JP7101912B1 - Water quality management equipment, water quality management methods, and water quality management programs - Google Patents

Abstract

Provided are a water quality control device, a water quality control method, and a water quality control program capable of preventing the concentration of residual chlorine in seawater from exceeding a standard value at a seawater outlet. A water quality control device for a seawater utilization plant, comprising: an attribute value acquisition unit for acquiring attribute values of seawater flowing through a condenser installed in the seawater utilization plant; A concentration prediction unit that predicts the residual chlorine concentration at the outlet of the discharge channel that discharges to the sea, and the required amount of neutralizing agent for residual chlorine to be injected into the discharge channel based on the predicted residual chlorine concentration A required amount calculation unit and a neutralizing agent injection unit for injecting the required amount of the neutralizing agent into the discharge channel are provided.

Description

The present invention relates to a water quality management device, a water quality management method, and a water quality management program. More specifically, it relates to a water quality management device, a water quality management method, and a water quality management program used in a seawater utilization plant such as a power plant.

Generates seawater electrolytic chlorine (sodium hypochlorite) as a countermeasure against barnacles, mussels, and other attached organisms that adhere to the seawater system of seawater utilization plants such as thermal power plants and nuclear power plants, and biofilms. Therefore, the technique of injecting into the water intake is widely practiced.

For example, Patent Document 1 produces sodium hypochlorite by electrolyzing natural seawater, and injects an electrolytic solution containing the sodium hypochlorite into the intake of seawater to prevent the adhesion of marine organisms. The technology used for is disclosed.

Japanese Patent No. 4923259

When injecting seawater electrolytic chlorine into the water outlet, it is necessary to inject it so as not to exceed the standard value of the residual chlorine concentration at the seawater outlet, but the residual chlorine concentration after injecting seawater electrolytic chlorine is the water temperature. Since the decay rate differs depending on the water quality and water quality, if an attempt is made to maintain a residual chlorine concentration that is effective in preventing attached organisms, the standard value may be temporarily exceeded at the outlet. Therefore, there is a demand for a technique for preventing the residual chlorine concentration from exceeding the standard value by injecting an appropriate amount of the neutralizing agent at an appropriate timing.

The present invention has been made in view of the above problems, and is a water quality management device, a water quality management method, and a water quality management method capable of preventing the reference value of the residual chlorine concentration in seawater from being exceeded at the outlet of seawater. The purpose is to provide a water quality management program.

The present invention is a water quality management device for a seawater utilization plant, which is based on an attribute value acquisition unit for acquiring an attribute value of seawater, which circulates a water condenser installed in the seawater utilization plant, and the attribute value. Then, a concentration predictor that predicts the residual chlorine concentration at the outlet of the discharge channel that discharges the seawater from the water concentrator to the sea, and a residue that is injected into the drainage channel based on the predicted residual chlorine concentration. The present invention relates to a water quality management device including a required amount calculation unit for calculating a required amount of a chlorine neutralizer and a neutralizer injection unit for injecting the required amount of the neutralizer into the flood bypass.

The attribute value includes the residual chlorine concentration of seawater at the intake of the condenser, the temperature of seawater at the outlet of the condenser, and the flow time of seawater from the intake to the discharge. It is preferable that the prediction unit predicts the residual chlorine concentration at the outlet by applying the attribute value to the Allenius equation.

Further, in the water quality management device, the concentration prediction unit inputs historical data of attribute values including the residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the inlet of the water recovery device. The input data acquisition unit that acquires the input data as a label, the label acquisition unit that acquires the history data of the residual chlorine concentration at the outlet as a label, and the pair of the input data and the label as learning data, the residual chlorine at the outlet. A learning unit that builds a learning model that estimates the concentration, and an estimation value generation unit that generates an estimated value of the residual chlorine concentration by applying a new attribute value to the learning model after the construction of the learning model. It is preferable to prepare.

Further, in the water quality management device, it is preferable that the learning unit constructs the learning model using a random forest.

Further, in the water quality management device, it is preferable that the learning unit constructs the learning model by using the generalized addition method (GAM method).

Further, the present invention is a water quality management method for a seawater utilization plant, in which an attribute value acquisition step for acquiring an attribute value of seawater that circulates a water concentrator installed in the seawater utilization plant and an attribute value acquisition step are used. Based on the concentration prediction step for predicting the residual chlorine concentration at the outlet of the discharge channel that discharges the seawater from the water concentrator to the sea, and the injection into the drainage channel based on the predicted residual chlorine concentration. The present invention relates to a water quality management method comprising a required amount calculation step for calculating a required amount of a neutralizing agent for residual chlorine, and a neutralizing agent injection step for injecting the required amount of the neutralizing agent into the flood channel.

Further, the present invention is a water quality management program for a seawater utilization plant, in which an attribute value acquisition step for acquiring an attribute value of seawater, which is distributed through a flood feeder installed in the seawater utilization plant, and the attribute value are used. Based on the concentration prediction step for predicting the residual chlorine concentration at the outlet of the drainage channel that discharges the seawater from the water concentrator to the sea, and the injection into the drainage channel based on the predicted residual chlorine concentration. Water quality management for causing a computer to perform a required amount calculation step for calculating a required amount of a neutralizing agent for residual chlorine and a neutralizing agent injection step for injecting the required amount of the neutralizing agent into the flood bypass. Regarding the program.

According to the present invention, it is possible to prevent the reference value of the residual chlorine concentration in seawater from being exceeded at the outlet of seawater.

It is an overall block diagram of the seawater utilization plant which concerns on embodiment of this invention. It is a functional block diagram of the water quality management apparatus which concerns on embodiment of this invention. It is a flowchart which shows the operation of the water quality management apparatus which concerns on embodiment of this invention. It is a functional block diagram of the concentration prediction part included in the water quality management apparatus which concerns on embodiment of this invention. It is a flowchart which shows the operation of the water quality management apparatus which concerns on embodiment of this invention. It is a flowchart which shows the operation of the water quality management apparatus which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the Arrhenius equation which concerns on embodiment of this invention. It is a figure which shows the comparison between the original data by the random forest which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the random forest which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result. It is a figure which shows the comparison between the original data by the generalized addition (GAM) which concerns on embodiment of this invention, and the prediction result.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[1 Outline of the invention]
FIG. 1 is an overall configuration diagram of a seawater utilization plant 100 including a water quality management device 1 and a condenser 2 according to the present embodiment. In the seawater utilization plant 100, the condenser 2 takes in seawater as cooling water from the sea 3A through the intake channel 21, and after cooling the steam turbine and the like, the cooling water is discharged from the outlet 221 of the condenser 2. Water is discharged to the sea 3B by the water channel 22. At the chlorine injection point of the intake channel 21, an electrolytic solution of seawater containing sodium hypochlorite is injected.

The water quality management device 1 predicts the residual chlorine concentration at the discharge port 222 of the flood bypass 22 based on the attribute value of the seawater flowing through the water recovery device 2, and compares it with the residual chlorine concentration of the seawater at the water recovery device inlet 211. After calculating the reduced amount, the required amount of the neutralizing agent is calculated based on the reduced amount, and the required amount of the neutralizing agent is injected into the flood channel 22.

[2 First Embodiment]
Hereinafter, the water quality management device 1 according to the first embodiment of the present invention will be described with reference to FIGS. 2 and 3.

[2.1 Configuration of Embodiment]
FIG. 2 is a functional block diagram of the water quality management device 1. The water quality management device 1 includes a control unit 11, a neutralizing agent injection unit 12, and a storage unit 13.

The control unit 11 is a part that controls the entire water quality management device 1, and by appropriately reading and executing various programs from a storage area such as a ROM, RAM, flash memory, or a hard disk (HDD), in the present embodiment. It realizes various functions. The control unit 11 may be a CPU. The control unit 11 includes an attribute value acquisition unit 111, a concentration prediction unit 112, and a required amount calculation unit 113.

In addition, the control unit 11 includes general functional blocks such as a functional block for controlling the entire water quality management device 1 and a functional block for performing communication. However, since these general functional blocks are well known to those skilled in the art, illustration and description thereof will be omitted.

The attribute value acquisition unit 111 acquires the attribute value of the seawater flowing through the condenser 2. More specifically, for example, seawater is pumped from the intake channel 21 by a pump (not shown) installed outside the water quality control device 1, and the pumped seawater is used as the water quality installed outside the water quality control device 1. Analyze with an analyzer (not shown). The attribute value acquisition unit 111 acquires the attribute value related to the water quality of the seawater analyzed by the water quality analyzer. Further, the attribute value acquisition unit 111 acquires, as attribute values, the water temperature of the seawater at the outlet 221 of the condenser 2, the flow time of the seawater from the condenser inlet 211 to the discharge port 222, and the like.

Here, as the attribute values related to the water quality, for example, in addition to the residual chlorine concentration and water temperature of the seawater at the return device inlet 211, the organic substance concentration, the amount of salt contained in the seawater, the pH, and the ORP are also used for the purpose of improving the accuracy. (Redox potential), including.

The concentration prediction unit 112 predicts the residual chlorine concentration at the seawater discharge port 222 based on the attribute value acquired by the attribute value acquisition unit 111. In particular, in the present embodiment, the concentration prediction unit 112 has attribute values such as the residual chlorine concentration of seawater at the inlet 211 of the condenser 2, the temperature of the seawater at the outlet 221 of the condenser 2, and the inlet 211 of the condenser 2. By applying the flow time of seawater from to the discharge port 222 to the discharge port 222 in the Arenius equation, the residual chlorine concentration at the discharge port 222 of the seawater is estimated.

Here, the "Arrhenius equation" is an equation that predicts the rate of a chemical reaction at a certain temperature, and the reaction constant k indicating the reaction rate is the temperature T as shown in the following equation 1. It is an equation showing that it becomes larger when the activation energy Ea is high and the activation energy _Ea is low.

Note that A is a constant (frequency factor) irrelevant to temperature, E _a is activation energy per 1 mol, R is a gas constant, and T is an absolute temperature. Here, the "frequency factor" is a factor that represents the number of collisions between molecules in a bimolecular reaction.

The required amount calculation unit 113 calculates the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel 22 based on the residual chlorine concentration predicted by the concentration prediction unit 112. More specifically, the required amount calculation unit 113 calculates the required amount of the neutralizing agent based on the amount of decrease in the estimated residual chlorine concentration at the outlet 222 as compared with the residual chlorine concentration at the intake port 211.

Here, the neutralizing agent may be an existing agent capable of rapidly neutralizing residual chlorine, such as 35% hydrogen peroxide solution, sodium sulfite, or sodium thiosulfate. The reaction by-products when 35% hydrogen peroxide solution is used are oxygen, water and chlorine ions, and the reaction by-products when sodium sulfite is used are sulfate ions and chlorine ions. All of these are abundant in seawater.

The neutralizing agent injection unit 12 injects the required amount of the neutralizing agent calculated by the required amount calculation unit 113 into the flood channel 22. In particular, it is preferable to automatically control the injection of the neutralizing agent into the drainage channel 22 by the neutralizing agent injection unit 12.

The storage unit 13 stores the attribute value acquired by the attribute value acquisition unit 111, the residual chlorine concentration predicted by the concentration prediction unit 112, and the required amount of the neutralizing agent calculated by the required amount calculation unit 113.

[2.2 Operation of the embodiment]
FIG. 3 is a flowchart showing the operation of the water quality management device 1.

In step S1, the attribute value acquisition unit 111 acquires the attribute value of the water quality at the intake port 211 of the seawater, the water temperature of the seawater at the outlet 221 of the condenser 2, the flow time of the seawater from the inlet 211 to the discharge port 222, and the like. do.

In step S2, the concentration prediction unit 112 predicts the residual chlorine concentration at the seawater discharge port 222 based on the attribute value acquired by the attribute value acquisition unit 111.

In step S3, the required amount calculation unit 113 calculates the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel 22 based on the residual chlorine concentration predicted by the concentration prediction unit 112.

In step S4, the neutralizing agent injection unit 12 injects the required amount of the neutralizing agent calculated by the required amount calculation unit 113 into the flood channel 22.

[2. Effects of the embodiment]
The water quality management device 1 according to the present embodiment is a water quality management device 1 for the seawater utilization plant 100, and is an attribute value for acquiring the attribute value of the seawater flowing through the water recovery device 2 installed in the seawater utilization plant 100. Based on the acquisition unit 111, the concentration prediction unit 112 that predicts the residual chlorine concentration at the discharge port 222 of the discharge channel 22 that discharges seawater from the water condensing device 2 to the sea based on the attribute value, and the estimated residual chlorine concentration. The required amount calculation unit 113 for calculating the required amount of the neutralizing agent for residual chlorine to be injected into the discharge channel 22, and the neutralizing agent injection unit 12 for injecting the required amount of the neutralizing agent into the discharge channel 22. To be equipped.

This makes it possible to prevent the reference value of the residual chlorine concentration in seawater from being exceeded at the outlet of seawater. In particular, by injecting an appropriate amount of the neutralizing agent before the reference value of the residual chlorine concentration in the outlet 222 is exceeded, it is possible to prevent the reference value from being exceeded. This makes it possible to adjust the base value of the electrolytic chlorine injection concentration to a higher level, and by more effectively suppressing the adhesion of attached organisms and biofilms that hinder power generation, the heat exchange efficiency of the condenser 2 Is improved, and a great cost reduction effect can be obtained.

Further, in the water quality management device 1, the above attribute values are the residual chlorine concentration of the seawater at the inlet 211 of the condenser 2, the water temperature of the seawater at the outlet 221 of the condenser 2, and the seawater from the outlet 221 to the discharge port 222. Including the flow time, the concentration predictor predicts the residual chlorine concentration at the outlet 222 by applying the above attribute value to the Arenius equation.

As a result, by using the Arrhenius equation, it is possible to prevent the residual chlorine concentration in seawater from exceeding the standard value.

[3 Second Embodiment]
Hereinafter, the water quality management device 1A according to the second embodiment of the present invention will be described with reference to FIGS. 4 to 6. In the following, for the sake of simplification of the description, the difference between the water quality management device 1A and the water quality management device 1 will be mainly described.

[Structure of 3.1 Embodiment]
The basic configuration of the water quality management device 1A is the same as that of the water quality management device 1 shown in FIG. However, the water quality management device 1A includes a concentration prediction unit 112A instead of the concentration prediction unit 112 included in the water quality management device 1. The concentration prediction unit 112 predicts the residual chlorine concentration at the outlet 222 mainly by using the Arrhenius equation, while the concentration prediction unit 112A predicts the residual chlorine concentration at the outlet 222 by machine learning.

FIG. 4 is a functional block diagram of the concentration prediction unit 112A. The density prediction unit 112A includes an input data acquisition unit 114, a label acquisition unit 115, a learning unit 116, and an estimated value generation unit 117.

The input data acquisition unit 114 uses the historical data of attribute values including the residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the intake port 211 from the storage unit 13 as input data for machine learning. Get as.

The label acquisition unit 115 acquires historical data of the residual chlorine concentration at the water discharge port 222 from the storage unit 13 as a label used for machine learning.

The learning unit 116 constructs a learning model for estimating the residual chlorine concentration at the outlet 222 by performing machine learning using a set of input data and a label as learning data, and stores the constructed learning model in the storage unit 13. ..

Here, when the set of the input data and the label is used as the training data, the flowing time of the seawater from the outlet 221 of the condenser 2 to the discharge port 222 is taken into consideration, and the condenser inlet data and the time after the flowing time are taken into consideration. Process so that it is paired with the outlet data. Further, for example, both the objective variable and the explanatory variable may be removed by regarding a negative value as an abnormal value.

Further, the machine learning executed by the learning unit 116 may be a random forest or a generalized addition method (GAM method). Here, "random forest" is one of the algorithms of machine learning, and it has a generalization ability by integrating a large number of weak learners learned by randomly sampled training data while using a decision tree as a weak learner. It is an algorithm that improves.
The "generalized addition method" is an algorithm that uses a model in which the linear predictor in the generalized linear model is the sum of non-linear functions. The non-linear functions at this time include a local regression function and a smoothing spline. , B spline, natural spline, etc. are used. Among them, an algorithm that uses a smoothing spline as a non-linear function is called a "GAM method".

The estimated value generation unit 117 acquires a learning model from the storage unit 13 after the learning model is constructed by the learning unit 116 and the constructed learning model is stored in the storage unit 13, and a new attribute value is obtained as the learning model. By applying to, an estimated value of the residual chlorine concentration at the outlet 222 is generated.

[3.2 Operation of the embodiment]
FIG. 5 is a flowchart showing the operation of the water quality management device 1A during machine learning.

In step S11, the input data acquisition unit 114 acquires historical data of attribute values including the residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), water temperature, and flow rate of seawater from the storage unit 13 as input data. ..

In step S12, the label acquisition unit 115 acquires historical data of the residual chlorine concentration at the outlet as a label.

In step S13, the learning unit 116 uses the set of the input data and the label as the learning data.

In step S14, the learning unit 116 performs machine learning using the learning data.

When the machine learning is completed in step S15 (S15: YES), the process proceeds to step S16. If the machine learning is not completed (S15: NO), the process proceeds to step S11.

In step S16, the learning unit 116 stores the constructed learning model in the storage unit 13.

FIG. 6 is a flowchart showing the operation of the water quality control device 1A at the time of injecting the neutralizing agent.

In step S21, the estimated value generation unit 117 acquires a learning model from the storage unit 13.

In step S22, the estimated value generation unit 117 acquires a new attribute value from the attribute value acquisition unit 111.

In step S23, the estimated value generation unit 117 generates an estimated value (predicted value) of the residual chlorine concentration at the outlet 222 by applying a new attribute value to the learning model.

In step S24, the required amount calculation unit 113 calculates the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel 22 based on the residual chlorine concentration predicted by the concentration prediction unit 112.

In step S25, the neutralizing agent injection unit 12 injects the required amount of the neutralizing agent calculated by the required amount calculation unit 113 into the flood channel 22.

[Effects of 3.3 embodiments]
In the water quality control device 1A, the concentration prediction unit 112A acquires historical data of attribute values including the residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the intake port 211 as input data. A learning model that estimates the residual chlorine concentration at the outlet using the acquisition unit 114, the label acquisition unit 115 that acquires the historical data of the residual chlorine concentration at the outlet 222 as a label, and the set of the input data and the label as learning data. It is provided with a learning unit 116 for constructing the learning model, and an estimation value generation unit 117 for generating an estimated value of the residual chlorine concentration by applying a new attribute value to the learning model after the construction of the learning model.

This makes it possible to inject the neutralizing agent into the flood channel 22 based on the predicted value of the residual chlorine concentration with higher accuracy.

Further, the learning unit 116 builds a learning model using a random forest.

This makes it possible to deal with a large number of explanatory variables and learn at high speed, and it is possible to calculate the importance (contribution) of the explanatory variables.

Further, the learning unit 116 constructs a learning model by using the generalized addition method.

As a result, by using a function having a complicated form, it is possible to explain something that is not a simple proportional relationship, and it is possible to improve the accuracy of prediction while maintaining the explanatory property of the linear model.

[4 Forecast data]
[4.1 Arrhenius equation]
(1) Analysis by data from June 29, 2018 to November 9, 2018 The measurement data of A power plant Unit 1 (maximum output 340 MW) from June 29, 2018 to November 9, 2018 was used.

(1-1) Relationship with condenser inlet concentration It is also known that the attenuation of the residual chlorine concentration is largely contributed by the initial concentration. Regarding the Arrhenius equation shown in (1-1), the residual chlorine concentration at the condenser inlet is 0.05 mg / L or more, 0.03 mg / L or more and less than 0.05 mg / L, and less than 0.03 mg / L for each power generation output. The Arrhenius equation in each case is shown in FIGS. 7A to 9C.

In each case of power generation output, the higher the residual chlorine concentration, the higher the coefficient of determination. The highest coefficient of determination was 0.589 when the power generation output was 200 MW or more and the residual chlorine concentration was 0.05 mg / L or more. When the residual chlorine concentration is less than 0.03, the coefficient of determination is less than 0.05 in any case of power generation output.

[4.2 Machine learning]
We examined the applicability of machine learning as a method for predicting the residual chlorine concentration at the outlet from the residual chlorine concentration at the condenser inlet.

(1) Data used For the data used for model construction and forecasting, the 1-minute values for June 29, 2018 to March 31, 2019 at Power Station A were used. The variables used and the data processing method are shown in the table below.

(2) Results by prediction model Using the processed data, prediction was performed using two prediction models, a random forest and a generalized additive model (GAM).

(2-1) Prediction result by random forest The data from June 29, 2018 to February 28, 2019 was used as training data, and the prediction was made from March 1, 2019 to March 31, 2019. 10A and 10B show a comparison between the original data and the prediction result.

The original data and the prediction result show similar behavior. Although it is within the range of 0.01 mg / L with respect to the original data, the coefficient of determination of the original data and the prediction result is as low as 0.096 due to the bias in the distribution of the error.

(2-2) Prediction results by generalized addition (GAM) Three types of predictions were made by changing the learning range and prediction range.

1) Prediction result 1
Similar to the prediction using the random forest, the data from June 29, 2018 to February 28, 2019 was used as the training data, and the prediction from March 1, 2019 to March 31, 2019 was made. 11A and 11B show a comparison between the original data and the prediction result.

Similar to the results from Random Forest, the original data and the predicted results behave similarly. Since it is within the range of 0.01 mg / L or less with respect to the original data and the error distribution is less biased, the coefficient of determination of the original data and the prediction result is 0.272, which is higher than the result by the random forest. There is.

2) Prediction result 2
The learning period and prediction period are different from the results shown in the previous section, and the data from June 29, 2018 to October 31, 2018 and December 1, 2018 to March 31, 2019 are used as training data, and 2018. Forecasts were made from November 1, 2018 to November 30, 2018. 12A and 12B show a comparison between the original data and the prediction result.

Similar to the prediction result 1, the original data and the prediction result show the same behavior. It was generally within the range of 0.01 mg / L with respect to the original data, and the coefficient of determination of the original data and the prediction result was 0.303.

3) Prediction result 3
The data from February 1, 2019 to February 14, 2019 was used as training data, and the forecast was made from February 15, 2019 to February 20, 2019. 13A and 13B show a comparison between the original data and the prediction result.

The behavior of the original data and the prediction result are almost the same, showing high reproducibility. It is within the range of about 0.005 mg / L with respect to the original data, and the coefficient of determination of the original data and the prediction result is as high as 0.505.

[4.3 Summary]
The coefficient of determination of the approximate curve by the Arenius equation is the highest when the power generation output is 200 MW or more and the residual chlorine concentration at the water return device inlet is 0.05 mg / L or more. It was confirmed that the coefficient of determination tends to decrease as both the power generation output and the residual chlorine concentration decrease. From this, it is considered that the prediction of the outlet concentration by the Arrhenius equation is more effective as the power generation output is higher and the condenser inlet concentration is higher.

As for the prediction of the outlet concentration by machine learning, as a result of verification using multiple models and conditions, the prediction accuracy that was generally practical was confirmed, and the applicability was shown.

The management method by the water quality management device 1 or 1A is realized by software. When realized by software, the programs constituting this software are installed in a computer (water quality management device 1 or 1A). In addition, these programs may be recorded on removable media and distributed to users, or may be distributed by being downloaded to a user's computer via a network. Further, these programs may be provided to the user's computer (water quality management device 1 or 1A) as a Web service via the network without being downloaded.

1, 1A Water quality control device 2 Condenser 11 Control unit 12 Neutralizer injection unit 13 Storage unit 21 Intake channel 22 Flood bypass 111 Attribute value acquisition unit 112, 112A Concentration prediction unit 113 Required amount calculation unit 114 Input data acquisition unit 115 Label acquisition unit 116 Learning unit 117 Estimated value generation unit

Claims (5)

A water quality management device for seawater utilization plants
The attribute value acquisition unit that acquires the attribute value of the seawater that circulates in the condenser installed in the seawater utilization plant, and the attribute value acquisition unit.
A concentration predictor that predicts the residual chlorine concentration at the outlet of the flood channel that discharges the seawater from the condenser to the sea based on the attribute value.
A required amount calculation unit that calculates the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel based on the predicted residual chlorine concentration.
A neutralizing agent injection unit for injecting the required amount of the neutralizing agent into the flood channel is provided.,
The concentration prediction unit
An input data acquisition unit that acquires historical data of attribute values including residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the inlet of the water condenser as input data.
A label acquisition unit that acquires historical data of the residual chlorine concentration at the outlet as a label,
A learning unit for constructing a learning model for estimating the residual chlorine concentration at the outlet, using the set of the input data and the label as learning data.
After constructing the learning model, it is provided with an estimated value generation unit that generates an estimated value of the residual chlorine concentration by applying a new attribute value to the learning model. Water quality management device.
The water quality management device according to claim 1 , wherein the learning unit constructs the learning model using a random forest. The water quality management device according to claim 1 , wherein the learning unit constructs the learning model by using generalized addition. A water quality management method for seawater utilization plants,
The attribute value acquisition step for acquiring the attribute value of the seawater flowing through the condenser installed in the seawater utilization plant, and the attribute value acquisition step.
A concentration prediction step for predicting the residual chlorine concentration at the outlet of the flood channel that discharges the seawater from the condenser to the sea based on the attribute value.
A required amount calculation step for calculating the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel based on the predicted residual chlorine concentration.
There is a neutralizing agent injection step of injecting the required amount of the neutralizing agent into the drainage channel.death,
The concentration prediction step,
An input data acquisition step for acquiring historical data of attribute values including residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the inlet of the water condenser as input data.
A label acquisition step for acquiring historical data of the residual chlorine concentration at the outlet as a label, and
A learning step for constructing a learning model for estimating the residual chlorine concentration at the outlet, using the set of the input data and the label as learning data.
After constructing the learning model, it has an estimated value generation step of generating an estimated value of the residual chlorine concentration by applying a new attribute value to the learning model. Water quality management method. A water quality management program for seawater utilization plants
The attribute value acquisition step for acquiring the attribute value of the seawater circulating in the condenser installed in the seawater utilization plant, and
A concentration prediction step for predicting the residual chlorine concentration at the outlet of the flood channel that discharges the seawater from the condenser to the sea based on the attribute value.
A required amount calculation step for calculating the required amount of the residual chlorine neutralizing agent to be injected into the drainage channel based on the predicted residual chlorine concentration.
A neutralizing agent injection step of injecting the required amount of the neutralizing agent into the flood channel,
Water quality management program to make a computer runAnd
The concentration prediction step is
An input data acquisition step for acquiring historical data of attribute values including residual chlorine concentration, salt content, pH, ORP (oxidation-reduction potential), and water temperature of seawater at the inlet of the water condenser as input data.
A label acquisition step for acquiring historical data of the residual chlorine concentration at the outlet as a label, and
A learning step for constructing a learning model for estimating the residual chlorine concentration at the outlet, using the set of the input data and the label as learning data.
A water quality management program having an estimated value generation step of generating an estimated value of the residual chlorine concentration by applying a new attribute value to the learning model after the construction of the learning model. ..

Images