JP7043831B2 - Plant management system, plant management server, plant management device, estimation model generation method, and learning data generation method - Google Patents

Description

The present disclosure relates to a plant management system, a plant management server, a plant management device, a method for generating an estimation model, and a method for generating learning data.

Patent Document 1 discloses a method for controlling a cement kiln. This control method uses fuzzy control to perform automatic operation of the cement kiln based on measured values such as temperature, oxygen concentration, and carbon monoxide concentration at each position of the cement kiln.

U.S. Pat. No. 4,910,684

By the way, in the above-mentioned cement kiln, the internal state of the plant may change irregularly. For example, when a cement kiln is used in a cement clinker firing step related to cement production, the raw material melted by the firing is repeatedly adhered to and peeled off from the inside of the cement kiln. The deposit (ring) of the raw material causes the raw material to stay, and affects the flow of the combustion gas, the combustion state of the fuel, and the like. If the state of the plant changes to the extent that the model as the control target changes significantly in this way, the control that has been executed may not be suitable and automatic operation may not be continued depending on the control method or the like described in Patent Document 1. there were.

On the other hand, skilled human operators maintain a high standard of continuous plant operation. Therefore, the present inventor has found that the automatic operation of the plant can be maintained at a high level by appropriately reflecting the state of the plant.

However, the state of the plant is totally influenced by many factors related to the operating state of the plant and changes, and the control process that keeps sufficiently reflecting the information of many of these factors in order to properly judge the state of the plant. Has become overly complicated.

It is an object of the present disclosure to provide a plant management system, a plant management server, a plant management device, a method for generating an estimation model, and a method for generating learning data, which are useful for determining the state of a plant.

The plant management system according to one aspect of the present disclosure is a time interval and a state in a two-dimensional space having a time axis including a plurality of time intervals arranged in a time series and an item axis including a plurality of state items related to the operating state of the plant. A map data generation unit that generates map data including data for each cell specified by a combination of items, a teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data, a map data, and the corresponding A model for estimating the state of the plant is created by machine learning based on the training data stored in the training data management unit and the training data management unit that stores the training data that combines the teacher data corresponding to the map data. It includes a model building unit to be constructed and an estimation unit that inputs map data to the estimation model and derives estimation data indicating the state of the plant.

According to this plant management system, the pattern of changes in multiple elements related to the operating conditions of the plant is converted into map data for machine learning, so the plant is based on comprehensive changes in operating conditions that allow noise such as instrument abnormalities to some extent. An estimation model that can reflect the changes in the state of the above with high accuracy is constructed. The estimation data obtained from the map data based on this estimation model can be used to appropriately judge the state of the plant. Therefore, this plant management system is useful for determining the state of the plant.

The plant management system inputs the map data of the training data into the estimation model, and when the deviation between the estimation data derived by inputting it into the estimation model and the teacher data of the training data exceeds a predetermined range, the training data The model construction unit further includes a training data selection unit that excludes May be configured to rebuild. In this case, for example, when the change in the teacher data does not correspond to the change in the state of the plant due to sampling error or the like, or when the training data contains data that is not suitable for the training data, the learning is concerned. Since machine learning is performed again by excluding those that are not suitable for the data, the estimation model can be brushed up. This makes it possible to determine the state of the plant more appropriately.

The map data generator may generate map data by applying a fuzzy set membership function to at least a part of the map data. In this case, by fuzzying the data for each cell of the map data, the characteristics of the changes in each operating state that a skilled human operator feels in the map data are likely to appear. Therefore, it is possible to appropriately determine the state of the plant while reducing the processing load of machine learning.

The map data generator may be configured to apply different membership functions to at least two state items. It is conceivable that the appearance of the characteristics of the change in the operating state of the plant differs from each other in a plurality of factors related to the operating state of the plant. By applying different membership functions to state items related to different operating states in this way, it is possible to make fuzzy to the extent that the processing load of machine learning can be reduced for each operating state.

The map data generator is configured to change the membership function applied to the map data data according to external factors, including at least one of the raw material conditions input to the plant and the plant operating environment. May be good. In this case, the characteristics of changes in each operating state are more likely to appear in the map data. Therefore, the processing load of machine learning can be further reduced.

The learning data management unit may correct the map data by shifting the time between the two state items so as to reduce the deviation of the data fluctuation time between at least two state items. In this case, it is possible to prevent the complex characteristics of a plurality of factors related to the operating state of the plant from appearing due to the deviation of the fluctuation time.

The item axis includes at least a first state item, a second state item, and a third state item having a stronger correlation with the first state item as compared with the second state item, and includes training data. The management unit may correct the map data so that the third state item is located between the first state item and the second state item. In this case, it is possible to make it easier for the complex characteristics of a plurality of factors related to the operating state of the plant to appear.

The item axis may include a state item indicating the derivative value of any state item. Thereby, for example, when the element related to the operating state of the plant includes a system error, the system error can be canceled, and the change tendency of the operating state can be displayed in the map data as a feature.

The map data generation unit can generate map data whose cell data format is a data format for pixels. In this case, since a machine learning engine for image processing can be used, it contributes to reducing the processing load of machine learning.

For example, the plant contains a cement kiln, the item axis contains a state item regarding the temperature in the cement kiln and a state item regarding the concentration of gas in the cement kiln, and the teacher data is the free lime in the cement kiln. It may include an evaluation value regarding the quantity.

The plant management server according to another aspect of the present disclosure is a time interval and a time axis in a two-dimensional space including a time axis including a plurality of time intervals arranged in a time series and an item axis including a plurality of state items related to the operating state of the plant. A learning data management unit that stores learning data that combines map data including data for each cell specified by a combination of state items and teacher data indicating the state of the plant corresponding to the map data, and learning data. It includes a model construction unit that constructs a model for estimating the state of the plant by machine learning based on the training data accumulated in the management unit.

Further, the plant management apparatus according to another aspect of the present disclosure is in a two-dimensional space having a time axis including a plurality of time intervals arranged in a time series and an item axis including a plurality of state items related to the operating state of the plant. At least one data acquisition unit that acquires data for generating map data including data for each cell specified by a combination of sections and status items, and teacher data indicating the status of the plant corresponding to the map data. , Map data is input to the plant state estimation model constructed by machine learning based on the training data that combines the map data and the teacher data corresponding to the map data, and the plant state estimation data is derived. It is equipped with an estimation unit.

The plant management server and the plant management device each contribute to an appropriate determination of the state of the plant similar to the above-mentioned plant management system.

The plant management device may further include a display unit that displays the map data input to the estimation model and the estimation data derived by inputting the map data to the estimation model as image data. In this case, by displaying the image data, it is possible to compare the change in the state of the plant based on the comprehensive change in the operating state indicated by the estimated data with the visual recognition of the human operator.

A method of generating an estimation model according to another aspect of the present disclosure is in a two-dimensional space of a time axis including a plurality of time intervals arranged in a time series and an item axis including a plurality of state items related to the operating state of a plant. Accumulation of learning data that combines map data including data for each cell specified by a combination of time intervals and state items and teacher data indicating the state of the plant corresponding to the map data, and for accumulated learning It involves generating a model for estimating the state of the plant by machine learning based on the data.

In addition, the method of generating the estimation model is when the discrepancy between the estimation data derived by inputting the map data of the training data into the estimation model and the teacher data of the training data exceeds a predetermined range. Further includes excluding the training data and reconstructing the estimation model by machine learning based on the remaining training data when any of the training data is excluded. May be good.

The method of generating these estimation models contributes to an appropriate judgment of the state of the plant similar to the plant management system described above.

The method for generating learning data according to another aspect of the present disclosure is a method of generating learning data in a two-dimensional space having a time axis including a plurality of time intervals arranged in a time series and an item axis including a plurality of state items related to the operating state of a plant. For machine learning, map data including data for each cell specified by a combination of time interval and state item is generated as input side data for machine learning, and teacher data regarding the state of the plant corresponding to the map data is generated. Includes generating as output side data.

Further, as a method of generating training data, map data may be generated by applying a fuzzy set membership function to at least a part of the map data.

The method of generating these learning data contributes to an appropriate judgment of the state of the plant similar to the above-mentioned plant management system.

According to the present disclosure, it is possible to provide a plant management system, a plant management server, a plant management device, a method for generating an estimation model, and a method for generating learning data, which are useful for determining the state of a plant.

FIG. 1 is a schematic diagram showing an overall configuration of a plant management system. FIG. 2 is a block diagram showing a functional configuration of the plant management device and the plant management server. FIG. 3 is a diagram showing an example of map data. FIG. 4 is a diagram showing an example of teacher data showing the state of the plant. FIG. 5 is a diagram showing an example of a membership function of a fuzzy set. FIG. 6 is a block diagram showing a hardware configuration of a plant management system. FIG. 7 is a flowchart showing a learning data generation process. FIG. 8 is a flowchart showing the map data generation process. FIG. 9 is a flowchart showing the model construction process. FIG. 10 is a flowchart showing the process of estimating the state of the plant.

Hereinafter, embodiments will be described in detail with reference to the drawings as an example. In the description, the same elements or elements having the same function may be designated by the same reference numerals, and duplicate description may be omitted.

[Plant management system]
The plant management system according to the present embodiment is a system for managing the plant and is used for determining the state of the plant. The plant management system 1 shown in FIG. 1 is applied to, for example, a plant for manufacturing cement clinker. The plant management system 1 includes a plant management device 100 and a plant management server 200.

The plant management device 100 acquires data for generating data for machine learning (hereinafter referred to as "learning data"), and is based on a model for estimating the state of the plant constructed by machine learning. Derivation of estimated data on the state of the plant and execution. The learning data includes the following input side data and output side data.

The input side data corresponds to the data input to the estimation model when deriving the estimation data. The input side data is specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. Map data including data for each cell.

The output side data is data for teaching in advance the data to be output by the estimation model according to the input of the input side data. The output side data is teacher data indicating the state of the plant corresponding to the map data. Corresponding to map data means that there is a relationship that correlates with the map data but is not directly derived from the map data. As a specific example of the teacher data, the data indicating the state of the plant in the future from the acquisition period of the data constituting the map data, or the state of the plant during the acquisition period of the data constituting the map data, and the map data can be used. Examples include data indicating a state that is not directly derived from the constituent data. It should be noted that "indicating the state of the plant" means "evaluating the state of the plant" and is not necessarily limited to the case where the state of the plant is appropriately indicated.

The plant management device 100 further generates map data as input side data for machine learning based on the acquired data, and generates teacher data as output side data for machine learning based on the acquired data. It may be configured to run. Further, the plant management device 100 is configured to further execute to display the map data input to the estimation model and the estimation data derived by inputting the map data to the estimation model as image data. May be.

The plant management server 200 accumulates learning data that combines map data and teacher data, and builds a model for estimating the state of the plant by machine learning based on the accumulated learning data. Execute.

Further, when the deviation between the estimated data derived by inputting the map data of the learning data into the estimation model and the teacher data of the learning data exceeds a predetermined range, the plant management server 200 is concerned. It is configured to further execute the exclusion of training data and the reconstruction of the estimation model by machine learning based on the remaining training data when any of the training data is excluded. It may have been done. Hereinafter, specific configurations of the plant management device 100 and the plant management server 200 will be illustrated in detail.

(Plant management equipment)
The plant management device 100 is a device for managing the state of the cement kiln 12 in the plant 10 for manufacturing cement clinker. The plant 10 has a preheater 11, a cement kiln 12, a cooler 13, and a plurality of sensors 14. In FIG. 1, the flow of the raw material of the cement clinker is indicated by a solid line arrow, and the flow of the high temperature gas is indicated by a broken line arrow.

The preheater 11 preheats the raw material of the cement clinker supplied from the apparatus (not shown) in the previous step (raw material crushing step). The preheater 11 has, for example, a plurality of cyclones 11a and a bottom cyclone 11c constituting a multi-stage cyclone. The preheater 11 sequentially drops the cement clinker raw material charged from the input port of the cyclone 11a (for example, the inlet of the uppermost cyclone) to the lower cyclone 11a so that the cement clinker raw material gradually becomes hot. Preheat.

The preheater 11 may further have a calciner 11b. That is, the preheater 11 may be an SP method (Suspension Preheater method: multi-stage cyclone preheating method) or an NSP method (New Suspension Preheater method: multi-stage cyclone preheating method provided with a calcining furnace). The calciner 11b is located at the gas side inlet of the bottom cyclone 11c, and the raw material of the cement clinker is calcined by the burner 11d. The bottom cyclone 11c is located between the calcining furnace 11b and the cement kiln 12, and supplies the raw material of the preheated cement clinker to the cement kiln 12.

The cement kiln 12 fires the raw material of the cement clinker supplied from the preheater 11. The cement kiln 12 is, for example, a rotary kiln, and has a drum body 12a inclined to the downstream side of the raw material along the rotation axis direction and a burner 12b. The drum body 12a includes a kiln tail 12c for receiving the raw material supply and a kiln front 12d for discharging the fired cement clinker. The drum body 12a rotates around a rotation axis inclined by 2 ° to 5 ° with respect to the horizontal plane so that the kiln front 12d is located below the kiln tail 12c, for example. As a result, the drum body 12a stirs the supplied raw material of the cement clinker while moving it, and fires it with the burner 12b to generate the cement clinker.

The cooler 13 cools the cement clinker discharged from the cement kiln 12. The cooler 13 is, for example, an air quenching cooler (clinker cooler), which cools the high-temperature cement clinker with air and supplies the high-temperature gas generated by the cooling to the calciner 11b.

The plurality of sensors 14 measure the value of each element with respect to the operating state of the plant 10. For example, the plurality of sensors 14 include sensors 14a, 14b, 14c, 14d, 14e. The sensors 14a and 14d are, for example, thermocouples and measure the temperature of the gas (air or bleed air). The sensors 14b and 14c are, for example, gas analyzers, and measure the oxygen concentration and the carbon monoxide concentration, respectively. The sensor 14e is, for example, a free lime detector, which measures the amount of free lime in the cement clinker. The free lime is a lime (CaO) that has not reacted to become a cement mineral. The sensors 14a, 14b, 14c may be arranged at the preheater 11 (for example, the outlet of the bottom cyclone 11c or the outlet of the calciner 11b or the outlet of the preheater 11 on the gas side), and the sensors 14a, 14b, 14c may be arranged in the kiln of the cement kiln 12. It may also be arranged in the buttock 12c. The sensor 14d is arranged, for example, in the flow path of the high temperature gas from the cooler 13 to the calciner 11b. The sensor 14e is arranged, for example, at the outlet of the cooler 13.

The plurality of sensors 14 are not limited to the above, but are a sensor for measuring the amount of the raw material of the cement clinker supplied to the preheater 11, a sensor for measuring the production amount of the cement clinker, and measuring the concentration of carbon dioxide. Sensor, sensor for measuring the concentration of nitrogen oxides, sensor for measuring the temperature of the raw material of the cement cleaner, sensor for detecting the power or torque or rotation speed of the rotation of the cement kiln 12, exhaust gas fan of the preheater 11 (not) It may include a sensor for detecting the number of revolutions (shown in the figure), a sensor for measuring the amount of fuel in the burners 11d and 12b, a sensor for measuring the weight of the cement cleaner, and the like.

As shown in FIG. 2, the plant management device 100 has a management unit 110, a data acquisition unit 101, a map data generation unit 103, and teacher data as a functional configuration (hereinafter referred to as “functional module”). It includes a generation unit 105, an estimation unit 106, and a display unit 107.

The management unit 110 manages the data. The management unit 110 stores the map data and the teacher data in the learning data storage unit 102, and stores the membership function to be applied to at least a part of the map data in the membership function storage unit 104. And execute.

The data acquisition unit 101 acquires data for generating map data and teacher data, and stores the acquired data in the data storage unit 101a. For example, the data acquisition unit 101 acquires data indicating the state of the plant from the sensor 14 or the like. The data acquisition unit 101 may acquire information that is not directly acquired by the sensor 14, such as weather information and operator identification information, based on the input of the operator. The data acquisition unit 101 stores the data of the data storage unit 101a in the learning data storage unit 102 of the management unit 110.

The map data generation unit 103 generates map data based on the data stored in the learning data storage unit 102, and stores the generated map data in the map data storage unit 103a.

FIG. 3 is a diagram illustrating map data generated by the map data generation unit 103. As shown in FIG. 3, the map data 51 is in a two-dimensional space having a time axis 51t including a plurality of time intervals arranged in a time series and an item axis 51p including a plurality of state items related to the operating state of the plant 10. Includes data for each cell 51c specified by a combination of time intervals and state items. In the map data 51, the time axis 51t is the horizontal axis and the item axis 51p is the vertical axis, but the time axis 51t is the vertical axis and the item axis 51p is the horizontal axis. May be good. In the time axis 51t, the direction away from the item axis 51p (for example, the right direction in the figure) is the time elapsed direction, but the time is not limited to this, and the direction approaching the item axis 51p (for example, the left direction in the figure) is the time elapsed direction. It may be said. Further, a plurality of time intervals may be arranged continuously or intermittently. The "time interval" may be one time indicating a specific moment, or may be a period specified by the start time and the end time.

The item axis 51p may include all state items related to the operating state of the plant, such as weather information, operator identification information, etc., in addition to including state items indicated by numerical data such as temperature. The item axis 51p may include a state item indicating a differential value of any of the state items. For example, the state item a of the item axis 51p is the temperature of the air at the point A (for example, the outlet of the preheater-11). The state item b is the concentration of the first gas (for example, oxygen) at the point A. The state item c is the concentration of the second gas (for example, carbon monoxide) at the point A. The state item d is a derivative value of the state item c. The state item e is the temperature of the bleed air at the point B (for example, the flow path of the high temperature gas from the cooler 13 to the calciner 11b). The state item f is the amount of free lime in the cement clinker.

If there is no data for a specific time interval as data related to a certain state item in the map data 51, the data itself for the time interval close to the time interval in the same state item, or a plurality of surrounding time intervals. The data obtained by the calculated value approximated from the data may be substituted. It may also be shown that there is no data. When indicating that there is no data, map data 51 may be generated in which the data in the corresponding cell 51c is represented by data indicating that there is no data (for example, data b1 described later), and the cell 51c is missing. The map data 51 of the outer shape may be generated.

The map data generation unit 103 may generate the map data 51 by applying a fuzzy set membership function to at least a part of the data of the map data 51. FIG. 5 is a diagram illustrating a membership function of a fuzzy set. The two graphs G in FIG. 5 both show the membership function F. For example, graph G1 shows a membership function F1 of a fuzzy set of data degrees (eg, a fuzzy set of high temperatures). Further, the graph G2 in FIG. 5 shows the membership function F2 of the fuzzy set (crisp set) regarding the presence / absence of data. The vertical axis of the graph G1 indicates the grade of the fuzzy set (the degree of agreement with respect to the high temperature), and the horizontal axis indicates the value of the data. The vertical axis of the graph G2 indicates the grade of the fuzzy set (the degree of coincidence with respect to the existence of data), and the horizontal axis indicates the presence or absence of data (positive when there is data and false when there is no data). .. There is no intermediate data between positive and false. As the grade range, the fuzzy set may use the entire range from 0 to 1, or may use only a part of the grade.

The map data generation unit 103 may generate map data 51 in which the data for each cell 51c specified by the combination of the time axis 51t and the item axis 51p is a data format for pixels. For example, the map data generation unit 103 may generate the map data 51 by using the membership function F indicating the grade in the data format for pixels. As the data format for pixels, the first two digits of a 6-digit hexadecimal number (000000 to FFFFFF) indicate the brightness (strength) of R (red), and the middle two digits indicate the brightness of G (green). An RGB color code indicating the brightness (strength) of B (blue) by the last two digits can be mentioned. As for the hue range corresponding to the grade range, if part or all of about 400 nm to about 700 nm, which corresponds to the wavelength of visible light, is used in ascending order of wavelength or vice versa, it becomes the same as a rainbow and is human. Easy to understand for some operators. When the hue is determined by linearly associating the wavelength and the grade, the color suddenly changes and looks unnatural to humans, so it may be determined by a method defined by some color spaces. At that time, the higher the saturation, the easier it is to understand the change.

At least one of saturation and lightness may be normalized from 0 to 1 and all or part of it may be pre-corresponded to the grade range. Since a general color display for a computer has a structure in which a range of lightness is adjusted and an additive color is mixed and displayed, it is easy to calculate the display if the grade corresponds to such a range of lightness. If you use the full range of brightness, the image tends to be dark, but once you get used to it, you can see it without any discomfort. In order to reduce this discomfort, only the high-brightness range may correspond to the grade range, not the full range. The number of hues can be set to an arbitrary number by preparing a memory for the number of hues so that the machine can recognize it as another primary color. Even in humans, in addition to the three primary colors of red, green, and blue, some people feel yellow as the fourth independent color, and the hue is not necessarily limited to the three primary colors. It has a high affinity for human vision in the range of colors that are a mixture of the three primary colors of red, green, blue, or cyan (light blue), magenta, and yellow (yellow), plus the colorless primary colors of white and black. be. Therefore, when considering human visibility, the number of hues is three of the three primary colors, or five of the five colors that correspond to the neutral color between the two that are created when the three primary colors are arranged on a straight line. You may. Specific examples of the five colors include red, yellow, green, cyan and blue.

A plurality of types of membership functions F corresponding to each of a plurality of types of hues may be applied to the same data, and a plurality of types of colors obtained thereby may be combined to derive data for each cell 51c. For example, the map data generation unit 103 applies the membership function F of three fuzzy sets of "large", "medium", and "small" corresponding to the three primary colors such as red, green, and blue to the same data. Then, the data for each cell 51c may be derived by synthesizing the three kinds of colors obtained by the synthesis. The three fuzzy sets may be any one that expresses the value (number) of the data sensuously (qualitatively), and in addition to large / medium / small, high / just good / low, and many / normal / few. And so on.

When using three fuzzy sets, the brightness and saturation are low between "large" and "medium", and between "medium" and "small". On the other hand, if five fuzzy sets such as "large", "slightly large", "medium", "slightly small", and "small" corresponding to five colors including three primary colors and intermediate colors are used, "large" is used. It is possible to increase the saturation and lightness between "medium" and "medium" and between "medium" and "small". The five fuzzy sets may be "very large", "large", "medium", "small" and "very small".

Further, the above membership functions F1 and F2 may be used in association with the color element. For example, in the example shown in FIG. 5, the map data generation unit 103 applies the membership functions F1 and F2 to the data of each cell, and the colors obtained from the charts C1 and C2 corresponding to the grades of the graphs G1 and G2. Map data 51 is generated by using data in which the elements c01 and c02 are mixed according to the data format as the color of each cell. The chart C1 is, for example, a color chart, and is configured by arranging the most saturated color wheels in the HSV (Hue, Saturation, Value) color space in a predetermined hue order (for example, from blue to red via green). Will be done. The chart C1 may be a continuous color chart or a discontinuous color chart. The chart C2 is, for example, a color chart, and is composed of a color element c02 (for example, transparent) and a designated color fill (for example, black b01). As an example, in FIG. 5, when the state item a (temperature at point A) in the time interval t3 is 380 ° C. (data value: 380, data is present: positive), the membership functions F1 and F2 are set. The applied data is evaluated on the charts C1 and C2, respectively, to obtain the color elements c01 and c02, and further, the color elements c01 and c02 are mixed to show that the cell pixel becomes the data c1. .. Further, as in the cell 51c specified by the time interval t8 and the state item f (amount of free lime in the cement clinker) of the map data 51 in FIG. 3, for example, when the corresponding data has not been acquired yet (data of the data). Value: None, Data exists: False), and by applying the membership functions F1 and F2, the pixel of the cell 51c becomes the data b1.

The map data generation unit 103 may be configured to apply different membership functions F to at least two state items. For example, for state items that require analog evaluation, such as high temperature, a membership function that gradually increases the grade as the value of temperature etc. rises is applied to "normal" and "normal". For state items that require binary evaluation, such as "abnormal", a membership function whose grade changes stepwise according to changes in the value may be applied.

The map data generation unit 103 changes the membership function F applied to the data of the map data 51 according to external factors including at least one of the state of the raw material input to the plant 10 and the product and the operating environment of the plant 10. It may be configured to do so. For example, the map data generation unit 103 acquires and uses the membership function F corresponding to the external factor from the membership function storage unit 104. The membership function storage unit 104 stores a plurality of types of membership functions preset for each state item so that they can be used properly according to external factors. In this case, by properly using the membership function F, it is possible to suppress the influence of the external environment and generate map data 51 having substantially similar tendencies in any external environment.

Returning to FIG. 2, the teacher data generation unit 105 evaluates the state of the plant 10 corresponding to the map data 51 and generates teacher data. For example, the teacher data generation unit 105 acquires map data 51 and future data from the map data 51 from the learning data storage unit 102, and generates teacher data indicating the state of the plant 10 corresponding to the map data 51. , The generated teacher data is stored in the teacher data storage unit 105a.

FIG. 4 is a diagram illustrating teacher data generated by the teacher data generation unit 105. The teacher data 52 shown in FIG. 4 is, for example, an evaluation value of the amount of free lime in the map data 51A in the future than the map data 51 (for example, a time interval following the time interval t1, t2, t3, t4, t5, t6, t7). Evaluation value of the amount of free lime at t8, t9, t10) is included. The evaluation value is a qualitative evaluation value such as "good", "normal" or "bad". The evaluation value may be a qualitative evaluation value for a change tendency such as "improvement tendency" or "deterioration tendency". Further, the evaluation value may be a quantitative evaluation value such as a value obtained by weighted averaging the amount of free lime itself or the amount of free lime over a predetermined number of time intervals. The evaluation target data may be data other than the amount of free lime. For example, the data to be evaluated may be the concentration of chlorine ( ^Cl- ) in the cement clinker, the concentration of sulfur ( ^SO3-2- ) in the cement clinker, etc., and the results will be obtained after a long time. Something like the strength of the cement produced by the cement clinker (eg, the strength of the mortar or concrete containing the cement at a given age) may be included.

Further, the teacher data 52 may include a plurality of the above evaluation values, and may include data obtained by evaluating the above evaluation values with something like an evaluation function. Further, the teacher data 52 does not have to include future data. Further, if a binary evaluation value such as "good" or "bad" is used as the teacher data 52, machine learning is performed except for the map data 51 of the evaluation of "bad", so that there is no so-called teacher. It is also possible to learn.

The teacher data 52 may not be suitable as learning data due to sampling errors or the like, and such data may be discarded from the learning data together with the map data by the learning data selection unit 204 described later. Therefore, it is not necessary to determine whether the teacher data generation unit 105 is an appropriate teacher.

Returning to FIG. 2, the estimation unit 106 outputs estimation data 53 indicating the state of the plant 10 based on the estimation model of the model storage unit 205 of the state of the plant 10 constructed by machine learning based on the learning data. .. The estimation model is a program module that outputs estimation data 53 indicating the state of the plant 10 in response to input of map data 51. Specific examples of the estimation model include a neural network connecting an input vector including map data 51 and an output vector including teacher data 52. For example, the estimation unit 106 acquires map data 51 different from the map data 51 used for deriving the estimation model from the learning data storage unit 102, and inputs the map data 51 into the estimation model (for example, described later). (Send to the estimation processing unit of) to derive the estimation data 53 (for example, acquire from the estimation processing unit described later).

The display unit 107 displays the map data 51 input by the estimation unit 106 into the estimation model and the estimation data 53 derived by the estimation unit 106 as image data. The display unit 107 may display a plurality of map data 51 arranged in chronological order as a moving image. The moving image display can encourage the operator to notice the change.

(Plant management server)
The plant management server 200 has a learning data management unit 210, a learning data acquisition unit 201, a data storage unit 202, a model construction unit 203, a learning data selection unit 204, and a model storage unit 205 as functional modules. And an estimation processing unit 206.

The learning data acquisition unit 201 acquires learning data in which map data 51 and teacher data 52 are combined from the learning data storage unit 102 of the plant management device 100.

The learning data management unit 210 stores the learning data acquired by the learning data acquisition unit 201 in the data storage unit 202 and corrects the map data 51.

The learning data management unit 210 corrects the map data 51 by shifting the time between the two state items so as to reduce the deviation of the fluctuation time of the data of the map data 51 between at least two state items. do. For example, the learning data management unit 210 sets the time so as to reduce the time lag of the data fluctuation of the map data 51 between two state items in which the data fluctuation due to a common phenomenon appears with a time difference. The map data 51 may be corrected by shifting. The time to be shifted (hereinafter referred to as “shift amount”) is preset as a standard delay time. For example, when the data of the state item e including the data c2 (the temperature of the point B) includes the standard delay time Δt (so-called waste time) with respect to the data of the state item a (the temperature of the point A), the delay time Δt. The entire data of the state item e may be shifted to the past side in. On the contrary, the entire data of the state item a may be shifted to the future side by the delay time Δt.

The shift amount may be set to a constant value by using statistical analysis or the like. While shifting one state item (for example, state item e), the shift amount may be changed so that the correlation with the other state item (for example, state item a) is maximized.

In the case of the cement kiln 12, for example, the standard delay time Δt may change significantly due to the weir of the ring (the deposit inside the kiln of the raw material melted by firing). In such a case, the magnitude of the delay time Δt itself can be an important index of the state of the cement kiln 12. When the magnitude of the delay time Δt is also used as an index of the state of the cement kiln 12, it may be effective to keep the shift amount constant without changing the shift amount according to the change of the delay time Δt.

If the time of the state item is shifted, the data at the end of the time axis 51t will be lost. Such missing data may be treated as no data (for example, the above data b1), or the outer shape of the map data 51 may be changed according to the missing data.

For learning when the item axis 51p includes a first state item, a second state item, and a third state item having a stronger correlation with the first state item than the second state item. The data management unit 210 may adjust the map data 51 so that the third state item is located between the first state item and the second state item. For example, the item axis 51p of the map data 51 in FIG. 3 has a correlation between the state item a (first state item), the state item b (second state item), and the state item a as compared with the state item b. It includes a strong state item e. In such a case, the learning data management unit 210 may adjust the map data 51 so that the state item e is located between the state item a and the state item b. Further, the learning data management unit 210 acquires a state item related to data acquired downstream from a state item related to data acquired upstream (for example, a state item a having a temperature acquired at point A) (for example, acquired at point B). The map data 51 may be adjusted so that the state items are arranged in the order of the state items e) which are the temperatures to be measured. Alternatively, the map data 51 may be adjusted in which state items related to information that trigger various phenomena are close to each other.

Further, the learning data management unit 210 stores the data used when the map data 51 is corrected (hereinafter, referred to as “correction parameter”) in the data storage unit 202.

The model building unit 203 builds a model for estimating the state of the plant 10 by machine learning based on the learning data stored in the data storage unit 202, and stores it in the model storage unit 205. For example, the model construction unit 203 tunes the weighting parameters of the nodes of the neural network by deep learning.

The learning data selection unit 204 inputs the map data 51 of the learning data into the estimation model stored in the model storage unit 205 through the model construction unit 203 to derive the estimation data 53, and the estimation data 53 and the estimation data 53. When the deviation of the learning data from the teacher data 52 exceeds a predetermined range, the learning data is excluded from the learning target. For example, the learning data selection unit 204 writes flag data indicating exclusion of the learning data to the data storage unit 202, and notifies the model construction unit 203 that the learning data has been excluded. In response to this, the model building unit 203 excludes the learning data to which the flag data is attached, and reconstructs the estimation model stored in the model storage unit 205.

The estimation processing unit 206 acquires map data 51 from the estimation unit 106 of the plant management device 100, inputs the map data 51 into the estimation model of the model storage unit 205, derives the estimation data 53, and derives the estimation data 53. Is transmitted to the estimation unit 106. For example, the estimation processing unit 206 applies the correction parameters stored in the learning data management unit 210 to the map data 51, inputs the corrected map data 51 to the estimation model, and estimates the derived estimation data 53. It is transmitted to the unit 106.

(Hardware configuration of plant management system)
FIG. 6 is a block diagram showing a hardware configuration of the plant management system 1. As shown in FIG. 6, the plant management device 100 has a circuit 121. The circuit 121 has at least one processor 122, a memory 123, a storage 124, a network adapter 125, an input / output port 126, a monitor 127, an input device 128, and a timer 129.

The storage 124 is a non-volatile storage medium (eg, hard disk or flash memory) that can be read by a computer. The storage 124 includes a storage area of a program for constituting each functional module of the plant management device 100, and a storage area allocated to a storage unit such as the learning data storage unit 102. The memory 123 temporarily records the program loaded from the storage 124, the calculation result by the processor 122, and the like. The processor 122 constitutes each functional module of the plant management device 100 by executing the above program in cooperation with the memory 123.

The network adapter 125 performs network communication via the network line NW in response to a command from the processor 122. The input / output port 126 inputs / outputs an electric signal to / from the sensor 14 and the like. The monitor 127 is, for example, an image display device such as a liquid crystal monitor, and is used for displaying information to an operator, for example, as the display unit 107 or the like. The input device 128 is, for example, an information input device such as a keypad, and is used, for example, for inputting external factors such as the state of raw materials input to the plant 10 to the data acquisition unit 101 and the operating environment of the plant 10. The timer 129 measures the elapsed time, for example, by counting a reference pulse having a fixed cycle.

The plant management server 200 has a circuit 221. The circuit 221 has at least one processor 222, a memory 223, a storage 224, a network adapter 225, and a timer 226.

The storage 224 is a non-volatile storage medium (eg, hard disk or flash memory) readable by a computer. The storage 224 includes a storage area of a program for constituting each functional module of the plant management server 200, and a storage area allocated to the data storage unit 202 and the model storage unit 205. The memory 223 temporarily records the program loaded from the storage 224, the calculation result by the processor 222, and the like. The processor 222 constitutes each functional module of the plant management server 200 by executing the above program in cooperation with the memory 223. The network adapter 225 performs network communication via the network line NW in response to a command from the processor 222. The timer 226 measures the elapsed time, for example, by counting a reference pulse having a fixed cycle.

The hardware configuration of the plant management system 1 is not necessarily limited to the one that configures each functional module by a program. For example, each functional module of the plant management device 100 and the plant management server 200 may be configured by a dedicated logic circuit or an ASIC (Application Specific Integrated Circuit) in which the logic circuit is integrated.

[Learning data generation method]
Subsequently, as an example of the learning data generation method, the contents of the learning data generation process executed by the plant management device 100 will be described.

FIG. 7 is a flowchart showing a learning data generation process. As shown in FIG. 7, the plant management device 100 first executes step S01. In step S01, the management unit 110 waits for the data for the map data generation unit 103 to generate the map data 51 to be accumulated in the learning data storage unit 102 from the data acquisition unit 101.

Next, the plant management device 100 executes step S02. In step S02, the map data generation unit 103 acquires the data for generating the map data 51 from the learning data storage unit 102, and generates and stores the map data 51. The details of the map data 51 generation process will be described later.

Next, the plant management device 100 executes step S03. In step S03, the teacher data generation unit 105 acquires data for generating the teacher data 52 from the learning data storage unit 102, generates and stores the teacher data 52.

The map data 51 and the teacher data 52 generated as described above are transmitted to the management unit 110 and stored in the learning data storage unit 102 to complete the learning data generation process. The plant management device 100 repeatedly executes the above learning data generation process.

(Map data generation process)
Subsequently, with reference to FIG. 8, the details of the map data generation process will be described. As shown in FIG. 8, the plant management device 100 first executes step S11. In step S11, the management unit 110 selects the membership function F to be applied to each state item. For example, the membership function F to be applied to the data of the map data 51 is selected according to an external factor including at least one of the state of the raw material input to the plant 10 and the operating environment of the plant 10.

Next, the plant management device 100 executes step S12. In step S12, the management unit 110 acquires all the data necessary for generating the map data 51 and the teacher data 52 from the data acquisition unit 101.

Next, the plant management device 100 executes steps S13 and S14 in order. In step S13, the map data generation unit 103 applies the membership function F to the data acquired in step S12 to generate the data in the corresponding cell 51c. In step S14, the data generated by each cell 51c is stored in the map data storage unit 103a of the map data generation unit 103.

Next, the plant management device 100 executes step S15. In step S15, the map data generation unit 103 confirms whether or not the data storage for the predetermined period (for example, the time interval t1 to t8) has been completed. When the plant management device 100 determines that the data storage for the predetermined period has not been completed, the process returns to step S13. After that, the plant management device 100 repeatedly executes the processes of steps S13 and S14 until it is determined that the data storage for the predetermined period is completed. As a result, all the data of each cell 51c specified by the combination of the time axis 51t and the item axis 51p is saved. This completes the map data generation process. The plant management device 100 repeatedly executes the above map data generation process.

[Model construction method]
Next, as an example of the model construction method, the contents of the model construction process executed by the plant management server 200 will be described. FIG. 9 is a flowchart showing the model construction process. As shown in FIG. 9, the plant management server 200 first executes step S21. In step S21, the learning data acquisition unit 201 acquires learning data from the plant management device 100 and stores the learning data in the data storage unit 202.

Next, the plant management server 200 executes step S22. In step S22, the model building unit 203 confirms whether or not a predetermined number of learning data has been accumulated in the data storage unit 202. When the plant management server 200 determines that a predetermined number of learning data has not been accumulated, the process returns to step S21. After that, the plant management server 200 repeatedly acquires and stores the learning data until it is determined that a predetermined number of learning data has been accumulated.

When it is determined in step S22 that a predetermined number of learning data has been accumulated, the plant management server 200 executes step S23. In step S23, the learning data management unit 210 adjusts the time between at least two state items of the map data 51 so as to reduce the deviation of the data fluctuation time. For example, between the state items a and e in FIG. 3, the entire data of the state item e including the data c2 is adjusted to a time interval shifted by the delay time Δt. The learning data management unit 210 may rewrite the map data 51 of the accumulated learning data to the map data 51 corrected by the adjustment.

Next, the plant management server 200 executes step S24. In step S24, when the item axis 51p includes a first state item, a second state item, and a third state item having a stronger correlation with the first state item as compared with the second state item. In addition, the learning data management unit 210 adjusts the map data 51 so that the third state item is located between the first state item and the second state item. For example, the item axis 51p of the map data 51 in FIG. 3 correlates with the state item a (first state item), the state item b (second state item), and the state item a as compared with the state item b. Includes a strong state item e. In such a case, the learning data management unit 210 adjusts the map data 51 so that the state item e is located between the state item a and the state item b. Further, the learning data management unit 210 acquires a state item related to data acquired downstream from a state item related to data acquired upstream (for example, a state item a having a temperature acquired at point A) (for example, acquired at point B). The map data 51 may be adjusted so that the state items are arranged in the order of the state items e) which are the temperatures to be measured. Alternatively, the learning data management unit 210 may adjust the map data 51 in which the state items related to the information that triggers various phenomena are close to each other. The learning data management unit 210 may rewrite the map data 51 of the accumulated learning data to the map data 51 corrected by the adjustment.

Further, the learning data management unit 210 stores data (the above-mentioned correction parameter) of the delay time and the order of the state items used for the correction of the map data 51 of the data storage unit 202 used in step S23 and step S24. I will do it.

Next, the plant management server 200 executes step S25. In step S25, the model building unit 203 constructs a model for estimating the state of the plant 10 by machine learning based on the learning data accumulated and corrected in the data storage unit 202, and stores it in the model storage unit 205.

Next, the plant management server 200 executes steps S26 and S27 in order. In step S26, the learning data selection unit 204 inputs the map data 51 of the learning data into the estimation model of the model storage unit 205 and derives the estimation data 53. In step S27, the learning data selection unit 204 determines whether or not the discrepancy between the estimated data 53 and the teacher data 52 of the learning data used for deriving the estimated data 53 exceeds a predetermined range.

When it is determined in step S27 that the deviation between the estimated data 53 and the teacher data 52 exceeds the range, the plant management server 200 executes step S28. In step S28, the learning data selection unit 204 excludes the learning data used for deriving the estimated data 53 from the learning target. For example, the learning data selection unit 204 writes flag data indicating exclusion of the learning data in the data storage unit 202.

Next, the plant management server 200 executes step S29. When it is determined in step S27 that the deviation between the estimated data 53 and the teacher data 52 does not exceed the range, the plant management server 200 omits step S28 and executes step S29. In step S29, the learning data selection unit 204 confirms whether or not the check of all the learning data stored in the data storage unit 202 is completed.

When it is determined in step S29 that the check of all the learning data stored in the data storage unit 202 has not been completed, the plant management server 200 returns the process to step S26. After that, the plant management server 200 repeatedly executes the processes of steps S26 to S28 until it is determined that the check of all the learning data stored in the data storage unit 202 is completed.

When it is determined in step S29 that the check of all the learning data stored in the data storage unit 202 is completed, the plant management server 200 executes step S30. In step S30, whether or not any of the learning data stored in the data storage unit 202 is excluded from the learning target by the model construction unit 203 (for example, whether or not there is a data exclusion notification from the learning data selection unit 204). Or) is judged.

When it is determined in step S30 that any of the learning data stored in the data storage unit 202 is excluded from the learning target, the plant management server 200 executes step S31. In step S31, the model building unit 203 reconstructs the estimation model by machine learning based on the remaining learning data. For example, the model building unit 203 excludes the learning data to which the flag data is attached, and reconstructs the estimation model by machine learning based on the learning data to which the flag data is not attached. After that, the plant management server 200 returns the process to step S26. After that, the exclusion of the training data and the reconstruction of the estimation model are repeated until there is no training data to be excluded. If it is determined in step S30 that there is no excluded learning data, the plant management server 200 completes the model construction process. The plant management server 200 repeatedly executes the above model construction process.

[Plant state estimation method]
Next, as an example of the method of estimating the state of the plant 10, the contents of the estimation process executed by the plant management device 100 will be described. FIG. 10 is a flowchart showing the process of estimating the state of the plant. As shown in FIG. 10, the plant management device 100 first executes step S41. In step S41, the map data generation unit 103 waits for the data for generating the map data 51 to be accumulated in the learning data storage unit 102.

Next, the plant management device 100 executes step S42. In step S42, the map data generation unit 103 acquires data for generating the map data 51 from the learning data storage unit 102, and generates the map data 51. The map data generation unit 103 stores the generated map data 51 in the learning data storage unit 102.

Next, the plant management device 100 executes step S43. In step S43, the estimation unit 106 transmits the map data 51 acquired from the learning data storage unit 102 to the estimation processing unit 206, inputs the map data 51 corrected by the correction parameters into the estimation model, and derives the map data 51. The estimated data 53 is acquired from the estimation processing unit 206. At this time, the estimation processing unit 206 applies the correction parameter stored in the learning data management unit 210 to the map data 51, inputs the corrected map data 51 to the estimation model, and derives the estimation data 53. Is transmitted to the estimation unit 106.

Next, the plant management device 100 executes step S44. In step S44, the estimation unit 106 receives the map data 51 input by the estimation processing unit 206 into the estimation model and the estimation data 53 derived by the estimation processing unit 206, and the display unit 107 displays them as image data. This completes the estimation process. The plant management device 100 repeatedly executes the above estimation process.

[Effect of this embodiment]
The plant management system 1 according to one aspect of the present disclosure includes a time axis 51t including a plurality of time intervals (for example, time intervals t1, t2, t3, t4, t5, t6, t7, t8) arranged in a time series, and a plant 10. In a two-dimensional space with an item axis 51p including a plurality of state items (for example, state items a, b, c, d, e, f) related to the operating state of, for each cell 51c specified by a combination of time intervals and state items. The map data generation unit 103 that generates the map data 51 including the data of the above, the teacher data generation unit 105 that generates the teacher data 52 indicating the state of the plant 10 corresponding to the map data 51, the map data 51, and the map data. The state of the plant 10 is estimated by machine learning based on the learning data stored in the learning data management unit 210 and the learning data management unit 210 that stores the learning data in combination with the teacher data 52 corresponding to the 51. It includes a model construction unit 203 for constructing a model for use, and an estimation unit 106 for inputting map data 51 into the estimation model and deriving estimation data 53 indicating the state of the plant 10.

According to this plant management system 1, since the pattern of change of a plurality of elements related to the operating state of the plant 10 is converted into map data and machine-learned, noise such as an instrument abnormality can be tolerated to some extent as a comprehensive change in the operating state. An estimation model capable of reflecting the change in the state of the plant 10 based on it with high accuracy is constructed. The state of the plant 10 can be appropriately determined from the estimation data 53 obtained from the map data based on this estimation model. Therefore, it is possible to appropriately determine the operating state such as normality, abnormality, or failure of operation. Therefore, the plant management system 1 is useful for determining the state of the plant 10.

The plant management system 1 receives when the deviation between the estimation data 53 derived by inputting the map data 51 of the training data into the estimation model and the teacher data 52 of the training data exceeds a predetermined range. A training data selection unit 204 for excluding the training data is further provided, and the model construction unit 203 is based on the remaining training data when any of the training data is excluded by the training data selection unit 204. It may be configured to reconstruct the estimation model by machine learning. In this case, for example, when the change in the teacher data does not correspond to the change in the state of the plant 10 due to sampling error or the like, or when the training data contains data that is not suitable for the training data, the relevant data is applicable. Since machine learning is performed again by excluding those that are not suitable for training data, the estimation model can be brushed up. This makes it possible to more appropriately determine the state of the plant 10.

The map data generation unit 103 may generate the map data 51 by applying the membership function F of the fuzzy set to at least a part of the data of the map data 51. In this case, by fuzzying the data for each cell 51c of the map data 51, the characteristics of changes in each operating state that a skilled human operator feels are likely to appear in the map data 51. Therefore, the state of the plant 10 can be appropriately determined while reducing the processing load of machine learning.

The map data generation unit 103 may be configured to apply different membership functions F to at least two state items. It is conceivable that in a plurality of factors relating to the operating state of the plant 10, the susceptibility to the appearance of the change in the operating state is different from each other. By applying different membership functions F to the state items related to the different operating states in this way, it is possible to make fuzzy according to the degree to which the processing load of machine learning can be reduced for each operating state.

The map data generation unit 103 uses a membership function F to be applied to the data of the map data 51 according to an external factor including at least one of the state of the raw material of the cement clinker input to the plant 10 and the operating environment of the plant 10. It may be configured to change. In this case, the characteristics of changes in each operating state are more likely to appear in the map data 51. Therefore, the processing load of machine learning can be further reduced.

The learning data management unit 210 takes time between the two state items a and e so as to reduce the deviation of the data fluctuation time between at least two state items (for example, the state items a and e). May be shifted to correct the map data 51. In this case, it is possible to prevent the complex characteristics of the plurality of factors related to the operating state of the plant 10 from appearing due to the deviation of the fluctuation time.

The item axis 51p has at least a correlation between the first state item (for example, state item a), the second state item (for example, state item b), and the first state item as compared with the second state item. Including a strong third state item (eg, state item e), the learning data management unit 210 maps data such that the third state item is located between the first state item and the second state item. 51 may be corrected. In this case, it is possible to make it easier for the complex characteristics of the plurality of elements related to the operating state of the plant 10 to appear.

The item axis 51p may include a state item (for example, state item d) indicating a differential value of any state item (for example, state item c). Thereby, for example, when the element related to the operating state of the plant 10 includes a system error, the system error can be canceled, and the change tendency of the operating state can be shown in the map data 51 as a feature.

The map data generation unit 103 may generate map data 51 whose data format is a data format for pixels. In this case, since a general-purpose machine learning engine for pixel recognition can be used, it contributes to reducing the processing load of machine learning.

The plant management device 100 further includes a display unit 107 that displays map data 51 input to the estimation model and estimation data 53 derived by inputting the map data 51 to the estimation model as image data. May be good. In this case, by displaying the image data, it is possible to compare the change in the state of the plant 10 based on the comprehensive change in the operating state indicated by the estimated data with the visual recognition of the human operator.

Although the embodiments have been described above, the present invention is not necessarily limited to the above-described embodiments, and various modifications can be made without departing from the gist thereof. For example, the present invention can be applied not only to the plant 10 including the cement kiln 12 but also to various plants.

The plant management system 1 may be configured so that the plant management server 200 executes a learning data generation process including the map data generation process described above. The model building unit 203 of the plant management server 200 builds a plurality of types of estimation models according to external factors including at least one of the state of the raw material input to the plant 10 and the raw material and the operating environment of the plant 10. The estimation unit 106 may be configured to use a plurality of types of estimation models properly according to external factors.

1 ... Plant management system, 10 ... Plant, 12 ... Cement kiln, 51, 51A ... Map data, 51t ... Time axis, 51p ... Item axis, 51c ... Cell, 52 ... Teacher data, 53 ... Estimated data, 100 ... Plant management Device, 101 ... data acquisition unit, 103 ... map data generation unit, 105 ... teacher data generation unit, 106 ... estimation unit, 107 ... display unit, 200 ... plant management server, 203 ... model construction unit, 204 ... learning data selection Unit, 210 ... Learning data management unit, t1 to t10 ... Time interval, a to f ... Status items.

Claims (9)

For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant .
The map data generation unit is a plant management system that generates the map data by applying a fuzzy set membership function to at least a part of the map data . The plant management system according to claim 1 , wherein the map data generation unit is configured to apply different membership functions to at least two state items. The map data generation unit changes the membership function applied to the data of the map data according to an external factor including at least one of the state of the raw material input to the plant and the operating environment of the plant. The plant management system according to claim 1 or 2 , which is configured as such. For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant.
The learning data management unit corrects the map data by shifting the time between the two state items so as to reduce the deviation of the fluctuation time of the data between at least two the state items . Plant management system. For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant.
The item axis includes at least a first state item, a second state item, and a third state item that has a stronger correlation with the first state item than the second state item.
The learning data management unit is a plant management system that corrects the map data so that the third state item is located between the first state item and the second state item. For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant.
The item axis is a plant management system that includes a state item indicating the derivative of any of the state items. For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant.
The map data generation unit is a plant management system that generates the map data in which the data format of the cell is a data format for pixels. For each cell specified by the combination of the time interval and the state item in the two-dimensional space of the time axis including a plurality of time intervals arranged in a time series and the item axis including a plurality of state items related to the operation state of the plant. A map data generator that generates map data including data,
A teacher data generation unit that generates teacher data indicating the state of the plant corresponding to the map data,
A learning data management unit that stores learning data that combines the map data and the teacher data corresponding to the map data.
A model building unit that builds a model for estimating the state of the plant by machine learning based on the learning data accumulated in the learning data management unit.
It is provided with an estimation unit that inputs the map data into the estimation model and derives estimation data indicating the state of the plant.
The plant contains a cement kiln and
The item axis includes a state item relating to temperature in the cement kiln and a state item relating to the concentration of gas in the cement kiln.
The teacher data is a plant management system that includes an assessment of the amount of free lime in the cement kiln. When the deviation between the estimated data derived by inputting the map data of the learning data into the estimation model and the teacher data of the learning data exceeds a predetermined range, the learning data It also has a learning data selection unit that excludes
The model building unit is configured to reconstruct the estimation model by machine learning based on the remaining learning data when any of the learning data is excluded by the learning data selection unit. The plant management system according to any one of claims 1 to 8 .

Images