KR-20260062266-A - METHOD FOR MONITORING WATER TREATMENT FACILITY BY GENERATING DIGITAL TWIN CORRESPONDING TO WATER TREATMENT FACILITY USING 3D MODELING IMAGES

KR20260062266AKR 20260062266 AKR20260062266 AKR 20260062266AKR-20260062266-A

Abstract

A method for monitoring a wastewater treatment facility comprises: a step of creating a digital twin environment for a wastewater treatment facility based on a 3D modeling image of the wastewater treatment facility including a collection tank, a plurality of biological water treatment devices, and a plurality of pumps; a step of collecting sensor data from a plurality of sensors installed in the wastewater treatment facility; a step of displaying an interface for selecting any one of the plurality of biological water treatment devices, the plurality of pumps, and the plurality of sensors in the digital twin environment; a step of obtaining predicted water quality data of treated water treated by the plurality of biological water treatment devices by inputting actual measurement data of wastewater stored in the collection tank into a plurality of artificial intelligence models; and a step of displaying predicted water quality data of treated water treated by the selected biological water treatment device when any one of the plurality of biological water treatment devices is selected through the interface.

Inventors

유기성
류황열
문영민
서준영

Assignees

재단법인 포항산업과학연구원

Dates

Publication Date: 20260507
Application Date: 20241028

Claims (20)

A step of generating a digital twin environment for a wastewater treatment facility based on a 3D modeling image of the wastewater treatment facility including a collection tank, a plurality of biological water treatment devices, and a plurality of pumps; A step of collecting sensor data from a plurality of sensors installed in the wastewater treatment facility; A step of displaying an interface for selecting any one of the plurality of biological water treatment devices, the plurality of pumps, and the plurality of sensors in the digital twin environment; A step of obtaining predicted water quality data of treated water treated by the plurality of biological water treatment devices by inputting actual measurement data of wastewater stored in the above collection tank into a plurality of artificial intelligence models; and A wastewater treatment facility monitoring method comprising the step of displaying predicted water quality data of treated water treated by the selected biological water treatment device when any one of the plurality of biological water treatment devices is selected through the interface.
In paragraph 1, The step of acquiring the above-mentioned predicted water quality data is, A step of obtaining simulation data of the wastewater supplied from the collection tank to the first biological water treatment device among the plurality of biological water treatment devices by inputting the above actual measurement data into a simulation artificial intelligence model; and A wastewater treatment facility monitoring method comprising the step of obtaining first predicted water quality data related to first treated water treated by the first biological water treatment device by inputting the above simulation data into a first predictive artificial intelligence model.
In paragraph 2, A step of modeling a process model corresponding to the wastewater treatment facility using a wastewater treatment simulator program; A step of obtaining learning simulation data related to the wastewater supplied from the collection tank to the first biological water treatment device by inputting the actual measurement data into the process model using the wastewater treatment simulator program; and A wastewater treatment facility monitoring method further comprising the step of training the simulation artificial intelligence model using the actual measurement data and the training simulation data as training data.
In paragraph 3, The step of training the above-mentioned simulation artificial intelligence model is, A step of removing redundant variables that have constant values regardless of the passage of time among the variables included in the above actual measurement data and the above training simulation data; and A wastewater treatment facility monitoring method comprising the step of training a simulation artificial intelligence model using the actual measurement data from which the aforementioned unnecessary variables have been removed and the training simulation data from which the aforementioned unnecessary variables have been removed as training data.
In paragraph 3, The method further includes the step of analyzing the correlation between variables included in the actual measurement data and the training simulation data, and determining multiple feature points by classifying multiple variables whose correlation is greater than or equal to a predetermined value into a single feature point. The step of training the above-mentioned simulation artificial intelligence model is, A wastewater treatment facility monitoring method comprising: a step of training a simulation artificial intelligence model using the plurality of feature points included in the actual measurement data and the training simulation data as training data.
In paragraph 2, A step of modeling a process model corresponding to the wastewater treatment facility using a wastewater treatment simulator program; A step of obtaining first learning simulation data related to the wastewater supplied to the first biological water treatment device and second learning simulation data related to the first treated water treated by the first biological water treatment device by inputting the actual measurement data into the process model using the wastewater treatment simulator program; A wastewater treatment facility monitoring method further comprising the step of training the first prediction artificial intelligence model using the first training simulation data and the second training simulation data as training data.
In paragraph 6, The step of training the above-mentioned first prediction artificial intelligence model is, A step of removing redundant variables that have constant values regardless of the flow of time among the variables included in the first training simulation data and the second training simulation data; and A wastewater treatment facility monitoring method comprising the step of training a first prediction artificial intelligence model using the first training simulation data from which unnecessary variables have been removed and the second training simulation data from which unnecessary variables have been removed as training data.
In paragraph 6, A step of determining a plurality of input feature points by analyzing the correlation between input variables included in the first learning simulation data and classifying a plurality of input variables whose correlation is greater than or equal to a predetermined value into a single input feature point; and The method further includes the step of analyzing the correlation between output variables included in the second learning simulation data and determining a plurality of output feature points by classifying a plurality of output variables whose correlation is greater than or equal to a predetermined value into a single output feature point. The step of training the above-mentioned first prediction artificial intelligence model is, A wastewater treatment facility monitoring method comprising the step of training a first prediction artificial intelligence model using the plurality of input feature points included in the first training simulation data and the plurality of output feature points included in the second training simulation data as training data.
In paragraph 6, The step of training the above-mentioned first prediction artificial intelligence model is, A method for monitoring a wastewater treatment facility comprising the step of using a plurality of variables included in the first training simulation data and the second training simulation data as training data after performing min-max normalization.
In paragraph 2, The step of acquiring the above-mentioned predicted water quality data is, A wastewater treatment facility monitoring method further comprising the step of acquiring second predicted water quality data related to second treated water treated by a second biological water treatment device that receives and processes the first treated water among the plurality of biological water treatment devices by inputting the first predicted data into a second predicted artificial intelligence model.
In Paragraph 10, A step of modeling a process model corresponding to the wastewater treatment facility using a wastewater treatment simulator program; A step of obtaining first learning simulation data related to the first treated water supplied from the first biological water treatment device to the second biological water treatment device and second learning simulation data related to the second treated water treated by the second biological water treatment device by inputting the actual measurement data into the process model using the wastewater treatment simulator program; and A wastewater treatment facility monitoring method further comprising the step of training the second prediction artificial intelligence model using the first training simulation data and the second training simulation data as training data.
In Paragraph 11, The step of training the above-mentioned second prediction artificial intelligence model is, A step of removing redundant variables that have constant values regardless of the flow of time among the variables included in the first training simulation data and the second training simulation data; and A wastewater treatment facility monitoring method comprising the step of training a second prediction artificial intelligence model using the first training simulation data from which unnecessary variables have been removed and the second training simulation data from which unnecessary variables have been removed as training data.
In Paragraph 11, A step of determining a plurality of input feature points by analyzing the correlation between input variables included in the first learning simulation data and classifying a plurality of input variables whose correlation is greater than or equal to a predetermined value into a single input feature point; and The method further includes the step of analyzing the correlation between output variables included in the second learning simulation data and determining a plurality of output feature points by classifying a plurality of output variables whose correlation is greater than or equal to a predetermined value into a single output feature point. The step of training the above-mentioned second prediction artificial intelligence model is, A wastewater treatment facility monitoring method comprising the step of training a second prediction artificial intelligence model using the plurality of input feature points included in the first training simulation data and the plurality of output feature points included in the second training simulation data as training data.
In Paragraph 11, The step of training the above-mentioned second prediction artificial intelligence model is, A method for monitoring a wastewater treatment facility comprising the step of using a plurality of variables included in the first training simulation data and the second training simulation data as training data after performing min-max normalization.
In paragraph 1, A wastewater treatment facility monitoring method further comprising the step of displaying sensor data collected by the selected sensor when any one of the plurality of sensors is selected through the interface.
In paragraph 1, A wastewater treatment facility monitoring method further comprising the step of, when any one of the plurality of pumps is selected through the interface, displaying at least one of the water supply unit flow rate of the selected pump and the failure state of the selected pump.
In Paragraph 16, The plurality of sensors above include an optical camera and a thermal imaging camera for photographing the pump, and The above wastewater treatment facility monitoring method is, A step of receiving an optical image and a thermal image from the optical camera and the thermal camera; A step of distinguishing a region of interest corresponding to the pump and the remaining background region based on the optical image above; A step of removing a portion corresponding to the background area from the thermal image; and A wastewater treatment facility monitoring method further comprising the step of identifying whether the pump is faulty by inputting the thermal image with the background area removed into a fault detection artificial intelligence model.
In paragraph 1, A step of detecting a fire in the wastewater treatment facility based on sensor data collected from the plurality of sensors; and A wastewater treatment facility monitoring method further comprising the step of: based on the detection of a fire in the wastewater treatment facility, marking the area where the fire was detected in the digital twin environment to distinguish it from other areas, and outputting a warning notification.
In paragraph 1, The above plurality of sensors include an optical camera and a thermal imaging camera, and The step of detecting the fire above is, A step of receiving optical image data and thermal image data from the optical camera and the thermal imaging camera, respectively; A step of identifying target pixels exceeding a predetermined safe temperature based on the processing of the above thermal image data; A step of calculating a fire risk rate based on the average value of temperatures corresponding to the above target pixels; A step of calculating a smoke detection rate by inputting the above optical image data into a smoke detection artificial intelligence model; and A wastewater treatment facility monitoring method comprising: a step of detecting that a fire has occurred in the wastewater treatment facility in response to the smoke detection rate exceeding a first predetermined value and the fire risk rate exceeding a second predetermined value.
In a storage medium for storing at least one instruction for monitoring a wastewater treatment facility, When the above at least one instruction is executed by a processor, the processor, A digital twin environment for a wastewater treatment facility is created based on a 3D modeling image of the wastewater treatment facility including a collection tank, a plurality of biological water treatment devices, and a plurality of pumps, and Collecting sensor data from a plurality of sensors installed in the above wastewater treatment facility, and Displays an interface for selecting any one of the plurality of biological water treatment devices, the plurality of pumps, and the plurality of sensors in the above digital twin environment, and By inputting actual measurement data of wastewater stored in the above collection tank into a plurality of artificial intelligence models, predicted water quality data of the treated water treated by the plurality of biological water treatment devices is obtained, and A storage medium that displays predicted water quality data of the treated water treated by the selected biological water treatment device when any one of the plurality of biological water treatment devices is selected through the interface.

Description

Method for monitoring a wastewater treatment facility by generating a digital twin corresponding to the wastewater treatment facility using 3D modeling images The present invention relates to a method for monitoring a wastewater treatment facility using a digital twin created using a 3D modeling image. Wastewater treatment facilities include various types of wastewater treatment equipment, and if any of the individual wastewater treatment facilities operates abnormally, the water quality of the treated water processed by the wastewater treatment facilities may not improve to the desired level. Operators rely on regular sampling and analysis to verify the quality of treated water from wastewater treatment facilities and maintain or control the equipment based on the results. However, this method fails to immediately reflect changes in water quality and is highly likely to miss unexpected contamination or abnormal conditions that occur between sampling cycles. Consequently, immediate response becomes difficult when urgent problems arise, and in severe cases, this can lead to equipment damage or massive maintenance costs. Therefore, in the absence of an efficient method to predict water quality in real time, it is difficult to guarantee the stability and efficiency of wastewater treatment facilities. This makes it difficult to respond quickly and appropriately to changes in water quality, ultimately having a negative impact on the overall operational reliability of the facilities. FIG. 1 is a flowchart illustrating an example of a wastewater treatment facility monitoring method according to one embodiment. FIG. 2 illustrates an example of an interface including a digital twin environment image of a wastewater treatment facility according to one embodiment. FIG. 3 illustrates an example of data displayed on an interface when a biological wastewater treatment device is selected among wastewater treatment facilities according to one embodiment. FIG. 4 illustrates an example of data displayed on an interface when a specific sensor is selected among wastewater treatment facilities according to one embodiment. FIG. 5 illustrates an example of a warning notification displayed on an interface according to one embodiment. FIG. 6 illustrates another example of a warning notification displayed on an interface according to one embodiment. FIG. 7 illustrates another example of a warning notification displayed on an interface according to one embodiment. FIG. 8 is a flowchart illustrating an example of a method for predicting the results of a biological wastewater treatment process according to one embodiment. FIG. 9 is a schematic diagram illustrating a method for predicting the results of a biological wastewater treatment process according to one embodiment, performed by a plurality of artificial intelligence models. FIG. 10 is a flowchart illustrating an example of a learning method for a plurality of artificial intelligence models used in a method for predicting the results of a biological wastewater treatment process according to one embodiment. Figure 11 illustrates an example of a process model corresponding to an actual wastewater treatment facility modeled using a wastewater treatment simulator. Figure 12 illustrates an example of training simulation data obtained using a wastewater treatment simulator. FIG. 13 is a flowchart illustrating an example of a method for preprocessing training simulation data according to one embodiment. FIG. 14 illustrates an example of an unnecessary variable that has a constant value regardless of the flow of time among the variables included in the training simulation data according to one embodiment. FIG. 15 illustrates an example of an analysis table in which the correlation between variables included in the training simulation data according to one embodiment is analyzed. FIG. 16 is a diagram illustrating variables that require maximum/minimum normalization among the variables included in the training simulation data according to one embodiment. FIG. 17 is a flowchart illustrating an example of a method for detecting a fire based on an optical image in a wastewater treatment facility monitoring method according to one embodiment. FIG. 18 illustrates an example of an output image when an optical image is input to a smoke detection artificial intelligence model according to one embodiment. FIG. 19 is a flowchart illustrating an example of a method for detecting a fire based on a thermal image in a wastewater treatment facility monitoring method according to one embodiment. FIG. 20 illustrates an example of target pixels identified in a thermal image according to one embodiment. FIG. 21 is a flowchart illustrating an example of a method for detecting a fire based on an optical image and a thermal image in a wastewater treatment facility monitoring method according to one embodiment. FIG. 22 is a flowchart illustrating an example of a method for detecting a failure of a target device