KR102085296B1 - The supporting method of determination of fouling control in reverse osmosis using classification algorithm - Google Patents

Abstract

One embodiment of the present invention comprises the steps of collecting the database monitoring data from the sensor of the reverse osmosis process; Removing noise of the databased data; Predicting membrane contamination by inputting the data from which the noise has been removed to a prediction algorithm; And classifying the degree of membrane contamination by inputting the membrane contamination prediction result into a classification algorithm, thereby providing a method for supporting decision support for membrane maintenance in the reverse osmosis process, wherein the driver of the reverse osmosis process The membrane monitoring prediction data obtained by inputting the collected monitoring data into the prediction algorithm is inputted into the classification algorithm without being analyzed separately, so that the membrane maintenance decision can be easily performed according to the grading result.

Description

The supporting method of determination of fouling control in reverse osmosis using classification algorithm}

The present invention relates to a method for supporting decision-making regarding membrane maintenance in a reverse osmosis process, and more particularly, to monitor membrane contamination by collecting monitoring data of a reverse osmosis process, and to classify it using a classification algorithm. Thereby assisting the process operator in determining membrane cleaning or replacement.

Seawater desalination technology is a technology that solves the water shortage problem that has recently begun to emerge, and refers to a technology for desalination of seawater capable of securing a large amount of water resources regardless of season or weather conditions. Since seawater desalination technology is shorter in construction period and lower initial investment cost than dam construction, it is positioned as an alternative technology for securing water resources, and pursues low energy, large size, stability, and eco-friendliness.

A seawater desalination plant is a general term for processes and facilities that produce fresh water for human use such as industrial water and drinking water by removing salts as well as many inorganic salts by evaporating seawater or passing through membranes.

Desalination can be classified into distillation / evaporation method using heat energy and water evaporation and reverse osmosis (RO) method using membrane differentiation and selective permeability. Reverse osmosis (RO), or Seawater Reverse Osmosis (SWRO), is one of the major construction methods for seawater desalination plants. Energy efficiency is higher than that of Evaporation (ME), which has recently increased commercial performance. Reverse osmosis is a method of separating fresh water by joining seawater to a reverse osmosis membrane that uses water to pass water but does not permeate solutes.

In the seawater desalination process using reverse osmosis membranes, membrane fouling occurs as reverse osmosis proceeds. Membrane contamination means that the substances present in the influent adhere to the membrane and degrade the performance of the membrane. In seawater desalination, the substances present in the influent are strongly accumulated in the membrane due to the high pressure condition (50 to 70 bar). Contamination of membranes dramatically reduces the efficiency of the overall process, so controlling membrane fouling is very important for reverse osmosis processes. As membrane contamination progresses, the membrane should be cleaned periodically, or the membrane should be replaced if it cannot be solved by cleaning.

The problem is that the degree of membrane contamination must be determined to determine when to clean or replace the membrane, and there is a limit to monitoring or predicting the extent of membrane contamination in real time. Although the progress of membrane contamination can be indirectly determined by measuring the flow of produced water or increasing the operating pressure, membrane contamination is not judged indirectly. Determining the timing will result in improper cleaning or replacement, resulting in unnecessary maintenance costs. In order to solve this problem, an algorithm-based approach has been sought to determine the extent of membrane contamination.

In the prior art, a database is established by measuring the water quality of inflow water flowing into the reverse osmosis process, the operating conditions of the process, and the quality of the produced water. It provides a technique for predicting changes in permeation and membrane resistance and predicting changes in process performance.

However, predicting membrane contamination and predicting changes in the performance of the process do not immediately determine when the membrane should be cleaned, how often it will be cleaned, or when the membrane will be replaced. It is left to the reverse osmosis process operator to determine the timing of cleaning by analyzing indicators of membrane contamination or predictive performance change of the process. Unless the accuracy of the membrane contamination prediction is 100%, determining the extent of the actual membrane contamination and determining the timing of cleaning will involve the operator's experience. Therefore, there is a need for a system that enables the reverse osmosis process operator to more easily analyze the membrane contamination prediction data and determine the timing or frequency of cleaning.

An object of the present invention is to solve the above problems, and to provide a method for supporting a membrane maintenance decision in the reverse osmosis process.

The technical problem to be achieved by the present invention is not limited to the technical problem mentioned above, and other technical problems not mentioned above may be clearly understood by those skilled in the art from the following description. There will be.

In order to achieve the above technical problem, an embodiment of the present invention comprises the steps of collecting the database monitoring data from the sensor of the reverse osmosis process; Removing noise of the databased data; Predicting membrane contamination by inputting the data from which the noise has been removed to a prediction algorithm; And classifying the degree of membrane contamination by inputting the membrane contamination prediction result into a classification algorithm. 2. The method of claim 1, further comprising: a membrane maintenance decision support method in a reverse osmosis process.

In one embodiment of the present invention, the monitoring data collected from the sensor of the reverse osmosis process is influent temperature, influent TDS concentration, operating pressure, influent flow rate, production water flow rate, production water TDS concentration, production water pressure and concentrated water pressure It may be any one or more of.

In one embodiment of the present invention, the database may include one year or more reverse osmosis process monitoring data.

In one embodiment of the present invention, the noise of the databased data may be removed by a Kalman Filter algorithm.

In one embodiment of the present invention, the prediction algorithm may be an artificial neural network (ANN).

In one embodiment of the present invention, the membrane contamination prediction may include membrane resistance prediction, production water flow rate prediction and pressure drop prediction.

In one embodiment of the present invention, the classification algorithm may be a support vector machine (SVM).

In one embodiment of the present invention, the step of grading the degree of contamination of the separator may be to classify the degree of contamination of the separator into a normal operation grade, a cleaning necessary grade and a membrane replacement grade.

In one embodiment of the present invention, it is possible to provide the reverse osmosis process driver with the result of grading the degree of contamination of the separator to assist the reverse osmosis process driver to determine the membrane cleaning or membrane replacement.

In one embodiment of the present invention, after the step of grading the degree of contamination of the separator, the method may further include the step of predicting the maintenance efficiency when the maintenance in accordance with the graded contamination of the separator.

In one embodiment of the present invention, the step of predicting the maintenance efficiency may be to predict the amount of electricity required for cleaning when the number of times of cleaning, respectively, depending on the degree of contamination of the graded membrane.

In one embodiment of the present invention, the maintenance efficiency prediction result may be provided to the reverse osmosis process driver to assist the reverse osmosis process driver in determining the separation membrane cleaning time and the separation membrane cleaning frequency.

According to an embodiment of the present invention, the operator of the reverse osmosis process inputs the monitoring data collected in the reverse osmosis process into the prediction algorithm, and inputs and grades the membrane contamination prediction data into the classification algorithm instead of analyzing it separately. As a result, membrane maintenance decisions can be easily performed.

In addition, according to an embodiment of the present invention, when the maintenance decision is made according to the grading result, the maintenance efficiency can be predicted, and the decision necessary for the membrane maintenance can be performed according to the prediction result.

The effects of the present invention are not limited to the above-described effects, but should be understood to include all the effects deduced from the configuration of the invention described in the detailed description or claims of the present invention.

1 is a flow chart illustrating a membrane maintenance decision support method according to an embodiment of the present invention.
2 is a structural diagram of an artificial neural network according to an embodiment of the present invention.
3 is a graph illustrating a film resistance prediction according to an embodiment of the present invention.
4 is a structural diagram of a support vector machine according to an embodiment of the present invention.
5 is a graph of membrane contamination prediction (product water flow rate-membrane resistance) according to an embodiment of the present invention.
6 is a graph of grading membrane contamination prediction (product water flow rate-membrane resistance) according to an embodiment of the present invention.
7 is a graph of membrane contamination prediction (membrane resistance-pressure drop) according to an embodiment of the present invention.
8 is a graph of grading membrane contamination prediction (membrane resistance-pressure drop) according to an embodiment of the present invention.
9 is a graph of grading membrane contamination prediction (product water flow rate-pressure drop) according to an embodiment of the present invention.
10 is a graph of maintenance efficiency prediction (electricity consumption) according to an embodiment of the present invention.
11 is a graph of maintenance efficiency prediction (production water flow rate) according to an embodiment of the present invention.
12 is a graph of maintenance efficiency prediction (membrane resistance) according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings will be described the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is said to be "connected (connected, contacted, coupled) with another part, it is not only" directly connected "but also" indirectly connected "with another member in between. "Includes the case. In addition, when a part is said to "include" a certain component, it means that it may further include other components, without excluding the other components unless otherwise stated.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprise" or "have" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof described in the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

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

Referring to Figure 1, the present invention comprises the steps of collecting the database monitoring data from the sensor of the reverse osmosis process (S100); Removing noise of the database data (S200); Predicting membrane contamination by inputting the data from which the noise is removed (S300); And inputting the separator contamination prediction result into a classification algorithm (S400) to classify the membrane contamination degree (S400).

Monitoring data collected from the sensor of the reverse osmosis process may be any one or more of influent temperature, influent TDS concentration, operating pressure, influent flow rate, production water flow rate, production water TDS concentration, production water pressure and concentrated water pressure.

For example, in a seawater desalination plant using reverse osmosis, the inflow water becomes seawater and the production water becomes freshwater. By adjusting the temperature, pressure or flow rate of the inflow water, the flow rate or water quality of the production water can be improved. Water quality is typically determined in units of Total Dissolved solids (TDS).

The flow rate and water quality data of the production water according to the control of the temperature, pressure or flow rate of the influent can be accumulated over several days or years, and it is used as a database for the present invention. The database preferably includes at least one year of reverse osmosis process monitoring data. Data less than one year is undesirable because the prediction algorithm cannot fully machine learn. Machine learning is advantageous with more data, and one year is the minimum time to make meaningful membrane contamination predictions. The type of monitoring data is not limited to the above, but may be extended or added to the extent necessary for predicting membrane contamination.

Noise of the databased data may be removed by a Kalman Filter algorithm.

The Kalman filter can be applied when the measured value includes stochastic error, and the state at a specific time point has a linear relationship with the state at the previous time point. Monitoring data collected from the sensor of the reverse osmosis process of the present invention may include an error in the measured value. Since the monitoring data have values that are measured continuously, the value at a particular measurement point is linearly related to the previous measurement, so the Kalman filter can be used to remove the noise.

The Kalman filter works recursively. That is, the Kalman filter estimates the current value based on the value estimated at the previous time. Also, the Kalman filter does not use the measured value or the estimated value other than the immediately previous time. Each estimation calculation is performed in two steps. First, for the state estimated at the previous time, the state expected when the input is applied in the state is calculated. This step is called the prediction step. It then calculates the exact state based on the predicted state and the actually measured state. This step is called the correction step.

Predicting membrane contamination by inputting the noise-free data into a prediction algorithm may be performed through machine learning, and the machine learning may be performed through an artificial neural network.

Machine learning is a branch of artificial intelligence that develops algorithms and technologies that enable computers to learn. To give computers the ability to learn without explicit programming, and neural network models are an area of machine learning.

2 is a structural diagram schematically showing the structure of an artificial neural network according to the present invention. Artificial neural networks are algorithms that simulate the way the human brain recognizes patterns. The circle shown in FIG. 2 simulates neurons constituting a human neural network and is called a node. Nodes are connected by synapses to form a network. Several nodes are arranged to form a layer, which is formed of an input layer, a hidden layer, and an output layer.

Computation actually takes place at the node, which is designed to simulate what happens in the neurons that make up the human neural network. A node responds when it receives more than a certain amount of stimulus. The magnitude of the response is approximately proportional to the product of the input value multiplied by the node's coefficient (or weight). In general, a node receives several inputs and has as many coefficients as the number of inputs. Thus, by adjusting this coefficient, different inputs can be given different weights. The final multiplications are added together and the sum is entered into the input of the active function. The result of the active function corresponds to the output of the node, which is ultimately used for classification or regression analysis.

The goal of neural network learning is to minimize errors in the output. Initialize all coefficients on the network before learning begins. And repeatedly send data to learn. If well learned, the coefficients will be updated to the appropriate values, and the neural network can be classified and predicted. Within the learning process, coefficient updates of the same principle occur repeatedly. The principle of coefficient updating is to first estimate the coefficients, measure the errors that occur when using them, and then update the coefficients slightly based on those errors. First, when the input enters the neural network, the result is output using the coefficient of the current state. And compare this estimate with the actual correct answer. The difference between the correct answer and the estimate is the error, and the neural network measures the error and corrects the coefficients by reflecting the error. Repeating this process is a learning process.

The membrane contamination prediction may include membrane resistance prediction, production water flow rate prediction, and pressure drop prediction. Chen, K. L., L. Song, S. L. Ong and W. J. Ng (2004). "The development of membrane fouling in full-scale RO processes." Referring to the Journal of Membrane Science 232 (1): 63-72.), The membrane resistance can be calculated according to the following equations (1) and (2), the value for the prediction accuracy using the NSE value, and the NSE The value can be calculated according to the following formula (3). 3 is a graph showing the prediction of the membrane resistance over time. The prediction accuracy of FIG. 3 is 0.86.

(1)

(One)

(2)

(2)

는 막 저항(Pa s/m),

은 막 자체의 저항(Pa s/m),

는 막 오염으로 인한 저항(Pa s/m),

는 유입수 파울링 포텐셜(m^-2),

는 생산수 유량(m/s)이다.) (In the above formulas (1) and (2)

Is the membrane resistance (Pa s / m),

The resistance of the film itself (Pa s / m),

Is the resistance due to membrane contamination (Pa s / m),

Is the influent fouling potential (m ^-2 ),

Is the production water flow rate (m / s).)

(3)

(3)

는 t에서 모의값,

는 t에서 관측값,

은 관측값의 평균값이다.)(In the above formula (3)

Is the simulated value at t,

Is an observation at t,

Is the mean value of the observations.)

The classification algorithm may be a support vector machine (SVM). 4 is a structural diagram of a support vector machine according to the present invention.

Support vector machines define the dividing line or hyperplane available for classification. If the hyperplane has a large difference from the closest training data point, the classification error is small. Therefore, in order to obtain a good classification, the hyperplane having the longest distance from the closest training data for any classified point should be found. In this process, there was a problem that data could not be linearly distinguished within the initial finite dimension, and to solve this problem, a method of easily classifying by applying a higher dimension was proposed. In order to prevent the increase of computation in the process, the support vector machine designed a structure that defines the kernel function appropriate for each problem so that the inner product operation of the vector can be effectively calculated using the variables of the initial problem. The hyperplane of high dimensional space is defined by the dot product of a set of points and a constant vector.

Referring to FIG. 4, the support vector machine defines and uses a kernel function. The kernel is the addition of additional dimensions to the data to make a distinction. Sending a dividing line or hyperplane in high dimensions creates an area to separate the data, and the data hanging on the edge is called a support vector. That is, according to the definition of the kernel function, the data closest to the hyperplane becomes the support vector, but as a result, the support vector is used to define the kernel function.

Support vector machines are often used in classification through pattern recognition. Since it is based on pattern recognition, it is possible to classify a target system based on monitoring data, and in the present invention, it is used as an algorithm for classifying and classifying membrane contamination prediction results.

As shown in FIG. 4, the degree of membrane contamination may be graded into a normal operation grade, a cleaning level, and a membrane replacement level. The pattern recognized by the support vector machine may be a criterion for grading, and in the present invention, membrane resistance, product flow rate, and pressure drop, which are values predicted through a prediction algorithm, may be used as criteria. For example, the criteria for dividing normal and cleaning required ranges include a range of 5 to 10% reduction in production water flow rate, 5 to 10% increase in membrane resistance, and 10 to 15% increase in pressure drop. The criteria may be changed and added by the reverse osmosis process operator.

The degree of separation of the membrane contamination may be provided to the reverse osmosis process driver to assist the reverse osmosis process driver in determining separation membrane cleaning or membrane replacement. In the prior art, the contamination of the separator is predicted through machine learning, the prediction data is analyzed in consideration of the accuracy of the prediction data and the driver's experience, and accordingly, the driver decides the maintenance of the membrane cleaning or the replacement of the membrane. In the present invention, since the membrane contamination prediction data is graded and provided to the driver, the driver can reduce the effort to analyze the membrane contamination prediction data. In addition, the operator can pre-set the classification criteria on the support vector machine to reflect the experience, further reducing the effort required to view the graded results and determine the timing of cleaning.

5-8 show grading the membrane contamination prediction results by the support vector machine.

5 is a graph of production water flow vs. membrane resistance before grading, FIG. 6 is a graph of production water flow vs. membrane resistance after grading, FIG. 7 is membrane resistance versus pressure drop before grading, and FIG. 8 is membrane resistance versus pressure after grading. Descent graph.

The reverse osmosis process driver can easily view maintenance grades by viewing the graded areas of FIGS. 6 and 8.

In the membrane osmosis management decision support method of the reverse osmosis process of the present invention, after the step of grading the degree of contamination of the membrane, the step of predicting the maintenance efficiency when each maintenance according to the graded membrane contamination degree It may further include. The step of predicting the maintenance efficiency may be to predict the amount of electricity required for cleaning when the number of cleaning cycles differs depending on the grade of the membrane contamination.

The maintenance efficiency prediction result may be provided to the reverse osmosis process driver to assist the reverse osmosis process driver in determining the separation membrane cleaning time and the separation membrane cleaning frequency. Choosing enough cleaning times while using the least amount of electricity can reduce unnecessary energy waste.

9 to 12 show maintenance efficiency when each maintenance is performed according to the grade of membrane contamination.

9 is a graph of product water flow versus pressure drop after grading. When the membrane cleaning time is determined at three points (A), (B) and (C) of FIG. 9, the electricity consumption, the flow rate of the produced water, and the change of the membrane resistance with time are shown in FIGS. 10 to 12, respectively.

According to FIGS. 10 to 12, when the cleaning point is determined at point (A), the number of cleaning is the highest but the electricity consumption is the smallest, and when the cleaning point is determined at point (C), the number of cleaning is decreased but the electricity consumption is increased. Can be. In addition (C) it can be seen that the pressure drop and the change in membrane resistance is also large. The reverse osmosis process operator may select the cleaning time point suitable for the process operation situation in consideration of this.

The above description of the present invention is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is represented by the following claims, and it should be construed that all changes or modifications derived from the meaning and scope of the claims and equivalent concepts thereof are included in the scope of the present invention.

Claims (12)

In the membrane maintenance management decision support method in the reverse osmosis process by the membrane maintenance management decision support system in the reverse osmosis process,
Collecting and database monitoring data from a sensor of a reverse osmosis process;
Removing noise of the databased data;
Predicting membrane contamination by inputting the data from which the noise has been removed to a prediction algorithm;
Inputting the membrane contamination prediction result into a classification algorithm to grade the membrane contamination into a normal operation grade, a cleaning necessity grade, and a membrane replacement necessity grade; And
After the step of grading the degree of contamination of the separator, the step of predicting the maintenance efficiency when the maintenance is performed by predicting the amount of electricity required for cleaning when the number of times of cleaning differs according to the degree of contamination of the separator Consists of including,
Separating membrane maintenance management support method in the reverse osmosis process, characterized in that to provide a reverse osmosis process driver to the result of grading the degree of contamination of the membrane to support the reverse osmosis process driver to clean the membrane or determine the replacement membrane. The method of claim 1,
Monitoring data collected from the sensor of the reverse osmosis process is any one or more of influent temperature, influent TDS concentration, operating pressure, influent flow rate, production water flow rate, production water TDS concentration, production water pressure and concentrated water pressure Method of supporting decision making for membrane maintenance in permeation process. The method of claim 1,
The method of claim 1, wherein the database comprises a reverse osmosis process monitoring data for more than one year. The method of claim 1,
Separating membrane maintenance decision support method in the reverse osmosis process, characterized in that the noise of the database data is removed by a Kalman Filter algorithm. The method of claim 1,
The prediction algorithm is a membrane maintenance decision support method in the reverse osmosis process, characterized in that the artificial neural network (Artificial Neural Network, ANN). The method of claim 1,
The membrane contamination prediction may include membrane resistance prediction, production water flow rate prediction, and pressure drop prediction. The method of claim 1,
The classification algorithm is a support vector machine (SVM) separation membrane maintenance decision support method in the reverse osmosis process, characterized in that. delete delete delete delete The method of claim 1,
Separating membrane maintenance decision support method in the reverse osmosis process, characterized in that to provide the reverse osmosis process driver to the result of the maintenance efficiency prediction to support the determination of the membrane cleaning time and the number of membrane cleaning of the reverse osmosis process driver.

Images