Information processing device, dehydrator, moisture content estimation method, and moisture content estimation program

Abstract

To maintain estimated accuracy of a water content of a dewatering cake in a normal case while alleviating reduction in the estimated accuracy when a driving condition is changed.SOLUTION: In an information processor (1), an estimation part (102) estimates a water content using a first estimation model (111), from measurement data acquired by an acquisition part (101), and an estimation value by a second estimation model (112) for a water content at a measurement time point of the measurement data. Measurement data used for learning by the first estimation model (111) is measurement data for a prescribed period of an actual operation period of a dehydrator. The second estimation model (112) is an estimation model learned by using the measurement data in a driving period in which the driving condition is different from that in the actual driving period as an explanatory variable, and an actually measured value of the water content at the measurement time point as an objective variable.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a technique for estimating the water content of a dewatered cake obtained by dehydrating a liquid containing suspended solids with a dehydrator.

Background Art

[0002] The sludge treatment carried out in wastewater treatment facilities such as sewage treatment plants includes a step of dehydrating sludge with a dehydrator. For efficient sludge treatment, it is important to maintain the water content of the dewatered cake obtained by dehydration within a predetermined range. However, when dehydration is performed with the operating conditions of the dehydrator kept constant, the water content of the dewatered cake fluctuates due to causes such as the non-constant properties of the supplied sludge, so it is not easy to maintain the water content of the dewatered cake within a predetermined range.

[0003] For this reason, the development of a technique for estimating the water content of a dewatered cake has been conventionally advanced. If the current water content can be estimated in real time, it becomes possible to maintain the water content within a predetermined range by various controls. For example, Patent Document 1 below discloses a technique for generating a water content estimation model using a plurality of parameters such as the amount of sludge supplied to a centrifugal dehydrator and values related to the centrifugal effect of the dehydrator, and estimating the water content.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Generally, the sludge treatment equipment described above is test-run under a wide variety of operating conditions to determine the optimal operating conditions. During this test-run period, the moisture content estimation model is generated using the diverse measurements of the various parameters measured under the wide variety of operating conditions as training data.

[0006] Subsequently, the device is put into actual operation based on the determined operating conditions. During this actual operation period, the moisture content estimation model is updated using the most recent measured values ​​of the above-mentioned parameters as training data. As a result, the moisture content estimation model becomes an estimation model that is adapted to the most recent operating conditions. Therefore, under normal circumstances where there are no changes in operating conditions, the accuracy of the estimation by the moisture content estimation model is maintained. However, there is a concern that the accuracy of the estimation by the moisture content estimation model will decrease if the operating conditions are changed.

[0007] In this regard, it is conceivable to update the moisture content estimation model using at least some of the above-mentioned diverse measurement values ​​and the most recent measurement values ​​as training data. However, due to the influence of the above-mentioned diverse measurement values, the accuracy of the estimation by the moisture content estimation model will usually decrease.

[0008] One aspect of the present invention aims to mitigate the decrease in the accuracy of the estimation of the moisture content when operating conditions are changed, while maintaining the accuracy of the estimation under normal circumstances. [Means for solving the problem]

[0009] To solve the above problems, an information processing device according to one aspect of the present invention provides a liquid containing suspended solids to which a chemical agent for coagulating the suspended solids is added, and the liquid discharged from the coagulation tank is used to process the liquid containing suspended solids. The system includes an acquisition unit that acquires measurement data related to the operation of a dewatering machine that dewaters while transporting, and an estimation unit that uses a first estimation model to estimate the moisture content at the time of measurement of the measurement data from the measurement data acquired by the acquisition unit and an estimated value by a second estimation model regarding the moisture content of the dewatered cake discharged from the dewatering machine at the time of measurement of the measurement data. The measurement data used by the first estimation model for learning is the measurement data acquired by the acquisition unit from a predetermined period before the learning time to the learning time during the actual operation period of the dewatering machine, and the second estimation model is an estimation model that has been learned with the measurement data acquired by the acquisition unit during an operation period in which the operating conditions differ from the actual operation period as explanatory variables, and the measured value of the moisture content at the time of measurement of the measurement data as the dependent variable.

[0010] Furthermore, another aspect of the present invention relates to a moisture content prediction method, which is a moisture content estimation method performed by one or more information processing devices, and includes an acquisition step of acquiring measurement data relating to the operation of a dewatering machine that dewaters a liquid while transporting the liquid discharged from a flocculation tank to which a chemical agent for flocculating suspended solids is added to a liquid containing suspended solids, and an estimation step of using a first estimation model to estimate the moisture content at the time of measurement of the measurement data from the measurement data acquired in the acquisition step and an estimated value by a second estimation model for the moisture content of the dewatered cake discharged from the dewatering machine at the time of measurement of the measurement data, wherein the measurement data used by the first estimation model for learning is the measurement data acquired in the acquisition step from before a predetermined period before the learning time to the learning time during the actual operation period of the dewatering machine, and the second estimation model is an estimation model that has been learned with the measurement data acquired in the acquisition step during an operation period in which the operating conditions differ from those of the actual operation period as explanatory variables and the measured value of the moisture content at the time of measurement of the measurement data as the objective variable. [Effects of the Invention]

[0011] According to one aspect of the present invention, the accuracy of the estimation of the moisture content can be maintained under normal circumstances while mitigating the decrease in accuracy of the estimation when operating conditions are changed, etc. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing an example of the main components of an information processing device according to one embodiment of the present invention. [Figure 2] This figure shows an example configuration of a control system including the above-mentioned information processing device. [Figure 3] This flowchart shows an example of the process for estimating the moisture content of a dehydrated cake in the above-mentioned information processing device. [Figure 4] This flowchart shows an example of the update process for the first estimation model in the above-mentioned information processing device. [Figure 5] This figure shows, in tabular form, the combinations of explanatory variables and target variables of the first estimation model in Embodiment 1 of the above-mentioned information processing device. [Figure 6] This figure shows, in tabular form, the combinations of explanatory and dependent variables in the second estimation model in Example 1. [Figure 7] This graph shows the variability between the moisture content estimates from the first estimation model in Example 1 and the actual moisture content measurements taken by the workers. [Figure 8] This graph shows the variability between the moisture content estimates obtained by the comparative models in Comparative Example 1 and Comparative Example 2 and the actual moisture content measurements taken by the workers. [Figure 9] This is a block diagram showing an example of the main components of an information processing device according to another embodiment of the present invention. [Figure 10] This is a model diagram illustrating the concept of creating training data for a predictive model in the above-mentioned information processing device. [Figure 11] This flowchart shows an example of the process for predicting the moisture content of a dehydrated cake in the above-mentioned information processing device. [Figure 12] This flowchart shows an example of the update process for the first estimation model and the prediction model in the above-mentioned information processing device. [Modes for carrying out the invention]

[0013] Hereinafter, embodiments of the present invention will be described in detail. For the sake of convenience of explanation, members having the same functions as those shown in each embodiment are denoted by the same reference numerals, and the description thereof will be omitted as appropriate.

[0014] [Embodiment 1] One embodiment of the present invention will be described with reference to FIGS. 1 to 4.

[0015] [System Configuration] FIG. 2 is a diagram showing a configuration example of the control system 100 according to the present embodiment. The control system 100 is a system used in a plant that adds a chemical for aggregating solid suspended matter to a liquid to be treated (liquid) in an aggregation tank to form flocs, and performs solid-liquid separation of the liquid to be treated in which the flocs are formed. Hereinafter, an example in which the liquid to be treated is sludge will be described, but the control system 100 is also applicable to a plant that treats a liquid to be treated other than sludge. Note that sludge is a liquid containing fine solids generated in wastewater treatment or the like, and can also be called slurry.

[0016] Although details will be described below, the control system 100 performs each treatment from the step of aggregating the sludge to be treated into aggregated sludge by aggregating the solid suspended matter in the sludge to be treated to form flocs to the step of dehydrating the aggregated sludge to obtain dehydrated sludge (also called dehydrated cake) and dehydrated filtrate in the sludge treatment process. As shown in FIG. 2, the control system 100 includes an information processing device 1, a control device 3, a flocculator 5, and a dehydrator 9.

[0017] The flocculator 5 is a device that forms flocs by adding a chemical for aggregating solid suspended matter to the liquid to be treated in an aggregation tank and stirring it appropriately. Specifically, the flocculator 5 uses sludge as the liquid to be treated, aggregates the solid suspended matter in the sludge to form flocs, and generates aggregated sludge. The flocculator 5 in FIG. 2 includes an aggregation tank 51, a stirring blade 52, a motor 53, and an inspection window 54. In addition, the flocculator 5 is provided with a sludge inlet 55, a chemical inlet 56, and a discharge port 57.

[0018] Furthermore, an imaging device 72 and an imaging lighting device 71 are attached to the inspection window 54. The imaging device 72 only needs to be capable of capturing at least still images. It is preferable that the coagulation tank 51 be opaque so that the way light hits the flocs does not change while the control system 100 is in operation. It is also preferable that the imaging device 72 and the lighting device 71 be housed in a light-shielding dark box with an opening on the inspection window 54 side, as shown in the illustrated example.

[0019] The dewatering machine 9 is a device that separates the solids and liquids of the treated liquid in which flocs have formed. Specifically, the dewatering machine 9 is installed downstream of the flocculator 5 and dewaters the flocculated sludge (liquid) discharged from the flocculator 5 to separate the solids and liquids. The dewatering machine 9 in Figure 2 is a screw press type dewatering machine equipped with an outer shell screen 91 and a screw 92. The dewatering machine 9 is also provided with a sludge inlet 93, a filtrate outlet 94, and a dewatered cake outlet 95. Although not shown in the figure, the dewatering machine 9 is also equipped with a motor to rotate the screw 92. Of course, the dewatering machine 9 can be any machine capable of dewatering flocculated sludge and is not limited to a screw press type. For example, a centrifugal dewatering machine, a filter press type dewatering machine, or a belt press dewatering machine can also be used.

[0020] In the control system 100, the sludge to be processed is continuously or intermittently supplied from the sludge inlet 55 into the flocculator 5's coagulation tank 51 by a supply device (not shown). The sludge supply rate is controlled according to the sludge processing rate of the flocculator 5 and the dewatering machine 9. The device or its control device 3 may be configured to be controlled automatically.

[0021] Then, a chemical agent (at least containing a coagulant) is introduced into the sludge in the coagulation tank 51 through the chemical inlet 56 to coagulate the sludge. In this state, the motor 53 is driven to rotate the stirring blade 52, stirring the sludge and chemical agent to form flocs. The coagulated sludge, which is a mixture of the formed flocs and the water contained in the sludge, is discharged from the discharge port 57.

[0022] Next, the coagulated sludge is supplied into the outer shell screen 91 from the sludge inlet 93 of the dewatering machine 9. Inside the dewatering machine 9, the coagulated sludge is dewatered under pressure from the screw 92, the filtrate is discharged from the filtrate outlet 94, and the dewatered cake, which is a mass of dewatered coagulated sludge, is discharged from the dewatered cake outlet 95.

[0023] The sludge is supplied to the coagulation tank 51 from the sludge inlet 55, and the coagulated sludge is pushed out from the discharge port 57 of the coagulation tank 51 and discharged, and the discharged coagulated sludge is supplied to the dewatering machine 9. For this reason, the flow rate of sludge supplied to the coagulation tank 51 and the flow rate of sludge supplied to the dewatering machine 9 coincide at the same time.

[0024] As will be explained in detail below, the information processing device 1 acquires measurement data related to the operation of the dewatering machine 9. Then, based on the acquired measurement data, the information processing device 1 estimates the moisture content of the dewatered cake.

[0025] Furthermore, the information processing device 1 can also control the operation of various devices that are components of the control system 100 (for example, the flocculator 5, the dewatering machine 9, and a sludge and chemical supply device not shown) via the control device 3. The control device 3 is a device that controls the operation of various devices that are components of the control system 100. The control device 3 may be, for example, a PLC (Programmable Logic Controller). In the example in Figure 2, the control device 3 controls the flocculator It includes a control device 3a for the flocculator that controls the equipment related to the data 5, and a control device 3b for the dewatering machine that controls the equipment related to the dewatering machine 9.

[0026] [Device configuration] The configuration of the information processing device 1 will be explained based on Figure 1. Figure 1 is a block diagram showing an example of the main components of the information processing device 1. As shown in the figure, the information processing device 1 includes a control unit 10 that controls all parts of the information processing device 1, and a storage unit 11 that stores various data used by the information processing device 1. The information processing device 1 also includes a communication unit 12 for the information processing device 1 to communicate with other devices, an input unit 13 that receives input of various data to the information processing device 1, and an output unit 14 for the information processing device 1 to output various data.

[0027] Furthermore, the control unit 10 includes an acquisition unit 101, an estimation unit 102, and an update unit 103. The update unit 103 will be explained later in the section "About the Update Unit".

[0028] The memory unit 11 contains a first estimation model 111 and a second estimation model 112. Details of these will be explained later in the sections "About the First Estimation Model" and "About the Second Estimation Model," respectively.

[0029] The acquisition unit 101 acquires measurement data related to the operation of the dehydrator 9. Details of this measurement data will be explained later in the section "About the First Estimation Model".

[0030] The estimation unit 102 uses the first estimation model 111 stored in the storage unit 11 to estimate the moisture content of the dewatered cake from the measurement data acquired by the acquisition unit 101. This dewatered cake is the dewatered cake discharged from the dewatering machine 9 at the time of measurement of the above measurement data. The estimation unit 102 Each time the acquisition unit 101 acquires new measurement data, it performs estimation using the first estimation model 111. Therefore, the estimation unit 102 can estimate the moisture content of the dehydrated cake in real time, that is, continuously at short time intervals (for example, every minute).

[0031] [Regarding the first estimation model] The first estimation model 111 is an estimation model that was learned using measurement data related to the operation of the dewatering machine 9, acquired by the acquisition unit 101, as the explanatory variable, and the moisture content of the dewatered cake discharged from the dewatered cake discharge port 95 of the dewatering machine 9 at the time of measurement of the said measurement data as the dependent variable. Various methods such as multivariate regression analysis and RandomForestRegressor can be used for the above learning. can.

[0032] In this embodiment, the first estimation model 111 has the estimated moisture content from the measurement data obtained by the second estimation model 112 added as one of the explanatory variables. The second estimation model 112 will be explained later in the section "About the Second Estimation Model".

[0033] The above measurement data consists of at least one of the following: the operating time of the dewatering machine 9, the rotational speed of the screw of the dewatering machine 9, the drive current value or torque value of the screw, the back pressure from the back pressure plate provided at the dewatering cake discharge port 95 (discharge section) of the dewatering machine 9 to compress the liquid, and the opening degree between the back pressure plate and the dewatering cake discharge port 95. These are obtained from the dewatering machine control device 3b. Of these, the measurement data that contribute most significantly (are of high importance) to the estimation accuracy of the first estimation model 111 are the drive current value of the screw and the opening degree (especially when the dewatering machine 9 is controlled so that the back pressure remains constant).

[0034] [Regarding the second estimation model] The second estimation model 112 is an estimation model learned using measurement data related to the operation of the dewatering machine 9 acquired by the acquisition unit 101 as explanatory variables, and the moisture content at the time of measurement of said measurement data as the dependent variable. In this embodiment, the second estimation model 112 is an estimation model learned during the trial run period, which is before the start of the actual operation period of the dewatering machine 9. The measurement data that serves as the explanatory variables for the second estimation model 112 may be the same as or different from the measurement data that serves as the explanatory variables for the first estimation model 111.

[0035] As described above, the information processing device 1 of this embodiment includes an acquisition unit 101 that acquires measurement data related to the operation of a dewatering machine 9 that dewaters the liquid discharged from a flocculation tank 51 to which a chemical agent for flocculating the suspended solids is added, and an estimation unit 102 that uses a first estimation model 111 to estimate the moisture content at the time of measurement of the measurement data from the measurement data acquired by the acquisition unit 101 and an estimated value of the moisture content, which is the moisture content of the dewatered cake discharged from the dewatering machine 9 at the time of measurement of the measurement data, by a second estimation model. The measurement data used by the first estimation model 111 for learning is the measurement data acquired by the acquisition unit 101 from before a predetermined period before the learning time to the learning time during the actual operation period of the dewatering machine 9. The second estimation model 112 is an estimation model that has been learned with the measurement data acquired by the acquisition unit 101 during a trial operation period, which is an operation period in which the operating conditions differ from the actual operation period, as the explanatory variable, and the moisture content at the time of measurement of the measurement data as the dependent variable.

[0036] According to the above configuration, the second estimation model 112 is an estimation model that was trained using measurement data acquired during the trial run period of the dewatering machine 9 under a wide variety of operating conditions compared to the operating conditions during the actual operating period as training data. Therefore, the estimate of the moisture content at the time of measurement by the second estimation model 112 is an estimate that is suitable for the wide variety of operating conditions mentioned above.

[0037] On the other hand, the first estimation model 111 is an estimation model that is learned using measurement data acquired from a predetermined period before the learning point to the learning point during the actual operation period of the dewaterer 9, and the estimated value of the moisture content at the time of measurement by the second estimation model 112 as explanatory variables, with the actual measured value of the moisture content at the time of measurement as the dependent variable. Since the first estimation model 111 is learned using the most recent measurement data, it becomes an estimation model that is suited to the operating conditions during the most recent actual operation period. Therefore, in normal cases where there are no changes in operating conditions, the accuracy of the estimation by the first estimation model 111 is maintained.

[0038] Furthermore, since the first estimation model 111 is trained using the estimated moisture content value estimated by the second estimation model 112 using the most recent measurement data, it becomes an estimation model that takes into account a wide variety of operating conditions. Therefore, even if operating conditions are changed after a long period of time without any changes, the decrease in the accuracy of the estimation by the first estimation model 111 can be mitigated.

[0039] Furthermore, it is desirable that the measurement data used by the first estimation model 111 for training be the measurement data acquired by the acquisition unit 101 from about two weeks before the training time to the training time. Alternatively, it is desirable that the measurement data used by the first estimation model 111 for training be the measurement data acquired by the acquisition unit 101 at approximately 48 measurement points immediately preceding the training time. This is because if the number of measurements is less than these, the accuracy will deteriorate, while if the number of measurements is too large, the influence of older measurement data will become apparent.

[0040] [Regarding the update section] The update unit 103 performs a process to update the first estimation model 111 during periods when the coagulation tank 51 and the dewatering machine 9 are stopped (downtime). Specifically, the update unit 103 uses the second estimation model 112 to calculate an estimated value of the moisture content at the time of measurement of the measurement data from the measurement data acquired by the acquisition unit 101 during the operation period of the dewatering machine 9. Then, the update unit 103 updates the first estimation model 111 using the set of the above measurement data, the estimated value of the moisture content, and the actual measured value of the moisture content of the dewatered cake measured by the operator as training data.

[0041] Typically, the above operating period is one set period within a day. Therefore, the update by the update unit 103 is performed once a day. Also, the actual measurement of the moisture content of the dewatered cake by the operator takes time. Therefore, instead of the actual moisture content measured by the operator, an estimated moisture content calculated using another model may be used as the target variable of the training data. Alternatively, the update unit 103 may update the first estimation model 111 using measurement data from the above operating period, for example, one to two days prior.

[0042] Therefore, by updating the first estimation model 111 using the latest measurement data, the estimation using the first estimation model 111 can be adapted to the latest conditions of the dehydrator 9.

[0043] [Estimation process] Figure 3 is a flowchart showing an example of the moisture content estimation process (moisture content estimation method) of the dewatered cake in the information processing device 1 with the above configuration. This estimation process is performed during the operating period, as described above. As shown in Figure 3, first, the acquisition unit 101 collects (acquires) various measurement data (S11, acquisition step). Next, the estimation unit 102 calculates an estimated value for the moisture content at the time of measurement of the measurement data using the second estimation model 112 from the measurement data, and estimates the moisture content at the time of measurement using the first estimation model 111 from the estimated value and the measurement data (S12, estimation step). After that, the process returns to step S11 and the above operation is repeated.

[0044] [Update process] Figure 4 is a flowchart showing an example of the update process for the first estimation model 111 in the information processing device 1. This update process is performed after each downtime, as described above.

[0045] As shown in Figure 4, first, the update unit 103 uses the second estimation model 112 to calculate an estimated value of the moisture content at a point in time after the residence time has elapsed from the measurement time of the measurement data, based on the measurement data acquired by the acquisition unit 101 during the operation period of the dewatering machine 9 (S21). Next, the update unit 103 updates the first estimation model 111 using the set of the above measurement data, the estimated value of the moisture content, and the actual measured value of the moisture content of the dewatered cake measured by the operator as training data (S22). After that, the update process is terminated.

[0046] [Variation] The entity executing each process described in the above-described embodiment is arbitrary and is not limited to the examples given above. For example, each step of the moisture content estimation method shown in Figure 3 can be divided among multiple information processing devices. In other words, the moisture content estimation method may be executed by one information processing device 1 or by multiple information processing devices.

[0047] [Additional Notes] In the embodiment described above, the explanatory variables of the first estimation model 111 include estimates of moisture content using the second estimation model 112, but they may also include estimates of moisture content using other models. In this way, multiple estimates using different models may be used as explanatory variables for the first estimation model 111.

[0048] Furthermore, although the above-described embodiment does not update the second estimation model 112, it is not limited to this. For example, if unusual operating conditions occur during the actual operating period, the acquisition unit 101 may store the acquired measurement data and use the stored measurement data to update the second estimation model 112.

[0049] Furthermore, in the above-described embodiment, a trial run period is used as an operating period in which the operating conditions differ from those of the actual operating period. However, the above-described different operating period is not limited to the trial run period and can be any operating period in which measurement data can be obtained under a wide variety of operating conditions, including the operating condition range of the actual operating period. For example, the test run period performed after maintenance of the dewatering machine 9 may be used as the above-described different operating period.

[0050] Furthermore, in the above-described embodiment, the acquisition unit 101 needs to acquire measurement data related to the dehydrator 9, but does not need to acquire measurement data related to the flocculator 5. Therefore, a dehydrator 9 equipped with the information processing device 1 with the above configuration can achieve the above-described effects.

[0051] [Examples] An embodiment and a comparative example of the information processing device 1 with the above configuration will be described with reference to Figures 5 to 8.

[0052] (Example 1) As described above, the first estimation model 111 in this embodiment is a model that estimates the moisture content at the time of measurement using measurement data at the time of measurement and an estimated value of the moisture content at the time of measurement calculated from the measurement data using the second estimation model 112. The first estimation model 111 is trained using the measurement data acquired during the most recent predetermined period in the actual operation period. The second estimation model 112 is an estimation model trained using the measurement data acquired during the trial operation period. Therefore, the second estimation model 112 is not updated.

[0053] Figure 5 is a table showing the combinations of explanatory and dependent variables of the first estimated model 111 in this embodiment. In the example of Figure 5, the measurement data at the time of measurement that will be the explanatory variables of the first estimated model 111 are the operating time of the dewatering machine 9, the rotational speed of the screw of the dewatering machine 9, the drive current value (or drive torque) of the screw, the back pressure from the back pressure plate that compresses the liquid and is provided at the dewatering cake discharge port 95 of the dewatering machine 9 (back pressure of the dewatering machine 9), and the opening degree between the back pressure plate and the dewatering cake discharge port 95 (back pressure opening degree of the dewatering machine 9).

[0054] Figure 6 is a table showing the combinations of explanatory and dependent variables for the second estimation model 112. The second estimation model 112 shown in Figure 6 differs from the first estimation model 111 shown in Figure 5 in that the estimated values ​​from the second estimation model 112 are omitted, but otherwise they are the same. In other words, in this embodiment, the measured data included in the explanatory variables of the first estimation model 111 and the measured data included in the explanatory variables of the second estimation model 112 are the same.

[0055] Figure 7 is a graph showing the variability between the moisture content estimates by the first estimation model 111 in Example 1 and the actual moisture content measurements taken by the workers. In the graph in Figure 7, the first estimation model 111 was trained using pairs of measurement data and actual moisture content values ​​from the most recent three-month period of actual operation as training data, and was further updated with new training data during periods of downtime. In addition, the graph in Figure 7 uses measurement data and actual moisture content values ​​from the seven-month period of actual operation since the first estimation model 111 was trained. In the graph shown in Figure 7, the mean absolute error (MAE) was 0.73%.

[0056] (Comparative Example 1) In Comparative Example 1, a comparative model is used in which the estimates from the second estimation model are omitted from the explanatory and dependent variables shown in Figure 5. Therefore, the second estimation model 112 is not used in Comparative Example 1. Furthermore, the comparative model was trained using pairs of measurement data and actual moisture content values ​​from the most recent three-month period of actual operation as training data. Moreover, the comparative model is not updated with new training data.

[0057] The upper and middle sections of Figure 8 are graphs showing the variability between the moisture content estimates from the above-described comparison model for Comparative Example 1 and the actual moisture content measurements taken by the workers. The upper graph in Figure 8 uses measurement data and actual moisture content values ​​from a two-month period of actual operation after the comparison model was trained. The middle graph in Figure 8 uses measurement data and actual moisture content values ​​from a seven-month period of actual operation after the comparison model was trained.

[0058] In the upper graph of Figure 8, the MAE was 0.73%, which was similar to that of Example 1. On the other hand, in the middle graph of Figure 8, the MAE was 1.95%. From this, it can be understood that the moisture content estimation of Comparative Example 1 was similar in accuracy to that of Example 1 immediately after training, but the accuracy deteriorated over time.

[0059] (Comparative Example 2) Comparative Example 2 differs from Comparative Example 1 in that the comparison model is updated with new training data during the shutdown period; otherwise, it is the same. The lower part of Figure 8 is a graph showing the variability of the moisture content estimates by the comparison model in Comparative Example 2 compared to the actual moisture content measurements taken by the workers. The graph in the lower part of Figure 8 uses measurement data and actual moisture content values ​​from a 7-month period of actual operation after the comparison model was trained.

[0060] In the lower graph of Figure 8, the MAE was 0.80%. From this, it can be understood that the deterioration in accuracy of the moisture content estimate for Comparative Example 2 over time is less than that of Comparative Example 1, but is greater than that of Example 1. From these comparison results, the moisture content estimate for Example 1 is This demonstrates that the accuracy is well maintained even as time passes.

[0061] [Embodiment 2] Other embodiments of the present invention will be described with reference to Figures 9 to 12. The control system 100 of this embodiment differs from the control system 100 shown in Figures 1 to 4 in the configuration of the information processing device 1, but the other configurations are the same.

[0062] Figure 9 is a block diagram showing an example of the main components of the information processing device 1 of this embodiment. The information processing device 1 shown in Figure 9 differs from the information processing device 1 shown in Figure 1 in that the control unit 10 includes an acquisition unit 104, an estimation unit 105, an update unit 106, and a prediction unit 107 instead of the acquisition unit 101, estimation unit 102, and update unit 103, and the storage unit 11 further includes a prediction model 113. The other components are the same. Details of the prediction model 113 will be explained later in the section "About the Prediction Model".

[0063] The acquisition unit 104 acquires at least one of the following: measurement data relating to the liquid supplied to the coagulation tank 51, measurement data relating to the chemicals supplied to the coagulation tank 51, measurement data relating to the liquid inside the coagulation tank 51, measurement data relating to the operation of the coagulation tank 51, and measurement data relating to the operation of the dewatering machine 9. Details of each measurement data will be explained later in the section "About the Prediction Model".

[0064] The estimation unit 105 differs from the estimation unit 102 shown in Figure 1 in that it further sends the estimated value of the moisture content of the dewatered cake, estimated using the first estimation model 111, to the update unit 106; otherwise, its configuration is the same.

[0065] The prediction unit 107 uses the prediction model 113 stored in the memory unit 11 to predict the moisture content of the dewatered cake from the measurement data acquired by the acquisition unit 104. This dewatered cake is the cake discharged from the dewatering machine 9 at the point when the retention time for the coagulated sludge has elapsed in the dewatering machine 9 from the time of measurement of the above measurement data (hereinafter referred to as the "elapsed time"). The prediction unit 107 performs a prediction using the prediction model 113 each time the acquisition unit 104 acquires new measurement data. As a result, the prediction unit 107 can predict the moisture content of the dewatered cake in real time, that is, continuously at short time intervals (for example, every minute). The specific method for calculating the moisture content of the dewatered cake will be explained later in the section "Method for predicting moisture content using a prediction model".

[0066] [About the predictive model] The prediction model 113 is a prediction model that has been trained with the measurement data at the time of measurement acquired by the acquisition unit 104 as the explanatory variable, and the moisture content of the dewatered cake discharged from the dewaterer 9 at the time of elapsed time (hereinafter referred to as "moisture content at the time of elapsed time") as the dependent variable. In this embodiment, the moisture content, which is the dependent variable, is an estimated value by the first estimation model 111. The measurement data acquired by the acquisition unit 104 and used for prediction by the prediction model 113 includes the following.

[0067] (1) Measurement data relating to the sludge supplied to the coagulation tank 51. This measurement data includes, for example, the supply flow rate of the sludge per unit time and at least one of the sludge concentration, and is measured before the sludge inlet 55.

[0068] (2) Measurement data relating to the chemical supplied to the coagulation tank 51. This measurement data is, for example, the supply flow rate of the chemical per unit time, and is measured before the chemical inlet 56.

[0069] (3) Measurement data concerning the sludge in the coagulation tank 51. The measurement data is the average concentration of the flocs. This value is at least one of the lightness value and the average unit area of ​​the gaps between flocs, and is obtained by image processing of a still image taken with the imaging device 72. The above average lightness value serves as an indicator of the sludge color tone (lightness / darkness). The above average unit area serves as an indicator of the floc diameter.

[0070] (4) Measurement data relating to the operation of the coagulation tank 51. This measurement data is the rotation speed of the stirring blade in the coagulation tank 51 and is obtained from the control device 3.

[0071] (5) Measurement data relating to the operation of the dewatering machine 9. This measurement data is obtained from the control device 3 and includes at least one of the following: the operating time of the dewatering machine 9, the rotational speed of the screw of the dewatering machine 9, the flow rate of flocculated sludge supplied to the dewatering machine 9 per unit time, and the input pressure of the flocculated sludge fed into the dewatering machine 9.

[0072] In this embodiment, various measurement data can be used as shown in (1) to (5) above. Therefore, the moisture content of the dewatered cake can be predicted with high accuracy using multifaceted explanatory variables. Among the measurement data shown in (1) to (5) above, the measurement data that contributes most to the prediction accuracy of the prediction model 113 are the operating time, screw rotation speed, and supply flow rate shown in (5) above, and the average density value of the flocs shown in (3) above.

[0073] [Methods for training predictive models] Next, we will explain the method for learning moisture content using the prediction model 113. As described above, the prediction model 113 is a model that uses measurement data at the time of measurement to predict the moisture content at a future time point beyond that measurement point.

[0074] Figure 10 is a model diagram illustrating the concept of creating training data for the prediction model 113. The horizontal axis of Figure 10 represents operating time (minutes). The vertical axis of Figure 10 represents the measured value of each measurement data, shown on an arbitrary scale for each measurement data.

[0075] The explanatory variables for the prediction model 113, as shown in the upper part of Figure 10, are the supply flow rate of liquid (sludge) per unit time, the concentration of liquid (sludge), the average density of flocs, the average unit area of ​​gaps between flocs, the rotation speed of the agitator blades in the coagulation tank 51, the operating time of the dewatering machine 9, the rotation speed of the screw of the dewatering machine 9, the input pressure of the liquid (coagulated sludge) fed into the dewatering machine 9, and the ratio of the supply flow rate of the chemical agent per unit time to the supply flow rate of liquid (sludge).

[0076] Furthermore, the dependent variable of the prediction model 113 is the moisture content at each elapsed time. If actual measurements by workers are used as this moisture content, there is a limit to the number of times a worker can take measurements in a day, as shown by the white circles in the lower part of Figure 10. Therefore, the number of training data, which are combinations of explanatory and dependent variables, is limited, making it difficult to create a highly accurate prediction model.

[0077] Therefore, in this embodiment, instead of using the measured values ​​by the worker as the target variable for the training data, the estimated values ​​from the first estimation model 111 described above are used. As a result, as shown by the gray circles in the lower part of Figure 10, the number of target variables is no longer limited by the number of times the worker can take measurements, so the amount of training data can be increased, and a highly accurate prediction model 113 can be created.

[0078] Incidentally, the estimated moisture content at the above time points is calculated from the measurement data at the above time points using the first estimation model 111 and the second estimation model 112. Therefore, the training of the prediction model 113 requires the measurement data at the measurement points, which are the explanatory variables of the prediction model 113, and the first The explanatory variables for the estimated model 111 and the second estimated model 112, which are measured data at elapsed time points, will be used.

[0079] [Method for predicting moisture content using a predictive model] The prediction unit 107 uses the trained prediction model 113 to predict the moisture content of the dewatered cake. Specifically, the prediction unit 107 inputs at least one of the explanatory variables—measurement data on the liquid supplied to the coagulation tank 51, measurement data on the chemicals supplied to the coagulation tank 51, measurement data on the liquid inside the coagulation tank 51, measurement data on the operation of the coagulation tank 51, and measurement data on the operation of the dewatering machine 9—into the prediction model 113 to predict the objective variable, which is the moisture content of the dewatered cake at the completion of dewatering.

[0080] As described above, the information processing device 1 of this embodiment, compared to the information processing device 1 shown in Figure 1, further includes a prediction unit 107 that uses a prediction model 113 to predict the moisture content of the dewatered cake discharged from the dewatering machine 9 at a point in time when the liquid has remained in the dewatering machine 9 for a period of time from the time of measurement of the measurement data to the point in time when the liquid has remained in the dewatering machine 9, based on the measurement data acquired by the acquisition unit 104. The prediction model 113 is a prediction model that has been learned with the measurement data acquired by the acquisition unit 104 as the explanatory variable and the estimated value by the first estimation model 111 for the moisture content of the dewatered cake discharged from the dewatering machine 9 at a point in time when the residence time has elapsed from the time of measurement of the measurement data as the objective variable.

[0081] According to the above configuration, the first estimation model 111 has the same measurement timing (at the completion of dewatering) for the explanatory variable, the measurement data related to the operation of the dewatering machine 9, and the same measurement timing (at the completion of dewatering) for the objective variable, the moisture content of the dewatered cake discharged from the dewatering machine 9. Therefore, even with a small amount of training data including measurement data and moisture content, the accuracy of estimation using the trained first estimation model 111 is good. Accordingly, the first estimation model 111 can be trained without increasing the workload on the operator in manual analysis of moisture content. Furthermore, the first estimation model 111 can be used to calculate estimated moisture content values ​​for as many measurements as there are measurements of the measurement data.

[0082] On the other hand, with respect to the prediction model 113, an estimated value of the moisture content at the elapsed time (when dewatering is completed) can be obtained using the first estimation model 111 from measurement data related to the operation of the dewatering machine 9, measured from the measurement time (when the liquid is supplied) to the elapsed time (when dewatering is completed) of the measurement data acquired by the acquisition unit 104. Then, training data can be created for the number of times the measurement data acquired by the acquisition unit 104 is measured, with the obtained estimated value as the dependent variable and the measurement data acquired by the acquisition unit 104 at the measurement time as the independent variables. Accordingly, using the prediction model 113 trained with the training data, the moisture content at the elapsed time (when the residence time has elapsed) from the measurement data acquired by the acquisition unit 104 can be predicted with high accuracy.

[0083] [Regarding the update section] The update unit 106 differs from the update unit 103 shown in Figure 1 in that it further performs the process of updating the prediction model 113 during the shutdown period, but is otherwise the same. Specifically, the update unit 106 updates the prediction model 113 during the shutdown period using as training data a pair of measurement data acquired by the acquisition unit 101 during the operation period of the coagulation tank 51 and the dewatering machine 9, and the estimated value by the first estimation model 111 for the moisture content of the dewatered cake discharged from the dewatering machine 9 at a point in time after the residence time has elapsed from the measurement time of the measurement data.

[0084] Furthermore, it is desirable to update the prediction model 113 using measurement data from the most recent operating period. In this case, the predictions made using the prediction model 113 can be adapted to the latest conditions of the coagulation tank 51 and the dewatering machine 9.

[0085] [Predictive processing] Figure 11 is a flowchart showing an example of the moisture content prediction process (moisture content prediction method) of the dewatered cake in the information processing device 1 with the above configuration. This prediction process is performed during the operating period, as described above. As shown in Figure 11, first, the acquisition unit 104 collects (acquires) various measurement data (S31, acquisition step). Next, the prediction unit 107 uses the prediction model 113 to calculate a predicted value of the moisture content at a point in time after the residence time has elapsed from the measurement point of the measurement data (S32, prediction step). After that, the process returns to step S31 and the above operation is repeated.

[0086] [Update process] Figure 12 is a flowchart showing an example of the update process for the first estimation model 111 and the prediction model 113 in the information processing device 1. This update process is performed after each downtime, as described above. The update process shown in Figure 12 is performed after the update processes S21 and S22 shown in Figure 4.

[0087] Specifically, the update unit 106 uses the first estimation model 111 and the second estimation model 112 to calculate an estimated value of the moisture content during the operating period from the measurement data during the operating period (S41). Next, the update unit 106 updates the prediction model 113 using the pair of the measurement data and the estimated value at a point in time after the residence time has elapsed from the measurement time as training data (S42). After that, the update process is terminated.

[0088] [Examples of implementation using software] The function of the information processing device 1 (hereinafter referred to as "the device") is a program that causes the device to function as a computer, and can be realized by a program (moisture content estimation program) that causes each control block of the device (particularly each part included in the control unit 10) to function as a computer.

[0089] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0090] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0091] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0092] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Explanation of Symbols]

[0093] 1. Information Processing Device 3. Control device 5. Flocculator 9 Dehydrator 10 Control Unit 11 Storage section 12 Communications Department 13 Input section 14 Output section 51 Coagulation tank 91 Outer shell screen 92 Screw 94 Liquid outlet 95 Dehydrated cake outlet 100 control systems 101, 104 Acquisition Department 102, 105 Estimation part 103, 106 Update section 107 Prediction Section 111 First Estimate Model 112 Second Estimated Model 113 Predictive Models

Claims

1. An acquisition unit acquires measurement data related to the operation of a dewatering machine that dewaters a liquid while transporting the liquid discharged from a coagulation tank to which a chemical agent that coagulates the suspended solids is added, The system includes an estimation unit that uses a first estimation model and a second estimation model to estimate the moisture content of the dewatered cake discharged from the dewatering machine at the time of measurement of the measurement data, based on the measurement data acquired by the acquisition unit. The second estimation model is an estimation model that has been learned using the measurement data acquired by the acquisition unit as explanatory variables during an operating period different from the actual operating period of the dewatering machine, and during an operating period in which the machine is operated under a wider range of operating conditions than those during the actual operating period, and the measured value of the moisture content at the time of measurement of the measurement data as the dependent variable. The first estimation model is an estimation model that has been learned with the following variables as explanatory variables: the measurement data acquired by the acquisition unit from a predetermined period prior to the learning time in the actual operating period included in the operational shutdown period when the coagulation tank and the dewatering machine are stopped, up to the learning time; and the estimated value of the moisture content estimated from the measurement data using the second estimation model; and the moisture content at the time of measurement of the measurement data as the dependent variable. The estimation unit estimates the moisture content of the measurement data at the time of measurement using the latest first estimation model, based on the measurement data and the estimated value of the moisture content of the measurement data at the time of measurement, which is estimated using the second estimation model from the measurement data acquired by the acquisition unit. Information processing device.

2. The aforementioned dehydrator is a screw press type dehydrator, The information processing apparatus according to claim 1, wherein the measurement data is at least one of the following: the operating time of the dehydrator, the rotational speed of the screw of the dehydrator, the drive current value or torque value of the screw, the back pressure by a back pressure plate provided in the discharge section of the dehydrator for compressing the liquid, and the opening degree between the back pressure plate and the discharge section.

3. The information processing apparatus according to claim 1, wherein the estimation unit performs the estimation each time the acquisition unit acquires new measurement data.

4. The system further includes a prediction unit that uses a prediction model to predict the moisture content of the dewatered cake discharged from the dewatering machine at a time when the liquid has remained in the dewatering machine for a period of time from the time of measurement of the measurement data to the time of residence of the liquid in the dewatering machine, based on the measurement data acquired by the acquisition unit. The information processing apparatus according to claim 1, wherein the prediction model is a prediction model learned with the measurement data acquired by the acquisition unit as explanatory variables and the estimated value by the first estimation model for the moisture content at the time the residence time has elapsed from the time of measurement of the measurement data as the objective variable.

5. A dehydrator equipped with an information processing device according to any one of claims 1 to 4.

6. A method for estimating moisture content, which is performed by one or more information processing devices, An acquisition step of acquiring measurement data related to the operation of a dewatering machine that dewaters a liquid while transporting the liquid discharged from a coagulation tank to which a chemical agent that coagulates the suspended solids is added, The process includes an estimation step in which, using a first estimation model and a second estimation model, the moisture content of the dewatered cake discharged from the dewatering machine at the time of measurement of the measurement data is estimated from the measurement data acquired in the acquisition step, The second estimation model is an estimation model that has been learned using the measurement data acquired in the acquisition step as an explanatory variable during an operating period different from the actual operating period of the dewatering machine, and during an operating period in which the machine is operated under a wider range of operating conditions than those during the actual operating period, and the measured value of the moisture content at the time of measurement of the measurement data as the dependent variable. The first estimation model is an estimation model that has been learned with the following variables as explanatory variables: the measurement data acquired in the acquisition step from a predetermined period before the learning time to the learning time in the actual operating period included in a predetermined period prior to the shutdown period when the coagulation tank and the dewatering machine are stopped, and the estimated value of the moisture content estimated from the measurement data using the second estimation model, and the moisture content at the time of measurement of the measurement data as the dependent variable. The estimation step involves estimating the moisture content of the measurement data at the time of measurement using the latest first estimation model, based on the measurement data and the estimated value of the moisture content of the measurement data at the time of measurement, which was estimated using the second estimation model from the measurement data acquired in the acquisition step. Moisture content estimation method.

7. A moisture content estimation program for causing a computer to function as an information processing device according to claim 1, wherein the computer functions as the acquisition unit and the estimation unit.

Images