support vector machine with an operation state database (not shown) preliminarily constructed in the storage device or the like of the image processing device 12. Is determined whether or not (S23).

  When the slag flow state is appropriate (S23; suitable), the molten slag flow state data is learned as a proper data by the learning algorithm and the operating state database is updated (S7), and the slag image is acquired (S21). Return to.

  On the other hand, when the slag flow condition is inappropriate (S23; unsuitable), the combustor temperature lower limit value is corrected among the operation constraint conditions as described above (S24), and the optimization calculation is performed according to the corrected operation constraint condition (formula 11) is performed (S25), and operation control parameters are obtained (S26). The calculation result is reported to the operator and fed back to the plant input device such as a burner as a control signal. Similarly, the molten slag flow state data is learned as inappropriate data by a learning algorithm, the operating state database is updated (S27), and the process returns to slag image acquisition (S21).

  The above molten slag monitoring process is performed in parallel with the process executed by the coal gasifier operation control program of the present embodiment shown in FIG. In the present embodiment, for example, the slag image acquisition process is executed once per second, and when it is determined that the slag flow situation is inappropriate, the process shown in FIG. 22 is interrupted. is there.

  The above-described embodiment is a preferred embodiment of the present invention, but is not limited thereto, and various modifications can be made without departing from the gist of the present invention. Moreover, the above arithmetic expression is an example, and various modifications can be made without departing from the scope of the present invention.

  For example, the coal gasification furnace is not limited to the shape and structure shown in FIGS. 1 and 3 and the like, and the present invention can be applied to furnaces having other shapes and structures. Moreover, although the example which applied this invention to the coal gasification furnace was demonstrated, the furnace which can apply this invention is not limited to a coal gasification furnace, Other gasification furnaces, sludge incineration ash are not used. A furnace that melts into slag may be used.

  The parameter setting method in the coal gasifier operation control program is not limited to the above example. For example, when the operator determines that the setting of the initial parameter (for example, gasifier load) is inappropriate, the subsequent processing may be performed after changing the initial parameter.

It is a block diagram which shows an example of a coal gasifier operation control apparatus. It is a figure which shows an example of the hardware constitutions of an optimization arithmetic unit. It is an example of the block diagram which shows the outline | summary of a coal gasification furnace. It is a graph which shows the relationship between coal flow volume, hopper weight, and elapsed time. It is a graph which shows the relationship between a gas flow rate and elapsed time. It is a graph which shows the relationship between the gas temperature data of each part of a gasifier, and elapsed time. It is the graph which plotted the instantaneous data of combustor temperature, and the air ratio corresponding to it. It is a graph which shows the time fluctuation of the 900 second average value of coal supply amount ratio. It is a graph which shows the time fluctuation of the 900-second average value of generated gas calorific value. It is a graph which shows the time fluctuation of the 900 second average value of the amount of recycled chars. It is a graph which shows the performance evaluation function of the combustor temperature which uses the online data (900 second average value) of combustor temperature and the air ratio corresponding to it as a parameter. It is a graph which shows the performance evaluation function of the combustor temperature which uses the online data (900 second average value) of combustor temperature and the air ratio and coal supply ratio corresponding to it as parameters. It is a histogram of the fluctuation component of a combustor temperature. 14 is a normal probability graph showing the histogram of FIG. It is a graph which shows the online data (900 second average value) of generated gas calorific value, and the performance evaluation function corresponding to it. It is a histogram of the fluctuation | variation component of generated gas calorific value. FIG. 17 is a normal probability graph representing the histogram of FIG. It is a graph which shows the online data (900 second average value) of the amount of recycled chars, and the performance evaluation function corresponding to it. It is a histogram of the fluctuation | variation component of the amount of recycled chars. 19 is a normal probability graph showing the histogram of FIG. It is the figure which described the normal distribution for demonstrating a risk factor in the graph which shows the performance evaluation function of generated gas calorific value. It is a flowchart which shows the outline | summary of the process which a coal gasifier operation control program performs. It is a flowchart which shows the detail of a molten slag flow monitoring process.

Explanation of symbols

1 Coal gasifier operation control device 2 Coal gasifier

Claims (7)

  Online data acquisition processing for acquiring at least the combustor temperature, the generated gas heat generation amount and the recycled char amount as online data from the operating coal gasification furnace, and the correlation between the online data and operation control parameters in advance for the coal gasification furnace A time interval in which the online data can be found, a statistical process for averaging the online data at the time interval, a performance evaluation function estimation process for estimating a performance evaluation function based on the averaged online data, and the performance It is assumed that an error between the evaluation function and the online data follows a normal distribution, and a probability of violating a driving constraint condition during preset driving is defined as a risk factor, and the online data is based on the normal distribution and the risk factor. The risk factor setting process for determining the set value of the And an operation control parameter determination process for setting an operation control parameter by solving an optimization problem using a constraint equation based on the set value as the operation constraint condition. .   The performance evaluation function is a first performance evaluation function, the normal distribution based on the first performance evaluation function is a first normal distribution, and the first normal distribution and the next performance evaluation function estimation process are estimated. When the second normal distribution based on the second performance evaluation function is determined to be significantly different from the second performance evaluation function, the performance evaluation function is reconstructed by changing the first performance evaluation function to the second performance evaluation function. The coal gasifier operation control method according to claim 1, wherein:   The molten slag flow monitoring process for monitoring the flow state of the molten slag flow is performed, and when the slag clogging is detected, the operation constraint condition is changed. Coal gasifier operation control method.   Online data acquisition means for acquiring at least the combustor temperature, the generated gas heat generation amount and the recycled char amount as online data from each measuring device installed in the operating coal gasification furnace, and storing it in a storage device, and for each of the coal gasification furnaces Preliminarily storing in the storage device a time interval at which the online data becomes correlated with the operation control parameter, and statistical means for averaging the online data at the time interval and storing it in the storage device; A performance evaluation function estimation unit that reads the averaged online data, obtains a performance evaluation function by regression analysis, and stores the performance evaluation function in a storage device; obtains a histogram of errors between the value of the performance evaluation function and the online data; Assumes that the histogram follows a normal distribution and violates driving constraints during preset driving A risk factor setting means for determining a set value of the online data based on the normal distribution and the risk factor and storing it in a storage device, and for the purpose of maximizing the generated gas calorific value, and An operation control parameter determination unit that solves an optimization problem using the constraint equation based on the set value as the operation constraint condition and obtains the operation control parameter is provided.   Further, the performance evaluation function is a first performance evaluation function, the normal distribution based on the first performance evaluation function is a first normal distribution, and the first normal distribution and the next performance evaluation function estimation process are estimated. When it is determined that the second normal distribution based on the second performance evaluation function is significantly different from the second performance evaluation function, the first performance evaluation function is changed to the second performance evaluation function. The coal gasifier operation control apparatus according to claim 4, further comprising a construction unit.   Online data acquisition processing for acquiring at least the combustor temperature, the generated gas heat generation amount and the recycled char amount as online data from each measuring device installed in the operating coal gasification furnace, and storing it in a storage device, and for each of the coal gasification furnaces A statistical process for storing the time interval in which the online data is correlated with the operation control parameter in advance in a storage device, averaging the online data at the time interval, and storing the average in the storage device; A performance evaluation function estimation process for reading the averaged online data, obtaining a performance evaluation function by regression analysis, and storing the performance evaluation function in a storage device; obtaining a histogram of errors between the value of the performance evaluation function and the online data; Assuming that the histogram follows a normal distribution, and driving constraints during preset driving For the purpose of maximizing the generated gas calorific value, the probability of violating the risk factor, determining the set value of the online data based on the normal distribution and the risk factor and storing it in a storage device; And an operation control parameter determination process for solving the optimization problem using the constraint equation based on the set value as the operation constraint condition and outputting the operation control parameter, and causing the computer to execute the operation. program.   Further, the performance evaluation function is a first performance evaluation function, the normal distribution based on the first performance evaluation function is a first normal distribution, and the first normal distribution and the next performance evaluation function estimation process are estimated. When the second normal distribution based on the second performance evaluation function is determined to be significantly different from the second normal performance distribution, the first performance evaluation function is changed to the second performance evaluation function. The coal gasifier operation control program according to claim 6, wherein the construction process is executed by a computer.