CN115660226A - Power load prediction model construction method and construction device based on digital twins - Google Patents
Power load prediction model construction method and construction device based on digital twins Download PDFInfo
- Publication number
- CN115660226A CN115660226A CN202211593076.0A CN202211593076A CN115660226A CN 115660226 A CN115660226 A CN 115660226A CN 202211593076 A CN202211593076 A CN 202211593076A CN 115660226 A CN115660226 A CN 115660226A
- Authority
- CN
- China
- Prior art keywords
- model
- coupling
- neural network
- historical
- value
- 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.)
- Granted
Links
- 238000010276 construction Methods 0.000 title claims abstract description 10
- 230000008878 coupling Effects 0.000 claims abstract description 125
- 238000010168 coupling process Methods 0.000 claims abstract description 125
- 238000005859 coupling reaction Methods 0.000 claims abstract description 125
- 238000003062 neural network model Methods 0.000 claims abstract description 78
- 238000000034 method Methods 0.000 claims abstract description 61
- 238000010606 normalization Methods 0.000 claims abstract description 41
- 238000012545 processing Methods 0.000 claims abstract description 32
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000011161 development Methods 0.000 claims description 31
- 238000012417 linear regression Methods 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 6
- 230000007613 environmental effect Effects 0.000 claims description 5
- 238000004880 explosion Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 46
- 230000008569 process Effects 0.000 description 14
- 241000272168 Laridae Species 0.000 description 12
- 230000015654 memory Effects 0.000 description 11
- 238000004891 communication Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 230000033001 locomotion Effects 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 241000272161 Charadriiformes Species 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
Provided herein are a method of constructing a prediction model of an electric power load and a construction apparatus based on a digital twin, wherein the method includes: acquiring historical influence factors of a target area; carrying out normalization processing on historical influence factors; substituting the processed historical influence factors into a preset coupling model to obtain coupling data; inputting the coupling data into a neural network model to obtain a predicted value of the neural network model; establishing a target function and a constraint condition of the target function according to the predicted value and the corresponding power load actual value; optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function; respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model; and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias. The future power load condition of the power system can be predicted by the aid of the built prediction model.
Description
Technical Field
The invention relates to the field of power systems, in particular to a prediction model construction method of a power load and a construction device based on digital twins.
Background
In the prior art, only visual monitoring on the current operation state of the power system can be realized, and the power load cannot be predicted, so that the subsequent development trend cannot be deduced accurately in time according to the current state of the power system, and early warning is carried out on the condition that overload possibly occurs, which brings great difficulty to the operation management and power scheduling control decision of the power system.
Therefore, a method for constructing a prediction model of a power load is needed, which can predict the future power load condition of a power system through the prediction model, and further early warn the condition that the power system is likely to overload in the future.
Disclosure of Invention
The embodiment of the invention aims to provide a method for constructing a prediction model of a power load and a construction device based on a digital twin, so as to predict the future power load condition of a power system through the prediction model and further early warn the condition that overload possibly occurs in the follow-up process of the power system.
To achieve the above object, in one aspect, an embodiment herein provides a method for building a prediction model of a power load, including:
acquiring historical influence factors of a target area;
carrying out normalization processing on the historical influence factors;
substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
inputting the coupling data into a neural network model to obtain a predicted value output by the neural network model;
establishing an objective function and a constraint condition of the objective function according to the predicted value and the corresponding actual value of the power load;
optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function;
respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model;
and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
Preferably, the historical influencing factors include: historical power loads, historical meteorological data, historical electrical data, historical event conditions, and regional development indices.
Preferably, the method for determining the preset coupling model includes:
establishing a thermal index coupling model according to historical meteorological data;
establishing a meteorological coupling model according to historical meteorological data and the thermal index coupling model;
establishing an electrical coupling model according to historical electrical data and a regional development index;
and establishing an event coupling model according to the historical event conditions.
Preferably, the establishing the thermal index coupling model according to the historical meteorological data further comprises:
the thermal index coupling model is expressed by the following formula:
wherein ,is the heat index; t is the environmental temperature after normalization treatment; r is the relative humidity after normalization treatment;all are thermal index coefficients.
Preferably, the establishing a meteorological coupling model according to historical meteorological data and the thermal index coupling model further comprises:
the meteorological coupling model is expressed by the following formula:
wherein ,in order to be the value of the meteorological coupling,is the thermal index, S is the wind speed after normalization treatment,is the coefficient corresponding to a non-negative thermal index,is the coefficient corresponding to a negative thermal index,is a coefficient corresponding to the wind speed,is the interference value.
Preferably, the establishing an electrical coupling model according to historical electrical data and a regional development index further comprises:
the electrical coupling model is expressed by the following formula:
wherein D (x) is an electrical coupling value, D is an area development index after normalization processing, I is a current value after normalization processing, E is a power value after normalization processing, f is a frequency value after normalization processing,for regional development index mappingThe coefficient of (a) is determined,is a coefficient corresponding to the current value,is a coefficient corresponding to the power value and,the coefficient is corresponding to the frequency value.
Preferably, the establishing an event coupling model according to the historical event situation further comprises:
the event coupling model is expressed by the following formula:
where s (x) event coupling values.
Preferably, the inputting the coupling data into the neural network model to obtain the predicted value output by the neural network model further includes:
inputting the coupling data into a first neural network model to obtain a first output value;
inputting the first output value into a second neural network model to obtain a second output value so as to eliminate gradient explosion;
and inputting the second output value to a linear regression layer to obtain a predicted value.
Preferably, the establishing an objective function according to the predicted value and the corresponding actual value and the constraint condition of the objective function further include:
the objective function and the constraint condition of the objective function are expressed by the following formula:
wherein f (y) is an objective function,is a predicted value at the ith moment,is the actual value of the ith moment, n is the number of moments in a period of time, s.t. is a constraint condition, m is a natural number which is more than 0.5 and less than 1,for the weights in the first neural network model,are the weights in the second neural network model, b is the bias in the first neural network model,is the bias of the linear regression layer.
In another aspect, embodiments herein provide an apparatus for constructing a digital twin-based power load prediction model, the apparatus including:
the acquisition module is used for acquiring historical influence factors of the target area;
the cloud control analysis module is used for carrying out normalization processing on the historical influence factors; substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
the virtual load prediction module is used for inputting the coupling data into a neural network model to obtain a predicted value output by the neural network model;
the load cycle deduction module is used for establishing a target function and a constraint condition of the target function according to the predicted value and the corresponding actual power load value; optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function; respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model; and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
According to the technical scheme provided by the embodiment, through the method provided by the embodiment, the historical influence factors of the target area are comprehensively considered, all the historical influence factors are coupled through the coupling model, the coupling data are input into the neural network model to obtain the predicted value, the target function is further optimized through the difference between the predicted value and the actual value, and the optimized neural network model is finally obtained. According to the embodiment of the invention, all historical influence factors which may influence the power load are taken into consideration comprehensively, and the internal influence factors and the relevance among the influence factors are taken into consideration to couple the influence factors, so that the optimized neural network model has higher accuracy, is more suitable for the actual situation of a target area, can predict the future power load condition of the power system, and further early warns the follow-up overload condition of the power system.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments or technical solutions in the prior art, the drawings used in the embodiments or technical solutions in the prior art are briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart illustrating a method for building a predictive model of an electrical load according to an embodiment of the disclosure;
fig. 2 is a flowchart illustrating a method for determining a preset coupling model provided in an embodiment of the present disclosure;
FIG. 3 is a schematic flow chart illustrating inputting of coupling data into a neural network model to obtain a predicted value of an output of the neural network model according to an embodiment of the present disclosure;
FIG. 4 is a schematic diagram illustrating a block structure of a digital twin-based power load prediction model construction device according to an embodiment of the present disclosure;
fig. 5 shows a schematic structural diagram of a computer device provided in an embodiment herein.
Description of the figures the symbols:
100. a sampling module is used;
200. a cloud control analysis module;
300. a virtual load prediction module;
400. a load cycle deduction module;
502. a computer device;
504. a processor;
506. a memory;
508. a drive mechanism;
510. an input/output module;
512. an input device;
514. an output device;
516. a presentation device;
518. a graphical user interface;
520. a network interface;
522. a communication link;
524. a communication bus.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below clearly and completely with reference to the drawings in the embodiments of the present invention, and it is obvious that the embodiments described are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments herein without making any creative effort, shall fall within the scope of protection.
In the prior art, only visual monitoring on the current operation state of the power system can be realized, and the power load cannot be predicted, so that the subsequent development trend cannot be deduced accurately in time according to the current state of the power system, and early warning is carried out on the condition that overload possibly occurs, which brings great difficulty to the operation management and power scheduling control decision of the power system.
In order to solve the above problem, embodiments herein provide a method for building a prediction model of a power load. Fig. 1 is a flow chart of a method for building a predictive model of an electrical load provided in an embodiment herein, and the present specification provides the method operation steps as described in the embodiment or the flow chart, but may include more or less operation steps based on conventional or non-inventive labor. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of sequences, and does not represent a unique order of performance. When an actual system or apparatus product executes, it can execute sequentially or in parallel according to the method shown in the embodiment or the figures.
It should be noted that the terms "first," "second," and the like in the description and claims herein and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments herein described are capable of operation in sequences other than those illustrated or described herein. Moreover, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or device that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or device.
Referring to fig. 1, there is provided a method for constructing a prediction model of a power load, including:
s101: acquiring historical influence factors of a target area;
s102: carrying out normalization processing on the historical influence factors;
s103: substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
s104: inputting the coupling data into a neural network model to obtain a predicted value output by the neural network model;
s105: establishing an objective function and a constraint condition of the objective function according to the predicted value and the corresponding actual value of the power load;
s106: optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function;
s107: respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model;
s108: and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
The historical influence factors may be influence factors of the past year, influence factors of the past month, and the like, and may include: historical power loads, historical meteorological data, historical electrical data, historical event conditions, and regional development indices. Specifically, the power load refers to the total power consumed by all the electric devices in the target area; meteorological data refers to ambient temperature, relative humidity, and wind speed within a target region; the electrical data refers to the total current, total power and total frequency of all electrical equipment in the target area; the historical event situation refers to whether an event occurs in a target area, generally, the smooth handling of the event needs the guarantee of an electric power system, so if the event occurs in the target area, extra attention needs to be paid to provide long-term effective electric power guarantee; the regional development index refers to the degree of development of a target region, and the more developed the target region is, the higher the requirement on an electric power system is, so that more stable electric power guarantee needs to be provided for the region with higher degree of development.
Further, the regional development index reflects the development degree of the target region, and may be determined according to an annual GDP ranking, for example, the regional development index of the target region is determined according to the ranking of provinces/prefectures of the target region in a national GDP ranking, for example, the national GDP ranking is sequentially divided into 3 levels from top to bottom, if the national GDP ranking is at level 1, the regional development index is 3, if the national GDP ranking is at level 2, the regional development index is 2, and if the national GDP ranking is at level 3, the regional development index is 1.
In acquiring the historical influence factors, taking one day as a unit, taking the historical power load of the past 10 months as an example, the total power consumed by all the electric equipment in the target area in 31 days of 10 months needs to be acquired. Because the magnitude of the acquired historical influence factors is large, the historical influence factors can be normalized so as to facilitate subsequent calculation. The historical power load, the historical meteorological data, the historical electrical data, the historical event situation and the regional development index can be respectively normalized in the normalization processing.
The preset coupling models comprise a thermal index coupling model, a meteorological coupling model, an electrical coupling model and an event coupling model, the coupling models are models corresponding to historical influence factors, and both historical meteorological data and historical electrical data are composed of a plurality of different data, so that the plurality of different data need to be coupled into one coupling data through the coupling model, and the coupling model can also couple two kinds of data, such as historical electrical data and regional development indexes, into one coupling data, so that the prediction of a subsequent neural network model is facilitated.
Specifically, referring to fig. 2, the method for determining the preset coupling model includes:
s201: establishing a thermal index coupling model according to historical meteorological data;
s202: establishing a meteorological coupling model according to historical meteorological data and the thermal index coupling model;
s203: establishing an electrical coupling model according to historical electrical data and a regional development index;
s204: and establishing an event coupling model according to the historical event conditions.
The coupling data is input into the neural network model, it should be noted that since the historical power load itself is one data, the coupling is not required, and when the neural network model is input, the historical power load also needs to be input into the neural network model, that is, the data input into the neural network model includes each coupling data obtained by each coupling model and the historical power load.
The neural network model may output a predicted value, which is a predicted power load, from the input data, and after an objective function is established from the predicted value and an actual value of the power load, the prediction effect may be evaluated by the objective function.
In order to realize better prediction of the power load, the objective function can be optimized through an optimization algorithm, and the weight and the bias corresponding to the optimized optimal solution obtained through optimization should be used as the optimal weight and the optimal bias of the neural network model, that is, after the optimal weight and the optimal bias are set, the predicted value of the neural network model obtained through prediction is the closest to the actual value, so that the optimized neural network model can be obtained.
By the method of the embodiment, historical influence factors of the target area are comprehensively considered, all the historical influence factors are coupled through the coupling model, the coupling data are input into the neural network model to obtain a predicted value, the target function is further optimized through the difference between the predicted value and the actual value, and the optimized neural network model is finally obtained. According to the embodiment of the invention, all historical influence factors which may influence the power load are thoroughly and comprehensively considered, the internal influence factors and the relevance among the influence factors are considered, and the influence factors are coupled, so that the optimized neural network model has higher accuracy, is more suitable for the actual situation of a target area, can predict the future power load situation of the power system, and further early warns the follow-up overload situation of the power system.
In this embodiment, the history influence factor may be normalized by the following formula:
wherein ,is the historical influence factor after the normalization process, x is the historical influence factor before the normalization process,is the minimum of the historical impact factors prior to normalization processing,is the maximum value among the historical influencing factors before the normalization process.
Since the historical influence factors comprise historical power loads, historical meteorological data, historical electrical data, historical event conditions and regional development indexes, the normalization processing process is only described by taking the historical power loads as examples, the historical power loads of 31 days are obtained by taking one day as a unit when the historical influence factors are obtained, and the formula is used when the historical power loads of a certain day are normalized, wherein the formula is used in the process of obtaining the historical influence factorsIs a power load value corresponding to a day having the smallest historical power load among 31 days, whereinAnd calculating the historical influence factor after normalization processing in a certain day by the power load value corresponding to the day with the maximum historical power load in 31 days.
In an embodiment herein, the building a thermal index coupling model from historical meteorological data further comprises:
the thermal index coupling model is expressed by the following formula:
wherein ,is the heat index; t is the environmental temperature after normalization treatment; r is relative humidity after normalization treatment;all are thermal index coefficients.
the environmental temperature and the relative humidity are two values which can affect each other, so that the environmental temperature and the relative humidity are coupled through the thermal index coupling model to obtain coupling data after the two values are coupled, and the coupling data can affect the prediction of the power load. The thermal index coefficient is obtained through historical data fitting, and the thermal index coefficient is applied to a thermal index coupling model and can better reflect the relation between the load and the climate.
In embodiments herein, said building a meteorological coupling model from historical meteorological data and said thermal index coupling model further comprises:
the meteorological coupling model is expressed by the following formula:
wherein ,is a value of the weather coupling value,is the thermal index, S is the wind speed after normalization treatment,is the coefficient corresponding to a non-negative thermal index,is the coefficient corresponding to a negative thermal index,is a coefficient corresponding to the wind speed,is the interference value.
Wherein the coefficients、Andand interference valueThe specific value can be determined according to the actual working condition, and is not described in detail herein.
The thermal index is obtained by coupling the ambient temperature and the relative humidity, and the ambient temperature, the relative humidity and the wind speed can also generate three values which influence each other, so that the ambient temperature, the relative humidity and the wind speed are coupled through a meteorological coupling model to obtain a meteorological coupling value after the three values are coupled, and the meteorological coupling value can influence the prediction of the power load. The climate factors are comprehensively considered, so that the relationship between the load and the climate can be better reflected.
In an embodiment herein, the modeling the electrical coupling according to the historical electrical data and the regional development index further comprises:
the electrical coupling model is expressed by the following formula:
wherein D (x) is an electrical coupling value, D is an area development index after normalization processing, I is a current value after normalization processing, E is a power value after normalization processing, f is a frequency value after normalization processing,for the coefficient corresponding to the regional development index,is a coefficient corresponding to the current value,the coefficients corresponding to the power values are,the coefficients correspond to the frequency values.
Wherein the coefficients、、Andthe specific value can be determined according to the actual working condition, and is not described in detail herein.
Generally, the area development index, the current value, the power value and the frequency value are four values which affect each other, for example, the higher the area development index is, the larger the total current value, the total power value and the total frequency value of the area are, so that the electric coupling value obtained by coupling the area development index, the current value, the power value and the frequency value through the electric coupling model is obtained, and the electric coupling value can affect the prediction of the power load. The relation between the load and the electrical data can be better reflected by comprehensively considering the electrical factors.
In this embodiment, the establishing an event coupling model according to the historical event situation further comprises:
the event coupling model is expressed by the following formula:
where s (x) event coupling values.
In the event, the presence or absence of the event affects the prediction of the power load, but since the event itself is not continuous and is sudden and accidental, the event can be considered alone to obtain the event coupling value.
In this embodiment, referring to fig. 3, the inputting the coupling data into the neural network model and obtaining the predicted value output by the neural network model further includes:
s301: inputting the coupling data into a first neural network model to obtain a first output value;
s302: inputting the first output value into a second neural network model to obtain a second output value so as to eliminate gradient explosion;
s303: and inputting the second output value to a linear regression layer to obtain a predicted value.
According to the above, the input of the first neural network model includes the historical power load in addition to the three coupling values, the three coupling values and the historical power load are combined into the four-dimensional vector, and the four-dimensional vector is input to the first neural network model to obtain the first output value.
The first neural network model can be an LSTM neural network model, and the four-dimensional vector is input and then sequentially passes through a forgetting gate, an input gate and an output gate.
The forgetting gate related calculation formula is specifically as follows:
wherein ,is the input information at time t;the state of the hidden layer at the moment t-1;activating a function for sigmoid;weight for forgetting gate;a bias for a forgetting gate;indicating a forget gate output at time t.
The input gate-related calculation formula is specifically:
wherein ,is the output of the input gate;is the bias of the input gate;the weight of the input gate.
The output gate related calculation formula is specifically as follows:
wherein ,is the output of the output gate;is the weight of the output gate;is the biasing of the output gate.
After passing through the forgetting gate, the input gate and the output gate in sequence, the state of the hidden layer at the next moment needs to be updated, and the specific formula is as follows:
wherein ,updating the state value of the neuron at the moment t before updating;is the state value of the neuron at the time t;is the state value of the neuron at the time t-1;representing multiplication of elements of a matrix;weights for neuron states;is a bias of a neuronal state;the state of the hidden layer at the moment t; tan h is the hyperbolic tangent activation function.
Wherein the second neural network model may be a GRU neural network modelThe input passes through the update gate, the reset gate and the gate control and commutation unit once.
The calculation formula related to the updated gate is specifically as follows:
The calculation formula related to the reset gate is specifically as follows:
The gate control circuit switching unit is used for calculating an implicit state value at the moment t, and a related calculation formula is as follows:
wherein ,is hidden at time tThe state value is contained in the data of the state,for the implicit state value at time t-1,for a candidate hidden state, the candidate hidden state is calculated by the following formula:
In order to realize regression prediction, a linear regression layer is added, a second output value is input into the linear regression layer, and a formula for obtaining a predicted value is as follows:
wherein ,、weights and offsets of the linear regression layers are respectively; and y is a predicted value.
After obtaining the predicted value, establishing an objective function according to the predicted value and the corresponding actual value, and further including:
the objective function and the constraint condition of the objective function are expressed by the following formula:
wherein f (y) is an objective function,is a predicted value at the ith moment,is the actual value of the ith moment, n is the number of moments in a period of time, s.t. is a constraint condition, m is a natural number which is more than 0.5 and less than 1,for the weights in the first neural network model,are the weights in the second neural network model, b is the bias in the first neural network model,is the bias of the linear regression layer.
wherein As a weight includes;Is weighted to include(ii) a b is an offset comprisingWhen the values in the constraint conditions are different, the corresponding predicted values are also different, andthe difference between the measured value and the actual value is different, and the solution of the objective function is different, so that the predicted value is closer to the actual value, the objective function can be optimized through an optimization algorithm, and the optimal solution of the objective function is obtained.
And when the objective function takes the optimal solution, the weight and the bias in the corresponding constraint condition are the optimal weight and the optimal bias of the neural network model, so that the optimal neural network model is obtained, and the optimal neural network model comprises an optimal first neural network model and an optimal second neural network model.
The optimization algorithm can be a gull optimization algorithm, and the gull optimization algorithm comprises the following specific steps:
step 1: and calculating the optimal solution and the optimal gull position of the objective function according to the objective function.
And 2, step: introducing an additional variable A to update the individual positions of the gulls to avoid colliding other seabirds:
wherein ,is the current position of the individual and is,representing a new position without conflict with other seagull positions; a represents the movement behavior of the seagull in a designated space; k is the current iteration number;to control the factor, set it to 2; m is the maximum number of iterations,as to the relative orientation of the optimal individual,representing the location of the optimal individual; wherein rd is [0,1]One in the otherObeying uniformly distributed random numbers.
And step 3: determining the relative distance to the optimal individual:
after the seagull successfully judges the relative direction with the optimal individual, the relative distance is determined:
in the formula ,representing the relative distance between the gull individual and the optimal individual;
and 4, step 4: in the attack behavior, the individual gull continuously changes the angle and speed in the air and makes spiral motion, and the position updating formula when the individual gull executes the attack behavior is as follows:
wherein u, v and w represent the motion behaviors of individual gulls in three-dimensional space when attacking prey, r represents the radius of each circle of spiral line, alpha represents the flight angle of gull when making spiral motion, h and k are constants defining the spiral shape, the base of e natural logarithm, and alpha is a random number in the range of [0,2 pi ].
And 5: performing iterative computation on the target function based on the constraint conditions and the gull optimization algorithm; in each iteration process, calculating the solution of each target function corresponding to different gull individuals by using a gull optimization algorithm, selecting the solution corresponding to each gull individual according to the solution of the target function, and updating the position of the gull individual; then, circulation is carried out, if the termination condition is met, the current solution of the objective function is output to be the optimal solution, and the program is ended; the termination condition is usually taken to terminate the algorithm when no new solution is accepted in a plurality of continuous chains or the maximum iteration number is reached; otherwise, returning to the step 1, namely, the accepted new solution is generated all the time, the result is converged to the optimal solution increasingly, the optimal solution is the minimum value of f (y), and the value of the constraint condition at the moment is the optimal weight and the optimal bias of the LSTM-GRU network.
Based on the method for constructing the prediction model of the power load, the method for predicting the power load can be provided, wherein the method for predicting the power load collects the influence factors on the day, and inputs the influence factors on the day into the model constructed by the method for constructing the prediction model to obtain the predicted value of the power load on the next day.
Of course, after the actual value of the power load is obtained in the next day, the objective function can be optimized according to the predicted value and the actual value, and the optimal weight and the optimal bias of the neural network model are further obtained.
It should be noted that the power load prediction method described in the embodiments herein may be implemented based on a prediction device, and the prediction device may be connected to a terminal interaction platform, where the terminal interaction platform is configured to visually display the collected influence factors of the current day and the predicted power load prediction value of the next day. The device can also be connected with the power system and used for controlling the power system according to the predicted value of the power load in the next day so as to provide more stable and effective power load.
It should be noted that, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party. In addition, the technical scheme described in the embodiment of the application conforms to relevant regulations of national laws and regulations in terms of data acquisition, storage, use, processing and the like.
Based on the above method for constructing the prediction model of the power load, the embodiment herein further provides a device for constructing the prediction model of the power load. The apparatus may include systems (including distributed systems), software (applications), modules, components, servers, clients, etc. that employ the methods described herein in embodiments, in conjunction with any necessary apparatus to implement hardware. Based on the same innovative concept, the apparatus in one or more of the embodiments provided in the embodiments herein is described in the following embodiments. Since the implementation scheme of the apparatus for solving the problem is similar to that of the method, the specific apparatus implementation in the embodiment of the present disclosure may refer to the implementation of the foregoing method, and repeated details are not repeated. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Specifically, fig. 4 is a schematic block diagram of an embodiment of a digital twin-based power load prediction model building apparatus provided in an embodiment of the present disclosure, and referring to fig. 4, the digital twin-based power load prediction model building apparatus provided in an embodiment of the present disclosure includes: the system comprises a sampling module 100, a cloud control analysis module 200, a virtual load prediction module 300 and a load cycle deduction module 400.
The acquisition module 100 is used for acquiring historical influence factors of a target area;
the cloud control analysis module 200 is used for carrying out normalization processing on the historical influence factors; substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
the virtual load prediction module 300 is configured to input the coupling data to a neural network model to obtain a predicted value output by the neural network model;
a load cycle deduction module 400, configured to establish an objective function and a constraint condition of the objective function according to the predicted value and the corresponding actual value of the power load; optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function; respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model; and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
Referring to fig. 5, a computer device 502 is further provided in an embodiment of the present disclosure based on the above-described method for constructing a prediction model of an electrical load, wherein the above-described method is executed on the computer device 502. Computer device 502 may include one or more processors 504, such as one or more Central Processing Units (CPUs) or Graphics Processors (GPUs), each of which may implement one or more hardware threads. The computer device 502 may also comprise any memory 506 for storing any kind of information, such as code, settings, data, etc., and in a particular embodiment a computer program running on the memory 506 and on the processor 504, which computer program, when being executed by the processor 504, may perform the instructions according to the above-described method. For example, and without limitation, memory 506 may include any one or combination of the following: any type of RAM, any type of ROM, flash memory devices, hard disks, optical disks, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent fixed or removable components of computer device 502. In one case, when the processor 504 executes the associated instructions, which are stored in any memory or combination of memories, the computer device 502 can perform any of the operations of the associated instructions. The computer device 502 also includes one or more drive mechanisms 508, such as a hard disk drive mechanism, an optical disk drive mechanism, etc., for interacting with any memory.
Corresponding to the methods in fig. 1-3, the embodiments herein also provide a computer-readable storage medium having stored thereon a computer program, which, when executed by a processor, performs the steps of the above-described method.
Embodiments herein also provide computer readable instructions, wherein when executed by a processor, a program thereof causes the processor to perform the method as shown in fig. 1-3.
It should be understood that, in various embodiments herein, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments herein.
It should also be understood that, in the embodiments herein, the term "and/or" is only one kind of association relation describing an associated object, meaning that three kinds of relations may exist. For example, a and/or B, may represent: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter associated objects are in an "or" relationship.
Those of ordinary skill in the art will appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the components and steps of the various examples have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided herein, it should be understood that the disclosed system, apparatus, and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments herein.
In addition, functional units in the embodiments herein may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present invention may be implemented in a form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The principles and embodiments of this document are explained herein using specific examples, which are presented only to aid in understanding the methods and their core concepts; meanwhile, for the general technical personnel in the field, according to the idea of this document, there may be changes in the concrete implementation and the application scope, in summary, this description should not be understood as the limitation of this document.
Claims (10)
1. A method for constructing a prediction model of a power load is characterized by comprising the following steps:
acquiring historical influence factors of a target area;
carrying out normalization processing on the historical influence factors;
substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
inputting the coupling data into a neural network model to obtain a predicted value output by the neural network model;
establishing an objective function and a constraint condition of the objective function according to the predicted value and the corresponding actual value of the power load;
optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function;
respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model;
and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
2. The method of constructing a predictive model of an electrical load according to claim 1, wherein the historical influencing factors include: historical power loads, historical meteorological data, historical electrical data, historical event conditions, and regional development indices.
3. The method of constructing a prediction model of an electric power load according to claim 2, wherein the method of determining the preset coupling model includes:
establishing a thermal index coupling model according to historical meteorological data;
establishing a meteorological coupling model according to historical meteorological data and the thermal index coupling model;
establishing an electrical coupling model according to historical electrical data and a regional development index;
and establishing an event coupling model according to the historical event conditions.
4. The method of building a predictive model of an electrical load according to claim 3, wherein said building a thermal index coupling model from historical meteorological data further comprises:
the thermal index coupling model is expressed by the following formula:
5. The method of claim 3, wherein said building a meteorological coupling model based on historical meteorological data and said thermal index coupling model further comprises:
the meteorological coupling model is expressed by the following formula:
wherein ,in order to be the value of the meteorological coupling,is the thermal index, S is the wind speed after normalization treatment,is the coefficient corresponding to a non-negative thermal index,is the coefficient corresponding to a negative thermal index,is a coefficient corresponding to the wind speed,is the interference value.
6. The method of claim 3, wherein the modeling the electrical coupling based on historical electrical data and regional growth indices further comprises:
the electrical coupling model is expressed by the following formula:
wherein D (x) is an electrical coupling value, D is an area development index after normalization processing, I is a current value after normalization processing, E is a power value after normalization processing, f is a frequency value after normalization processing,for the coefficient corresponding to the regional development index,is a coefficient corresponding to the current value,is a coefficient corresponding to the power value and,the coefficients correspond to the frequency values.
8. The method of claim 1, wherein inputting the coupling data to a neural network model to obtain a predicted value of a neural network model output further comprises:
inputting the coupling data into a first neural network model to obtain a first output value;
inputting the first output value into a second neural network model to obtain a second output value so as to eliminate gradient explosion;
and inputting the second output value to a linear regression layer to obtain a predicted value.
9. The method of claim 8, wherein the establishing an objective function and the constraint condition of the objective function according to the predicted value and the corresponding actual value further comprises:
the objective function and the constraint condition of the objective function are expressed by the following formula:
wherein f (y) is an objective function,is a predicted value at the ith moment,is the actual value of the ith moment, n is the number of moments in a period of time, s.t. is a constraint condition, m is a natural number which is more than 0.5 and less than 1,for the weights in the first neural network model,are the weights in the second neural network model, b is the bias in the first neural network model,is the bias of the linear regression layer.
10. A digital twin-based power load prediction model construction apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring historical influence factors of the target area;
the cloud control analysis module is used for carrying out normalization processing on the historical influence factors; substituting the history influence factors after the normalization processing into a preset coupling model to obtain coupling data, wherein the coupling model is a model corresponding to the history influence factors of the target area;
the virtual load prediction module is used for inputting the coupling data into a neural network model to obtain a predicted value output by the neural network model;
the load cycle deduction module is used for establishing a target function and a constraint condition of the target function according to the predicted value and the corresponding actual value of the power load; optimizing the objective function through an optimization algorithm to obtain an optimal solution of the objective function; respectively taking the weight and the bias corresponding to the optimal solution as the optimal weight and the optimal bias of the neural network model; and determining the optimized neural network model as a prediction model of the power load according to the optimal weight and the optimal bias.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211593076.0A CN115660226B (en) | 2022-12-13 | 2022-12-13 | Power load prediction model construction method and digital twin-based construction device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211593076.0A CN115660226B (en) | 2022-12-13 | 2022-12-13 | Power load prediction model construction method and digital twin-based construction device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115660226A true CN115660226A (en) | 2023-01-31 |
CN115660226B CN115660226B (en) | 2023-04-25 |
Family
ID=85017277
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211593076.0A Active CN115660226B (en) | 2022-12-13 | 2022-12-13 | Power load prediction model construction method and digital twin-based construction device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115660226B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116702978A (en) * | 2023-06-07 | 2023-09-05 | 西安理工大学 | Electric vehicle charging load prediction method and device considering emergency characteristics |
CN116956594A (en) * | 2023-07-25 | 2023-10-27 | 广州锐兴科技有限公司 | Rural power grid optimization method, device and equipment based on topological structure |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112116153A (en) * | 2020-09-18 | 2020-12-22 | 上海电力大学 | Park multivariate load joint prediction method for coupling Copula and stacked LSTM network |
AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
US20210203159A1 (en) * | 2019-12-27 | 2021-07-01 | North China Electric Power University | Power load forecasting method in multi-energy coupling mode |
CN113610328A (en) * | 2021-08-25 | 2021-11-05 | 浙江浙能台州第二发电有限责任公司 | Power generation load prediction method |
CN113762387A (en) * | 2021-09-08 | 2021-12-07 | 东北大学 | Data center station multi-load prediction method based on hybrid model prediction |
CN115276006A (en) * | 2022-09-26 | 2022-11-01 | 江苏永鼎股份有限公司 | Load prediction method and system for power integration system |
CN115293326A (en) * | 2022-07-05 | 2022-11-04 | 深圳市国电科技通信有限公司 | Training method and device of power load prediction model and power load prediction method |
CN115293400A (en) * | 2022-06-23 | 2022-11-04 | 国网浙江省电力有限公司营销服务中心 | Power system load prediction method and system |
CN115423161A (en) * | 2022-08-19 | 2022-12-02 | 国网山东省电力公司电力科学研究院 | Digital twin-based multi-energy coupling optimization scheduling method and system |
-
2022
- 2022-12-13 CN CN202211593076.0A patent/CN115660226B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20210203159A1 (en) * | 2019-12-27 | 2021-07-01 | North China Electric Power University | Power load forecasting method in multi-energy coupling mode |
CN112116153A (en) * | 2020-09-18 | 2020-12-22 | 上海电力大学 | Park multivariate load joint prediction method for coupling Copula and stacked LSTM network |
AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
CN113610328A (en) * | 2021-08-25 | 2021-11-05 | 浙江浙能台州第二发电有限责任公司 | Power generation load prediction method |
CN113762387A (en) * | 2021-09-08 | 2021-12-07 | 东北大学 | Data center station multi-load prediction method based on hybrid model prediction |
CN115293400A (en) * | 2022-06-23 | 2022-11-04 | 国网浙江省电力有限公司营销服务中心 | Power system load prediction method and system |
CN115293326A (en) * | 2022-07-05 | 2022-11-04 | 深圳市国电科技通信有限公司 | Training method and device of power load prediction model and power load prediction method |
CN115423161A (en) * | 2022-08-19 | 2022-12-02 | 国网山东省电力公司电力科学研究院 | Digital twin-based multi-energy coupling optimization scheduling method and system |
CN115276006A (en) * | 2022-09-26 | 2022-11-01 | 江苏永鼎股份有限公司 | Load prediction method and system for power integration system |
Non-Patent Citations (2)
Title |
---|
DONGXIAO NIU 等: "Short-term multi-energy load forecasting for integrated energy systems based on CNN-BiGRU optimized by attention mechanism" * |
姚程文等: "基于CNN-GRU混合神经网络的负荷预测方法" * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116702978A (en) * | 2023-06-07 | 2023-09-05 | 西安理工大学 | Electric vehicle charging load prediction method and device considering emergency characteristics |
CN116702978B (en) * | 2023-06-07 | 2024-02-13 | 西安理工大学 | Electric vehicle charging load prediction method and device considering emergency characteristics |
CN116956594A (en) * | 2023-07-25 | 2023-10-27 | 广州锐兴科技有限公司 | Rural power grid optimization method, device and equipment based on topological structure |
CN116956594B (en) * | 2023-07-25 | 2024-02-09 | 广州锐兴科技有限公司 | Rural power grid optimization method, device and equipment based on topological structure |
Also Published As
Publication number | Publication date |
---|---|
CN115660226B (en) | 2023-04-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115660226A (en) | Power load prediction model construction method and construction device based on digital twins | |
CN112862012A (en) | Operation and maintenance system abnormity early warning method, device and equipment based on LSTM model | |
CN111428874A (en) | Wind control method, electronic device and computer readable storage medium | |
AU2020418399A1 (en) | Method and apparatus for predicting power consumption, device and readiable storage medium | |
CN112734135B (en) | Power load prediction method, intelligent terminal and computer readable storage medium | |
CN110264270A (en) | A kind of behavior prediction method, apparatus, equipment and storage medium | |
US20190385045A1 (en) | Systems And Methods For Generalized Adaptive Storage Endpoint Prediction | |
CN107977748B (en) | Multivariable distorted time sequence prediction method | |
US11481598B2 (en) | Auto scaling a distributed predictive analytics system with machine learning | |
EP3985460B1 (en) | Controller, control system and method to control a facility and to generate a control model by learning | |
CN111027672A (en) | Time sequence prediction method based on interactive multi-scale recurrent neural network | |
CN115238969A (en) | Trend prediction method and device for termite prevention, electronic equipment and medium | |
EP4009239A1 (en) | Method and apparatus with neural architecture search based on hardware performance | |
CN114648170A (en) | Reservoir water level prediction early warning method and system based on hybrid deep learning model | |
US20220318615A1 (en) | Time-aligned reconstruction recurrent neural network for multi-variate time-series | |
CN115687882A (en) | Method and device for estimating probability distribution of transmission section limit transmission capacity | |
Sarıkaya et al. | Training Sigma-Pi neural networks with the grey wolf optimization algorithm | |
CN114818484A (en) | Training method of driving environment model and prediction method of driving environment information | |
WO2021181605A1 (en) | Machine learning model determination system and machine learning model determination method | |
Kara | A deep learning framework with convolutional long short‐term memory for influenza‐like illness trend estimation | |
CN109947749B (en) | Intelligent home control behavior habit mining method with forgetting learning capability | |
Zhao et al. | Compact cat swarm optimization algorithm | |
US20220253687A1 (en) | Generic discriminative inference with generative models | |
CN114928477B (en) | Network intrusion detection method and device, readable storage medium and terminal equipment | |
CN116434973A (en) | Infectious disease early warning method, device, equipment and medium based on artificial intelligence |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |