JP7645123B2 - BOILER OPERATION ASSISTANCE DEVICE, BOILER OPERATION CONTROL SYSTEM, BOILER OPERATION ASSISTANCE METHOD, AND BOILER OPERATION ASSISTANCE PROGRAM - Google Patents

Description

The present invention relates to a boiler operation support device, a boiler operation control system, a boiler operation support method, and a boiler operation support program.

In power plants, semi-automatic/fully automatic operation is being considered using operation control devices incorporating AI (artificial intelligence) technology such as machine learning to optimize operation and reduce labor. There is a need to visually understand the benefits of introducing such operation control devices in order to make decisions about their introduction (updates) and to confirm the effects after introduction.

As an example of a technology for evaluating the benefits of introducing a new system, for example, Patent Document 1 describes a system that calculates the difference between the amount of power consumed (actual measured value) in an energy-saving operating mode for a specified period and the amount of power consumed (actual measured value) in a fixed-capacity basic operating mode for another period, and calculates and notifies the user of a merit fee based on this difference.

Patent No. 3747798

In a power plant, conditions such as fuel properties, combustion state, and weather conditions change from moment to moment, and the optimal operating conditions (settings of operating instruction values) also change, so it is difficult to say that the benefits of introducing an operation control device are being appropriately evaluated. In addition, the validity of the base values to be compared when making an evaluation is also necessary. This point is not taken into consideration in Patent Document 1.

The present invention was made in consideration of the above situation, and aims to provide operational support by more appropriately evaluating operation when controlling boiler operation using a learning model.

In order to achieve the above object, the present invention has the configuration described in the claims. As an example, the present invention is a boiler operation support device, which includes an optimization unit that calculates an estimated evaluation index based on the results of an operation simulation using a learning model that calculates a predicted value of a process value that occurs when the boiler is operated, and derives an optimized control terminal parameter corresponding to the optimal value of the estimated evaluation index, a first estimated cost calculation unit that sets a period that is the subject of operation evaluation among the periods during which the boiler is operated using the optimized control terminal parameters as an evaluation period and calculates a first estimated cost that is estimated to occur during the evaluation period, a second estimated cost calculation unit that calculates a second estimated cost that is estimated to occur during the evaluation period when the boiler is operated using a base control terminal parameter for comparison with the optimized control terminal parameter, and an operation evaluation unit that outputs an operation evaluation result based on the difference between the first estimated cost and the second estimated cost.

According to the present invention, operation support can be provided by more appropriately evaluating operation when controlling boiler operation using a learning model. Problems, configurations, and effects other than those described above will become clear from the description of the embodiment below.

FIG. 1 is a schematic diagram of a boiler. FIG. 1 is a schematic diagram of a boiler operation control system. FIG. 2 is a diagram showing a hardware configuration of a driving assistance device. 11 is a flowchart showing an overview of a cumulative evaluation of AI driving control. FIG. 4 is a diagram showing an example of a learning model created by a model construction unit. FIG. 13 is a diagram showing process values used when evaluating AI operation control from the viewpoint of operation costs. FIG. 13 is a diagram showing the calculation process of the AI-on prediction data calculation unit. FIG. 13 is a diagram showing the calculation process of the AI-off prediction data calculation unit. FIG. 13 is a diagram showing process values used when evaluating AI operation control from the perspective of an environmental score. 11 is a flowchart showing the flow of a data processing process in the cumulative evaluation process of the AI driving control in step S4. FIG. 13 is a diagram showing an example of calculation of accumulated costs. FIG. 13 is a diagram showing an example of calculation of a prediction error.

Below, a preferred embodiment of the present invention will be described in detail with reference to the attached drawings. Note that the present invention is not limited to this embodiment, and when there are multiple embodiments, the present invention may be configured by combining the respective embodiments. The same components and steps are given the same reference numerals throughout the drawings, and duplicate explanations will be omitted.

Figure 1 is a schematic diagram of boiler 1.

The boiler 1 has a furnace 11, a combustion device 12, and a flue 13. The furnace 11 is, for example, a hollow rectangular cylinder and is installed along the vertical direction. The wall surface of the furnace 11 is composed of evaporation tubes (heat transfer tubes) and fins connecting the evaporation tubes, and the temperature rise of the furnace wall is suppressed by heat exchange between the feed water and steam flowing in the evaporation tubes and the combustion gas in the furnace 11. Specifically, on the side wall surface of the furnace 11, multiple evaporation tubes are arranged, for example, along the vertical direction and arranged side by side in the horizontal direction. The fins block the space between the evaporation tubes. The furnace 11 has an inclined surface 62 at the bottom of the furnace, and the bottom evaporation tube 70 is provided on the inclined surface 62 to form the bottom surface.

The combustion device 12 is provided on the vertically lower side of the furnace wall that constitutes the furnace 11. In this embodiment, the combustion device 12 has multiple combustion burners (e.g., 21, 22, 23, 24, 25) attached to the furnace wall. For example, the combustion burners (burners) 21, 22, 23, 24, 25 are arranged at equal intervals along the circumferential direction of the furnace 11. However, the shape of the furnace, the arrangement of the burners, the number of combustion burners in one stage, and the number of stages are not limited to this embodiment.

Each of these combustion burners 21, 22, 23, 24, 25 is connected to pulverizers (coal pulverizers/mills: equivalent to auxiliary equipment) 31, 32, 33, 34, 35 via pulverized coal supply pipes 26, 27, 28, 29, 30. When coal is transported by a conveying system (not shown) and fed into these pulverizers 31, 32, 33, 34, 35, it is pulverized to a predetermined powder size, and the pulverized coal (pulverized coal) can be supplied to the combustion burners 21, 22, 23, 24, 25 from the pulverized coal supply pipes 26, 27, 28, 29, 30 together with conveying air (primary air).

Furnace 11 also has a wind box 36 at the mounting position of each combustion burner 21, 22, 23, 24, 25, and one end of air duct 37b is connected to this wind box 36, and the other end is connected to air duct 37a that supplies air at connection point 37d.

In addition, a flue 13 is connected vertically above the furnace 11, and multiple heat exchangers (41, 42, 43, 44, 45, 46, 47) for generating steam are arranged in the flue 13. Therefore, a flame is formed by the combustion burners 21, 22, 23, 24, 25 injecting a mixture of pulverized fuel and combustion air into the furnace 11, and combustion gas is generated and flows into the flue 13. The combustion gas then heats the feedwater and steam flowing through the furnace wall and the heat exchangers (41-47) to generate superheated steam, which is supplied to rotate a steam turbine (not shown), which then rotates and drives a generator (not shown) connected to the rotating shaft of the steam turbine to generate electricity. In addition, this flue 13 is connected to an exhaust gas passage 48, and is provided with a denitration device 50 for purifying the combustion gas, an air heater 49 for exchanging heat between the air sent from the blower 38 to the air duct 37a and the exhaust gas sent through the exhaust gas passage 48, a dust treatment device 51, an induced draft blower 52, and the like, and a chimney 53 is provided at the downstream end. Note that the denitration device 50 does not need to be provided if the exhaust gas standards can be met.

The furnace 11 of this embodiment is a so-called two-stage combustion furnace, in which after excess fuel combustion using pulverized coal transport air (primary air) and combustion air (secondary air) fed into the furnace 11 from the wind box 36, new combustion air (after air) is fed to perform lean fuel combustion. For this reason, the furnace 11 is provided with an after-air port 39, one end of an air duct 37c is connected to the after-air port 39, and the other end is connected to an air duct 37a that supplies air at a connection point 37d. Note that if the two-stage combustion method is not adopted, the after-air port 39 does not need to be provided.

The air sent from the blower 38 to the air duct 37a is heated by heat exchange with the combustion gas in the air heater 49, and at the connection point 37d, it branches into secondary air that is led to the air box 36 via the air duct 37b, and after-air that is led to the after-air port 39 via the air duct 37c.

Figure 2 is a schematic diagram of the operation control system 10 for the boiler 1. The operation control system 10 is configured by communicatively connecting the boiler 1, an operation support device 100 for the boiler 1, and an operation control device 200 for the boiler 1.

The operation assistance device 100 constructs a learning model to predict various process values that arise when the boiler 1 is operated, and optimizes the operating conditions of the boiler 1 using AI operation control based on the predicted values of the various process values, thereby assisting in the operation control of the boiler 1.

Furthermore, the driving assistance device 100 evaluates the cost benefits of introducing AI driving control over cumulative time (hereinafter referred to as "cumulative evaluation").

The driving assistance device 100 includes a data acquisition unit 110, an RTC 112, a driving data storage unit 114, a data extraction unit 116, a data pre-processing unit 118, a model construction unit 120, a model storage unit 122, an optimization unit 124, a driving evaluation unit 132, an alternative control logic storage unit 134, and an input/output control unit 136.

The operation evaluation unit 132 includes an AI-on predicted data calculation unit 132a that calculates predicted data of the operation cost when the boiler 1 is operating under AI operation control (hereinafter referred to as "AI-on predicted data"); an AI-off predicted data calculation unit 132b that calculates predicted data of the operation cost when the boiler 1 is operating under operation control by an alternative control logic rather than AI operation control (hereinafter referred to as "AI-off predicted data"); a prediction error calculation unit 132c that calculates the prediction error included in the AI prediction data included in each of the AI-on predicted data and the AI-off predicted data; and a cumulative evaluation unit 132d that calculates and outputs a cumulative evaluation result that reflects the prediction error using the AI-on predicted data and the AI-off predicted data. In this specification, "prediction data" is data that includes one or more predicted values, and does not necessarily match the data of the "prediction value" alone. The AI-on predicted data calculation unit 132a corresponds to the first estimated cost calculation unit, and the AI-off predicted data calculation unit 132b corresponds to the second estimated cost calculation unit.

The input/output control unit 136 accepts input such as the cumulative evaluation period and cumulative evaluation conditions from the input device 311 and outputs them to the driving evaluation unit 132.

The input/output control unit 136 also outputs the cumulative evaluation result from the driving evaluation unit 132 to the display 312 for display. The functions of the other units will be described later.

Figure 3 is a diagram showing the hardware configuration of the driving assistance device 100.

The driving assistance device 100 includes a processor 301, a RAM (Random Access Memory) 302, a ROM (Read Only Memory) 303, a HDD (Hard Disk Drive) 304, an input I/F 305, an output I/F 306, and a communication I/F 307, which are connected to each other via a bus 308, and is configured using a computer.

The processor 301 may be a GPU (Graphics Processing Unit) or a CPU (Central Processing Unit), and may be of any type as long as it is a device that executes a calculation function. Furthermore, the hardware configuration of the driving assistance device 100 is not limited to the above, and may be configured by a combination of a control circuit and a storage device. The driving assistance device 100 is configured by the processor 301 executing a driving assistance program that realizes each function of the driving assistance device 100, or by the control circuit performing calculations.

An input device 311 such as a mouse, keyboard, or touch panel is connected to the input I/F 305.

A display 312 consisting of an LCD, an organic panel, or the like is connected to the output I/F 306. An external communication device may also be used as the output device, and the cumulative evaluation results may be transmitted to a remote location.

The boiler 1 and the operation control device 200 are each connected to the communication I/F 307.

The boiler 1 is equipped with sensors 1, 2, ..., M that detect various process values, and control terminals 1, 2, ..., N that set the operating conditions.

The driving assistance device 100 receives actual measurements of various process values from sensors 1, 2, ..., M.

The driving assistance device 100 transmits data indicating the driving conditions optimized by AI driving control to the driving control device 200. The driving control device 200 determines and sets the setting values to be set for each control element 1, 2, ..., N based on the driving conditions. The setting values for each control element are control element parameters. The driving control device 200 transmits the control element parameters to the driving assistance device 100.

Figure 4 is a flowchart showing an overview of the cumulative evaluation of AI driving control.

As a preliminary step for AI operation control, the model construction unit 120 inputs the boiler 1 operation data preprocessed by the data preprocessing unit 118, specifically various input parameters indicating the operation conditions, performs machine learning, and constructs a learning model that outputs a predicted value of the process value (S1). The constructed learning model is stored in the model storage unit 122.

Figure 5 is a diagram showing an example of a learning model created by the model construction unit 120. As input parameters, operation parameters (e.g., boiler load), and other parameters (e.g., environmental conditions such as temperature) are used.

A learning model is set up for each process value.

In the example of Figure 5, the process values are "boiler outlet steam temperature," "exhaust gas properties (NOx, etc.)," "gas temperature," "unburned content in ash," and "temperature inside boiler 1 (metal temperature)." The model construction unit 120 constructs learning model 1, learning model 2, learning model 3, learning model 4, and learning model 5 corresponding to each process value. Learning model 4 corresponds to a first learning model that calculates a predicted value of a specific non-measured process value, and learning models 1, 2, 3, and 5 correspond to second learning models that calculate predicted values of other process values.

Returning to FIG. 4, the optimization unit 124 uses the learning model to calculate a predicted value of the process value (the predicted value of the process value corresponds to the result of the operation simulation), and optimizes the operating conditions to improve the process value (S2). Operation control that optimizes the operating conditions using the learning model is called "AI operation control." In addition, the operating end parameters included in the optimized operating conditions are called optimized operating end parameters.

The boiler 1 is actually operated in one operation mode selected from multiple operation modes. The purpose of AI operation control is to improve the process value to suit that operation mode. Therefore, the optimization unit 124 uses a learning model to change the operation conditions (input parameters) and calculates predicted values of the process value under multiple operation conditions. Then, the optimization unit 124 selects the operation condition under which a predicted value of the process value that suits the operation mode is obtained. The optimization unit 124 obtains an estimated evaluation index according to the operation mode for the multiple operation conditions under which the operation simulation was performed, and selects the operation condition corresponding to the optimal value of the estimated evaluation index as the optimal condition. The optimization control unit 200 transmits the optimization control terminal parameters included in the optimization conditions to the operation control unit 200. The operation control unit 200 sets the optimization control terminal parameters to each of the operation terminals 1, 2, ..., N.

Operation data of the boiler 1 is collected in chronological order (S3). The operation data includes operating conditions and detected process values associated with time data (e.g., timestamps) obtained from the RTC 112, and is stored in the operation data storage unit 114.

Perform a cumulative evaluation of AI driving control (S4).

First, the calculation of the operating costs used in the cumulative evaluation will be explained with reference to Figures 6 to 8. Figure 6 shows the process values used when evaluating the AI operation control from the viewpoint of operating costs.

The evaluation method of this embodiment is called a "prediction method" because it uses not only actual measured values but also predicted values calculated from a learning model for the cumulative evaluation of AI driving control.

First, in FIG. 6, "AI operation state: ON" indicates a state in which the boiler 1 is operating under AI operation control. In "AI operation state: ON," the process values that can be measured while the boiler 1 is operating are actual measured values, and for process values that cannot be measured while the boiler 1 is operating (non-measured process values), a learning model that models the process values is used to calculate predicted values of the non-measured process values that occur during AI operation control.

The "cost-related process values" that are calculation factors for "operating cost" include "unburned content in ash" and "exhaust gas temperature, etc." Of these, "unburned content in ash" is a non-measured process value. Therefore, the AI-on prediction data calculation unit 132a calculates the predicted value of "unburned content in ash" using a learning model.

In addition, "exhaust gas temperature, etc." can be measured while boiler 1 is operating, so actual measured values are used.

Therefore, the "operating cost" is calculated including the predicted value of "unburned content in ash," so the "operating cost" is also obtained as a predicted value. Since the prediction error of the predicted value of "unburned content in ash" also affects the "operating cost," the prediction error of the predicted value is taken into consideration when cumulatively evaluating the operating cost. Details will be given later.

Figure 7 shows the calculation process of the AI-on prediction data calculation unit 132a.

The AI-on prediction data calculation unit 132a applies the input parameters, fuel parameters (coal properties, etc.), operation parameters (boiler load, etc.), other parameters (environmental conditions such as temperature, etc.), and optimization control terminal parameters set in the control terminal when the boiler 1 is operated under AI operation control, to the learning model 4 (see Figure 5) to calculate a prediction value of the unburned content in ash. This prediction value of the unburned content in ash is calculated not only in cost calculations but also in AI operation control and is used in optimization settings. The prediction values of other process values are used in optimization settings for AI operation control, but when calculating costs, actual measured values from the boiler 1 operating under AI operation control are used for cost calculations.

On the other hand, "AI operation state: OFF" indicates a state in which the boiler 1 is operating with an alternative control logic rather than AI operation control. The AI-off prediction data calculation unit 132b reads the alternative control logic from the alternative control logic storage unit 134 and uses the alternative control logic to determine the base operating end parameter. The base operating end parameter and other parameters are then applied as input parameters to all learning models to calculate a predicted value. Therefore, even for process values such as "exhaust gas temperature, etc." for which actual measured values are obtained during the operation of the boiler 1, in "AI operation state: OFF", predicted values are obtained for all process values as shown in FIG. 6. The AI-off prediction data calculated by the AI-off prediction data calculation unit 132b is compared (base value) with the prediction data under AI operation control.

Figure 8 shows the calculation process of the AI off prediction data calculation unit 132b.

When AI is off, there are no control end parameters set by AI operation control. Therefore, the AI-off prediction data calculation unit 132b reads the alternative control logic from the alternative control logic storage unit 134, and applies input parameters other than the control end parameters, specifically, the operation parameters and other parameters, to the alternative control logic to calculate the base control end parameters. Also included is the case where the base control end parameters when AI is off are obtained directly from the operation control device 200 instead of from the alternative control logic storage unit.

Next, the AI off prediction data calculation unit 132b inputs the base operating terminal parameters and other input parameters, specifically fuel parameters, operating parameters, and other parameters, into all learning models that model the process values required for cumulative evaluation, i.e., learning model 4 for unburned fuel in ash and learning models 1, 2, 3, and 5 corresponding to the other process values, and calculates prediction values for all required process values.

Figure 9 shows the process values used when evaluating AI operation control from the perspective of environmental scores.

The "environment-related process values" that are used to calculate the "environmental score" include "unburned fuel in ash" and "NOx, CO, etc."

The AI-on prediction data calculation unit 132a and the AI-off prediction data calculation unit 132b both use learning model 4 to calculate the prediction value of "unburned fuel in ash."

For "NOx, CO, etc.", the AI-on prediction data calculation unit 132a uses actual measured values and does not calculate a prediction value. On the other hand, after determining the base operating terminal parameters, the AI-off prediction data calculation unit 132b applies all input parameters to the process values required for calculating the "environmental score" for each corresponding learning model to determine a prediction value.

Therefore, since the prediction error of the "unburned fuel in ash" when AI is on, and the prediction error of the predicted values of all related learning models when AI is off, also affect the "environmental score," the prediction error of the predicted values is taken into consideration when cumulatively evaluating the "environmental score."

Figure 10 is a flowchart showing the flow of the data processing process in the cumulative evaluation process of AI driving control in step S4.

The AI-off prediction data calculation unit 132b reads preprocessed data from the data preprocessing unit 118. This preprocessed data is time-series data.

The AI-off prediction data calculation unit 132b extracts evaluation target data that satisfies the prediction conditions from the preprocessed data, and performs averaging processing (S401). The "prediction conditions" here include settling time and model learning conditions (e.g., operating load and number of operating mills).

The AI-off prediction data calculation unit 132b applies the same load and coal supply amount as when evaluating the AI operation control to the alternative control logic, and calculates the setting value to be set in the control element when the AI is off, i.e., in operation control using the alternative control logic (S402).

The "AI-off prediction data" that the AI-off prediction data calculation unit 132b uses to calculate the AI-off prediction data is prepared by adding the setting value of the control element calculated using the alternative control logic and input parameters that are not subject to optimization, such as atmospheric conditions.

The AI-off prediction data calculation unit 132b uses the "AI-off prediction data" as an explanatory variable to calculate the "AI-off prediction data" (shown as "modeling" in FIG. 10) (S403).

In addition, in S403, the AI-on prediction data calculation unit 132a calculates "AI-on prediction data" using input parameters that are not subject to optimization, such as operating parameters, fuel parameters, and atmospheric conditions, as explanatory variables.

The cumulative evaluation unit 132d performs cost calculation (S404). Specifically, it calculates the operating cost during AI operation control using the actual measurement data and the AI-on prediction data.

The cumulative evaluation unit 132d also calculates the operating cost (base value) assuming operation with alternative control logic using the AI off prediction data.

The cumulative evaluation unit 132d calculates the difference between the operating cost during AI operation control and the base value (corresponding to the operating evaluation value), accumulates it over the accumulation time, and outputs the cumulative evaluation result (S405). When performing the cumulative evaluation, the prediction error is taken into account.

Figure 11 shows an example of cumulative cost calculation.

The operating cost trend graph is shown in the upper part of Figure 11. The operating cost trend graph shows the time series transition between AI-off predicted data and AI-on predicted data for operating costs.

The middle section of Figure 11 shows a graph of time vs. load and number of operating mills. This graph shows the changes in operating load and number of operating mills during the same time period as the operating cost trend graph.

The cumulative merit is evaluated using an evaluation value (accumulated operating cost) obtained by accumulating the operating cost over time. In this case, it is desirable that the operating cost converted into the cumulative evaluation should be the sum of only the operating cost when the boiler 1 is operated while the evaluation conditions for the cumulative merit are satisfied. It is desirable that the evaluation conditions for the cumulative merit are the same as or within an acceptable range of the conditions satisfied by the operating data used to construct the learning model. The cumulative merit is merely one way of expressing the evaluation of the merit of introducing AI operation control, and other ways of expression, such as the time series change in the difference in operating cost between AI operation control and operation control using alternative control logic, may also be output.

In the upper graph of Figure 11, the learning conditions for the prediction model, i.e., the predictable operating conditions, are "90% or more, 5 operating mills in operation," and when the operational load range of the AI operation control is 95% or more, the shaded time period in Figure 11 does not meet the evaluation conditions for the cumulative merit. Therefore, the AI-off prediction data and AI-on prediction data in the shaded time period are excluded from the cumulative evaluation. In other words, only the non-shaded time periods, i.e., the time periods that meet the evaluation conditions, are used as the evaluation period for the cumulative evaluation.

Furthermore, if the operating conditions that define the evaluation conditions for the cumulative merit include fuel parameters that indicate the properties of the fuel used to operate boiler 1, the data extraction unit 116 extracts actual measurement data for a period in which fuel with the same fuel properties as those in the evaluation period was used. The same properties do not only mean completely the same properties, but also fuel with properties that can be considered to be the same. By using fuel with the same properties as those used in the evaluation period and extracting actual measurement data obtained under the same load as in the evaluation period, the evaluation accuracy for the evaluation period can be further improved.

The lower part of Figure 11 shows a graph illustrating the time series change in the difference in operating costs. This graph is one form of output of the operating evaluation results. The vertical axis of the graph shows the difference in operating costs (corresponding to the evaluation value) between AI on and off, and the horizontal axis shows time.

In addition, in the graph showing the time series change in the operating cost difference, the change in the operating cost difference is shown as a solid line, and the range of prediction error in operating cost caused by the prediction error contained in the forecast data is shown as a dashed line around it.

Figure 12 shows an example of how to calculate prediction error.

The prediction error calculation unit 132c performs cross-validation using data measuring the unburned fuel in the ash (data from which actual operating costs can be calculated). However, as shown in FIG. 12, data measuring the unburned fuel in the ash before the evaluation period is always included in the learning data in the cross-validation.

The prediction error calculation unit 132c calculates a prediction value of the unburned fuel in ash using a part of the data a, b, and c from before the evaluation period, for example a and b, as learning data, and obtains the difference between this and the actual value of the unburned fuel in ash contained in c as verification data as the prediction error μ1.

Similarly, the prediction error calculation unit 132c uses b and c as training data and a as validation data to obtain the prediction error μ2.

Furthermore, the prediction error calculation unit 132c uses c and a as training data and b as validation data to obtain the prediction error μ3.

The prediction error calculation unit 132c corrects the predicted operating cost for the evaluation period with the average error μ _ave . FIG. 12 shows an embodiment in which the trend graph of the operating cost is shifted downward by the average error μ _ave as an output embodiment taking into account the prediction error. In the case of AI-off prediction, it is necessary to take into account the prediction errors of all learning models related to the operating cost, so in the above description of FIG. 12, the unburned fuel in ash is replaced with the operating cost, and the prediction error is calculated in the same manner.

On the other hand, a form in which the prediction error range is displayed with a dashed line, such as the cumulative operation time-cumulative cost merit graph in the lower part of Figure 11, is also an output form that takes prediction error into account.

The contents described in each of the above embodiments can be understood, for example, as follows:

(1) A boiler operation support device according to one aspect includes:
an optimization unit that calculates an estimated evaluation index based on an operation simulation result using a learning model that calculates a predicted value of a process value that occurs when the boiler is operated, and derives an optimized operating end parameter corresponding to an optimal value of the estimated evaluation index;
a first cost estimate calculation unit that calculates a first cost estimate that is estimated to occur during an evaluation period, the evaluation period being a period during which the boiler is operated using the optimized operating end parameters;
a second cost estimate calculation unit that calculates a second cost estimate that is estimated to be incurred during the evaluation period when the boiler is operated using a base control parameter for comparison with the optimized control parameter;
a driving evaluation unit that outputs a driving evaluation result based on a difference between the first estimated cost and the second estimated cost;
The present invention is a boiler operation support device equipped with the above-mentioned.

According to the above aspect (1), the merits are calculated by comparing values before and after optimization based on the estimated values of the evaluation index (cost). As a result, since both values compared are estimated (calculated values), the calculations can be performed under the same conditions, and only the effect of optimization can be extracted, allowing for a clear evaluation of the merits. In other words, if actual measured values are used, it is difficult to grasp only the effect of optimization due to the influence of disturbances and measurement errors, but by using estimated values for both, the influence of disturbances and measurement errors can be eliminated.

Furthermore, when calculating the operating costs of AI operation control, actual values are used for process values that can be measured during operation, and predicted values are used only for process values that cannot be measured during operation, minimizing deviations in merit assessment due to prediction errors in the learning model.

(2) A boiler operation support device according to one aspect,
the second estimated cost calculation unit calculates the second estimated cost based on a predicted value obtained by applying the base operating end parameter derived from an alternative control logic that does not use the learning model to the learning model;
The boiler operation support device according to the above aspect (1).

According to the above aspect (2), the base operating end parameters can be derived from the alternative control logic, and can be evaluated based on the operating state before the introduction of AI operation control as a case in which AI operation control is not performed, and the benefits of introducing AI operation control can be evaluated more accurately.

(3) A boiler operation support device according to one aspect includes:
the learning model includes a first learning model that calculates a predicted value of a specific unmeasured process value and a second learning model that calculates a predicted value of another process value;
the first estimated cost calculation unit inputs the optimized control input parameter into the first learning model to calculate a predicted value of the specific unmeasured process value, and calculates the first estimated cost based on the predicted value;
the second estimated cost calculation unit inputs the base operating end parameters derived by applying the operating parameters applied during the evaluation period and the environmental parameters to the alternative control logic into the first learning model and the second learning model to calculate predicted values of the specific unmeasured process value and the other process values, and calculates the second estimated cost based on these predicted values.
The boiler operation support device according to the above aspect (2).

According to the above aspect (3), parameters other than the operation terminal parameters that change depending on whether or not AI operation control is introduced can be made common, and the benefits of introducing AI operation control can be evaluated more accurately.

(4) A boiler operation support device according to one aspect,
The specific non-measured process value is unburned carbon in ash;
The boiler operation support device according to the above aspect (3).

According to the above aspect (4), the benefits of introducing AI operation control can be evaluated using a predicted process value of the unburned fuel in ash, which cannot be measured while the boiler is operating.

(5) A boiler operation support device according to one aspect,
an operation data storage unit that stores actual measurement data including operating conditions of the boiler used in the actual operation of the boiler;
a data extraction unit that extracts actual measurement data for calculating the first estimated cost and the second estimated cost from the operation data storage unit,
The data extraction unit extracts the actual measurement data associated with the same operating conditions as the evaluation period,
the first cost estimate calculation unit and the second cost estimate calculation unit calculate the first cost estimate and the second cost estimate by using the extracted actual measurement data, respectively.
The boiler operation support device according to the above aspect (1).

According to the above aspect (5), the calculation conditions for the second estimated cost, which is the base value, are aligned with the evaluation period, so that the benefits of introducing AI operation control can be evaluated more accurately.

In addition, by only evaluating the AI driving control for time periods that meet the evaluation conditions and only evaluating the conditions that the learning model has learned (for that model), it is possible to avoid a decrease in accuracy due to predictions in the extrapolated range.

(6) A boiler operation support device according to one aspect,
The driving evaluation result is an economic merit or an environmental merit based on at least one of NOx emissions and _CO2 emissions;
The boiler operation support device according to the above aspect (1).

According to the above aspect (6), by changing the process value used for the operation evaluation, it is possible to evaluate the economic merit of AI operation control in terms of cost merit, for example, or to evaluate the environmental merit by displaying the merits regarding NOx emissions and _CO2 emissions instead of the operating cost.

(7) A boiler operation support device according to one aspect,
The boiler operation support device according to the above aspect (1),
an operation control device that sets the optimized operation terminal parameters in the boiler and controls the operation of the boiler;
The boiler operation control system is provided with:

According to the above aspect (7), optimized operation can be performed using AI operation control.

(8) A boiler operation support method according to one aspect,
A step of calculating an estimated evaluation index based on an operation simulation result using a learning model that calculates a predicted value of a process value that occurs when the boiler is operated, and deriving an optimized operating end parameter corresponding to an optimal value of the estimated evaluation index;
A step of operating the boiler using the optimized operating end parameters, setting a period to be subject to operation evaluation as an evaluation period, and calculating a first estimated cost estimated to be incurred during the evaluation period;
calculating a second estimated cost that would be incurred during the evaluation period if the boiler were operated using a base control parameter for comparison with the optimized control parameter;
outputting a driving evaluation result based on a difference between the first estimated cost and the second estimated cost;
The present invention relates to a boiler operation support method including the steps of:

According to the above aspect (8), the merits are calculated by comparing the values before and after optimization based on the estimated values of the evaluation index (cost). As a result, since both values compared are estimated (calculated values), the calculations can be performed under the same conditions, and only the effect of optimization can be extracted, allowing for a clear evaluation of the merits. In other words, if actual measured values are used, it is difficult to grasp only the effect of optimization due to the influence of disturbances and measurement errors, but by using estimated values for both, the influence of disturbances and measurement errors can be eliminated.

Furthermore, when calculating the operating costs of AI operation control, actual values are used for process values that can be measured during operation, and predicted values are used only for process values that cannot be measured during operation, minimizing deviations in merit assessment due to prediction errors in the learning model.

(9) A boiler operation assistance program according to one embodiment,
A boiler operation support program for executing the boiler operation support method of the above aspect (8).

According to the above aspect (9), the driving assistance method of the above aspect (8) can be realized by a computer.

The above embodiment does not limit the present invention, and various modifications are possible without departing from the spirit of the present invention.

For example, the functional blocks of the operation evaluation unit 132 are merely an example. Furthermore, the display examples of each graph in FIG. 11 and the process values used in the explanation are not limited to those described above.

10: Operation control system 11: Furnace 12: Combustion device 13: Flue 21, 22, 23, 24, 25: Combustion burner 26, 27, 28, 29, 30: Pulverized coal supply pipe 31, 32, 33, 34, 35: Crusher 36: Wind box 37a, 37b, 37c: Air duct 37d: Connection point 38: Blower 39: After-air port 48: Exhaust gas passage 49: Air heater (AH)
50: Denitrification device 51: Dust treatment device 52: Induced draft fan 53: Chimney 62: Inclined surface 70: Furnace bottom evaporation tube 100: Operation support device 110: Data acquisition unit 112: RTC
114: Driving data storage unit 116: Data extraction unit 118: Data pre-processing unit 120: Model construction unit 122: Model storage unit 124: Optimization unit 132: Driving evaluation unit 132a: AI-on prediction data calculation unit (first estimated cost calculation unit)
132b: AI-off prediction data calculation unit (second estimated cost calculation unit)
132c: Prediction error calculation unit 132d: Accumulation evaluation unit 134: Alternative control logic storage unit 136: Input/output control unit 200: Operation control device 301: Processor 305: Input I/F
306: Output I/F
307: Communication I/F
308: Bus 311: Input device 312: Display

Claims (7)

A boiler operation support device, comprising:
an optimization unit that calculates an estimated evaluation index based on an operation simulation result using a learning model that calculates a predicted value of a process value that occurs when the boiler is operated, and derives an optimized operating end parameter corresponding to an optimal value of the estimated evaluation index;
a first cost estimate calculation unit that calculates a first cost estimate that is estimated to occur during an evaluation period, the evaluation period being a period during which the boiler is operated using the optimized operating end parameters;
a second cost estimate calculation unit that calculates a second cost estimate that is estimated to be incurred during the evaluation period when the boiler is operated using a base control parameter for comparison with the optimized control parameter;
a driving evaluation unit that outputs a driving evaluation result based on a difference between the first estimated cost and the second estimated cost;
Equipped with
the learning model includes a first learning model that calculates a predicted value of a specific unmeasured process value and a second learning model that calculates a predicted value of another process value;
the first estimated cost calculation unit inputs the optimized control input parameter into the first learning model to calculate a predicted value of the specific unmeasured process value, and calculates the first estimated cost based on the predicted value;
the second estimated cost calculation unit inputs the base operating end parameters derived by applying the operating parameters and environmental parameters applied during the evaluation period to an alternative control logic that does not use the learning model into the first learning model and the second learning model to calculate predicted values of the specific unmeasured process value and the other process values, and calculates the second estimated cost based on these predicted values.
Boiler operation support device. The boiler operation support device according to claim 1 ,
The specific non-measured process value is unburned carbon in ash;
Boiler operation support device. The boiler operation support device according to claim 1,
an operation data storage unit that stores actual measurement data including operating conditions of the boiler used in the actual operation of the boiler;
a data extraction unit that extracts actual measurement data for calculating the first estimated cost and the second estimated cost from the operation data storage unit,
The data extraction unit extracts the actual measurement data associated with the same operating conditions as the evaluation period,
the first cost estimate calculation unit and the second cost estimate calculation unit calculate the first cost estimate and the second cost estimate by using the extracted actual measurement data, respectively.
Boiler operation support device. The boiler operation support device according to claim 1,
The driving evaluation result is an economic merit or an environmental merit based on at least one of NOx emissions and CO2 emissions.
Boiler operation support device. The boiler operation support device according to claim 1 ;
an operation control device that sets the optimized operation terminal parameters in the boiler and controls the operation of the boiler;
A boiler operation control system comprising: A boiler operation support method, comprising:
A step of calculating an estimated evaluation index based on an operation simulation result using a learning model that calculates a predicted value of a process value that occurs when the boiler is operated, and deriving an optimized operating end parameter corresponding to an optimal value of the estimated evaluation index;
A step of setting a period during which the boiler is operated using the optimized operating end parameters as an evaluation period and calculating a first estimated cost estimated to be incurred during the evaluation period;
calculating a second estimated cost that would be incurred during the evaluation period if the boiler were operated using a base control parameter for comparison with the optimized control parameter;
outputting a driving evaluation result based on a difference between the first estimated cost and the second estimated cost;
Including,
the learning model includes a first learning model that calculates a predicted value of a specific unmeasured process value and a second learning model that calculates a predicted value of another process value;
The step of calculating the first estimated cost includes inputting the optimized control input parameter into the first learning model to calculate a predicted value of the specific unmeasured process value, and calculating the first estimated cost based on the predicted value;
The step of calculating the second estimated cost includes inputting the base operating end parameter derived by applying the operating parameters and the environmental parameters applied during the evaluation period to an alternative control logic that does not use the learning model into the first learning model and the second learning model to calculate predicted values of the specific unmeasured process value and the other process values, and calculating the second estimated cost based on these predicted values.
A method for supporting boiler operation. A boiler operation support program for executing the boiler operation support method according to claim 6 .

Images