CN111476419A - Planned value prediction method of energy storage system and energy storage coordination control device - Google Patents

Planned value prediction method of energy storage system and energy storage coordination control device Download PDF

Info

Publication number
CN111476419A
CN111476419A CN202010269028.0A CN202010269028A CN111476419A CN 111476419 A CN111476419 A CN 111476419A CN 202010269028 A CN202010269028 A CN 202010269028A CN 111476419 A CN111476419 A CN 111476419A
Authority
CN
China
Prior art keywords
day
load
value
predicted
time point
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010269028.0A
Other languages
Chinese (zh)
Inventor
徐成斌
陈锐
黄植炜
楚启超
周国琦
肖声远
朱小帆
丁凯
宋代伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
CYG Sunri Co Ltd
Original Assignee
CYG Sunri Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by CYG Sunri Co Ltd filed Critical CYG Sunri Co Ltd
Priority to CN202010269028.0A priority Critical patent/CN111476419A/en
Publication of CN111476419A publication Critical patent/CN111476419A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention provides a method for predicting a planned value of an energy storage system, which comprises the following steps: step one, obtaining a local peak-valley price difference and historical load data of an energy storage system, and predicting a next-day load value through the historical load data to obtain a next-day load value; the historical load data comprises a load value at each time point in N days; step two, setting a constraint condition, taking the next day load value as the input of a KNN algorithm, taking the peak-to-valley price difference as the constraint condition, and predicting the planned value of the next day through the KNN algorithm to obtain predicted next day planned value data; the predicted next-day plan value data comprises a predicted value of each time point in the next day; and step three, charging and discharging the stored energy according to the predicted value of the time point in the next-day planning value data corresponding to the current time point. Compared with the prior art, the method achieves the purpose of realizing the constraint management of daily electricity consumption by utilizing the predicted plan value, improves the control accuracy and brings better economic benefit to users.

Description

Planned value prediction method of energy storage system and energy storage coordination control device
Technical Field
The present invention relates to power grid control, and in particular, to a method for predicting a planned value of an energy storage system and an energy storage coordination control apparatus.
Background
Today, the shortage of energy is increasing, people pay more and more attention to the reasonable and effective utilization of energy, and renewable energy plays more and more important role in the utilization. The power generation has random and intermittent renewable energy sources such as wind energy, solar energy and the like, and can generate impact on a power grid, and a large-scale malignant accident is caused in serious cases, so that a certain amount of stored energy is required to be provided in a direct current bus or an alternating current system to track the change of load. Therefore, an energy storage system (as shown in fig. 1) composed of distributed power sources can be matched with the capacity of a wind power/photovoltaic generator set, the rapid switching of charging and discharging states is supported, the safety and stability of the system are ensured, renewable energy sources can be fully utilized, the cascade utilization of the energy sources is realized, the environmental pollution is reduced, and the economical efficiency of the system is improved. Under the encouragement of national policies, distributed photovoltaic is rapidly developed, and the total installed capacity of photovoltaic power generation is estimated to reach 150GW by the end of 2020, wherein the cumulative installed capacity of distributed photovoltaic power generation reaches 70GW, which is close to half of the total installed capacity of photovoltaic power generation.
The energy storage application prospect is wide, the country greatly promotes the commercialized application of the energy storage, but the large-scale application of the energy storage is limited due to the high cost of the energy storage, so that the energy storage optimization configuration at the power grid side and the construction of a distributed electric energy storage facility at the user side can be mainly realized. On the other hand, the user side is mainly concerned with his/her own profit, and the idea of reducing the battery cost cannot be realized in a short time due to technical limitations. At present, the electricity charges paid by large industrial users are charged according to electricity consumption and capacity, for some manufacturing or heavy industrial users, the required amount of the electricity charges accounts for about 20% of the electricity charges, and a measurement rule adopted by a power supply company is that an intelligent electric meter enters a measurement period (15min) every 1 min. Thus, 1440 measurements per day, the smart meter only accessed the maximum value for the day, and if the value measured the next day was greater than the value recorded the previous day, the smart meter recorded the value that was automatically overwritten with the new value. The basic electricity fee the user needs to pay is the maximum demand reported in the month multiplied by the unit price. On this basis, if the portion of the maximum demand actually measured by the user exceeds the contractually agreed value by more than 5%, the excess portion is penalized by doubling it. Most energy storage systems all only have adopted simple peak clipping to fill valley mode at present, carry out the replenishment of electric energy through the energy storage discharge in power consumption peak period, charge the energy storage in power consumption valley period, but can't carry out comparatively accurate control to user's numerical value, so often can not make the user reach better income.
Energy storage mainly refers to the storage of electrical energy. The stored energy can be used as emergency energy, can also be used for storing energy when the load of the power grid is low, and can be used for outputting energy when the load of the power grid is high, so that the energy can be used for clipping peaks and filling valleys and reducing the fluctuation of the power grid.
Disclosure of Invention
The invention aims to provide a plan value prediction method of an energy storage system and an energy storage coordination control device, aiming at solving the technical problem of accurately carrying out peak clipping and valley filling on the maximum demand of a user by energy storage and improving the control accuracy.
In order to solve the problems, the invention adopts the following technical scheme that the method for predicting the planned value of the energy storage system comprises the following steps:
step one, obtaining a local peak-valley price difference and historical load data of an energy storage system, and predicting a next-day load value through the historical load data to obtain a next-day load value; the historical load data comprises a load value at each time point in N days;
step two, setting a constraint condition, taking the next day load value as the input of a KNN algorithm, taking the peak-to-valley price difference as the constraint condition, and predicting the planned value of the next day through the KNN algorithm to obtain predicted next day planned value data; the predicted next-day plan value data comprises a predicted value of each time point in the next day;
and step three, charging and discharging the stored energy according to the predicted value of the time point in the next-day planning value data corresponding to the current time point.
Further, after the second step, updating iteration is also performed on the predicted next-day planned value, and when an actual value at a certain time point is obtained, updating iteration is performed on the predicted value at a time point subsequent to the certain time point.
Further, the updating iteration of the predicted values of the time points after the time point is realized by a KNN algorithm.
Further, the obtaining of the historical load data in the first step is realized by adopting the following method:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t；
Wherein i represents the previous ith day, t represents the serial number corresponding to the time point in the day, and N is the total number of days of taking N days in total;
step S12, calculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure BDA0002442387030000021
wherein i represents the ith day, M represents the total number of time points in the day, t represents the serial number of the time points in the day, and Zi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line, wherein the fitting method adopts a least square method;
Figure BDA0002442387030000031
wherein the content of the first and second substances,
Figure BDA0002442387030000032
the method is characterized in that the daily average load after the least square method is represented, a is y-axis intercept, b is slope, and is quantity representing the inclination degree of a straight line relative to the abscissa axis, i represents day i, and the least square method can enable the error level to be small;
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Wherein a is the y-intercept, b is the slope representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the i-th day, XiCalculated by step S12;
adding the square of the error of N days
Figure BDA0002442387030000033
Figure BDA0002442387030000034
Wherein the content of the first and second substances,
Figure BDA0002442387030000035
is the sum of the squares of the errors over N days, a is the y-intercept, b is the slope, and is a quantity representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the day i, X represents the day iiCalculated by step S12;
to find
Figure BDA0002442387030000036
A and b are now also two unknown variables, and
Figure BDA0002442387030000037
the minimum is made by making its partial derivatives for a and b zero,
Figure BDA0002442387030000041
Figure BDA0002442387030000042
the coefficients a and b of the linear model are solved by the above equation:
Figure BDA0002442387030000043
step S14, predicting daily average load of the next day:
Figure BDA0002442387030000044
wherein a is y-axis intercept, b is slope, and is a quantity representing the inclination degree of a straight line about the axis of abscissa, the values of a and b are calculated above, namely the straight line is fitted, then the average load of the next day is predicted,
Figure BDA0002442387030000045
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation coefficient Si，tI.e. the load value at each time point divided by the daily average load on the day,
Figure BDA0002442387030000051
wherein i represents the i-th day, t represents the number of time points in the day, and Zi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure BDA0002442387030000052
wherein i represents the number of days, t represents the number of time points in the day,
Figure BDA0002442387030000053
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure BDA0002442387030000054
wherein the content of the first and second substances,
Figure BDA0002442387030000055
the predicted value of the load the next day,
Figure BDA0002442387030000056
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day.
Further, in the second step, the next-day plan value is predicted through a KNN algorithm to obtain a predicted next-day plan value, and the method is specifically realized by adopting the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points on an abscissa, and the test points are time points;
euclidean distance:
Figure BDA0002442387030000057
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
The invention also discloses an energy storage coordination control device, which comprises a power supply module, a main CPU module, a display module, a data acquisition module and an auxiliary CPU module, wherein the auxiliary CPU module sets constraint conditions for historical load data and real-time load data received from the main CPU module, predicts the next-day load value through the historical load data and uses the next-day load value as the input of a KNN algorithm, uses the peak-valley price difference as the constraint conditions, predicts the next-day plan value through the KNN algorithm, and feeds back the predicted next-day plan value data to the main CPU module after obtaining the predicted next-day plan value data; the predicted next-day plan value comprises a predicted value of each time point in the next day;
and the main CPU module charges and discharges the stored energy according to the predicted next-day plan value.
Further, the auxiliary CPU module updates and iterates the predicted next-day plan value according to the real-time load data obtained on the current day.
Further, updating and iterating the predicted value of the time point after the time point is realized by a KNN algorithm.
Further, the obtaining of the historical load data is realized by adopting the following method:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t；
Wherein i represents the previous ith day, t represents the serial number corresponding to the time point in the day, and N is the total number of days of taking N days in total;
step S12, countCalculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure BDA0002442387030000061
wherein i represents the ith day, M represents the total number of time points in the day, t represents the serial number of the time points in the day, and Zi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line, wherein the fitting method adopts a least square method;
Figure BDA0002442387030000062
wherein the content of the first and second substances,
Figure BDA0002442387030000063
the method is characterized in that the daily average load after the least square method is represented, a is y-axis intercept, b is slope, and is quantity representing the inclination degree of a straight line relative to the abscissa axis, i represents day i, and the least square method can enable the error level to be small;
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Wherein a is the y-intercept, b is the slope representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the i-th day, XiCalculated by step S12;
adding the square of the error of N days
Figure BDA0002442387030000071
Figure BDA0002442387030000072
Wherein the content of the first and second substances,
Figure BDA0002442387030000073
is the sum of the squares of the errors over N days, a is the y-intercept, b is the slope, and is a quantity representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the day i, X represents the day iiCalculated by step S12;
to find
Figure BDA0002442387030000074
A and b are now also two unknown variables, and
Figure BDA0002442387030000075
the minimum is made by making its partial derivatives for a and b zero,
Figure BDA0002442387030000076
Figure BDA0002442387030000077
the coefficients a and b of the linear model are solved by the above equation:
Figure BDA0002442387030000081
step S14, predicting daily average load of the next day:
Figure BDA0002442387030000082
wherein a is y-axis intercept, b is slope, and is a quantity representing the inclination degree of a straight line about the axis of abscissa, the values of a and b are calculated above, namely the straight line is fitted, then the average load of the next day is predicted,
Figure BDA0002442387030000083
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation systemNumber Si，tI.e. the load value at each time point divided by the daily average load on the day,
Figure BDA0002442387030000084
wherein i represents the i-th day, t represents the number of time points in the day, and Zi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure BDA0002442387030000085
wherein i represents the number of days, t represents the number of time points in the day,
Figure BDA0002442387030000086
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure BDA0002442387030000091
wherein the content of the first and second substances,
Figure BDA0002442387030000092
the predicted value of the load the next day,
Figure BDA0002442387030000093
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day.
Further, the next-day plan value is predicted through the KNN algorithm to obtain a predicted next-day plan value, and the method is specifically realized by adopting the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points on an x-axis coordinate, and the test points are time points;
euclidean distance:
Figure BDA0002442387030000094
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
Compared with the prior art, the method and the device have the advantages that the planned value of the stored energy is predicted through the KNN algorithm, the stored energy is charged and discharged through the peak-valley price difference and the predicted planned value, the peak clipping and valley filling effects are achieved, the constraint management on daily electricity quantity is achieved through the predicted planned value, the control accuracy is improved, and better economic benefits are brought to users.
Drawings
Fig. 1 is an application scenario of the present invention.
Fig. 2 is a processing diagram of the energy storage coordination control device of the invention.
FIG. 3 is a flow chart of a projected value prediction method of the present invention.
FIG. 4 is a flow chart of the implementation of the load prediction value of the present invention.
FIG. 5 is a flow chart of a projected value prediction implementation of the present invention.
Fig. 6 is a block diagram of the energy storage coordination control device according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
As shown in fig. 1, the energy storage coordination control device directly acquires voltage and current of a grid-connected point (PCC grid-connected point switch, that is, a point where the energy storage system is connected to a large power grid) of the system through a hard cable, and when detecting that frequency deviation or voltage amplitude deviation of the grid-connected point is out of limit, automatically and rapidly coordinates and controls the magnitude and direction of active and reactive power flows of the whole energy storage system, so as to support the power grid with emergency frequency and voltage. The main functions of the coordination control of the energy storage coordination control device are communication to the upper part (EMS, upper SCADA and power grid dispatching center), communication to the lower part (PCS (energy storage converter) and energy storage coordination controller submachine), alternating current sampling and switch control, support of dual-machine dual-network operation and peak clipping and valley filling.
The energy storage coordination control device is in real-time communication with each PCS managed under the jurisdiction through a serial port or an Ethernet modbus protocol, and the device adopts a set distribution strategy according to the capacity, SOC and other states of each PCS to realize the distribution and the rapid closed-loop tracking control of the total active and reactive power output targets of the energy storage system among the PCS.
The peak load shifting and the load adjusting are carried out by taking a measure of adjusting the power load, wherein an energy storage system absorbs electric quantity to store energy by utilizing the difference price between the peak price and the valley price of a power grid and releases the electric quantity at the peak value so as to reduce the peak load and fill the low valley load.
SOC (state of charge), which is a state of charge, is used to reflect the remaining capacity of the battery, and is numerically defined as a ratio of the remaining capacity to the battery capacity, and is usually expressed by a percentage, and the value ranges from 0 to 1, and indicates that the battery is completely discharged when the SOC is equal to 0, and indicates that the battery is completely charged when the SOC is equal to 1.
As shown in fig. 2 and fig. 3, the present invention discloses a planned value prediction method (planned value prediction method or method of an energy storage system) of an energy storage system based on a KNN algorithm, which includes the following steps:
step one, obtaining a local peak-valley price difference and historical load data of an energy storage system, and predicting a next-day load value through the historical load data to obtain a next-day load value; the peak-to-valley price difference refers to the difference between the fixed peak and valley periods of electricity consumption, such as 9: 00-12: 00 is the peak of electricity consumption, the charge is 1.6/degree, the valley period is 24: 00-8: 00, the charge is 0.6/degree, and the difference value between the two prices is the peak-valley price difference; the historical load data comprises a load value at each time point in N days;
step two, setting constraint conditions, namely using the next day load value as the input of a KNN (K nearest neighbor) algorithm, using the peak-valley price difference as the constraint conditions, and predicting the next day plan value through the KNN algorithm to obtain predicted next day plan value data; the predicted next-day planning value data comprises a predicted value of each time point in the next day, and each predicted value is a numerical value of each time point; the time points are taken every 5 minutes from 0 to 24;
and step three, charging and discharging the stored energy according to the predicted value of the time point in the next-day planning value data corresponding to the current time point.
After the second step, updating and iterating the predicted next-day plan value, and when an actual value of a certain time point is obtained, updating and iterating the predicted value of the time point after the certain time point, specifically, updating and iterating the predicted value of the time point after the certain time point is realized by a KNN algorithm, and the specific implementation steps are the same as the second step, and are not repeated here.
The acquisition of the historical load data in the first step is realized by adopting the following method: as shown in fig. 4:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t(ii) a Where i represents the previous ith day (i ═ 1,2, … … N), and t represents the ordinal number corresponding to the time point in the day, the ordinal number of the time point being defined by the ratio of each day from 0: 00 starting from 24: 00 minutes, and obtaining a time point every 5 minutes and sequencing the time points in sequence, wherein when t is 1, the corresponding time point is 0: when t is 2 at 05 minutes, the corresponding time point is 0: 10.… …, when t is 288, the corresponding time points are 24: 00 minutes, namely 288 total time points in the day, and so on; the above-mentionedN is the total number of days of the previous total N days;
step S12, calculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure BDA0002442387030000111
wherein i represents the ith day (i is 1,2, … … N), M represents the total number of time points in the day, t represents the serial number of the time points in the day (if t is 0: 05 minutes corresponding to the time point of 1), and Z represents the time point of the dayi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line by using a least square method (linear model);
Figure BDA0002442387030000112
wherein the content of the first and second substances,
Figure BDA0002442387030000113
the least square method can make the error level small, and the least square method is characterized in that the average daily load after the least square method is expressed, a is a y-axis intercept, b is a slope, and is a quantity (namely a relation between a time point and the average daily load) expressing the inclination degree of a straight line relative to a (horizontal) coordinate axis, i expresses the ith day (i is 1,2, … … N);
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Where a is the y-intercept, b is the slope, which is the amount of inclination of a straight line about the (abscissa) axis (i.e. the relationship between time point and daily average load), i is day i (i ═ 1,2, … … N), X isiCalculated by step S12;
adding the square of the error of N days
Figure BDA0002442387030000121
Figure BDA0002442387030000122
Wherein the content of the first and second substances,
Figure BDA0002442387030000123
the sum of the squares of the errors over N days, a being the y-intercept, b being the slope, is the amount by which a straight line is tilted about the (abscissa) axis (i.e. the relationship between time point and daily average load), i being the ith day (i ═ 1,2, … … N), XiCalculated by step S12;
to find
Figure BDA0002442387030000124
A and b are now also two unknown variables, and
Figure BDA0002442387030000125
the minimum is made by making its partial derivatives for a and b zero,
Figure BDA0002442387030000126
Figure BDA0002442387030000127
the coefficients a and b of the linear model are solved by the above equation:
Figure BDA0002442387030000131
step S14, predicting daily average load of the next day:
Figure BDA0002442387030000132
wherein a is the y-intercept and b is the slope, which is a quantity representing the degree of inclination of a straight line with respect to the (abscissa) axis, as already mentioned aboveAfter the values of a and b are calculated, the straight line fitting is completed, then the average load of the next day is predicted,
Figure BDA0002442387030000133
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation coefficient Si，tI.e. the load value at each time point is divided by the daily average load for the day (for example 288 coefficients are calculated per day, i.e. one value for every 5 minutes),
Figure BDA0002442387030000134
wherein i represents the ith day (i ═ 1,2, … … N), t represents the time point number (total number M) on that day, and Z represents the number of time points on that dayi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure BDA0002442387030000135
wherein i represents the number of days (i is 1,2, … … N), t represents the number of time points in the day (the total number of time points per day is M),
Figure BDA0002442387030000136
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure BDA0002442387030000141
wherein the content of the first and second substances,
Figure BDA0002442387030000142
the predicted value of the load the next day,
Figure BDA0002442387030000143
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day (the total number of time points per day is M).
As shown in fig. 5, in the second step, the planned value of the next day is predicted by the KNN algorithm to obtain a predicted planned value of the next day, and the method is specifically implemented by the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points (time points) on an x-axis (horizontal) coordinate, for example, 288 test points;
euclidean distance:
Figure BDA0002442387030000144
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
for example, sample coordinates (x, y) are trained, and then a test point (i.e., time point) coordinate (x) is given10), find the corresponding y on the regression value1The value is obtained. With KNN, the procedure is to take k discrete x1The most recent sample coordinates, then average their y values to get y1，x1Is variable, say for each time point;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
As shown in fig. 6, the present invention discloses an energy storage coordination control device, which mainly comprises a power module, a main CPU module, an auxiliary CPU module, a display module and a data acquisition module, wherein:
the main CPU module is respectively connected with the display module, the serial port expansion board, the data acquisition module and the auxiliary CPU module, and the power supply module is respectively connected with the main CPU module and the auxiliary CPU module;
the power module provides working power supply for each module, the input of the power module is AC220V, DC220V and DC110V which are compatible and adaptive, the power module outputs 24V power as the device input operation power supply, outputs 5V power to supply power for each plug-in, and outputs positive and negative 12V power for AD sampling of the alternating current board. And a remote control outlet power supply of the IO interface is opened through the starting relay and is controlled by the main CPU module.
The main CPU module is responsible for the coordination control of each module; sending the historical load data and the real-time load data acquired by the data acquisition module to the auxiliary CPU module, and simultaneously charging and discharging the stored energy according to the predicted next-day plan value, specifically, charging and discharging the stored energy according to the predicted value corresponding to the current time point;
the auxiliary CPU module sets constraint conditions for historical load data and real-time load data received from the main CPU module, predicts the next-day load value through the historical load data and uses the next-day load value as the input of a KNN (K nearest neighbor) algorithm, uses the peak-valley price difference as the constraint conditions, predicts the next-day plan value through the KNN algorithm, and feeds back the predicted next-day plan value data to the main CPU module after obtaining the predicted next-day plan value data; the predicted planning value of the next day comprises a predicted value of each time point in the next day, and each predicted value is a numerical value of each time point; the time points are taken every 5 minutes from 0 to 24;
the display module is responsible for displaying an interface;
the data acquisition module is used for acquiring and sending historical load data and real-time load data of the energy storage system to the main CPU module;
the alternating-current board is used for communication interaction between the main CPU module and equipment (such as charging piles) in the energy storage system, obtains alternating-current AD sampling data and sends the alternating-current AD sampling data to the main CPU module
The serial port expansion board expands the serial port, so that the energy storage system can be accessed to more devices and realize control and data interaction;
the auxiliary CPU module also updates and iterates the predicted next-day plan value according to the real-time load data obtained on the current day; specifically, updating and iterating the predicted value of the time point after the time point is realized by the KNN algorithm, and the specific implementation steps are the same as the method for predicting the planned value of the next day by the KNN algorithm described below, and are not described herein again.
The historical load data is obtained by adopting the following method: as shown in fig. 4:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t(ii) a Where i represents the previous ith day (i ═ 1,2, … … N), and t represents the ordinal number corresponding to the time point in the day, the ordinal number of the time point being defined by the ratio of each day from 0: 00 starting from 24: 00 minutes, and obtaining a time point every 5 minutes and sequencing the time points in sequence, wherein when t is 1, the corresponding time point is 0: when t is 2 at 05 minutes, the corresponding time point is 0: 10.… …, when t is 288, the corresponding time points are 24: 00 minutes, namely 288 total time points in the day, and so on; n is the total number of days of the previous total N days;
step S12, calculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure BDA0002442387030000161
wherein i represents the ith day (i is 1,2, … … N), M represents the total number of time points in the day, t represents the serial number of the time points in the day (if t is 0: 05 minutes corresponding to the time point of 1), and Z represents the time point of the dayi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line by least square method(linearized model);
Figure BDA0002442387030000162
wherein the content of the first and second substances,
Figure BDA0002442387030000163
the least square method can make the error level small, and the least square method is characterized in that the average daily load after the least square method is expressed, a is a y-axis intercept, b is a slope, and is a quantity (namely a relation between a time point and the average daily load) expressing the inclination degree of a straight line relative to a (horizontal) coordinate axis, i expresses the ith day (i is 1,2, … … N);
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Where a is the y-intercept, b is the slope, which is the amount of inclination of a straight line about the (abscissa) axis (i.e. the relationship between time point and daily average load), i is day i (i ═ 1,2, … … N), X isiCalculated by step S12;
adding the square of the error of N days
Figure BDA0002442387030000164
Figure BDA0002442387030000165
Wherein the content of the first and second substances,
Figure BDA0002442387030000166
the sum of the squares of the errors over N days, a being the y-intercept, b being the slope, is the amount by which a straight line is tilted about the (abscissa) axis (i.e. the relationship between time point and daily average load), i being the ith day (i ═ 1,2, … … N), XiCalculated by step S12;
to find
Figure BDA0002442387030000171
A and b are now also two unknown variables, and
Figure BDA0002442387030000172
the minimum is made by making its partial derivatives for a and b zero,
Figure BDA0002442387030000173
Figure BDA0002442387030000174
the coefficients a and b of the linear model are solved by the above equation:
Figure BDA0002442387030000175
step S14, predicting daily average load of the next day:
Figure BDA0002442387030000176
wherein a is the y-axis intercept, b is the slope, and is the amount representing the degree of inclination of a straight line with respect to the (abscissa) axis, the values of a and b have been calculated above, i.e. the fitting of the straight line has been completed, and then the average load of the next day is predicted,
Figure BDA0002442387030000177
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation coefficient Si，tI.e. the load value at each time point is divided by the daily average load for the day (for example 288 coefficients are calculated per day, i.e. one value for every 5 minutes),
Figure BDA0002442387030000181
wherein i represents day i (i ═ 1,2, … … N),t represents the number of time points (total number M) in the day, Zi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure BDA0002442387030000182
wherein i represents the number of days (i is 1,2, … … N), t represents the number of time points in the day (the total number of time points per day is M),
Figure BDA0002442387030000183
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure BDA0002442387030000184
wherein the content of the first and second substances,
Figure BDA0002442387030000185
the predicted value of the load the next day,
Figure BDA0002442387030000186
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day (the total number of time points per day is M);
as shown in fig. 5, the predicting the planned value of the next day by the KNN algorithm to obtain the predicted planned value of the next day specifically includes the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points (time points) on an x-axis (horizontal) coordinate, for example, 288 test points;
euclidean distance:
Figure BDA0002442387030000187
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
for example, sample coordinates (x, y) are trained, and then a test point (i.e., time point) coordinate (x) is given10), find the corresponding y on the regression value1The value is obtained. With KNN, the procedure is to take k discrete x1The most recent sample coordinates, then average their y values to get y1，x1Is variable, say for each time point;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
According to the method, the planned value of the stored energy is predicted through the KNN algorithm, the stored energy is charged and discharged through the peak-valley price difference and the predicted planned value, the peak clipping and valley filling effects are achieved, the constraint management on daily electricity consumption is achieved through the predicted planned value, the control accuracy is improved, and better economic benefits are brought to users.

Claims (10)

1. A planned value prediction method of an energy storage system comprises the following steps:
step one, obtaining a local peak-valley price difference and historical load data of an energy storage system, and predicting a next-day load value through the historical load data to obtain a next-day load value; the historical load data comprises a load value at each time point in N days;
step two, setting a constraint condition, taking the next day load value as the input of a KNN algorithm, taking the peak-to-valley price difference as the constraint condition, and predicting the planned value of the next day through the KNN algorithm to obtain predicted next day planned value data; the predicted next-day plan value data comprises a predicted value of each time point in the next day;
and step three, charging and discharging the stored energy according to the predicted value of the time point in the next-day planning value data corresponding to the current time point.
2. The planned value prediction method for an energy storage system according to claim 1, characterized in that: and after the second step, updating and iterating the predicted next-day plan value, and when the actual value of a certain time point is obtained, updating and iterating the predicted value of the time point after the certain time point.
3. The planned value prediction method of an energy storage system according to claim 2, characterized in that: and updating and iterating the predicted values of the time points after the time point through a KNN algorithm.
4. The planned value prediction method for an energy storage system according to claim 1, characterized in that: the acquisition of the historical load data in the first step is realized by adopting the following method:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t；
Wherein i represents the previous ith day, t represents the serial number corresponding to the time point in the day, and N is the total number of days of taking N days in total;
step S12, calculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure FDA0002442387020000011
wherein i represents day i and M represents the total of time points in the dayThe number, t, represents the number of time points in the day, Zi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line, wherein the fitting method adopts a least square method;
Figure FDA0002442387020000021
wherein the content of the first and second substances,
Figure FDA0002442387020000022
the method is characterized in that the daily average load after the least square method is represented, a is y-axis intercept, b is slope, and is quantity representing the inclination degree of a straight line relative to the abscissa axis, i represents day i, and the least square method can enable the error level to be small;
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Wherein a is the y-intercept, b is the slope representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the i-th day, XiCalculated by step S12;
adding the square of the error of N days
Figure FDA0002442387020000023
Figure FDA0002442387020000024
Wherein the content of the first and second substances,
Figure FDA0002442387020000025
is the sum of the squares of the errors over N days, a is the y-intercept, b is the slope, and is a quantity representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the day i, X represents the day iiCalculated by step S12;
to find
Figure FDA0002442387020000026
A and b are now also two unknown variables, and
Figure FDA0002442387020000027
the minimum is made by making its partial derivatives for a and b zero,
Figure FDA0002442387020000028
Figure FDA0002442387020000029
the coefficients a and b of the linear model are solved by the above equation:
Figure FDA0002442387020000031
step S14, predicting daily average load of the next day:
Figure FDA0002442387020000032
wherein a is y-axis intercept, b is slope, and is a quantity representing the inclination degree of a straight line about the axis of abscissa, the values of a and b are calculated above, namely the straight line is fitted, then the average load of the next day is predicted,
Figure FDA0002442387020000033
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation coefficient Si，tI.e. the load value at each time point divided by the daily average load on the day,
Figure FDA0002442387020000034
wherein i represents the i-th day, t represents the number of time points in the day, and Zi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure FDA0002442387020000035
wherein i represents the number of days, t represents the number of time points in the day,
Figure FDA0002442387020000036
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure FDA0002442387020000041
wherein the content of the first and second substances,
Figure FDA0002442387020000042
the predicted value of the load the next day,
Figure FDA0002442387020000043
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day.
5. The planned value prediction method for an energy storage system according to claim 1, characterized in that: predicting the planned value of the next day through a KNN algorithm in the second step to obtain a predicted planned value of the next day, and specifically adopting the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points on an abscissa, and the test points are time points;
euclidean distance:
Figure FDA0002442387020000044
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
6. The utility model provides an energy storage coordination control device, includes power module, main CPU module, display module and data acquisition module, its characterized in that: the auxiliary CPU module sets constraint conditions for historical load data and real-time load data received from the main CPU module, predicts the next-day load value through the historical load data and uses the next-day load value as the input of a KNN algorithm, predicts the next-day plan value through the KNN algorithm by using the peak-valley price difference as the constraint conditions, and feeds back the predicted next-day plan value data to the main CPU module after obtaining the predicted next-day plan value data; the predicted next-day plan value comprises a predicted value of each time point in the next day;
and the main CPU module charges and discharges the stored energy according to the predicted next-day plan value.
7. The energy storage coordination control device according to claim 6, characterized in that: and the auxiliary CPU module also updates and iterates the predicted next-day plan value according to the real-time load data obtained on the current day.
8. The energy storage coordination control device according to claim 7, characterized in that: and updating and iterating the predicted values of the time points after the time point by a KNN algorithm.
9. The energy storage coordination control device according to claim 6, characterized in that: the historical load data is obtained by adopting the following method:
step S11, setting the load value at the t-th time point in the previous N days as Zi,t；
Wherein i represents the previous ith day, t represents the serial number corresponding to the time point in the day, and N is the total number of days of taking N days in total;
step S12, calculating daily average load X of each dayiThe average value is obtained after all load values of each day are added:
Figure FDA0002442387020000051
wherein i represents the ith day, M represents the total number of time points in the day, t represents the serial number of the time points in the day, and Zi,tRepresents the load value at the t time point of the ith day;
step S13, average daily load X of N days to be obtainediFitting into a straight line, wherein the fitting method adopts a least square method;
Figure FDA0002442387020000052
wherein the content of the first and second substances,
Figure FDA0002442387020000053
the average daily load after least square method, a is y-intercept, b is slope, and is quantity representing inclination degree of a straight line with respect to abscissa axis, i is day i, and the least square method can beThe error level is small;
then, an error r between the daily average load obtained in step S12 and the daily average load of the least square method corresponding to the straight line obtained in step S13 is calculatedi，
ri＝a+bi-Xi
Wherein a is the y-intercept, b is the slope representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the i-th day, XiCalculated by step S12;
adding the square of the error of N days
Figure FDA0002442387020000054
Figure FDA0002442387020000055
Wherein the content of the first and second substances,
Figure FDA0002442387020000061
is the sum of the squares of the errors over N days, a is the y-intercept, b is the slope, and is a quantity representing the degree of inclination of a straight line with respect to the abscissa axis, i represents the day i, X represents the day iiCalculated by step S12;
to find
Figure FDA0002442387020000062
A and b are now also two unknown variables, and
Figure FDA0002442387020000063
the minimum is made by making its partial derivatives for a and b zero,
Figure FDA0002442387020000064
Figure FDA0002442387020000065
the coefficients a and b of the linear model are solved by the above equation:
Figure FDA0002442387020000066
step S14, predicting daily average load of the next day:
Figure FDA0002442387020000067
wherein a is y-axis intercept, b is slope, and is a quantity representing the inclination degree of a straight line about the axis of abscissa, the values of a and b are calculated above, namely the straight line is fitted, then the average load of the next day is predicted,
Figure FDA0002442387020000068
represents the daily average load of the following day; i is the first day;
step S15, obtaining daily load cycle variation coefficient Si，tI.e. the load value at each time point divided by the daily average load on the day,
Figure FDA0002442387020000071
wherein i represents the i-th day, t represents the number of time points in the day, and Zi,tDenotes the load value, X, at the t-th time point on the i-th dayiRepresents the daily average load on day i;
step S16, calculating the average value of the daily load cycle variation coefficient N balance, namely adding the obtained daily load cycle variation coefficients of N days and then obtaining the average value as the predicted value of the load cycle variation coefficient of the next day;
Figure FDA0002442387020000072
wherein i represents the number of days, t represents the number of time points in the day,
Figure FDA0002442387020000073
the daily load cycle change coefficient corresponding to each predicted time point;
step S17, predicting the next day load value, namely multiplying the predicted average daily load of the next day by the daily load periodic variation coefficient corresponding to each predicted time point to obtain t predicted next day load values;
Figure FDA0002442387020000074
wherein the content of the first and second substances,
Figure FDA0002442387020000075
the predicted value of the load the next day,
Figure FDA0002442387020000076
coefficient of daily load cycle variation, X, for each time point representing the next day predictionN+1Daily average load for the next day; t represents the number of time points in the day.
10. The energy storage coordination control device according to claim 6, characterized in that: predicting the planned value of the next day through a KNN algorithm to obtain a predicted planned value of the next day, and specifically adopting the following steps:
step S21, firstly, the predicted next-day load value is used as a training sample in training data, and the distance between the test data and each training data is calculated through Euclidean distance; the test data is composed of test points on an x-axis coordinate, and the test points are time points;
euclidean distance:
Figure FDA0002442387020000077
wherein d isnRepresenting the distance, x, of the training data from the test sample0For the abscissa, x, of the test specimennTo train the abscissa, y, of the specimennIs the ordinate of the training sample;
s22, sorting according to the increasing relation of the distances;
s23, selecting K points with the minimum distance; 5 of the K are taken;
and step S24, averaging the K values to obtain a predicted next-day plan value.
CN202010269028.0A 2020-04-08 2020-04-08 Planned value prediction method of energy storage system and energy storage coordination control device Pending CN111476419A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010269028.0A CN111476419A (en) 2020-04-08 2020-04-08 Planned value prediction method of energy storage system and energy storage coordination control device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010269028.0A CN111476419A (en) 2020-04-08 2020-04-08 Planned value prediction method of energy storage system and energy storage coordination control device

Publications (1)

Publication Number Publication Date
CN111476419A true CN111476419A (en) 2020-07-31

Family

ID=71750168

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010269028.0A Pending CN111476419A (en) 2020-04-08 2020-04-08 Planned value prediction method of energy storage system and energy storage coordination control device

Country Status (1)

Country Link
CN (1) CN111476419A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116231696A (en) * 2023-04-04 2023-06-06 重庆跃达新能源有限公司 Energy storage switching control method and system based on load prediction

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102624017A (en) * 2012-03-22 2012-08-01 清华大学 Battery energy storage system peak clipping and valley filling real-time control method based on load prediction
CN102868155A (en) * 2012-08-28 2013-01-09 深圳蓝波幕墙及光伏工程有限公司 Power load forecasting control system and method
CN107248751A (en) * 2017-07-26 2017-10-13 武汉大学 A kind of energy storage station dispatch control method for realizing distribution network load power peak load shifting
CN107370170A (en) * 2017-06-23 2017-11-21 浙江大学 A kind of energy storage system capacity collocation method for considering capacity price of electricity and load prediction error
CN107846034A (en) * 2017-11-28 2018-03-27 甘肃省电力公司风电技术中心 A kind of energy storage for stabilizing wind power plant climbing rate goes out force control method
CN108734355A (en) * 2018-05-24 2018-11-02 国网福建省电力有限公司 A kind of short-term electric load method of parallel prediction and system applied to power quality harnessed synthetically scene
CN110176765A (en) * 2019-05-30 2019-08-27 东北电力大学 A kind of energy storage peak shaving of time-of-use tariffs driving Optimization Scheduling a few days ago
CN110460075A (en) * 2019-08-21 2019-11-15 国网河南省电力公司电力科学研究院 A kind of hybrid energy-storing for stabilizing power grid peak-valley difference goes out force control method and system

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102624017A (en) * 2012-03-22 2012-08-01 清华大学 Battery energy storage system peak clipping and valley filling real-time control method based on load prediction
CN102868155A (en) * 2012-08-28 2013-01-09 深圳蓝波幕墙及光伏工程有限公司 Power load forecasting control system and method
CN107370170A (en) * 2017-06-23 2017-11-21 浙江大学 A kind of energy storage system capacity collocation method for considering capacity price of electricity and load prediction error
CN107248751A (en) * 2017-07-26 2017-10-13 武汉大学 A kind of energy storage station dispatch control method for realizing distribution network load power peak load shifting
CN107846034A (en) * 2017-11-28 2018-03-27 甘肃省电力公司风电技术中心 A kind of energy storage for stabilizing wind power plant climbing rate goes out force control method
CN108734355A (en) * 2018-05-24 2018-11-02 国网福建省电力有限公司 A kind of short-term electric load method of parallel prediction and system applied to power quality harnessed synthetically scene
CN110176765A (en) * 2019-05-30 2019-08-27 东北电力大学 A kind of energy storage peak shaving of time-of-use tariffs driving Optimization Scheduling a few days ago
CN110460075A (en) * 2019-08-21 2019-11-15 国网河南省电力公司电力科学研究院 A kind of hybrid energy-storing for stabilizing power grid peak-valley difference goes out force control method and system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王士政: "电力*** 运行控制与调度自动化", vol. 1, 河海大学出版社, pages: 182 - 184 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116231696A (en) * 2023-04-04 2023-06-06 重庆跃达新能源有限公司 Energy storage switching control method and system based on load prediction
CN116231696B (en) * 2023-04-04 2023-09-12 重庆跃达新能源有限公司 Energy storage switching control method and system based on load prediction

Similar Documents

Publication Publication Date Title
Li et al. The capacity allocation method of photovoltaic and energy storage hybrid system considering the whole life cycle
Yang et al. Optimal two-stage dispatch method of household PV-BESS integrated generation system under time-of-use electricity price
Liao et al. BESS-sizing optimization for solar PV system integration in distribution grid
CN104158203A (en) Micro-grid power supply capacity optimization configuration method
CN111969655A (en) Multi-source multi-load coordination scheduling method and system considering load response uncertainty
CN114884108B (en) Multi-time scale energy management optimization method for source network charge storage integrated micro-grid
CN115765015A (en) Source network load storage cooperative interaction scheme making method oriented to power grid practical application scene
CN109995025A (en) A kind of energy storage control method based on two-part rate system price system
CN116526452A (en) Electric energy prediction metering monitoring mains load light stores up integrative control system
CN109617066A (en) Energy storage configuration method based on system of electricity price
CN111476419A (en) Planned value prediction method of energy storage system and energy storage coordination control device
Tian et al. Coordinated RES and ESS Planning Framework Considering Financial Incentives Within Centralized Electricity Market
Kyrylenko et al. Power Systems Research and Operation: Selected Problems III
Bai et al. An online multi-level energy management system for commercial building microgrids with multiple generation and storage systems
KR102471609B1 (en) Photovoltaics generation forecasting system
Zhang et al. Optimal configuration of user-side energy storage considering load response and demand management
Roy et al. Stochastic power allocation of distributed tri-generation plants and energy storage units in a zero bus microgrid with electric vehicles and demand response
Hafiz et al. Network constraints consideration for grid-edge energy management system
Yang et al. Energy Storage Configuration Optimization Method for Industrial Park Microgrid Based on Demand Side Response
Xue et al. Flow battery control strategy implementation and cost benefit analysis for microgrid under updated time-of-use periods
Xu et al. Smart grid dispatching strategy considering the difference of electric vehicle demand
Chen et al. Optimal scheduling strategy with multi-time scales for PV-storage based charging station
Luo et al. Two-stage optimal scheduling of virtual power plant considering demand response and forecast errors
Zeng et al. Active residential load management based on dynamic real time electricity price of carbon emission
KR102472630B1 (en) Power generation management system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination