JP4734295B2 - Apparatus and method for creating rainwater / pollutant inflow calculation model - Google Patents

Description

  The present invention relates to a creation device and a creation method for a rainwater / pollutant inflow calculation model for predicting the amount of rainwater inflow and the amount of pollutant in rainy weather for safe and efficient operation of a sewerage facility.

  As is well known, sewerage facilities are installed in urban sewers for the purpose of draining rainwater flowing down sewer pipes and drains into rivers when raining. Sewerage facilities are indispensable not only for the treatment of sewage but also for preventing disasters caused by rainwater and for maintaining the safety and environment of urban sanitation. For example, Patent Literature 1 and Patent Literature 2 disclose methods for predicting the amount of rainwater flowing into the sewerage facility and the amount of contaminants. Patent Documents 1 and 2 include a method of solving numerically by combining a hydrologically simulated model of a stormwater runoff process and a partial differential equation model based on hydraulics, as well as known pollution loads such as the “Hydraulic Formulas”. A runoff analysis model is disclosed.

  In addition to these approaches for modeling the phenomenon of stormwater runoff and pollutant load runoff in detail from the physical aspect, the input / output relationship can be determined only from the time series data of the input rainfall and output inflow. There is a black-box approach to identify. Specifically, for example, a technique using a neural network is disclosed in Patent Document 3. Further, Patent Document 4 discloses a method for identifying a block-oriented model, which is a kind of nonlinear model, using a system identification technique. Furthermore, Non-Patent Document 1 describes a pollution runoff analysis model in rainy weather using a hydraulic formula.

Generally, the mathematical model constructed by the latter black box method is an on-line system that requires a short response time with limited computer resources compared to the former physical model because of the simplicity of arithmetic processing. Suitable for the case.
Japanese Patent Laid-Open No. 11-36422 Japanese Unexamined Patent Publication No. 6-322808 Japanese Patent Laid-Open No. 5-134715 JP 2000-56835 A Japan Society of Civil Engineers, "Hydrologic Official Collection", pp450-452, 1994

  However, when the input / output relationship of a model is identified as a black box model, in order to be able to reproduce the target phenomenon, data of an appropriate type, number, and pattern is necessary to capture the characteristics of the process. For example, in a model identified by data of only light rain, the prediction accuracy during heavy rain is significantly deteriorated. Therefore, when identifying the input / output relationship between rainfall and pumping station inflow as a black box based on actual data, if data collection is insufficient or impossible, or there is a pattern of rainfall that is not in the collected data There was concern about the reliability of the forecast.

  In order to solve the above-mentioned problems, the present invention virtually generates a desired rainfall, a discharge phenomenon to a sewer pipe, and a flow phenomenon to a pump station at the stage of creating a prediction model in which onlineness is not required. In addition to the input / output data for blacksbox model construction, it aims to expand predictable inflow patterns.

(1) A rainwater inflow calculation model creation device according to the present invention is a device for creating a calculation model for predicting an inflow amount of rainwater into a sewerage facility as described in claim 1. Or, it is composed of a number of routes, and is composed of a first calculation model that predetermines the coefficient or constant value according to the target sewerage facility, and a single or multiple number of routes, and the value of the coefficient or constant A virtual calculation unit that virtually generates an inflow phenomenon into a sewer facility using the first calculation model, a virtual rainfall data generation unit that virtually generates desired rainfall data, and a virtual calculation unit that uses the first calculation model. Inflow data generation means, and means for determining a value of an undetermined coefficient or constant in the second calculation model from virtual input / output data comprising at least a set of virtual rainfall time series data and virtual inflow time series data The rain time series data in the virtual rain time series data includes a plurality of types and seasons of light rain data, heavy rain data, all day rain data, and temporary rain data. It is data of different virtual rainfall phenomena, and data for constructing the second calculation model is expanded using the first calculation model.

  (2) According to the second aspect of the present invention, there is provided an apparatus for creating a stormwater inflow calculation model according to the second aspect, wherein in (1) the virtual input comprising a set of virtual rainfall time-series data and virtual inflow-volume time-series data. In addition to output data, actual rainfall time-series data is used to determine undetermined coefficients or constant values in the second calculation model.

  (3) According to the present invention, in the rainwater inflow calculation model creation device according to the present invention, in (1) or (2), the first calculation model is a rainwater outflow based on hydrology. It is composed of single or multiple equations describing the process, the rainwater flow process in the sewer pipe based on hydraulics, or a combination thereof, and is characterized by being solved numerically.

  (4) According to the fourth aspect of the present invention, the apparatus for creating a stormwater inflow calculation model is characterized in that, in (1) to (3), the second calculation model is a past input / output time series. This is a black box model that defines the functional relationship only from data. The current amount of rainfall, the amount of rainfall before the present time, the current amount of inflow and the amount of inflow before the present time are used as input variables. Is an output variable, and is a polynomial function composed of these variables and coefficients or constants.

(5) A pollutant inflow calculation model creation device according to the present invention is a device for creating a calculation model for predicting the amount of pollutants flowing into a sewerage facility as described in claim 5. The first operational model is composed of a single or multiple number of routes, and the coefficient or constant value is determined in advance according to the target sewerage facility, and the number of single or multiple routes, and the coefficient Alternatively, using the second calculation model whose constant value is undetermined, virtual rain data generation means for virtually generating desired rainfall data, and the first calculation model, the inflow phenomenon of the amount of pollutants into the sewer facility Virtual pollutant data generating means for generating a virtual turbidity data, and virtual input / output data comprising at least a set of virtual rainfall time-series data and virtual muddy-volume time-series data. Has means for determining a constant value, and the rain time series data in the virtual rain time series data includes light rain data, heavy rain data, all day rain data, and temporary rain time data. A plurality of types of data, which are data of virtual rainfall phenomena with different seasons, and the data for constructing the second calculation model is expanded using the first calculation model.

  (6) The pollutant inflow amount calculation model creation device according to the present invention is the virtual device comprising the combination of the virtual rainfall time series data and the virtual pollutant amount time series data in (5) as described in claim 6. In addition to the input / output data, the undetermined coefficient or constant value in the second calculation model is determined using actual rainfall time-series data.

  (7) The pollutant inflow calculation model creation device according to the present invention is the hydrodynamic calculation model according to (7), wherein the first calculation model is hydrologically based on (5) or (6). It consists of single or multiple equations that describe the stormwater runoff process, the stormwater flow process in the sewer pipe based on hydraulics, the sediment deposition process, and the advection process, and is characterized by being solved numerically.

  (8) The pollutant inflow amount calculation model creation device according to the present invention is as described in claim 8, wherein, in (5) to (7), the second calculation model is a past input / output time series. This is a black box model that defines the functional relationship from data alone, and includes at least the current rainfall, the rainfall before the current, the current pollutant amount, and the previous pollutant amount as input variables. It is an arithmetic model using the amount of contaminants as an output variable, and is characterized by a polynomial function consisting of these variables and coefficients or constants.

(5) A pollutant inflow calculation model creation device according to the present invention is a device for creating a calculation model for predicting the amount of pollutants flowing into a sewerage facility as described in claim 5. The first operational model is composed of a single or multiple number of routes, and the coefficient or constant value is determined in advance according to the target sewerage facility, and the number of single or multiple routes, and the coefficient Alternatively, using the second calculation model whose constant value is undetermined, virtual rain data generation means for virtually generating desired rainfall data, and the first calculation model, the inflow phenomenon of the amount of pollutants into the sewer facility Virtual pollutant data generating means for generating a virtual turbidity data, and virtual input / output data comprising at least a set of virtual rainfall time-series data and virtual muddy-volume time-series data. Has means for determining a constant value, and the rain time series data in the virtual rain time series data includes light rain data, heavy rain data, all day rain data, and temporary rain time data. A plurality of types of data, which are data of virtual rainfall phenomena with different seasons, and the data for constructing the second calculation model is expanded using the first calculation model.

(10) The method for creating a pollutant inflow calculation model according to the present invention is a method for creating a calculation model for predicting the amount of pollutant in rainy weather flowing into a sewerage facility as described in claim 10. A step of creating a first calculation model composed of a single or a plurality of measures and predetermining the coefficient or constant value in accordance with the target sewerage facility, and virtually generating desired rainfall data A step of creating virtual rainfall data, a step of creating virtual pollutant data that virtually causes an inflow phenomenon of the pollutant amount to the sewer facility using the first calculation model, and at least virtual rainfall time-series data; comprising the step of determining the value of the undetermined coefficients or constants in the second calculation model from the virtual output data formed of a pair of virtual Pollution quantity time series data, in the virtual rainfall time-series data Amount of time series data is data of hypothetical rainfall phenomena with different types and seasons, such as data during light rain, data during heavy rain, data during rain all day, and data during temporary rain. and characterized in that it consists of several more single or multiple people, the data for the values the second arithmetic model construction is undetermined of the coefficient or constant, to expand with the first calculation model To do. In the above creation method, in addition to virtual input / output data consisting of a set of virtual rainfall time-series data and virtual pollutant amount time-series data, an undecided coefficient or constant value in the second calculation model using actual rainfall time-series data. Can be determined.

  Even if machine data is not collected or insufficient, such as a new pump station, the rainwater inflow and inflow pollutant data to the sewerage facility output by the first model for the input rainfall data are trusted. Under the conditions, it is possible to create a black box inflow pollutant amount prediction model that is expected to be able to reproduce these virtual rainwater outflow phenomena and pollutant outflow phenomena. Therefore, the present invention is effective for creating a prediction model for use in an online system that requires a short-time response.

Next, embodiments of the present invention will be described with reference to the drawings. In addition, a present Example is not limited to what is described below.
(Example 1) (corresponding to claims 1 to 4 and 9)
FIG. 1 is a block diagram showing a configuration of a device for creating a rainwater inflow calculation model according to Embodiment 1 of the present invention, and FIG. 2 is an explanatory diagram of each data of FIG.
The apparatus for creating a stormwater inflow calculation model according to the first embodiment includes a first calculation model 1, a second calculation model 2, virtual rainfall data generation means 3, virtual inflow data generation means 4, and undetermined coefficients. -An unconstant determining means 5 is provided. The first calculation model 1 is composed of a single equation or a plurality of equations, and the coefficient or constant value is predetermined according to the target sewerage facility. The second calculation model 2 is composed of a single equation or a plurality of equations, and the coefficient or constant value is not yet determined. The virtual rainfall data generation means 3 has a function of virtually generating desired rainfall data. The virtual inflow data generation means 4 has a function of virtually generating an inflow phenomenon to the sewer facility using the first calculation model 1. The undetermined coefficient / unconstant determining means 5 is tentative in the second calculation model 2 from the virtual input / output data composed of a set of the virtual rainfall time series data 6 and the virtual inflow time series data 7 and the actual rainfall time series data 8. It has a function to determine the value of coefficient or constant. In addition, the number 9 in a figure shows performance inflow amount time series data.

The first calculation model 1 is composed of a single equation or a plurality of equations describing a rainwater runoff process based on hydrology, a rainwater flow process in a sewer pipe based on hydraulics, or a combination thereof. For example, in the known method disclosed in Japanese Patent Laid-Open No. 6-322808, the following equations 1 to 4 are used.

However, in the formulas 1 to 4, the respective symbols are as follows.
K, n: constant by basin, S: rainwater storage [m ³ ], Q: rainwater outflow [m ³ / s],
t: time [s], P: rainwater inflow [mm / s], A: water channel cross-sectional area [m ² ],
g: gravitational acceleration [m / s ² ], h: water depth [m], i: riverbed gradient [0/00],
n: Manning roughness coefficient, q: tributary combined flow rate per unit length [m ³ / s],
Q: flow rate [m ³ / s], R: diameter depth [m], t: time [sec],
U: Average flow velocity [m / s], x: Downward coordinate [m]
The second calculation model 2 will be described with reference to a known block-oriented model in Japanese Patent Application Laid-Open No. 2000-56835. The block-oriented model is shown in FIG. As constants that are not yet determined, there are the transfer function orders n, m, and l and the nonlinear order p. Further, the coefficients of the transfer functions A (z), B (z), C (z) and the coefficients of the polynomial f (u) representing the nonlinear term (a _i | i = 1, 2,..., N), (b _i | I = 1, 2,..., M) and (c _i | i = 1, 2,..., L) are also undecided. In FIG. 3, symbol u represents rainfall data, symbol y represents runoff data, and symbol e represents a measurement error that cannot be observed.
In the first embodiment, the block-oriented model is used as the second calculation model, but other known examples may be a black box model using ARX, ARMAX, a neural network, or the like.

Next, the operation of the first embodiment will be described with reference to FIGS.
(1) First, virtual rainfall time series data 6 as shown in FIG. In order to create a black box rainwater inflow prediction model, it is preferable to collect rainfall time-series data for as long a period as possible, for example, one year to several years. The data is preferably as different as possible in the type and season. For example, data during light rain, data during heavy rain, data during rain all day, and rain during the day (eg, evening) And the like. Of such rainfall data, the virtual data supplements what is lacking in the actual data.

(2) Next, the virtual inflow data generating means 4 obtains the virtual inflow time series data 7 (shown in FIG. 2B) to the sewer facility by numerical calculation using the first calculation model 1. As the method, there are a known RRL method, a modified RRL method and the like in addition to using the above formulas 1 to 4.
(3) Subsequently, generation of virtual input / output data composed of a set of virtual rainfall time series data 6 and virtual inflow quantity time series data 7 is repeated until desired data is obtained.

(4) If the actual rainfall time series data 8 (shown in FIG. 2C) is obtained, in addition to the virtual input / output data, the undetermined coefficient and undetermined constant are determined by the following method. The order of the transfer function is obtained by, for example, the AIC standard (Akaike information standard). The nonlinear order is determined in consideration of the strength of the nonlinearity and the number of estimation parameters. Then, the coefficient of the transfer function and the coefficient of the polynomial representing the nonlinear term (a _i | i = 1, 2,..., N), (b _i | i = 1, 2,..., M), (c _i | i = , 1,..., L) are estimated so as to be closest to the relationship between the past rainfall and the inflow data as seen from the prediction target time point. As the estimation method, for example, the least square method is used.

  According to the first embodiment, the inflow amount data to the pumping station output by the first calculation model with respect to the input rainfall data is obtained even if the machine field data is not collected or is insufficient, such as a new pumping station. Under the condition of trust, it is possible to create a black box inflow prediction model that is expected to reproduce these virtual rainfall phenomena.

(Example 2) (corresponding to claims 5 to 8 and 10)
FIG. 4 is a block diagram showing the configuration of the apparatus for creating a pollutant inflow calculation model according to the second embodiment of the present invention. However, the same members as those in FIG. The difference from the first embodiment is that the virtual inflow amount generating means is replaced with the virtual pollutant amount generating means 11, the virtual inflow amount time series data is replaced with the virtual pollutant amount time series data 12, and the first calculation. In addition to the stormwater runoff process based on hydrology and the stormwater flow process in the sewer pipe based on hydrology, the sedimentation and advection process, or a combination of them, is described in the single or multiple equations that make up model 1 It is the point where the equation to do is added. In addition, the number 13 in a figure shows track record amount time series data. The virtual pollutant amount time-series data 12 and the actual pollutant amount time-series data are data indicating the relationship between the time and the pollutant amount, respectively.

The first calculation model 1 can be described by, for example, a well-known pollutant load runoff analysis model in rainy weather in “Hydraulic formula collection” of Non-Patent Document 1. Specifically, it is expressed as the following formula 5 to formula 8.

However, in Equations 5 to 8, the symbols are as follows.
Q _S: sewer outflow _{load, Q SS:} outflow load components in the tube accumulation amount _{Q S, Q SR:} outflow load component of a road surface such as a deposition quantity _{S R,} Q: the sewer flow, _{Q C:} limit Flow rate, R: rainfall intensity, DWF: load during fine weather, t: time, f ₁ and f ₂ : function form, K ₁ : constant.

  FIG. 6 is a flowchart showing the operation of the second embodiment. The difference from the first embodiment is that the virtual inflow generation means is replaced with virtual pollutant generation means.

  According to the second embodiment, the inflow pollutant to the sewerage facility that the first calculation model 1 outputs to the input rainfall data even if the machine data is not collected or insufficient, such as a new pump station. Under the condition of trusting the data, it is possible to create a black box inflow pollutant amount prediction model that is expected to be able to reproduce these virtual pollutant outflow phenomena.

  Note that the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiments. Furthermore, the constituent elements in different embodiments may be appropriately combined. Specifically, in the case of the above-described embodiment, the case where the second calculation model whose coefficient / constant is undetermined is formed based on the actual rainfall time series data in addition to the virtual rainfall time series data and the actual inflow time series data will be described. However, the present invention is not limited to this, and the second calculation model with undecided coefficients and constants may be formed only from the virtual rainfall time series data and the actual inflow time series data.

It is a block diagram which shows the structure of the preparation apparatus of the rainwater inflow amount calculating model which concerns on Example 1 of this invention. It is explanatory drawing of the virtual rainfall time series data, the virtual inflow time series data, the actual rainfall time series data, and the actual inflow time series data in the creation apparatus of FIG. It is explanatory drawing of a well-known Block-Oriented model. It is a block diagram which shows the structure of the preparation apparatus of the pollutant water inflow amount calculating model which concerns on Example 2 of this invention. It is a flowchart which shows the effect | action of the preparation apparatus of Example 1 of this invention. It is a flowchart which shows the effect | action of the preparation apparatus of Example 2 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... 1st calculation model, 2 ... 2nd calculation model, 3 ... Virtual rain amount data generation means, 4 ... Virtual flow volume data generation means, 5 ... Undecided coefficient / undetermined constant determination means, 6 ... Virtual rain time series data, 7 ... virtual inflow time series data, 8 ... actual rainfall time series data, 9 ... actual inflow time series data, 11 ... virtual pollutant amount data generating means, 12 ... virtual pollutant amount time series data, 13 ... actual pollutant amount time Series data.

Claims (10)

An apparatus for creating a calculation model for predicting the amount of inflow of rainwater into a sewerage facility,
A first calculation model composed of a single or a plurality of units, and having predetermined coefficients or constant values according to the target sewerage facility;
A second arithmetic model composed of single or plural numbers, the coefficient or constant value of which is undetermined;
Virtual rainfall data generation means for virtually generating desired rainfall data;
Virtual inflow data generating means for virtually generating an inflow phenomenon into a sewer facility using the first calculation model;
Means for determining a value of an undetermined coefficient or constant in the second computation model from virtual input / output data comprising at least a set of virtual rainfall time series data and virtual inflow time series data;
The rain time series data in the virtual rain time series data includes virtual rain data having different types and seasons, such as light rain data, heavy rain data, all day rain data, and temporary rain data. It ’s the data of a heavy rainfall phenomenon,
An apparatus for creating a rainwater inflow calculation model, wherein data for constructing the second calculation model is expanded using the first calculation model. In addition to virtual input / output data consisting of a combination of virtual rainfall time-series data and virtual inflow time-series data, an undecided coefficient or constant value in the second calculation model is determined using actual rainfall time-series data. The apparatus for creating a rainwater inflow calculation model according to claim 1. The first calculation model is composed of a single or a plurality of numbers that describe a stormwater runoff process based on hydrology, a stormwater flow process in a sewer pipe based on hydrology, or a combination thereof. The apparatus for creating a rainwater inflow calculation model according to claim 1 or 2, wherein the rainwater inflow calculation model is solved. The second calculation model is a black box model that determines the functional relationship only from past input / output time series data. 4. An inflow quantity at a time later than the present time as an output variable with a quantity as an input variable, and a polynomial function comprising these variables and coefficients or constants. A device for creating the rainwater inflow calculation model described. An apparatus for creating a calculation model for predicting the amount of pollutants flowing into a sewer facility during rainy weather,
A first calculation model composed of a single or a plurality of units, and having predetermined coefficients or constant values according to the target sewerage facility;
A second arithmetic model composed of single or plural numbers, the coefficient or constant value of which is undetermined;
Virtual rainfall data generation means for virtually generating desired rainfall data;
Using the first calculation model, virtual pollutant data generating means for virtually generating an inflow phenomenon of the pollutant amount to the sewerage facility,
Means for determining a value of an undetermined coefficient or constant in the second calculation model from virtual input / output data comprising a set of at least virtual rainfall time series data and virtual turbidity time series data
The rain time series data in the virtual rain time series data includes virtual rain data having different types and seasons, such as light rain data, heavy rain data, all day rain data, and temporary rain data. It ’s the data of a heavy rainfall phenomenon,
An apparatus for creating a pollutant inflow calculation model, wherein the data for constructing the second calculation model is expanded using the first calculation model. In addition to virtual input / output data consisting of a combination of virtual rainfall time-series data and virtual pollutant-volume time-series data, the unspecified coefficient or constant value in the second calculation model is determined using actual rainfall time-series data. 6. The apparatus for creating a pollutant inflow calculation model according to claim 5. The first calculation model is composed of a single or multiple numbers describing a stormwater runoff process based on hydrology, a stormwater flow process in a sewer pipe based on hydraulics, a sediment deposition process, and an advection process. 7. The apparatus for creating a pollutant inflow calculation model according to claim 5 or 6, which is solved by numerical calculation. The second calculation model is a black box model that determines the functional relationship only from past input / output time series data, and at least the current rainfall and the rainfall before the current time, the current pollutant amount and the current 6. An arithmetic model including a pollutant amount as an input variable and a pollutant amount at a time later than the present time as an output variable, wherein the calculation model is a polynomial function composed of these variables and coefficients or constants. The apparatus for creating the pollutant inflow calculation model according to claim 7. A method for creating an arithmetic model for predicting the amount of inflow of rainwater into a sewerage facility,
Creating a first calculation model that is composed of a single or a plurality of measures and predetermines the value of the coefficient or constant according to the target sewerage facility;
Creating virtual rainfall data for virtually generating desired rainfall data;
Creating virtual inflow data that virtually causes an inflow phenomenon to a sewer facility using the first calculation model;
A step of determining a value of an undetermined coefficient or constant in the second calculation model from virtual input / output data comprising at least a set of virtual rainfall time series data and virtual inflow time series data,
The rain time series data in the virtual rain time series data includes virtual rain data having different types and seasons, such as light rain data, heavy rain data, all day rain data, and temporary rain data. It ’s the data of a heavy rainfall phenomenon,
Data for constructing a second operation model that is composed of a single or a plurality of units and whose coefficient or constant value is undetermined is expanded using the first operation model. How to create a rainwater inflow calculation model. A method for creating a calculation model for predicting the amount of pollutants flowing into a sewer facility during rainy weather,
Creating a first calculation model that is composed of a single or a plurality of measures and predetermines the value of the coefficient or constant according to the target sewerage facility;
Creating virtual rainfall data for virtually generating desired rainfall data;
Using the first calculation model to create virtual pollutant data that virtually causes an inflow phenomenon of the pollutant amount to the sewer facility; and
A step of determining a value of an undetermined coefficient or a constant in the second calculation model from virtual input / output data consisting of a set of at least virtual rainfall time series data and virtual turbidity time series data,
The rain time series data in the virtual rain time series data includes virtual rain data having different types and seasons, such as light rain data, heavy rain data, all day rain data, and temporary rain data. It ’s the data of a heavy rainfall phenomenon,
The data for constructing the second arithmetic model, which is composed of a single or plural number of coefficients and whose coefficient or constant value is undetermined, is expanded using the first arithmetic model. How to create a pollutant inflow calculation model.

Images