JPS5947930B2 - Power demand forecasting method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention is a demand generator that counts power pulses for each unit of power sent from a power meter, and when a calculation time comes, predicts and calculates the power amount at the end of the demand period at that time. The present invention relates to a method for predicting power demand of a monitoring device.

The mission of the demand monitoring device is to monitor and control the building to keep the power consumption within the contracted power, so the method of calculating the predicted power is important.

Therefore, if the current state of power usage is considered to exceed the contracted power, that is, if the predicted power exceeds the contracted power, the circuit will be cut off to reduce the power used and the contracted power will be exceeded. Control to keep the power within the range. Furthermore, when the current power consumption is considered to be within the contracted power, that is, when the predicted power has a margin compared to the contracted power, the power is used as is. Therefore, if a prediction is made incorrectly, the shearing command will be delayed and the contracted power will be exceeded, or the shedding command will be issued even though it is unnecessary, and the demand monitoring device will not function properly. become unable. Now, let ptN be the amount of power used from the demand start time to minutes, let Δ be Δ in minutes before the minute, let ΔP be the average power used in minutes, and let tr be the remaining time of demand time Td in minutes.
(Ta=tftr), 5 conventional predicted power Pd
was calculated using equation (1) or equation (2).

Pd=Pt+ΔP×− ・・・・・・・・・・・・(1
) Δ as Pd=Pt×−・・・・・・・・・・・・・・・・・・
・・・・・・(2) Figure 1 shows time (1), (2)
The predicted power according to the formula is shown.

P is the final value of the actual power used, Pd,j5 is the predicted power at time according to formula (1), and Pd2 is the predicted power at time according to formula (2). If Δ
The predicted power is calculated based on the average power usage ΔP for the time Δt without knowing the state change of the current power before the time. Also(
In the case of Equation 2, the predicted power is calculated based on the amount of power used Pt up to time t. However, these methods are suitable as an initial pre-loss method for demand time, which should be insensitive to loads where fluctuations in average power usage are averaged for each method number AΔt, or to fluctuations in load. Forecasting when there is only a short amount of time remaining has the disadvantage that it is not possible to completely grasp trends, resulting in false alarms and erroneous control. The purpose of the present invention is to count the power pulses for each unit of power sent from the watt-hour meter, calculate the power usage for a certain period of time, and calculate the changes in the power usage as the most recent power usage. It is an object of the present invention to provide a power demand forecasting method that obtains an appropriate predicted power value in a time when the remaining time within the demand time is short by taking into account the following. The invention will be explained below based on an embodiment shown in the drawings.

First, the principle of the method of the present invention will be explained with reference to FIG.

Figure 2 shows the time Δt before the demand forecast time t.
Each current power ΔPl, ΔP,, ΔP
3 is shown. Analyzing the straight line A in Fig. 2, we find that the average power ΔP per unit time Δt before time t
,, ΔP2, ΔP3 continue to increase. In other words, it continues to change in the same direction, increasing u and selling, i.e. changing u in the same direction.
I understand that it's selling. This state is considered to continue for a short time after time t. Therefore, each average power ΔP, ΔP2 for each unit time Δt before the predicted time t
, ΔP, change Δα,. ScoldingP2屹P,,Δα. , 3P3-ΔP2 to find the rate of change -Y immediately before the predicted time t. This is multiplied by the change ゜α12Δα23 immediately before the predicted time t to find the change in the next unit time Δt after the predicted time t, that is, ``(6a゛゜)'' and Δα12.

This value is calculated as the average flying power ΔP3 at the predicted time t.
is added to the unit time Δt after the predicted time t
The average power per Therefore, in relation to the remaining time Tr, the power amount Pd3 at the end of the demand is predicted by calculating the equation (3) shown below.

However, if the above equation (3) is applied from the beginning of the demand time, the predicted value will change drastically, so the above equation (1) or (2) is applied at the beginning of the demand time.

Therefore, equation (3) is applied when the remaining time is short and the changes Δα,2, Δα·23 before the predicted time t have the same sign. Here, the difference between equations (1) and (3) is shown in FIGS. 3 and 4.

In the case of FIG. 3, the average power ΔP, ΔP2, ΔP3 for each unit time Δt before the predicted time t continues to decrease, and there is a very strong possibility that it will become zero after the time t.

However, according to equation (1), it is predicted that the average power ΔP3 of t-Δt immediately before time t will continue as it is, so at time t
The current prediction at is the straight line σ. Therefore, the contract power is F
s, the predicted power P'D, becomes P'D,>Ys, and it becomes necessary to issue a shedding command to the load. However, in reality, as mentioned above, the amount of electricity used continues to decrease.
The above-mentioned shunting directive is not necessary. On the other hand, according to equation (3), the average power Δ per unit time Δt before time t
Find the rate of change just before time t from P,, ΔP2, ΔP3, multiply it by Δα12 the change just before time t, Δα23, and calculate (Aa23
)2 and convert it into the current average power ΔP at time t
3, the prediction follows the decreasing trend of the load before time t, and the current prediction at time t is a straight line σ.

Therefore, the relationship between contract power Ys and predicted power P'D3 is Yd
3<Ys, and no shearing command is issued to the load. In other words, it is possible to accurately control the power demand without causing unnecessary load interruptions that occur when using the conventional formula (C).Also, in the case of FIG. 4, as described above, (According to Equation C, the current prediction is a straight line σ, so the predicted power Pd,'<contracted power Ps', and no shearing command is issued to the load.

However, in reality, the amount of power used is increasing and is expected to increase gradually, and unless the load is cut off, there is a strong possibility that it will exceed the contracted power Ys. However, according to equation (3), the current prediction is a straight line c, so the predicted power Pd'. 〉
The contracted power becomes Ys, and a shedding command is issued to the load. Therefore, as in the case using the conventional formula (1), the contract power Ys
Exceeding the ゛゛-ノ hawk-゛-&m parable + ＾ flue field digging 1 nail fuzumasu. ? ．． First, with reference to FIG. 5, an embodiment of a demand monitoring device for carrying out the above-mentioned method of the invention will be described. Under the control of the central processing unit 2, the demand monitoring device 1 performs scanning and processing at a fixed time according to a program stored in advance in the ROM 3. Here, a unit power pulse and a unit time pulse, which are sufficiently long pulses compared to the scan time of the CPU, are input to the digital input device 5 having a latch function, and the power meter 8 and the unit time generator 9 output the unit power pulse and unit time pulse.
Input from The digital input device 5 then latches and holds this signal. Then the CPU sends that signal to RA
At the same time as input to M3, a reset signal is output to the latch circuit to prepare for the next input pulse. Then, calculations are performed according to the program of the RCM 4, and the results are outputted from the digital output device 7 to the display 11. Also, when it is thought that the amount of power used exceeds the contracted power, the digital output device 6 issues a power cut command to the connected loads 10A to 10E, reduces the load, and controls the amount of power used to be kept within the contracted power. do. In the above configuration, the central processing unit 2 uses the constant area shown in FIG. 7b according to the block diagram in FIG. 6 to perform calculations and calculations in the variable area shown in FIG. .

The above process will be explained below with reference to FIG. The reference numeral ABCDE in the figure represents the connection relationship of the flowcharts shown separately due to space constraints, and has no particular meaning. First, a general explanation will be given. When the demand monitoring device 1 is powered on, it is initialized, and in step 21, the demand time Td is entered in the remaining time area Tr. At step 22, it is checked whether or not a unit power pulse is input.
At step 27, it is checked whether or not a unit time pulse is input. Then, in step 40, the predicted power Pd is displayed on the display 11 shown in FIG. 5, and in step 41, the predicted power Pd and the contracted power Ps are compared. This result Pd≧P8
When it is considered that the amount of power used exceeds the contracted power, in step 42 the loads 10A-10E shown in FIG.
Issuing a Nishiyadan order. In such a scan process, when a unit power pulse is input, it is checked in step 22, and then in step 23 a latch reset signal for the unit power pulse is output.

Upon input of the above unit power pulse, in step 24 it is added to the number of pulses N for calculating the amount of power used Pt within the demand time and the number of pulses DN for calculating the current power Pd, respectively, and the unit pulses are added to the number of pulses DN for calculating the current power Pd. Replaced as N and DN up to power pulse input. Then, in step 26, the unit power amount P, which is a constant, is multiplied by the cumulative number N of pulses from the start of the demand to calculate the power consumption Pt. By multiplying, the current power ΔP is calculated. On the other hand, when a unit time pulse is input, it is checked in step 27 and then a latch reset signal for the unit time pulse is outputted in step 28. Then, in step 29, the calculation cycle (in this case, 6
0 second) to the count timer Tc and the judgment timer TΔ of the current power calculation cycle Δt, and subtract the remaining time Tr to calculate the Tc and time until the next unit time pulse input, respectively.
Replace as TΔ, Tr. At step 30, a check is made to determine whether or not the current power calculation cycle has arrived, and if the calculation time Δt has elapsed, the current power data is shifted at step 31.

Immediately replace the current power at that point with the latest current power ΔP shown in Figure 2, and change what was previously ΔP to Δ
Similarly, ΔP3 is replaced with ΔP1 in P2. Then, in step 32, the current power calculation timer TΔ and the number of pulses D are set.
Reset N. In step 33, it is checked whether or not the calculation cycle (60 seconds) for demand prediction has elapsed. If the calculation cycle has elapsed, the calculation cycle timer Tc is reset in step 34, and then the remaining time is is less than 5 minutes. That is, in step 34, if the remaining time is 5 minutes or more or ΔP, 〈ΔP2〈
If ΔP, ΔP1>ΔP2ΔF does not hold, the predicted power Pd is calculated in step 37 using the above-mentioned equation (1). The remaining time is less than 5 minutes, and ΔP1〈ΔP2〈
When ΔP3 or ΔP1>ΔP2>ΔP3 holds true, in step 36, the above (3
) Calculate the predicted power Pa using the equation. Next step 38
Checks whether the remaining time Tr is O or not, and if it is O, that is, if the demand period Td has elapsed, the power consumption Pt is changed to step 39 in order to make the demand start state again.
The number of pulses N for calculation is reset, and the remaining time Tr
The demand time Td is inputted to step 40. If the request period T:d has not elapsed, the process directly advances to step 40. Then, the predicted power Pd calculated in step 40 is displayed on the digital output device shown in FIG. The output is displayed on the display 11. Furthermore, in step 41, the predicted power Pa is compared with the contracted power Ps, and as a result, when it is considered that the amount of power used exceeds the contracted power Ps, that is, Pd
When ≧Ps, in step 42, a shearing signal is issued to the loads 10A to IOE to reduce the current load. We have given an example in which the predicted power is calculated using equation (1) before the remaining 5 minutes of the demand time and using equation (3) after that, but this time can be set to a value suitable for each load. By taking the equation (2) and equation (3) together, the predicted power can be calculated.

As described above, according to the present invention, at the beginning of the demand time, a relatively slow-response prediction method is used to obtain a predicted value based on the average power before the predicted time, and when the remaining time becomes short, Because it uses a sensitive prediction method that calculates the predicted value based on the rate of change immediately before the predicted time, it is possible to efficiently control the power within the set power value without performing unnecessary load control over the entire demand period. .

[Brief explanation of the drawing]

Figure 1 is a diagram showing the relationship between power consumption and demand time, Figure 2 is an enlarged view of the Δt portion of Figure 1, Figures 3 and 4 are comparison diagrams of predicted power between the conventional method and the method of the present invention, and Figure 5 FIG. 6 is a block diagram showing an example of a demand monitoring device for carrying out the method of the present invention, FIGS. Figures S and b are diagrams showing data areas used in the present invention. 1...Tode monitoring device, 2...Central processing unit, 3
...RAM) 4...ROMI5...Digital input device with latch function, 6,? ...Digital output device with latch function, 8...Water consumption meter, 9...
Unit time pulse oscillator, 10A to 10E... load or disconnection, 11... indicator, t... expected time, ΔP,,
ΔP2, ΔP3...average power, Δt...unit time.

Claims (1)

[Scope of Claims] 1. In order to control the amount of power used within a certain period of time to a predetermined value, in predicting the final amount of power used at a predetermined time within said time, a unit time before the predicted time is used. Find the rate of change in the average power immediately before the prediction time from the average power at each time, multiply this by the change in the average power immediately before the prediction time to obtain a value, and add this value to the average power at the prediction time to make the prediction. A power demand forecasting method characterized by calculating average power after a time. 2 In order to control the amount of electricity used within a certain period of time to a predetermined value, when predicting the final amount of electricity used at a predetermined time within the said period, the elapsed demand time is counted and the amount of electricity used is calculated until the scheduled elapsed time. calculates the average power before the prediction time as the average power after the prediction time, and after the scheduled elapsed time, calculates the rate of change in the average power immediately before the prediction time from the average power for each unit time before the prediction time. A power demand forecasting method characterized by multiplying this by the change in average power immediately before the prediction time to obtain a value, and adding this value to the average power at the average time to obtain the average power after the prediction time.