Intelligent power supply station work order scheduling method and system, storage medium and electronic terminal

Abstract

The invention discloses an intelligent power supply station work order scheduling method and system, a storage medium and an electronic terminal. The method comprises the following steps: collecting electric energy data, load data and renewable energy data in a power system in real time; pre-processing the electric energy data and the load data, wherein the pre-processing comprises data interpolation and abnormal data elimination; screening out electric energy data and load data with high attention weight from the preprocessed electric energy data and load data by adopting an attention mechanism; predicting work order data to be scheduled by using a recurrent neural network (RNN) based on the attention mechanism, and outputting a first prediction result; predicting the renewable energy source data by adopting a Benders decomposition algorithm, and outputting a second prediction result; and making a power supply station work order scheduling strategy by referring to the first prediction result and the second prediction result. By utilizing the method, the importance and the processing efficiency of the data are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of work order dispatching of intelligent power supply stations, and in particular to a work order dispatching method, system, storage medium and electronic terminal of an intelligent power supply station. Background Art

[0002] With the continuous growth of power supply and the increasing demand of users, the optimization and dispatching of work orders of intelligent power supply stations has become a key task in the power system. The traditional dispatching method of power supply station work orders often relies on manual processing and rule making, which has problems such as low efficiency, easy errors and inability to adapt to the complex and changing power supply environment. Therefore, a more intelligent and efficient method is needed to achieve the optimization and dispatching of power supply station work orders.

[0003] At present, deep learning technology has made significant progress in the field of power systems, especially recurrent neural networks (RNNs). RNNs are a type of neural network that can process sequence data and have a memory function that can capture the temporal dependencies between data, so they perform well in processing time series data.

[0004] However, the existing RNN-based work order scheduling method still has some problems, such as the inability to accurately handle a large amount of abnormal data and the lack of accurate judgment on the priority and urgency of work orders. Summary of the invention

[0005] In view of the technical problems pointed out in the background technology, the purpose of the present invention is to provide a method, system, storage medium and electronic terminal for scheduling work orders of an intelligent power supply station.

[0006] To achieve the purpose of the present invention, the technical solution provided by the present invention is specifically as follows:

[0007] First aspect

[0008] The present invention provides a method for dispatching work orders of an intelligent power supply station, comprising the following steps:

[0009] Step S1: real-time collection of electric energy data, load data and renewable energy data in the power system;

[0010] Step S2: preprocessing the electric energy data and load data, wherein the preprocessing includes data interpolation and abnormal data elimination;

[0011] Step S3: using the attention mechanism to filter out the preprocessed power data and load data with high attention weights;

[0012] Step S4: using a recurrent neural network (RNN) based on the attention mechanism to predict the work order data to be dispatched, and outputting a first prediction result focusing on power data and load data;

[0013] Step S5: using the Benders decomposition algorithm to predict the renewable energy data collected in step S1, and outputting a second prediction result focusing on the renewable energy data;

[0014] Step S6: Referring to the first prediction result and the second prediction result, a power supply station work order dispatching strategy is formulated, including the following:

[0015] Supply and demand balancing: adjusting the output of conventional power plants to meet the forecasted load, taking into account the forecasted output of renewable energy sources and using them to reduce reliance on fossil fuels;

[0016] Energy storage management: If forecasts show high renewable energy production and low load in the short term, the excess power can be stored in the energy storage system and released during peak demand periods;

[0017] Flexible scheduling: For those loads that can be adjusted, their usage time can be adjusted according to the forecast results to alleviate the pressure during peak hours;

[0018] Emergency preparedness: If forecasts show that supply shortages are likely, emergency preparedness measures need to be put in place in advance;

[0019] Market trading decisions: Power companies can also decide whether to buy or sell electricity in the electricity market based on the forecast results to optimize costs and profits.

[0020] Second aspect

[0021] Corresponding to the above method, the present invention also provides a smart power supply station work order dispatching system, comprising the following units: a data acquisition unit, a data preprocessing unit, an attention mechanism processing unit, a first prediction result output unit and a second prediction result output unit, and a work order dispatching strategy unit;

[0022] The data acquisition unit is used to collect electric energy data, load data and renewable energy data in the power system in real time;

[0023] The data preprocessing unit is used to preprocess the electric energy data and load data, and the preprocessing includes data interpolation and abnormal data elimination;

[0024] The attention mechanism processing unit is used to use the attention mechanism to filter out the power data and load data with high attention weights from the pre-processed power data and load data;

[0025] The first prediction result output unit is used to use the recurrent neural network RNN ​​based on the attention mechanism to predict the dispatch work order data, and output the first prediction result focusing on the power data and the load data.

[0026] The second prediction result output unit is used to use the Benders decomposition algorithm to predict the renewable energy data and output a second prediction result focusing on the renewable energy data.

[0027] The work order dispatching strategy unit is used to formulate a work order dispatching strategy for the power supply station by referring to the first prediction result and the second prediction result, including the following:

[0028] Supply and demand balancing: adjusting the output of conventional power plants to meet the forecasted load, taking into account the forecasted output of renewable energy sources and using them to reduce reliance on fossil fuels;

[0029] Energy storage management: If forecasts show high renewable energy production and low load in the short term, the excess power can be stored in the energy storage system and released during peak demand periods;

[0030] Flexible scheduling: For those loads that can be adjusted, their usage time can be adjusted according to the forecast results to alleviate the pressure during peak hours;

[0031] Emergency preparedness: If forecasts show that supply shortages are likely, emergency preparedness measures need to be put in place in advance;

[0032] Market trading decisions: Power companies can also decide whether to buy or sell electricity in the electricity market based on the forecast results to optimize costs and profits.

[0033] The third aspect

[0034] The present invention also provides a storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the smart power supply station work order scheduling method.

[0035] The fourth aspect

[0036] The present invention also provides a terminal, which includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the smart power supply station work order scheduling method.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention proposes a method, system, storage medium and electronic terminal for dispatching work orders of intelligent power supply stations. The scheme combines attention mechanism and recurrent neural network to realize intelligent optimization of the dispatching process of work orders of power supply stations. Through the attention mechanism, when screening power data, it will focus on important information and reduce the weight of unimportant data. Combined with the prediction results of high attention weight electric energy data, load data and renewable energy data, the importance and processing efficiency of data are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of a method flow chart provided in an embodiment of the present application;

[0040] Figure 2 A schematic diagram of a prediction curve of electric energy data provided in an embodiment of the present application. DETAILED DESCRIPTION

[0041] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0042] Example 1

[0043] like Figure 1 As shown, this embodiment provides a method for scheduling work orders of a smart power supply station, comprising the following steps:

[0044] Step S1: real-time collection of electric energy data, load data and renewable energy data in the power system;

[0045] It should be noted that by deploying smart meters, sensors and other equipment in power supply stations, real-time data on power, load and renewable energy in the power system can be collected. These data include historical power consumption of power loads, energy supply at different times and the volatility of renewable energy.

[0046] Step S2: preprocessing the electric energy data and load data, wherein the preprocessing includes data interpolation and abnormal data elimination;

[0047] It should be noted that in this step, the power data and load data are preprocessed. Data collection and processing is an indispensable step in subsequent predictions. Since some errors in data collection will lead to errors in the data analysis stage, the collected data needs to be preprocessed before the prediction analysis to achieve the purpose of eliminating abnormal data and stabilizing the historical power curve.

[0048] First, data interpolation was performed using the curve fitting method.

[0049] Secondly, periodicity is a major characteristic of power load changes. This feature can be used to first compare the data longitudinally, that is, the amount of charge required at the same time point on adjacent dates is similar. Usually, the data at that time point is circled within a period. If there is data outside the range, it is considered bad data and removed from the data set, and the data set is corrected. The detailed algorithm is as follows:

[0050] |y(d,t)|-m(t)>γ(t) (1)

[0051] y(d,t)=m(t)+γ(t) y(d,t)>m(t) (2)

[0052] y(d,t)=γ(t)-m(t) y(d,t) <m(t) (3)

[0053] Among them, m(t) represents the average value of electric energy data at time t in the vertical direction, γ(t) is the threshold of its variation range, and y(d, t) represents the load data value at time t on the dth day.

[0054] Step S3: using the attention mechanism to filter out the preprocessed power data and load data with high attention weights;

[0055] It should be noted that the attention mechanism is used to analyze and screen important power data and load data. The mechanism encodes the pre-processed power data and load data in step S2, calculates the attention weights of all power data and load data, and then screens out important data with higher attention weights. This screening process enables the system to focus on key data and give priority to the scheduling and processing of these data, thereby improving the accuracy and efficiency of work order scheduling. The adaptive learning characteristics of the attention mechanism enable it to dynamically adjust the attention weight according to different situations to adapt to the operating environment and business needs of the power supply station.

[0056] Through this intelligent attention scheduling, the method can effectively optimize the work order processing flow of the power supply station, realize intelligent and efficient work order scheduling, and improve the operating efficiency and service quality of the power supply station.

[0057] Among them, the formula for using the attention mechanism to screen out important power data and load data with high attention weights for the pre-processed power data and load data is as follows:

[0058] e ij =V T tanh(Wh i +Uy j +b) (4)

[0059]

[0060]

[0061] Among them, h i is the ith load data feature vector preprocessed in step S2, y j is the jth electric energy data feature vector after preprocessing in step S2; W and U are weight matrices used to map the input feature vector to a new representation space; b is a bias vector; V is a weight vector used to convert the result of the activation function tanh(·) into a scalar score; e ij is the attention score; a ij is the attention weight; c i According to the attention weight h i The result of weighted summation, T transpose matrix.

[0062] Step S4: Use the recurrent neural network RNN ​​based on the attention mechanism to predict the work order data to be dispatched, and output the first prediction result (focusing on the prediction of power data and load data), which usually includes the following contents:

[0063] Energy demand forecasts for a specific time period in the future: This can be given at hourly or finer-grained intervals, helping power companies understand upcoming demand peaks and troughs.

[0064] Load forecasting: This refers to the prediction of electricity consumption by different regions or users, which helps identify areas that may experience power shortages.

[0065] Power quality prediction: This may also include predictions of power quality (such as voltage fluctuations), which is crucial to ensuring stable operation of the power grid.

[0066] It should be noted that the work order scheduling data is represented by the electric energy data, load data and renewable energy data. Therefore, by referring to the optimized processing of the electric energy data, load data and renewable energy data, the optimization of the work order scheduling can be achieved.

[0067] It should be noted that the recurrent neural network RNN ​​is used to predict and analyze future work order scheduling information. The RNN network is a sequential data neural network processor because it has an internal memory that can update the state of each neuron in the network with previous inputs. RNN is trained using back propagation, and more layers are stacked in the RNN architecture, which provides more significant advantages. When the sequence time is represented from t to τ, the computational graph of the RNN has N sequential x i Input and sequence y i Output. In the mathematical representation of the RNN model calculation graph below, the number of layers from 1 to N is represented by l, and the activation function corresponding to l is represented by hl :

[0068]

[0069]

[0070]

[0071] Among them, x (t) is the work order data to be scheduled, y(t) is the predicted output, and the encoder is used to match the generated result with the power data and load data at the corresponding time point for similarity, and then the decoder is used to obtain the predicted power data and load data values. W and U are the weight matrices trained in step S3. Figure 2 A schematic diagram of a prediction curve of electric energy data provided in an embodiment of the present application.

[0072] Step S5: Use the Benders decomposition algorithm to predict the renewable energy data collected in step S1, and output a second prediction result (focusing on the prediction of renewable energy), which usually includes the following contents:

[0073] Renewable energy production forecasts for a specific time period in the future: This could be the expected output from a solar panel or wind turbine, again given in hourly or shorter time intervals.

[0074] Trends in renewable energy production: This helps assess the impact of weather changes on electricity generation.

[0075] Renewable energy availability forecast: that is, the actual availability of renewable energy after taking weather conditions into account.

[0076] It should be noted that the Benders decomposition algorithm is used for prediction to ensure the stability and reliability of power supply. It is used to solve the basic two-stage power system scheduling problem. The specific algorithm formula is as follows:

[0077]

[0078] Among them, s represents a probability scenario, which usually describes the uncertainty of renewable energy generation; p s represents the probability of the occurrence of probability scenario s; S is the set of all probability scenarios; the decision variable y describes the on / off decision of the thermal power unit; in the two-stage stochastic optimization problem, it is called the here-and-now decision z; A, C, B, D, b are learnable matrices, x s is the renewable energy data under the probability scenario s, Z is an integer set, R is a real number set, f is a constraint variable, n1 and n2 are the lengths of the vectors.

[0079] Step S6: refer to the first prediction result and the second prediction result to formulate a power supply station work order dispatching strategy. Combining the above two prediction results, a more reasonable power dispatching strategy can be formulated:

[0080] Supply and demand balance: Adjust the output of traditional power plants according to the forecast of power demand to meet the load forecast. At the same time, taking into account the forecast output of renewable energy, use these green energy sources as much as possible to reduce dependence on fossil fuels.

[0081] Energy storage management: If forecasts show high renewable energy production and low load in the short term, excess electricity can be stored in energy storage systems and released during peak demand periods.

[0082] Flexible scheduling: For those loads that can be adjusted (such as industrial users or electric vehicle charging stations), their usage time can be adjusted according to the forecast results to alleviate the pressure during peak hours.

[0083] Emergency preparedness: If forecasts show that supply shortages are likely, emergency preparedness measures need to be taken in advance, such as activating backup power sources or implementing power rationing measures.

[0084] Market trading decisions: Power companies can also decide whether to buy or sell electricity in the electricity market based on the forecast results to optimize costs and profits.

[0085] In addition, for different types of work orders, the work orders are also classified and managed according to their nature and urgency. For example, work orders of different professions such as line loss, metering, collection, energy efficiency, station services, and non-urgent matters can be classified into different categories. Each type of work order has specific attributes and processing requirements. Through classification management, work orders can be more accurately assigned to the corresponding responsible personnel or professional teams to ensure that each type of work order is handled professionally.

[0086] Secondly, the smart power supply station work order scheduling method also adopts a solution for automatically merging duplicate work orders. In actual operation, there may be multiple work orders involving the same substation, meter or the same task. In order to avoid repeated processing, the system will automatically detect and merge similar work orders into a unified work order. Through the matching and screening of intelligent algorithms, the same or similar tasks are merged into one work order, avoiding unnecessary repeated operations and reducing the burden of work order processing. This strategy of classification management and automatic merging of duplicate work orders effectively simplifies the work order processing process and improves the efficiency and response speed of the power supply station work order scheduling. By realizing one-time task dispatching, the work order can be directly dispatched to the responsible person responsible for handling such tasks, reducing the links of information transmission and task scheduling, making work order processing more efficient and timely.

[0087] Thirdly, the solution of the present invention supports user participation, allowing users to monitor and adjust energy consumption, so as to improve users' awareness and participation in energy use, and promote energy conservation and environmental awareness. Users can monitor their own electricity consumption and energy consumption in real time through devices such as smart meters, and understand the power load and energy supply. At the same time, users can also participate in the scheduling decision-making of the power system, adjust the power consumption mode according to their personal energy needs and energy-saving intentions, and contribute to energy optimization and environmental protection. Through user participation and feedback, the smart grid can more flexibly meet the needs of different users and improve the efficiency and sustainability of the overall energy system.

[0088] Example 2

[0089] This embodiment provides a smart power supply station work order dispatching system, including the following units: a data acquisition unit, a data preprocessing unit, an attention mechanism processing unit, a first prediction result output unit and a second prediction result output unit, and a work order dispatching strategy unit;

[0090] The data acquisition unit is used to collect electric energy data, load data and renewable energy data in the power system in real time;

[0091] The data preprocessing unit is used to preprocess the electric energy data and load data, and the preprocessing includes data interpolation and abnormal data elimination;

[0092] The attention mechanism processing unit is used to use the attention mechanism to filter out the power data and load data with high attention weights from the pre-processed power data and load data;

[0093] The first prediction result output unit is used to predict the work order scheduling data to be tested by using the recurrent neural network RNN ​​based on the attention mechanism, and output a first prediction result (focusing on the prediction of power data and load data);

[0094] The second prediction result output unit is used to predict the renewable energy data using a Benders decomposition algorithm and output a second prediction result (focusing on the prediction of renewable energy data);

[0095] The work order dispatching strategy unit is used to formulate a work order dispatching strategy for the power supply station with reference to the first prediction result and the second prediction result.

[0096] The data collection unit includes a smart meter and a sensor deployed in a power supply station.

[0097] Wherein, the data interpolation adopts the fitting curve method.

[0098] The formula for the attention mechanism processing unit to use the attention mechanism to screen out important power data and load data with high attention weights for the pre-processed power data and load data is as follows:

[0099] e ij =V T tanh(Wh i +Uy j +b)

[0100]

[0101]

[0102] Among them, h i is the ith load data feature vector preprocessed in step S2, y j is the jth electric energy data feature vector after preprocessing in step S2; W and U are weight matrices used to map the input feature vector to a new representation space; b is a bias vector; V is a weight vector used to convert the result of the activation function tanh(·) into a scalar score; e ij is the attention score; a ij is the attention weight; c i According to the attention weight h i The result of weighted summation, T transpose matrix.

[0103] Example 3

[0104] This embodiment provides a storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the smart power supply station work order scheduling method.

[0105] Example 4

[0106] This embodiment provides an electronic terminal, which includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the smart power supply station work order scheduling method.

[0107] The optional implementation modes of the embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation modes. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical scheme of the embodiments of the present invention, and these simple modifications all belong to the protection scope of the embodiments of the present invention.

Claims

1. A method for dispatching work orders for an intelligent power supply station, characterized in that: The steps include: Step S1: real-time collection of electric energy data, load data and renewable energy data in the power system; Step S2: preprocessing the electric energy data and load data, wherein the preprocessing includes data interpolation and abnormal data elimination; Step S3: using the attention mechanism to filter out the preprocessed power data and load data with high attention weights; Step S4: using the recurrent neural network RNN ​​based on the attention mechanism to predict the work order data to be dispatched, and outputting a first prediction result focusing on the prediction of power data and load data; Step S5: using the Benders decomposition algorithm to predict the renewable energy data collected in step S1, and outputting a second prediction result focusing on the prediction of renewable energy data; Step S6: Referring to the first prediction result and the second prediction result, a power supply station work order dispatching strategy is formulated, including the following: Supply and demand balancing: adjusting the output of conventional power plants to meet the forecasted load, taking into account the forecasted output of renewable energy sources and using them to reduce reliance on fossil fuels; Energy storage management: If forecasts show high renewable energy production and low load in the short term, the excess power can be stored in the energy storage system and released during peak demand periods; Flexible scheduling: For those loads that can be adjusted, their usage time can be adjusted according to the forecast results to alleviate the pressure during peak hours; Emergency preparedness: If forecasts show that supply shortages are likely, emergency preparedness measures need to be put in place in advance; Market trading decisions: Power companies can also decide whether to buy or sell electricity in the electricity market based on the forecast results to optimize costs and profits.

2. A method for dispatching work orders for a smart power supply station according to claim 1, characterized in that: In step S1, the electric energy data, load data and renewable energy data in the power system are collected in real time through the smart meters and sensors deployed in the power supply station.

3. The method for dispatching work orders of a smart power supply station according to claim 1, characterized in that: In step S2, the data interpolation adopts the curve fitting method.

4. The method for dispatching work orders of a smart power supply station according to claim 1, characterized in that: In step S3, the formula for using the attention mechanism to screen out important power data and load data with high attention weights is as follows: e ij =V T fishy i +Yes j +b) Among them, h i is the ith load data feature vector preprocessed in step S2, y j is the jth electric energy data feature vector after preprocessing in step S2; W and U are weight matrices used to map the input feature vector to a new representation space; b is a bias vector; V is a weight vector used to convert the result of the activation function tanh(·) into a scalar score; e ij is the attention score; a ij is the attention weight; c i It is the result of weighted summation of hi according to the attention weight, T is the transposed matrix.

5. An intelligent power supply station work order dispatching system, characterized in that: It includes the following units: a data collection unit, a data preprocessing unit, an attention mechanism processing unit, a first prediction result output unit, a second prediction result output unit, and a work order scheduling strategy unit; The data acquisition unit is used to collect electric energy data, load data and renewable energy data in the power system in real time; The data preprocessing unit is used to preprocess the electric energy data and load data, and the preprocessing includes data interpolation and abnormal data elimination; The attention mechanism processing unit is used to use the attention mechanism to filter out the power data and load data with high attention weights from the pre-processed power data and load data; The first prediction result output unit is used to use the recurrent neural network RNN ​​based on the attention mechanism to predict the dispatch work order data, and output the first prediction result focusing on the prediction of power data and load data. The second prediction result output unit is used to predict the renewable energy data using the Benders decomposition algorithm and output a second prediction result focusing on the prediction of renewable energy data. The work order dispatching strategy unit is used to formulate a work order dispatching strategy for the power supply station by referring to the first prediction result and the second prediction result, including the following: Supply and demand balancing: adjusting the output of conventional power plants to meet the forecasted load, taking into account the forecasted output of renewable energy sources and using them to reduce reliance on fossil fuels; Energy storage management: If forecasts show high renewable energy production and low load in the short term, the excess power can be stored in the energy storage system and released during peak demand periods; Flexible scheduling: For those loads that can be adjusted, their usage time can be adjusted according to the forecast results to alleviate the pressure during peak hours; Emergency preparedness: If forecasts show that supply shortages are likely, emergency preparedness measures need to be put in place in advance; Market trading decisions: Power companies can also decide whether to buy or sell electricity in the electricity market based on the forecast results to optimize costs and profits.

6. The intelligent power supply station work order dispatching system according to claim 5, characterized in that: The data collection unit includes a smart meter and a sensor deployed in a power supply station.

7. The intelligent power supply station work order dispatching system according to claim 5 is characterized in that: The data interpolation adopts the fitting curve method.

8. The intelligent power supply station work order dispatching system according to claim 5 is characterized in that: The attention mechanism processing unit uses the attention mechanism to filter out important power data and load data with high attention weights using the following formula: e ij =V T fishy i +Yes j +b) Among them, h i is the ith load data feature vector preprocessed in step S2, y j is the jth electric energy data feature vector after preprocessing in step S2; W and U are weight matrices used to map the input feature vector to a new representation space; b is a bias vector; V is a weight vector used to convert the result of the activation function tanh(·) into a scalar score; e ij is the attention score; a ij is the attention weight; c i According to the attention weight, h i The result of the weighted summation, T transposed matrix.

9. A storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the smart power supply station work order scheduling method as described in any one of claims 1-4.

10. An electronic terminal, characterized in that: The electronic terminal includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the smart power supply station work order scheduling method as described in any one of claims 1-4.

Images