A distribution scheduling decision method, system and storage medium based on LSTM algorithm

Abstract

The present invention discloses a distribution scheduling decision method, system and storage medium based on LSTM algorithm, and relates to the technical field of distribution network scheduling decision. The present invention first obtains the historical power load of each power unit in the distribution area; based on the historical power load of each power unit, the daily power load forecast value of each power unit is predicted based on the LSTM algorithm to obtain the total daily power load forecast value of the distribution area; the daily dispatchable capacity of each type of distributed power source in the distribution network is obtained; according to the daily power load forecast value of each power unit, the total daily power load forecast value of the distribution area and the daily dispatchable capacity of each type of distributed power source, the objective function and constraint conditions of the daily distribution scheduling are constructed; under the constraint conditions, the objective function is solved using a multi-objective optimization algorithm to obtain the daily distribution scheduling decision. The present invention can improve the distribution efficiency of distributed power sources, reduce the distribution cost, and optimize the distribution scheduling decision plan.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network dispatching decision-making, and more specifically to a power distribution dispatching decision-making method, system and storage medium based on LSTM algorithm. Background Art

[0002] With the advancement of smart distribution network construction, the requirements for optimal allocation of distribution network resources are increasing, which puts higher requirements on dispatching operation and dispatching optimization. The formulation of dispatching strategies needs to comprehensively consider multiple factors such as load forecasting, power supply quality, and clean energy consumption. Formulate reasonable dispatching plans to ensure that the distribution network can operate stably in various operating scenarios and meet the electricity needs of users.

[0003] In the existing scheduling decision-making schemes, usually only factors such as electricity purchase costs and power constraints are considered. In addition, there are usually multiple distributed power sources and multiple electricity users in the distribution area. The existing solutions usually determine the responsibility scope of each distributed power source through clustering and other methods, without considering the daily dispatchable capacity of each distributed power source, and even more lack consideration of the distance factor between the distributed power sources and the electricity users.

[0004] Therefore, how to improve the distribution efficiency of distributed power sources and optimize distribution scheduling decisions while ensuring the safe and stable operation of the distribution network is an urgent problem that technical personnel in this field need to solve. Summary of the invention

[0005] In view of this, the present invention provides a distribution scheduling decision method, system and storage medium based on the LSTM algorithm, which is used to obtain an optimized distribution scheduling decision plan according to the daily power load forecast value of the power user, the total daily power load forecast value of the distribution area and the daily dispatchable capacity of various types of distributed power sources.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A distribution scheduling decision method based on LSTM algorithm includes the following steps:

[0008] Step 1: Obtain the historical power load of each power user in the power distribution area;

[0009] Step 2: According to the historical power load of each power user, the daily power load forecast value of each power user is predicted based on the LSTM algorithm to obtain the total daily power load forecast value of the distribution area;

[0010] Step 3: Obtain the daily dispatchable capacity of each type of distributed power source in the distribution network;

[0011] Step 4: According to the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of each type of distributed power source, the objective function and constraint conditions of the daily distribution dispatch are constructed;

[0012] Step 5: Under the constraints, a multi-objective optimization algorithm is used to solve the objective function to obtain a daily distribution scheduling decision.

[0013] Optionally, in step 2, the method for predicting the daily power load forecast value of each power user based on the LSTM algorithm is:

[0014] Step 2.1, collect data, obtain the actual value of historical power load, and obtain historical power load series data;

[0015] Step 2.2, model training, using the historical power load sequence data as training data, training the LSTM neural network model to obtain a power load prediction model;

[0016] Step 2.3, model prediction, inputting the day-ahead power load sequence data of each power consumer into the power load prediction model to obtain the daily power load prediction value of each power consumer.

[0017] Optionally, in step 4, the objective function of intraday power distribution scheduling includes:

[0018] Electricity purchase cost objective function:

[0019]

[0020] Among them, c 1t represents the electricity price at time t, T t represents the amount of electricity purchased at time t;

[0021] Distribution distance objective function:

[0022]

[0023] Where i represents the i-th power consumption unit in the distribution area, N represents the number of power consumption units in the distribution area; j represents the j-th distributed power source in the distribution area, M represents the number of distributed power sources in the distribution area; c ij represents the power supply relationship between the i-th power user and the j-th distributed generation. When the j-th distributed generation supplies power to the i-th power user, c ij The value is 1, otherwise the value is 0; l ij Represents the power distribution distance between the i-th power consumer and the j-th distributed generation.

[0024] Optionally, in step 4, the constraints of intraday power distribution scheduling include:

[0025] Constraints on purchased power:

[0026] P min ≤P ij ≤P max ;

[0027] Among them, P ij represents the power purchased between the i-th power user and the j-th distributed generation, P min and P max are the lower and upper limits of the purchased power respectively;

[0028] Voltage constraints:

[0029] U min ≤U ij ≤U max ;

[0030] Among them, U ij represents the three-phase node voltage between the i-th power consumption unit and the j-th distributed generation, U min and U max are the lower and upper limits of the voltage of the three-phase nodes of the distribution network respectively;

[0031] Line power constraints:

[0032] Y ij ≤Y max ;

[0033] Among them, Y ij represents the power distribution line power between the i-th power user and the j-th distributed generation, Y max Indicates the maximum power allowed by the line.

[0034] Optionally, the constraint condition also includes:

[0035] Distributed generation capacity constraints during the day:

[0036]

[0037] Among them, X j represents the dispatchable capacity of the jth distributed generation within a day, T ij It indicates the power supply provided by the j-th distributed generation to the i-th power user.

[0038] Optionally, in step 5, the process of solving the objective function using a multi-objective optimization algorithm is:

[0039] Step 5.1: Encode the parameters in the objective function, randomly assign values ​​to the encoded parameters, generate an initial population, decode each set of encoded values ​​in the initial population, and obtain the intraday distribution scheduling decision parameters;

[0040] Step 5.2, according to the intraday power distribution dispatch decision parameters, dispatch according to the rules to obtain the dispatch result;

[0041] Step 5.3: Filter and sort the obtained scheduling results, and select the solutions that are not dominated by any other results as the non-inferior solution set generated by this scheduling;

[0042] Step 5.4, randomly extracting a preset number of solutions from the non-inferior solution set, and encoding the parameter sets corresponding to these solutions as offspring;

[0043] Step 5.5, performing hybridization and mutation operations on the offspring to generate a new population, and decoding to obtain new intraday distribution scheduling decision parameters, and scheduling according to the rules to obtain the final scheduling result;

[0044] Step 5.6, repeat steps 5.2 to 5.5 until the difference between the dispatch result obtained in this round and the dispatch result obtained in the previous round is within the preset range, and the new intraday distribution dispatch decision parameters generated in the last round are used as the final intraday distribution dispatch decision.

[0045] A distribution dispatch decision system based on LSTM algorithm, including the following modules:

[0046] The historical data collection module is used to obtain the historical power load of each power user in the power distribution area;

[0047] The power load forecasting module is used to predict the daily power load forecast value of each power user based on the LSTM algorithm according to the historical power load of each power user, and obtain the total daily power load forecast value of the distribution area;

[0048] A daily dispatchable capacity determination module is used to obtain the daily dispatchable capacity of various types of distributed power sources in the distribution network;

[0049] The objective function determination module is used to construct the objective function and constraint conditions of the daily distribution scheduling according to the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of various types of distributed power sources;

[0050] The scheduling decision solving module is used to solve the objective function under the constraints using a multi-objective optimization algorithm to obtain the intra-day distribution scheduling decision.

[0051] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the computer program implements any of the above-mentioned distribution scheduling decision methods based on the LSTM algorithm.

[0052] It can be seen from the above technical solutions that the present invention provides a distribution scheduling decision method, system and storage medium based on LSTM algorithm, which have the following beneficial effects compared with the prior art:

[0053] The present invention first predicts the daily power load forecast value of each power user and the total daily power load forecast value of the entire distribution area based on the LSTM algorithm. According to the daily power load forecast value of the power user, the total daily power load forecast value of the distribution area and the daily dispatchable capacity of each type of distributed power source, multiple objective functions and constraints are determined by considering factors such as the daily dispatchable capacity of the distributed power source and the distance between the distributed power source and the power user, and a multi-objective optimization genetic algorithm is used to obtain the optimal distribution scheduling decision plan. Through the scheme of the present invention, it is ensured that under the condition of safe and stable operation of the distribution network, the distribution efficiency of the distributed power source is improved, the distribution cost is reduced, and the distribution scheduling decision plan is optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0055] Figure 1 It is a schematic diagram of the method steps of the present invention;

[0056] Figure 2 It is a schematic diagram of the system module of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] The embodiment of the present invention discloses a distribution scheduling decision method based on LSTM algorithm, see Figure 1 , including the following steps:

[0059] Step 1: Obtain the historical power load of each power user in the distribution area.

[0060] Step 2: According to the historical power load of each power consumer, the daily power load forecast value of each power consumer is predicted based on the LSTM algorithm to obtain the total daily power load forecast value of the distribution area.

[0061] The method for predicting the daily power load forecast value of each power user based on the LSTM algorithm is as follows:

[0062] Step 2.1, collect data, obtain the actual value of historical power load, and obtain the historical power load sequence data. In the specific implementation process, the actual value of the historical power load data collected is the data from 30 days to 5 days ago. The data from 30 days to 15 days ago is used as the training set, and the data from 15 days to 5 days ago is used as the test set for subsequent training and testing.

[0063] After collecting historical data and before model training, data preprocessing is performed on the collected historical data. The data preprocessing process includes outlier removal and filtering to obtain normalized historical data.

[0064] Step 2.2: Model training: using the historical power load sequence data as training data, the LSTM neural network model is trained to obtain a power load prediction model.

[0065] The core of the LSTM neural network model lies in the "memory unit" and three special "gate" structures: input gate, forget gate and output gate. Among them, the input gate is responsible for deciding when to write the input information of the current time step into the memory unit. The forget gate is used to control the forgetting of information. The output gate is used to control the output of information. In this embodiment, the training data, that is, the training set, is input into the LSTM neural network model through the input gate, the predicted power load is output from the output gate, the predicted power load value is compared with the actual power load value in the training data, and the parameters of the LSTM neural network model are adjusted according to the difference, and finally a power load prediction model that meets the preset accuracy is obtained.

[0066] Step 2.3, model prediction, inputting the day-ahead power load sequence data of each power consumer into the power load prediction model to obtain the daily power load prediction value of each power consumer.

[0067] By summing up the daily power load forecast values ​​of each power user, the total daily power load forecast value of the distribution area can be obtained.

[0068] Step 3: Obtain the daily dispatchable capacity of each type of distributed generation in the distribution network.

[0069] Step 4: Construct the objective function and constraints of daily distribution scheduling based on the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of various types of distributed power sources.

[0070] (1) The objective function of intraday distribution scheduling includes:

[0071] 1) Electricity purchase cost objective function:

[0072]

[0073] Among them, c 1t represents the electricity price at time t, T t represents the amount of electricity purchased at time t;

[0074] 2) Distribution distance objective function:

[0075]

[0076] Where i represents the i-th power consumption unit in the distribution area, N represents the number of power consumption units in the distribution area; j represents the j-th distributed power source in the distribution area, M represents the number of distributed power sources in the distribution area; c ij represents the power supply relationship between the i-th power user and the j-th distributed generation. When the j-th distributed generation supplies power to the i-th power user, c ij The value is 1, otherwise the value is 0; l ij Represents the power distribution distance between the i-th power consumer and the j-th distributed generation.

[0077] (2) The constraints of intraday distribution scheduling include:

[0078] 1) Power purchase constraints:

[0079] P min ≤P ij ≤P max ;

[0080] Among them, P ij represents the power purchased between the i-th power user and the j-th distributed generation, P min and P max are the lower and upper limits of the purchased power respectively;

[0081] 2) Voltage constraints:

[0082] U min ≤U ij ≤U max ;

[0083] Among them, U ijrepresents the three-phase node voltage between the i-th power consumption unit and the j-th distributed generation, U min and U max are the lower and upper limits of the voltage of the three-phase nodes of the distribution network respectively;

[0084] 3) Line power constraints:

[0085] Y ij ≤Y max ;

[0086] Among them, Y ij represents the power distribution line power between the i-th power user and the j-th distributed generation, Y max Indicates the maximum power allowed by the line.

[0087] 4) Dispatchy capacity constraints of distributed generation in a day:

[0088]

[0089] Among them, X j represents the dispatchable capacity of the jth distributed generation within a day, T ij It indicates the power supply provided by the j-th distributed generation to the i-th power user.

[0090] 5) Daily power load constraints of power users:

[0091]

[0092] Among them, Z i It represents the daily power load forecast value of the i-th power consumer.

[0093] 6) Daily power load constraints in the distribution area:

[0094]

[0095] Among them, Z represents the total predicted value of daily power load in the distribution area.

[0096] Step 5: Under the constraints, use a multi-objective optimization algorithm to solve the objective function and obtain the intraday distribution scheduling decision. Specifically:

[0097] Step 5.1: Encode the parameters in the objective function, randomly assign values ​​to the encoded parameters, generate an initial population, decode each set of encoded values ​​in the initial population, and obtain the intraday distribution scheduling decision parameters;

[0098] Step 5.2, according to the intraday power distribution dispatch decision parameters, dispatch according to the rules to obtain the dispatch result;

[0099] Step 5.3: Filter and sort the obtained scheduling results, and select the solutions that are not dominated by any other results as the non-inferior solution set generated by this scheduling;

[0100] Step 5.4, randomly extracting a preset number of solutions from the non-inferior solution set, and encoding the parameter sets corresponding to these solutions as offspring;

[0101] Step 5.5, performing hybridization and mutation operations on the offspring to generate a new population, and decoding to obtain new intraday distribution scheduling decision parameters, and scheduling according to the rules to obtain the final scheduling result;

[0102] Step 5.6, repeat steps 5.2 to 5.5 until the difference between the dispatch result obtained in this round and the dispatch result obtained in the previous round is within the preset range, and the new intraday distribution dispatch decision parameters generated in the last round are used as the final intraday distribution dispatch decision.

[0103] Another embodiment of the present invention further provides a distribution scheduling decision system based on LSTM algorithm, see Figure 2 , used to implement the above method embodiment, including the following modules:

[0104] The historical data collection module is used to obtain the historical power load of each power user in the power distribution area;

[0105] The power load forecasting module is used to predict the daily power load forecast value of each power user based on the LSTM algorithm according to the historical power load of each power user, and obtain the total daily power load forecast value of the distribution area;

[0106] A daily dispatchable capacity determination module is used to obtain the daily dispatchable capacity of various types of distributed power sources in the distribution network;

[0107] The objective function determination module is used to construct the objective function and constraint conditions of the daily distribution scheduling according to the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of various types of distributed power sources;

[0108] The scheduling decision solving module is used to solve the objective function under the constraints using a multi-objective optimization algorithm to obtain the intra-day distribution scheduling decision.

[0109] As for the system module disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0110] Another embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the distribution scheduling decision method based on the LSTM algorithm described in any one of the above items is implemented.

[0111] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0112] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A distribution scheduling decision method based on LSTM algorithm, characterized in that: The following steps are involved: Step 1: Obtain the historical power load of each power user in the power distribution area; Step 2: According to the historical power load of each power user, the daily power load forecast value of each power user is predicted based on the LSTM algorithm to obtain the total daily power load forecast value of the distribution area; Step 3: Obtain the daily dispatchable capacity of each type of distributed power source in the distribution network; Step 4: According to the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of each type of distributed power source, the objective function and constraint conditions of the daily distribution dispatch are constructed; In step 4, the objective function of intraday power distribution scheduling includes: Electricity purchase cost objective function: Among them, c 1t represents the electricity price at time t, T t represents the amount of electricity purchased at time t; Distribution distance objective function: Where i represents the i-th power consumption unit in the distribution area, N represents the number of power consumption units in the distribution area; j represents the j-th distributed power source in the distribution area, M represents the number of distributed power sources in the distribution area; c ij represents the power supply relationship between the i-th power user and the j-th distributed generation. When the j-th distributed generation supplies power to the i-th power user, c ij The value is 1, otherwise the value is 0; l ij represents the power distribution distance between the i-th power user and the j-th distributed generation; In step 4, the constraints of intraday power distribution scheduling include: Constraints on purchased power: P min ≤P ij ≤P max ; Among them, P ij represents the power purchased between the i-th power user and the j-th distributed generation, P min and P max are the lower and upper limits of the purchased power respectively; Voltage constraints: IN min ≤U ij ≤U max ; Among them, U ij represents the three-phase node voltage between the i-th power consumption unit and the j-th distributed generation, U min and U max are the lower and upper limits of the voltage of the three-phase nodes of the distribution network respectively; Line power constraints: AND ij ≤And max ; Among them, Y ij represents the power distribution line power between the i-th power user and the j-th distributed generation, Y max Indicates the maximum power allowed by the line; Step 5: Under the constraints, a multi-objective optimization algorithm is used to solve the objective function to obtain a daily distribution scheduling decision.

2. According to a distribution scheduling decision method based on LSTM algorithm according to claim 1, it is characterized in that: In step 2, the method for predicting the daily power load forecast value of each power user based on the LSTM algorithm is: Step 2.1, collect data, obtain the actual value of historical power load, and obtain historical power load series data; Step 2.2, model training, using the historical power load sequence data as training data, training the LSTM neural network model to obtain a power load prediction model; Step 2.3, model prediction, inputting the day-ahead power load sequence data of each power consumer into the power load prediction model to obtain the daily power load prediction value of each power consumer.

3. According to a distribution scheduling decision method based on LSTM algorithm according to claim 1, it is characterized in that: The constraints also include: Distributed generation capacity constraints during the day: Among them, Xj represents the dispatchable capacity of the jth distributed generation within a day, T ij It indicates the power supply provided by the j-th distributed generation to the i-th power user.

4. The power distribution scheduling decision method based on LSTM algorithm according to claim 1 is characterized in that: In step 5, the process of solving the objective function using the multi-objective optimization algorithm is as follows: Step 5.1: Encode the parameters in the objective function, randomly assign values ​​to the encoded parameters, generate an initial population, decode each set of encoded values ​​in the initial population, and obtain the intraday distribution scheduling decision parameters; Step 5.2, according to the intraday power distribution dispatch decision parameters, dispatch according to the rules to obtain the dispatch result; Step 5.3: Filter and sort the obtained scheduling results, and select the solutions that are not dominated by any other results as the non-inferior solution set generated by this scheduling; Step 5.4, randomly extracting a preset number of solutions from the non-inferior solution set, and encoding the parameter sets corresponding to these solutions as offspring; Step 5.5, performing hybridization and mutation operations on the offspring to generate a new population, and decoding to obtain new intraday distribution scheduling decision parameters, and scheduling according to the rules to obtain the final scheduling result; Step 5.6, repeat steps 5.2 to 5.5 until the difference between the dispatch result obtained in this round and the dispatch result obtained in the previous round is within the preset range, and the new intraday distribution dispatch decision parameters generated in the last round are used as the final intraday distribution dispatch decision.

5. A distribution dispatch decision system based on LSTM algorithm, characterized in that: Includes the following modules: The historical data collection module is used to obtain the historical power load of each power user in the power distribution area; The power load forecasting module is used to predict the daily power load forecast value of each power user based on the LSTM algorithm according to the historical power load of each power user, and obtain the total daily power load forecast value of the distribution area; A daily dispatchable capacity determination module is used to obtain the daily dispatchable capacity of various types of distributed power sources in the distribution network; The objective function determination module is used to construct the objective function and constraint conditions of the daily distribution scheduling according to the daily power load forecast value of each power user, the total daily power load forecast value of the distribution area, and the daily dispatchable capacity of various types of distributed power sources; The objective functions of intraday distribution scheduling include: Electricity purchase cost objective function: Among them, c 1t represents the electricity price at time t, T t represents the amount of electricity purchased at time t; Distribution distance objective function: Where i represents the i-th power consumption unit in the distribution area, N represents the number of power consumption units in the distribution area; j represents the j-th distributed power source in the distribution area, M represents the number of distributed power sources in the distribution area; c ij represents the power supply relationship between the i-th power user and the j-th distributed generation. When the j-th distributed generation supplies power to the i-th power user, c ij The value is 1, otherwise the value is 0; l ij represents the power distribution distance between the i-th power user and the j-th distributed generation; The constraints of intraday distribution scheduling include: Constraints on purchased power: P min ≤P ij ≤P max ; Among them, P ij represents the power purchased between the i-th power user and the j-th distributed generation, P min and P max are the lower and upper limits of the purchased power respectively; Voltage constraints: IN min ≤U ij ≤U max ; Among them, U ij represents the three-phase node voltage between the i-th power consumption unit and the j-th distributed generation, U min and U max are the lower and upper limits of the voltage of the three-phase nodes of the distribution network respectively; Line power constraints: AND ij ≤And max ; Among them, Y ij represents the power distribution line power between the i-th power user and the j-th distributed generation, Y max Indicates the maximum power allowed by the line; The scheduling decision solving module is used to solve the objective function under the constraints using a multi-objective optimization algorithm to obtain the intra-day distribution scheduling decision.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, it implements a distribution scheduling decision method based on the LSTM algorithm as described in any one of claims 1 to 4.

Images