CN117638939A - Water-light complementary optimization scheduling method based on Adam algorithm consideration - Google Patents
Water-light complementary optimization scheduling method based on Adam algorithm consideration Download PDFInfo
- Publication number
- CN117638939A CN117638939A CN202311614599.3A CN202311614599A CN117638939A CN 117638939 A CN117638939 A CN 117638939A CN 202311614599 A CN202311614599 A CN 202311614599A CN 117638939 A CN117638939 A CN 117638939A
- Authority
- CN
- China
- Prior art keywords
- period
- power
- water
- photovoltaic
- representing
- 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
Links
- 230000000295 complement effect Effects 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 50
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 title claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 42
- 238000005457 optimization Methods 0.000 title claims abstract description 33
- 238000010248 power generation Methods 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 9
- 230000001105 regulatory effect Effects 0.000 claims abstract description 6
- 230000000452 restraining effect Effects 0.000 claims abstract description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 73
- 238000011156 evaluation Methods 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 16
- 238000005315 distribution function Methods 0.000 claims description 12
- 238000011144 upstream manufacturing Methods 0.000 claims description 12
- 230000000694 effects Effects 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000013459 approach Methods 0.000 claims description 3
- 230000003750 conditioning effect Effects 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000005611 electricity Effects 0.000 claims description 3
- 238000002310 reflectometry Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 3
- 229910052799 carbon Inorganic materials 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000029087 digestion Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a water-light complementary optimization scheduling method based on Adam algorithm consideration, which relates to the technical field of power scheduling technology and comprises the steps of establishing a peak regulation model of water-light complementary optimization operation; constraining the peak regulation model based on a constraint condition of hydroelectric generation; restraining the peak regulation model based on the constraint condition of photovoltaic power generation; optimizing the constraint conditions of hydroelectric generation and photovoltaic power generation by using an Adam algorithm; evaluating the stability of the power station according to the parameters of the peak regulation model; and solving the peak shaving model, and obtaining a scheduling method. By using the Adam algorithm, the convergence speed is high, the training time is reduced, the prediction error is reduced, the super parameters are few, the method is simple to realize, the processing memory is reduced, the coefficient gradient problem is solved, the operation of a power station can be quickly and effectively regulated, and the stable and effective operation of a power grid is ensured.
Description
Technical Field
The invention relates to the technical field of power dispatching technologies, in particular to a water-light complementary optimization dispatching method based on Adam algorithm consideration.
Background
Under the background of double carbon, in consideration of the pattern of large-scale high-proportion new energy development of future electric power systems in China, clean energy sources such as hydropower, photovoltaics and the like can be used as reliable power sources for bearing flexible regulation functions for a long time, play a key role in realizing the goals of carbon peak, carbon neutralization and building a novel electric power system, and contribute irreplaceable strength for safe and stable operation of the electric power system and electric power consumption of the new energy sources.
However, as the photovoltaic power generation has volatility and uncertainty, the direct access to the power grid can cause damage to the safe and stable operation of the photovoltaic power generation, and the peak shaving difficulty is increased. Hydropower stations have the advantage of being able to quickly start, stop or regulate power generation. Therefore, the flexible adjustment system using water-light complementation is a feasible method for realizing grid-connected digestion. The scholars at home and abroad have made a great deal of research on generating and describing the prediction of the power generation under the general scene of the random factors, but are difficult to predict the power generation under all scenes. The water-light complementary form in the prior art has poor adjustment capability, high photovoltaic light rejection rate, light energy waste and high peak shaving difficulty.
Disclosure of Invention
The invention solves the technical problems that: the water-light complementary form in the prior art has poor adjustment capability, high photovoltaic light rejection rate, light energy waste and high peak shaving difficulty.
In order to solve the technical problems, the invention provides the following technical scheme: a water-light complementary optimization scheduling method based on Adam algorithm consideration comprises the steps of establishing a peak shaving model of water-light complementary optimization operation; constraining the peak regulation model based on a constraint condition of hydroelectric generation; restraining the peak regulation model based on the constraint condition of photovoltaic power generation; optimizing the constraint conditions of hydroelectric generation and photovoltaic power generation by using an Adam algorithm; evaluating the stability of the power station according to the parameters of the peak regulation model; and solving the peak shaving model, and obtaining a scheduling method.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: the method for establishing the peak shaving model of the water-light complementary optimization operation comprises the following steps:
the mathematical expression of the peak shaving model is as follows:
P max ＝max<P h,i +P s,i >i∈[1,T]
wherein P is max Representing the maximum load borne by the system during the conditioning periodValue, P h,i Representing the efficacy value, P, of the hydro-generator set h during the ith period s,i The efficacy value of the photovoltaic generator set s in the ith period is represented, and T represents the total period length of the regulating period.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: constraining the peak shaving model based on the constraint condition of hydroelectric generation comprises:
the constraint conditions of hydroelectric generation comprise power balance constraint, water balance constraint, start-stop duration constraint, inter-step hydraulic connection constraint, water head calculation constraint and water head loss calculation constraint of hydroelectric generation;
the power balance constraints of hydroelectric generation include:
N y,i ＝N h,i +N s,i
wherein N is y,i Representing the load value of the power system born by the virtual hydroelectric generating set y in the ith period;
the water balance constraint includes:
wherein,representing the reservoir capacity of hydropower station m at the beginning of period t,/day>Representing the storage capacity of the hydroelectric power station m at the end of the t period, m 3 Indicating the flow of the hydroelectric power plant m in the warehouse during the t-th period,/->Indicating the flow of hydropower station m out of the warehouse during period t,/->Representing the flow of a hydroelectric power plant m in storage in period t, < >>Represents the flow of electricity generated by hydropower station m during period t, < >>The flow of the water discarded by the hydroelectric power station m during the period t, deltat representing the calculated time step,/->Represents the flow rate of the hydroelectric power station m in the t period, m 3 /s，/>Representing the delivery flow of the power station m-1 immediately upstream of the hydroelectric power station m in the period t, m 3 /s；
The on-off duration constraint includes:
wherein if itIndicating that the power station m is started up from the t-th period, and at least continuously operating at I m The machine can be stopped only in a certain time period;
if it isIndicating that the plant m is shut down from the t-th period, then at least O is continued m Starting up only in a certain time period;
the inter-ladder hydraulic link constraints include:
wherein,indicating the section flow of hydropower station m in period t, +.>The kth direct upstream power station representing hydropower station m is in period T-T o,i Time period of delivery flow, T o,i Represents the number of water flow delay time periods from the upstream power station k to the power station m, phi IUm Representing a collection of directly upstream power stations of hydropower station m;
the water head calculation includes:
wherein Z is m,t Representing the dam front water level, Z, of a hydropower station m in a period t m,t Is the dam front water level of the hydropower station m in the period of t+1,for the tail water level of the hydropower station m in the period t, h m,t Represents the head of hydropower station m during period t, < >>Representing the head loss of the hydropower station m in a period t;
the head loss calculation includes:
wherein,a function representing the head loss of hydropower station m with respect to the flow out of the warehouse, +.>Represents the ex-warehouse flow of the hydropower station m in the period t, m 3 /s。
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: the restraining of the peak shaving model based on the constraint condition of photovoltaic power generation comprises the following steps:
the constraint condition of the photovoltaic power generation is specifically solar irradiance constraint, solar irradiance obeys Beta distribution in a certain period, probability density function of solar irradiance is expressed as follows:
wherein s represents the current irradiance sor t Maximum output power from photovoltaicCorresponding irradiance sor max Ratio of (i.e. s=sor) t /sor max Alpha and Beta represent shape parameters of Beta distribution, r represents gamma function;
wherein pv represents abbreviation of distributed light source, μ represents average value, σ represents standard deviation;
s represents the area of a photovoltaic power station, N represents the number of the photovoltaic power stations, eta represents the surface reflectivity, solar irradiance historical data are obtained according to local meteorological data of a virtual power station, a solar irradiance distribution function is determined, photovoltaic output distribution functions and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver for optimization prediction, and the photovoltaic power generation power under the current constraint is obtained;
in the model training process, an optimization algorithm is needed to calculate and update the network parameters trained by the model, so that the network parameters approach to an optimal value as much as possible, an optimal model is obtained, and accurate prediction of the data change trend is realized.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: optimizing the constraint condition of the hydroelectric generation and the constraint condition of the photovoltaic generation by using an Adam algorithm comprises the following steps:
according to the local meteorological data of the virtual power plant, solar irradiance historical data are obtained, a solar irradiance distribution function is determined, photovoltaic output distribution function and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver, and optimization prediction is carried out on the data to obtain photovoltaic power generation power under the current constraint.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: and correcting the paranoid of the constraint condition by using an Adam algorithm, wherein the mathematical expression is as follows:
m t ＝β 1 m t-1 +(1-β 1 )g t
wherein m is t Representing first order motion terms, V t Represent the second order motion term, g t Representing beta 1 And beta 2 For the power values, default values are 0.9 and 0.999, m t And V t The deviation correction of (a) is respectively as followsAnd->The mathematical expression is as follows:
at time t+1, i.e. the parameter θ of the model at time t+1st iteration t+1 The mathematical expression of (2) is:
wherein θ t The parameter representing the t time, namely the t time iteration model, eta represents the super parameter and defaults to 10 -5 Epsilon is a small number (10 -8 ) The t th iteration function is related to theta t The gradient calculation formula of (2) is:
for the loss function L Huber And (3) performing calculation, wherein the mathematical expression is as follows:
wherein Y represents a predicted value, f (x) represents a true value, when L Huber The iteration times and training times when the model reaches the optimum can be obtained in the minimum time;
the parameter is updated by estimating self-adaptive adjustment learning rate through the first-order and second-order motion terms of the gradient, and after the threshold is reached, the optimal solution is considered to be close to, and the updated parameter theta is returned t 。
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: the evaluation of the stability of the power station according to the parameters of the peak shaving model comprises the following steps:
the effects on the complementary system include effects on the power system and reservoir effects;
the influence on the power system takes the output difference coefficient of the complementary system and the photovoltaic output prediction error as judging indexes;
the influence on the reservoir is judged by using the change of the reservoir water level as an index.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: the output difference coefficient of the complementary system is used as an index for judging the stability of the photovoltaic output and the water output, and the mathematical expression is as follows:
wherein n represents the total time period number of evaluation, P represents the output of the complementary system, and P z' (i) Representing the actual output of the complementary system during the ith evaluation period,representing the average force of the complementary system.
Photovoltaic power error prediction includes:
and analyzing the influence of the photovoltaic uncontrollable power supply access system on the system output by using an average absolute error (MAE) and a Root Mean Square Error (RMSE), wherein the mathematical expression is as follows:
wherein P' (i) represents the difference value between the predicted output and the actual output of the photovoltaic power generation and the complementary power generation in different evaluation periods;
the reservoir water level fluctuation error prediction comprises the following steps:
selecting reservoir water level fluctuation as a judging index, selecting water level fluctuation of the maximum water level fluctuation day in a period of time for daily fluctuation analysis, and further evaluating the influence of system output on water level change of the reservoir, wherein the mathematical expression is as follows:
wherein Z (i) represents the water level in the period before evaluation, Z (i-1) represents the water level in the period after evaluation, and Δt represents the evaluation period.
As a preferable scheme of the water-light complementary optimization scheduling method based on Adam algorithm, the invention comprises the following steps: solving the peak shaving model and obtaining a scheduling method comprises the following steps:
solving the peak shaving model to obtain respective output conditions of the photovoltaic and the water output in each period, and adjusting the respective output of the photovoltaic and the water output in each period to carry out peak shaving.
The invention has the beneficial effects that: by using the Adam algorithm, the convergence speed is high, the training time is reduced, the prediction error is reduced, the super parameters are few, the method is simple to realize, the processing memory is reduced, the coefficient gradient problem is solved, the operation of a power station can be quickly and effectively regulated, and the stable and effective operation of a power grid is ensured.
Drawings
Fig. 1 is a basic flow diagram of a water-light complementary optimization scheduling method based on Adam algorithm consideration according to an embodiment of the present invention.
Fig. 2 is a graph of a dispatching result of a water-light complementary power generation system based on a water-light complementary optimization dispatching method considered by Adam algorithm according to an embodiment of the invention.
Fig. 3 is a basic flow diagram of a water-light complementary optimization scheduling method based on Adam algorithm consideration according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of the respective output of water and light in each period of time of a water-light complementary optimization scheduling method based on Adam algorithm consideration according to an embodiment of the present invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings.
Referring to fig. 1 and 2, for one embodiment of the present invention, there is provided a water-light complementary optimization scheduling method based on Adam algorithm, including:
step one: the method for establishing the peak shaving model of the water-light complementary optimization operation comprises the following steps:
the mathematical expression of the peak shaving model is as follows:
P max ＝max<P h,i +P s,i >i∈[1,T]
wherein P is max Representing the maximum load value, P, assumed by the system during the conditioning period h,i Representing the efficacy value, P, of the hydro-generator set h during the ith period s,i The efficacy value of the photovoltaic generator set s in the ith period is represented, and T represents the total period length of the regulating period.
Step two: constraining the peak shaving model based on the constraint condition of hydroelectric generation comprises:
the constraint conditions of hydroelectric generation comprise power balance constraint, water balance constraint, start-stop duration constraint, inter-step hydraulic connection constraint, water head calculation constraint and water head loss calculation constraint of hydroelectric generation;
the power balance constraints of hydroelectric generation include:
N y,i ＝N h,i +N s,i
wherein N is y,i Representing the load value of the power system born by the virtual hydroelectric generating set y in the ith period;
the water balance constraint includes:
wherein,representing the reservoir capacity of hydropower station m at the beginning of period t,/day>Representing the storage capacity of the hydroelectric power station m at the end of the t period, m 3 Indicating the flow of the hydroelectric power plant m in the warehouse during the t-th period,/->Indicating the flow of hydropower station m out of the warehouse during period t,/->Representing the flow of a hydroelectric power plant m in storage in period t, < >>Represents the flow of electricity generated by hydropower station m during period t, < >>The flow of the water discarded by the hydroelectric power station m during the period t, deltat representing the calculated time step,/->Represents the flow rate of the hydroelectric power station m in the t period, m 3 /s，/>Representing the delivery flow of the power station m-1 immediately upstream of the hydroelectric power station m in the period t, m 3 /s；
The on-off duration constraint includes:
wherein if itRepresentation ofThe power station m is started from the t time period and at least continues to be I m The machine can be stopped only in a certain time period;
if it isIndicating that the plant m is shut down from the t-th period, then at least O is continued m Starting up only in a certain time period;
the inter-ladder hydraulic link constraints include:
wherein,indicating the section flow of hydropower station m in period t, +.>The kth direct upstream power station representing hydropower station m is in period T-T o,i Time period of delivery flow, T o,i Represents the number of water flow delay time periods from the upstream power station k to the power station m, phi IUm Representing a collection of directly upstream power stations of hydropower station m;
the water head calculation includes:
wherein Z is m,t Representing the dam front water level, Z, of a hydropower station m in a period t m,t Is the dam front water level of the hydropower station m in the period of t+1,for the tail water level of the hydropower station m in the period t, h m,t Represents the head of hydropower station m during period t, < >>Representing the head loss of the hydropower station m in a period t;
the head loss calculation includes:
wherein,a function representing the head loss of hydropower station m with respect to the flow out of the warehouse, +.>Represents the ex-warehouse flow of the hydropower station m in the period t, m 3 /s。
Step three: the restraining of the peak shaving model based on the constraint condition of photovoltaic power generation comprises the following steps:
solar irradiance is the source of uncertainty in photovoltaic output. The constraint condition of the photovoltaic power generation is specifically solar irradiance constraint:
the solar irradiance obeys Beta distribution in a certain period, the probability density function of the solar irradiance is expressed as follows:
wherein s represents the current irradiance sor t Maximum output power from photovoltaicCorresponding irradiance sor max Ratio of (i.e. s=sor) t /sor max Alpha and Beta represent shape parameters of Beta distribution, r represents gamma function;
wherein pv represents abbreviation of distributed light source, μ represents average value, σ represents standard deviation;
s represents the area of a photovoltaic power station, N represents the number of the photovoltaic power stations, eta represents the surface reflectivity, solar irradiance historical data are obtained according to local meteorological data of a virtual power station, a solar irradiance distribution function is determined, photovoltaic output distribution functions and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver for optimization prediction, and the photovoltaic power generation power under the current constraint is obtained;
in the model training process, an optimization algorithm is needed to calculate and update the network parameters trained by the model, so that the network parameters approach to an optimal value as much as possible, an optimal model is obtained, and accurate prediction of the data change trend is realized.
The magnitude of the gradient values was observed with attention paid to the problem of gradient extinction and gradient explosion. When the gradient value is close to zero, it is explained that the magnitude of the weight update becomes small, and the optimum value is already approached until the magnitude of the weight update is 0, thereby obtaining the optimum value.
Step four: optimizing the constraint condition of the hydroelectric generation and the constraint condition of the photovoltaic generation by using an Adam algorithm comprises the following steps:
according to the local meteorological data of the virtual power plant, solar irradiance historical data are obtained, a solar irradiance distribution function is determined, photovoltaic output distribution function and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver, and optimization prediction is carried out on the data to obtain photovoltaic power generation power under the current constraint.
Step five: and correcting the paranoid of the constraint condition by using an Adam algorithm, wherein the mathematical expression is as follows:
m t ＝β 1 m t-1 +(1-β 1 )g t
wherein m is t Representing first order motion terms, V t Represent the second order motion term, g t Representing beta 1 And beta 2 For the power values, default values are 0.9 and 0.999, m t And V t The deviation correction of (a) is respectively as followsAnd->The mathematical expression is as follows:
at time t+1, i.e. the parameter θ of the model at time t+1st iteration t+1 The mathematical expression of (2) is:
wherein θ t The parameter representing the t time, namely the t time iteration model, eta represents the super parameter and defaults to 10 -5 Epsilon is a small number (10 -8 ) The objective is to avoid denominator 0, the t-th iteration function is related to θ t The gradient calculation formula of (2) is:
for the loss function L Huber And (3) performing calculation, wherein the mathematical expression is as follows:
wherein Y represents a predicted value, f (x) represents a true value, when L Huber The iteration times and training times when the model reaches the optimum can be obtained in the minimum time;
the parameter is updated by estimating self-adaptive adjustment learning rate through the first-order and second-order motion terms of the gradient, and after the threshold is reached, the optimal solution is considered to be close to, and the updated parameter theta is returned t 。
The evaluation of the stability of the power station according to the parameters of the peak shaving model comprises the following steps:
the complementary system is provided with an uncontrollable power source such as photovoltaic, and the instability of the output of the uncontrollable power source can have a certain influence on the complementary system. The influence on the complementary system is mainly divided into an influence on the power system and an influence on the reservoir, and the influence on the power system takes the output difference coefficient of the complementary system and the photovoltaic output prediction error as judgment indexes; the influence on the reservoir is judged by using the change of the reservoir water level as an index.
Coefficient of difference of complementary output A c ：
The complementarity index is the basis of the operation characteristic judgment of the combined power generation system, and the complementary degree of various energy sources can be primarily judged.
The effects on the complementary system include effects on the power system and reservoir effects;
the influence on the power system takes the output difference coefficient of the complementary system and the photovoltaic output prediction error as judging indexes;
the influence on the reservoir is judged by using the change of the reservoir water level as an index.
The photovoltaic output power has stronger time complementarity, meanwhile, the hydropower can further stabilize photovoltaic fluctuation by utilizing the self rapid adjustment capability, so that the complementary system is more stable, and the complementary system is quantized in power complementation capability, so that the output difference coefficient of the complementary system is used as an index for judging the output stability of the photovoltaic water, and the mathematical expression is as follows:
wherein n represents the total time period number of evaluation, P represents the output of the complementary system, and P z' (i) Representing the actual output of the complementary system during the ith evaluation period,representing the average force of the complementary system.
Photovoltaic power error prediction includes:
and analyzing the influence of the photovoltaic uncontrollable power supply access system on the system output by using an average absolute error (MAE) and a Root Mean Square Error (RMSE), wherein the mathematical expression is as follows:
wherein P' (i) represents the difference value between the predicted output and the actual output of the photovoltaic power generation and the complementary power generation in different evaluation periods;
the reservoir water level fluctuation error prediction comprises the following steps:
in order to stabilize the fluctuation of the photovoltaic power generation, the water power output is required to have the characteristic of rapid adjustment, and the rapid increase and decrease of the water power output can bring frequent change to the water level of the reservoir, and if the water level of the reservoir fluctuates too much, the safety operation of the reservoir can be risked. Therefore, the water level fluctuation of the reservoir is selected as a judging index, the water level fluctuation of the maximum water level fluctuation day in the period is selected for daily fluctuation analysis, and the influence of the output of the system on the water level change of the reservoir is further evaluated. Selecting reservoir water level fluctuation as a judging index, selecting water level fluctuation of the maximum water level fluctuation day in a period of time for daily fluctuation analysis, and further evaluating the influence of system output on water level change of the reservoir, wherein the mathematical expression is as follows:
wherein Z (i) represents the water level in the period before evaluation, Z (i-1) represents the water level in the period after evaluation, and Δt represents the evaluation period.
Step six: solving the peak shaving model and obtaining a scheduling method comprises the following steps:
solving the peak shaving model to obtain respective output conditions of the photovoltaic and the water output in each period, and adjusting the respective output of the photovoltaic and the water output in each period to carry out peak shaving.
By using the Adam algorithm, the convergence speed is high, the training time is reduced, the prediction error is reduced, the super parameters are few, the method is simple to realize, the processing memory is reduced, the coefficient gradient problem is solved, the operation of a power station can be quickly and effectively regulated, and the stable and effective operation of a power grid is ensured.
It should be appreciated that embodiments of the invention may be implemented or realized by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer readable memory. The methods may be implemented in a computer program using standard programming techniques, including a non-transitory computer readable storage medium configured with a computer program, where the storage medium so configured causes a computer to operate in a specific and predefined manner, in accordance with the methods and drawings described in the specific embodiments. Each program may be implemented in a high level procedural or object oriented programming language to communicate with a computer system. However, the program(s) can be implemented in assembly or machine language, if desired. In any case, the language may be a compiled or interpreted language. Furthermore, the program can be run on a programmed application specific integrated circuit for this purpose.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (9)
1. The water-light complementary optimal scheduling method based on Adam algorithm consideration is characterized by comprising the following steps of:
establishing a peak regulation model of water-light complementary optimization operation;
constraining the peak regulation model based on a constraint condition of hydroelectric generation;
restraining the peak regulation model based on the constraint condition of photovoltaic power generation;
optimizing the constraint conditions of hydroelectric generation and photovoltaic power generation by using an Adam algorithm;
evaluating the stability of the power station according to the parameters of the peak regulation model;
and solving the peak shaving model, and obtaining a scheduling method.
2. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 1 is characterized in that:
the method for establishing the peak shaving model of the water-light complementary optimization operation comprises the following steps:
the mathematical expression of the peak shaving model is as follows:
P max ＝max<P h，i +P s，i >i∈[1，T]
wherein P is max Representing the maximum load value, P, assumed by the system during the conditioning period h，i Representing the efficacy value, P, of the hydro-generator set h during the ith period s，j The efficacy value of the photovoltaic generator set s in the ith period is represented, and T represents the total period length of the regulating period.
3. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 2, which is characterized in that:
constraining the peak shaving model based on the constraint condition of hydroelectric generation comprises:
the constraint conditions of hydroelectric generation comprise power balance constraint, water balance constraint, start-stop duration constraint, inter-step hydraulic connection constraint, water head calculation constraint and water head loss calculation constraint of hydroelectric generation;
the power balance constraints of hydroelectric generation include:
N y，i ＝N h，i +N s，i
wherein N is y，i Representing the load value of the power system born by the virtual hydroelectric generating set y in the ith period;
the water balance constraint includes:
wherein,representing the reservoir capacity of hydropower station m at the beginning of period t,/day>Representing the storage capacity of the hydroelectric power station m at the end of the t period, m 3 Indicating the flow of the hydroelectric power plant m in the warehouse during the t-th period,/->Indicating the flow of hydropower station m out of the warehouse during period t,/->Representing the flow of a hydroelectric power plant m in storage in period t, < >>Representing the flow rate of the hydroelectric power plant m generating electricity during the period t,the flow of the water discarded by the hydroelectric power station m during the period t, deltat representing the calculated time step,/->Represents the flow rate of the hydroelectric power station m in the t period, m 3 /s，/>Representing the delivery flow of the power station m-1 immediately upstream of the hydroelectric power station m in the period t, m 3 /s；
The on-off duration constraint includes:
wherein if itIndicating that the power station m is started up from the t-th period, and at least continuously operating at I m The machine can be stopped only in a certain time period;
if it isIndicating that the plant m is shut down from the t-th period, then at least O is continued m Starting up only in a certain time period;
the inter-ladder hydraulic link constraints include:
wherein,indicating the section flow of hydropower station m in period t, +.>The kth direct upstream power station representing hydropower station m is in period T-T o，i Time period of delivery flow, T o，i Represents the number of water flow delay time periods from the upstream power station k to the power station m, phi IUm Representing a collection of directly upstream power stations of hydropower station m;
the water head calculation includes:
wherein Z is m，t Representing the dam front water level, Z, of a hydropower station m in a period t m，t Is the dam front water level of the hydropower station m in the period of t+1,for the tail water level of the hydropower station m in the period t, h m，t Represents the head of hydropower station m during period t, < >>Representing the head loss of the hydropower station m in a period t;
the head loss calculation includes:
wherein,a function representing the head loss of hydropower station m with respect to the flow out of the warehouse, +.>Represents the ex-warehouse flow of the hydropower station m in the period t, m 3 /s。
4. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 3, wherein the method comprises the following steps:
the restraining of the peak shaving model based on the constraint condition of photovoltaic power generation comprises the following steps:
the constraint condition of the photovoltaic power generation is specifically solar irradiance constraint, solar irradiance obeys Beta distribution in a certain period, probability density function of solar irradiance is expressed as follows:
wherein s represents the current irradiance sort and the maximum output power of the photovoltaicCorresponding irradiance sor max Ratio of (i.e. s=sor) t /sor max Alpha and Beta represent shape parameters of Beta distribution, r represents gamma function;
wherein pv represents abbreviation of distributed light source, μ represents average value, σ represents standard deviation;
s represents the area of a photovoltaic power station, N represents the number of the photovoltaic power stations, eta represents the surface reflectivity, solar irradiance historical data are obtained according to local meteorological data of a virtual power station, a solar irradiance distribution function is determined, photovoltaic output distribution functions and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver for optimization prediction, and the photovoltaic power generation power under the current constraint is obtained;
in the model training process, an optimization algorithm is needed to calculate and update the network parameters trained by the model, so that the network parameters approach to an optimal value as much as possible, an optimal model is obtained, and accurate prediction of the data change trend is realized.
5. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 4, which is characterized in that:
optimizing the constraint condition of the hydroelectric generation and the constraint condition of the photovoltaic generation by using an Adam algorithm comprises the following steps:
according to the local meteorological data of the virtual power plant, solar irradiance historical data are obtained, a solar irradiance distribution function is determined, photovoltaic output distribution function and photovoltaic power generation power are calculated, the photovoltaic power generation power obtained in all scenes is used as data to be input into an Adam solver, and optimization prediction is carried out on the data to obtain photovoltaic power generation power under the current constraint.
6. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 5, which is characterized in that:
and correcting the paranoid of the constraint condition by using an Adam algorithm, wherein the mathematical expression is as follows:
m t ＝β 1 m t-1 +(1-β 1 )g t
wherein m is t Representing first order motion terms, V t Represent the second order motion term, g t Representing beta 1 And beta 2 For the power values, default values are 0.9 and 0.999, m t And V t The deviation correction of (a) is respectively as followsAnd->The mathematical expression is as follows:
at time t+1, i.e. the parameter θ of the model at time t+1st iteration t+1 The mathematical expression of (2) is:
wherein θ t The parameter representing the t time, namely the t time iteration model, eta represents the super parameter and defaults to 10 -5 Epsilon is a small number (10 -8 ) The t th iteration function is related to theta t The gradient calculation formula of (2) is:
for the loss function L Huber And (3) performing calculation, wherein the mathematical expression is as follows:
wherein Y represents a predicted value, f (x) represents a true value, when L Huber The iteration times and training times when the model reaches the optimum can be obtained in the minimum time;
the parameter is updated by estimating self-adaptive adjustment learning rate through the first-order and second-order motion terms of the gradient, and after the threshold is reached, the optimal solution is considered to be close to, and the updated parameter theta is returned t 。
7. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 6, wherein the method comprises the following steps:
the evaluation of the stability of the power station according to the parameters of the peak shaving model comprises the following steps:
the effects on the complementary system include effects on the power system and reservoir effects;
the influence on the power system takes the output difference coefficient of the complementary system and the photovoltaic output prediction error as judging indexes;
the influence on the reservoir is judged by using the change of the reservoir water level as an index.
8. The method for optimizing and scheduling the water-light complementation based on Adam algorithm consideration according to claim 7, wherein the method comprises the following steps:
the output difference coefficient of the complementary system is used as an index for judging the stability of the photovoltaic output and the water output, and the mathematical expression is as follows:
wherein n represents the total time period number of evaluation, P represents the output of the complementary system, and P z′ (i) Representing the actual output of the complementary system during the ith evaluation period,representing the average force output of the complementary system;
photovoltaic power error prediction includes:
and analyzing the influence of the photovoltaic uncontrollable power supply access system on the system output by using an average absolute error (MAE) and a Root Mean Square Error (RMSE), wherein the mathematical expression is as follows:
wherein P' (i) represents the difference value between the predicted output and the actual output of the photovoltaic power generation and the complementary power generation in different evaluation periods;
the reservoir water level fluctuation error prediction comprises the following steps:
selecting reservoir water level fluctuation as a judging index, selecting water level fluctuation of the maximum water level fluctuation day in a period of time for daily fluctuation analysis, and further evaluating the influence of system output on water level change of the reservoir, wherein the mathematical expression is as follows:
wherein W (i) represents the water level in the period before evaluation, W (i-1) represents the water level in the period after evaluation, and Δt represents the evaluation period.
9. The method for optimizing scheduling of the water-light complementation based on Adam algorithm consideration according to claim 8, wherein the method comprises the following steps: solving the peak shaving model and obtaining a scheduling method comprises the following steps:
solving the peak shaving model to obtain respective output conditions of the photovoltaic and the water output in each period, and adjusting the respective output of the photovoltaic and the water output in each period to carry out peak shaving.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311614599.3A CN117638939A (en) | 2023-11-29 | 2023-11-29 | Water-light complementary optimization scheduling method based on Adam algorithm consideration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311614599.3A CN117638939A (en) | 2023-11-29 | 2023-11-29 | Water-light complementary optimization scheduling method based on Adam algorithm consideration |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117638939A true CN117638939A (en) | 2024-03-01 |
Family
ID=90022905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311614599.3A Pending CN117638939A (en) | 2023-11-29 | 2023-11-29 | Water-light complementary optimization scheduling method based on Adam algorithm consideration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117638939A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117933677A (en) * | 2024-03-25 | 2024-04-26 | 南方电网调峰调频(广东)储能科技有限公司 | Regional frequency modulation-based associated data prediction method and device for hydropower plant |
-
2023
- 2023-11-29 CN CN202311614599.3A patent/CN117638939A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117933677A (en) * | 2024-03-25 | 2024-04-26 | 南方电网调峰调频(广东)储能科技有限公司 | Regional frequency modulation-based associated data prediction method and device for hydropower plant |
| CN117933677B (en) * | 2024-03-25 | 2024-05-24 | 南方电网调峰调频(广东)储能科技有限公司 | Regional frequency modulation-based associated data prediction method and device for hydropower plant |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105846461B (en) | Control method and system for large-scale energy storage power station self-adaptive dynamic planning | |
| CN112583017B (en) | Hybrid micro-grid energy distribution method and system considering energy storage operation constraint | |
| CN113393054B (en) | Optimal scheduling method and optimal scheduling system for wind-storage combined system | |
| CN112103994B (en) | Layered coordination control method and device for wind-hydrogen coupling system based on MPC | |
| CN109103929B (en) | Power distribution network economic optimization scheduling method based on improved dynamic kriging model | |
| CN106058917A (en) | Dynamic economic dispatching method for power system considering photovoltaic power generation | |
| CN104299173B (en) | It is a kind of to optimize dispatching method a few days ago suitable for the robust that various energy resources are accessed | |
| CN110912207B (en) | Hydropower station in-plant economic operation method and system considering multi-branch-plant constraint | |
| CN110957717A (en) | Multi-target day-ahead optimal scheduling method for multi-power-supply power system | |
| CN117638939A (en) | Water-light complementary optimization scheduling method based on Adam algorithm consideration | |
| CN115940292B (en) | Optimization scheduling method and system for wind-containing power storage system based on circle search algorithm | |
| CN113659620A (en) | Water-wind complementary power generation system day-ahead scheduling method based on dynamic frequency constraint | |
| CN111817296A (en) | Electric energy scheduling method and system for micro-grid | |
| CN115795992A (en) | Park energy Internet online scheduling method based on virtual deduction of operation situation | |
| CN113363976A (en) | Scene graph-based mid-term optimized scheduling method for wind, light and water complementary power generation system | |
| Hashmi et al. | Power energy management for a grid-connected PV system using rule-base fuzzy logic | |
| CN117013522A (en) | Comprehensive energy system scheduling optimization method considering distributed power supply and gas-electricity cooperation | |
| CN116865358A (en) | Multi-time long-scale power system wind power waste and load fluctuation tracking method and equipment | |
| CN110943452B (en) | Method for optimizing and scheduling power system | |
| CN115146936A (en) | Dynamic adjustment learning factor algorithm for solving cascade water-light storage complementary scheduling model | |
| CN114204552A (en) | Comprehensive energy scheduling method for intelligent park | |
| CN114676921A (en) | Method for calculating wind power receptibility of system by considering source load storage coordination optimization | |
| Nassourou et al. | Robust optimization based energy dispatch in smart grids considering demand uncertainty | |
| Oshnoei et al. | Optimal model predictive control of photovoltaic plants for frequency regulation in an interconnected power system | |
| Zhang et al. | Optimal economic dispatch for wind farm considering uncertainty of wind power |
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 |