JP2022100265A - Abnormality detection method of soaking type membrane separator - Google Patents

Abstract

To provide an abnormality detection method of a soaking type membrane separator capable of mitigating a monitoring person's load and quickly and accurately diagnosing a condition of a filtration film even when a monitoring person is absent.SOLUTION: An abnormality detection method of a soaking type membrane separator that alternately repeats filtration running and relaxation running includes a data acquisition step of measuring and acquiring an inter-membrane differential pressure of a soaking type membrane separator at intervals of a predetermined time, a condition separation step of separating a group of inter-membrane differential pressure measured values for a first period, which encompasses an object period during which an abnormality is detected. out of acquired time-sequential inter-membrane differential pressure measured values into a first group of inter-membrane differential pressure measured values for filtration running and a second group of inter-membrane differential pressure measured values for relaxation running, a noise removing step of removing noise data from each of the groups of inter-membrane differential pressure measured values, and an abnormality determination step of calculating a dispersing condition from a group of differential pressure measured values for a second period shorter than the first period from the group of inter-membrane differential pressure measured values resulting from noise removal, and determining based on the dispersing condition whether the soaking type membrane separator is abnormal.SELECTED DRAWING: Figure 4

Description

The present invention relates to an abnormality detection method for an immersion type membrane separation device.

In various water treatment plants such as sewage treatment, an immersion type membrane separation device is arranged in the treatment tank, and the water to be treated in the treatment tank is suction-filtered by the filtration membrane incorporated in the immersion type membrane separation device. The treated water separated into solid and liquid can be obtained.

In the immersion type membrane separation device, an air diffuser is installed below the membrane separation device, and a filtration operation is executed in which suction filtration is performed using a filtration membrane while the air is diffused by the air diffuser. However, with the passage of time, clogging and accumulation of solid components occur in the filtration membrane, and the filtration efficiency decreases. Therefore, normally, the filtration operation and the relaxation to stop the suction filtration while maintaining the air diffusion from the air diffuser A management device that repeats the operation at predetermined time intervals and manages the operation so that the filtration film surface is cleaned by the upward flow of the gas-liquid mixed phase of the air bubbles from the air diffuser and the water to be treated even in the relaxation operation as in the filtration operation. It is equipped with.

Then, the management device measures the pressure difference between the front surface (primary side surface) side and the back surface (secondary side surface) side of the filtration membrane, that is, the intermembrane differential pressure, and the intermembrane differential pressure exceeds the control threshold or suddenly increases. It is configured to determine that reverse pressure cleaning or chemical cleaning is necessary at the timing.

For example, Patent Document 1 proposes a filtration member cleaning system capable of performing back pressure cleaning on a filtration member at an effective timing while suppressing the occurrence of blockage on the filtration member in the filtration processing apparatus. ing.

The filtration member cleaning system includes a filtration member that allows the water to be treated to pass from the primary side to the secondary side to filter the water to be treated, and a primary side region and a secondary side region partitioned by the filtration member. A filtration treatment device that performs filtration treatment of the water to be treated by passing the water to be treated through the filtration member from the primary side region toward the secondary side region, and from the primary side region to the secondary side region. To measure the pressure difference between the primary side region and the secondary side region, the treatment water distribution means for circulating the water to be treated, the cleaning liquid supply means for supplying the cleaning liquid from the secondary side to the primary side to the filtration member. A pressure difference measuring means to be performed, a determination means that can set a predetermined threshold and determine the activation of the cleaning liquid supply means based on the set threshold, and an activation of the cleaning liquid supply means based on the determination result by the determination means. The determination means calculates the threshold value based on the measured pressure difference value, which is the pressure difference measured by the pressure difference measuring means, after the supply of the cleaning liquid by the cleaning liquid supply means is completed. It is configured to reset to a value obtained by adding a predetermined offset value to the pressure difference value after cleaning.

By the way, in recent years, from the viewpoint of improving the efficiency of filtration membrane management, a management device installed in a water treatment plant that manages a membrane separation device so as to switch between filtration operation and relaxation operation at predetermined time intervals, A remote monitoring system is being constructed equipped with a remote monitoring device that is communicably connected to the management device and that collects and manages the intermembrane differential pressure of each filter membrane sampled by each management device at predetermined time intervals. ..

Based on the time-series intermembrane differential pressure of each membrane separation device collected by the remote monitoring device, a trend graph of the intermembrane differential pressure is displayed on the monitor, and multiple observers visually observe the trend graph. It is configured to determine whether or not an abnormality has occurred in the filtration membrane of the membrane separation device.

Japanese Unexamined Patent Publication No. 2011-31145

However, as the number of membrane separation devices to be monitored increases, the processing load of the observer increases, and the observer is allowed to visually check the individual trend graphs to judge whether they are abnormal or normal. Time was limited, and it was becoming difficult in terms of accurate and quick judgment.

Therefore, it is conceivable to increase the number of observers, but it takes a long time to train many observers to make stable judgments of the same quality, and the judgment results will inevitably fluctuate due to individual differences. There was a problem. Since the operation mode of the membrane separation device is different for each facility and the tendency of the trend graph of the intermembrane differential pressure displayed on the display device is also different, it is normal or abnormal in a short time. This is because sufficient experience is required to be able to properly judge whether or not.

In addition, there was an inconvenience that even if an abnormality occurred at night or on holidays when the guard was absent, it could not be found.

An object of the present invention is to reduce the burden on the observer in view of the above-mentioned problems, and to make an abnormality of the immersion type membrane separation device capable of quickly and accurately diagnosing the state of the filtration membrane even in the absence of the observer. The point is to provide a detection method.

In order to achieve the above object, the first characteristic configuration of the abnormality detection method of the immersion type membrane separation device according to the present invention is the abnormality detection method of the immersion type membrane separation device in which the filtration operation and the relaxation operation are alternately repeated. Of the data collection step of measuring and collecting the intermembrane differential pressure of the immersion type membrane separation device at predetermined time intervals and the time-series intermembrane differential pressure measurement value collected in the data collection step, the target for detecting an abnormality. A state separation step of separating the intermembrane differential pressure measurement value group of the first period including the period into the first intermembrane differential pressure measurement value group during the filtration operation and the second intermembrane differential pressure measurement value group during the relaxation operation. And / or the noise removing step of removing noise data from the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group, and the first intermembrane noise removing noise in the noise removing step. For the differential pressure measurement value group and / or the second membrane differential pressure measurement value group, the dispersion state is calculated from the differential pressure measurement value group in the second period shorter than the first period, and the dispersion state is obtained. It is a point including an abnormality determination step of determining whether or not the immersion type membrane separation device is abnormal based on the above.

One type of time-series data called the intermembrane differential pressure measurement value group, which is a set of intermembrane differential pressure measurement values obtained by measuring the intermembrane differential pressure at predetermined time intervals, is collected in the data collection process, and the state separation process is executed. As a result, the intermembrane differential pressure measurement value group corresponding to the first period, that is, the period including the target period for detecting an abnormality, corresponds to the first intermembrane differential pressure measurement value group during the filtration operation and the second membrane corresponding to the relaxation operation. It is separated from the differential pressure measurement value group. Abnormality after noise data that fluctuates the abnormality determination is removed by executing the noise removal step for the separated first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group. The determination process is executed. In the abnormality determination step, the dispersion state is calculated from the intermembrane differential pressure measurement value group in the second period shorter than the first period, and it is determined whether or not the immersion type membrane separation device is abnormal based on the dispersion state. Will be done.

In the second feature configuration, in addition to the first feature configuration described above, the noise reduction step is performed in each of the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group. On the other hand, it is a process of removing data deviating from the upper and lower predetermined ranges.

By removing data that deviates from the upper and lower predetermined ranges for each of the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group, for example, the output of a sensor that measures the intermembrane differential pressure. Noise data that fluctuates the abnormality determination, such as electromagnetic noise mixed in the signal line and data of the transition time when the filtration operation and the relaxation operation are switched, is effectively removed.

The third characteristic configuration is a method for detecting an abnormality in an immersion type membrane separation device that alternately repeats a filtration operation and a relaxation operation, and measures and collects the intermembrane differential pressure of the immersion type membrane separation device at predetermined time intervals. The data collection step, the noise removal step of removing noise data from the time-series differential pressure measurement values collected in the data collection step, and the intermembrane differential pressure measurement value of removing noise in the noise removal step. Of the normalization processing step that normalizes based on the maximum and minimum measured values in the normal state sampled in advance and the intermembrane differential pressure measurement value normalized in the normalization processing step, the target period for detecting an abnormality is set. A state separation step of separating the intermembrane differential pressure measurement value group in the first period including the first intermembrane differential pressure measurement value group during the filtration operation and the second intermembrane differential pressure measurement value group during the relaxation operation. The differential pressure measurement value in the second period shorter than the first period with respect to the first membrane differential pressure measurement value group and / or the second membrane differential pressure measurement value group separated in the state separation step. The point includes an abnormality determination step of calculating a dispersion state from the group and determining whether or not the immersion type membrane separation device is abnormal based on the dispersion state.

For the immersion type membrane separation device that is the target of abnormality detection, the intermembrane pressure is measured at predetermined time intervals in the data collection process, the noise data is removed in the noise removal process, and the noise data is removed. The normalization processing step is executed based on the maximum measured value and the minimum measured value of the intermembrane differential pressure during normal operation of the immersion type membrane separation device sampled in advance with respect to the pressure measured value. The state separation step is executed for the normalized intermembrane differential pressure measurement value, and the intermembrane differential pressure measurement value group in the first period including the target period for detecting an abnormality is the first membrane during the filtration operation. It is separated into a differential pressure measurement value group and a second intermembrane differential pressure measurement value group during relaxation operation, and an abnormality is determined for the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group. The process is carried out. That is, the dispersion state is calculated from the differential pressure measurement value group in the second period shorter than the first period, and it is determined whether or not the immersion type membrane separation device is abnormal based on the dispersion state. In the intermembrane differential pressure measurement value group measured in the data acquisition process, a certain amount of fluctuation different from noise is seen, and if the fluctuation is within the normal range, it is erroneously judged as abnormal. It is possible to avoid such a situation.

The fourth feature configuration is that, in addition to the third feature configuration described above, the noise reduction step is a step of removing data deviating from the upper and lower predetermined ranges with respect to the intermembrane differential pressure measurement value. be.

Appropriate abnormalities are obtained by removing data that deviates from the upper and lower predetermined ranges by the noise reduction process as noise that is not the target of the original abnormality judgment for multiple intermembrane differential pressure measurement values measured in the data collection process. You will be able to judge.

In the fifth feature configuration, in addition to any of the first to fourth feature configurations described above, the state separation step is performed by performing a non-hierarchical cluster analysis based on unsupervised machine learning. The point is that it is a step of separating the intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group.

By adopting non-hierarchical cluster analysis based on unsupervised machine learning, it becomes possible to appropriately separate the states into the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group.

In the sixth feature configuration, in addition to any of the first to fifth feature configurations described above, the dispersion state calculated in the abnormality determination step is the intermembrane differential pressure measurement value group in the second period. It is an output value of a predetermined evaluation function having a variance or standard deviation as a variable, and when the output value exceeds a preset threshold value, it is determined that the operating state is abnormal.

As the dispersion state used for abnormality determination, it is preferable to adopt the output value of a predetermined evaluation function whose variable is the dispersion or standard deviation with respect to the intermembrane differential pressure measurement value group in the second period, and the output value is set in advance. When the set threshold is exceeded, it becomes possible to appropriately determine that the operating state is abnormal. It is noted that if the stability is impaired for some reason in the second period, the variance or standard deviation of the differential pressure measurement value group in the second period becomes large. The evaluation function is not particularly limited, and a linear function or the like having a variance or a standard deviation as a variable can be appropriately used.

In the seventh feature configuration, in addition to any of the first to sixth feature configurations described above, a series of steps from the state separation step to the abnormality determination step is repeatedly executed every third period. The first period is set to be longer than the sum of the second period and the third period.

Since the abnormality determination step is repeated from the state separation step every time a third period shorter than the first period elapses in the first period including the target period for detecting an abnormality, the first period longer than the third period Monitoring is performed by executing an abnormality determination process every time a third period shorter than the first period elapses, while ensuring the accuracy of the state separation by the state separation process using the intermembrane differential pressure measurement value group of the period. Abnormalities can be determined according to the timing of monitoring by personnel.

In the eighth feature configuration, in addition to any of the first to seventh feature configurations described above, in the abnormality determination step, each period in which the second period is shifted in time series at predetermined time intervals. It is in the point of executing with.

By shifting the second period in time series at predetermined time intervals, it becomes possible to determine the abnormality in the entire target period for detecting the abnormality.

As described above, according to the present invention, there is an abnormality detection method for an immersion type membrane separation device that can reduce the burden on the observer and can quickly and accurately diagnose the state of the filtration membrane even in the absence of the observer. Can now be provided.

Explanatory drawing of remote monitoring system to which this invention is applied The first aspect of the abnormality detection method of the immersion type membrane separation apparatus is shown, (a) is an explanatory diagram of the membrane differential pressure measurement value group collected in a data collection step, and (b) is state separation in a state separation step. Explanatory drawing of the first intermembrane differential pressure measurement value group during the filtration operation and the second intermembrane differential pressure measurement value group during the relaxation operation The first aspect of the abnormality detection method of the immersion type membrane separation apparatus is shown, FIG. An explanatory diagram of the arithmetic processing of the distributed state, (c) is an explanatory diagram of the evaluation value calculated based on the distributed state. A flowchart showing the processing procedure of the first aspect of the abnormality detection method of the immersion type membrane separation device. The first aspect of the abnormality detection method of the immersion type membrane separation apparatus is shown, (a) is an explanatory diagram of the membrane differential pressure measurement value group collected in a data collection step, and (b) is state separation in a state separation step. An explanatory diagram of the first intermembrane differential pressure measurement value group during the filtration operation and the second intermembrane differential pressure measurement value group during the relaxation operation, (c) is the output characteristics and abnormality determination of the evaluation function determined in the abnormality determination process. Explanatory diagram of the threshold A flowchart showing the processing procedure of the second aspect of the abnormality detection method of the immersion type membrane separation device. The second aspect of the abnormality detection method of the immersion type membrane separation device is shown, (a) is an explanatory diagram of a group of measured differential pressures between membranes collected in a data collection step, and (b) is a state separation in a state separation step. An explanatory diagram of the first intermembrane differential pressure measurement value group during the filtration operation and the second intermembrane differential pressure measurement value group during the relaxation operation, (c) is the output characteristics and abnormality determination of the evaluation function determined in the abnormality determination process. Explanatory diagram of the threshold

Hereinafter, embodiments of the abnormality detection method of the immersion type membrane separation device according to the present invention will be described.
FIG. 1 shows a remote monitoring system to which an abnormality detection method of an immersion type membrane separation device is applied.

The remote monitoring system 100 is an immersion type membrane separation device 2 (2A, 2B, ..., 2N) installed in the water treatment plant 1 (1A, 1B, ..., 1N) (hereinafter, simply "membrane separation device". 2 ”) is managed by the management device 3 (3A, 3B, ..., 3N) and the management device 3 (3A, 3B, ..., 3N) are connected to each other via the Internet or the like. It is equipped with a remote monitoring device 4 which is a cloud server and a diagnostic device 10. The diagnostic device 10 may be a remote monitoring device 4 having a function thereof, or may be connected so as to be communicable via the Internet or the like like the remote management device 4.

Each control device 3 (3A, 3B, ..., 3N) is a single or multiple series of membrane separation devices 2 (2A, 2B) installed in the water treatment plant 1 (1A, 1B, ..., 1N). , ..., 2N), for example, 9 minutes of filtration operation and 1 minute of relaxation operation are repeated as one unit.

During the filtration operation, the treated water is suction-filtered from the filter membrane in a state of being diffused by the air diffuser installed under the membrane separation device 2, and during the relaxation operation, the treated water is suction-filtered from the filter membrane while maintaining the state of being diffused by the air diffuser during the relaxation operation. Stop suction filtration of the treated water. Even during the relaxation operation, the filtration membrane surface is cleaned by the upward flow of the gas-liquid mixed phase of the air bubbles from the air diffuser installed below the membrane separation device 2 and the water to be treated.

Each management device 3 measures the intermembrane differential pressure at 1-minute intervals via a pressure sensor and stores it in an internal storage unit, and at the same time, sampling data of the intermembrane differential pressure measured via the Internet or the like at predetermined time intervals. It is configured to transmit the operation information including the above to the remote monitoring device 4.

The remote monitoring device 4 stores the sampling data of the intermembrane differential pressure transmitted from each management device 3, that is, the measured value of the intermembrane differential pressure in the database 4A, and also browses using an information terminal such as a smartphone from a person in charge at the site. Upon request, it is configured to provide sampling data of the intermembrane differential pressure stored in the database 4A, for example, in the form of a trend graph.

The remote monitoring device 4 includes an immersion type membrane separation device 2 (2A, 1N) installed in each water treatment plant 1 (1A, 1B, ..., 1N) based on the measured differential pressure between membranes stored in the database 4A. An abnormality detection program that detects whether 2B, ..., 2N) is operating normally or some abnormality has occurred is installed.

That is, the remote monitoring device 4 also serves as the diagnostic device 10, and each management device 3 and the remote monitoring device 4 execute the abnormality detection method of the immersion type membrane separation device of the present invention. The diagnosis result is stored in the database 4A and is browsed in response to a browsing request using an information terminal such as a smartphone from a person in charge at the site. The diagnostic device 10 that executes the abnormality detection method of the immersion type membrane separation device is configured separately from the remote monitoring device 4, and is composed of a general-purpose personal computer that can communicate with the remote monitoring device 4 via the Internet or the like. May be good.

Hereinafter, the first aspect of the abnormality detection method of the immersion type membrane separation device will be described in detail.
The abnormality detection method of the immersion type membrane separation device includes a data collection step, a state separation step, a noise removal step, and an abnormality determination step.

The data collection step is a step of measuring and collecting the intermembrane differential pressure of each immersion type membrane separation device 2 at predetermined time intervals. Intermembrane differential pressure sampled at 1-minute intervals is collected by each management device 3 via a pressure sensor and stored in an internal storage unit. The stored differential pressure between the membranes is transmitted to the remote monitoring device 4 via the Internet at a predetermined time, for example, about once every half a day, and is stored in the database 4A provided in the remote monitoring device 4.

Other data include an ID code uniquely identifying the water treatment plant 1, an ID code uniquely identifying the membrane separation device, and a sampling time of the differential pressure between the membranes. When a plurality of systems of membrane separation devices 2 are installed in the water treatment plant 1, an ID code for each system of membrane separation devices 2 is included.

FIG. 2A exemplifies a plurality of intermembrane differential pressure measurement values sampled at 1-minute intervals, that is, a group of intermembrane differential pressure measurement values. The figure shows a state in which the measured differential pressure between membranes sampled in time series and plotted is connected by a straight line, and the measured differential pressure between membranes during both filtration operation and relaxation operation are mixed. In addition, a noise signal mixed in the signal line of the pressure sensor is superimposed. The measured differential pressure between membranes is a negative value, and the lower the value in the figure, the larger the absolute value of the measured differential pressure between membranes. Therefore, in the following description, it is expressed that the lower the intermembrane differential pressure is, the higher the intermembrane differential pressure is.

In the state separation step, among the time-series intermembrane differential pressure measurement values collected in the data collection step, the intermembrane differential pressure measurement value group of the first period P1 including the target period for detecting an abnormality is filtered during the filtration operation. This is a step of separating the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group during relaxation operation.

The state separation process is configured to separate into two clusters, a first intermembrane differential pressure measurement group and a second intermembrane differential pressure measurement group, by performing non-hierarchical cluster analysis based on unsupervised machine learning. Has been done. The k-means method can be preferably used as a non-hierarchical cluster analysis.

In the k-means method, first, the temporary center of gravity positions of the two clusters are initially set, and those that are close to each other based on the Euclidean distance between the measured differential pressure between the membranes and the temporary center of gravity are separated as clusters belonging to each temporary center of gravity position and are the same. A new center of gravity is calculated from the measured value of the differential pressure between the membranes contained in the cluster, then the Euclidean distance between the new center of gravity and the measured value of the differential pressure between the membranes is calculated, and the measured value of the differential pressure between the membranes is on the side closer to the distance. By repeating the process of separating into the cluster of the center of gravity until it converges, it is separated into the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group.

FIG. 2B shows the intermembrane differential pressure measurement value group of FIG. 2A, the first intermembrane differential pressure measurement value group during the filtration operation and the second membrane during the relaxation operation using the k-means method. The results of separation into the differential pressure measurement value group are shown. The high-concentration portion indicates the second intermembrane differential pressure measurement value group, and the low-concentration portion indicates the first intermembrane differential pressure measurement value group. Isolated points discrete from the mass are scattered in each of the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group. The isolated point becomes noise. The noise includes a noise signal mixed in the signal line of the pressure sensor and an intermembrane differential pressure detected at the transition from the filtration operation to the relaxation operation or at the transition from the relaxation operation to the filtration operation.

The noise removal step is a step of removing noise data from the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group in the second period P2 shorter than the first period P1. Then, it is a step of removing data deviating from the upper and lower predetermined ranges for each of the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group.

For example, the data of the upper 95% or more with a high value and the data of the lower 5% with a low value with respect to the number of data of the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group It is removed as noise. Further, for example, it is possible to remove data deviating up and down from an appropriate range set to 5% to 95% of the range in which the measured value of the differential pressure between membranes can be taken as noise. It should be noted that this range is not particularly limited to the range of 5% to 95%, and may be appropriately set.

The abnormality determination step is a dispersed state from the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group of the second period P2 shorter than the first period P1 in which noise is removed in the noise removal step. Is a step of determining whether or not the immersion type membrane separation device is abnormal based on the dispersed state.

Abnormality diagnosis focused on filtration operation by separating into a cluster that is a group of first intermembrane differential pressure measurement values during filtration operation and a cluster that is a group of second intermembrane differential pressure measurement values during relaxation operation. , You will be able to perform focused abnormality diagnosis during relaxation driving.

FIG. 3A exemplifies a group of measured values of the first intermembrane differential pressure during the filtration operation separated in the state separation step, and FIG. 3B shows the noise removal step and the dispersion state is calculated. The second period P2 to be used is indicated by a dashed line or a solid rectangle. In the first intermembrane differential pressure measurement value group, the dispersion state is calculated after the noise reduction step is performed on the intermembrane differential pressure measurement value group included in the time width of the quadrangle of the broken line or the solid line. The broken line rectangle indicates the second period P2 executed in the past, and the solid line rectangle indicates the second period P2 currently being executed.

In the present embodiment, the first period P1 is set to 24 hours (1 day) and the second period P2 is set to 1 hour, but the value is not limited to this. Further, the noise reduction step is performed and the dispersion state is calculated in each period in which the second period P2 is time-series shifted at predetermined time intervals (for example, every minute) shorter than the second period P2. For example, when the dispersion state is calculated in each period in which the second period P2 is shifted in time series in 1-minute units, 60 dispersion states are calculated in 1 hour. The intermembrane differential pressure measurement value group used in the noise reduction step of each period of the second period P2 is the original data before noise reduction.

As shown in FIG. 3C, as the dispersion state used for abnormality determination, the output value of a predetermined evaluation function whose variable is the dispersion or standard deviation with respect to the intermembrane differential pressure measurement value group in the second period P2. That is, it is preferable to adopt an evaluation value, and when the output value exceeds a preset threshold value Ref (abnormality determination threshold value), it is determined that the operating state is abnormal.

The evaluation function is not particularly limited, and a linear function having a variance or a standard deviation as a variable can be appropriately used. In this embodiment, a linear function that multiplies the variance or standard deviation by a predetermined coefficient is adopted. When the coefficient is 1, the variance or standard deviation is the evaluation value.

According to FIG. 3B, it is determined that the evaluation value exceeds the abnormality determination threshold value, that is, an abnormality has occurred in the region R in which the variation in the measured value of the differential pressure between the membranes gradually increases. It is noted that if the stability is impaired in the second period P2 for some reason, the variance or standard deviation of the differential pressure measurement value group in the second period P2 becomes large.

As shown by the broken line arrow in FIG. 3B, it is preferable to configure the series of steps from the state separation step to the abnormality determination step to be repeatedly executed every time the third period P3 elapses. In this case, the target first period P1 (24 hours) is the first period P1 (24 hours) after the lapse of the third period P3 (6 hours) from the previously targeted first period P1. ). The first period P1 is set to a period longer than the sum of the second period P2 and the third period P3.

Since the abnormality determination step is repeated from the state separation step every time the third period P3, which is shorter than the first period P1, elapses in the first period P1 including the target period for detecting the abnormality, it is longer than the third period P3. Predetermined such as adjusting the third period P3 to the timing of monitoring by the observer while ensuring the accuracy of the state separation by the state separation process using the first membrane differential pressure measurement value group of the set first period P1. By setting the period of, even when the number of immersion type membrane separation devices to be diagnosed increases, the calculation load of the diagnostic device 10 can be distributed. In this example, the third period P3 is set to 6 hours. Therefore, a series of steps from the state separation step to the abnormality determination step is repeated every 6 hours.

Further, the same processing is performed on the second intermembrane differential pressure measurement value group, and abnormality determination is performed for each of the filtration operation and the relaxation operation. Since the type of abnormality may differ between the abnormality captured from the first intermembrane differential pressure measurement value group and the abnormality captured from the second intermembrane differential pressure measurement value group, the first and second intermembrane differential pressures may be different. It is not essential to execute a series of steps for both of the measured value groups, and it is sufficient that a series of steps are executed for any of the abnormal states to be captured.

FIG. 4 shows the procedure of the abnormality detection method of the above-mentioned immersion type membrane separation device.
That is, the abnormality detection method of the immersion type membrane separation device that alternately repeats the filtration operation and the relaxation operation includes a data collection step of measuring and collecting the intermembrane differential pressure of the immersion type membrane separation device at predetermined time intervals (S1). Among the time-series intermembrane differential pressure measurement values collected in the data collection process, the intermembrane differential pressure measurement value group of the first period P1 including the target period for detecting an abnormality is the first intermembrane difference during the filtration operation. The state separation step (S2) of separating into the pressure measurement value group and the second membrane differential pressure measurement value group during the relaxation operation is executed.

After that, a second period P2 shorter than the first period P1 is set (S3), and noise data is obtained from the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group in the second period P2. The noise removal step of removing noise is executed (S4), and the dispersion state is calculated for the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group in the second period P2 in which the noise is removed. Processing is executed (S5).

In order to execute each process from step S3 to step S5 in the entire period of the target period, the second period P2 is shifted in chronological order at predetermined time intervals and repeated (S6), and based on each distributed state obtained. An abnormality determination step of determining whether or not the immersion type membrane separation device is abnormal is executed based on a predetermined abnormality determination threshold value (S7, S8, S9, S10).

In the flowchart shown in FIG. 4, the second period P2 is shifted at predetermined time intervals so as to cover the entire period of the first period P1, and the evaluation value is determined to be abnormal in any one of the second period P2. An example of determining that an abnormality has occurred when the threshold value is exceeded has been shown, but it is configured to determine that an abnormality has occurred when the number of times the evaluation value exceeds the abnormality determination threshold value reaches a predetermined number of times in the second period P2. May be good.

Further, as described with reference to the broken line arrow in FIG. 3B, the above-mentioned abnormality determination process may be repeated in a third period shorter than the first period P1.

Hereinafter, the second aspect of the abnormality detection method of the immersion type membrane separation device will be described in detail. In the following description, the differences from the first aspect will be mainly described in detail.
The abnormality detection method of the immersion type membrane separation device includes a data collection step, a noise removal step, a normalization processing step, a state separation step, and an abnormality determination step.

In the data collection step, the intermembrane differential pressure of the immersion type membrane separation device is measured and collected at predetermined time intervals. In the noise reduction step, noise data is removed from the time-series differential pressure measurement values collected in the data collection step. In the noise reduction step, data deviating from the upper and lower predetermined ranges with respect to the measured value of the differential pressure between the films is removed. Similar to the first aspect, for example, the data of the upper 95% or more having a high value and the data of the lower 5% having a low value with respect to the number of data of the intermembrane differential pressure measurement value group are removed as noise. Further, for example, it is possible to remove data deviating up and down from an appropriate range set to 5% to 95% of the range in which the measured value of the differential pressure between membranes can be taken as noise. It should be noted that this range is not particularly limited to the range of 5% to 95%, and may be appropriately set.

In the normalization process, the intermembrane differential pressure measurement value collected in the data acquisition process based on the maximum and minimum measured values in the normal state sampled in advance for the immersion type membrane separation device that is the target of abnormality determination. The group is normalized. For example, 0 kPa is set during the relaxation operation and -1 kPa is set during the filtration operation.

In the state separation step, among the intermembrane differential pressure measurement values normalized in the normalization processing step, the intermembrane differential pressure measurement value group in the first period P1 including the target period for detecting an abnormality is the first in the filtration operation. It is separated into one intermembrane differential pressure measurement value group and a second intermembrane differential pressure measurement value group during relaxation operation. Similar to the first aspect, in the state separation step, the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group are performed by performing non-hierarchical cluster analysis based on unsupervised machine learning. It is configured to be separated into two clusters. The k-means method can be preferably used as a non-hierarchical cluster analysis.

In the abnormality determination step, as in the first aspect, it is shorter than the first period with respect to the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group separated in the state separation step. The dispersion state is calculated from the differential pressure measurement value group in the second period, and it is determined whether or not the immersion type membrane separation device is abnormal based on the dispersion state.

As shown in FIG. 5A, depending on the immersion type membrane separation device, suction by a suction pump corresponds to a time zone in which the amount of water to be treated is large during the filtration operation in which the treated water is suction-filtered from the filtration membrane. The amount may be switched in two stages. In this example, it is −20 kPa during the relaxation operation, about −65 kPa during the filtration operation with a small pump suction amount, and about −85 kPa during the filtration operation with a large pump suction amount.

As shown in FIG. 5B, when the state separation step described in the first aspect is executed for the intermembrane differential pressure measured value exhibiting such characteristics, the first intermembrane differential pressure during the filtration operation is executed. In the measured value group, the data varies due to the switching of the suction amount by the suction pump. Therefore, as shown in FIG. 5 (c), even if the behavior is originally normal, in the regions R1 and R2 where the variation is large. In the abnormality determination step, the evaluation value exceeds the abnormality determination threshold value, which may lead to erroneous determination as an abnormality.

Further, even in the immersion type membrane separation device in which the suction amount by the pump is maintained constant, when the water level of the treatment tank in which the filtration membrane is immersed fluctuates, the differential pressure between the first membranes during the filtration operation is measured accordingly. The value group may fluctuate, leading to the same misjudgment. Further, even when the intermediate value of the differential pressure between the membranes during the filtration operation and the relaxation operation is measured at the time of switching between the filtration operation and the relaxation operation, the same erroneous determination may occur.

Even in such a case, if the abnormality detection method of the immersion type membrane separation device according to the second aspect described above is adopted, the abnormality can be appropriately determined without causing an erroneous determination.

FIG. 6 shows a procedure for detecting an abnormality in the immersion type membrane separation device according to the second aspect.
That is, the abnormality detection method of the immersion type membrane separation device, which repeats the filtration operation and the relaxation operation alternately, measures the intermembrane differential pressure with respect to the normal immersion type membrane separation device for the normalization process required later. The normal intermembrane differential pressure measurement data acquisition step of acquiring the maximum and minimum measured values is executed in advance (S100), and then the membrane of the immersion type membrane separation device is used for abnormality determination. A data collection step of measuring and collecting the differential pressure at predetermined time intervals is executed (S101).

A noise removal step of removing noise data from the time-series interfilm differential pressure measurement values collected in the data collection step is executed (S102), and the intermembrane differential pressure measurement values of which noise is removed in the noise removal step are measured at normal times. A normalization processing step of normalizing based on the maximum measured value and the minimum measured value sampled in advance in the intermembrane differential pressure measurement data collecting step is executed (S103).

Next, among the intermembrane differential pressure measurement values normalized in the normalization processing step, the intermembrane differential pressure measurement value group of the first period P1 including the target period for detecting an abnormality is the first membrane during the filtration operation. A state separation step of separating into a differential pressure measurement value group and a second membrane differential pressure measurement value group during relaxation operation is executed (S104).

Further, a second period P2 shorter than the first period P1 is set (S105), and the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group of the second period P2 is set. The process of calculating the distributed state is executed (S106).

In order to execute each process of step S105 and step S106 in the entire period of the target period, the second period P2 is shifted in chronological order at predetermined time intervals and repeated (S107), and based on each distributed state obtained. An abnormality determination step of determining whether or not the immersion type membrane separation device is abnormal is executed based on a predetermined abnormality determination threshold value (S108, S109, S110, S111).

In the state separation step, as in the first aspect, the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group are performed by performing non-hierarchical cluster analysis based on unsupervised machine learning. The dispersion state calculated in the abnormality determination step is the output value of a predetermined evaluation function whose variable is the dispersion or standard deviation with respect to the intermembrane differential pressure measurement value group in the second period. When the value exceeds a preset threshold, it is determined that the operating state is abnormal.

Further, as in the first aspect, a series of steps from the state separation step to the abnormality determination step are repeatedly executed every third period P3, and the first period P1 is the second period P2. The third period is set to be longer than the period obtained by adding P3. Further, the abnormality determination step is executed in each period in which the second period P2 is shifted in time series at predetermined time intervals.

FIG. 7A shows a characteristic diagram in which the intermembrane differential pressure measurement value group of FIG. 5A processed in the first aspect is normalized. It is normalized to 0.0 kPa during relaxation operation and -1.0 kPa during filtration operation when the suction amount of the pump is large. FIG. 7B shows the result of state separation of the data shown in FIG. 7A. By the normalization process, the variance or standard deviation calculated for the fluctuation of the suction amount of the pump does not fluctuate so much, and by setting the abnormality judgment threshold value appropriately, erroneous judgment does not occur. ..

It is necessary to determine which of the first aspect and the second aspect described above is preferable based on the installation environment of each immersion type membrane separation device. Is it an environment where multiple suction pumps for drawing out treated water are provided and the number of operating units is switched, or is it an environment where even one suction pump is operated by switching the suction amount up or down? It is necessary to determine in advance according to the individual environment, such as whether the environment is such that the water level of the water tank in which the filter membrane is immersed fluctuates greatly even if the value is constant.

The above description is an example of the abnormality detection method of the immersion type membrane separation device according to the present invention, and the specific embodiment of each step can be appropriately modified and designed within the range in which the action and effect of the present invention are exhibited. Needless to say.

100: Remote monitoring system 1 (1A, 1B, 1C): Water treatment plant 2: Immersion type membrane separation device 3: Management device 4: Remote monitoring device (diagnostic device)
4A: Database

Claims (8)

It is an abnormality detection method for an immersion type membrane separation device that alternately repeats filtration operation and relaxation operation.
A data acquisition step of measuring and collecting the intermembrane differential pressure of the immersion type membrane separation device at predetermined time intervals, and
Among the time-series intermembrane differential pressure measurement values collected in the data acquisition step, the intermembrane differential pressure measurement value group in the first period including the target period for detecting an abnormality is the first intermembrane difference during the filtration operation. A state separation process that separates the pressure measurement value group and the second membrane differential pressure measurement value group during relaxation operation, and
The first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group with respect to the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group. Noise removal process to remove noise data from
The differential pressure in the second period shorter than the first period with respect to the first intermembrane differential pressure measurement value group and / or the second intermembrane differential pressure measurement value group from which noise was removed in the noise removal step. An abnormality determination step of calculating the dispersion state from the measured value group and determining whether or not the immersion type membrane separation device is abnormal based on the dispersion state, and
Anomaly detection method for immersion type membrane separation devices including. The noise removing step is a step of removing data deviating from the upper and lower predetermined ranges with respect to each of the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group. Abnormality detection method for immersion type membrane separation device. It is an abnormality detection method for an immersion type membrane separation device that alternately repeats filtration operation and relaxation operation.
A data acquisition step of measuring and collecting the intermembrane differential pressure of the immersion type membrane separation device at predetermined time intervals, and
A noise reduction step of removing noise data from the time-series differential pressure measurement values collected in the data collection step, and a noise reduction step.
A normalization process that normalizes the measured differential pressure between membranes from which noise has been removed in the noise reduction step based on the maximum and minimum measured values in the normal state sampled in advance.
Among the intermembrane differential pressure measurement values normalized in the normalization processing step, the intermembrane differential pressure measurement value group in the first period including the target period for detecting an abnormality is the first intermembrane differential pressure during the filtration operation. A state separation process that separates the measured value group into the second intermembrane differential pressure measured value group during relaxation operation, and
The differential pressure measurement value in the second period shorter than the first period with respect to the first membrane differential pressure measurement value group and / or the second membrane differential pressure measurement value group separated in the state separation step. An abnormality determination step of calculating the dispersion state from the group and determining whether or not the immersion type membrane separation device is abnormal based on the dispersion state, and
Anomaly detection method for immersion type membrane separation devices including. The abnormality detection method for an immersion type membrane separation device according to claim 3, wherein the noise removing step is a step of removing data deviating from a predetermined range above and below the measured value of the differential pressure between the membranes. The state separation step is a step of separating into the first intermembrane differential pressure measurement value group and the second intermembrane differential pressure measurement value group by performing non-hierarchical cluster analysis based on unsupervised machine learning. Item 4. The method for detecting an abnormality in the immersion type membrane separation device according to any one of Items 1 to 4. The dispersion state calculated in the abnormality determination step is an output value of a predetermined evaluation function whose variable is the dispersion or standard deviation with respect to the intermembrane differential pressure measurement value group in the second period, and the output value is provided in advance. The method for detecting an abnormality in the immersion type membrane separation device according to any one of claims 1 to 5, wherein it is determined that the operating state is abnormal when the threshold value is exceeded. A series of steps from the state separation step to the abnormality determination step are repeatedly executed every third period, and the first period is the sum of the second period and the third period. The method for detecting an abnormality in the immersion type membrane separation device according to any one of claims 1 to 6, which is set to a period longer than the period. The abnormality detection method for an immersion type membrane separation device according to any one of claims 1 to 7, wherein the abnormality determination step is executed in each period in which the second period is shifted in time series at predetermined time intervals.

Images