Coagulant injection control device, coagulant injection control method, and computer program

Abstract

To provide a coagulant injection control device that minimizes an operating cost of a water treatment plant while maintaining appropriate water quality.SOLUTION: A coagulant injection control device according to an embodiment includes: a calculation unit that calculates an evaluation value of an evaluation function that changes with a first operation amount of a control target and indicates an index for optimization of the control target by using multiple measured values obtained from the control target; a phase delay estimation unit that calculates a phase delay estimation value from the first operation amount in the control target to the evaluation value by using the values of multiple components included in the evaluation value and a phase compensation parameter corresponding to each component; a search unit that calculates gradient information of the evaluation function being the change rate with respect to the first operation amount of the evaluation value by using the phase delay estimation value and information on the evaluation value, and determines an optimal operation amount of the first operation amount by integrating the gradient information; and a control unit that sets the optimal operation amount as the target value of a control amount of the control target, calculates and outputs a second operation amount so that the control amount follows the target value.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] An embodiment of the present invention relates to a coagulant injection control device, a coagulant injection control method, and a computer program. [Background technology]

[0002] In water treatment plants, a flocculant injection control device is used to control the amount of flocculant injected into the raw water to be treated. Conventionally, for example, in water purification plants, sewage treatment plants, or industrial wastewater treatment facilities, a flocculant is injected into the raw water to be treated to flocculate suspended solids contained in the raw water. The generated flocs are then separated by settling to remove suspended solids contained in the raw water. Furthermore, the sedimentation-treated water after this operation is filtered by a sand filter to remove fine flocs that were not removed by settling.

[0003] In coagulant injection control, if the amount of coagulant injected into the raw water is small, the water quality after water treatment cannot be properly maintained. If the amount of coagulant injected into the raw water is large, the water quality after water treatment can be properly maintained, but it is difficult to suppress the increase in plant operating costs, such as the cost of the coagulant, the cost required for sludge treatment, and the cost of cleaning the tank called the sand filter basin. For this reason, it is desirable to appropriately control the injection rate of the coagulant.

[0004] In controlling the coagulant injection rate, for example, a method has been proposed in which the average value of the measured electrophoretic velocity of flocs and a target value of the charge state are used to control the coagulant injection rate. The control method is a general feedback control, such as P control, PI control, or PID control. In the feedback control, a control target value of the floc charge state is set.

[0005] In setting the control target value of the floc charge state, the use of "extreme value control or extremum seeking control (ESC)", which has been used in recent years as a means of solving the problem in optimal control of plants, has been proposed. Extremum value control (ESC) calculates the value of the evaluation function to be optimized from online sensor information of the plant that can directly measure it, and adaptively searches for the optimum value (minimum or maximum) by changing the manipulated variable so as to maintain the evaluation function value at the optimum value. Extremum value control has attracted attention as a practical optimal control technique for processes involving complex phenomena that are difficult to model using mathematical formulas (e.g. water purification plant processes, sewage treatment processes, combustion processes, petrochemical processes, etc.). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2017-033104 A [Patent Document 2] JP 2020-142188 A Summary of the Invention [Problem to be solved by the invention]

[0007] The embodiments of the present invention have been made in consideration of the above-mentioned circumstances, and aim to provide a coagulant injection control device, a coagulant injection control method, and a computer program that minimize the operating costs of a water treatment plant while maintaining appropriate water quality. [Means for solving the problem]

[0008] A coagulant injection control device according to an embodiment of the present invention includes a process measurement value acquisition unit that acquires a plurality of measurement values ​​from a controlled object; a process evaluation value calculation unit that uses the measurement values ​​to calculate an evaluation value of an evaluation function that changes depending on a first manipulated variable of the controlled object and indicates an index regarding optimization of the controlled object; a phase lag estimation unit that calculates a phase lag estimate from the first manipulated variable in the controlled object to the evaluation value using the values ​​of each of a plurality of components included in the evaluation value and a phase compensation parameter corresponding to each of the plurality of components; an evaluation function gradient estimation unit that uses the phase lag estimate and information on the evaluation value to calculate gradient information of the evaluation function, which is the rate of change of the evaluation value relative to the first manipulated variable; an extreme value search unit that determines an optimal manipulated variable for the first manipulated variable by integrating the gradient information; and a control unit that calculates and outputs a second manipulated variable such that the optimal manipulated variable is a target value of a controlled variable of the controlled object so that the controlled variable follows the target value. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing an example of a water treatment plant in which a coagulant injection rate is controlled by a coagulant injection control device of the first embodiment. [Diagram 2] FIG. 2 is a diagram showing an example of a cell provided in a separation flow path in the water treatment plant shown in FIG. [Diagram 3] FIG. 3 is a diagram showing an example of a distribution map of the movement speed of flocs calculated by the analysis unit shown in FIG. [Figure 4] FIG. 4 is a diagram illustrating an example of the configuration of the control target value determination unit illustrated in FIG. [Diagram 5] FIG. 5 is a diagram showing in more detail a part of the control target value determination unit shown in FIG. [Figure 6] FIG. 6 is a schematic diagram showing an example of a system for adjusting a target value of the charge state of flocs by extreme value control in feedback control of the coagulant injection rate in a water purification plant. [Figure 7] FIG. 7 is a diagram for explaining an example of a result of verifying the application example of FIG. 6 by simulation. [Figure 8] FIG. 8 is a diagram for explaining an example of a result of verifying the application example of FIG. 6 by simulation. [Figure 9] FIG. 9 is a diagram for explaining the principle of extremum search by extremum control. [Figure 10] FIG. 10 is a diagram showing an example of changes in each cost included in the operating cost of a controlled plant and a control target value of the flocculant injection control device of one embodiment. [Figure 11] FIG. 11 is a diagram showing an example of changes in each cost included in the operating cost of a controlled plant and a control target value of the flocculant injection control device of one embodiment. [Figure 12] FIG. 12 is a diagram showing an example of changes in the operating cost (evaluation value) and the control target value of the controlled plant of the flocculant injection control device of one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, a flocculant injection control device, a flocculant injection control method, and a computer program according to embodiments will be described with reference to the drawings. FIG. 1 is a diagram showing an example of a water treatment plant in which a coagulant injection rate is controlled by a coagulant injection control device of the first embodiment.

[0011] A water treatment plant is a facility that realizes a solid-liquid separation process in which solids such as suspended solids contained in the water to be treated are flocculated by a flocculant and the flocculated solids are separated from the water to be treated by gravity settling. In the following, the water to be treated by the water treatment plant or the water being treated by the water treatment plant is referred to as "water to be treated," and water that has been treated by the water treatment plant and can be discharged or reused is referred to as "treated water." In the following, of the water to be treated that flows into the solid-liquid separation process from inside and outside the water treatment plant, the water to be treated immediately after the flow is referred to as "raw water." In other words, raw water is the water to be treated in an initial state, into which no flocculant has been injected.

[0012] Furthermore, the application of the flocculant injection control device of the embodiment described below is not limited to a specific water treatment plant or water treatment facility as long as it realizes the solid-liquid separation process as described above. For example, the application of the flocculant injection control device of the embodiment may be a water treatment plant such as a water purification plant, or may be a water treatment facility installed in various factories such as a paper factory or a food factory. For example, in a water purification plant, the raw water may be river water, dam lake water, groundwater, rainwater, sewage, etc. Furthermore, in industrial plants such as a paper factory or a food factory, the raw water may be industrial wastewater from these factories.

[0013] FIG. 1 shows a water treatment plant 100 that realizes a solid-liquid separation process in a water purification plant, as an example of an application of the above-mentioned coagulant injection control device. The water treatment plant 100 includes various facilities for realizing a solid-liquid separation process, and a coagulant injection control device 1. For example, the water treatment plant 100 includes, as facilities for realizing a solid-liquid separation process, a receiving well 3, a rapid mixing basin (mixing basin) 4, a flocculation basin 5, a settling basin 6, a filtration basin 7, a coagulant injection device 8, an analysis unit 9, a pH adjuster injection device 10, and a sludge basin 62. The water to be treated is conveyed from a river or a dam lake, and first arrives at the receiving well 3, and then is sent in this order to the rapid mixing basin 4, the flocculation basin 5, the settling basin 6, and the filtration basin 7. That is, the facility located most upstream in terms of the flow of the water to be treated is the receiving well 3, and the facility located most downstream is the filtration basin 7.

[0014] The receiving well 3 is a water tank that stores raw water flowing into the water treatment plant 100. In the receiving well 3, solid matter with a relatively large specific gravity, such as plants and soil, settles by gravity, and the supernatant water is sent to the downstream rapid mixing basin 4 as the water to be treated.

[0015] The receiving well 3 is equipped with a raw water quality meter 31. The raw water quality meter 31 measures the quality of the raw water that has arrived at the receiving well 3. Specifically, the raw water quality meter 31 measures index values ​​of water quality that may affect the treatment results of the solid-liquid separation process. For example, the raw water quality meter 31 measures various quantities of the raw water, such as turbidity, color, water temperature, conductivity, pH (hydrogen ion concentration index), alkalinity, and ultraviolet light absorbance. The ultraviolet light absorbance of the raw water can be used as an index value of the amount of organic matter contained in the raw water, and for example, ultraviolet light with a wavelength of 260 nm is used for this measurement. The various index values ​​measured by the raw water quality meter 31 are input to the coagulant injection control device 1 as plant data.

[0016] In addition, a flow meter 32 is provided in the water distribution pipe between the receiving well 3 and the rapid mixing basin 4. The flow meter 32 measures the flow rate of the water to be treated sent from the receiving well 3 to the rapid mixing basin 4. The flow rate measured by the flow meter 32 is input to the coagulant injection control device 1 as plant data.

[0017] The pH adjuster injector 10 injects a pH adjuster into the water to be treated between the receiving well 3 and the rapid mixing basin 4. The pH adjuster is, for example, an acid such as sulfuric acid or hydrochloric acid, carbon dioxide, or a base such as caustic soda or hydrated lime. By injecting the pH adjuster into the water to be treated sent from the receiving well 3 to the rapid mixing basin 4, the pH of the treated water is adjusted so that it falls within an appropriate pH range for the coagulant injected in the downstream rapid mixing basin 4.

[0018] In the rapid mixing basin 4, a coagulant is injected into the water to be treated that has been sent from the receiving well 3. The rapid mixing basin 4 is a water storage tank for rapidly mixing the water to be treated (hereinafter also referred to as "mixed water") to which the coagulant has been injected.

[0019] The flocculant injector 8 injects a flocculant, which is a chemical such as polyaluminum chloride (PAC) or aluminum sulfate (sulfate band), into the mixed water in the rapid mixing basin 4. The flocculant injector 8 adjusts the amount of flocculant to be injected into the rapid mixing basin 4 according to the value of the flocculant injection amount (or flocculant injection rate) supplied from the flocculant injection control device 1. For example, aluminum-based inorganic flocculants such as polyaluminum chloride (PAC) and aluminum sulfate (sulfate band) are widely used as flocculants. In addition, iron-based flocculants and cationic and anionic organic polymer flocculants may be used as flocculants. For example, suspended matter in raw water from rivers and dam lakes used in water purification plants is usually negatively charged and exists in the water while electrically repelling each other. By adding a positively charged flocculant and stirring, the suspended matter and the flocculant are electrically attracted to each other, and the negative charge of the suspended matter is neutralized. This reduces the repulsive force, making it easier for the suspended matter to bond and flocculation to proceed.

[0020] The rapid mixing basin 4 is equipped with a rapid mixer 41. The rapid mixer 41 mixes the water to be treated into which a coagulant has been injected in the rapid mixing basin 4. For example, the rapid mixer 41 is a flash mixer. The rapid mixer 41 may be one that operates at a constant mixing speed, or one that can adjust the mixing speed by controlling a motor. In the rapid mixing basin 4, minute flocs are formed in the water being treated by injection of a coagulant and stirring by the rapid mixer 41. The water being treated containing such minute flocs is sent to the subsequent flocculation basin 5, where further agglomeration of the flocs is promoted in the facilities following the flocculation basin 5.

[0021] A mixed water quality meter 42 is provided in the distribution pipe between the rapid mixing basin 4 and the flocculation basin 5. The mixed water quality meter 42 measures the quality of the mixed water. Specifically, the mixed water quality meter 42 measures the electrophoretic velocity of flocs in the mixed water as an index value of the flocculation state. The value measured by the mixed water quality meter 42 is input to the coagulant injection control device 1 as plant data.

[0022] Normally, the surface of suspended solids present in water is negatively charged, and they remain stable in the water due to the repulsive force between the negative charges. This causes the water to become turbid. On the other hand, flocculants are positively charged in water. Therefore, when a flocculant is injected into the water to be treated that contains suspended solids, the flocculant adheres to the suspended solids. The flocculant that adheres to the suspended solids cancels out the negative charge of the suspended solids (hereinafter referred to as "neutralizing"), and brings the surface charge of the suspended solids closer to 0 [mV]. When the surface charge of the suspended solids approaches 0 [mV], the zeta potential also approaches 0 [mV]. Therefore, the flocculant weakens the repulsion between suspended solids, increasing the number of collisions. Due to the action of this flocculant, the colliding flocs gradually agglomerate together, forming larger flocs.

[0023] The flocculation basin 5 is a water storage tank for forming larger flocs in the water to be treated. The flocculation basin 5 is equipped with slow speed agitators 54, 55, and 56, and further agglomeration of the flocs is promoted by agitating the water to be treated with the slow speed agitators 54, 55, and 56. For example, the flocculation basin 5 is divided into three agitation basins 51, 52, and 53 as shown in FIG. 1, and the slow speed agitators 54, 55, and 56 are installed in the agitation basins 51, 52, and 53, respectively. For example, the slow speed agitators 54, 55, and 56 are flocculators. Of the three agitation basins 51, 52, and 53, the agitation basin 51 is located most upstream with respect to the flow of the water to be treated, and the agitation basin 53 is located most downstream.

[0024] The untreated water sent from the rapid mixing basin 4 flows into the mixing basin 51. In the mixing basin 51, the untreated water is mixed by the slow speed mixer 54, causing fine flocs to repeatedly collide with each other, forming flocs with larger particle sizes. The untreated water in the mixing basin 51 is mixed for a predetermined period of time before being sent to the subsequent mixing basin 52.

[0025] The water to be treated sent from agitation basin 51 flows into agitation basin 52. In agitation basin 52, flocs with larger particle size are formed by agitating the water to be treated by slow speed agitator 55. Here, since the agglomerated flocs are destroyed if the agitation strength is too strong, slow speed agitator 55 agitates the water to be treated with a weaker strength than slow speed agitator 54. This promotes further agglomeration of the flocs. The water to be treated in agitation basin 52 is sent to a downstream agitation basin 53 after a predetermined period of agitation.

[0026] The water to be treated sent from the agitation basin 52 flows into the agitation basin 53. In the agitation basin 53, the water to be treated is agitated by the slow speed agitator 56, thereby forming flocs with larger particle sizes. Here too, the slow speed agitator 56 agitates the water to be treated with less strength than the slow speed agitator 55 so as not to destroy the agglomerated flocs. This promotes further agglomeration of the flocs. The water to be treated in the agitation basin 53 is sent to the downstream settling basin 6 after being agitated for a predetermined period of time.

[0027] The settling basin 6 is a water tank that stores the water to be treated that flows in from the flocculation basin 5. By storing the water to be treated in the settling basin 6 for a predetermined time, flocs with large particle size formed in the flocculation basin 5 settle by gravity. For example, the water to be treated is stored in the settling basin 6 for about three hours. As a result, the flocs are separated from the water to be treated, and the supernatant water is sent to the subsequent filtration basin 7. The most downstream part of the settling basin 6 may be equipped with equipment that performs additional treatment such as ozone treatment or biological activated carbon treatment on the water to be treated that is sent to the filtration basin 7. The flocs that have settled in the settling basin 6 are extracted as sludge by a sludge extraction pump 63 and sent to a sludge treatment facility such as a sludge extraction basin 62.

[0028] Further, a sedimentation basin water quality meter 61 is provided downstream of the sedimentation basin 6. The sedimentation basin water quality meter 61 measures the water quality of the water to be treated sent to the filtration basin 7. Specifically, the sedimentation basin water quality meter 61 measures various index values ​​related to the treatment results of the solid-liquid separation process. For example, the sedimentation basin water quality meter 61 measures the turbidity and particle count of the water to be treated as index values ​​related to the treatment results of the solid-liquid separation process. The particle count refers to, for example, the number of particles per particle size fraction. The various index values ​​measured by the sedimentation basin water quality meter 61 are input to the coagulant injection control device 1 as plant data.

[0029] The filtration basin 7 is a reservoir equipped with a filtration system that filters the water to be treated that flows in from the sedimentation basin 6. In the filtration basin 7, minute solid matter remaining in the water to be treated is separated by filtration. The filtered water to be treated is stored in a clean water reservoir or the like as treated water and used as tap water. A filter water quality meter 71 is provided downstream of the filter basin 7. The filter water quality meter 71 measures the water quality of the water to be treated filtered by the filter basin 7. Specifically, the filter water quality meter 71 measures various index values ​​related to the processing results of solid-liquid separation of the water to be treated in the final stage treated in the water treatment plant 100. For example, the filter water quality meter 71 measures the turbidity and particle count of the water to be treated as index values ​​related to the processing results of the solid-liquid separation process. The various index values ​​measured by the filter water quality meter 71 are input to the coagulant injection control device 1 as plant data.

[0030] The filter basin 7 is also equipped with a water level gauge 72 that measures the clogging of the filter basin. When the filter basin becomes clogged, the water level at the top of the filter basin rises, and when it reaches a certain level, filtration is stopped and cleaning is performed. Therefore, the faster the water level rises in the filter basin, the more frequently cleaning is required, and the higher the cleaning costs. Factors that promote clogging of the filter basin include the turbidity and aluminum concentration at the outlet of the settling basin. Aluminum concentration exists as a result of residual aluminum-based coagulants such as polyaluminum chloride that are added as a coagulant. The various index values ​​measured by the filter basin water level gauge 72 are input to the coagulant injection control device 1 as plant data.

[0031] For example, in water purification plants, aluminum-based inorganic coagulants with a large positive charge are added to the water to be treated, causing suspended matter to flocculate and settle. Factors that affect the turbidity at the outlet of the settling basin include the turbidity of the raw water, the amount of coagulant injected, pH, water temperature, alkalinity, the mixing strength in the mixing basin, the shape of the settling basin, whether or not a tilt plate is installed in the settling basin, and the retention time. Appropriate coagulation conditions for maintaining a good turbidity at the outlet of the settling basin (for example, turbidity of about 0.5 degrees) in a water purification plant are constantly changing due to the influence of fluctuations in the water quality of the raw water. Here, the time until the result of coagulant injection is reflected in the turbidity at the outlet of the settling basin depends on the retention time of the water to be treated flowing through the water treatment process, and is generally about 3 to 6 hours.

[0032] The analysis unit 9 is an example of a specific means for measuring the electrophoretic velocity of flocs in the mixed water, which is the aggregation state index value. The analysis unit 9 is a device for measuring the aggregation state index value by analyzing a part of the separated mixed water. In the example shown in FIG. 1, a configuration is shown in which a part of the mixed water is separated from a water distribution pipe between the rapid mixing basin 4 and the flocculation basin 5. This configuration is an example, and the mixed water does not necessarily have to be separated from the water distribution pipe between the rapid mixing basin 4 and the flocculation basin 5. For example, the mixed water may be collected from the rapid mixing basin 4 or may be obtained from the flocculation basin 5. In addition, when a separation flow path cannot be provided, the mixed water may be separated manually. In this embodiment, the separation of the mixed water is realized by flowing a part of the mixed water flowing through the water distribution pipe into a flow path (hereinafter referred to as a "separation flow path") other than the water distribution pipe.

[0033] Here, the flocculant injection control device 1 of this embodiment employs a feedback control method for controlling the flocculant injection rate based on the charge state of the flocs (flocculation state index value) measured by electrophoresis. Information representing the electrophoretic speed of each floc is used as an index of the charge state of the flocs used in the feedback control.

[0034] An analysis unit 9, which will be described later, calculates the average value of the electrophoretic velocities (average migration speed) of flocs in mixed water based on information indicating the electrophoretic speed of each floc. The time required to obtain information for one measurement is about 1 to 5 minutes. During this time, several tens to several hundreds of individual flocs are observed. In this embodiment, the electrophoretic speed of each floc is statistically processed to determine the average value of the electrophoretic speed in one measurement.

[0035] The analysis unit 9 includes a light source unit 91, an imaging unit 92, and a speed measurement unit 93 as components for measuring the aggregation state index value. The light source unit 91 irradiates light onto the mixed water flowing through the separation flow path. The light source unit 91 is a light source that irradiates, for example, laser light or visible light. The light source unit 91 may be configured to be able to change the intensity and wavelength of the irradiated light. A part of the light irradiated from the light source unit 91 is scattered by the surfaces of the flocs in the mixed water, and the rest passes through the mixed water to be received by the optical system of the imaging unit 92.

[0036] The imaging unit 92 is configured using an imaging device such as a camera. The imaging unit 92 is disposed at a position where it can image the mixed water flowing through the separation flow path. For example, a transparent container called a cell is disposed midway through the separation flow path, and the light source unit 91 and the imaging unit 92 are disposed so as to sandwich the cell in a direction perpendicular to the flow of the mixed water and face each other. With such an arrangement, light is irradiated onto the mixed water flowing through the cell in a direction perpendicular to the flow, and the imaging unit 92 receives the light transmitted through the cell. The imaging unit 92 generates image data of the mixed water passing through the cell by converting the intensity of the received light into a digital value. The imaging unit 92 images the mixed water passing through the cell at a predetermined imaging period (for example, 1 / 3 second period) and outputs the generated image data to the speed measurement unit 93 in chronological order.

[0037] The velocity measurement unit 93 measures an index value (aggregation state index value) indicating the aggregation state of flocs in the mixed water based on the image data output from the imaging unit 92. Specifically, the velocity measurement unit 93 measures the electrophoretic velocity of flocs as the aggregation state index value. At this time, the flow path is blocked to prevent the flow of water from moving, and voltage is applied to electrodes installed on the left and right sides of the flow path to electrophorese the enclosed flocs in the mixed water. The velocity measurement unit 93 measures the electrophoretic velocity of flocs in the mixed water using the time-series image data output from the imaging unit 92, and outputs the measurement data to the flocculant injection control unit 12.

[0038] FIG. 2 is a diagram showing an example of a cell provided in a separation flow path in the water treatment plant shown in FIG. FIG. 2 shows an example of a cell 300 in which the mixed water flowing in from the negative direction of the y-axis passes in the positive direction of the y-axis when the mixed water is passed through the cell. The cell 300 is provided with valves on the inlet and outlet sides to seal the mixed water in the cell after the mixed water is passed through. The cell 300 is provided with a positive electrode 220 and a negative electrode 210 that form an electric field in a direction perpendicular to the flow of the mixed water when the mixed water is passed through, and a power source 230 that applies a voltage to the positive electrode 220 and the negative electrode 210. When the mixed water is sealed in the cell 300, the power source 230 applies a voltage to the positive electrode 220 and the negative electrode 210, causing electrophoresis of flocs in the mixed water in the cell 300. When a voltage is applied, the flow path in the y-axis direction is blocked by an electromagnetic valve or the like, and the flocs are in a substantially stationary state before the voltage is applied.

[0039] To specifically explain the movement during electrophoresis, flocs with a negative surface charge move toward the positive electrode 220 (i.e., the negative direction on the x-axis) when a voltage is applied. Therefore, the average value of the electrophoretic velocity of flocs with a negative surface charge is negative. On the other hand, flocs with a positive surface charge move toward the negative electrode 210 (i.e., the positive direction on the x-axis) when a voltage is applied. Therefore, the average value of the electrophoretic velocity of flocs with a positive surface charge is positive.

[0040] In contrast, flocs with neutralized surface charges are hardly affected by the electric field. Therefore, the direction of movement of flocs with neutralized surface charges is not constant even when voltage is applied. Therefore, the variation in the movement speed of individual flocs increases, and the movement speed toward the electrode decreases. Therefore, the absolute value of the electrophoretic speed of the flocs decreases because the surface charge approaches 0.

[0041] The velocity measurement unit 93 detects flocs in an image by performing software image analysis processing on an image of mixed water containing flocs electrophoresing in the cell 300, and calculates the migration speed of each detected floc. The migration speed is calculated based on the position of the floc between successively captured images and the imaging period. The velocity measurement unit 93 measures the migration speed of each detected floc and calculates a distribution diagram (histogram) of the migration speed of each floc. At the same time, the velocity measurement unit 93 calculates an average value (hereinafter referred to as "average migration speed") from the migration speeds of each floc.

[0042] FIG. 3 is a diagram showing an example of a distribution map of the movement speed of flocs calculated by the analysis unit shown in FIG. In Figure 3, the migration speed on the horizontal axis is the electrophoretic velocity, and the larger the absolute value of the electrophoretic velocity, the more electrically charged the floc is. The closer the absolute value is to 0, the more neutral the charge is. From the example shown in Figure 3, it can be seen that the charge state of each floc is not uniform, but varies.

[0043] The speed measurement unit 93 statistically processes the movement speeds obtained for each floc to obtain an average value, and uses the calculated average value as a representative value for one batch of measurements. The speed measurement unit 93 may take a moving average of the average movement speeds obtained for each floc over several batches, and calculate this moving average value as the average movement speed. The speed measurement unit 93 outputs the average movement speed thus calculated to the flocculant injection control unit 12 as an aggregation state index value (control amount). This average movement speed is used as a control amount (PV) in feedback control.

[0044] In a water treatment plant 100 having such various water treatment facilities, the flocculant injection control device 1 controls the injection rate of the flocculant injected by the flocculant injection device 8 (hereinafter referred to as the "flocculant injection rate") based on input plant data. In general, the flocculant injection rate is the amount of flocculant injected per unit time relative to the flow rate of the water to be treated per unit time. The flocculant injection rate is converted into the injection amount of the flocculant (hereinafter referred to as the "flocculant injection amount") using the flow rate of the water to be treated per unit time. The configuration of the flocculant injection control device 1 of this embodiment will be described in detail below, but the flocculant injection rate can be replaced with the flocculant injection amount as appropriate.

[0045] In the flocculant injection control device 1 of this embodiment, the flocculant injection rate is controlled by quantitatively expressing the quality of the flocculation state in the flocculation reaction using a water quality index in the treatment process and an index indicating the properties of the flocculant. This method is a feedback control method that adjusts the flocculant injection rate based on the flocculation state after the flocculant is injected. When determining the optimal flocculant injection rate, the charge state of the flocs after the flocculant is given as a discrimination index of the flocculation state that is not affected by the water quality of the raw water. In the flocculant injection control device 1 of this embodiment, as a method for measuring the charge state, a method for universally setting the flocculant injection rate from the charge state of the flocs measured by the electrophoresis method and image processing as described above is used. In controlling the flocculant injection rate, the flocculant injection rate is controlled using the average value of the measured electrophoretic velocity of the flocs and the target value of the charge state.

[0046] The flocculant injection control device 1 includes at least one processor such as a CPU (Central Processing Unit) and a storage unit (memory or auxiliary storage device) in which a program executed by the processor is stored, and executes the program. The flocculant injection control device 1 functions as a device including a control target value determination unit 11 and a flocculant injection control unit 12 by executing the program. All or part of the functions of the flocculant injection control device 1 may be realized using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The program may be recorded in a computer-readable recording medium. The computer-readable recording medium is, for example, a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, or a storage device such as a hard disk built into a computer system. The program may be transmitted via an electric communication line.

[0047] The control target value determination unit 11 determines a control target value when determining the flocculant injection rate of the flocculant injector 8 by a feedback control method. In this embodiment, the control target value determination unit 11 determines a target value of the flocculation state index value (hereinafter referred to as the "flocculation state target value") as the control target value, with the aim of maintaining the turbidity of the settling basin 6 below a predetermined management target value while minimizing the operating cost even when the raw water quality gradually changes seasonally. The control target value determination unit 11 outputs the determined flocculation state target value to the flocculant injection control unit 12.

[0048] The flocculant injection control unit 12 performs feedback control, which is a control method for making the control amount follow the control target value by varying the operation amount (second operation amount) based on the deviation between the control amount and the control target value. For example, the flocculant injection control unit 12 performs feedback control such as P control (Proportional Controller), PI control (Proportional-Integral Controller), and PID control (Proportional-Integral-Differential Controller). The flocculant injection control unit 12 determines the flocculant injection rate of the flocculant injection device 8 as the operation amount (second operation amount, MV) of the controlled plant based on the flocculation state target value (control target value) determined by the control target value determination unit 11 and the input plant data (specifically, the measurement data of the flow meter 32 and the mixed water quality meter 42). The flocculant injection control unit 12 notifies the flocculant injection device 8 of the determined flocculant injection rate. The control period of this flocculant injection rate is, for example, 5 minutes.

[0049] FIG. 4 is a diagram illustrating an example of the configuration of the control target value determination unit illustrated in FIG. FIG. 5 is a diagram showing in more detail a part of the control target value determination unit shown in FIG. The control target value determination unit 11 sets a control target value for the floc charge state by using a technique called extreme value control. First, the controlled plant is any plant having an operation input U and a process output Z. In this embodiment, a water purification plant including a coagulant injection process is taken as an example of the controlled plant.

[0050] The manipulated variable input U is the control target value in the coagulant injection process, i.e., the target value of the movement speed. The actual manipulated variable (second manipulated variable) is the coagulant injection rate, but because this control system has a two-stage cascade configuration, the first manipulated variable (manipulated variable command value) adjusted by extreme value control is the target value (control target value) of the movement speed. The outputs Y are the PAC (coagulant) injection rate, the amount of sludge generated, the frequency of cleaning the filter basin, the turbidity at the outlet of the settling basin, and the turbidity at the outlet of the filter basin.

[0051] The control target value determination unit 11 includes a process measurement value acquisition unit 110, a process evaluation value calculation unit 120, a demodulation dither signal generation unit 130, an evaluation function gradient estimation unit 140, a gradient estimation amount normalization unit 160, an extreme value search unit (optimum manipulated variable adaptive adjustment unit) 170, a modulation dither signal generation unit 180, an output unit 190, and a phase lag estimation unit 1000.

[0052] The process measurement value acquisition unit 110 first acquires measurement information required to calculate evaluation quantities and constraint conditions for the above-mentioned controlled plant. When setting an operating cost including a total value of components of chemical costs, sludge disposal costs, and filter cleaning costs as an example of an evaluation function, the process measurement value acquisition unit 110 acquires, for example, the injection rate of PAC (flocculant), the amount of sludge generated, and the frequency of cleaning the filter as measurement information. When setting constraints on the turbidity at the outlets of the settling basin 6 and the filtration basin 7, the process measurement value acquisition unit 110 also measures the turbidity values ​​at the outlets of the settling basin 6 and the filtration basin 7 at a predetermined period and saves them as time-series data in a predetermined format.

[0053] The process evaluation value calculation unit 120 calculates the evaluation value of a preset evaluation function in real time using the information acquired by the process measurement value acquisition unit 110. As an evaluation function, for example, the operating cost is defined as the sum of the chemical cost (PAC cost), the sludge disposal cost, and the filter basin cleaning cost. The chemical cost is calculated by multiplying the time series data of the PAC injection amount by coefficients such as the chemical unit price and dilution rate, and can be obtained as time series data. The sludge disposal cost is calculated by multiplying the amount of sludge generated by the sludge disposal unit price, and time series data of the sludge disposal cost can be obtained.

[0054] The filter basin 7 is usually cleaned when the water level or resistance (filtration resistance) of the filter basin exceeds a predetermined threshold value, so the timing of cleaning can be known from the time series data of the resistance and cleaning history data, and the cleaning cost at the timing of cleaning can be obtained from the history data. The cleaning cost of the filter basin 7 is considered to be the cost from the previous cleaning to the next cleaning, so the cleaning cost can be converted into time series data by distributing this cleaning cost evenly over that period. However, while this conversion can be easily performed for past performance data, in real time the timing of the next cleaning is uncertain, so some ingenuity is required, such as estimating the next cleaning timing while monitoring the rate of change of the water level (resistance) and converting it into time series data of the cleaning cost in real time. The sum of the various expenses converted into time-series data as described above is the operating cost, and the process evaluation value calculation unit 120 can obtain the operating cost in real time using the above method.

[0055] The process evaluation value calculation unit 120 uses water quality constraints as constraint conditions for the evaluation function. In this case, the process measurement value acquisition unit 110 acquires the turbidity at the outlet of the settling basin 6 and the outlet of the filtration basin 7 of the water treatment plant shown in Fig. 1 as time-series data, and the process evaluation value calculation unit 120 incorporates these turbidity constraint values, for example, constraint conditions such as a settling basin turbidity of 0.8 degrees or less, into the evaluation function.

[0056] In extreme value control, constraint conditions cannot be handled directly, but they can be converted so that they can be handled as an evaluation function. Here, they are handled as an evaluation function using the idea of ​​a penalty function, which is well known in the optimization field. That is, for example, the following conversion is performed. Wcost=max(0,a×(exp(T-Tlim)-1)) (1) Here, T is the measured turbidity value, Tlim is the upper limit of turbidity, and a (>0) is a design parameter. Wcost is a converted evaluation function, and the water quality cost is calculated by converting the turbidity at the outlet of the settling basin 6 and the turbidity at the outlet of the filtration basin 7 using equation (1) and adding them up. Equation (1) means that when turbidity T is less than or equal to the upper limit of turbidity Tlim, the cost Wcost becomes 0, and when turbidity T exceeds the upper limit of turbidity Tlim, the cost Wcost rises rapidly (exponentially), and is a type of penalty function.

[0057] By adding the water quality cost considered by the penalty function above to the operating cost mentioned above, the total cost J can be defined as follows: Total cost J = Chemical cost + Sludge disposal cost + Filtration pond cleaning cost + Water quality cost (2) Once the evaluation function has been properly set by the above-mentioned processing, the value of the evaluation function can be measured and calculated in real time at a predetermined control period to obtain the evaluation quantity that changes from moment to moment.

[0058] The demodulation dither signal generator 130 generates a periodic signal (demodulation dither signal) required to estimate the gradient (first-order differential, Jacobian) of the evaluation function by extreme value control. A sine wave is typically used as the demodulation dither signal, but it does not necessarily have to be a sine wave, and any periodic signal, such as a square wave or a triangular wave, may be used. The demodulation dither signal must have the same period and waveform shape as the modulation dither signal described later. Here, the demodulation dither signal D(t) when a sine wave is used as the dither signal, is shown in equation (3). D(t)=sin(ωt+φ) (3)

[0059] Here, ω is the frequency of the dither signal, and φ is the phase lag (phase lag estimate value). In this embodiment, the demodulation dither signal D(t) has a phase shift of φ due to the phase lag estimate value φ calculated by the phase lag estimator 1000 described later, as shown in equation (3). By performing phase compensation as shown in equation (3), the performance of the extreme value control is improved.

[0060] The evaluation function gradient estimator 140 estimates the gradient of the evaluation value with respect to the manipulated variable (first manipulated variable). As shown in Fig. 5, the evaluation function gradient estimator 140 includes a high-pass filter HPF and a low-pass filter LPF, and includes an adder between the blocks of the high-pass filter HPF and the low-pass filter LPF that adds the demodulation dither signal D(t) generated by the demodulation dither signal generator 130 to the output of the high-pass filter HPF.

[0061] The extreme value (local optimum) of the evaluation index value is the object of search, and is therefore naturally unknown. The high-pass filter HPF is a configuration introduced to forcibly set the extreme value to approximately 0 on the assumption that this unknown extreme value will not change, or if it does change, it will change very slowly. The high-pass filter HPF is not essential in the theoretical configuration of extreme value control, but is usually incorporated because it contributes to improving control performance.

[0062] It is also known from theoretical analysis that the periodic average value of the evaluation value or the signal obtained by passing the evaluation value through a high-pass filter, adding a demodulation dither signal to the signal, and then passing it through a low-pass filter, is a signal (gradient information) proportional to the gradient of the evaluation function. For this reason, the output signal of the low-pass filter LPF can be regarded as the gradient of the evaluation function value (more precisely, a signal proportional to the gradient).

[0063] The gradient estimator normalization unit 160 normalizes the gradient estimate using a normalization signal generated by a normalization signal generation unit included therein. That is, the gradient estimator normalization unit 160 generates G_n(t) by applying the normalization signal to the gradient estimate G(t) estimated by the evaluation function gradient estimator 140. Note that in this embodiment, the gradient estimator normalization unit 160 is not an essential component and may be omitted.

[0064] The extremum search unit 170 searches for an extremum of the evaluation function using an integrator. That is, the extremum search unit 170 can perform an extremum search by multiplying the gradient estimate G(t) or the normalized gradient estimate G_n(t) by a constant called integral gain K through an integrator. The integral gain K is an important parameter that determines the convergence speed of the extremum search, and the value of the integral gain K can be increased by introducing phase lag compensation using the phase lag estimate φ. By increasing the integral gain K, the control performance (convergence speed) can be improved. That is, if the value of the integral gain K is made too large without phase lag compensation, the stability of the extremum control system may be impaired. On the other hand, by performing phase lag compensation, it becomes possible to increase the integral gain K, and as a result, the control performance can be improved.

[0065] The configuration of multiplying the integral gain K through an integrator is almost the same as an optimization algorithm called the gradient method, and the integral gain K corresponds to a parameter called the learning rate in the gradient method.

[0066] The modulation dither signal generating unit 180 generates a dither signal for modulation. This modulation dither signal may be any periodic signal, but must have the same waveform as the demodulation dither signal generated by the demodulation dither signal generating unit 130. When a sine wave signal corresponding to the formula (3) is used, the modulation dither signal M(t) of the formula (4) is used. M(t)=A sin(ωt) (4)

[0067] Here, A is the amplitude of the dither signal and is an adjustable parameter. This dither signal M(t) for modulation is an essential element in the dither signal driven extremum search control. In this embodiment, there is a phase difference φ between the signal of formula (3) and the signal of formula (4), and the phase delay is compensated. Due to the effect of this phase compensation, it becomes possible to set the frequency ω of the dither signal to a large value (=the dither period to a small value) or to set the above-mentioned integral gain K to a large value, thereby improving the control performance (convergence speed) of the extremum search.

[0068] The output unit 190 outputs an operation amount command (control target value) for the controlled object to the coagulant injection control unit 12. This operation amount command signal is obtained by adding the output (optimum operation amount) of the extreme value search unit 170 to the modulation dither signal M(t), whereby the first operation amount is periodically driven by the modulation dither signal M(t) and at the same time an extreme value is searched for.

[0069] Next, an example of application of extreme value control (ESC) in which the flock charge state is set as the control target value will be described. Fig. 6 is a schematic diagram showing an example of a system that adjusts the target value of the charge state of flocs by extreme value control in feedback control of the coagulant injection rate at a water purification plant. The example shown in Fig. 6 is a system that can switch between a manually set target value of the charge state (SV) and a target value of the charge state (SV) that is automatically adjusted in real time using extreme value control.

[0070] At water purification plants, coagulants called PAC are injected to coagulate and settle the turbidity in rivers and other water bodies, and there is a demand to maintain the turbidity of the water flowing out of tanks called settling basins and sand filters below a certain standard value while minimizing the cost of coagulant injection (e.g. chemical costs) and other operational costs as much as possible.

[0071] In the application of extreme value control in Figure 6, the amount of coagulant to be injected (injection rate) is not directly determined by extreme value control, but the charge state (average electrophoretic migration speed) of a substance called floc is set as the controlled variable (PV), and the coagulant injection rate is adjusted by PI (proportional-integral) control so that it follows the target value of the migration speed. The SV, which is the control target value (target value of the migration speed) of PI control, is adjusted by extreme value control. By automatically adjusting the control target value (SV) in real time by extreme value control, the coagulant injection rate is controlled so that the controlled variable (PV) follows the real-time target value (SV) by feedback control, making it possible to respond more quickly to changes in the coagulation state due to changes in the water quality of the raw water.

[0072] In addition, the evaluation function is set as the total cost, which is the sum of the costs (components) related to operation, such as the cost of cleaning the downstream filter basin and the cost of sludge disposal, in addition to the chemical cost of the coagulant, and is designed as an extreme value control system with constraints set to maintain the turbidity of the settling basin and the turbidity of the filter basin below the control value.By applying extreme value control to the setting of the control target value of the floc charge state in this way, it is possible to realize automatic control of the plant that minimizes operating costs while maintaining the control value (constraint) of water quality.

[0073] 7 and 8 are diagrams for explaining an example of a result of verifying the application example of FIG. 6 by simulation. Figure 7 shows an example of the results of a simulation constructed to simulate the coagulation and sedimentation process. In this simulation, a simulator is used to simulate the charge neutralization of flocs accompanying the reaction of a coagulant (PAC), the subsequent growth and sedimentation of flocs, and changes in water quality such as the turbidity of the settled water.

[0074] The manually set SV in Fig. 7 is an example pattern in which the SV is changed seasonally, taking into consideration the annual fluctuations in water temperature, as shown in Fig. 8. In contrast, the same graph shows an example of the simulation results when automatic SV adjustment was performed using extreme value control. Fig. 7 shows that SV setting using extreme value control follows a similar trend to manual SV setting, and also finely adjusts the SV in response to changes in water temperature. The sedimentation water turbidity, which is the treated water quality at this time, was almost maintained at 0.8 degrees, which was set as a water quality constraint, and the simulation confirmed that automatic SV adjustment using extreme value control is effective.

[0075] FIG. 9 is a diagram for explaining the principle of extremum search by extremum control. In FIG. 9, an example of the relationship between the manipulated variable and the evaluation variable in the principle of extreme value control is shown on the left side, and an example of the relationship between the manipulated variable and the evaluation variable when the extreme value control is applied to an actual control target is shown on the right side. In the left and right diagrams of Fig. 9, the horizontal axis is the manipulated variable operated by the extreme value control, which in this embodiment is the target value of the moving speed (first manipulated variable). The vertical axis is the evaluation function (evaluation variable) to be optimized (minimized) by the extreme value control, which in this embodiment is the operating cost (total cost). It is assumed in advance that there is some relationship between the manipulated variable and the evaluation function, and that there is an extreme value (local minimum value), and in Fig. 9, it is assumed that there is a functional relationship that is convex downward.

[0076] Here, in Fig. 9, the function shape is shown as a graph for the purpose of explanation, but when extremum control is actually performed, the evaluation value for each manipulated variable at the time of control execution can be obtained in real time, but the shape of the evaluation function cannot be grasped in real time and is in an unknown state. In such a situation, extremum control provides a control algorithm for searching for an extremum (local optimum, global minimum in Fig. 9) of a function with a downward convex shape, for example.

[0077] When the manipulated variable is driven by a periodic dither signal such as a sine wave, if the manipulated variable is on the right side of the extreme value, the movement of the manipulated variable driven by the dither signal and the movement of the evaluation variable (evaluation value) obtained in real time move in sync with each other to the right of the extreme value of the downward convex function, that is, they move in phase. In other words, this means that the operating cost increases when the target value of the movement speed is increased. On the other hand, if the manipulated variable is on the left side of the extreme value of the downward convex function, the movement of the manipulated variable driven by the dither signal and the movement of the evaluation variable (evaluation value) obtained in real time move in opposite directions, that is, they move in antiphase. In other words, this means that the operating cost decreases when the target value of the movement speed is increased.

[0078] By using this information, it is possible to determine which side of the extreme value the manipulated variable is during the control operation, and accordingly, it is possible to search for the extreme value by decreasing the manipulated variable if it is to the right of the extreme value, and increasing the manipulated variable if it is to the left of the extreme value.

[0079] To explain more precisely, the relationship between the manipulated variable and the evaluation variable is actually as shown on the right side of Figure 9. That is, when the controlled object has dynamics, there is a time lag (phase lag if driven by a sine wave) between the manipulated variable on the horizontal axis and the evaluation variable on the vertical axis, so even if the manipulated variable is on the right side of the extreme value of a downwardly convex function, the manipulated variable and the evaluation variable do not move in phase, but the evaluation variable changes with a phase lag relative to the manipulated variable. Specifically, in a water purification plant, for example, there is a lag time until the change is transmitted by the flow of water, so there is a delay between changing the target value of the movement speed and the change being transmitted to the downstream side.

[0080] Similarly, when the manipulated variable is on the left side of the extreme value of a downward convex function, the manipulated variable and the evaluation variable do not move in antiphase, but the evaluation variable changes with a phase lag relative to the manipulated variable. According to the diagram on the right side of Figure 6, when the phase lag becomes 90 degrees, the relationship that was in phase in the diagram on the left side becomes antiphase, and the relationship that was antiphase in the diagram on the left side becomes inphase.

[0081] This phase delay needs to be as small as possible. To achieve this, the controlled object needs to be a fast enough process that it can be considered as an almost static process. If the controlled object responds fast enough and can be considered as an approximately static process, the relationship between the manipulated variable and the evaluation variable shown in the left diagram of Figure 8 can be realized in an actual control algorithm.

[0082] On the other hand, an actual controlled object does not necessarily respond quickly enough to be considered a static process. In order to deal with such a case, focusing on the fact that the response speed is relative, it is possible to approximately consider the actual controlled object as static by adjusting the frequency of the dither signal. In other words, since the response speed of the actual controlled object is a characteristic specific to the controlled object, it is necessary to design the frequency of the dither signal to be slow so that the response speed can be determined to be sufficiently fast. By doing so, even if there is a response delay due to the controlled object, it can be made sufficiently small in terms of phase delay. This causes the trade-off problem between "stability" and "control performance" mentioned at the beginning. In other words, if stability is achieved by designing the frequency of the dither signal to be slow so that the phase shift does not affect it, the control performance will decrease because the control will be slow.

[0083] In order to achieve both "stability" and "control performance (convergence speed)" in extreme value control, for example, phase compensation can be performed to search for a (local) optimum value that maximizes the convergence speed while maintaining stability. For example, by estimating the phase lag in extreme value control and compensating for the phase lag, it is possible to improve the convergence speed while maintaining the stability of extreme value control. This makes it possible to improve the deterioration of extreme value search performance caused by a slow convergence speed. For example, by simply performing a simple response test such as a step response test used when adjusting PID control, which is widely used in industry, it becomes possible to configure a phase compensation device and configure an extreme value control system that can be used in practice.

[0084] In a control system that continuously measures the charge state of flocs using the electrophoresis method and image processing described above and feedback controls the coagulant injection rate, it is effective to determine the target value of the movement speed using extreme value control and to improve the issues of "stability" and "control performance (convergence speed)" that are inherent to extreme value control using a phase compensation function.

[0085] Here, the breakdown (components) of the operating cost, such as the coagulant cost, sludge disposal cost, and filter basin cleaning cost, have different delay times after changing the target value of the moving speed, which is the manipulated variable (first manipulated variable) of extreme value control, in other words, they have different phase shifts, so it is desirable to consider these multiple phase shifts simultaneously. For example, the locations where each cost included in the operating cost of the controlled plant (chemical cost, filter basin cleaning cost, sludge disposal cost, water quality constraints) occurs can be upstream or downstream in the flow of the treated water in the treatment plant. Therefore, the time at which a response to a change in the control target value SV appears also differs for each cost.

[0086] 10 and 11 are diagrams showing an example of changes in each cost included in the operating cost of a controlled plant and a control target value of the coagulant injection control device of one embodiment. 10 and 11, a target value of the average value (average movement speed) calculated from the movement speed of flocks is shown as a value corresponding to the control target value SV. The coagulant injection rate corresponding to the chemical cost responds relatively quickly to changes in the control target value SV operated by extreme value control, for example, with a delay of about 30 to 60 minutes.

[0087] The filter basin cleaning cost is measured after the effect on the water level of the filter basin 7 has been felt, and therefore responds to a change in the control target value SV with a delay of, for example, 3 to 6 hours. Note that in Fig. 10, the change in the turbidity at the outlet of the settling basin 6 is shown as the value corresponding to the change in the filter basin cleaning cost. This is because the water level of the filter basin 7 is affected by the turbidity at the outlet of the settling basin 6, and the change in the control target value SV only changes after the effect on the turbidity at the outlet of the settling basin 6 has been felt.

[0088] The sludge disposal cost (amount of sludge generated) responds to a change in the control target value SV with a delay of, for example, the midpoint between the response time of the coagulant injection rate and the response time of the filter basin cleaning cost, depending on the timing of the sludge being withdrawn from the settling basin 6. Alternatively, if the amount of sludge generated is calculated from the coagulant injection rate, the response to a change in the control target value SV is delayed by the same amount of time as the coagulant injection rate (for example, about 30 to 60 minutes).

[0089] 12 is a diagram showing an example of changes in the operating cost (evaluation value) of a plant to be controlled by the coagulant injection control device of one embodiment and the control target value SV. Here, the target value of the average value (average movement speed) calculated from the movement speed of flocs is shown as a value corresponding to the control target value SV. As described above, each cost included in the operating cost of the controlled plant has a different time to respond to a change in the control target value SV. Therefore, the phase shift in the evaluation value calculated by adding up each cost becomes even larger. Considering the phase shift of each cost, it is necessary to set the period of the dither signal longer. This is because, when the phase shift between the control target value SV and the evaluation value is π / 2 or more, the minimum value control by the extreme value control becomes the maximum value control. However, when the period of the dither signal is longer, it is difficult to improve the control performance (convergence speed) of the extreme value search, and it is difficult to adjust the control target value SV to minimize the operating cost while maintaining the water quality constraints.

[0090] On the other hand, in the breakdown of the operating costs, for example, in the case where the water quality management value constraint is converged upon, the chemical cost and sludge disposal cost are the main components of the evaluation value, and the phase shift in the evaluation value is small. In addition, in the breakdown of the operating costs, for example, in the case where the water level of the filtration basin 7 based on the turbidity at the outlet of the settling basin 6 (the cost of cleaning the filtration basin) is the main component, the residence time of the water to be treated up to the filtration basin 7 is long, and the phase shift in the evaluation value is large. The inventors of the present application have experimentally confirmed that there are multiple cases of phase shift in the evaluation value, depending on the operation of the controlled plant.

[0091] Therefore, the coagulant injection control device 1 of this embodiment includes a phase lag estimator 1000 that adjusts the phase lag estimated value φ according to the breakdown of the cost, which is a plurality of components included in the operating cost (evaluation value) of the controlled plant. Specifically, the phase lag estimator 1000 sets the phase lag estimated value φ using the values ​​of each of the plurality of components (costs) included in the evaluation value and the phase lag values ​​corresponding to each of the plurality of components.

[0092] For example, if the principal component among the multiple components (chemical cost, sludge disposal cost, filter basin cleaning cost) included in the operating cost, which is the evaluation value, has a small phase shift, a smaller phase lag estimated value φ is set. Conversely, if the principal component among the multiple components included in the operating cost has a large phase shift, the phase lag estimator 1000 sets a larger phase lag estimated value φ. Note that the number of principal components among the multiple components is not limited to one, and there may be multiple.

[0093] The phase lag estimate value φ may be switched manually by the operator, or may be automatically selected from, for example, a table of the relationship between each cost and the phase compensation parameter (phase lag estimate value φ) in accordance with the ratio of each cost in the operating cost. This makes it possible to automatically adjust the phase compensation parameter even if the operating state of the plant, the unit cost, etc. change.

[0094] The phase lag estimating unit 1000 may take a cost having a large proportion in a breakdown of a plurality of costs in the operating cost as a main component, and set the phase lag estimated value φ based on the phase shift of the cost having a large proportion.

[0095] For example, when the breakdown of operating costs is 15% chemical cost, 35% sludge disposal cost, and 50% filtration basin cleaning cost, and the phase compensation parameter for the chemical cost is φ1, the phase compensation parameter for the sludge disposal cost is φ2, and the phase compensation parameter for the filtration basin cleaning cost is φ3, the phase lag estimation unit 1000 may set the phase lag estimated value φ to φ3, which is the phase compensation parameter for the filtration basin cleaning cost, using the filtration basin cleaning cost, which has the largest proportion, as the main component.

[0096] The phase lag estimator 1000 may calculate a phase lag estimated value φ by combining phase compensation parameters of each cost in accordance with the ratio of the operation cost among a plurality of costs.

[0097] For example, when the breakdown of operating costs is 15% chemical costs, 35% sludge disposal costs, and 50% filtration basin cleaning costs, and the phase compensation parameter for the chemical costs is φ1, the phase compensation parameter for the sludge disposal costs is φ2, and the phase compensation parameter for the filtration basin cleaning costs is φ3, the phase lag estimation unit 1000 may set the phase lag estimated value φ to a value calculated by φ=φ1*0.15+φ2*0.35+φ3*0.5 according to the proportion of each cost.

[0098] By setting the phase lag estimate φ based on the principal component based on the cost ratio, the phase lag estimate φ can be adjusted without taking into account other cost fluctuations, i.e., phase shifts of components that have little effect on the fluctuation of the evaluation value.

[0099] Furthermore, the phase lag estimator 1000 may focus on a cost with large fluctuations in the breakdown of the operating cost as a main component, and set the phase lag estimate value φ based on the phase shift of the cost with large fluctuations.

[0100] For example, the phase lag estimation unit 1000 can calculate a plurality of operating costs, record each calculated value in association with time information, and calculate the amount of change or rate of change of each cost in a predetermined period, the ratio of the amount of change of the plurality of costs, etc. The phase lag estimation unit 1000 may use a cost with a large amount of change (or rate of change) as the main component, or may use the cost with the largest amount of change as the main component based on the ratio of the amounts of change of the plurality of costs.

[0101] For example, when the ratios of the changes in the operating costs are 10% for the change in the chemical cost, 15% for the change in the sludge disposal cost, and 75% for the filter basin cleaning cost, and the phase compensation parameter for the chemical cost is φ1, the phase compensation parameter for the sludge disposal cost is φ2, and the phase compensation parameter for the filter basin cleaning cost is φ3, the phase lag estimation unit 1000 may set the phase lag estimated value φ to φ3, which is the phase compensation parameter for the filter basin cleaning cost, using the filter basin cleaning cost, which has the largest change ratio, as the main component.

[0102] The phase lag estimator 1000 may calculate a phase lag estimated value φ by combining phase compensation parameters of each cost in accordance with a ratio of the amount of change in a plurality of operating costs.

[0103] For example, when the breakdown of operating costs is such that the ratio of change in chemical costs is 10%, the ratio of change in sludge disposal costs is 15%, and the filtration basin cleaning costs is 75%, and the phase compensation parameter for chemical costs is φ1, the phase compensation parameter for sludge disposal costs is φ2, and the phase compensation parameter for filtration basin cleaning costs is φ3, the phase lag estimation unit 1000 may set the phase lag estimated value φ to a value calculated by φ=φ1*0.1+φ2*0.15+φ3*0.75 according to the ratio of change in each cost.

[0104] By setting the phase lag estimate φ based on the principal component due to cost fluctuation, when there is a fixed-value cost that accounts for a high proportion of the cost but hardly changes, it is possible to adjust the phase lag estimate φ based on the phase shift of the cost, which fluctuates greatly and is the principal component of the fluctuation of the evaluation value, without taking into account the phase shift of the fixed value.

[0105] The phase lag estimation value φ is not limited to the above value. For example, the phase lag estimation unit 1000 may take the cost with the largest ratio among the multiple operating costs as the first principal component, take the cost with the largest fluctuation as the second principal component, and calculate the phase lag estimation value φ by combining the phase compensation parameter of the first principal component and the phase compensation parameter of the second principal component. For example, the phase lag estimation unit 1000 may select multiple costs from among multiple operating costs in descending order of ratio, further select at least one cost with large fluctuation from the selected multiple costs, and calculate the phase lag estimation value φ using a phase compensation parameter corresponding to the at least one selected cost.

[0106] As described above, according to the present embodiment, it is possible to provide a coagulant injection control device, a coagulant injection control method, and a computer program that minimize the operating costs of a water treatment plant while maintaining appropriate water quality.

[0107] In other words, according to this embodiment, it is possible to provide a coagulant injection control device, a coagulant injection control method, and a computer program that can more appropriately control the amount of coagulant injected even when there are fluctuations in the quality or quantity of raw water to be treated, the water quality control values ​​at a specified location in the plant, and the unit price related to costs such as the unit price of the chemical.

[0108] In particular, according to the present embodiment, in a system that feeds back the amount of coagulant injection based on the floc charge state, it is possible to automatically calculate the control target value (SV) of the floc charge state in real time, and it is possible to provide a coagulant injection control device, a coagulant injection control method, and a computer program that realize operations that minimize operating costs while maintaining the water quality control value.

[0109] Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, and are included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0110] 1...coagulant injection control device, 3...receiving well, 4...rapid mixing basin, 5...flocculation basin, 6...sedimentation basin, 7...filtration basin, 8...coagulant injection device, 9...analysis unit, 10...pH adjuster injection device, 11...control target value determination unit, 12...coagulant injection control unit (control unit), 31...raw water quality meter, 32...flow meter, 41...rapid mixer, 42...mixed water quality meter, 51-53...mixing basin, 54-36...slow mixer, 61...sedimentation basin water quality meter, 62...sludge discharge basin, 63...sludge extraction pump, 71...filtration basin Water quality meter, 72...filter pond water level meter, 91...light source unit, 92...imaging unit, 93...speed measurement unit, 100...water treatment plant, 110...process measurement value acquisition unit, 120...process evaluation value calculation unit, 130...demodulation dither signal generation unit, 140...evaluation function gradient estimator, 160...gradient estimator normalizer, 170...extreme value search unit, 180...modulation dither signal generation unit, 190...output unit, 210...negative electrode, 220...positive electrode, 230...power source, 300...cell, 1000...phase lag estimation unit

Claims

1. a process measurement value acquisition unit for acquiring a plurality of measurement values ​​from a control target; a process evaluation value calculation unit that calculates, using the measurement values, an evaluation value of an evaluation function that changes depending on a first manipulated variable of the controlled object and indicates an index related to optimization of the controlled object; a phase lag estimator that calculates a phase lag estimate from the first manipulated variable in the controlled object to the evaluation value by using values ​​of each of a plurality of components included in the evaluation value and a phase compensation parameter corresponding to each of the plurality of components; an evaluation function gradient estimator that calculates gradient information of the evaluation function, which is a rate of change of the evaluation value with respect to the first manipulated variable, by using the phase lag estimate value and the evaluation value; an extremum seeker for determining an optimal manipulated variable for the first manipulated variable by integrating the gradient information; A coagulant injection control device comprising: a control unit that calculates and outputs a second operation quantity so that the optimal operation quantity is a target value of a control quantity of the controlled object and the control quantity follows the target value of the control quantity.

2. The coagulant injection control device according to claim 1, wherein the phase lag estimation unit regards the component having a large proportion among the multiple components included in the evaluation value as the principal component, and sets the phase lag estimation value based on the phase compensation parameter corresponding to the principal component.

3. The coagulant injection control device according to claim 1 , wherein the phase lag estimation unit calculates the phase lag estimation value by combining the phase compensation parameters of each of the multiple components according to a ratio of the multiple components included in the evaluation value.

4. The coagulant injection control device according to claim 1, wherein the phase lag estimation unit regards the component having the greatest variation among the multiple components included in the evaluation value as a principal component, and sets the phase lag estimation value based on the phase lag value corresponding to the principal component.

5. The coagulant injection control device according to claim 1 , wherein the phase lag estimation unit calculates the phase lag estimation value by combining the phase compensation parameters of each of the multiple components in accordance with fluctuations of the multiple components included in the evaluation value.

6. The control object is a water treatment plant including a coagulant injection process, the first manipulated variable is a target value of a flocculation state index value of flocs contained in the treatment water into which a flocculant has been injected by the flocculant injection process in the water treatment plant; the evaluation value is a total cost including a plurality of costs, each of which has a different time from a change in the first manipulated variable to a response in the water treatment plant, as a plurality of the components; The flocculant injection control device according to claim 1 , wherein the control unit calculates and outputs the flocculant injection rate, which is the second manipulated variable, so that the flocculation state index value of the flocs follows a target value of the controlled variable.

7. Acquire multiple measurement values ​​from the control target, calculating an evaluation value of an evaluation function that changes depending on a first manipulated variable of the controlled object and indicates an index related to optimization of the controlled object, using the measured value; calculating an estimated phase lag value from the first manipulated variable in the controlled object to the evaluation value by using values ​​of each of a plurality of components included in the evaluation value and a phase compensation parameter corresponding to each of the plurality of components; calculating gradient information of the evaluation function, which is a rate of change of the evaluation value with respect to the first manipulated variable, using information on the phase lag estimate and the evaluation value; determining an optimal manipulated variable for the first manipulated variable by integrating the gradient information; A coagulant injection control method, comprising: calculating and outputting a second manipulated variable such that the optimal manipulated variable is a target value of a controlled variable of the controlled object and the controlled variable follows the target value of the controlled variable.

8. A computer program for causing a computer to execute the coagulant injection control method according to claim 7.

Images