JP7263020B2 - Treated water quality estimation device, treated water quality estimation method and program - Google Patents

Description

TECHNICAL FIELD Embodiments of the present invention relate to a treated water quality estimation device, a treated water quality estimation method, and a program.

In a water treatment plant, there is a demand for estimating time-series changes in treated water quality based on operation values and control values in order to operate the plant efficiently and stably. Specific technologies for estimating treated water quality include a technology for predicting treated water quality based on physical and chemical models (Patent Document 1, Patent Document 2) and a technology for predicting treated water quality using a machine learning technique (Patent Reference 3) and others have been proposed.

JP 2017-140595 A JP 2017-121595 A JP 2016-195974 A

Techniques for estimating treated water quality based on physical/chemical models and techniques for estimating treated water quality using machine learning may be affected by fluctuations in pumping station characteristics, etc. There are strengths and weaknesses depending on the given situation, such as when the prediction accuracy drops when data outside the data range is input. Therefore, operators of water treatment plants must select and apply estimation techniques at the appropriate timing according to the situation in an environment in which one can be selected from multiple estimation techniques, and experience and proficiency are required. It was a complicated task.

The present embodiment has been made in view of the actual situation as described above, and its purpose is to appropriately and automatically switch or combine a plurality of estimation models according to a given situation to obtain a highly accurate estimation result. To provide a treated water quality estimation device, a treated water quality estimation method, and a program capable of

The treated water quality estimation apparatus according to the present embodiment includes an input information acquisition unit that acquires statistical information obtained by measuring the quality of water that needs to be treated in a plant to be controlled and operation value information as input information, and the input information acquisition unit. An input information storage unit that stores the input information acquired in , an estimation model storage unit that stores a plurality of estimation models that calculate an estimated value of treated water quality using input information with different algorithms, and the input information storage unit An estimation calculation unit for calculating an estimated value of treated water quality using the stored input information and a plurality of estimation models stored in the estimation model storage unit; and storing the estimated value of treated water quality calculated by the estimation calculation unit. an estimation result storage unit that stores a plurality of aggregate calculation models for aggregating a plurality of treated water quality estimation values; an aggregate operation unit that aggregates using an aggregate operation model stored in a model storage unit; and an aggregate operation command that commands the aggregate operation model used by the aggregate operation unit based on the input information stored in the input information storage unit. and

The block diagram which shows the system configuration|structure of the whole plant to be controlled which concerns on 1st Embodiment. The block diagram which shows the specific example of a functional structure of the treated-water-quality estimation apparatus which concerns on 1st Embodiment. 4 is a flow chart showing a process for determining an operation value using the treated water quality estimating device according to the first embodiment; 4 is a flow chart showing processing contents for commanding an aggregate computation model according to the first embodiment; The block diagram which shows the specific example of a functional structure of the treated water quality estimation apparatus which concerns on 2nd Embodiment. 10 is a flow chart showing the details of processing for updating an aggregate computation model according to the second embodiment; The block diagram which shows the specific example of a functional structure of the treated water quality estimation apparatus which concerns on 3rd Embodiment. 10 is a flowchart showing the details of parallel processing of a plurality of estimation operations and processing of aggregation operations according to the third embodiment; The block diagram which shows the specific example of a functional structure of the treated water quality estimation apparatus which concerns on 4th Embodiment. The figure which illustrates the graph display screen which concerns on 4th Embodiment. The figure which illustrates the graph display screen which concerns on 4th Embodiment.

Embodiments will be described below with reference to the drawings.
[First embodiment]
FIG. 1 is a block diagram showing the overall system configuration of a controlled plant 100 according to the first embodiment. The controlled plant 100 is a system that includes a treated water quality estimation device 1 , a monitoring control system 3 , and one or more measuring devices 4 , treats influent water 6 in a water and sewage plant 5 , and supplies treated water 7 .

When applied as a water supply facility, inflow water 6 from a river is supplied as treated water 7 to water distribution plants and consumers. When applied as a sewerage facility, inflow water 6 from a sewage pipe network is supplied to a river as treated water 7 .

The treated water quality estimating device 1 receives statistical information from the monitoring and control system 3 and is obtained by measurement at the water and sewage plant 5, and an operation value input by the user 2, which is composed of an operation item and an operation value. Based on the information, estimation result information indicating the water quality such as the concentration of various substances after treatment is calculated and presented to the user 2 .

The operation items are labels of operation values, such as "aeration air volume", "return rate", and "excess sludge removal rate". The operation numerical value is, for example, a numerical value such as "400 [m ³ /h]" for the operation item "aeration air volume".

Further, specific estimation result information output by the treated water quality estimation apparatus 1 includes, for example, information such as "phosphate phosphorus concentration", "nitrate nitrogen concentration", and "ammonia nitrogen concentration".

The statistical information that the monitoring control system 3 transmits to the water quality estimation device 1 includes operation value commands and measurement information for a certain period of time up to the present. The measurement information includes an operation value command (also referred to as a “management value” as an operation value managed at the time of measurement), and, for example, “phosphate phosphorus concentration”, “nitrate nitrogen concentration”, “ammonia and treated water quality values such as "nitrogen concentration".

The user 2 inputs the operation value information necessary for the estimation calculation of the treated water quality to the treated water quality estimation apparatus 1, and obtains the estimation result information based on the input operation value information. The user 2 determines an operation value according to estimation result information obtained by inputting various pieces of operation value information, and inputs the determined result to the monitoring control system 3 as an operation value command. The process in which the user 2 inputs the operation value information to the treated water quality estimation device 1 and acquires the estimation result information from the treated water quality estimation device 1 can be repeatedly executed until the user 2 determines the operation value.

The monitoring control system 3 has a function of monitoring and controlling the water and sewage plant 5, and transmits control information for controlling each part in the plant 5 to the water and sewage plant 5 according to the operation value command from the user 2. do. In addition, the monitoring control system 3 acquires the measurement information of the water and sewage plant 5 measured by the measurement device 4, and transmits the operation value command and the measurement information for a certain time up to the present to the treated water quality estimation device 1 as statistical information. .

The measuring device 4 measures the quality value of treated water in the water and sewage plant 5, collects the time at the time of measurement, the operation value command (management value), and the measured quality value of the treated water, and outputs it to the monitoring control system 3 as measurement information.

FIG. 2 is a block diagram showing a specific example of the functional configuration of the treated water quality estimation device 1. As shown in FIG. The treated water quality estimation device 1 includes a CPU (Central Processing Unit), a memory, an auxiliary storage device, etc., which are connected via a bus, and executes a program. By executing a program, the treated water quality estimation apparatus 1 has an input information acquisition unit 11, an input information storage unit 12, an estimation calculation unit 13, an aggregation operation commanding unit 14, an estimation model storage unit 15, an estimation result storage unit 16, an aggregation operation unit 17 and an aggregate computation model storage unit 18 .

All or part of each function of the treated water quality estimation device 1 is realized using hardware such as ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), and FPGA (Field Programmable Gate Array). Also good.

The program may be recorded on a computer-readable recording medium. The computer-readable recording medium includes, for example, a magneto-optical disk, a portable medium such as a ROM (Read Only Memory), and a storage device such as a hard disk device built in or externally attached to the computer. Furthermore, the program may be received via a wired or wireless communication line and stored in a medium.

The input information storage unit 12, the estimation model storage unit 15, the estimation result storage unit 16, and the aggregate computation model storage unit 18 are configured using storage devices such as hard disk devices and semiconductor storage devices.

The input information acquisition unit 11 includes a communication interface, and communicates with the supervisory control system 3 via the communication interface. The input information acquisition unit 11 transmits the statistical information obtained from the monitoring control system 3 and the operation value information input by the user 2 as input information to the input information storage unit 12 for storage. The input information acquisition unit 11 repeatedly acquires the statistical information at a predetermined periodic timing, thereby accumulating the acquired statistical information in the input information storage unit 12 as part of the input information.

The input information storage unit 12 stores statistical information in the input information acquired via the input information acquisition unit 11 as time-series data. The time-series statistical information and the operation value information stored in the input information storage unit 12 are read by the estimation calculation unit 13 , the aggregation commanding unit 14 and the aggregation calculation unit 17 .

The estimation calculation unit 13 calculates, for example, time Calculate future treated water quality estimated values such as "phosphate phosphorus concentration", "nitrate nitrogen concentration", "ammonia nitrogen concentration" in units, and store the calculated treated water quality estimated values in the estimation result storage unit 16 Send and store.

The estimation model storage unit 15 stores a plurality of estimation models with mutually different algorithms. Each of the plurality of estimation models stored in the estimation model storage unit 15 inputs one or more pieces of statistical information and outputs one or more water treatment estimated values. These multiple estimation models may store multiple models using static analysis methods based on physical and chemical models, and multiple models using machine learning. It is generated in advance by an automatic generation method such as machine learning using As for the estimation models stored in the estimation model storage unit 15, a plurality of different estimation models may be prepared for each target treated water quality, or a plurality of estimation models for estimating the same treated water quality may be provided.

The estimation result storage unit 16 stores a plurality of treated water quality estimation values obtained by the estimation calculation unit 13 . A plurality of treated water quality estimation values stored in the estimation result storage unit 16 are read out to the aggregation calculation unit 17 .

The aggregate calculation unit 17 corresponds to how much the statistical information stored in the input information storage unit 12 deviates from the steady-state value, and stores the aggregate calculation model from the plurality of treated water quality estimated values stored in the estimation result storage unit 16. Based on one of the plurality of aggregated calculation models stored in the unit 18, one aggregated highly accurate estimated value of treated water quality is calculated and presented to the user 2 as estimation result information.

The aggregate computation model storage unit 18 stores a plurality of aggregate computation models for aggregating a plurality of identical estimated values of treated water quality into one. Each of the plurality of aggregated computational models stored in the aggregated computational model storage unit 18 inputs one or more identical estimated values of treated water quality and outputs one identical estimated value of treated water quality. Aggregate calculation models include a selection model that selects any one of multiple estimation results, such as the median value, as an output value, and a simple function that calculates a combination of multiple estimation results and outputs an average value, etc. It includes a model, a black box model that calculates one output value by complexly combining multiple estimation results using machine learning using pre-learned data, etc.

The aggregate computation commanding unit 14 refers to the operation value included in the input information given from the input information storage unit 12, and selects a plurality of aggregate computation models stored in the aggregate computation model storage unit 18 to be used for the aggregate computation. An aggregate operation model is automatically and appropriately selected, and an aggregate operation unit 17 executes an operation based on the selected aggregate operation model.

The operation of this embodiment in the configuration as described above will be described.
FIG. 3 is a flow chart showing the processing steps when the user 2 who is the operator determines the operation value using the treated water quality estimation device 1 based on the highly accurate estimated values calculated from a plurality of estimation models under the same conditions. . In addition, in this embodiment, it is assumed that there are a plurality of operation values to be determined.

At the beginning of the process, the user 2 inputs and sets the quality of treated water to be evaluated and the estimated time (step S101).
In response to this, the input information acquisition unit 11 acquires the statistical information generated by the monitoring control system 3 including the measurement information from the measuring device 4 as the input information necessary for all estimation models, and associates it with the operation value information. and store it in the input information storage unit 12 (step S102).

In order to perform calculations with all the estimation models, the estimation calculation unit 13 first determines whether or not the quality of all treated water satisfies the standard by calculating with all the estimation models using one operation value (step S103).

If it is determined that the quality of all treated water does not satisfy the standard (NO in step S103), then the estimation calculation unit 13 changes the operation value being selected at that time so that the quality of all treated water meets the standard. It is determined whether or not (step S104).

If it is determined that the quality of all the treated water will meet the standard by changing the selected operation value (YES in step S104), the estimation calculation unit 13 presents to the user 2 to change the selected operation value. Then, after receiving and setting the change operation (step S105), the process proceeds to step S103 to perform determination processing for confirmation.

Further, in step S104, when it is determined that the quality of all treated water does not meet the standard even if the operation value being selected is changed (NO in step S104), the estimation calculation unit 13 changes the operation item to be selected. After presenting it to the user 2 and accepting and setting the change operation (step S106), the process returns to the process from step S103.

In this way, by appropriately changing and setting the selected operation values and their numerical values, each operation value that satisfies the standard for all treated water quality in all estimation models is explored.

In step S103, when it is determined that the quality of all treated water in the currently selected estimation model satisfies the standard (YES in step S103), the estimation calculation unit 13 determines the set operation value ( Step S107), the determined operation value is stored in the estimation result storage unit 16 (step S108), and the processing of FIG. 3 mainly by the estimation calculation unit 13 ends.

FIG. 4 is a flow chart showing the processing details when the aggregate computation commanding unit 14 commands the aggregate computation unit 17 to use the aggregate computation model in a state in which the estimation results of all the estimation models are stored in the estimation result storage unit 16. be.

The example of processing shown in FIG. 4 shows a case where the number of aggregate computation models is limited to two. Actually, it is not limited to this, and may be a case of switching three or more models. Also, a value other than the steady-state value may be used as the judgment criterion.

At the beginning of the processing, the aggregate calculation commanding unit 14 reads and acquires all input information from the input information storage unit 12 (step S201). The aggregate computation command unit 14 refers to the contents of the operation value information and statistical information included in all the acquired input information, and selects an optimal aggregate computation model according to their characteristics.

Specifically, it is determined whether or not all the input information is out of the range of steady values set in advance (step S202).

If it is determined that all the input information is within the steady-state value range (YES in step S202), the estimation result of the static analysis estimation model using the physical / chemical model and the estimation model of machine learning Both estimation results are considered to be highly reliable. Therefore, the aggregate operation command unit 14 stores an aggregate operation model using both the estimation result of the static analysis estimation model using the physical/chemical model and the estimation result of the machine learning inference model. 18 to execute the calculation by the selected aggregate calculation model (step S203), and the process of FIG. 4 is finished.

Also, in step S202, when it is determined that at least one of all the input information is out of the range of steady values (NO in step S202), the reliability of the estimation result of the machine learning estimation model is low. It is considered to be Therefore, the aggregate computation command unit 14 does not use the estimation result of the machine learning estimation model, and uses only the estimation result of the static analysis estimation model using the physical/chemical model to perform the aggregate computation. The intensive calculation unit 17 is instructed to select from the model storage unit 18, the calculation is executed by the selected intensive calculation model (step S204), and the process of FIG. 4 is terminated.

In the description of FIG. 4, the aggregation operation commanding unit 14 clearly determines whether or not to use the estimation result of the estimation model based on machine learning based on the setting of the steady-state value, and the aggregation performed by the aggregation operation unit 14 Although the calculation model is commanded, internal processing may be performed to automatically determine whether or not to use the estimation result of the estimation model based on machine learning within the aggregate calculation model.

In this case, internal processing in the aggregation calculation unit 17 using an aggregation calculation model, which is a black box model, that calculates one output value by combining multiple estimation results in a complicated manner using machine learning using pre-learned data. By doing so, the configuration of the aggregate calculation commanding unit 14 can be omitted.

When performing such internal processing, the criteria for judging deviation from the steady-state value can be expressed continuously by automatic adjustment according to the degree of deviation, instead of being discontinuous based on uniform or step-by-step threshold settings. becomes. In addition, not only judgment criteria based on simple steady-state values, but also complex conditions based on the characteristics of the machine field, such as unsteady conditions only when condition A and condition B overlap, can be included in the aggregate calculation model. is.

Aggregation operation unit 17 reads an aggregation operation model commanded by aggregation operation commanding unit 14 or an aggregation operation model provided with a mechanism for determining by internal processing from aggregation operation model storage unit 18, and executes the operation to perform aggregation. One highly accurate estimated value of treated water quality is calculated and presented to the user 2 as estimation result information.

As described in detail above, according to this embodiment, it is possible to automatically and appropriately select an aggregate computation model according to the data characteristics of input information. As a result, the user 2 can acquire highly accurate estimation result information without having to manually select an aggregate computation model according to the situation.

[Second embodiment]
FIG. 5 is a block diagram showing a specific example of the functional configuration of the treated water quality estimation device 1 according to the second embodiment instead of the functional configuration of FIG. Since the basic configuration is almost the same, the same parts are denoted by the same reference numerals, and the description thereof is omitted.

In this embodiment, it becomes the structure which adds the aggregation calculation model update part 21 to the treated water quality estimation apparatus 1. FIG. This aggregate computation model update unit 21 updates the current treated water quality measurement value included in the statistical information in the input information stored in the input information storage unit 12 and the most recent past treated water quality stored in the estimation result storage unit 16 Based on the estimated value, the aggregate computation model stored in the aggregate computation model storage unit 18 is updated as necessary.

Next, the operation of this embodiment will be described.
FIG. 6 is a flow chart showing the processing contents for updating the aggregate computation model by the aggregate computation model update unit 21. As shown in FIG. This process is executed at a constant period equal to or longer than the statistical information update period, for example, every 24 hours. is obtained (step S301).

Next, the aggregate computation model updating unit 21 acquires the treated water quality measurement values from 24 hours ago to the present, which are included in the input information stored in the input information storage unit 12 (step S302).

The aggregate computation model update unit 21 uses each of the acquired information to determine how much error the treated water quality estimated value calculated in the past for the simultaneous series has as a reference, the treated water quality measured value from 24 hours ago to the present. is calculated (step S303).

The aggregate computation model update unit 21 updates the aggregate computation model based on whether or not the calculated error in the estimated value of treated water quality from 24 hours ago to the present exceeds a preset error threshold. It is determined whether or not there is a need to do so (step S304).

If it is determined that the error in the estimated treated water quality from 24 hours ago to the present is within a range that does not exceed the preset error threshold value (NO in step S304), the aggregate computation model updating unit 21 Assuming that there is no need to update the aggregate computation model, the process of FIG. 6 ends.

On the other hand, when it is determined that the error in the treated water quality measurement value from 24 hours ago to the present exceeds the preset threshold value (YES in step S304), the aggregate computation model updating unit 21 updates the current After updating the aggregate computation model stored in the aggregate computation model storage unit 18 based on the treated water quality measurement value (step S305), the processing in FIG. 6 ends.

Regarding the update processing of the aggregate computation model stored in the aggregate computation model storage unit 18, a method of updating by weighting the linear characteristics according to the magnitude and direction of the error, a method using machine learning, etc. can be considered.

As described above, according to the present embodiment, by comparing the estimation result of each past estimation model and the current measured value of treated water quality, which should be the correct value corresponding to it, The reliability of each estimation model corresponding to the characteristics of minute input information can be automatically evaluated, and the aggregate computational model can be updated based on the evaluation results. As a result, the accuracy of the final estimation result obtained by the computation in the aggregate computation model can be made higher (effect of claim 3).

[Third embodiment]
FIG. 7 is a block diagram showing a specific example of the functional configuration of the treated water quality estimation device 1 according to the third embodiment instead of the functional configuration of FIG. Since the basic configuration is almost the same, the same parts are denoted by the same reference numerals, and the description thereof is omitted.

In this embodiment, instead of the estimation calculation unit 13 that serially executes calculations of a plurality of estimation models, estimation calculation units 131, 132, . . . that execute calculations of a plurality of estimation models in parallel are provided. Also, an estimation operation commanding unit 22 is provided in order to allocate estimation models to be executed by these estimation operation units 131, 132, .

The estimation calculation units 131, 132, . . . may be arranged in parallel on hardware, or may be logically parallelized on the same hardware. Furthermore, the plurality of estimation calculation units 131, 132, . . . may be computer resources with different performance in terms of hardware or software. configuration can be changed.

The estimation calculation commanding unit 22 can speed up processing by sequentially allocating estimation models in parallel to the plurality of estimation calculation units 131, 132, . . . and executing calculations.

Further, the estimation computation commanding unit 22 refers to the estimation result used for the subsequent aggregation computation stored in the aggregation computation model storage unit 18, thereby allocating only necessary estimation computations to the estimation computation units 131, 132, . . . It is also possible to shorten the time until the final estimation result information is output.

In this case, in addition, the estimation calculation commanding unit 22 causes the incomplete estimation results to use provisional values such as past estimation results even before all the estimation calculations are completed by the aggregation calculation unit 17. It is possible to make the provisional aggregate operation result closer to the final aggregate operation result at an early stage.

Next, the operation of this embodiment will be described.
In this operation example, all of the plurality of estimation models are not simply assigned to the estimation calculation units 131, 132, . will be described.

FIG. 8 is a flow chart showing details of control processing for parallel processing of a plurality of estimation computations and aggregation computation by the estimation computation commanding unit 22 . Past estimation results are used for provisional values.

When the estimation calculation units 131, 132, . , and refer to the estimation result required when executing the aggregate computation model in the subsequent process (step S401).

Based on this reference result, the estimation computation commanding unit 22 limits the estimation models to be executed by the estimation computation units 131, 132, . It is instructed to execute the estimated models in parallel (step S402).

Note that the degree of importance here is set by the degree of reference of the estimation result in the aggregate computation model, for example, weighting in linear sums, calculation load, and the like.

Upon receiving a command from the estimation calculation commanding unit 22, the estimation calculation units 131, 132, . Store.

After that, the estimation calculation commanding unit 22, which monitors the estimation calculation units 131, 132, .

If it is determined that the execution of all commanded estimation model calculations has not been completed (NO in step S403), the estimation calculation commanding unit 22 is used for input in the aggregate calculation model stored in the aggregate calculation model storage unit 18. Regarding the input information, regarding the estimation result for which the estimation result has not yet been calculated, the past estimation result stored in the estimation result storage unit 16 is read out and replaced (step S404), and the provisional aggregation operation is executed by the aggregation operation model. (step S405). Then, after outputting the calculation result as temporary estimation result information at any time (step S406), the process returns to step S403.

In this way, the processing of steps S403 to S406 is repeatedly executed until the execution of the calculations of all the commanded estimation models is completed, and the estimation results of the provisional aggregate calculation model are continuously output. The calculation of the estimation model by the estimation calculation units 131, 132, .

Therefore, before the execution of all the calculations of the estimation model is completed, the estimation result information whose accuracy is gradually increased is output even though it is a provisional aggregation calculation.

Then, at step S403, when it is determined that the execution of all commanded estimation model calculations has been completed (YES in step S403), the estimation calculation commanding unit 22 executes the final aggregate calculation model using the estimation results. (step S407).

Based on this command, the aggregation calculation unit 17 executes the final aggregation calculation and outputs the calculated estimation result information (step S408), thereby ending the processing in FIG.

As described above, according to this embodiment, by executing estimation operations in parallel, it is possible to speed up the execution of operations in all estimation models.

In particular, when executing calculations in the estimation model by allocating them to multiple calculation units, it is necessary to limit the calculations to only the calculations in the estimation model that are deemed necessary, and to perform calculations after setting priorities according to their importance. can further reduce the computation time.

In addition, during the execution of computations in the parallel estimation models, the aggregation computation unit 17 is caused to read out provisionally corresponding past estimation results from the estimation result storage unit 16 for estimation results that have not yet been calculated. Since the computation of the aggregate computation model is executed by the temporary aggregate computation model, it is possible to bring the estimation result calculated by the computation of the temporary aggregate computation model closer to the finally obtained value at an earlier stage.

[Fourth embodiment]
FIG. 9 is a block diagram showing a specific example of the functional configuration of the treated water quality estimation device 1 according to the fourth embodiment instead of the functional configurations of FIGS. Since the basic configuration is almost the same, the same parts are denoted by the same reference numerals, and the description thereof is omitted.

In this embodiment, the estimation result information output from the aggregation calculation unit 17 is provided to a GUI (Graphical User Interface) display unit 23 . The GUI display unit 23 displays the numerical information contained in the estimation result information given from the aggregation calculation unit 17 as a graph display developed on the screen of the display in a visually easy-to-understand form using coordinates and color shading. express.

FIG. 10 shows an example of a graph display screen on the GUI display section 23 presented to the user 2. As shown in FIG. On the right side of the center of the screen, as items that can be selected by the user 2, the operation index items "aeration air volume", "return rate", and "excess sludge removal rate", and the management index item "MLSS (Mixed Liquor Suspended Solids: Activated sludge suspended matter), DO (Dissolved Oxygen), and SRT (Sludge Retention Time) are listed. Here, an example is shown in which the item "aeration air volume" in the operation index is selected and displayed in reverse so as to be distinguished from other items.

Correspondingly, on the left side including the center of the screen, "PO4-P (phosphate phosphorus)", "NH4-N (nitrate nitrogen)", and "NO3-N (ammonia nitrogen)" Estimated graph corresponding to "aeration air volume" is displayed.

In each graph, the vertical axis is the aeration air volume [m ³ /h], which is the operation value, and the horizontal axis is the future time axis with the current time as the left end. -N" and "NO3-N" are three-dimensional graphs that display estimated values for each concentration. In each graph, the dashed line indicates the current aeration air volume of slightly less than 300 [m ³ /h].

Each graph shows estimation results from the current time "6:00" to tomorrow's "6:00" 24 hours later at the maximum, as shown at the bottom. For example, in the graph of "PO4-P" in the upper part, the PO4-P concentration increases when the aeration air volume is 400 [m ³ /h] or more, while the aeration air volume is 200 [m ³ /h] or less. It visually expresses that the PO4-P concentration is expected to increase after 12 hours even if the dose is reduced.

In addition, instead of such a three-dimensional graph, the display can be easily switched to a two-dimensional graph representing the estimation results for the current operating values.

FIG. 11 is a diagram exemplifying a two-dimensional graph that is switched from the current state in which the graph button BT2 is operated in FIG. 10 when the button BT1 is operated. The vertical axis is the concentration of the treated water quality, and the horizontal axis is the time axis.

In the same figure, the operation of one button BT1 shows the case where the graphs of three treated water qualities "PO4-P", "NH4-N" and "NO3-N" are switched in tandem. It is also possible to selectively manipulate the contents of the graph, as buttons are provided in .

As described above, according to the present embodiment, it is possible to compare and display a plurality of treated water quality estimation results for operation values within one screen. can do.

Although several embodiments of the invention have been described above, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the scope of the invention described in the claims and equivalents thereof.

DESCRIPTION OF SYMBOLS 1... Treated water quality estimation apparatus, 2... User, 3... Monitoring and control system, 4... Measuring device, 5... Water supply and sewage plant, 6... Influent water, 7... Treated water, 11... Input information acquisition part, 12... Input information storage Part 13, 131, 132... Estimation calculation unit 14... Aggregation calculation command unit 15... Estimation model storage unit 16... Estimation result storage unit 17... Aggregation calculation unit 18... Aggregation calculation model storage unit 21... Aggregation Calculation model updating unit 22 Estimated calculation command unit 23 GUI display unit 100 Plant to be controlled.

Claims (9)

an input information acquisition unit that acquires as input information statistical information obtained by measuring the quality of water that needs to be processed in the plant to be controlled, and operation value information;
an input information storage unit that stores the input information acquired by the input information acquisition unit;
an estimation model storage unit that stores a plurality of estimation models that calculate an estimated value of treated water quality using input information with different algorithms;
an estimation calculation unit that calculates an estimated value of treated water quality using input information stored in the input information storage unit and a plurality of estimation models stored in the estimation model storage unit;
an estimation result storage unit that stores the estimated value of treated water quality calculated by the estimation calculation unit;
an aggregate calculation model storage unit that stores a plurality of aggregate calculation models that aggregate a plurality of treated water quality estimation values;
an aggregation computation unit that aggregates a plurality of treated water quality estimation values stored in the estimation result storage unit using an aggregation computation model stored in the aggregation computation model storage unit;
an aggregate computation commanding unit that commands an aggregate computation model used by the aggregate computation unit based on the input information stored in the input information storage unit;
Treated water quality estimation device. The estimated model storage unit stores an estimated model based on static analysis of one or more physical/chemical models and an estimated model based on data analysis of one or more machine learning models,
The aggregate calculation command unit performs estimation based on static analysis of a physical/chemical model based on whether or not the input information stored in the input information storage unit is out of a preset steady-state value range. select and command an aggregate computation model that uses the estimation results of the model and an aggregate computation model that uses the estimation results of the estimation model based on the data analysis of the machine learning model;
The treated water quality estimation device according to claim 1. Store the aggregate calculation model based on the current treated water quality measurement value in the statistical information included in the input information stored in the input information storage unit and the past treated water quality estimated value stored in the estimation result storage unit 2. The treated water quality estimation device according to claim 1, further comprising an aggregate computation model updating unit that updates the aggregate computation model stored in the unit. The estimation calculation unit performs a calculation using the estimation model in parallel with a plurality of
further comprising an estimation computation commanding unit that allocates the estimation model stored in the estimation model storage unit to each of the plurality of estimation computation units and causes each of the estimation computation units to execute computation;
The treated water quality estimation device according to claim 1. The estimation computation command unit limits the estimation model to be allocated to the estimation computation unit based on the estimation result used in the aggregate computation model stored in the aggregate computation model storage unit and the importance thereof, and selects an estimation model according to the importance. 5. The treated water quality estimation device according to claim 4, wherein a priority is set and an estimation model is assigned to each of said plurality of estimation calculation units to execute calculation. The estimation computation commanding unit causes the aggregation computation unit to read the corresponding past estimation result from the estimation result storage unit instead of the estimation result that has not yet been calculated by the estimation computation unit, and computes the aggregation computation model. The treated water quality estimation device according to claim 4 or 5, which executes 2. The treated water quality estimation device according to claim 1, further comprising a graph display unit for displaying the estimated treated water quality values aggregated by the aggregation calculation unit in a graph showing the relationship between the operation value and the prediction time. A computer-executed method for estimating treated water quality,
an input information acquisition step of acquiring statistical information obtained by measuring the quality of water that needs to be treated in the plant to be controlled, and manipulated value information as input information;
an input information storing step of storing the input information acquired in the input information acquiring step;
an estimation model storing step of storing a plurality of estimation models for calculating an estimated value of treated water quality using input information with different algorithms;
an estimation calculation step of calculating an estimated value of treated water quality using the input information stored in the input information storing step and a plurality of estimation models stored in the estimation model storing step;
an estimation result storing step of storing the treated water quality estimated value calculated in the estimation calculation step;
an aggregate calculation model storing step of storing a plurality of aggregate calculation models that aggregate a plurality of treated water quality estimated values;
an aggregation operation step of aggregating the plurality of treated water quality estimation values stored in the estimation result storing step using the aggregation operation model stored in the aggregation operation model storage step;
an aggregate computation command step of commanding an aggregate computation model used in the aggregate computation step based on the input information stored in the input information storing step;
A method for estimating treated water quality. A program executed by a computer, the computer comprising:
an input information acquisition unit that acquires as input information statistical information obtained by measuring the quality of water that needs to be processed in the plant to be controlled, and operation value information;
an input information storage unit that stores the input information acquired by the input information acquisition unit;
an estimation model storage unit that stores a plurality of estimation models that calculate an estimated value of treated water quality using input information with different algorithms;
an estimation calculation unit that calculates an estimated value of treated water quality using the input information stored in the input information storage unit and a plurality of estimation models stored in the estimation model storage unit;
an estimation result storage unit that stores the estimated value of treated water quality calculated by the estimation calculation unit;
an aggregate calculation model storage unit that stores a plurality of aggregate calculation models that aggregate a plurality of treated water quality estimation values;
an aggregation computation unit that aggregates a plurality of treated water quality estimation values stored in the estimation result storage unit using an aggregation computation model stored in the aggregation computation model storage unit;
an aggregate computation commanding unit that commands an aggregate computation model used in the aggregate computation unit based on the input information stored in the input information storage unit;
A program that works as

Images