KR102403493B1 - Integrated management system of sewage and wastewater treatment plants using artificial intelligence - Google Patents

Abstract

The present invention provides an integrated management system of sewage and wastewater treatment plants which may block influence of disturbance by predicting outflow water quality using a prediction model derived from artificial intelligence to apply a control operation at a suitable time requiring control. The integrated management system of sewage and wastewater treatment plants comprises: a sensor unit installed at sewage and wastewater treatment plants to measure inflow water quality, inflow water quantity, a driving factor, outflow water quality, and outflow water amount in real time or at predetermined time intervals; a prediction unit mounted therein with an artificial intelligence program for collecting data from the sensor unit and analyzing the data through a learning algorithm to generate prediction models by sewage and wastewater treatment plants, and predicting the outflow water quality of a selected sewage and wastewater treatment plant based on the prediction models; and a treatment process controller selecting at least one control variable from the inflow water quality, the inflow water quantity, or the driving factor to control a process of the corresponding sewage and wastewater treatment plants when the outflow water quality predicted from the prediction unit exceeds the reference outflow water quality.

Description

Integrated management system of sewage and wastewater treatment plants using artificial intelligence

The present invention relates to a system for constructing a predictive model for the quality of runoff at each sewage and wastewater treatment plant using artificial intelligence, and to appropriately control each sewage and wastewater treatment plant accordingly.

The sewage and wastewater treatment process has a great difficulty in stably treating the influent, which fluctuates daily, weekly, and seasonally, to meet the effluent quality standards. The most optimal way to cope with fluctuations in influent flow rate or load fluctuations of pollutants in influent is to predict such fluctuations in advance, take measures not to affect the process, and then reduce the processing capacity of the process to the quality of the effluent. It is monitored and confirmed as In this respect, the existing feedback control (FB control) technique that alleviates the effect of disturbance on the process after the effect of disturbance occurs, such as the DO control traditionally used in the sewage and wastewater treatment process, is There is a disadvantage in that a lag time occurs until the time when the control operation affects the process, so that the complete removal of the disturbance cannot be performed.

On the other hand, the Feedforward control (FF control) technique, which detects disturbances affecting the process in advance and performs a control operation that can cancel the disturbance in advance, predicts the effect of disturbances on the process in advance. It is easier to secure process stability than the FB control method in that it blocks In order to apply this FF control technique to the sewage and wastewater treatment process, a prerequisite is that disturbance must be measured or predicted in advance.

The control technology of the sewage and wastewater treatment process has been mentioned in various ways, from simple DO control to real-time control technology using an automatic nutrient analysis device. However, looking at the form of the existing technologies that have been applied in the latest automated control system, most had a feedback control (FB) structure that controls the effect of disturbance on the process after it occurs, which does not prevent the influence of disturbance in advance. There were frequent cases of difficulties in stable process operation because of Recently, a method of predicting and controlling COD or T-N items has also been reported, but since it is limited to unit item control and does not have an FF control structure based on a pre-planned scheduling, the driving timing of an appropriate adjustment variable for disturbance control and its fluctuations The value was not clearly displayed, so there was a limit that the disturbance effect could not be completely controlled.

Korean Patent Registration No. 1890574

The present invention is to solve the above problems, and as the runoff water quality is predicted using a prediction model derived by artificial intelligence, a control operation is applied in advance at an appropriate time when control is required to block the effect of disturbance in advance. Its purpose is to provide an integrated management system for sewage and wastewater treatment plants.

As a means for achieving the above object, the integrated management system for sewage and wastewater treatment plants using artificial intelligence (hereinafter referred to as the "system of the present invention") is installed in each sewage and wastewater treatment plant to provide inflow water quality in real time or at predetermined time intervals. , a sensor unit for measuring inflow water quantity, driving factors, outflow water quality, and outflow water amount; It is equipped with an artificial intelligence program that collects the data of the sensor unit, analyzes the data through a learning algorithm, generates a predictive model for each sewage and wastewater treatment plant, and predicts the quality of the effluent from the selected sewage and wastewater treatment plant based on the predictive model predicted part; When the effluent quality predicted by the prediction unit is out of the standard effluent quality, a treatment process control unit that selects one or more control variables among influent water quality, influent water quantity, and operation factors to control the process of the corresponding sewage/wastewater treatment plant; characterized by including; do it with

As an example, the influent water quality and the outflow water quality items include at least one of temperature, BOD, COD, SS, T-N, and T-P, which are water quality analysis items.

As an example, when the effluent quality predicted by the prediction unit is out of the standard effluent quality, the treatment process control unit virtually changes one or more of influent water quality, influent water quantity, and operating factors, and applies the predictive model to determine the predicted effluent quality It is characterized in that the process of the corresponding sewage/wastewater treatment plant is controlled by selecting one or more of influent water quality, influent water quantity, and operation factors to be within the standard outflow water quality range.

As an example, the prediction unit may include: a data module for receiving and storing data from the sensor unit of each sewage and wastewater treatment plant; a feature deriving module for extracting one or more features for learning from the data collected from the data module; a learning module for performing deep learning learning based on the features extracted from the feature deriving module; a model generation module for receiving the result information learned from the learning module and generating a predictive model for each sewage and wastewater treatment plant based on this; and a prediction module for predicting the effluent quality of each sewage and wastewater treatment plant based on the prediction model of each sewage and wastewater treatment plant generated by the model generation module.

As an example, the sewage and wastewater treatment plant includes a bioreactor, wherein the bioreactor has a floating port formed in the lower portion and a collection space formed therein, and is connected to the floating collection port and is collected in the floating collection port A gas discharge line for discharging gas, and a by-product gas measurement unit comprising a second sensor unit for measuring the amount of oxygen and carbon dioxide mixed from the gas discharged to the gas discharge line is further included, and the prediction unit receives the data of the second sensor unit Collect, analyze the data through a learning algorithm to generate a predictive model for each sewage and wastewater treatment plant, predict the allowable load of the selected sewage and wastewater treatment plant based on the predictive model, and in the treatment process control unit, in the prediction unit When the predicted allowable load is out of the standard allowable load, it is characterized in that one or more control variables of operation of the Fenton process and control of the air flow are selected to control the process of the corresponding sewage/wastewater treatment plant.

As an example, the floating port is configured in a shape with an open lower portion, and the floating port forms a predetermined distance from the lower end to the upper side of the side wall, so that the lower end of the floating port in the floating port is a blocking border that is locked in the water surface. do.

As an example, the floating collection port further includes a water removal port, wherein the water removal port is divided into a collection chamber at a lower portion and a separation chamber at an upper portion by a partition wall, and the gas collected in the floating collection port flows into the collection chamber. A housing comprising a gas inlet line and a water discharge line, the separation chamber having the gas discharge line formed therein, a plurality of perforated tubes upright in the separation chamber, the lower end communicating with the collection chamber and forming pores; It is characterized in that it includes a suction pump configured in the discharge line.

The system of the present invention has the advantage of enabling integrated management by enabling efficient control of each sewage and wastewater treatment plant through the establishment of a predictive model using artificial intelligence.

1 is a block diagram showing a system of the present invention;
Figure 2 is a block diagram showing the detailed configuration of the prediction unit shown in Figure 1,
3 is a schematic diagram showing an embodiment of the present invention,
Figure 4 is a schematic view showing the floating trap shown in Figure 3,
5 is an operation state diagram showing the water removal port shown in FIG.

Hereinafter, the present invention will be described in more detail with reference to the drawings and examples. Since the following description is for a specific example of the present invention, it is not intended to limit the scope of the rights defined in the claims, even if there are assertive and limiting expressions.

As shown in FIG. 1, the system 100 of the present invention is installed in each sewage and wastewater treatment plant (a) and measures inflow water quality, inflow water quantity, driving factor, outflow water quality, and outflow water quantity in real time or at predetermined time intervals. (101); The data of the sensor unit 101 is collected, and the data are analyzed through a learning algorithm to generate a predictive model for each sewage and wastewater treatment plant (a), and the quality of the effluent from the selected sewage and wastewater treatment plant based on the predictive model Prediction unit 102 equipped with an artificial intelligence program to predict; When the effluent quality predicted by the prediction unit 102 deviates from the standard effluent quality, the treatment process control unit 103 selects one or more control variables among influent water quality, influent water quantity, and operation factors to control the process of the corresponding sewage and wastewater treatment plant. It is characterized in that it contains;

As shown in FIG. 1, the sensor unit 101 is installed in a plurality of sewage and wastewater treatment plants (a) to measure inflow water quality, inflow water quantity, driving factors, outflow water quality, and outflow water quantity in real time or at predetermined time intervals. , in the sewage and wastewater treatment plant (a), the inflow water quality, the inflow water quantity and the operating factors of unit processes in the treatment plant, the outflow water quality and the outflow water quantity, etc. are measured in real time or at predetermined time intervals and transmitted to the prediction unit 102 . The sensor unit 101 for measuring water quality includes sensors capable of measuring the flow rate and water quality (temperature, BOD, COD, SS, TN, TP, etc.) of influent and outflow water, and operating factors (DO, It is a sensor that can measure MLSS, sludge conveyance flow rate, internal conveyance flow rate, temperature, etc.).

In addition, when a sewage/wastewater treatment plant is configured with a plurality of reaction tanks, it is natural that a plurality of these sensors may be installed in each reaction tank or the like. In addition, the sensor may measure in real time, and may measure at intervals of a predetermined time, for example, minutes or hours.

The prediction unit 102 collects the data of the sensor unit 101, analyzes the data through a learning algorithm to generate a prediction model for each sewage and wastewater treatment plant, and a sewage and wastewater treatment plant selected based on the prediction model It is characterized in that it is equipped with an artificial intelligence program that predicts the quality of runoff. Here, various known technologies may be applied to the artificial intelligence program, but the present invention provides an example in which a deep learning-based learning algorithm is applied as an example.

To explain this in more detail, the prediction unit 102 includes a data module 1021 that receives and stores data from the sensor unit 101 of each sewage and wastewater treatment plant as shown in FIG. 2 ; a feature deriving module 1022 for extracting one or more features for learning from the data collected from the data module 1021; a learning module 1023 for performing deep learning learning based on the features extracted from the feature deriving module 1022; a model generation module 1024 for receiving the result information learned from the learning module 1023 and generating a predictive model for each sewage and wastewater treatment plant based on this; and a prediction module 1025 for predicting the effluent quality of each sewage and wastewater treatment plant based on the prediction model of each sewage and wastewater treatment plant generated by the model generation module 1024.

The data module 1021 may collect data from the sensor unit 101 of at least one or more sewage and wastewater treatment plants through online.

The feature deriving module 1022 may extract and separate one or more features for learning from the data collected from the data module 1021 .

That is, the feature derivation module 1022 extracts major factors necessary for learning from the vast amount of information collected from the data module 1021, for example, inflow water quality, inflow water quantity, operation factors, It is possible to extract characteristics such as effluent quality and effluent quantity.

The learning module 1023 may perform learning based on the features extracted from the feature deriving module 1022 . Learning by the learning module 1023 may be applied to one or more selected among known learning algorithms.

The learning module 1023 may use a deep learning-based learning algorithm, and preferably, may perform learning using a Convolutional Neural Network (CNN) algorithm belonging to a deep learning learning neural network series.

The model generation module 1024 may receive the result information learned from the learning module 1023 and generate a predictive model for each sewage and wastewater treatment plant based on the received information. That is, the predictive model is a predictive model for each sewage and wastewater treatment plant, and this predictive model may be updated in real time by the collection information of the data module 1021 .

The prediction module 1025 may predict the effluent quality of the selected sewage/wastewater treatment plant based on the prediction model of each sewage/wastewater treatment plant generated by the model generation module 1024 .

The treatment process control unit 103 corresponds to a configuration for controlling the process of the sewage and wastewater treatment plant by selecting a control variable when the effluent quality predicted by the prediction unit 102 is out of the reference effluent quality. The control variable is characterized in that at least one of influent water quality, influent water quantity, and operating factors.

To explain this in more detail, the treatment process control unit 103 virtually changes one or more of influent water quality, influent water quantity, and operating factors when the effluent quality predicted by the prediction unit 102 deviates from the standard effluent quality, and the prediction model is applied to select one or more of influent quality, influent water quantity, and operation factors to ensure that the predicted effluent quality is within the standard effluent quality range to control the process of the relevant sewage/wastewater treatment plant.

Here, the operating factor as the control variable is not limited thereto, but may include an external carbon source flow rate, a reaction tank internal return flow rate, a sludge return flow rate, a waste sludge flow rate, and an aerobic tank dissolved oxygen concentration.

For example, when the concentration value of NOX-N among the predicted effluent concentration values exceeds the reference concentration range, the treatment process control unit 103 may select an external carbon source flow rate and an internal conveyance flow rate as control variables. That is, by increasing the external carbon source flow rate and the internal return flow rate, it is possible to prevent the effluent concentration value from exceeding the reference range in the near future. It is possible to further include other control variables, but there is a burden of additional processing time and cost.

In addition, when the concentration value of NH4+-N exceeds the standard concentration range, control variables including internal return flow rate and dissolved oxygen concentration in the aerobic tank can be selected. Control parameters including sludge return flow rate and aerobic dissolved oxygen concentration can be selected.

The prediction unit 102 and the processing process control unit 103 may be configured in the central server (b).

With these configurations, the system 100 of the present invention builds a predictive model of the effluent water quality of each device with reference to the data on the past influent water quality, influent water quantity and operating factors of unit processes in the treatment plant, effluent water quality and effluent quantity, , the time when a control operation is required is detected in advance by predicting the quality of the effluent from the sewage and wastewater treatment plant selected by this prediction model, and the type of control operation (selection of the control variable) and to choose its size.

On the other hand, in FIG. 3, an example of a sewage and wastewater treatment plant (a) is shown. The sewage and wastewater treatment plant (a) of this embodiment includes a bioreactor 1 in which sewage and wastewater are introduced and a biological reaction is performed, and thus In addition, an example in which the by-product gas measurement unit 2 is further included in the bioreactor 1 is presented.

As shown in FIG. 3, the by-product gas measuring unit 2 has a floating port 211 formed at the lower portion and is connected to a floating collecting port 21 having a collecting space therein, and the floating collecting port 21, A gas discharge line 24 for discharging the gas collected in the floating collection port 21, and a second sensor unit 22 for measuring the amount of oxygen and carbon dioxide mixed from the gas discharged to the gas discharge line 24 characterized in that

As shown in FIG. 3 , the floating port 21 has an open lower portion, and the floating port 221 is configured while forming a predetermined distance from the lower end of the side wall to the upper side. The lower end of the floating port 221 is such that a blocking border 212 that is locked in the water surface is formed.

The reason that the blocking border 212 is formed is to block the mixing of the air outside the floating collection port 21 into the off-gas collected inside the floating collection port 21, so that the second sensor unit This is to increase the sensing accuracy of (22). That is, this is to prevent the collection space formed by the floating collection port 21 from being exposed to the outside air even when the floating collection port 21 is shaken due to various causes.

The second sensor unit 22 is connected to the gas discharge line 24 and is configured to sense the mixing amount of oxygen and carbon dioxide from the by-product gas collected in the floating collection port 21 .

For this, of course, the second sensor unit 22 should include an oxygen measuring sensor 221 and a carbon dioxide measuring sensor 222 .

In conjunction with this by-product gas measurement unit 2, the prediction unit 102 collects the data of the second sensor unit 22, and analyzes the data through a learning algorithm to generate a predictive model for each sewage and wastewater treatment plant. and the allowable load of the selected sewage and wastewater treatment plant is predicted based on the prediction model.

In this case as well, as mentioned above, the prediction unit 102 collects the data of the second sensor unit 22 and analyzes the data through a learning algorithm to generate a predictive model for each sewage and wastewater treatment plant, It is characterized in that an artificial intelligence program is installed to predict the allowable load of the selected sewage and wastewater treatment plant based on the prediction model.

Then, when the allowable load predicted by the prediction unit 102 is out of the standard allowable load, the treatment process control unit 103 selects one or more control variables among operation of the Fenton process and control of the air flow rate to control the process of the corresponding sewage and wastewater treatment plant. characterized in that

That is, the concentration of O ₂ , CO ₂ gas in each sewage and wastewater treatment plant is analyzed in real time to determine the nitrification point and to efficiently operate the blower to increase the treatment efficiency and achieve energy savings.

In addition, in the present invention, an example is provided in which a water removal port 23 is further included in the floating collection port 21 so that the sensing is performed more accurately by the second sensor unit 22 .

The water removal port 23 is divided into a collecting chamber 231-3 in the lower part and a separation chamber 231-2 in the upper part by a partition wall 231-1, and the floating collecting chamber 231-3 in the collecting chamber 231-3. A gas inlet line 231-4 and a water outlet line 231-5 through which the gas collected in the sphere 21 is introduced are configured, and the gas outlet line 24 is formed in the separation chamber 231-2 A housing 231, a plurality of perforated pipes 232 upright in the separation chamber 231-2, the lower ends communicating with the collecting chamber 231-3 and forming pores, and the gas discharge line 24 It is characterized in that it includes a suction pump 233 configured to.

The gas collected in the floating collection port 21 is introduced into the collection chamber 231-3 through the gas inlet line 231-4, and the gas collected in the collection chamber 231-3 is separated by a partition wall ( 231-1) through the perforated tube 232 is guided upward.

To this end, although not shown in the drawings, the partition wall 231-1 is provided with a through hole facing the perforated tube 232 so that the lower end of the perforated tube 232 and the collecting chamber 231-3 communicate with each other. do.

As described above, when the suction pump 233 is operated to apply a negative pressure to the separation chamber 231-2 while the gas ascends through the perforated tube 232, the gas having a relatively small specific gravity in the gas is transferred to the perforated tube ( 232), the separated water is discharged to the outside through the pores of the perforated tube 232, and is stored in the collection chamber 231-3 downwards on the inside of the perforated tube 232, and is discharged to the outside through the water discharge line 231-5). to be expelled.

More preferably, a lipophilic coating layer is applied to the inner periphery of the perforated tube 232 to increase the separation efficiency of water from the gas by negative pressure.

In this way, moisture is separated from the collected gas to increase the sensing accuracy in the second sensor unit 22, and foreign substances mixed in the gas are filtered out by the perforated tube 232 in the second sensor unit. It is to control the deposition in the oxygen measuring sensor 221 and the carbon dioxide measuring sensor 222 of (22).

Those skilled in the art from the above description will be able to see that various changes and modifications are possible without departing from the technical spirit of the present invention. Accordingly, the technical scope of the present invention should be defined by the claims rather than being limited to the contents described in the detailed description of the specification.

101: sensor unit 102: prediction unit
103: control unit 1021: data module
1022: feature derivation module 1023: learning module
1024: model generation module 1025: prediction module

Claims (7)

a sensor unit installed at each sewage and wastewater treatment plant to measure inflow water quality, inflow water quantity, driving factors, outflow water quality, and outflow water quantity in real time or at predetermined time intervals;
It is equipped with an artificial intelligence program that collects data from the sensor unit, analyzes the data through a learning algorithm, generates a predictive model for each sewage and wastewater treatment plant, and predicts the effluent quality of the selected sewage and wastewater treatment plant based on the predictive model predicted part; and
When the effluent quality predicted by the prediction unit is out of the standard effluent quality, a treatment process control unit that selects one or more control variables among influent water quality, influent water quantity, and operating factors to control the process of the corresponding sewage and wastewater treatment plant; includes,
The sewage and wastewater treatment plant includes a bioreactor in which sewage and wastewater are introduced and a biological reaction occurs,
In the bioreactor, a floating port is formed in the lower portion and a collection space is formed therein; a gas discharge line connected to the floating collection port and exhausting the gas collected in the floating collection port; A by-product gas measurement unit including a second sensor unit for measuring the mixed amount of oxygen and carbon dioxide from the gas is included,
The prediction unit collects the data of the second sensor unit, analyzes the data through a learning algorithm to generate a prediction model for each sewage and wastewater treatment plant, and predicts the allowable load of the selected sewage and wastewater treatment plant based on the prediction model do,
When the allowable load predicted by the prediction unit is out of the standard allowable load, the treatment process control unit selects one or more control variables among operation of the Fenton process and airflow control to control the process of the corresponding sewage and wastewater treatment plant,
The floating collection port,
In the form of an open lower portion, the floating port is formed while forming a certain distance from the lower end of the side wall to the upper side, and as the floating port is submerged in the water surface with the lower end of the floating port in the floating port, the by-product gas collected inside the floating port A blocking border is formed to block the mixing of air from the outside of the floating collection port,
The floating collection port further includes a water removal port,
The water removal port is divided into a collecting chamber at the lower part and a separation chamber at the upper part by a partition wall, and the collecting chamber includes a gas inlet line and a water outlet line through which the gas collected in the floating collecting port flows in,
A housing in which the gas discharge line is formed in the separation chamber; a plurality of perforated tubes installed upright in the separation chamber, the lower end communicating with the collection chamber, forming pores, and an inner periphery of which a lipophilic coating layer is applied; a suction pump configured in the discharge line, wherein a plurality of through-holes are formed in the partition wall to face the perforated tube so that the lower end of each perforated tube communicates with the collection chamber;
The gas collected in the floating collection port is introduced into the collection chamber through the gas inlet line, and the suction pump is operated while the gas collected in the collection chamber is upwardly guided to each perforated tube through the partition wall to operate the separation chamber. When a negative pressure is formed at The moisture removal unit separates moisture from the gas that is discharged to the outside through the collected gas and improves the sensing accuracy of the second sensor unit, and prevents foreign substances mixed in the gas from being deposited in the second sensor unit by filtering foreign substances by the perforated tube. An integrated management system for sewage and wastewater treatment plants using artificial intelligence, characterized in that
The method of claim 1,
The influent water quality and the outflow water quality items include at least one of temperature, BOD, COD, SS, TN, and TP, which are water quality analysis items.
The method of claim 1,
The treatment process control unit, when the effluent quality predicted by the prediction unit deviates from the standard effluent quality, virtually changes one or more of influent water quality, influent water quantity, and operating factors, and applies the predictive model to determine the estimated effluent quality is the standard outflow An integrated management system for a sewage and wastewater treatment plant using artificial intelligence, characterized in that the process of the corresponding sewage and wastewater treatment plant is controlled by selecting one or more of influent water quality, influent water quantity, and operation factors to be within the water quality range.
The method of claim 1,
The prediction unit,
a data module for receiving and storing data from the sensor unit of each sewage and wastewater treatment plant;
a feature derivation module for extracting one or more features for learning from the data collected from the data module;
a learning module for performing deep learning learning based on the features extracted from the feature deriving module;
a model generation module for receiving the result information learned from the learning module and generating a predictive model for each sewage and wastewater treatment plant based on this;
a prediction module for predicting effluent quality of each sewage and wastewater treatment plant based on the prediction model of each sewage and wastewater treatment plant generated by the model generation module;
Sewage and wastewater treatment plant integrated management system using artificial intelligence, characterized in that it comprises a.
delete delete delete

Images