JPS6323391B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to sewage/rainwater pump equipment in a sewer system, and relates to a method of operating a sewage/rainwater pump group constituted by a plurality of pumps of different capacities.

Sewage/rainwater pumping equipment for sewerage systems consists of pump wells and pump groups. The pump wells are connected to underground pipes, the suction side of the pump group is connected to the pump well, and the discharge side is connected to the initial subsidence of the sewage treatment plant. Both are connected to the pond by a conduit. Rainwater falling in the city and sewage from households flow into the pump wells through pipes, but the discharge volume of the pump group is changed so that the water level in the pump wells does not exceed the upper and lower limits established. . Conventionally, the water level in a pump well was measured, and if the water level rose, the pump group discharge amount was increased, and conversely, if the water level fell, the pump group discharge amount was decreased. In this case, the pump group is composed of pumps of different capacities, and the discharge amount of the group can be adjusted stepwise by controlling the number of pumps or continuously by partially controlling the speed.

In general, household sewage has a somewhat fixed time series, with changes occurring on a weekly or daily basis depending on the activities of human groups. However, rainwater changes from time to time, both in the form of time-series changes in rainfall by region, and also in the form of time-series changes in the amount of rainwater flowing into pump wells through pipes. It is highly unpredictable and has almost no reproducibility.

Therefore, for sewage/rainwater pump groups that target rainwater, the conventional method of determining the discharge amount of the pump group by measuring only the water level has the effect that the rate of change in the amount of rainwater inflow at one point at the current point is This corresponds to thinking that the rate of change will continue, and it cannot be said that it corresponds to an object whose rate of change changes.
Specifically, this results in poor water level prediction accuracy and unnecessary starting and stopping.

The present invention has been made based on the above reasons, and takes into consideration the time change in the rate of change of the pump well inflow during rainfall.
We predict the value of the inflow at a certain point in time, calculate the pump well water level, and if it falls within the upper and lower water level limits, we decide not to change the operating status of the pump group and avoid unnecessary pump starts and stops. By reducing the
It is an object of the present invention to provide a method of operating a group of sewage/rainwater pumps that does not have the above-mentioned drawbacks and has the minimum necessary number of starts and stops.

The present invention will be explained below with reference to the drawings. FIG. 1 shows the configuration of the target system. Pipe 1
The sewage (sewage and rainwater) flowing from the pump passes through a settling basin 2 and a screen 3, and then flows into a pump well 4. The pump group 5 sends the sewage stored in the pump well 4 to the sewage treatment facility 6. A pump well water level gauge 7 is installed in the pump well 4, and the pump group is operated so that the water level is maintained within a range between an upper limit value 10 and a lower limit value 11.

Further, a discharge flow meter 8 is attached to the discharge side of the pump group 5, and the value of the water level gauge 7 is calculated by a computer 9, and the pump selected by the built-in algorithm is further calculated using the result. Start or stop the pump to match operating conditions. Since the relationship between the pump well bottom area or cross-sectional area and height is known, the pump well water level is proportional to the difference between the inflow amount and the pump group discharge amount. Since the water level and the discharge amount are obtained as actual measured values at each discrete time, in order to predict the water level at the next discrete time, both the inflow amount and the discharge amount must be predicted.

Now, if the discharge amount is constant, that is, if the pump operating state does not need to be changed, then only the inflow amount needs to be predicted. When the inflow amount is predicted by the method described below, the water level can be predicted using the actual measured value of the discharge amount. If this predicted water level is within the upper and lower limits of the pump well, there will be no problem, but if it exceeds this range, the pump group discharge amount must be changed. Therefore, this operating method mainly consists of two parts. The first is () a pump well inflow prediction calculation unit,
The second is () a pump operation state changing section.

(); Pump well inflow prediction calculation unit At discrete time n, time series data y of inflow
(k) (k=1, 2, . . . , n) has already been obtained, and an autoregressive model based on this can be obtained by the following equation.

y(n+1)= _o 〓 ^i=n-m+1 xi(n)・y(i)...(1) Here, the coefficient xi(n), (i=n-m+1, n-
m+2,...,n) are determined by the method of least squares. Also, regarding the number of terms m of explanatory variables, for example,
It is determined in advance using data related to the subject using AIC (Akaike Information Criterion). The coefficient xi(n) of the autoregressive model in the first equation is expressed as the state vector
(n), and from the interpretation that the inflow amount including rainwater includes noise, the state transition formula and output observation formula of this prediction system are as follows.

X(n+1)=Φ(n)・X(n)+U(n) y(n)=H(n)・X(n)+v(n) (2) Here, y(n) is discrete time n observed value of inflow,
Coefficient vector H(n) = (y(n-1), y(n-2)
，
..., y(n-m)) coefficient Φ(n) is a transition matrix of X(n), and vector U(n) and scalar v(n) are normal white noise that is mutually uncorrelated.

That is, each probability density function F is: F(U(n))=N(o,Q(n)) F(v(n))=N(o,r(n)) However, N(A, B) means mean value A variance B. Since this falls within the framework of Kalman's predictive estimation theory, the best estimate X^(n+1|n) of the state quantity X(n+1) at discrete time n can be obtained by the following calculation.

X^(n+1|n) =Φ(n)X^(n|n-1)+G(n)・{y(
n)-H(n)・Φ(n)X^(n|n-1)} Here, G(n+1) = [P(n+1|n)・H ^T (n)] [H(n)・P(
n+1｜n)・H ^T (n)+r(n+1)] P(n+1｜n)=Φ(n)・P(n|n)Φ ^T (n
)+Q(n+1) P(n+1｜n+1)=[I-G(n)・H(n)]
P(n+1|n) where G(n) is the Kalman gain, P(n+1|
n) is the estimation error X^ of the state quantity vector X(n+1)
(n+1|n) (i.e. x〓(n+1|n)=△=x
(n+1)−x^(n+1|n)). In the end, the predicted value y(n+1) of the observed value y^(n) is given by the following equation.

y^(n+1)=H(n+1)・X^(n+1｜n) (); Pump operating state change section When it is necessary to change the pump operating state, the following two auxiliary rules should be taken into consideration when making a decision. .

(-a): Reduce the number of times the pump starts and stops as much as possible.

(-b); Equalization of pump operation time.

The term (-a) is basically satisfied by using the current pump operating state as the base point.
In other words, when making individual selections, for example, if it is predicted that the water level will exceed the upper limit and one pump must be started for this reason, the one that is currently in operation should be left as is, the one that is stopped, and the one that satisfies the water level condition should be selected. Select as a candidate. If a candidate is not found, a candidate is found as a combination of two stopped machines.

In addition, the term (-b) means that different capacity pumps, that is, m types of capacity pumps, are each ni (i=1, 2,...,
m) units, m≧1 and ni≧1 (i=1, 2,
. The overall flowchart of the above algorithm is shown in FIG. The point to be noted here is whether the amount of sewage/rainwater flowing into the pump well satisfies the condition of normal white noise, and it is generally thought that natural phenomena satisfy this condition. The characteristics of rainwater become apparent when the amount is several times that of sewage, so it can be assumed that the amount of inflow of sewage rainwater satisfies the characteristics of rainwater as a natural phenomenon, that is, the condition of regular white noise. .

One unit of discrete time is generally 5 or 10 minutes, but this is determined by considering various points such as pump well volume, pump capacity, annual rainfall, target area population, computer performance, etc. . By accurately predicting the pump well inflow rate at each discrete time, the pump well water level can also be predicted accurately. This effect is especially noticeable when there is a lot of rainfall (for example, Figure 3).
It becomes possible to prevent unnecessary starting and stopping of the pump. Further, in determining the pump operating state, by using the integrated value of the pump operating time as an auxiliary rule, the operating time of the pump is equalized, and as a result, the life of the pump equipment is extended.

As mentioned above, this operating method is known to be most effective when there is a lot of rainfall, but it is not ineffective even when there is little rainfall. However, if you cannot find an advantage when evaluating the ratio of the calculation and its effect, the part that determines the amount of rainfall should be placed before the ``Calculation of predicted inflow amount using Kalman filter ()'' in Figure 2. It is also practical to provide one and omit this calculation.

As described above, the present invention applies the Kalman prediction and estimation theory that treats changes as noise in order to predict the amount of rainfall that changes from moment to moment, making it possible to predict pump well water levels with high accuracy. . As a result of improved prediction accuracy, it is now possible to eliminate the wasteful starting and stopping of pumps, and by taking into account the equalization of pump operating times,
This is a pump operating method that also extends the life of pump equipment.

[Brief explanation of the drawing]

The drawings show an embodiment of the present invention. Fig. 1 is a configuration diagram of the target system of the present invention, Fig. 2 is an overall flowchart of the pump group operation algorithm, and Fig. 3 is a simulation result of pump group operation during rain. This is an example. 1...Buried pipe, 2...Sand basin, 3...Screen,
4... Pump well, 5... Pump group, 6... Sewage treatment equipment, 7... Pump well water level meter, 8... Pump group discharge meter, 9... Computer, 10... Pump well upper limit water level, 11
... Pump well lower limit water level, 12... Pump well inflow time series curve (actual data), 13... Pump group discharge amount time series curve (determined by this algorithm), 14... Pump well water level time series curve.

Claims (1)

[Claims] 1. In the operation of a group of sewage/rainwater pumps that mainly pump sewage/rainwater from urban areas to a sewage treatment plant, two types of measuring instruments, a pump well water level meter and a pump group discharge flow meter, are used to measure the sewage and rainwater at predetermined time intervals. Predict the pump well inflow flow rate in time by applying a Kalman filter, calculate the pump well water level, and if this calculated value occurs within the upper and lower limits set for the pump well water level, it will fall within the pump well water level limit range. A method of operating a group of sewage/rainwater pumps is characterized in that the combination of pumps to be operated is determined so as not to change the current operating state as much as possible, thereby minimizing the number of times the pumps start and stop.