Water Treatment Systems

Abstract

This water treatment system comprises: a main system for purifying water being treated; a subordinate system for purifying the water being treated; water quality control units (107, 207) for controlling the quality of the water being treated; a subordinate system water quality data acquisition unit (209) for acquiring water quality items pertaining to the subordinate system; a main system water quality data acquisition unit (110) for acquiring water quality items of the same type as the water quality items acquired by the subordinate system water quality data acquisition unit (209) and water quality items different from the water quality items acquired by the subordinate system water quality data acquisition unit (209); a main system water quality fluctuation model derivation unit (302) for deriving a main system water quality fluctuation model; a subordinate system water quality fluctuation model derivation unit (303) for deriving a subordinate system water quality fluctuation model; and an estimation unit (304) for estimating the water quality items different from the water quality items acquired by the subordinate system water quality data acquisition unit (209), said water quality items being included in main system water quality data, on the basis of the main system water quality fluctuation model, the subordinate system water quality fluctuation model, and data saved in a data-saving unit.

Description

[Technical Field]

[0001] The present disclosure relates to a water treatment system that measures the quality of water to be treated, such as sewage, and performs a purification treatment. [Background technology]

[0002] The activated sludge process is a widely used method for sewage treatment. The activated sludge process activates aerobic microorganisms, such as nitrifying bacteria, present in sewage by supplying air using a blower or other device, thereby treating water pollutants such as ammonia nitrogen in the sewage.

[0003] In the activated sludge process, to sufficiently remove water pollutants from sewage, it is necessary to measure the water quality, such as the concentration of water pollutants to be treated in the sewage, and control the blower based on the measured values. However, in a conventional water treatment system consisting of multiple lines, if sensors to measure water quality are installed in every line, the number of sensors required is proportional to the number of lines, which poses a problem of high installation costs.

[0004] To address this issue, in a conventional water treatment system consisting of multiple systems, when controlling the water quality of the water to be treated in a certain system, the water quality of the system to be controlled is as In this system, the number of sensors that need to be installed is reduced by referencing water quality data from other systems rather than directly referencing the measurement values ​​of that system. For example, in Patent Document 1, an ammonia nitrogen concentration meter and a dissolved oxygen concentration meter are installed in one system, and only a dissolved oxygen concentration meter is installed in another system. Water quality control for the system in which an ammonia nitrogen concentration meter is installed is performed using the measurement values ​​of that ammonia nitrogen concentration meter, and water quality control for the system in which only a dissolved oxygen concentration meter is installed is performed by referring to the ammonia nitrogen concentration value of the system in which the ammonia nitrogen concentration meter is installed, rather than just the measurement values ​​of the dissolved oxygen concentration meter of that system, thereby reducing the number of sensors that need to be installed. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2005-199115 Summary of the Invention [Problem to be solved by the invention]

[0006] However, while the conventional water treatment system described above performs control using measurable values ​​from other systems, it does not estimate the water quality of the system itself that is the target of control. As a result, individual differences in the devices and sensors in each system are not taken into consideration, which can lead to unstable control.

[0007] The present disclosure has been made in consideration of the above, and aims to provide a water treatment system that can estimate the water quality of a system in which no sensor is installed based on the measurement values ​​of sensors installed in other systems, while taking into account individual differences between systems, and can control water quality. [Means for solving the problem]

[0008] In order to solve the above-mentioned problems and achieve the object, the present disclosure provides a representative system that purifies water to be treated, a subordinate system that purifies water to be treated, a water quality control unit that is provided in the representative system and the subordinate system and controls the water quality of the water to be treated, a subordinate system water quality data acquisition unit that is provided in the subordinate system and acquires at least one type of water quality item from the water to be treated as subordinate system water quality data, a representative system water quality data acquisition unit that is provided in the representative system and acquires from the water to be treated as representative system water quality data at least one first water quality item of the same type as the water quality items acquired by the subordinate system water quality data acquisition unit and at least one second water quality item of a different type from the water quality items acquired by the subordinate system water quality data acquisition unit, a data storage unit that stores control data indicating the control content performed by the water quality control unit, the representative system water quality data acquired by the representative system water quality data acquisition unit and the subordinate system water quality data acquired by the subordinate system water quality data acquisition unit, and a first water quality item included in the representative system water quality data. Water quality itemsa representative series water quality fluctuation model deriving unit that derives a representative series water quality fluctuation model, which is a mathematical model showing fluctuations of the first and second water quality items, based on the data stored in the data storage unit; Water quality items and the second water quality item The values ​​of the same types of water quality items, It is a mathematical model that shows the fluctuations in a dependent series. It is a water quality fluctuation model based on a representative series. a dependent series water quality fluctuation model derivation unit that derives a dependent series water quality fluctuation model based on the representative series water quality fluctuation model and data stored in the data storage unit; and a second water quality item included in the representative series water quality data in the dependent series. Fluctuations in the values ​​of the same water quality items and an estimation unit that estimates the representative series water quality fluctuation model, the dependent series water quality fluctuation model, and the data stored in the data storage unit. [Effects of the Invention]

[0009] According to the present disclosure, it is possible to obtain a water treatment system that can estimate the water quality of a system in which no sensor is installed based on the measurement values ​​of sensors installed in other systems, while taking into account individual differences between systems, and can control the water quality. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a water treatment system according to a first embodiment. [Figure 2] FIG. 10 is a diagram showing a schematic configuration of a water treatment system according to a second embodiment. [Figure 3] FIG. 1 is a diagram illustrating an example of a hardware configuration of a data storage unit and a processing unit according to the first and second embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0011] A water treatment system according to an embodiment of the present disclosure will be described in detail below with reference to the drawings.

[0012] Embodiment 1 Fig. 1 is a diagram showing a schematic configuration of a water treatment system according to embodiment 1. Fig. 1 shows a water treatment system according to the present disclosure that is configured with two or more lines, and for ease of explanation, a water treatment system that is configured with two lines, line 1 and line 2, will be described as an example.

[0013] System 1 is a system including an anaerobic tank 102, an anoxic tank 103, an aerobic tank 104, and a settling tank 105. Sewage, the water to be treated, flows into system 1 from the outside by a pump 101. The inflowing sewage is then treated through the anaerobic tank 102, anoxic tank 103, aerobic tank 104, and settling tank 105 before being discharged. System 1 is also equipped with a water quality control unit 107 that controls the quality of the water to be treated in system 1. In this first embodiment, the water quality control unit 107 is connected to a blower 106 installed in the aerobic tank 104. The water quality control unit 107 sends an aeration volume command value to the blower 106 to aerate the water to be treated in the aerobic tank 104 and remove water pollutants such as ammonia nitrogen. In addition, a sensor is installed in the aerobic tank 104 to measure water quality items of the water to be treated contained in the aerobic tank 104. In this example, two sensors are installed: a dissolved oxygen concentration meter (sensor 108) that measures the concentration of oxygen dissolved in the water to be treated, and an ammonia nitrogen concentration meter (sensor 109) that measures the concentration of ammonia nitrogen contained in the water to be treated.

[0014] Similar to system 1, system 2 includes an anaerobic tank 202, an anoxic tank 203, an aerobic tank 204, and a settling tank 205. Sewage, the water to be treated, flows into system 2 from the outside via pump 201. The inflowing sewage is then treated through anaerobic tank 202, anoxic tank 203, aerobic tank 204, and settling tank 205 before being discharged. System 2 also includes a water quality control unit 207 that controls the quality of the water to be treated within system 2. In this first embodiment, the water quality control unit 207 is connected to a blower 206 installed in the aerobic tank 204. The water quality control unit 207 sends an aeration volume command value to the blower 206 to aerate the water to be treated within the aerobic tank 204 and remove water pollutants such as ammonia nitrogen. Additionally, sensors are installed in the aerobic tank 204 to measure the water quality of the water to be treated within the aerobic tank 204. Unlike Series 1, Series 2 is equipped with only one sensor, that is, a dissolved oxygen concentration meter (sensor 208) that measures the concentration of oxygen dissolved in the water to be treated.

[0015] In this first embodiment, a series capable of measuring two or more water quality items among the water quality items contained in the water to be treated is treated as a series representing all series and is referred to as a representative series. Furthermore, of the water quality items measurable in the representative series, a series capable of measuring at least one water quality item among the water quality items contained in the water to be treated within each series is treated as a series subordinate to the representative series and is referred to as a subordinate series. At least one or more of the water quality items measurable in the representative series is different from the water quality items obtainable in the subordinate series. Specifically, series 1 is treated as a representative series because it is equipped with a dissolved oxygen concentration meter (sensor 108) and an ammonia nitrogen concentration meter (sensor 109), while series 2 is treated as a subordinate series because it is equipped with only a dissolved oxygen concentration meter (sensor 208).

[0016] The representative series is equipped with a representative series water quality data acquisition unit 110 that acquires water quality items measurable in the representative series, and each subordinate series is equipped with a subordinate series water quality data acquisition unit 209 that acquires water quality items measurable in the respective subordinate series. The water quality data collected by the representative series water quality data acquisition unit 110 and the subordinate series water quality data acquisition unit 209 is transmitted to and stored in the data storage unit 301. The data storage unit 301 also stores not only the water quality data for each series, but also control data indicating the control details applied to the water to be treated in each series. This control data is stored by the water quality control units (water quality control unit 107 and water quality control unit 207) installed in each series transmitting it to the data storage unit 301. In the first embodiment, the aeration volume command values ​​sent by the water quality control units 107 and 207 to the connected blowers (blower 106 and blower 206) to control them are also transmitted to and stored in the data storage unit 301. In addition, the data storage unit 301 also stores data on the parameters included in the representative series water quality fluctuation model, which will be described later, and the parameter correction amounts included in the dependent series water quality fluctuation model, as well as the estimated water quality values ​​of the dependent series.

[0017] Water treatment facility The processing system includes a processing unit 300. The processing unit 300 includes a representative series water quality fluctuation model derivation unit 302, a dependent series water quality fluctuation model derivation unit 303, and an estimation unit 304. Fluctuation The model derivation unit 302 derives a representative series water quality fluctuation model, which is a mathematical model that shows fluctuations in the representative series of water quality items that can be obtained in the representative series, based on the various data stored in the data storage unit 301. In the first embodiment, the water quality items that can be obtained in the representative series are the dissolved oxygen concentration obtained by sensor 108 and the ammonia nitrogen concentration obtained by sensor 109, so the representative series water quality fluctuation model derivation unit 302 derives a mathematical model that shows fluctuations in the representative series of these two types of water quality items.

[0018] The dependent series water quality fluctuation model derivation unit 303 derives a dependent series water quality fluctuation model. The dependent series water quality fluctuation model is a mathematical model that indicates fluctuations in each dependent series of water quality items that can be obtained in the representative series, based on the representative series water quality fluctuation model and various data stored in the data storage unit 301. In the first embodiment, an example is shown in which there is one dependent series, but there may be multiple dependent series, and the number of dependent series water quality fluctuation models derived is not necessarily limited to one.

[0019] In this embodiment 1, as mentioned above, the water quality items that can be obtained in the representative series are two types: dissolved oxygen concentration and ammonia nitrogen concentration, and the only dependent series is series 2, so the dependent series water quality fluctuation model derivation unit derives a mathematical model that shows the fluctuations of these two types of water quality items in series 2.

[0020] The estimation unit 304 estimates the parameters included in the representative series water quality fluctuation model based on the representative series water quality fluctuation model, each dependent series water quality fluctuation model, and various data stored in the data storage unit 301, and calculates the water quality fluctuation model for each dependent series of water quality items that can be obtained from the representative series. Fluctuations The estimation is performed on the parameter correction amounts included in each dependent series water quality fluctuation model. The estimation results are stored in the data storage unit 301.

[0021] The output unit 305 outputs the various estimation results estimated by the estimation unit 304 and stored in the data storage unit 301. The output means may be any means, such as outputting as an image on a display, outputting on paper by a printer, or outputting to another device via a network.

[0022] Next, the operation will be described. The representative series water quality fluctuation model derivation unit 302 derives a representative series water quality fluctuation model. As described above, in this first embodiment, a mathematical model of the dissolved oxygen concentration and ammonia nitrogen concentration for series 1, which is the representative series, is derived. In this first embodiment, the activated sludge model (ASM) is an example of a mathematical model that can be applied to this representative series water quality fluctuation model. The activated sludge model is a series of mathematical models for the main water quality items contained in sewage, proposed by the International Water Association (IWA). Mathematical models such as ASM1, ASM2, ASM2D, and ASM3 are described in "Activated Sludge Models - ASM1, ASM2, ASM2D, ASM3" by Shun Ajino, published by Kankyo Shimbunsha on January 31, 2005.

[0023] For example, the mathematical model for the ammonia nitrogen concentration in ASM1 is proposed as follows:

[0024]

number

[0025] The variables and parameters in equation (1) are as follows:

[0026] [Table 1]

[0027] Furthermore, a simpler linear model such as the following can also be applied to the first embodiment.

[0028]

number

[0029] The variables and parameters in equation (2) are as follows:

[0030] [Table 2]

[0031] Additionally, any mathematical model can be applied as the representative series water quality fluctuation model in this embodiment 1. In this embodiment 1, the explanation will be based on the case where the linear model shown in Equation (2) is applied as the representative series water quality fluctuation model.

[0032] In the representative series, the ammonia nitrogen concentration at time t is x (r) NH (t), the dissolved oxygen concentration is x (r) 0(t), aeration volume u (r) NH If (t) is set, the representative series water quality fluctuation model in the first embodiment is expressed as follows:

[0033]

number

[0034]

number

[0035] In the representative series water quality fluctuation model derivation section, the initial estimated values ​​of the parameters in the representative series water quality fluctuation model are derived using the data stored in the data storage section. In this example, the parameters in the representative series water quality fluctuation model shown in Equation (3) and Equation (4), θ (r) NH,NH , θ (r) NH,0 , θ (r) NH,u,b , θ (r) NH,B , θ (r) 0,0 , θ (r) 0,NH , θ (r) 0,u,b and θ (r) 0,B ofInitial estimate θ^ (r) NH,NH , θ^ (r) NH,0 , θ^ (r) NH,u,b , θ^ (r) NH,B , θ^ (r) 0,0 , θ^ (r) 0,NH , θ^ (r) 0,u,b and θ^ (r) 0,B The following is derived. Note that the ^ indicates the initial estimate.

[0036] In the first embodiment, the parameters in the linear model as shown in Equation (3) and Equation (4) are derived by the least squares method. First, the initial estimated values ​​of the parameters in Equation (3) are calculated as θ̂ (r) NH,NH , θ^ (r) NH,0 , θ^ (r) NH,u,b , θ^ (r) NH,B When deriving by the least squares method, the data stored in the data storage unit from a time T to T+N in the past T Using the representative series of ammonia nitrogen concentration, dissolved oxygen concentration, and aeration volume in -1, matrix A(T:T+N-1) is

[0037]

number

[0038] Let us assume that the data stored in the data storage unit are from time T+1 to time T+N. T Using the representative series of ammonia nitrogen concentration, dissolved oxygen concentration, and aeration volume, vector b(T+1:T+N T )of,

[0039]

number

[0040] Then, we have a vector of initial estimates of the parameters we want to derive.

[0041]

number

[0042] can be derived as follows: In the formula, a matrix with -1 in the upper right corner indicates the inverse matrix of the target matrix, and a matrix with T in the upper right corner indicates the transpose matrix of the target matrix.

[0043]

number

[0044] Similarly, the initial estimates of the parameters in Eq. (4), θ̂ (r) 0,0 , θ^ (r) 0,NH , θ^ (r) 0,u,b and θ^ (r) 0,B , can also be derived. In this first embodiment, the case where the representative series water quality fluctuation model handles only two variables, ammonia nitrogen concentration and dissolved oxygen concentration, has been described. However, even if the number of variables increases, the above-mentioned method can be naturally extended to derive initial estimates of the parameters in the representative series water quality fluctuation model in a similar manner. Furthermore, even if the representative series water quality fluctuation model is a nonlinear model such as Equation (1) rather than a linear model such as Equation (2), it is possible to derive initial estimates of the parameters using nonlinear least squares methods such as the Gauss-Newton method. Therefore, regardless of the content of the representative series water quality fluctuation model, initial estimates of the parameters included in the representative series water quality fluctuation model can be derived without loss of generality using the data stored in the data storage unit 301.

[0045] Next, the operation of the dependent series water quality fluctuation model derivation unit 303 will be described. The dependent series water quality fluctuation model derivation unit 303 derives a water quality fluctuation model for each dependent series. As mentioned above, in this example, mathematical models of dissolved oxygen concentration and ammonia nitrogen concentration are derived for series 2, which is a dependent series. When deriving the water quality fluctuation model for each dependent series, considering the differences between the series, each series constitutes the same water treatment system, and therefore sewage of the same water quality flows into each series, and the sewage treatment environment, such as water temperature, can be considered to be approximately the same between the series. Therefore, although individual differences occur between devices and sensors between the series, when a certain series is used as a reference, the differences between that reference series and other series can be considered small. In this first embodiment, this reference series is considered to be the representative series, and it is assumed that the water quality fluctuations in each dependent series will show similar fluctuations to the water quality fluctuations in the representative series. It is believed that each dependent series water quality fluctuation model can be derived by adding a sufficiently small parameter correction amount to the parameters in the representative series water quality fluctuation model.

[0046] The process of deriving the water quality fluctuation model of the dependent series 2 by the dependent series water quality fluctuation model derivation unit 303 will be described below, taking the case where the representative series water quality fluctuation model is the formula (3) and the formula (4) as an example. As mentioned above, the parameter θ (r) NH,NH , θ (r) NH,0 , θ (r) NH,u,b , θ (r) NH,B , θ (r) 0,0 , θ (r) 0,NH , θ (r) 0,u,b and θ (r) 0,B The dependent series water quality fluctuation model is derived using the parameter correction amount for series 2. Here, the ammonia nitrogen concentration at time t is x (s,2) NH (t), the dissolved oxygen concentration is x (s,2) 0(t), aeration volume u (s,2)0(t), and the parameter correction amount for each parameter in the representative series water quality fluctuation model is Δθ (s,2) NH,NH , Δθ (s,2) NH,0 , Δθ (s,2) NH,u,b , Δθ (s,2) NH,B , Δθ (s,2) 0,0 , Δθ (s,2) 0,NH , Δθ (s,2) 0,u,b and Δθ (s,2) 0,B In this case, the dependent series water quality fluctuation model for series 2 can be expressed as follows:

[0047]

number

[0048]

number

[0049] In the dependent series water quality fluctuation model derivation section, the initial estimates of the parameter correction amounts included in the dependent series water quality fluctuation model are derived. In this example, the parameter correction amounts Δθ in the dependent series water quality fluctuation model shown in Equation (9) and Equation (10) are (s,2) NH,NH , Δθ (s,2) NH,0 , Δθ (s,2) NH,u,b , Δθ (s,2) NH,B , Δθ (s,2) 0,0 , Δθ (s,2) 0,NH , Δθ (s,2) 0,u,b and Δθ (s,2) 0,B The initial estimate of Δθ^ (s,2) NH,NH , Δθ^ (s,2) NH,0 , Δθ^ (s,2) NH,u,b , Δθ^ (s,2)NH,B , Δθ^ (s,2) 0,0 , Δθ^ (s,2) 0,NH , Δθ^ (s,2) 0,u,b and Δθ^ (s,2) 0,B However, unlike the representative series, there is no data for uniquely determining the initial estimated values ​​of these parameter correction amounts. Therefore, these initial estimated values ​​are all set to 0, and the estimation unit 304 estimates the water quality items that can be measured in the representative series for each dependent series, while updating the estimated values ​​of these parameter correction amounts to bring them closer to the true values.

[0050] Next, the operation of the estimation unit 304 will be described. As described above, the estimation unit 304 estimates water quality items that can be measured in the representative series for each dependent series, and at the same time, estimates parameters in the representative series water quality fluctuation model and parameter correction amounts included in each dependent series water quality fluctuation model. In this embodiment 1, the estimation unit 304 estimates the ammonia nitrogen concentration of series 2, which is a dependent series, and the parameters and parameter correction amounts in the various models.

[0051] Book In the first embodiment, this estimation process is performed using a particle filter. A particle filter is an algorithm that assumes a large number of virtual state variables called particles and uses these particles to perform estimation by approximating the probability distribution of the state variable to be estimated. In estimation using a particle filter, it is first necessary to define the state equation and observation equation of the system to be estimated. Regarding the state equation, in this system, the water quality items to be measured in the representative series and each dependent series, as well as the various parameters and parameter correction amounts required to express their water quality fluctuations, constitute the system's state equation. Regarding the observation equation, in this system, the water quality items that can be measured in the representative series and each dependent series constitute the system's observation equation.

[0052] In the present embodiment 1, the variables and parameters constituting the equation of state are the ammonia nitrogen concentration and the dissolved oxygen concentration in the representative series 1 and the subordinate series 2, respectively, and the parameter θ required to express these. (r) NH,NH , θ (r) NH,0 , θ (r) NH,u,b , θ (r) NH,B , θ (r) 0,0 , θ (r) 0,NH , θ (r) 0,u,b , θ (r) 0,B and the parameter correction amount Δθ (s,2) NH,NH , Δθ (s,2) NH,0 , Δθ (s,2) NH,u,b , Δθ (s,2) NH,B , Δθ (s,2) 0,0 , Δθ (s,2) 0,NH , Δθ (s,2) 0,u,b and Δθ (s,2) 0,B The variables that make up the observation equation are the dissolved oxygen concentration and ammonia nitrogen concentration of series 1, which can be measured by sensors 108 and 109, and the dissolved oxygen concentration of series 2, which can be measured by sensor 208. These state equations and observation equations can be expressed as follows using the representative series water quality fluctuation model and dependent series water quality fluctuation model described above.

[0053]

number

[0054]

number

[0055] Equation (11) expresses the state equation, and equation (12) expresses the observation equation for the water treatment system. In equation (11), x(t) is a vector composed of the water quality items to be measured in the representative series and each dependent series at time t, as well as various parameters and parameter correction amounts required to express their water quality fluctuations, and is composed of the following vectors.

[0056]

number

[0057]

number

[0058]

number

[0059]

number

[0060]

number

[0061] Furthermore, the vector u(t) in equation (11) is a vector composed of control command values ​​for the system. In the first embodiment, the control command values ​​are aeration volume command values ​​for blower 106 or blower 206 output by water quality control unit 107 of series 1 and water quality control unit 207 of series 2, so vector u(t) can be expressed as follows:

[0062]

number

[0063] Furthermore, f(x(t), u(t)) in Equation (11) is a vector function corresponding to each element constituting the left-hand side x(t+1), and in this example, is configured as follows:

[0064]

number

[0065] The vector function f that constitutes equation (19) (x(r)) (x(t),u(t)) corresponds to the representative series water quality fluctuation model, and the first dimension is equal to the right side of equation (3), and the second dimension is equal to the right side of equation (4). Vector function f (x(s,2)) (x(t),u(t)) corresponds to the dependent series water quality fluctuation model of series 2, and the first dimension is equal to the right side of equation (9), and the second dimension is equal to the right side of equation (10). The vector function f (θ(r)) (x(t),u(t)) corresponds to the parameters included in the representative series water quality fluctuation model, and the vector function f (Δθ(s,2)) (x(t), u(t)) correspond to the parameter correction amounts included in the dependent series water quality fluctuation model of series 2. Since these values ​​are constants, the respective vector functions can be expressed as follows:

[0066]

number

[0067]

number

[0068] Also, the vector w included in equation (11) x (t) is the system noise added to the system state at time t. In this example, w x It is assumed that each element of (t) is normally distributed with a mean of 0, and the variance of the normal distribution of each element can be set to any value by the system user.

[0069] The vector y(t) in equation (12) represents the observation vector, which is a vector composed of elements that can be measured at time t. The vector function h(x(t)) is a vector function that represents a mathematical model for the observation vector y(t) obtained from the state x(t). In this example, the ammonia nitrogen concentration and dissolved oxygen concentration of series 1, and the dissolved oxygen concentration of series 2, can be measured from the state x(t), so the observation vector y(t) and vector function h(x(t)) can be expressed as follows:

[0070]

number

[0071]

number

[0072]

number

[0073]

number

[0074] Also, the vector w included in equation (12) y (t) is the observation noise added to the observation at time t. In this example, w y It is assumed that each element of (t) is normally distributed with a mean of 0, and the variance of the normal distribution of each element can be set to any value by the system user.

[0075] Based on the state equation shown in Equation (11) and the observation equation shown in Equation (12), the particle filter estimates the state of the system at each time. Before performing estimation processing using the particle filter, it is necessary to define and initialize the particle state. The number of assumed particles is N p Let particle i(1≦i≦N p) state as a vector x (i) In particle filters, each particle is treated as a sample of the system's state vector x(t), so the state vector x (i) (t) has the same elements as the system's state vector x(t). In addition, since state transitions and observations can be performed according to the same state equations and observation equations as the system, shown in Equations (11) and (12), the following equations hold. In the equations below, the state vector x of particle i (i) (t), the observation vector y (i) (t), and system noise w x (i) (t), observation noise w y (i) (t) has the same configuration as the actual system described above, so its explanation will be omitted.

[0076]

number

[0077]

number

[0078] The initial state of each particle is determined by taking a sample from a normal distribution with the measured value or initial estimated value as the mean. (r) NH (0), x (r) 0(0), x (s,2) 0(0) and the initial guess θ^ (r) 0,0 (0), θ^ (r) 0,NH (0), θ^ (r) 0,u,b (0), θ^ (r) 0,B (0), θ^ (r) NH,NH (0), θ^ (r) NH,0 (0), θ^ (r) NH,u,b (0), θ^ (r) NH,B(0) , Δθ^ (s,2) 0,0 (0), Δθ^ (s,2) 0,NH (0), Δθ^ (s,2) 0,u,b (0), Δθ^ (s,2) 0,B (0), Δθ^ (s,2) NH,NH (0), Δθ^ (s,2) NH,0 (0), Δθ^ (s,2) NH,u,b (0), Δθ^ (s,2) NH,B (0) A normal distribution with mean x is generated, and the sampled values ​​obtained from these normal distributions are set as the initial state of each particle. However, the variance of these normal distributions can be set by the system user at will, and for cases where the measured value and initial estimated value are not available, x (s,2) NH The system user can also set any value for the mean and variance of (0).

[0079] The first step of the estimation process in a particle filter is to make the state of each particle transition according to the state equation. (i) (t), the input u(t) applied to the system, and the system noise w for each particle. (i) x (t) is generated by sampling, and the transitioned particle state x is calculated based on Eq. (26). (i) Calculate (t+1).

[0080] The second step of the estimation process is to estimate the state x of the transitioned particle. (i) The observation vector y for each particle at (t+1) (i) (t+1). For the prediction of the observation vector, the state x of the transitioned particle is (i) (t+1) and the observation noise w for each particle (i) y (t) is generated by sampling and expressed as (2 7) based on the predicted value y of the observation vector (i) Calculate (t+1).

[0081] The third step of the estimation process is performed after the actual observed value y(t+1) is obtained, and the predicted observed value y (i) (t+1) and the actual observed value y(t+1), we calculate the likelihood of each particle. In this example, we use a normal distribution as the likelihood function, and the likelihood l of particle i at time t+1 is (i) (t+1) is calculated as follows:

[0082]

number

[0083] σ in Equation (28) ^2(σ squared) is a parameter that indicates the variance of the likelihood function, and is set to an arbitrary value by the system user. (i) (t+1)-y(t+1)| is the vector y (i) This is a function that returns the Euclidean norm of (t+1)-y(t+1), and in this example, it is calculated as follows:

[0084]

number

[0085] The fourth step of the estimation process is to normalize the calculated likelihood, weight the state of each particle by the normalized likelihood, and calculate the average value. The vector obtained by this calculation is the estimated value x^(t+1) of the system x(t+1) at time t+1. This calculation is expressed as follows:

[0086]

number

[0087] The fifth step of the estimation process is to resample the particles. A particle filter is an algorithm that estimates by approximating the probability distribution of the state vector to be estimated using a large number of samples (particles). The larger the number of particles, the more accurately it is possible to approximate the desired probability distribution. However, when the first to fourth steps described above are repeated, degeneration occurs, where only a small number of particles have a high likelihood. When degeneration occurs, the desired probability distribution of Only a small number of particles are used for approximation, and the accuracy of the approximation deteriorates dramatically. Resampling prevents this degradation by reducing the likelihood of each particle to 1 / N p In this process, particles are regenerated so that they are close to N based on the likelihood of each particle. p Particles are regenerated by sampling with replacement.

[0088] By sequentially executing the first to fifth steps, the estimated values ​​of the system at each time, i.e., the water quality items to be estimated in each dependent series, the parameters included in the representative series water quality fluctuation model, and the parameters included in each dependent series water quality fluctuation model, are calculated. Rupa Parameter Correction amount In the first embodiment, the system estimation method was described using an example in which there are two systems, an ammonia nitrogen concentration meter and a dissolved oxygen concentration meter are installed in the representative system, and only a dissolved oxygen concentration meter is installed in the subordinate system, and the only control input is the aeration volume command value for each system. However, this estimation method can be implemented regardless of the number of systems, the type of water quality item to be measured, or the content of the control input. Furthermore, in the first embodiment, the processing content was described using a simple linear model as an example, but it can also be implemented without loss of generality even with a complex nonlinear model such as ASM1. Furthermore, although a particle filter was used as the estimation algorithm, other state estimation algorithms such as an extended Kalman filter can also be applied, and the present disclosure is not limited by these factors.

[0089] Water quality control algorithms using estimated values ​​may be performed using classical control methods such as PID (Proportional-Integral-Differential) control using one type of estimated value and two-degree-of-freedom PID control using two types of estimated values, or may be performed using modern control methods such as optimal control using multiple estimated values ​​and a derived water quality fluctuation model.In addition, when performing the above control, actual measured values ​​may be used in addition to estimated values.

[0090] Embodiment 2 2 is a diagram showing a schematic configuration of a water treatment system according to embodiment 2. In the above embodiment 1, among the systems constituting the water treatment system, a system capable of measuring two or more types of water quality items is defined as a representative system, and systems other than the representative system that can acquire at least one type of water quality item among the water quality items measured in the representative system are defined as dependent systems, and the differences between each dependent system and the representative system are set to a small and minute value. Napa Parameter Correction amount This paper describes a method for estimating water quality items that are measured in the representative series but not in the respective subordinate series by expressing them as follows:

[0091] On the other hand, in this second embodiment, as shown in Fig. 2, a similar estimation method is applied to the reaction tanks included in a certain series, rather than to the difference between series. That is, among the reaction tanks included in the same series, a reaction tank capable of measuring two or more types of water quality items is defined as a representative reaction tank, and a reaction tank other than the representative reaction tank that is capable of acquiring at least one or more types of water quality items among the water quality items measured in the representative reaction tank is defined as a subordinate reaction tank, and the difference between each subordinate reaction tank and the representative reaction tank is calculated by using a small number of minute differences. Napa Parameter Correction amount By expressing it as follows, it is possible to estimate water quality items that are the measurement targets of the representative reaction tank but not the measurement targets of each of the subordinate reaction tanks.

[0092] As shown in Fig. 2, the water treatment system according to the second embodiment of the present disclosure includes one or more lines, one of which is composed of two or more reaction tanks. For simplicity of explanation, the second embodiment of the present disclosure will be described taking as an example a water treatment system having one line and five reaction tanks in that line.

[0093] System 1 is a system including an anaerobic tank 402, an anoxic tank 403, an aerobic tank 404, an aerobic tank 405, and a settling tank 406. Sewage, which is the water to be treated, flows into system 1 from the outside by pump 401, and then the inflowing sewage is treated through anaerobic tank 402, anoxic tank 403, aerobic tank 404, aerobic tank 405, and settling tank 406 before being discharged. System 1 is also equipped with a water quality control unit 409 that controls the quality of the water to be treated in system 1. In this second embodiment, water quality control unit 409 is connected to blowers 407 and 408. Blower 407 aerates aerobic tank 404, and blower 408 aerates aerobic tank 405. This water quality control unit 409 sends aeration volume command values ​​to blowers 407 and 408, thereby aerating the water to be treated in aerobic tanks 404 and 405 and removing water pollutants such as ammonia nitrogen. In addition, a sensor is installed in aerobic tank 404 to measure the water quality items of the water to be treated contained in aerobic tank 404. In this second embodiment, two sensors are installed in aerobic tank 404: a dissolved oxygen concentration meter (sensor 410) that measures the concentration of oxygen dissolved in the water to be treated, and an ammonia nitrogen concentration meter (sensor 411) that measures the concentration of ammonia nitrogen contained in the water to be treated. total A sensor (sensor 411) is installed in the aerobic tank 405. The aerobic tank 405 is also installed with a sensor for measuring water quality items of the water to be treated contained in the aerobic tank 405. Unlike the aerobic tank 404, the aerobic tank 405 in this second embodiment is installed with one sensor, and a dissolved oxygen concentration meter (sensor 413) is installed to measure the concentration of oxygen dissolved in the water to be treated.

[0094] In this second embodiment, a reaction tank capable of measuring two or more water quality items contained in the water to be treated in each reaction tank in the series is treated as a reaction tank representing all reaction tanks and is referred to as a representative reaction tank. Furthermore, a reaction tank capable of measuring at least one water quality item among the water quality items measurable in the representative reaction tank and contained in the water to be treated in each reaction tank is treated as a reaction tank subordinate to the representative reaction tank and is referred to as a subordinate reaction tank. At least one of the water quality items measurable in the representative reaction tank is different from the water quality items measurable in the subordinate reaction tanks. In this second embodiment, aerobic tank 404 is treated as a representative reaction tank because it is equipped with a dissolved oxygen concentration meter (sensor 410) and an ammonia nitrogen concentration meter (sensor 411). Furthermore, aerobic tank 405 is treated as a subordinate reaction tank because it is equipped with only a dissolved oxygen concentration meter (sensor 413).

[0095] The representative reaction tank is equipped with a representative reaction tank water quality data acquisition unit 412 that acquires measurable water quality items in the representative reaction tank, and each subordinate reaction tank is equipped with a subordinate reaction tank water quality data acquisition unit 414 that acquires measurable water quality items. The water quality data collected by the representative reaction tank water quality data acquisition unit 412 and the subordinate reaction tank water quality data acquisition unit 414 is sent to and stored in the data storage unit 501. The data storage unit also stores not only the water quality data of the representative reaction tank and each subordinate reaction tank, but also control data indicating the control details applied to the water to be treated in each reaction tank. This control data is stored by being sent to the data storage unit by the water quality control unit 409, which transmits the control details for each reaction tank. In this example, the water quality control unit controls blowers 407 and 408, so the aeration volume command values ​​sent to each blower are also sent to the data storage unit and stored. In addition, the data storage unit also stores data on the parameters included in the representative reaction tank water quality fluctuation model described below, the parameter correction amounts included in each subordinate reaction tank water quality fluctuation model, and the estimated water quality values ​​of each subordinate reaction tank.

[0096] The representative reaction tank water quality fluctuation model derivation unit 502 derives a representative reaction tank water quality fluctuation model, which is a mathematical model showing fluctuations in water quality items that can be obtained in the representative reaction tank, based on various data stored in the data storage unit. In this example, the water quality items that can be obtained in the representative reaction tank are the dissolved oxygen concentration obtained by sensor 410 and the ammonia nitrogen concentration obtained by sensor 411, so the representative reaction tank water quality fluctuation model derivation unit derives a mathematical model showing fluctuations in these two types of water quality items in the representative reaction tank.

[0097] Subordinate reactor water quality Fluctuation Model derivation part 50 3 Based on the representative reactor water quality fluctuation model and various data stored in the data storage unit, a subordinate reactor water quality fluctuation model is derived, which is a mathematical model that shows the fluctuation of water quality items that can be obtained in the representative reactor in each subordinate reactor. There is only one representative reactor. do However, since there can be multiple subordinate reactors, the derived subordinate reactor water quality Fluctuation In this example, the water quality items that can be obtained in the representative reactor are the dissolved oxygen concentration and the ammonia nitrogen concentration, as mentioned above, and the only subordinate reactor is the aerobic tank 405, so the subordinate reactor water quality Fluctuation Model derivation part 503 Now, we will derive a mathematical model that shows the fluctuations of these two types of water quality items in the aerobic tank 405.

[0098] Based on the representative reaction tank water quality fluctuation model, the water quality fluctuation models of each subordinate reaction tank, and various data stored in the data storage unit, the estimation unit 504 estimates the parameters included in the representative reaction tank water quality fluctuation model, estimates the water quality items that can be obtained in the representative reaction tank in each subordinate reaction tank, and estimates the parameter correction amounts included in each subordinate reaction tank water quality fluctuation model, and the estimation results are stored in the data storage unit.

[0099] The output unit 505 outputs the various estimation results estimated by the estimation unit and stored in the data storage unit. The output means may be any means, such as outputting as an image on a display, outputting on paper by a printer, or outputting to another device via a network.

[0100] Next, the operation will be described. Water treatment facility The management system includes a processing unit 500. The processing unit 500 includes a representative reaction tank water quality fluctuation model derivation unit 502, a subordinate reaction tank water quality fluctuation model derivation unit 503, and an estimation unit 504. The representative reaction tank water quality fluctuation model derivation unit 502 derives a representative reaction tank water quality fluctuation model. As described above, in this example, a mathematical model of the dissolved oxygen concentration and ammonia nitrogen concentration in the aerobic tank 404, which is the representative reaction tank, is derived. As in the first embodiment, this mathematical model may be any mathematical model, such as ASM1 shown by equation (1) or a linear model shown by equation (2). In this example, the ammonia nitrogen concentration at time t in the representative reaction tank is deduced as x (r) NH (t), dissolved oxygen concentration x0 (r) NH (t), aeration volume u (r) NH If (t) is set, the representative reactor water quality fluctuation model becomes the same as equations (3) and (4).

[0101] In the representative reaction tank water quality fluctuation model derivation unit 502, initial estimated values ​​of the parameters in the representative reaction tank water quality fluctuation model are derived using the data stored in the data storage unit. ,number The parameter θ in the representative reactor water quality fluctuation model, which is the same as Equation (3) and Equation (4), (r) NH,NH , θ (r) NH,0 , θ (r) NH,u,b , θ (r) NH,B , θ (r) 0,0 , θ (r) 0,NH , θ (r)0,u,b and θ (r) 0,B The initial estimate of θ^ (r) NH,NH , θ^ (r) NH,0 , θ^ (r) NH,u,b , θ^ (r) NH,B , θ^ (r) 0,0 , θ^ (r) 0,NH , θ^ (r) 0,u,b and θ^ (r) 0,B is derived. As for the derivation method, if a linear model such as Equation (3) and Equation (4) is adopted, it can be derived by the least squares method as shown in embodiment 1, and if a nonlinear model such as ASM1 is adopted, it is possible to derive initial estimates of the parameters by nonlinear least squares methods such as the Gauss-Newton method. Therefore, regardless of the content of the representative reaction tank water quality fluctuation model, it is possible to derive initial estimates of the parameters included in the representative reaction tank water quality fluctuation model without loss of generality using the data stored in the data storage unit 501.

[0102] Next, the operation of the sub-reaction tank water quality fluctuation model derivation unit 503 will be described. The sub-reaction tank water quality fluctuation model derivation unit 503 derives a sub-reaction tank water quality fluctuation model. As described above, in the second embodiment, the sub-reaction tank water quality fluctuation model is derived. 4Mathematical models of the dissolved oxygen concentration and ammonia nitrogen concentration are derived in 05. When deriving the water quality fluctuation model for each subordinate reaction tank, considering the differences between the reaction tanks, the same discussion as in embodiment 1 can be used to consider that, although individual differences may occur between devices and sensors, if a certain reaction tank is used as a reference, the differences between that reference reaction tank and other reaction tanks are small. In this embodiment, this reference reaction tank is considered to be the representative reaction tank, and it is considered that the water quality fluctuations in each subordinate reaction tank will be similar to those in the representative reaction tank, and it is considered that the water quality fluctuation models for each subordinate reaction tank can be derived by adding a sufficiently small parameter correction amount to the parameters in the representative reaction tank water quality fluctuation model.

[0103] The process of deriving the water quality fluctuation model of the reaction tank 405, which is a subordinate reaction tank, by the subordinate reaction tank water quality fluctuation model derivation unit will be described below, taking as an example the case where the representative reaction tank water quality fluctuation model is the same as Equation (3) and Equation (4). As mentioned above, the parameter θ (r) NH,NH , θ (r) NH,O , θ (r) NH,u,b , θ (r) NH,B , θ (r) O,O , θ (r) O,NH , θ (r) O,u,b and θ (r) O,B The sub-reactor water quality fluctuation model is derived using the parameter correction amount for the ammonia nitrogen concentration at time t in the reactor 405. (s,405) NH (t), the dissolved oxygen concentration is x (s,405) O (t), aeration volume u (s,405) O (t) and representative Reactor The parameter correction amount for each parameter in the water quality change model is Δθ (s,405) NH,NH , Δθ (s,405) NH,O , Δθ (s,405) NH,u,b , Δθ(s,405) NH,B , Δθ (s,405) O,O , Δθ (s,405) O,NH , Δθ (s,405) O,u,b and Δθ (s,405) O,B In this case, the dependency of the reactor 405 Reactor The water quality fluctuation model can be expressed in the same format as Equations (9) and (10).

[0104] The slave reaction tank water quality fluctuation model derivation unit 503 derives an initial estimated value of the parameter correction amount included in the slave reaction tank water quality fluctuation model. (s,405) NH,NH , Δθ (s,405) NH,O , Δθ (s,405) NH,u,b , Δθ (s,405) NH,B , Δθ (s,405) O,O , Δθ (s,405) O,NH , Δθ (s,405) O,u,b and Δθ (s,405) O,B The initial estimate of Δθ^ (s,405) NH,NH , Δθ^ (s,405) NH,O , Δθ^ (s,405) NH,u,b , Δθ^ (s,405) NH,B , Δθ^ (s,405) O,O , Δθ^ (s,405) O,NH , Δθ^ (s,405) O,u,b and Δθ^ (s,405) O,B However, unlike the representative reaction tank, there is no data for uniquely determining the initial estimated values ​​of these parameter correction amounts. Therefore, these initial estimated values ​​are all set to 0, and the estimation unit 504 estimates the water quality items in each subordinate reaction tank that can be measured in the representative reaction tank, while updating the estimated values ​​of these parameter correction amounts and performing processing to bring them closer to the true values.

[0105] Next, the operation of the estimation unit 504 will be described. As described above, the estimation unit 504 estimates the water quality items in each subordinate reaction tank that can be measured in the representative reaction tank, and at the same time, estimates the parameters in the representative reaction tank water quality fluctuation model and the parameter correction amounts included in each subordinate reaction tank water quality fluctuation model. In this second embodiment, the ammonia nitrogen in the reaction tank 405, which is the subordinate reaction tank, concentration and estimate the parameters and parameter correction amounts in the various models.

[0106] In the second embodiment, this estimation process is performed using a particle filter. As in the first embodiment, estimation using a particle filter requires defining a state equation and an observation equation for the system to be estimated. Regarding the state equation, in this system, the water quality items to be measured in the representative reaction tank and each of the subordinate reaction tanks, as well as the various parameters and parameter correction amounts required to express the fluctuations in these water qualities, constitute the state equation of the system. Furthermore, regarding the observation equation, in this system, the water quality items that can be measured in the representative reaction tank and each of the subordinate reaction tanks constitute the observation equation of the system.

[0107] In the second embodiment, the variables and parameters constituting the equation of state are the ammonia nitrogen concentration and the dissolved oxygen concentration in the reaction tank 404, which is the representative reaction tank, and the reaction tank 405, which is the subordinate reaction tank, and the parameter θ required to express these. (r) NH,NH , θ (r) NH,O , θ (r) NH,u,b , θ (r) NH,B , θ (r) O,O , θ (r) O,NH , θ (r) O,u,b , θ (r) O,B and the parameter correction amount Δθ (s,405) NH,NH , Δθ (s,405) NH,O , Δθ (s,405) NH,u,b , Δθ (s,405) NH,B, Δθ (s,405) O,O , Δθ (s,405) O,NH , Δθ (s,405) O,u,b and Δθ (s,405) O,B The variables that make up the observation equation are the dissolved oxygen concentration and ammonia nitrogen concentration in the reaction tank 404, which can be measured by sensors 410 and 411, and the dissolved oxygen concentration in the reaction tank 405, which can be measured by sensor 413. These equations of state and observation equations can be expressed as follows using the representative reaction tank water quality fluctuation model and subordinate reaction tank water quality fluctuation model described above.

[0108]

number

[0109]

number

[0110] Equation (31) expresses the state equation, and equation (32) expresses the observation equation of this system. In equation (31), x(t) is a vector composed of the water quality items to be measured in the representative reactor and each subordinate reactor at time t, as well as the various parameters and parameter correction amounts required to express their water quality fluctuations, and is composed of the following vectors.

[0111]

number

[0112]

number

[0113]

number

[0114]

number

[0115]

number

[0116] Furthermore, the vector u(t) in equation (31) is a vector composed of control command values ​​for the system. In the second embodiment, the control command value is an aeration volume command value for blower 407 or blower 408 output by water quality control unit 409, so vector u(t) can be expressed as follows:

[0117]

number

[0118] Furthermore, f(x(t), u(t)) in Equation (31) is a vector function corresponding to each element constituting the left side x(t+1), and in the second embodiment, is configured as follows:

[0119]

number

[0120] The vector function f(x(t),u(t)) in Equation (39) corresponds to the representative reactor water quality fluctuation model, and the vector function f (x(s,405)) (x(t),u(t)) is the dependent variable of the reactor 405 Reactor This corresponds to the water quality fluctuation model. The content of the function is the same as in the first embodiment, so it will not be described here. The vector function f (θ(r)) (x(t),u(t)) is the representative Reactor It corresponds to the parameters included in the water quality variation model and is a vector function f (Δθ(s,405)) (x(t),u(t)) is the dependent variable of the reactor 405 ReactorThese correspond to the parameter correction amounts included in the water quality variation model. Since these values ​​are constants, the respective vector functions can be expressed as follows:

[0121]

number

[0122]

number

[0123] Also, the vector w included in equation (31) x (t) is the system noise added to the system state at time t. In this example, w x It is assumed that each element of (t) is normally distributed with a mean of 0, and the variance of the normal distribution of each element can be set to any value by the system user.

[0124] The vector y(t) in equation (32) represents an observation vector, which is a vector composed of elements that can be measured at time t. Furthermore, the vector function h(x(t)) is a vector function that represents a mathematical model for the observation vector y(t) obtained from the state x(t). In this example, since the ammonia nitrogen concentration and dissolved oxygen concentration in the reaction tank 404 and the dissolved oxygen concentration in the reaction tank 405 can be measured from the state x(t), the observation vector y(t) and the vector function h(x(t)) can be expressed as follows:

[0125]

number

[0126]

number

[0127]

number

[0128]

number

[0129] Also, the vector w included in equation (32) y (t) is the observation noise added to the observation at time t. In this example, w y It is assumed that each element of (t) is normally distributed with a mean of 0, and the variance of the normal distribution of each element can be set to any value by the system user.

[0130] As described above, in the second embodiment, the state equation is defined as formula (31), and the observation equation is defined as formula (32). Since formulas (31) and (32) are the same as formulas (11) and (12) of the first embodiment, it is possible to sequentially estimate the ammonia nitrogen concentration in the reaction tank 405, the parameters included in the representative reaction tank water quality fluctuation model, and the parameter correction amounts included in each subordinate reaction tank water quality fluctuation model, which are included in the state vector x(t), using the same method as in the first embodiment. In this example, the number of reaction tanks that are the target of the system is two, and the representative reaction tank has an ammonia nitrogen concentration total and dissolved oxygen concentration total The subordinate reactor is equipped with a dissolved oxygen concentration total The system estimation method was explained using a situation in which only one reactor is installed and the only control inputs are aeration volume command values ​​for the representative and subordinate reactors. However, this method can be implemented regardless of the number of reactors, the type of water quality items to be measured, or the content of the control input. Furthermore, in this second embodiment, the processing content was explained using a simple linear model as an example, but it can also be implemented without loss of generality even with a complex nonlinear model such as ASM1. Furthermore, although a particle filter was used as the estimation algorithm, other state estimation algorithms such as an extended Kalman filter can also be applied, and the present disclosure is not limited by these factors.

[0131] Water quality control algorithms using estimated values ​​may be performed using classical control methods such as PID (Proportional-Integral-Differential) control using one type of estimated value and two-degree-of-freedom PID control using two types of estimated values, or may be performed using modern control methods such as optimal control using multiple estimated values ​​and a derived water quality fluctuation model.In addition, when performing the above control, actual measured values ​​may be used in addition to estimated values.

[0132] 3 is a diagram illustrating an example of the hardware configuration of the data storage unit and processing unit according to the first and second embodiments. The data storage units 301, 501 and processing units 300, 500 each include a processor 11 that executes various processes and a memory 12 that stores information. The processor 11 and the memory 12 can transmit and receive information to and from each other via a bus 13.

[0133] The data storage unit 301 is realized by the memory 12. The processor 11 reads and executes the program stored in the memory 12 to execute the representative series water quality fluctuation model derivation unit 302. , Representative reactor water quality fluctuation model derivation part 502, dependent series water quality fluctuation model derivation part 303 , Sub-reactor water quality fluctuation model derivation part 503, and the estimation units 304 and 504.

[0134] The processor 11 is an example of a processing circuit, and includes one or more of a CPU (Central Processing Unit), a DSP (Digital Signal Processor), and a system LSI (Large Scale Integration).

[0135] The memory 12 includes one or more of RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), and EEPROM (Electrically Erasable Programmable Read Only Memory). The memory 12 also includes a recording medium on which a computer-readable program is recorded. Such a recording medium includes one or more of non-volatile or volatile semiconductor memory, a magnetic disk, a flexible memory, an optical disk, a compact disk, and a DVD (Digital Versatile Disc). The processing unit 300 , 500 may include integrated circuits such as ASICs (Application Specific Integrated Circuits) and FPGAs (Field Programmable Gate Arrays).

[0136] The water quality control units 107, 207, and 409 are connected to the blowers 106, 206, and 40 7,408 The control unit is not limited to a device that controls the above, but may be a device that controls at least one of pumps 101, 201, 401, a valve (not shown), and a chemical addition device (not shown) to control water quality. The valve is, for example, a valve that is provided in a pipe through which the water to be treated flows into the system and adjusts the inflow rate of the water to be treated. The chemical addition device is, for example, a device that adds chemicals to the water to be treated in order to purify it.

[0137] Furthermore, it is desirable that the water quality items acquired by each water quality data acquisition unit 110, 209, 412, 414 include at least one of the following: dissolved oxygen concentration, ammonia nitrogen concentration, nitrite nitrogen concentration, nitrate nitrogen concentration, total nitrogen concentration, total phosphorus concentration, pH (hydrogen ion concentration), alkalinity, water temperature, MLSS (Mixed Liquor Suspended Solids), ORP (Oxidaton reduction potential), turbidity, color, conductivity, BOD (Biochemical Oxygen Demand), and COD (Chemical Oxygen Demand).

[0138] The configurations shown in the above embodiments are merely examples, and may be combined with other known technologies, or different embodiments may be combined with each other. It is also possible to omit or modify parts of the configurations as long as they do not deviate from the gist of the invention. [Explanation of symbols]

[0139] 101,201,401 Pumps, 102,202,402 Anaerobic tanks, 103,203,403 Anoxic tanks, 104,204,404,405 Aerobic tanks, 105,205 ,406 Settling tank, 106,20 6, 407,408 Blower, 107,207,409 Water quality control unit, 108,109,208,410,411,413 Sensor, 110 Representative series water quality data acquisition unit, 209 Dependent series water quality data acquisition unit, 300,500 Processing unit, 301,501 Data storage unit, 30 2 Representative series water quality fluctuation model derivation part, 30 3 Dependent series water quality fluctuation model derivation unit, 304, 504 estimation unit, 305, 505 output unit, 412 representative reactor water quality data acquisition unit, 414 subordinate reactor water quality data acquisition unit , 502 Representative reactor water quality fluctuation model derivation part, 503 Subordinate reactor water quality fluctuation model derivation part .

Claims

1. A representative series for purifying and treating the water to be treated; a subordinate system for purifying the water to be treated; a water quality control unit provided in the representative line and the subordinate line to control the water quality of the water to be treated; a dependent series water quality data acquisition unit provided in the dependent series and configured to acquire at least one water quality item from the water to be treated as dependent series water quality data; a representative series water quality data acquisition unit provided in the representative series, which acquires from the treated water as representative series water quality data at least one first water quality item of the same type as the water quality items acquired by the dependent series water quality data acquisition unit, and at least one second water quality item of a different type from the water quality items acquired by the dependent series water quality data acquisition unit; a data storage unit that stores control data indicating the control details performed by the water quality control unit, the representative series water quality data acquired by the representative series water quality data acquisition unit, and the dependent series water quality data acquired by the dependent series water quality data acquisition unit; a representative series water quality fluctuation model derivation unit that derives a representative series water quality fluctuation model, which is a mathematical model that shows fluctuations of the first water quality item and the second water quality item included in the representative series water quality data, based on the data stored in the data storage unit; a dependent series water quality fluctuation model derivation unit that derives a dependent series water quality fluctuation model, which is a mathematical model that shows fluctuations in the dependent series of values ​​of water quality items of the same type as the first water quality item and the second water quality item acquired by the representative series water quality data acquisition unit, and is a water quality fluctuation model based on the representative series, based on the representative series water quality fluctuation model and the data stored in the data storage unit; A water treatment system characterized by comprising an estimation unit that estimates fluctuations in the values ​​of water quality items in the dependent series that are of the same type as the second water quality item included in the representative series water quality data based on the representative series water quality fluctuation model, the dependent series water quality fluctuation model, and data stored in the data storage unit.

2. The water treatment system according to claim 1 , further comprising an output unit that outputs the estimation result estimated by the estimation unit.

3. The dependent series water quality fluctuation model derivation unit The water treatment system described in claim 1, characterized in that the dependent series water quality fluctuation model is derived by deriving a parameter correction amount for each parameter included in the representative series water quality fluctuation model, and adding the derived parameter correction amount to a parameter included in the representative series water quality fluctuation model that corresponds to the parameter correction amount, thereby deriving each parameter included in the dependent series water quality fluctuation model.

4. The dependent series water quality fluctuation model derivation unit The water treatment system described in claim 2, characterized in that the dependent series water quality fluctuation model is derived by deriving a parameter correction amount for each parameter included in the representative series water quality fluctuation model, and adding the derived parameter correction amount to a parameter included in the representative series water quality fluctuation model that corresponds to the parameter correction amount, thereby deriving each parameter included in the dependent series water quality fluctuation model.

5. The water treatment system described in Claim 3, characterized in that the parameter correction amount is derived by sequentially updating a predetermined estimated value based on a state equation and an observation equation corresponding to the dependent series water quality fluctuation model.

6. The water treatment system described in Claim 5, characterized in that the estimation unit estimates fluctuations in the values ​​of water quality items in the dependent series that are of the same type as the second water quality item included in the representative series water quality data, while updating the estimated values.

7. The water treatment system described in Claim 4, characterized in that the parameter correction amount is derived by sequentially updating a predetermined estimated value based on a state equation and an observation equation corresponding to the dependent series water quality fluctuation model.

8. The water treatment system described in Claim 7, characterized in that the estimation unit estimates fluctuations in the values ​​of water quality items in the dependent series that are of the same type as the second water quality item included in the representative series water quality data, while updating the estimated values.

9. a representative reaction tank for purifying the water to be treated; A subordinate reaction tank for purifying the water to be treated; a water quality control unit provided in the representative reaction tank and the subordinate reaction tank to control the water quality of the water to be treated; A subordinate reaction tank water quality data acquisition unit provided in the subordinate reaction tank and acquiring at least one water quality item from the treated water as subordinate reaction tank water quality data; A representative reaction tank water quality data acquisition unit is provided in the representative reaction tank and acquires at least one first water quality item of the same type among the water quality items acquired by the subordinate reaction tank water quality data acquisition unit and at least one second water quality item different from the water quality items acquired by the subordinate reaction tank water quality data acquisition unit as representative reaction tank water quality data from the treated water; A data storage unit that stores control data indicating the control content performed by the water quality control unit, the representative reaction tank water quality data acquired by the representative reaction tank water quality data acquisition unit, and the subordinate reaction tank water quality data acquired by the subordinate reaction tank water quality data acquisition unit; A representative reaction tank water quality fluctuation model derivation unit derives a representative reaction tank water quality fluctuation model, which is a mathematical model showing fluctuations of the first water quality item and the second water quality item included in the representative reaction tank water quality data, based on the data stored in the data storage unit; A subordinate reaction tank water quality fluctuation model derivation unit derives a subordinate reaction tank water quality fluctuation model, which is a mathematical model that shows fluctuations in the subordinate reaction tank of values ​​of water quality items of the same type as the first water quality item and the second water quality item acquired by the representative reaction tank water quality data acquisition unit, and is a water quality fluctuation model based on the representative reaction tank, based on the representative reaction tank water quality fluctuation model and data stored in the data storage unit; and an estimation unit that estimates fluctuations in the values ​​of water quality items in the subordinate reaction tanks that are the same type as the second water quality item included in the representative reaction tank water quality data based on the representative reaction tank water quality fluctuation model, the subordinate reaction tank water quality fluctuation model, and data stored in the data storage unit.

10. The water treatment system according to claim 9 , further comprising an output unit that outputs the estimation result estimated by the estimation unit.

11. The subordinate reaction tank water quality fluctuation model derivation unit The water treatment system according to claim 9, wherein the dependent reaction tank water quality fluctuation model is derived by deriving a parameter correction amount for each parameter included in the representative reaction tank water quality fluctuation model, and adding the derived parameter correction amount to a parameter included in the representative reaction tank water quality fluctuation model that corresponds to the parameter correction amount, thereby deriving each parameter included in the dependent reaction tank water quality fluctuation model.

12. The subordinate reaction tank water quality fluctuation model derivation unit The water treatment system according to claim 10, wherein the dependent reaction tank water quality fluctuation model is derived by deriving a parameter correction amount for each parameter included in the representative reaction tank water quality fluctuation model, and adding the derived parameter correction amount to a parameter included in the representative reaction tank water quality fluctuation model that corresponds to the parameter correction amount, thereby deriving each parameter included in the dependent reaction tank water quality fluctuation model.

13. The water treatment system described in Claim 11, characterized in that the parameter correction amount is derived by sequentially updating a predetermined estimated value based on a state equation and an observation equation corresponding to the dependent reaction tank water quality fluctuation model.

14. The water treatment system described in Claim 13, characterized in that the estimation unit estimates fluctuations in the values ​​of water quality items in the subordinate reaction tank that are of the same type as the second water quality item included in the representative reaction tank water quality data, while updating the estimated values.

15. The water treatment system described in Claim 12, characterized in that the parameter correction amount is derived by sequentially updating a predetermined estimated value based on a state equation and an observation equation corresponding to the dependent reaction tank water quality fluctuation model.

16. The water treatment system described in Claim 15, characterized in that the estimation unit estimates fluctuations in the value of a water quality item in the subordinate reaction tank that is of the same type as the second water quality item included in the representative reaction tank water quality data, while updating the estimated value.

17. 17. The water treatment system according to claim 1, wherein the water quality control unit controls at least one device selected from the group consisting of a blower, a pump, a valve, and a chemical addition device.

18. 17. The water treatment system according to claim 1, wherein the water quality items include at least one of dissolved oxygen concentration, ammonia nitrogen concentration, nitrite nitrogen concentration, nitrate nitrogen concentration, total nitrogen concentration, total phosphorus concentration, hydrogen ion concentration, alkalinity, water temperature, activated sludge suspended solids, oxidation-reduction potential, turbidity, color, conductivity, biochemical oxygen demand, and chemical oxygen demand.

19. A water treatment system described in any one of claims 1 to 16, characterized in that the estimation unit performs estimation processing using a particle filter with a predetermined number of particles.

20. The estimation unit calculates likelihoods indicating the likelihoods of each particle based on the observed predicted value and the actual predicted value of each particle in the particle filter, calculates a weighted average of the states of each particle using the normalized likelihoods as weights, and updates the states of each particle using the value of the weighted average, The water treatment system of claim 19, characterized in that the estimation unit performs a re-extraction of each particle based on the likelihood of each particle and regenerates the same number of particles as the number of particles, thereby bringing the likelihood of each particle closer to the reciprocal of the number of particles.

Images