JP6845703B2 - Plant status information presentation system and plant status information presentation method - Google Patents

Description

The present invention relates to a plant status information presenting system and a plant status information presenting method for presenting information on the operating status of a plant based on values measured by a plurality of sensors that measure the state of the plant.

Patent Documents 1 and 2 are documents relating to a system for estimating the state of a plant. In the system disclosed in Patent Document 1, in the learning phase, a normal model in which the normal state is modeled using the sensor signal output from the plant is created, and in the evaluation phase, a feature vector is created from the sensor signal output from the plant. Is generated, and the obtained feature vector is compared with the normal model to determine the abnormality. In the system disclosed in Patent Document 2, the locus of the data space is divided into a plurality of clusters based on the temporal change of the data of the sensor provided in the plant, and the divided cluster group is subjected to a regression analysis method or the like. Model by the subspace method of, and calculate outliers as abnormal candidates.

JP 2015-172945 Japanese Unexamined Patent Publication No. 2012-230703

However, in a plant where it is difficult to estimate the internal environment, such as a blast furnace where measurement inside the furnace is difficult, it is difficult to create a model with high accuracy. In addition, when modeling a large-scale plant with a large number of data items by the subspace method, it is common to reduce the dimensions by principal component analysis, etc. in order to shorten the calculation time and improve the accuracy. If the dimension does not match the dimension of the measured value, the cause cannot be identified even if an abnormality is caught.

The present invention has been made in view of such circumstances, and its main purpose is to operate with high accuracy even in a plant in which it is difficult to estimate the internal environment and a large-scale plant in which many data items are detected by sensors. The purpose of the present invention is to provide a plant status information presentation system and a plant status information presentation method capable of estimating the status.

In order to solve the above-mentioned problems, the plant status information presentation system according to one aspect of the present invention continuously measures the state of the plant from each time series data of the measured values by a plurality of different types of sensors. It was determined by the extraction means for extracting the time variation of the value, the determination means for separately determining the pattern showing the characteristics of the time variation for each of the plurality of the time variations extracted by the extraction means, and the determination means. An output means for outputting operation status information regarding the operation status of the plant based on the plurality of patterns, a pattern database for storing a plurality of patterns of time fluctuations of measured values by the plurality of sensors for each type of the sensor, and a pattern database. Each of the patterns stored in the pattern database is provided with a classification means for classifying each of the patterns into a pattern in a steady time and a pattern in a non-steady time based on the number of occurrences thereof, and the determination means is said by the classification means. From the pattern classified into the pattern in the non-stationary time, the pattern similar to the time variation extracted by the extraction means is determined .

In this embodiment, the plant status information presenting system further includes an estimation means for estimating the operation status of the plant based on the plurality of patterns determined by the determination means, and the output means is the operation status information. As a result, the estimation result by the estimation means may be output.

Further, in the above aspect, the plant status information presentation system further includes a past record database that stores the time variation pattern of each measured value of the plant in the past and the operation status information in association with each other, and the estimation means The operation status information corresponding to the pattern similar to the pattern determined by the determination means is acquired from the past performance database, and the output means is the said output means acquired by the estimation means. It may be configured to output operation status information.

Further, in the above aspect, the extraction means is configured to extract each of the time fluctuations in a plurality of time widths different from each other, and the determination means is the plurality of time widths extracted by the extraction means. It may be configured to determine the pattern for each of the time variations in.

Further, in the above aspect, the plant status information presentation system further includes a pattern generation means for generating the pattern based on the time variation of the measured value extracted from the time series data of the measured value, and the pattern database. May be configured to store the pattern generated by the pattern generating means.

Further, in the above aspect, the plant status information presentation system further includes a clustering means for classifying each of the plurality of the time fluctuations into a plurality of clusters based on the waveform of the time fluctuations, and the pattern generation means is the clustering means. For each cluster in which the time variation is classified by means, the pattern indicating the characteristics of the time variation belonging to the cluster may be generated.

Further, the plant status information presentation method of another aspect of the present invention includes a step of extracting the time variation of the measured value from each time series data of the measured value by a plurality of sensors that continuously measure the state of the plant. For each of the extracted time fluctuations, a step of separately determining a pattern showing the characteristics of the time fluctuation, and a step of outputting operation status information regarding the operation status of the plant based on the plurality of determined patterns. , The pattern of the time variation of the measured value by the plurality of sensors is stored for each type of the sensor. Each of the patterns stored in the pattern database is stored in the pattern in the steady state and in the non-steady time based on the number of appearances. possess a step of classifying into a pattern, in said determining step, from the classified the pattern, to determine the pattern similar to the extracted the time variation.

According to the plant status information presentation system and the plant status information presentation method according to the present invention, even in a large-scale plant where it is difficult to estimate the internal environment and there are many data items detected by sensors, the operation status is highly accurate. Can be estimated.

The schematic diagram which shows the structure of the plant status information presentation system which concerns on Embodiment 1. The block diagram which shows the internal structure of the plant status information presentation system which concerns on Embodiment 1. The flowchart which shows the operation procedure of the plant status information presentation system which concerns on Embodiment 1. A flowchart showing the procedure of the learning process. A graph showing an example of multidimensional time series data. The figure which schematically explains the extraction of the time variation of the measured value. The figure which schematically explains the generation of the time variation pattern of the measured value. The schematic diagram which shows the structure of the past performance database. A flowchart showing the procedure of the estimation process. The figure schematically explaining the generation of state information. A flowchart showing the procedure of the similar state identification process. The flowchart which shows the operation procedure of the plant status information presentation system which concerns on Embodiment 2. A flowchart showing a procedure of information presentation processing.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. It should be noted that the embodiments shown below exemplify methods and systems for embodying the technical idea of the present invention, and the technical idea of the present invention is not limited to the following. .. The technical idea of the present invention can be modified in various ways within the technical scope described in the claims.

(Embodiment 1)
In the present embodiment, the plant status information presentation system extracts the time variation of the measured value from each time series data of the measured value by a plurality of sensors, determines the pattern of each extracted time variation, and determines the current time. The past pattern that most closely resembles the fluctuation pattern is specified, and the operation status information of the plant at the past time when the specified pattern is obtained is output.

<Configuration of plant status information presentation system>
FIG. 1 is a schematic diagram showing a configuration of a plant status information presentation system according to the present embodiment. The plant status information presentation system 100 estimates the operating status of the blast furnace 10. The blast furnace 10 continuously executes a heat treatment process of burning coke charged into the furnace together with iron ore in order to continuously produce pig iron. The blast furnace 10 is provided with a temperature sensor, a pressure sensor, and other sensors for measuring physical quantities. In the present embodiment, a plant status information presentation system that estimates the operating status of the blast furnace 10 using the measured values output by the two temperature sensors 11a and 11b and the two pressure sensors 12a and 12b installed in the blast furnace 10. 100 will be described.

The plant status information presentation system 100 is composed of a computer. The plant status information presentation system 100 is connected to temperature sensors 11a and 11b and pressure sensors 12a and 12b, and is configured to receive measured values from these.

FIG. 2 is a block diagram showing an internal configuration of the plant status information presentation system 100. The plant status information presentation system 100 is realized by the computer 200. As shown in FIG. 2, the computer 200 includes a main body 300, an input unit 400, and a display unit 500. The main body 300 includes a CPU 301, ROM 302, RAM 303, a read device 304, a hard disk 305, an input / output interface 306, and an image output interface 307, and includes a CPU 301, ROM 302, RAM 303, a read device 304, a hard disk 305, and an input / output interface 306. And the image output interface 307 is connected by a bus.

The CPU 301 executes the computer program loaded in the RAM 303. Then, when the CPU 301 executes the plant status information presentation program 310, which is a computer program for presenting the plant status information, the computer 200 functions as the plant status information presentation system 100.

The ROM 302 records a computer program executed by the CPU 301, data used for the program, and the like. The RAM 303 is used to read the plant status information presentation program 310 recorded on the hard disk 305. Further, the RAM 303 is used as a work area of the CPU 301 when the CPU 301 executes a computer program.

The hard disk 305 is installed with various computer programs such as an operating system and an application program for the CPU 301 to execute, and data used for executing the computer programs. The plant status information presentation program 310 is also installed on the hard disk 305.

Further, the hard disk 305 is provided with a past record database (past record DB) 320, a pattern database (pattern DB) 330, and a measured value database (measured value DB) 340. The past record DB 320 stores past record data regarding the operating status of the blast furnace 10. The pattern DB 330 stores pattern data showing a pattern of time fluctuations of values measured by the temperature sensors 11a and 11b and the pressure sensors 12a and 12b. Further, the measured value DB 340 stores multidimensional time series data of the measured values measured by the temperature sensors 11a and 11b and the pressure sensors 12a and 12b in the past.

An input unit 400 including a keyboard and a mouse is connected to the input / output interface 306. Further, the temperature sensors 11a and 11b and the pressure sensors 12a and 12b described above are connected to the input / output interface 306, and are configured to receive output data from these sensors.

The image output interface 307 is connected to a display unit 500 composed of an LCD, a CRT, or the like, and outputs a video signal corresponding to the image data given by the CPU 301 to the display unit 500. The display unit 500 displays an image (screen) according to the input video signal.

<Operation of plant status information presentation system>
Next, the operation of the plant status information presentation system 100 will be described. FIG. 3 is a flowchart showing an operation procedure of the plant status information presentation system 100 according to the present embodiment. The plant status information presentation system 100 executes the learning process S100 and the estimation process S200 in order to estimate the operating status of the blast furnace 10. The learning process S100 is a process for generating pattern data showing a characteristic time fluctuation pattern of the measured value from the past results of the operation data, and the estimation process S200 is a process for generating the pattern data of the blast furnace 10 using the pattern data. This is a process for estimating the situation. The learning process S100 and the estimation process S200 may be continuously executed, or may be independently and discontinuously executed. However, since the estimation process S200 requires pattern data, it is necessary to execute the learning process S100 at least once before the estimation process S200.

The learning process S100 will be described in detail. The temperature sensors 11a and 11b and the pressure sensors 12a and 12b of the blast furnace 10 continuously output multidimensional time series data of measured values. The plant status information presentation system 100 receives the multidimensional time series data and stores it in the measured value DB 340. In the learning process S100, the multidimensional time series data stored in the measured value DB 340 is used.

FIG. 4 is a flowchart showing the procedure of the learning process S100. In the learning process S100, the CPU 301 first acquires each multidimensional time series data in the learning period, which is a past period, from the measured value DB 340 (step S101). FIG. 5 is a graph showing an example of each multidimensional time series data. The period in which the multidimensional time series data is shown is a common learning period. Since the multidimensional time series data may include a missing portion, the CPU 301 repairs the missing portion by replacing the missing portion with data before and after the missing portion (step S102).

Next, the CPU 301 extracts the time variation of the measured value from the time series data (step S103). FIG. 6 is a diagram schematically illustrating extraction of time fluctuations of measured values. The time variation is extracted from the time series data in each of the plurality of time widths. FIG. 6 shows an example of extracting time fluctuations in two time widths of t1 and t2. Further, the time variation is extracted from the time series data with the same time widths t1 and t2 by changing the extraction time. Specifically, as shown in FIG. 6, the CPU 301 extracts the time variation with the time widths t1 and t2 having the time t as the end point, and extracts the time variation with the time widths t1 and t2 having the time t + Δ as the end point. Extract. In this way, a plurality of time fluctuations are extracted from all the multidimensional time series data by changing the time and time width. That is, the time fluctuations of the time widths t1 and t2 are extracted from the multidimensional time series data of the temperature sensor 11a at a plurality of times, and similarly, a plurality of time fluctuations of the temperature sensor 11b and the pressure sensors 12a and 12b are also obtained. Time fluctuations of time widths t1 and t2 are extracted at the time.

Next, the CPU 301 unifies the number of dimensions of the extracted time fluctuation data by taking an average in the vicinity (step S104).

Further, the CPU 301 collects the extracted time variation separately for temperature and pressure, and creates a temperature data set and a pressure data set (step S105). The CPU 301 clusters each data set based on the time-varying waveform and classifies it into n clusters (step S106). Further, the CPU 301 generates a time variation pattern belonging to each cluster for each cluster (step S107). FIG. 7 is a diagram schematically illustrating pattern generation. In the process of step S107, the center-of-gravity vector of the total time variation belonging to the cluster of interest is calculated, and this is defined as a pattern representing the characteristics of the time variation. The value of n is determined in consideration of what kind of time fluctuation may occur based on the experience with respect to the measured value. In addition, in order to detect an abnormal operating condition, it is necessary to prepare a pattern of time fluctuations that appears when the furnace condition is abnormal. In addition to the blast furnace 10, many plants have stable operating conditions during most of the operating conditions. Therefore, in order to secure the time fluctuation pattern that appears when the furnace conditions are abnormal, the period in which the operating conditions are unstable is learned in advance. It is preferable to acquire multidimensional time series data as a period.

Here, the pattern was generated by dividing the time fluctuation into temperature and pressure, because the characteristics of the time fluctuation are different between the two, such as the pressure fluctuating sharply compared to the temperature. is there. Even if the physical quantity is the same, if the characteristics of the time fluctuation differ greatly depending on the measurement position, the pattern may be generated separately for the measurement position. On the contrary, if the characteristics of time fluctuations are similar even if the physical quantities are different, a pattern may be generated by combining them.

The CPU 301 registers the generated pattern in the pattern DB 330 (step S108). After that, the CPU 301 ends the learning process S100.

Here, the data stored in the past record DB 320 will be described. FIG. 8 is a schematic diagram showing the configuration of the past record DB 320. The past record DB 320 stores past record data including operation status information and status information of the blast furnace 10 at a past time point. The operation status information includes the type of product produced in the blast furnace 10 at the past time, the raw material, the production amount, the set value and other operating conditions, the type of abnormality if an abnormality has occurred, and the like. Further, the state information includes a time variation pattern for each measured value of the temperature sensors 11a and 11b and the pressure sensors 12a and 12b at the past time points. The pattern of each measured value is for each of the above time widths t1 and t2. Such state information is the same as the state information generated in step S205 described later. The past performance data is a record including the operation status information and the status information, and the operation status information and the status information in one record are related to each other. The past record database may be composed of a plurality of tables. For example, a first table containing a field of a past time and a field of operation status information at that time, and a second table including a field of a past time and a field of status information at that time are displayed in the past. DB 320 may be included. In this case, the operation status information and the status information are associated with each other via the data of the past time.

FIG. 9 is a flowchart showing the procedure of the estimation process S200. In the estimation process S200, the CPU 301 acquires the current time, that is, each multidimensional time series data in the latest predetermined period from the measured value DB 340 (step S201). Next, the CPU 301 repairs the missing portion included in the multidimensional time series data by replacing it with the data before and after it (step S202).

The CPU 301 extracts the time variation of the measured value from the time series data (step S203). In this process, the time variation is extracted from each time series data by using the time widths t1 and t2 obtained by cutting out the time variation in step S103. However, in step S203, the time widths t1 and t2 for extracting the time variation are only the period including the current time (latest time) as the end point. Further, the CPU 301 unifies the number of dimensions of the extracted time fluctuation data by taking an average in the vicinity (step S204).

Next, the CPU 301 compares the time variation with each pattern stored in the pattern DB 330 for each time variation at the current time, determines the pattern most similar to the time variation at the current time, and determines each time variation. A set of patterns similar to is generated as state information of the blast furnace 10 at the current time (step S205). FIG. 10 is a diagram schematically illustrating the generation of state information. In this process, the similarity of two-dimensional vectors such as the Euclidean distance is calculated, and the pattern having the highest similarity is determined. Similar patterns are determined for each measured value (temperature, pressure) and each time width t1 and t2, and a set of these patterns is defined as state information.

See FIG. 9 again. The CPU 301 executes a similar state identification process for specifying past state information similar to the state information at the current time (step S206). In this similar state identification process, the past state information most similar to the state information at the current time is specified by comparing the state information at the current time with each state information stored in the past record DB 320. In this process, the similarity between the pattern included in the state information at the current time and the pattern included in the past state information is calculated for each measured value (temperature, pressure) and time widths t1 and t2.

Hereinafter, the similar state identification process will be described in detail. FIG. 11 is a flowchart showing the procedure of the similar state identification process. First, the CPU 301 classifies the patterns stored in the pattern DB 330 into a pattern in a steady state and a pattern in a non-steady state (step S2601). Here, the steady state means a state in which the blast furnace 10 is operating normally and stably (that is, a state in which the furnace condition is good), and an unsteady state means a state in which the operation of the blast furnace 10 is unstable (that is, a state in which the operation of the blast furnace 10 is unstable). The state of poor furnace condition). The non-steady state is an abnormal time in many cases, but also includes a case where the furnace condition is poor even in a normal state. As a method of classifying the patterns, in the clustering process of step S106, the pattern corresponding to the clusters in which the number of classified time fluctuations is a certain number or more is set as the steady state pattern, and the number of the classified time fluctuations is less than a certain number. The pattern corresponding to the cluster can be the non-stationary pattern. In addition, it is considered that the classification results of the patterns in the steady state and the non-steady state change depending on the temperature, pressure, their measurement positions, and the time width. For this reason, it is preferable to classify all patterns by temperature, pressure, measurement position, and time width, instead of classifying them all at once. That is, for one time width of the measured value output from one sensor, the number of classifications of time fluctuation is obtained for each cluster, and if the number of classifications is equal to or more than a certain value, it is set as a steady-state pattern, and if it is less than a certain value. The pattern at the non-stationary time is used.

Next, the CPU 301 compares the state information at the current time with the state information stored in the past record DB 320 for each pattern having the same measured value and time width, and calculates the degree of similarity between them (step S2602). In this process, a similarity measure of a two-dimensional vector such as the Euclidean distance is calculated. Further, the CPU 301 extracts only those for the non-stationary pattern from all the calculated similarities, and calculates the average as the inter-state similarity (step S2603). By excluding the similarity for the pattern in the steady state in the calculation of the similarity between states, a characteristic time fluctuation pattern in the non-steady state appears in the time series data of each sensor 11a, 11b, 12a, 12b. It is possible to accurately detect what is being done.

The CPU 301 identifies the state information of the past record having the highest calculated similarity between states (step S2604), and ends the similarity state identification process.

See FIG. 9 again. The CPU 301 reads the operation status information corresponding to the status information of the past record specified in the similar state identification process from the past record DB 320 (step S207), and displays this on the display unit 500 (step S208). This completes the estimation process S200. It is presumed that the past actual operation status information displayed on the display unit 500 is similar to the current operation status. Therefore, the operator can grasp the current operating status by checking the operating status information displayed on the display unit 500, and can appropriately deal with this.

By configuring as described above, if the time width is set appropriately according to the plant and the time fluctuation patterns that can be taken by each measured value are sufficiently prepared, a set of time fluctuation patterns of the measured values is prepared. The state information can hold sufficient information for determining the operation status of the plant, and the operation status of the plant can be accurately estimated by using the status information. Therefore, for example, when the operating condition changes from a steady state to a non-steady state, the cause and grounds for the change become clear.

Further, by extracting the time variation with the short time width t1 and the long time width t2, it is possible to capture each of the short-term variation and the long-term tendency of the measured value.

In addition, since the past operating status corresponding to the status information similar to the status information at the current time is estimated as the operating status at the current time using the status information, the past operating status is stored in the past record DB320. If it is accumulated, it can be used to estimate the current operating status. In a large-scale plant, the state of the plant is complicated and the number is enormous, so it is not easy to create a database that systematizes the operation status for each state of the plant. With the above configuration, it is possible to estimate the operating status without using such a database.

(Embodiment 2)
In the present embodiment, the plant status information presentation system extracts the time variation of the measured value from each time series data of the measured value by a plurality of sensors, and generates the state information including the extracted pattern of each time variation. , The sensor in the unsteady state is identified from these patterns, and the information of the identified sensor is output as the operation status information.

<Configuration of plant status information presentation system>
Since the configuration of the plant status information presentation system according to the present embodiment is the same as the configuration of the plant status information presentation system 100 according to the first embodiment, the same components are designated by the same reference numerals and the description thereof is omitted. To do. However, in the present embodiment, since the past record DB 320 is not used, the plant status information presentation system 100 may be configured not to include the past record DB 320.

<Operation of plant status information presentation system>
FIG. 12 is a flowchart showing an operation procedure of the plant status information presentation system 100 according to the present embodiment. The plant status information presentation system 100 executes the learning process S100 and the information presentation process S300. Since the learning process S100 is the same as that described in the first embodiment, the description thereof will be omitted.

FIG. 13 is a flowchart showing the procedure of the information presentation process S300. Since the processes of steps S301 to S305 in the information presentation process S300 are the same as the processes of steps S201 to S205 in the estimation process S200 described in the first embodiment, the description thereof will be omitted.

When the state information is generated in step S305, the CPU 301 classifies the pattern stored in the pattern DB 330 into a pattern in the steady state and a pattern in the non-steady state (step S306). Since the process of step S306 is the same as the process of step S2601 described in the first embodiment, the description thereof will be omitted.

Next, the CPU 301 uses the classification result in step S306 to specify the pattern in the non-steady state among the time variation patterns included in the state information at the current time generated in step S305 (step S307). Next, the CPU 301 identifies the sensor corresponding to the pattern in the non-steady state among the sensors 11a, 11b and 12a, 12b as the sensor that has detected the non-steady state (step S308). Each of the temperature sensors 11a and 11b and the pressure sensors 12a and 12b is assigned a sensor ID for identifying the sensor. Further, the temperature sensors 11a and 11b are installed at different positions in the blast furnace 10, and the pressure sensors 12a and 12b are also installed at different positions in the blast furnace 10. The CPU 301 causes the display unit 500 to display information (sensor ID, sensor type, sensor mounting position, etc.) of the sensor that has detected the unsteady state (step S309). This completes the information presentation process S300.

As described above, by presenting the information of the sensors in the unsteady state, the operator can judge the approximate operating status of the blast furnace 10 from the number of sensors in the unsteady state, and the blast furnace from the position of the sensors in the unsteady state. The cause of the unsteady state that occurred in 10 can be estimated.

(Other embodiments)
In the first embodiment described above, as the operation status information, the type of the produced product, the raw material, the production amount, the set value and other operating conditions, and the type of the abnormality when an abnormality has occurred are output. Further, in the second embodiment, the information of the sensor in the unsteady state is output as the operation status information, but the present invention is not limited to these. As the operation status information, the presence / absence of an abnormality in the plant, the location of the abnormality, the cause of the abnormality, the countermeasures thereof, or a part of the information may be output. In this case, the presence or absence of an abnormality in the plant, the location of the abnormality, the cause of the abnormality, and how to deal with it, or some of these information are stored in the database in association with the status information, and from the plant status information at the current time. , The presence or absence of the corresponding abnormality, the location of the abnormality, the cause of the abnormality, and the coping method thereof, or some information thereof can be specified.

Further, in the above-described first and second embodiments, each time the estimation process S200 or the information presentation process S300 is performed, all the patterns stored in the pattern DB 330 are classified into a steady pattern and a non-stationary pattern. However, it is not limited to this. All the patterns stored in the pattern DB 330 in advance are classified into a steady state pattern and a non-stationary time pattern, and for each pattern stored in the pattern DB 330, a steady state or a non-steady state is identified. Information can also be associated. In this case, in the estimation process S200 or the information presentation process S300, it is possible to omit the process of classifying the pattern in the stationary state and the pattern in the non-stationary state.

Further, in the above-described first and second embodiments, a configuration is described in which a plurality of time-varying centroid vectors belonging to the same cluster are calculated and determined as a time-varying pattern, but the present invention is not limited to this. Absent. For example, multiple pattern templates are prepared, multiple templates belonging to the same cluster that are most similar to the time variation are identified, and this is determined as the time variation pattern. A characteristic pattern may be determined.

Further, in the above-described embodiment, the configuration in which all the processes of the plant status information presentation program 310 are executed by a single computer 200 has been described, but the present invention is not limited to this, and the plant status information is not limited thereto. It is also possible to make a distributed system in which the same processing as the presentation program 310 is distributed and executed by a plurality of devices (computers). For example, in the first embodiment, the plant status information presentation system includes a learning device and an estimation device, and the learning device can execute the learning process S100, and the estimation device can execute the estimation process S200. In the second embodiment, the plant status information presentation system includes a learning device and an information presenting device, the learning device executes the learning process S100, and the information presenting device executes the information presentation process S300.

The plant status information presentation system and the plant status information presentation method of the present invention present the plant status information presentation system and the plant status information presentation that present information on the operation status of the plant based on the measured values by a plurality of sensors that measure the plant status. It is useful as a method.

10 Blast furnace 11a, 11b Temperature sensor 12a, 12b Pressure sensor 100 Plant status information presentation system 200 Computer 300 Main unit 301 CPU
303 RAM
305 Hard disk 310 Plant status information presentation program 320 Past performance database 330 Pattern database 340 Measured value database 500 Display unit

Claims (7)

An extraction means for extracting the time variation of the measured value from each time series data of the measured value by a plurality of different types of sensors that continuously measure the state of the plant.
For each of the plurality of time variations extracted by the extraction means, a determination means for separately determining a pattern showing the characteristics of the time fluctuations, and
An output means for outputting operation status information regarding the operation status of the plant based on the plurality of patterns determined by the determination means, and an output means .
A pattern database that stores a plurality of patterns of time fluctuations of values measured by the plurality of sensors for each type of the sensor, and
A classification means for classifying each of the patterns stored in the pattern database into a pattern in a stationary state and a pattern in a non-stationary time based on the number of occurrences thereof.
Equipped with a,
The determination means is configured to determine the pattern similar to the time variation extracted by the extraction means from the pattern classified into the pattern in the non-stationary time by the classification means.
Plant status information presentation system. Further provided with an estimation means for estimating the operation status of the plant based on the plurality of patterns determined by the determination means.
The output means is configured to output the estimation result by the estimation means as the operation status information.
The plant status information presentation system according to claim 1. It is further equipped with a past record database that stores the pattern of time fluctuations of each measured value of the plant in the past in association with the operation status information.
The estimation means is configured to acquire the operation status information corresponding to the pattern similar to the pattern determined by the determination means from the past performance database.
The output means is configured to output the operation status information acquired by the estimation means.
The plant status information presentation system according to claim 2. The extraction means is configured to extract each of the time fluctuations in a plurality of time widths different from each other.
The determination means is configured to determine the pattern for each of the time variations in the plurality of time widths extracted by the extraction means.
The plant status information presentation system according to any one of claims 1 to 3. Further provided with a pattern generation means for generating the pattern based on the time variation of the measured value extracted from the time series data of the measured value.
The pattern database is configured to store the pattern generated by the pattern generating means.
The plant status information presentation system according to any one of claims 1 to 4. Further provided with a clustering means for classifying each of the plurality of time fluctuations into a plurality of clusters based on the waveform of the time fluctuations.
The pattern generation means is configured to generate the pattern showing the characteristics of the time variation belonging to the cluster for each cluster in which the time variation is classified by the clustering means.
The plant status information presentation system according to claim 5. A step of extracting the time variation of the measured value from each time series data of the measured value by a plurality of sensors that continuously measure the state of the plant, and
For each of the extracted time fluctuations, a step of separately determining a pattern showing the characteristics of the time fluctuations, and
Based on a plurality of said pattern determined, and outputting the operational status information about the operating status of the plant,
Multiple patterns of time fluctuations of values measured by the plurality of sensors are stored for each type of the sensor. Each of the patterns stored in the pattern database is stored as a pattern in the steady state and a pattern in the non-steady state based on the number of appearances. possess a step of classification to the door,
A method for presenting plant status information for determining a pattern similar to the time variation extracted from the classified patterns in the determination step.

Images