Power load forecasting method, device, electronic device and storage medium

Abstract

The present application provides a method, device, electronic device and storage medium for predicting electric load, which relates to the field of microgrid management technology. The method for predicting electric load includes: obtaining user characteristic information of a first user in a microgrid, the electric load of the first user on the current date, and environmental information of the first user, wherein the environmental information includes: geographical location, climate information and / or date type of the current date; using the user characteristic information of the first user, the electric load of the first user on the current date and the environmental information of the first user as a trained electric load prediction model to obtain the predicted electric load of a specified day output by the electric load prediction model. The method can solve the problem that the load law is difficult to grasp and the prediction is easy to be inaccurate in the prior art, and is helpful to improve the safety and stability of the microgrid system.

Description

Technical Field

[0001] The present application relates to the field of microgrid management technology, and in particular to a method, device, electronic device and storage medium for power load forecasting. Background Art

[0002] With the improvement of my country's industrial structure and the improvement of people's overall living standards, the demand for electricity has increased year by year, and due to the simultaneity of electricity production and consumption, higher requirements have been put forward for smart distribution networks, smart microgrids and layouts. The short-term load forecasting results of the power system in the microgrid environment are related to the scheduling and operation of the power system, the formulation of production plans, and the cost of power generation. However, the factors affecting the power load include weather factors such as temperature and precipitation, as well as the influence of human factors such as holidays and industrial economic activities. The randomness and uncertainty of these factors lead to great difficulties in the prediction of power load. At present, in the load forecasting of the power system in the microgrid environment, the uncertainty of the influencing factors makes it difficult to grasp the load law, and inaccurate load forecasting can easily affect the safety and stability of the microgrid system. Summary of the invention

[0003] In order to solve the above technical problems, the purpose of the embodiments of the present application is to provide a power load forecasting method, device, electronic device and storage medium.

[0004] In order to solve the above technical problems, in a first aspect, an embodiment of the present application provides a method for predicting power load, comprising:

[0005] Acquire user characteristic information of a first user in the microgrid, the power load of the first user on the current date, and environmental information of the first user, wherein the environmental information includes: geographic location, climate information of a specified day, and / or date type of the specified day;

[0006] The user characteristic information of the first user, the power load of the first user on the current date and the environmental information of the first user are used as inputs of the trained power load prediction model to obtain the predicted power load on the specified day output by the power load prediction model.

[0007] In the above implementation process, by obtaining the characteristic information of users in the microgrid, the power load and environmental information of the current date, the environmental information includes the user's geographical location, climate information and date type, and inputting the obtained information into the trained power load prediction model, the power load of the user's specified day is predicted by the power load prediction model. It can be seen that the power load of the microgrid user's specified day can be predicted based on the influence of various factors such as user characteristics, geographical location, climate, date type, etc. on the user's power load, which can solve the problem of difficult to grasp the load law and easy inaccurate prediction in the prior art, and help to improve the safety and stability of the microgrid system, thereby facilitating the dispatching and operation of the power system and the formulation of production plans, and reducing the cost of power generation.

[0008] Optionally, before acquiring user characteristic information of a first user in the microgrid, the power load of the first user on the current date, and environmental information of the first user, the method further includes:

[0009] Acquire a training data set, the training data set comprising: power loads of multiple users with different environmental information within a preset time period and user characteristic information of the multiple users;

[0010] Acquire daily feature vectors of users with different user feature information under various environmental information according to the training data set to obtain multiple daily feature vectors;

[0011] The multiple daily feature vectors are used as input vectors of a preset power load forecasting initial model, and the power load forecasting initial model is trained to obtain the power load forecasting model after the training is completed.

[0012] In the above implementation process, the power load of multiple users with different environmental information within a preset time period and the user characteristic information of the multiple users are obtained as a training data set, and the daily characteristic vectors of users with different user characteristic information under various environmental information are obtained according to the training data set to obtain multiple daily characteristic vectors, and the multiple daily characteristic vectors are used as input vectors of the pre-set power load prediction initial model to train the power load prediction initial model to obtain a trained power load prediction model. The above training data set can reflect the impact of various factors such as user characteristics, geographical location, climate, date type, etc. on the user's power load. Therefore, the power load prediction model obtained by the above-mentioned large amount of historical data on the power load prediction initial model can improve the accuracy of power load prediction for microgrid users.

[0013] Optionally, the climate information includes temperature information, weather information and season information, the date types include working days, weekends and statutory holidays, the power loads of the multiple users with different environmental information within a preset time period include: the power consumption and daily average power consumption of each user in the multiple users in different temperature ranges, different weather, different seasons and different date types in each period of a day within N years, and the user characteristic information of the multiple users includes the industry type, power consumption scale and power consumption characteristics of each user in the multiple users;

[0014] The step of obtaining daily feature vectors of users with different user feature information under various environmental information according to the training data set to obtain multiple daily feature vectors includes:

[0015] Determine multiple scenario modes according to the power consumption and daily average power consumption of each user in each period of a day in different temperature ranges, different weather information, different season information and different date types within N years, and the industry type, power consumption scale and power consumption characteristics of each user; wherein the same industry type, power consumption scale, power consumption characteristics, and at least one of the temperature range, weather information, season information and date type are the same scenario mode;

[0016] A daily feature vector corresponding to each scenario mode is generated according to the industry type, electricity consumption scale, electricity consumption characteristics, temperature range, weather information, season information and / or date type corresponding to each scenario mode, as well as the electricity consumption in each time period and the daily average electricity consumption on the date that conforms to each scenario mode.

[0017] In the above implementation process, by dividing the factors affecting the user's power load into different scenario modes, such as industry type, power consumption scale, power consumption characteristics, temperature range, weather information, season information and / or date type, the daily feature vectors corresponding to different scenario modes are determined, and these daily feature vectors are used as the input of the initial power load prediction model. The impact of different scenario modes on the user's power load can be taken into account when training the model. Therefore, the power load prediction model obtained by the above method can improve the accuracy of power load prediction for microgrid users and improve the accuracy and stability of power prediction.

[0018] Optionally, the user characteristic information of the first user includes the industry type, electricity consumption scale and electricity consumption characteristics of the first user, the electricity load of the first user on the current date includes the electricity consumption of the first user at various times on the current date and the daily average electricity consumption, the climate information of the specified day includes temperature information, weather information, and season information, and the date type of the specified day includes working days, weekends and statutory holidays;

[0019] The method of using the user characteristic information of the first user, the power load of the first user on the current date, and the environmental information of the first user as inputs of the trained power load prediction model to obtain the predicted power load at each time of the specified day output by the power load prediction model comprises:

[0020] Generate a daily feature vector according to the industry type of the first user, the electricity consumption scale and the electricity consumption characteristics, the electricity consumption at each time of the current date, the temperature information, the weather information, the season information and / or the date type;

[0021] The daily feature vector is used as an input vector of the power load forecasting model to obtain the predicted power load of a specified day output by the power load forecasting model, wherein the predicted power load of the specified day includes the predicted power consumption at each time.

[0022] In the above implementation process, the daily feature vector is generated as the input vector of the power load forecasting model based on the user's industry type, power consumption scale and power consumption characteristics, power consumption at various times of the current date, temperature information, weather information, season information and / or date type, and the predicted power load of the specified day output by the power load forecasting model is obtained. The influence of factors such as user characteristics, temperature information, weather information, season information and / or date type on power consumption can be considered in the prediction, thereby improving the accuracy of power load forecasting for microgrid users.

[0023] Optionally, the temperature information includes the minimum temperature, the maximum temperature, and the average temperature. The season information includes: heating season and cooling season. The heating season includes the early cold period, the late cold period and the severe cold period. The cooling season includes the early summer period, the late summer period and the hot summer period. The statutory holidays include domestic statutory holidays and foreign statutory holidays.

[0024] In the above implementation process, since climate conditions have a great impact on the user's power load, the factors affecting climate conditions are further divided, and the climate information is further divided. The temperature information is divided into the lowest temperature, the highest temperature and the average temperature, and the season information is divided into the heating season and the cooling season. The heating season includes the early cold period, the late cold period and the severe cold period, and the cooling season includes the early summer period, the late summer period and the hot summer period. Further division by the factors affecting climate conditions can make the power load forecast more accurate.

[0025] Optionally, the power load prediction model is a neural network model.

[0026] In the above implementation process, a neural network model can be used as the power load prediction model. Since the impact of various factors on the user's power load is nonlinear, the use of a neural network model can map out these nonlinear impacts. At the same time, the neural network model has excellent learning and adaptability, and can cope with the complex impact of various influences on the user's power load. At the same time, it also has a certain fault tolerance capability, which can improve the accuracy of power load prediction for microgrid users.

[0027] In a second aspect, an embodiment of the present application provides a power load prediction device, including: a data acquisition module, used to obtain user characteristic information of a first user in a microgrid, the power load of the first user on a current date, and environmental information of the first user, wherein the environmental information includes: geographic location, climate information and / or date type of the current date;

[0028] The prediction module is used to use the user characteristic information of the first user, the power load of the first user on the current date and the environmental information of the first user as a trained power load prediction model to obtain the predicted power load on the specified day output by the power load prediction model.

[0029] In the above implementation process, the characteristic information of users in the microgrid, the power load of the current date and environmental information are obtained through the data acquisition module. The environmental information includes the user's geographical location, climate information and date type. The obtained information is input into the trained power load prediction model through the prediction module to predict the power load of the user's specified day. It can be seen that the power load of the microgrid user on the specified day can be predicted based on the influence of various factors such as user characteristics, geographical location, climate, date type, etc. on the user's power load, which can solve the problem of difficult to grasp the load law and easy inaccurate prediction in the prior art, and help to improve the safety and stability of the microgrid system, thereby facilitating the dispatching and operation of the power system and the formulation of production plans, and reducing the cost of power generation.

[0030] Optionally, the device further comprises:

[0031] A data acquisition module, configured to acquire a training data set before acquiring user characteristic information of a first user in the microgrid, the power load of the first user on a current date, and the environmental information of the first user, wherein the training data set includes: power loads of multiple users with different environmental information within a preset time period and user characteristic information of the multiple users;

[0032] A vector acquisition module, used to acquire daily feature vectors of users with different user feature information under various environmental information according to the training data set, so as to obtain multiple daily feature vectors;

[0033] The training module is used to use the multiple daily feature vectors as input vectors of a preset power load forecasting initial model to train the power load forecasting initial model, so as to obtain the power load forecasting model after completing the training.

[0034] In the above implementation process, the power load of multiple users with different environmental information within a preset time period and the user characteristic information of the multiple users are obtained as a training data set, and the daily characteristic vectors of users with different user characteristic information under various environmental information are obtained according to the training data set to obtain multiple daily characteristic vectors, and the multiple daily characteristic vectors are used as input vectors of the pre-set power load prediction initial model to train the power load prediction initial model to obtain a trained power load prediction model. The above training data set can reflect the impact of various factors such as user characteristics, geographical location, climate, date type, etc. on the user's power load. Therefore, the power load prediction model obtained by the above-mentioned large amount of historical data on the power load prediction initial model can improve the accuracy of power load prediction for microgrid users.

[0035] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the memory stores program instructions, and when the processor runs the program instructions, the steps of any one of the methods described in the first aspect are executed.

[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising: computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement the steps of any one of the methods described in the first aspect above. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A flow chart of a method for predicting power load provided in an embodiment of the present application;

[0039] Figure 2 A flowchart of a power load prediction model training method provided in an embodiment of the present application;

[0040] Figure 3 A flowchart of a method for generating a daily feature vector provided in an embodiment of the present application;

[0041] Figure 4A flowchart of another power load forecasting method provided in an embodiment of the present application;

[0042] Figure 5 A structural diagram of a power load prediction device provided in an embodiment of the present application;

[0043] Figure 6 A structural diagram of another power load prediction device provided in an embodiment of the present application;

[0044] Figure 7 A structural diagram of a vector acquisition module provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0046] Figure 1 For a flow chart of a power load forecasting method provided in an embodiment of the present application, please refer to Figure 1 , the power load forecasting method may include the following steps:

[0047] In step S101, user characteristic information of a first user in a microgrid, the power load of the first user on a current date, and environmental information of the first user are obtained.

[0048] For example, the information of the first user in the microgrid can be obtained through the database of the power grid power consumption information collection system and the database of related influencing factors. Among them, the user characteristic information may include: the type of application industry in which the user is located, the scale of power consumption of the user and the power consumption characteristics of the user. The type of application industry in which the user is located may include: manufacturing, service industry and tourism industry, etc. The scale of power consumption of the user can usually be divided into: large, medium and small, and the power consumption characteristics of the user can usually be divided into: centralized power consumption type and segmented power consumption type.

[0049] The above-mentioned environmental information may include: geographical location, climate information of the specified day and / or date type of the specified day, the climate information may include temperature information, weather information and season information, wherein the temperature information may include: the highest temperature of the day, the lowest temperature of the day, the average temperature of the day, or the temperature information may include the temperature corresponding to each time of the day; the weather information includes: various weather conditions such as sunny days, cloudy days, rainy days, snowy days, etc., and may also include extreme climates such as typhoon days; the season information may include: spring, summer, autumn and winter, and the date types may include weekdays, weekends and statutory holidays.

[0050] In step S102, the user characteristic information of the first user, the power load of the first user on the current date and the environmental information of the first user are used as inputs of the trained power load prediction model to obtain the predicted power load of the specified day output by the power load prediction model.

[0051] The power load prediction model may be a trained neural network model, and the designated date may be any date after the current date, such as the next day.

[0052] The above implementation method can predict the power load of microgrid users on a specified day based on the impact of various factors such as user characteristics, geographical location, climate, date type, etc. on user power load. It can solve the problems in the prior art that load patterns are difficult to grasp and inaccurate predictions are easy to make, and help improve the safety and stability of the microgrid system, thereby facilitating the scheduling and operation of the power system and the formulation of production plans, and reducing power generation costs.

[0053] Figure 2 This is a flowchart of a power load prediction model training method provided in an embodiment of the present application. The power load prediction model training method can be executed before step S101. Figure 2 , the method comprises the following steps:

[0054] In step S201, a training data set is obtained.

[0055] The training data set may include: power loads of multiple users in different environmental information within a preset time period and user characteristic information of the multiple users.

[0056] For example, the preset duration can be selected as a relatively long time, for example, it can be in units of years, to obtain the power load of multiple users in different environmental information within the preset duration in the previous few years and the user characteristic information of the multiple users. Among them, the number of multiple users can be in units of hundreds, thousands, tens of thousands or more, so that a relatively large amount of data can be obtained, thereby providing sufficient training data for the training of the model. Different environmental information may include different geographical locations, climate information and / or date types. For example, the user characteristic information of multiple users in the microgrid can be obtained through the power grid power consumption information collection system database and the relevant influencing factor database. For example, the power load of the user within the preset duration can be obtained from the power grid power consumption information collection system database, the geographical location can be obtained through the database of the map service agency, and the climate information can be obtained through the database of the meteorological monitoring agency.

[0057] In step S202, daily feature vectors of users with different user feature information under various environmental information are obtained according to the training data set to obtain a plurality of daily feature vectors.

[0058] In step S203, the plurality of daily feature vectors are used as input vectors of a preset initial model for power load prediction, and the initial model for power load prediction is trained to obtain the power load prediction model after the training is completed.

[0059] Through the above implementation, after obtaining the power load of multiple users under different environmental information within a preset time period and the user characteristic information of the multiple users, the user characteristic information of the multiple users is processed and converted into multiple daily feature vectors, and the multiple daily feature vectors are input into the initial model of power load prediction, and the initial model of power load prediction is trained so that the model reaches the target state, thereby obtaining a trained power load prediction model.

[0060] Furthermore, the climate information may include temperature information, weather information and season information, the date type may include working days, weekends and statutory holidays, and the user characteristic information of the multiple users may include the industry type, power consumption scale and power consumption characteristics of each user among the multiple users. Accordingly, the power load of the multiple users in different environmental information within a preset time period includes: the power consumption and daily average power consumption of each user among the multiple users in different temperature ranges, different weather, different seasons and different date types in each period of the day within N years. Figure 3 For a flowchart of a method for generating a daily feature vector provided in an embodiment of the present application, please refer to Figure 3 , the above step S202 may include the following steps:

[0061] In step S2021, multiple scenario modes are determined based on the electricity consumption and daily average electricity consumption in different time periods of a day for each of the multiple users in different temperature ranges, different weather information, different seasonal information and different date types within N years, and the industry type, electricity consumption scale and electricity consumption characteristics of each user.

[0062] Among them, the same scenario mode refers to the same industry type, electricity consumption scale, electricity consumption characteristics, temperature range, weather information, season information and date type. In practical applications, N can be set to 5, that is, to obtain user data in the last 5 years, so that there are enough samples. If the user's data in the power grid electricity consumption information collection system database is less than 5 years, all the user's data will be obtained.

[0063] For example, the above season information and date information can be divided in more detail. For example, in addition to early summer, autumn and winter, the season information can be further divided into cooling season and heating season on the basis of the four seasons, wherein the cooling season can include: early summer, late summer and hot summer, and the heating season can include: early cold period, late cold period and severe cold period. Legal holidays can include: domestic legal holidays and foreign legal holidays.

[0064] Taking the early summer as an example, the power consumption and daily average power consumption of each user in each period of each day in the early summer within 5 years can be obtained respectively, which can include: the temperature information, weather information, season information, and date type (working day, weekend or statutory holiday) of each user in the early summer every day, as well as the power consumption and daily average power consumption of each period of each day, where the power consumption of each period can be in the form of a curve. The methods for the late summer, hot summer, early cold, late cold and severe cold are the same and will not be listed one by one.

[0065] A more detailed division of the above-mentioned influencing factors will help to more accurately analyze the impact of the above-mentioned influencing factors on the power load, making the power load forecasting model more accurate in predicting user power consumption.

[0066] In step S2022, a daily feature vector corresponding to each scenario mode is generated based on the industry type, electricity consumption scale, electricity consumption characteristics, temperature range, weather information, season information and / or date type corresponding to each scenario mode, as well as the electricity consumption in each time period and the daily average electricity consumption in the date that conforms to each scenario mode.

[0067] Among them, each scenario mode may correspond to multiple daily feature vectors. In this case, the multiple daily feature vectors corresponding to the scenario mode can be used for model training, or the average vector of the multiple daily feature vectors can be taken as the daily feature vector corresponding to the scenario mode during model training.

[0068] Correspondingly, the user characteristic information of the first user may include the industry type, electricity consumption scale and electricity consumption characteristics of the first user, the electricity load of the first user on the current date may include the electricity consumption of the first user at various times of the current date and the daily average electricity consumption, the climate information of the specified day includes the temperature information, weather information and season information of the specified day, and the date type of the specified day includes working days, weekends and statutory holidays. For example, if the specified day to be predicted is the next day, the climate information of the next day may be obtained, including the temperature information, weather information and season information of the next day, and the date type of the next day, including the working day, weekend and statutory holidays of the next day. Figure 4 For a flowchart of another power load forecasting method provided in an embodiment of the present application, please refer to Figure 4, the above step S102 may include the following steps:

[0069] In step S1021, a daily feature vector is generated based on the industry type of the first user, the electricity consumption scale and the electricity consumption characteristics, the electricity consumption at each time of the current date, the temperature information of the specified day, the weather information, the season information and / or the date type of the specified day.

[0070] Among them, since the power load forecasting model is trained based on multiple users in N years, the above model can be used to predict any date after the current date (it should be understood that it is under the premise that the temperature information of the specified day, the weather information, the season information and / or the date type of the specified day can be obtained). If a day needs to be predicted, a day feature vector can be generated based on the temperature information, weather information, season information and / or date type of that day. For example, when predicting the next day of the current date, a day feature vector can be generated based on the industry type of the first user, the scale of electricity consumption and the characteristics of electricity consumption, the electricity consumption at each moment of the current date, the temperature information of the next day, the weather information, the season information and / or the date type of the next day.

[0071] In step S1022, the daily feature vector is used as an input vector of the power load forecasting model to obtain the predicted power load of a specified day output by the power load forecasting model, wherein the predicted power load of the specified day includes the predicted power consumption at each time.

[0072] Among them, it should be noted that the input is the power consumption at each moment, and the output is the power consumption at each moment of the specified day. Taking the predicted specified day as the next day as an example, the output is the power consumption at each moment of the next day, that is, the power consumption at each moment of the next day. According to the power consumption at each moment of the next day, the daily average power of the next day, or the average power of each time period, can be further determined, or the power consumption (kWh) of each time period of the next day can be calculated according to the power consumption at each moment of the next day. For example, each time period can be divided into one-hour intervals, and a day can be divided into 24 time periods, or the time period division can be performed at a larger or smaller time granularity.

[0073] Exemplarily, during the prediction, the above season information and date information can also be divided in more detail. As in the above training method, the season information can be further divided into cooling season and heating season, wherein the cooling season can include: early summer, late summer and hot summer, and the heating season can include: early cold period, late cold period and severe cold period. Legal holidays can include: domestic legal holidays and foreign legal holidays. After the above influencing factors are divided in more detail, it is conducive to more accurate analysis of the above influencing factors, so that the power load forecasting model can more accurately predict the user's electricity consumption.

[0074] Optionally, the power load prediction model can be a neural network model. Since the various factors affecting power load mentioned above have a nonlinear effect on the user's power load, the use of a neural network model can map out these nonlinear effects. At the same time, the neural network model has excellent learning and adaptability, and can cope with the complex impacts of various influences on the user's power load. It also has a certain fault tolerance capability, which can improve the accuracy of power load prediction for microgrid users.

[0075] For example, in the neural network model, neuron i can be used to represent the actual electrical equipment of a certain user, such as large machine tools, refrigeration or heating equipment, etc. The input of neuron i can represent the impact of the above-mentioned various influencing factors on the electrical equipment, and the output of neuron i represents the change in the user's electrical load caused by the influencing factors. The relationship between the input and output of neuron i can be expressed as:

[0076]

[0077]

[0078] Among them, net i represents the net activation of neuron i, i represents the user's electrical device i, i = 0, 1, 2, 3, ..., n, if net i If the value of is positive, it means that the neuron i is in an activated or excited state, indicating that the influencing factor will increase the load of the corresponding electrical equipment i, such as net iIf the value of is negative, it means that the neuron is in an inhibited state, indicating that the influencing factor will reduce the load of the corresponding power-consuming device i; x represents the input of the neuron, representing the above-mentioned influencing factors, namely the user's industry type, power consumption scale and power consumption characteristics, power consumption at various times of the current date, temperature information, weather information, season information, date type and other influencing factors; j represents each power-consuming device under a certain user or each sub-user under a user system of a certain user; n represents n power-consuming devices or n sub-users under the user system; w represents the connection weight, representing the proportion of the value of the input information affecting the neuron during the input process. The larger the value of w of the influencing factor, the greater the impact of the influencing factor on the user's power load; θ is a threshold value, which means that the maximum or minimum value of w cannot exceed the threshold value, which is used to illustrate that in practical applications, the impact of each influencing factor on the user's power load is limited and cannot be unlimited; Represents the output of neuron i, that is, the predicted power load of the corresponding electrical equipment i, where function f represents the activation function. Since the influence of the above factors on the user's power load is nonlinear, the use of activation function can map out these nonlinear influences.

[0079] For example, if the threshold θ is regarded as an input x of neuron i 0 The weight of , the above formula can be simplified to:

[0080]

[0081]

[0082] Wherein, X represents the input vector, which is the daily feature vector mentioned above;

[0083] That is: X = [x 0 ,x 1 ,x 2 ,…,x n ], where x 0 ,x 1 ,x 2 ,…,x n Represent the above-mentioned influencing factors respectively.

[0084]

[0085] Let W represent the weight vector, and the above formula can be expressed as vector multiplication:

[0086] net i =XW

[0087]

[0088] The nonlinear effects of the above-mentioned influencing factors on the power load can be mapped out through the daily feature vector, the weight vector, and the activation function, so that the power load of a certain user's power device i at each time on a specified day can be predicted according to the above formula.

[0089] Through the above method, the predicted power load of multiple power-consuming devices 0 to n under a user can be predicted one by one, and the predicted power load of the user can be obtained by summing the predicted power load of the multiple power-consuming devices.

[0090] In the above formula, the input factor can be an influencing factor for a device, and the predicted power load is the predicted load of the device. The input factor can also be an influencing factor for a user, that is, it includes multiple devices, and the predicted power load is the predicted load of the user.

[0091] For the entire microgrid system, after predicting the power load of each user on a specified day, the power load of the entire microgrid system on the specified day can be obtained.

[0092] Based on the same inventive concept, the embodiment of the present application also provides a power load prediction device, Figure 5 For a structural diagram of a power load prediction device provided in an embodiment of the present application, please refer to Figure 5 .

[0093] The power load prediction device 50 may include:

[0094] The data acquisition module 501 is used to obtain user characteristic information of a first user in the microgrid, the power load of the first user on the current date, and environmental information of the first user, wherein the environmental information includes: geographic location, climate information of a specified day, and / or date type of a specified day.

[0095] The prediction module 502 is used to use the user characteristic information of the first user, the power load of the first user on the current date and the environmental information of the first user as inputs to the trained power load prediction model to obtain the predicted power load on the specified day output by the power load prediction model.

[0096] Optional, Figure 6 For a structural diagram of another power load prediction device provided in an embodiment of the present application, please refer to Figure 6 , the power load prediction device 50 may further include:

[0097] The data acquisition module 503 is used to acquire a training data set before acquiring the user characteristic information of the first user in the microgrid, the power load of the first user on the current date, and the environmental information of the first user, wherein the training data set includes: the power load of multiple users with different environmental information within a preset time period and the user characteristic information of the multiple users;

[0098] A vector acquisition module 504 is used to acquire daily feature vectors of users with different user feature information under various environmental information according to the training data set to obtain multiple daily feature vectors;

[0099] The training module 505 is used to use the multiple daily feature vectors as input vectors of a preset power load forecasting initial model to train the power load forecasting initial model, so as to obtain the power load forecasting model after the training is completed.

[0100] Optionally, the climate information includes temperature information, weather information and season information, the date types include weekdays, weekends and statutory holidays, the power load of the multiple users with different environmental information within a preset time period includes: the power consumption and daily average power consumption of each user among the multiple users in different temperature ranges, different weather, different seasons and different date types in each time period of the day within N years, and the user characteristic information of the multiple users includes the industry type, power consumption scale and power consumption characteristics of each user among the multiple users. Figure 7 For a structural diagram of a vector acquisition module provided in an embodiment of the present application, please refer to Figure 7 , the vector acquisition module 504 may include:

[0101] The scenario determination submodule 5041 is used to determine multiple scenario modes according to the power consumption and daily average power consumption of each user in different temperature ranges, different weather information, different season information and different date types in N years in each period of a day, and the industry type, power consumption scale and power consumption characteristics of each user; wherein the same industry type, power consumption scale, power consumption characteristics, and at least one of the temperature range, weather information, season information and date type are the same scenario mode;

[0102] The vector generation submodule 5042 is used to generate a daily feature vector corresponding to each scenario mode based on the industry type, electricity consumption scale, electricity consumption characteristics, temperature range, weather information, season information and / or date type corresponding to each scenario mode, as well as the electricity consumption in each time period and the daily average electricity consumption in the date that conforms to each scenario mode.

[0103] Optionally, the user characteristic information of the first user includes the industry type, electricity consumption scale and electricity consumption characteristics of the first user, the electricity load of the first user on the current date includes the electricity consumption of the first user at various times on the current date and the daily average electricity consumption, the climate information of the specified day includes temperature information, weather information, and season information, and the date type of the specified day includes working days, weekends and statutory holidays; the prediction module 502 is used to:

[0104] Generate a daily feature vector according to the industry type of the first user, the electricity consumption scale and the electricity consumption characteristics, the electricity consumption at each time of the current date, the temperature information, the weather information, the season information and / or the date type;

[0105] The daily feature vector is used as an input vector of the power load forecasting model to obtain the predicted power load of a specified day output by the power load forecasting model, wherein the predicted power load of the specified day includes the predicted power consumption at each time.

[0106] Among them, the industry types may include: manufacturing, service and tourism, etc.; the electricity consumption scale may include: large, medium and small; the electricity consumption characteristics may include: centralized electricity consumption type and segmented electricity consumption type; the weather information may include various weather conditions such as sunny days, cloudy days, rainy days, snowy days, etc., and may also include extreme climates such as typhoon days; the seasonal information may include: heating season and cooling season, the heating season may include the early cold period, the late cold period and the severe cold period, the cooling season may include the early summer period, the late summer period and the hot summer period, and the date types may include working days, weekends and statutory holidays.

[0107] It can be seen that by obtaining the characteristic information of users in the microgrid, the power load and environmental information of the current date, the environmental information includes the user's geographical location, climate information and date type, and inputting the obtained information into the trained power load prediction model, the power load of the user's specified day is predicted by the power load prediction model. It can be seen that the power load of the microgrid user's specified day can be predicted based on the influence of various factors such as user characteristics, geographical location, climate, date type, etc. on the user's power load, which can solve the problem of difficult to grasp the load law and easy inaccurate prediction in the prior art, and help improve the safety and stability of the microgrid system, thereby facilitating the dispatching and operation of the power system and the formulation of production plans, and reducing the cost of power generation.

[0108] An embodiment of the present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores program instructions, and when the processor runs the program instructions, the steps of the power load forecasting method described in any of the above embodiments are executed.

[0109] An embodiment of the present application also provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the steps of the power load forecasting method described in any of the above embodiments are executed.

[0110] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0111] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0112] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0113] It should be noted that if the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM) random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0114] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0115] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting power load, characterized in that: include: Acquire user characteristic information of a first user in the microgrid, the power load of the first user on a current date, and environmental information of the first user, wherein the environmental information includes: geographic location, climate information of a specified day, and / or date type of the specified day; Using the user characteristic information of the first user, the power load of the first user on the current date, and the environmental information of the first user as inputs of the trained power load forecasting model to obtain the predicted power load on the specified day output by the power load forecasting model; The user characteristic information of the first user includes the industry type, electricity consumption scale and electricity consumption characteristics of the first user, the electricity load of the first user on the current date includes the electricity consumption of the first user at various times on the current date and the daily average electricity consumption, the climate information of the specified day includes temperature information, weather information, and season information, and the date type of the specified day includes working days, weekends and statutory holidays; The method of using the user characteristic information of the first user, the power load of the first user on the current date, and the environmental information of the first user as inputs of the trained power load prediction model to obtain the predicted power load at each time of the specified day output by the power load prediction model comprises: Generate a daily feature vector according to the industry type of the first user, the electricity consumption scale and the electricity consumption characteristics, the electricity consumption at each time of the current date, the temperature information, the weather information, the season information and / or the date type; The daily feature vector is used as an input vector of the power load forecasting model to obtain a predicted power load of a specified day output by the power load forecasting model, wherein the predicted power load of the specified day includes predicted power consumption at each time; The power load prediction model is a neural network model; The input and output relationship of the neural network model is: Among them, i is a neuron, representing the electrical equipment of the first user, represents the output of the neuron i, net i represents the net activation of the neuron i, w ij represents the connection weight, x n represents the component of the daily feature vector, and j represents a sub-user of the first user.

2. The method according to claim 1, characterized in that Before acquiring user characteristic information of a first user in the microgrid, the power load of the first user on the current date, and environmental information of the first user, the method further includes: Acquire a training data set, the training data set comprising: power loads of multiple users with different environmental information within a preset time period and user characteristic information of the multiple users; Acquire daily feature vectors of users with different user feature information under various environmental information according to the training data set to obtain multiple daily feature vectors; The multiple daily feature vectors are used as input vectors of a preset power load forecasting initial model, and the power load forecasting initial model is trained to obtain the power load forecasting model after the training is completed.

3. The method according to claim 2, characterized in that The climate information includes temperature information, weather information and season information, the date types include working days, weekends and statutory holidays, the power loads of the multiple users with different environmental information within a preset time period include: the power consumption and daily average power consumption of each user in the multiple users in different temperature ranges, different weather, different seasons and different date types in each period of a day within N years, and the user characteristic information of the multiple users includes the industry type, power consumption scale and power consumption characteristics of each user in the multiple users; The step of obtaining daily feature vectors of users with different user feature information under various environmental information according to the training data set to obtain multiple daily feature vectors includes: Determine multiple scenario modes according to the power consumption and daily average power consumption of each user in each period of a day in different temperature ranges, different weather information, different season information and different date types within N years, and the industry type, power consumption scale and power consumption characteristics of each user; wherein the same industry type, power consumption scale, power consumption characteristics, and at least one of the temperature range, weather information, season information and date type are the same scenario mode; A daily feature vector corresponding to each scenario mode is generated according to the industry type, electricity consumption scale, electricity consumption characteristics, temperature range, weather information, season information and / or date type corresponding to each scenario mode, as well as the electricity consumption in each time period and the daily average electricity consumption on the date that conforms to each scenario mode.

4. The method according to claim 1 or 3, characterized in that: The temperature information includes the minimum temperature, the maximum temperature, and the average temperature. The season information includes: heating season and cooling season. The heating season includes the early cold period, the late cold period, and the severe cold period. The cooling season includes the early summer period, the late summer period, and the hot summer period. The statutory holidays include domestic statutory holidays and foreign statutory holidays.

5. A power load forecasting device, characterized in that: include: A data acquisition module, used to obtain user characteristic information of a first user in the microgrid, the power load of the first user on the current date, and environmental information of the first user, wherein the environmental information includes: geographic location, climate information and / or date type of the current date; A prediction module, configured to use the user characteristic information of the first user, the power load of the first user on the current date, and the environmental information of the first user as a trained power load prediction model to obtain a predicted power load on a specified day output by the power load prediction model; The user characteristic information of the first user includes the industry type, electricity consumption scale and electricity consumption characteristics of the first user, the electricity load of the first user on the current date includes the electricity consumption of the first user at various times on the current date and the daily average electricity consumption, the climate information of the specified day includes temperature information, weather information, and season information, and the date type of the specified day includes working days, weekends and statutory holidays; In the process of using the user characteristic information of the first user, the power load of the first user on the current date and the environmental information of the first user as inputs of the trained power load forecasting model to obtain the predicted power load at each moment of the specified day output by the power load forecasting model, the forecasting module is specifically used to: generate a daily feature vector according to the industry type of the first user, the power consumption scale and the power consumption characteristics, the power consumption at each moment of the current date, the temperature information, the weather information, the season information and / or the date type; use the daily feature vector as the input vector of the power load forecasting model to obtain the predicted power load of the specified day output by the power load forecasting model, wherein the predicted power load of the specified day includes the predicted power consumption at each moment; The power load prediction model is a neural network model; The input and output relationship of the neural network model is: Among them, i is a neuron, representing the electrical equipment of the first user, represents the output of the neuron i, net i represents the net activation of the neuron i, w ij represents the connection weight, x n represents the component of the daily feature vector, and j represents a sub-user of the first user.

6. The device according to claim 5, characterized in that The device also includes: A data acquisition module, configured to acquire a training data set before acquiring user characteristic information of a first user in the microgrid, the power load of the first user on a current date, and the environmental information of the first user, wherein the training data set includes: power loads of multiple users with different environmental information within a preset time period and user characteristic information of the multiple users; A vector acquisition module, used to acquire daily feature vectors of users with different user feature information under various environmental information according to the training data set, so as to obtain multiple daily feature vectors; The training module is used to use the multiple daily feature vectors as input vectors of a preset power load forecasting initial model to train the power load forecasting initial model, so as to obtain the power load forecasting model after completing the training.

7. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores program instructions, and when the processor runs the program instructions, the steps in the method according to any one of claims 1 to 4 are executed.

8. A computer-readable storage medium, characterized in that: include: Computer program instructions are stored thereon, and when the computer program instructions are executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images