CN114912546A - Energy data aggregation method and device - Google Patents

Abstract

The invention discloses a method and device for aggregating energy data. The method includes: acquiring renewable energy data and energy data to be measured, and establishing an individual physical model and an aggregated cluster model of renewable energy based on the renewable energy data , divide the renewable energy data into training set data and test set data, and train the individual physical model and the aggregated cluster model based on the K-means clustering algorithm, the training set data and the test set data and verification, obtain the target individual physical model and the target aggregation cluster model, input the energy data to be measured into the target individual physical model and the target aggregation cluster model, and obtain the predicted aggregation index data of the energy data to be measured . It is conducive to solving the technical problem that the social idle distributed renewable resources are difficult to aggregate, and unifies the access standards of renewable resources.

Description

A kind of aggregation method and device of energy data

technical field

The present invention relates to the technical field of distributed energy in power systems, and in particular, to a method and device for aggregating energy data.

Background technique

With the increasing demand for energy in society, the proposal of the "Dual-Carbon Goal" has promoted the transformation of the energy structure, and distributed renewable energy has also ushered in massive growth. Various types of heterogeneous and massive distributed renewable resources, including distributed photovoltaics, distributed wind turbines, distributed energy storage, electric vehicles and flexible loads, provide the grid with economical and clean energy, and can give full play to different regions according to local conditions. The unique advantages of wind and solar resources can also ensure the safe and economic operation of the power grid through peak shaving, frequency regulation, and demand response. At present, massive distributed renewable energy is rapidly increasing in the proportion of energy, with a huge number and a wide variety of types. However, the geographical distribution is scattered, the subordinate rights are not unified, and its power generation characteristics are uncertain due to climate and human factors. If it is directly connected to the power grid in a disorderly manner, it will not only cause management difficulties and bring impact to the power grid, on the other hand. It will also cause waste of idle resources in the society, and cannot guarantee the full consumption of renewable resources. In view of this, combined with the characteristics of small volume, scattered distribution, and different subordinate rights of massive distributed renewable resources, some intermediate institutions can be aggregated for unified access, thereby enhancing the flexibility of the power system.

For the aggregation and access of massive distributed renewable resources, the current mainstream practice at home and abroad is mainly to access through aggregators, virtual power plants and other forms. Respond to scheduling commands as a whole. For the effective use of various distributed renewable energy, it is necessary to establish a standardized model, which not only needs to meet the external controllability of various resources to shield the underlying physical characteristics to ensure the openness attribute, but also conform to the heterogeneous characteristics of various resources themselves. .

At present, the distributed renewable energy connected to the power grid is mostly large-scale wind farms and photovoltaic units, which lack effective means of utilizing the huge amount of idle distributed resources in the society.

Therefore, in order to unify the access standards of renewable resources and solve the current technical problem that the social idle distributed renewable resources are difficult to aggregate, it is urgent to build an energy data aggregation method.

SUMMARY OF THE INVENTION

The present invention provides a method and device for aggregating energy data, which solves the technical problem that the existing social idle distributed renewable resources are difficult to aggregate.

In a first aspect, the present invention provides a method for aggregating energy data, including:

Obtain renewable energy data and energy data to be measured;

Based on the renewable energy data, establish an individual physical model and an aggregated cluster model of renewable energy;

dividing the renewable energy data into training set data and test set data;

Based on the K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are trained and verified to obtain the target individual physical model and the target aggregated cluster model;

Inputting the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated index data of the energy data to be measured.

Optionally, obtain renewable energy data and energy data to be measured, including:

Obtain initial data on renewable energy;

The initial data of renewable energy is cleaned and repaired to obtain the renewable energy data and the energy data to be measured.

Optionally, based on the K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are trained and verified to obtain the target individual physical model and the target aggregated cluster model. ,include:

Using the K-means clustering algorithm, in combination with the training set data, the individual physical model and the aggregated cluster model are trained to obtain the aggregated cluster model after training and the aggregated cluster model after training;

Based on the test set data, the trained aggregate cluster model and the trained aggregate cluster model are verified to obtain the target individual physical model and the target aggregate cluster model.

Optionally, using the K-means clustering algorithm, in combination with the training set data, the individual physical model and the aggregated cluster model are trained to obtain a trained aggregated cluster model and a trained aggregated cluster model, include:

Inputting the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data;

Determine the training error according to the data label corresponding to the training set data and the predicted aggregate value;

Based on the training error, through the K-means clustering algorithm, the individual physical model and the aggregated cluster model are adjusted to obtain optimal parameters, and the optimal parameters are used to optimize the individual physical model and the aggregated cluster model. For the aggregated cluster model, the trained aggregated cluster model and the trained aggregated cluster model are obtained.

Optionally, before inputting the training set data into the individual physical model and the aggregated cluster model to obtain the predicted aggregated value of the corresponding renewable energy data, the method further includes:

The parameters of the individual physical model and the parameters of the aggregated cluster model are initialized.

In a second aspect, the present invention provides a device for aggregating energy data, including:

The acquisition module is used to acquire renewable energy data and energy data to be measured;

establishing a module for establishing an individual physical model and an aggregated cluster model of renewable energy based on the renewable energy data;

a dividing module, configured to divide the renewable energy data into training set data and test set data;

A training module for training and verifying the individual physical model and the aggregated cluster model based on the K-means clustering algorithm, the training set data and the test set data, to obtain a target individual physical model and a target aggregated cluster Model;

The indicator module is used for inputting the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated indicator data of the energy data to be measured.

Optionally, the obtaining module includes:

Obtain sub-module for obtaining initial data of renewable energy;

The repairing submodule is used for cleaning and repairing the initial data of renewable energy to obtain the renewable energy data and the energy data to be measured.

Optionally, the training module includes:

A training sub-module for using the K-means clustering algorithm, in combination with the training set data, to train the individual physical model and the aggregated cluster model to obtain a trained aggregated cluster model and a trained aggregated cluster Model;

A verification submodule, configured to verify the trained aggregate cluster model and the trained aggregate cluster model based on the test set data, and obtain the target individual physical model and the target aggregate cluster model.

Optionally, the training submodule includes:

a prediction unit, configured to input the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data;

an error unit, configured to determine a training error according to the data label corresponding to the training set data and the predicted aggregate value;

An optimization unit, configured to adjust the individual physical model and the aggregated cluster model through the K-means clustering algorithm based on the training error to obtain optimal parameters, and use the optimal parameters to optimize all the parameters. The individual physical model and the aggregated cluster model are obtained to obtain the trained aggregated cluster model and the trained aggregated cluster model.

Optionally, the training submodule also includes:

A parameter unit for initializing parameters of the individual physical model and parameters of the aggregated cluster model.

It can be seen from the above technical solutions that the present invention has the following advantages: the present invention provides a method for aggregating energy data, by acquiring renewable energy data and energy data to be measured, and based on the renewable energy data, establishes a renewable energy data The individual physical model and aggregated cluster model of energy, divide the renewable energy data into training set data and test set data, based on the K-means clustering algorithm, the training set data and the test set data, the individual physical The model and the aggregation cluster model are trained and verified to obtain the target individual physical model and the target aggregation cluster model, and the energy data to be measured is input into the target individual physical model and the target aggregation cluster model, and the target individual physical model and the target aggregation cluster model are obtained. The forecast aggregation index data of the measured energy data, through an energy data aggregation method, solves the current technical problem that the social idle distributed renewable resources are difficult to aggregate, and unifies the access standards of renewable resources.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

FIG. 1 is a flowchart of a first embodiment of a method for aggregating energy data according to the present invention;

Fig. 2 is a flow chart of the second embodiment of a method for aggregating energy data according to the present invention;

FIG. 3 is a flow chart for updating a standardized model of renewable energy according to the present invention;

FIG. 4 is a structural block diagram of an embodiment of an energy data aggregation apparatus according to the present invention.

Detailed ways

Embodiments of the present invention provide a method and device for aggregating energy data, which are used to solve the technical problem that it is difficult to aggregate social idle distributed renewable resources.

In order to make the purpose, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the following The described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Embodiment 1, please refer to FIG. 1. FIG. 1 is a flow chart of Embodiment 1 of a method for aggregating energy data according to the present invention, including:

Step S101, acquiring renewable energy data and energy data to be measured;

It should be noted that the energy data includes individual photovoltaic array data, wind turbine data, distributed energy storage data, electric vehicle data, flexible load data, as well as intra-cluster characteristic characterization data and out-of-cluster characteristic characterization data.

The internal characteristic data of the cluster include capacity support capability (including reference value and upper and lower boundaries), power support capability (including reference value and upper and lower boundaries), dynamic response capability and effective convergence time ratio.

The characteristic data outside the cluster include stability factor, reliability factor, network loss factor, unit regulation power and dynamic sensitivity factor.

In the embodiment of the present invention, the initial data of renewable energy is acquired, the initial data of renewable energy is cleaned and repaired, and the data of renewable energy and the energy data to be measured are obtained.

Step S102, based on the renewable energy data, establish an individual physical model and an aggregated cluster model of renewable energy;

It should be noted that the individual physical models of renewable energy include individual photovoltaic array models, wind turbine models, distributed energy storage models, electric vehicle models, and flexible load models.

In the embodiment of the present invention, according to the renewable energy data, an individual physical model and an aggregated cluster model of renewable energy are established.

Step S103, dividing the renewable energy data into training set data and test set data;

Step S104, based on the K-means clustering algorithm, the training set data and the test set data, train and verify the individual physical model and the aggregated cluster model to obtain a target individual physical model and a target aggregated cluster model;

In the embodiment of the present invention, the K-means clustering algorithm is used, combined with the training set data, to train the individual physical model and the aggregated cluster model, to obtain a trained aggregated cluster model and a post-trained aggregated cluster model The cluster model, based on the test set data, verifies the trained aggregated cluster model and the trained aggregated cluster model to obtain the target individual physical model and the target aggregated cluster model.

Step S105: Input the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated index data of the energy data to be measured.

In the method for aggregating energy data provided by the embodiment of the present invention, by acquiring renewable energy data and energy data to be measured, and based on the renewable energy data, an individual physical model and an aggregated cluster model of renewable energy are established, The renewable energy data is divided into training set data and test set data, and based on the K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are trained and analyzed. Verify, obtain the target individual physical model and the target aggregation cluster model, input the energy data to be measured into the target individual physical model and the target aggregation cluster model, and obtain the predicted aggregation index data of the energy data to be measured, Through an energy data aggregation method, the existing technical problem that the social idle distributed renewable resources are difficult to aggregate is solved, and the access standard of renewable resources is unified.

Embodiment 2, please refer to FIG. 2. FIG. 2 is a flowchart of a method for aggregating energy data according to the present invention, including:

Step S201, obtaining initial data of renewable energy;

In the embodiment of the present invention, the initial data of renewable energy is obtained, including individual photovoltaic array data, wind turbine data, distributed energy storage data, electric vehicle data, flexible load data, as well as in-cluster characteristic characterization data and out-of-cluster characteristic characterization data data.

The internal characteristic data of the cluster include capacity support capability (including reference value and upper and lower boundaries), power support capability (including reference value and upper and lower boundaries), dynamic response capability and effective convergence time ratio.

The characteristic data outside the cluster include stability factor, reliability factor, network loss factor, unit regulation power and dynamic sensitivity factor.

Step S202, cleaning and repairing the initial data of renewable energy to obtain renewable energy data and energy data to be measured;

Step S203, based on the renewable energy data, establish an individual physical model and an aggregated cluster model of renewable energy;

In the embodiment of the present invention, based on the renewable energy data, an individual physical model and an aggregated cluster model of renewable energy are established, and the individual physical model of renewable energy includes an individual photovoltaic array model, a wind turbine model, and a distributed energy storage model. model, electric vehicle model and flexible load model.

In the specific implementation, please refer to Fig. 3, Fig. 3 is a flow chart of updating a standardized model of renewable energy according to the present invention, wherein, t is the time, p is the power generation, e is the accumulated power generation, and r is the ramp speed , D is the external characteristic factor;

Distributed renewable resources consider individual photovoltaic arrays, wind turbines, distributed energy storage, electric vehicles and flexible loads. The physical level modeling content is mainly about its electrical quantities and constraints, as follows:

描述。

分别为聚合商节点i下，第j类且编号为k的灵活资源在第t个时间段的发电功率、累积发电量和爬坡速率的上下界。The power generation characteristics of a general flexible resource at time period t can be represented by the hexatuple

describe.

Under the aggregator node i, the upper and lower bounds of the power generation, cumulative power generation and ramp rate of the j-th type of flexible resources numbered k in the t-th time period.

通过电力电子器件，能够从0到

连续调节机组出力，PV功率和累计发电量的上下界为：For the photovoltaic unit, the Maximum Power Point Tracking (MPPT) technology it is equipped with can ensure the best match between the parameters of the photovoltaic cell and the load, and the output power is always kept at the maximum power point P _m . Based on this, the predicted value of P _m in each period is regarded as the maximum output prediction as

Through power electronics, it is possible to go from 0 to

Continuously adjust the output of the unit, the upper and lower bounds of PV power and cumulative power generation are:

连续调节机组出力，风电机组功率和累计发电量的上下界为：The control characteristics of wind turbines are similar to those of photovoltaic units, which can be adjusted from 0 to

The upper and lower bounds of wind turbine power and cumulative power generation are as follows:

进行预测，因此采用风电小时平均出力变化率衡量：Compared with the generation of photovoltaic units, the output of the generator set fluctuates more and has no regularity, which is more difficult to correct.

Therefore, the average hourly output change rate of wind power is used to measure:

为风电小时出力变化率，

为风电小时出力波动量，

为风电额定装机容量。in,

is the hourly output change rate of wind power,

For the hourly output fluctuation of wind power,

It is the rated installed capacity of wind power.

The energy storage operating characteristics can be described based on the state of charge (SOC) of the energy storage device at time t:

Among them, SOC(t) is the state of charge of the energy storage at the end of the t-th period; η _c and η _d are the charging and discharging efficiencies of the energy storage, respectively; E _r is the rated capacity of the energy storage; σ is the self-discharge of the energy storage rate; P _c and P _d are the charging and discharging power of the current period, respectively.

Energy storage device operating constraints:

The first is the charge and discharge power limit of the energy storage device, and the charge and discharge of the energy storage device in a unit time does not exceed 20% of its rated capacity, that is:

In order to prevent the overcharge and discharge of the energy storage device and have sufficient margin for emergency dispatch, the minimum charge amount of the energy storage device is set as SOC _min and the maximum charge amount as SOC _max , namely:

SOC _min ≤SOC _ess (t)≤SOC _max ;

Usually take SOC _min as 0.2 and SOC _max as 0.9.

The allowable charge and discharge power per unit time of the energy storage device changes continuously with the SOC, namely:

The essence of electric vehicles is still energy storage, but as a mobile energy storage with stronger user attributes, it brings greater uncertainty, and its user characteristics reflect its social attributes. to describe the travel rules of electric vehicles to describe their social attributes;

The probability distribution of the time when the electric vehicle leaves home and when it arrives at home obeys the generalized extreme value distribution, and the probability distribution of the travel distance obeys the Weibull distribution. The Monte Carlo stochastic simulation method can be used to simulate the travel of the electric vehicle randomly. The inverse function of the time of leaving home, the time of arriving home, and the travel distance distribution function is:

Among them, x _out is the travel time of the electric vehicle, f _PEV,out is the inverse function of the probability density of the electric vehicle travel time, y _out is the travel time probability of the electric vehicle, μ _PEV,out , σ _PEV,out , ξ _PEV,out is the electric vehicle The location parameters, scale parameters and shape parameters of the probability distribution of the travel time; x _in , f _PEV,in , y _in , μ _PEV,in , σ _PEV,in , ξ _PEV,in are the corresponding electric vehicle arrival time, the probability of the arrival time Density inverse function, probability of arriving home, location parameters, scale parameters and shape parameters of the probability distribution of electric vehicles when they arrive home; x _dis , f _dis , y _dis , k _dis , λ _dis are the inverse functions of electric vehicle travel distance and travel distance probability density , the travel distance probability, the shape and scale parameters of the travel distance distribution function.

Flexibility loads are characterized in terms of interruptible, translatable:

Mathematical model for interruptible load resources:

Resource maximum invocation constraints:

Resource availability period constraints:

Resource minimum and maximum usage time constraints:

The relationship between the actual operating power of the resource and whether to start the response:

Response cost of interruptible load resource:

Among them, S _DR is the set of interruptible load resources; y _i,t is the use state of the interruptible load resource; zi _,t is the startup variable of the resource, and the above two sets of variables are both 0-1 variables; N is the resource Ψ _use is the set of available moments; T _min and T _max are the minimum and maximum response times of the resource, respectively; p _DR,i,t is the actual power of resource i; p _i,t,0 and P _{DR, i, t} are the initial power and response power of the resource, respectively; c _i is the unit power response cost of the interruptible load resource.

Mathematical model for translatable load resources:

Resource maximum invocation constraints:

Resource availability period constraints:

The relationship between the amount of shiftable load resources moved out and in:

Resources can be moved into the period constraint:

The actual operating power of the resource:

The relationship between the input variable and the response variable:

Response cost of translatable load resource:

Among them, S _TR is the set of translatable load resources; y _{i, OUT, t} is the use state of the translatable load resource i; z _{i, OUT, t} is the startup variable of the resource, y _{i, IN, t} represents whether the Period shift-in load, the above three groups of variables are all 0-1 variables; Ψ _in is the set formed by the shift-in time; p _TR,i,t is the actual power of resource i; p _i,t,0 and P _TR,i respectively are the initial power and response power of the resource.

The cluster model has a two-layered property, which aggregates down to the properties of the individual, similar to the individual model. However, the cluster layer no longer distinguishes source-storage loads, so the lower model is the unity of the three types of source-storage-load models. The cluster attributes, the physical part, of the upper layer facing scheduling or market, should be characterized by its internal characteristics, namely the power up and down capabilities and power baselines of the aggregated virtual.

表示接入集群节点i下，第j类且编号为k的灵活资源在第t个时间段的发电功率。若功率值为正，表示该资源向电网发出有功功率，若其为负值，表示该资源从电网吸收有功功率。采用

表示接入集群节点i下，第j类且编号为k的灵活资源在第1到第t个时间段的累积发电量。具体表现如下表所示：use

Indicates the generation power of the j-th type of flexible resource numbered k in the t-th time period under the access cluster node i. If the power value is positive, it means that the resource sends active power to the grid, and if it is negative, it means that the resource absorbs active power from the grid. use

Indicates the cumulative power generation of the j-th type and number k of flexible resources under the access cluster node i in the first to t-th time periods. The specific performance is shown in the following table:

The parameters of this six-tuple characterize the active power boundary of the flexible resource in the period t, the cumulative power generation boundary and the ramp rate boundary from t to t+1, namely:

分别为集群节点i下，第j类且编号为k的灵活资源在第t个时间段的发电功率、累积发电量和爬坡速率的上下界。in,

are the upper and lower bounds of the power generation, cumulative power generation, and ramp rate of the j-th type of flexible resources numbered k in the t-th time period under cluster node i, respectively.

Among them, the baseline power is the aggregate power when the cluster does not take control measures on it, which is used to confirm the control effect of the cluster and needs to be calculated in a rolling period of 5 to 30 minutes. The up-regulation and down-regulation boundaries were calculated using the individual model boundary calculation method. At the same time, considering the external characteristics of cluster aggregation evaluation and cluster social attribute formation, the final unity is:

A={C,P,D,T,Sta,Se,L,K,Sen,…}

Sta=f ₁ (C,D,P,T)

Se=f ₂ (C,D,P,T)

L=f ₃ (C,D,P,T,Site(Z))

K=f ₄ (C, D, P, T, Site(Z));

Among them, C, P, D, and T are the internal characteristics of the cluster, which are capacity support capability (including reference value and upper and lower boundaries), power support capability (including reference value and upper and lower boundaries), dynamic response capability, and effective convergence time ratio. . St, Se, L, K, and Sen represent the characteristics outside the cluster, which are stability factor, reliability factor, network loss factor, unit adjustment power, and dynamic sensitivity factor.

Step S204, dividing the renewable energy data into training set data and test set data;

Step S205, using the K-means clustering algorithm, in combination with the training set data, to train the individual physical model and the aggregated cluster model to obtain a trained aggregated cluster model and a trained aggregated cluster model;

In an optional embodiment, the K-means clustering algorithm is used to train the individual physical model and the aggregated cluster model in combination with the training set data, to obtain a trained aggregated cluster model and a trained aggregated cluster model. Aggregate cluster model, including:

initializing the parameters of the individual physical model and the parameters of the aggregated cluster model;

Inputting the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data;

Determine the training error according to the data label corresponding to the training set data and the predicted aggregate value;

Based on the training error, through the K-means clustering algorithm, the individual physical model and the aggregated cluster model are adjusted to obtain optimal parameters, and the optimal parameters are used to optimize the individual physical model and the aggregated cluster model. For the aggregated cluster model, the trained aggregated cluster model and the trained aggregated cluster model are obtained.

In the embodiment of the present invention, the parameters of the individual physical model and the parameters of the aggregated cluster model are initialized, and the training set data is input into the individual physical model and the aggregated cluster model to obtain corresponding renewable energy data According to the data label corresponding to the training set data and the predicted aggregate value, determine the training error, and based on the training error, through the K-means clustering algorithm, the individual physical model and the The aggregated cluster model is adjusted to obtain optimal parameters, and the optimal parameters are used to optimize the individual physical model and the aggregated cluster model to obtain the trained aggregated cluster model and the trained aggregated cluster model .

Step S206, based on the test set data, verify the trained aggregate cluster model and the trained aggregate cluster model, and obtain the target individual physical model and the target aggregate cluster model;

Step S207, inputting the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated index data of the energy data to be measured;

In the embodiment of the present invention, the energy data to be measured is input into the target individual physical model and the target aggregated cluster model, the predicted aggregated index data of the energy data to be measured is obtained, and the energy data to be measured is determined. Corresponds to the degree of aggregation of energy.

In the specific implementation, the aggregation characteristics of the cluster depend on the characteristics of various distributed resources participating in the combination, as well as the internal resource regulation strategy. The cluster server needs to adjust the interactive resource output or working state within the allowable range, so that the overall power of the system is close to the target curve. Since the interactive resources respond to the regulation commands of the cluster and have certain power regulation characteristics, after aggregation, the cluster power regulation capability will have a scale effect, and the power curve will be smoother.

The space area to which the user belongs is divided in units of grid connection points to obtain the spatial aggregation characteristics of the cluster:

After aggregating the response features distributed in different geographic locations, the temporal aggregation features of the cluster are obtained:

In the formula, Vi ^,j,k are 0-1 variables, when Vi ^,j,k is 1, it means participating in the response, ΔP _Group,t means the response value provided by the cluster in the t period, P _Group,t means the post-response The actual aggregated load value of the cluster at time period t.

In the method for aggregating energy data provided by the embodiment of the present invention, by acquiring renewable energy data and energy data to be measured, and based on the renewable energy data, an individual physical model and an aggregated cluster model of renewable energy are established, The renewable energy data is divided into training set data and test set data, and based on the K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are trained and analyzed. Verify, obtain the target individual physical model and the target aggregation cluster model, input the energy data to be measured into the target individual physical model and the target aggregation cluster model, and obtain the predicted aggregation index data of the energy data to be measured, Through an energy data aggregation method, the existing technical problem that the social idle distributed renewable resources are difficult to aggregate is solved, and the access standard of renewable resources is unified.

Please refer to FIG. 4. FIG. 4 is a structural block diagram of an embodiment of an energy data aggregation apparatus according to the present invention, including:

an acquisition module 401, used for acquiring renewable energy data and energy data to be measured;

A building module 402, configured to build an individual physical model and an aggregated cluster model of renewable energy based on the renewable energy data;

A dividing module 403, configured to divide the renewable energy data into training set data and test set data;

A training module 404, configured to train and verify the individual physical model and the aggregated cluster model based on the K-means clustering algorithm, the training set data and the test set data, to obtain a target individual physical model and a target aggregate cluster model;

The indicator module 405 is configured to input the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated indicator data of the energy data to be measured.

In an optional embodiment, the obtaining module 401 includes:

Obtain sub-module for obtaining initial data of renewable energy;

The repairing submodule is used for cleaning and repairing the initial data of renewable energy to obtain the renewable energy data and the energy data to be measured.

In an optional embodiment, the training module 404 includes:

A training sub-module for using the K-means clustering algorithm, in combination with the training set data, to train the individual physical model and the aggregated cluster model to obtain a trained aggregated cluster model and a trained aggregated cluster Model;

A verification submodule, configured to verify the trained aggregate cluster model and the trained aggregate cluster model based on the test set data, and obtain the target individual physical model and the target aggregate cluster model.

In an optional embodiment, the training submodule includes:

a prediction unit, configured to input the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data;

an error unit, configured to determine a training error according to the data label corresponding to the training set data and the predicted aggregate value;

An optimization unit, configured to adjust the individual physical model and the aggregated cluster model through the K-means clustering algorithm based on the training error to obtain optimal parameters, and use the optimal parameters to optimize all the parameters. The individual physical model and the aggregated cluster model are obtained to obtain the trained aggregated cluster model and the trained aggregated cluster model.

In an optional embodiment, the training submodule further includes:

A parameter unit for initializing parameters of the individual physical model and parameters of the aggregated cluster model.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above may refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the method and apparatus disclosed in the present invention may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware, or may be implemented in the form of software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as independent products, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present invention can be embodied in the form of a software product in essence, or the part that contributes to the prior art, or the whole or part of the technical solution, and the computer software product is stored in a readable storage The medium includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned readable storage medium includes: U disk, removable hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes.

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. an aggregation method of energy data, is characterized in that, comprises: Obtain renewable energy data and energy data to be measured; Based on the renewable energy data, establish an individual physical model and an aggregated cluster model of renewable energy; dividing the renewable energy data into training set data and test set data; Based on the K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are trained and verified to obtain the target individual physical model and the target aggregated cluster model; Inputting the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated index data of the energy data to be measured. 2. The aggregation method of energy data according to claim 1, wherein obtaining renewable energy data and energy data to be measured, comprising: Obtain initial data on renewable energy; The initial data of renewable energy is cleaned and repaired to obtain the renewable energy data and the energy data to be measured. 3. The method for aggregating energy data according to claim 1, wherein, based on K-means clustering algorithm, the training set data and the test set data, the individual physical model and the aggregated cluster model are Perform training and verification to obtain the target individual physical model and the target aggregated cluster model, including: Using the K-means clustering algorithm, in combination with the training set data, the individual physical model and the aggregated cluster model are trained to obtain the aggregated cluster model after training and the aggregated cluster model after training; Based on the test set data, the trained aggregate cluster model and the trained aggregate cluster model are verified to obtain the target individual physical model and the target aggregate cluster model. 4. The aggregation method of energy data according to claim 3, characterized in that, using the K-means clustering algorithm, in combination with the training set data, the individual physical model and the aggregation cluster model are trained, Get the trained aggregate cluster model and the trained aggregate cluster model, including: Inputting the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data; Determine the training error according to the data label corresponding to the training set data and the predicted aggregate value; Based on the training error, through the K-means clustering algorithm, the individual physical model and the aggregated cluster model are adjusted to obtain optimal parameters, and the optimal parameters are used to optimize the individual physical model and the aggregated cluster model. For the aggregated cluster model, the trained aggregated cluster model and the trained aggregated cluster model are obtained. 5 . The energy data aggregation method according to claim 4 , wherein the training set data is input into the individual physical model and the aggregation cluster model, and before the predicted aggregation value of the corresponding renewable energy data is obtained. 6 . ,Also includes: The parameters of the individual physical model and the parameters of the aggregated cluster model are initialized. 6. A device for aggregation of energy data, comprising: The acquisition module is used to acquire renewable energy data and energy data to be measured; establishing a module for establishing an individual physical model and an aggregated cluster model of renewable energy based on the renewable energy data; a dividing module, configured to divide the renewable energy data into training set data and test set data; A training module for training and verifying the individual physical model and the aggregated cluster model based on the K-means clustering algorithm, the training set data and the test set data, to obtain a target individual physical model and a target aggregated cluster Model; The indicator module is used for inputting the energy data to be measured into the target individual physical model and the target aggregated cluster model to obtain predicted aggregated indicator data of the energy data to be measured. 7. The apparatus for aggregating energy data according to claim 6, wherein the acquisition module comprises: Obtain sub-module for obtaining initial data of renewable energy; The repairing submodule is used for cleaning and repairing the initial data of renewable energy to obtain the renewable energy data and the energy data to be measured. 8. The apparatus for aggregating energy data according to claim 6, wherein the training module comprises: A training sub-module for using the K-means clustering algorithm, in combination with the training set data, to train the individual physical model and the aggregated cluster model to obtain a trained aggregated cluster model and a trained aggregated cluster Model; A verification submodule, configured to verify the trained aggregate cluster model and the trained aggregate cluster model based on the test set data, and obtain the target individual physical model and the target aggregate cluster model. 9. The apparatus for aggregating energy data according to claim 8, wherein the training submodule comprises: a prediction unit, configured to input the training set data into the individual physical model and the aggregated cluster model to obtain a predicted aggregated value of the corresponding renewable energy data; an error unit, configured to determine a training error according to the data label corresponding to the training set data and the predicted aggregate value; An optimization unit, configured to adjust the individual physical model and the aggregated cluster model through the K-means clustering algorithm based on the training error to obtain optimal parameters, and use the optimal parameters to optimize all the parameters. The individual physical model and the aggregated cluster model are obtained to obtain the trained aggregated cluster model and the trained aggregated cluster model. 10. The apparatus for aggregating energy data according to claim 9, wherein the training submodule further comprises: A parameter unit for initializing parameters of the individual physical model and parameters of the aggregated cluster model.

Images