A substation safe and economic operation regulation and control method and device based on machine learning

Abstract

The present application relates to the technical field of power systems, and particularly relates to a substation safe and economic operation regulation method and device based on machine learning. In the substation safe and economic operation regulation method, the substation transformer predicted load rate and the predicted power supply load of each loop in each time period are predicted according to the date properties of the target date and the hourly meteorological data. When the substation transformer predicted load rate in a certain time period exceeds the rated load rate of the substation transformer, the power-off planning of part of the loops needs to be selected in the time period, early warning planning is realized, so as to avoid the overload operation of the substation and improve the operation safety of the substation. When the substation transformer predicted load rate in a certain time period does not exceed the rated load rate of the substation transformer, the optimal transformer operation combination is selected as the operation plan of the time period, the transformer operation loss is reduced, and economic operation is realized.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method and equipment for safe and economical operation control of substations based on machine learning. Background Technology

[0002] As a critical node in the power system for power transmission, distribution, and voltage transformation, the safety and economy of substation operation directly affect the reliable power supply and operational efficiency of the regional power grid. Currently, substation operation and control strategies largely rely on historical load data, fixed-mode switching operations, or passive responses based on real-time monitoring. Ignoring factors such as future weather conditions and date characteristics (whether it is a holiday) and lacking accurate predictions of short-term load changes, control measures are often delayed, which is detrimental to the safe and economical operation of substations. Summary of the Invention

[0003] The main objective of this invention is to provide a machine learning-based method and equipment for the safe and economical operation control of substations. This method aims to proactively predict the hourly load rate of substation transformers by utilizing future meteorological data and date characteristics, thereby enabling the early construction of safe and economical operation plans to facilitate the safe and economical operation of substations.

[0004] To achieve the above objectives, this invention proposes a machine learning-based method for the safe and economical operation control of substations, comprising the following steps: Obtain the date nature of the target date and the hourly meteorological data corresponding to the target date, wherein the date nature is whether it is a holiday; Input the date characteristics of the target date and the hourly meteorological data corresponding to the target date into the trained substation transformer load rate prediction model to predict the hourly load rate of the substation transformer and obtain the predicted load rate of the substation transformer for each time period. Input the date nature of the target date and the hourly meteorological data corresponding to the target date into the trained substation circuit power supply load prediction model to predict the hourly power supply load of each circuit, and obtain the predicted power supply load of each circuit in each time period; For each time period, the predicted load rate of the substation transformer is determined to determine whether it exceeds the rated load rate of the substation transformer. If it does, some circuits are selected for power outage planning based on the predicted power supply load of each circuit during that time period, so that the sum of the predicted power supply load of the remaining circuits is less than the rated total power supply load of the substation. If not, the optimal transformer operation combination is selected as the operation plan for that time period.

[0005] In the aforementioned machine learning-based substation safe and economical operation control method, the step of selecting some circuits for power outage planning based on the predicted power supply load of each circuit during the time period includes the following steps: The predicted power supply load of each circuit is traversed and combined to obtain multiple power supply combinations. Summing up the loads of each power supply combination yields the sum of the power supply loads for that power supply combination. Select power supply combinations whose sum of power supply loads is less than the substation's rated total power supply load to obtain a candidate set; Select the power supply combination with the largest sum of power supply loads from the candidate set, and select the power supply circuit other than the power supply combination as the proposed power outage circuit.

[0006] In the aforementioned machine learning-based substation safe and economical operation control method, selecting the optimal transformer operation combination as the operation plan for that period includes the following steps: The total load demand is obtained based on the predicted load rate of the substation transformers. By iterating through the combined transformers, a transformer combination is obtained, wherein the total rated capacity of the transformer combination is not less than the total load requirement; In each transformer assembly, the total load demand is allocated to each transformer with the goal of minimizing total losses. Based on the allocation results and the efficiency curves of each transformer, the total loss value of each transformer combination is calculated. The transformer combination with the minimum total loss is selected as the optimal transformer operating combination.

[0007] In the aforementioned machine learning-based substation safe and economical operation control method, the hourly meteorological data includes temperature parameters, relative humidity parameters, solar radiation parameters, and wind speed parameters.

[0008] The aforementioned machine learning-based substation safe and economical operation control method includes establishing a substation transformer load rate prediction model, which comprises the following steps: Acquire historical hourly meteorological data and hourly load rate data of substation transformers based on historical operation; Local historical hourly meteorological data and substation transformer historical hourly load rate data are divided into holidays and working days based on time characteristics to obtain the first holiday data and the first working day data. Data sets were created for the data from the first holiday and the data from the first working day, and the data were cleaned and divided into a first training set and a first test set. Based on the first training set and the first test set, a load rate prediction model for substation transformers under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the substation transformer load rate prediction model.

[0009] The aforementioned machine learning-based substation safe and economical operation control method includes establishing a power supply load prediction model for each circuit of the substation. This power supply load prediction model for each circuit of the substation includes the following steps: Obtain local historical hourly meteorological data and hourly power supply load data of each circuit of the substation during historical operation; Local historical hourly meteorological data and the historical hourly power supply load data of each circuit of the substation are divided into holidays and working days based on time characteristics, to obtain the second holiday data and the second working day data of each circuit of the substation. Data sets were created for the second holiday data and the second working day data of each circuit in the substation, and the data were cleaned. A second training set and a second test set were also created for each circuit in the substation. Based on the second training set and the second test set of each circuit in the substation, the power supply load prediction model of each circuit in the substation under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the power supply load prediction model of each circuit in the substation.

[0010] In the aforementioned machine learning-based substation safe and economical operation control method, the data cleaning includes missing value processing, which is used to process hourly load rate data of the substation transformer's historical operation; the missing value processing includes the following steps: Determine the proportion of missing values ​​within a set time period. If the proportion of missing values ​​is less than the first set threshold, then use the nearest neighbor values ​​or interpolation to fill in the missing values. If the proportion of missing values ​​is greater than the first set threshold but less than the second set threshold, then use data from dates with similar meteorological and temporal characteristics to fill in the missing values. If the proportion of missing values ​​is greater than the second set threshold, then discard the data for the current time period.

[0011] In the above-mentioned machine learning-based substation safe and economical operation control method, the data cleaning also includes outlier processing, which is used to process the hourly load rate data of the substation transformer's historical operation. The outlier handling method is to take the change in value at each sampling time as the amplitude at that sampling time, and determine whether a sudden change has occurred based on the magnitude of the amplitude at that time. If a sudden change has occurred, the value at the previous sampling time is taken as the value at the current sampling time.

[0012] This application also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-mentioned machine learning-based substation safe and economical operation control method.

[0013] This application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described machine learning-based substation safe and economical operation control method.

[0014] The technical solution provided by this invention may include the following beneficial effects: The substation safe and economical operation control method based on machine learning provided in this invention predicts the substation transformer load rate and the predicted power supply load of each circuit for each time period based on the date characteristics of the target date and hourly meteorological data. When the substation transformer load rate exceeds the rated load rate of the substation transformer in a certain time period, power outage planning is required for some circuits during that time period to achieve advance planning and early warning, thereby avoiding substation overload operation and improving substation operational safety. When the substation transformer load rate does not exceed the rated load rate of the substation transformer in a certain time period, the optimal transformer operation combination is selected as the operation plan for that time period to reduce transformer operating losses and achieve economical operation. Therefore, the substation safe and economical operation control method based on machine learning provided in this application can proactively predict the hourly load rate of the substation transformer using future meteorological data and date characteristics, thereby constructing a safe and economical operation plan in advance to facilitate the safe and economical operation of the substation. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a machine learning-based substation safe and economical operation control method according to the present invention. Figure 2 This is a schematic diagram illustrating the process of selecting partial circuits for power outage planning in this invention; Figure 3 A flowchart illustrating the process of selecting the optimal transformer operating combination as the operating plan for this period in this invention; Figure 4 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0019] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0020] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the word "and / or" throughout the text means including three parallel solutions; taking "A and / or B" as an example, it includes solution A, solution B, or a solution that simultaneously satisfies A and B. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0021] The following is combined Figure 1 This invention describes a machine learning-based method for controlling the safe and economical operation of a substation, comprising the following steps: The process involves acquiring the date characteristics of a target date and the corresponding hourly meteorological data. The date characteristics refer to whether it falls on a holiday or public holiday. Specifically, the target date is a future date, such as the next day. Hourly meteorological data refers to meteorological data for each hour; for example, if there are 24 hours in a day, there are 24 time periods corresponding to the meteorological data. The hourly meteorological data corresponding to the target date can be obtained from a meteorological platform. In this application, the meteorological data is the meteorological data corresponding to the area where the substation is located. In a specific embodiment, the hourly meteorological data includes temperature parameters, relative humidity parameters, solar radiation parameters, and wind speed parameters.

[0022] Input the date characteristics of the target date and the hourly meteorological data corresponding to the target date into the trained substation transformer load rate prediction model to predict the hourly load rate of the substation transformer, and obtain the predicted load rate of the substation transformer for each time period; the predicted load rate of the substation transformer refers to the ratio of the total load in the substation to the sum of the rated capacity of all transformers in that time period.

[0023] Input the date nature of the target date and the hourly meteorological data corresponding to the target date into the trained substation circuit power supply load prediction model to predict the hourly power supply load of each circuit, and obtain the predicted power supply load of each circuit in each time period; For each time period, the predicted load rate of the substation transformer is determined to determine whether it exceeds the rated load rate of the substation transformer. If it does, some circuits are selected for power outage planning based on the predicted power supply load of each circuit during that time period, so that the sum of the predicted power supply load of the remaining circuits is less than the rated total power supply load of the substation. If not, the optimal transformer operation combination is selected as the operation plan for that time period.

[0024] The substation safe and economical operation control method based on machine learning provided in this invention predicts the substation transformer load rate and the predicted power supply load of each circuit for each time period based on the date characteristics of the target date and hourly meteorological data. When the substation transformer load rate exceeds the rated load rate of the substation transformer in a certain time period, power outage planning is required for some circuits during that time period to achieve advance planning and early warning, thereby avoiding substation overload operation and improving substation operational safety. When the substation transformer load rate does not exceed the rated load rate of the substation transformer in a certain time period, the optimal transformer operation combination is selected as the operation plan for that time period to reduce transformer operating losses and achieve economical operation. After each time period is determined, the substation transformer operation plan and the power supply outage plan for each circuit for the target date can be generated. Therefore, the substation safe and economical operation control method based on machine learning provided in this application can proactively predict the hourly load rate of the substation transformer using future meteorological data and date characteristics, thereby constructing a safe and economical operation plan in advance.

[0025] Specifically, taking the formulation of the next day's operation and control plan as an example, the first step is to obtain the date characteristics of the next day, i.e., whether the next day is a holiday, and the hourly meteorological data for the next day. The date characteristics and hourly meteorological data for the next day are then input into the substation transformer load rate prediction model and the substation circuit power supply load prediction model to obtain the substation transformer predicted load rate and the predicted power supply load of each circuit for each time period. This yields the substation transformer predicted load rate and the predicted power supply load of each circuit for a 24-hour period. Starting from the substation transformer predicted load rate of the first hour, it is determined whether the substation transformer predicted load rate exceeds the substation transformer rated load rate. If so, some circuits are selected from the predicted power supply load of each circuit in the first hour as the planned outage circuits. If not, the optimal transformer operation combination is selected as the operation plan for that period. This process is repeated for the remaining hourly periods to formulate the operation and control plan for the next day.

[0026] Furthermore, such as Figure 2 As shown, the step of selecting some circuits for power outage planning based on the predicted power supply load of each circuit during this time period includes the following steps: The predicted power supply load of each circuit is traversed and combined to obtain multiple power supply combinations. Summing up the loads of each power supply combination yields the sum of the power supply loads for that power supply combination. Select power supply combinations whose sum of power supply loads is less than the substation's rated total power supply load to obtain a candidate set; Select the power supply combination with the largest sum of power supply loads from the candidate set, and select the power supply circuit other than the power supply combination as the proposed power outage circuit.

[0027] For example, in some specific embodiments, the rated total power supply load of the substation is 1000KW, the predicted power supply load of circuit A is 100KW, the predicted power supply load of circuit B is 200KW, the predicted power supply load of circuit C is 350KW, and the predicted power supply load of circuit D is 400KW. In this case, traversing and combining circuits A, B, C, and D yields 16 possible combinations, as shown in the table below: By selecting combinations other than combination 9, a candidate set is obtained. From the candidate set, combination 13 is selected, and circuit A is designated as the outage circuit. In this way, the impact of the power outage can be reduced while ensuring that the substation is not overloaded.

[0028] Furthermore, such as Figure 3 As shown, selecting the optimal transformer operating combination as the operating plan for this period includes the following steps: The total load demand is obtained based on the predicted load rate of the substation transformers; specifically, the calculation formula is: Total load demand = Sum of the rated capacity of all transformers in the substation × Predicted load rate of the substation transformers.

[0029] By iterating through the combined transformers, a transformer combination is obtained, wherein the total rated capacity of the transformer combination is not less than the total load requirement; In each transformer assembly, the total load demand is allocated to each transformer with the goal of minimizing total losses. Specifically, the optimization model is as follows: Objective function: Min L(Pi); Pi is the load of the i-th transformer; L(Pi) is the loss of the i-th transformer under load Pi; n is the number of transformers in the transformer combination; The constraints are: Pi = Ptotal, and 0 ≤ Pi ≤ Si; where Ptotal is the total load demand, and Si is the rated capacity of the i-th transformer. By solving the optimization model, the total load demand can be allocated to each transformer with the goal of minimizing total losses.

[0030] Based on the allocation results and the efficiency curves of each transformer, the total loss value of each transformer combination is calculated. The optimal transformer operating combination is selected based on the transformer combination with the lowest total loss. This determines how many transformers should be in operation during a given period and how much load each transformer should be allocated. This ensures efficient operation of the substation transformers and reduces operational losses.

[0031] Furthermore, the hourly meteorological data includes temperature parameters, relative humidity parameters, solar radiation parameters, and wind speed parameters.

[0032] Furthermore, this includes establishing a substation transformer load rate prediction model, which comprises the following steps: Acquire historical hourly meteorological data and historical hourly load rate data of substation transformers. Specifically, the historical hourly meteorological data refers to meteorological data for the area where the substation is located, including hourly temperature, relative humidity, solar radiation, and wind speed parameters, and is labeled with the current date. Similarly, the historical hourly load rate data of substation transformers refers to the load rate of the substation transformers for each hour, labeled with the current date. More specifically, acquire one month, one quarter, or half a year's worth of historical hourly meteorological data and historical hourly load rate data of substation transformers as data sources.

[0033] Local historical hourly meteorological data and substation transformer historical hourly load rate data are divided into holidays and workdays based on time characteristics, resulting in first holiday data and first workday data. Specifically, the first holiday data includes the substation transformer historical hourly load rate data for each holiday, as well as the local historical hourly meteorological data for each holiday. The historical hourly load rate data for each holiday is correlated one-to-one with the local historical hourly meteorological data for the holiday. For example, if yesterday was a holiday, the first holiday data includes the substation transformer hourly load rate data for yesterday, as well as yesterday's hourly meteorological data. The hourly load rate data and hourly meteorological data are correlated one-to-one in terms of time; for example, the load rate data at 2 PM yesterday corresponds to the meteorological data at 2 PM yesterday. Similarly, the first workday data includes the substation transformer historical hourly load rate data for each workday, as well as the local historical hourly meteorological data for each workday. The historical hourly load rate data for each workday is correlated one-to-one with the local historical hourly meteorological data for the workday. For example, if the day before yesterday was a working day, the data for the first working day includes the hourly load rate data of the substation transformers operating the day before yesterday, as well as the hourly meteorological data of the day before yesterday. The hourly load rate data and the hourly meteorological data correspond one-to-one in terms of time. For example, the load rate data at 14:00 the day before yesterday corresponds to the meteorological data at 14:00 the day before yesterday.

[0034] Data sets were created for the data from the first holiday and the data from the first working day, and the data were cleaned and divided into a first training set and a first test set; specifically, the data can be divided in a 7:3 or 8:2 ratio.

[0035] Based on the first training set and the first test set, a load rate prediction model for substation transformers under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the substation transformer load rate prediction model.

[0036] Specifically, the machine learning method can be a random forest regression model, an LSTM model, or other reinforcement learning models. For example, in an optional embodiment of this application, a random forest regression model is used for load rate prediction. The random forest regression model is an ensemble learning algorithm that completes classification or regression tasks by constructing multiple decision trees. The algorithm first uses the Bootstrap method to repeatedly randomly sample the original training set (size K), generating multiple training subsets, each containing k samples, allowing for sample duplication. Each decision tree is trained independently based on one subset, and only a subset of features is randomly selected for computation when a node splits. Finally, the random forest obtains the final prediction by majority voting or averaging the outputs of multiple decision trees. Commonly used metrics in model evaluation and error analysis include mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE). Simultaneously, the coefficient of determination R² (R-Square) is used to measure the goodness of fit between the predicted and actual values; the closer R² is to 1, the better the model fit.

[0037] Furthermore, this includes establishing a power supply load prediction model for each circuit of the substation, which includes the following steps: Acquire local historical hourly meteorological data and historical hourly power load data for each circuit of the substation. Specifically, the historical hourly meteorological data refers to meteorological data for the area where the substation is located, including hourly temperature, relative humidity, solar radiation, and wind speed parameters, and is tagged with the current date. Similarly, the historical hourly power load data for each circuit of the substation refers to the power load of each circuit for each hour, tagged with the current date. More specifically, acquire one month, one quarter, or half a year's historical hourly meteorological data and historical hourly power load data for each circuit of the substation as data sources.

[0038] Local historical hourly meteorological data and the historical hourly power supply load data of each circuit in the substation are divided into holidays and weekdays based on time characteristics, resulting in the second holiday data and the second weekday data for each circuit of the substation. Specifically, for each circuit, the second holiday data includes the hourly power supply load data of that circuit on each holiday, as well as the hourly meteorological data for each holiday. The hourly power supply load data for holidays is correlated one-to-one with the hourly meteorological data for the holiday. For example, if yesterday was a holiday, the second holiday data for each circuit includes the hourly power supply load data of that circuit yesterday, as well as the hourly meteorological data for yesterday. The hourly power supply load data and the hourly meteorological data correspond one-to-one in terms of time; for example, the power supply load data of that circuit at 14:00 yesterday corresponds to the meteorological data at 14:00 yesterday.

[0039] Similarly, in each circuit, the second working day data includes the hourly power load data of that circuit on each working day, as well as the hourly meteorological data of that working day. The hourly power load data of the working day is correlated one-to-one with the hourly meteorological data of the working day according to the time. For example, if the day before yesterday was a working day, the second working day data of each circuit includes the hourly power load data of that circuit on the day before yesterday, as well as the hourly meteorological data of the day before yesterday. The hourly power load data and the hourly meteorological data are correlated one-to-one in terms of time, that is, the power load data of that circuit at 14:00 on the day before yesterday corresponds to the meteorological data at 14:00 on the day before yesterday.

[0040] Data sets were created for the second holiday and second working day data of each circuit in the substation, and the data were cleaned. A second training set and a second test set were then established for each circuit in the substation; that is, each circuit in the substation has its own independent second training set and second test set. Specifically, the data can be divided in a 7:3 or 8:2 ratio.

[0041] Based on the second training set and the second test set of each circuit in the substation, a power load prediction model for each circuit under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the power load prediction model for each circuit in the substation. Specifically, the machine learning method can be a random forest regression model, an LSTM model, or other reinforcement learning models. For example, in an optional embodiment of this application, a random forest regression model is used for load rate prediction.

[0042] Furthermore, the data cleaning includes missing value processing, which is used to process hourly load rate data of the substation transformer's historical operation; the missing value processing includes the following steps: Determine the proportion of missing values ​​within a set time period. If the proportion of missing values ​​is less than the first set threshold, then use the nearest neighbor values ​​or interpolation to fill in the missing values. If the proportion of missing values ​​is greater than the first set threshold but less than the second set threshold, then use data from dates with similar meteorological and temporal characteristics to fill in the missing values. If the proportion of missing values ​​is greater than the second set threshold, then discard the data for the current time period.

[0043] For example, the first threshold is set to 0.1, and the second threshold is set to 0.3. Both thresholds can be adjusted according to actual needs. Taking missing hourly load rate data as an example, there are 24 historical hourly load rate data points for a substation transformer in a day. If two data points are missing, the missing percentage is 1 / 12, which is less than the first threshold of 0.1. In this case, nearest neighbor values ​​or interpolation are used to fill the missing values. If four data points are missing, the missing percentage is 1 / 6, which is greater than the first threshold but less than the second threshold. In this case, a date with similar characteristics and meteorological data is selected as a reference, and data from the corresponding time period is used to fill the missing values. If eight data points are missing, the missing percentage is 1 / 3, which is greater than the second threshold. In this case, the data for that date is discarded. By identifying and handling missing values, the accuracy of the data can be improved, thereby increasing the prediction accuracy.

[0044] Furthermore, the data cleaning also includes outlier processing, which is used to process hourly load rate data of the historical operation of substation transformers. The outlier handling method is to take the change in value at each sampling time as the amplitude at that sampling time, and determine whether a sudden change has occurred based on the magnitude of the amplitude at that time. If a sudden change has occurred, the value at the previous sampling time is taken as the value at the current sampling time.

[0045] For example, consider the anomalies in hourly load factor data. Throughout the day, the hourly load factor data (0-23 hours) is 52%, 57%, 61%, 65%, 58%, 63%, 65%, 71%, 73%, 80%, 82%, 84%, 87%, 55%, 79%, 78%, 80%, 72%, 83%, 81%, 77%, 72%, 63%, and 58%. For anomalies in hourly load factor data, assuming a threshold of 20%, the hourly amplitude is calculated sequentially, where the hourly amplitude is the load factor at the current hour minus the load factor at the previous hour. When an hourly amplitude exceeds the threshold, the load factor at the current hour is adjusted to the load factor at the previous hour. In the above sequence, the load factor at 13:00 is an abrupt change, and it is corrected to 87%. By identifying and correcting outliers, the accuracy of the data can be improved, thereby increasing prediction precision.

[0046] like Figure 4 As shown, a third aspect of the present invention also discloses an electronic device 100. Figure 4The illustrated electronic device 100 includes a processor 101 and a memory 102. The processor 101 and memory 102 are connected, for example, via a bus 103. Further, the electronic device 100 may also include a transceiver 104. It should be noted that in practical applications, the transceiver 104 is not limited to one, and the structure of the electronic device 100 does not constitute a limitation on the embodiments of this application. The processor 101 is used in the embodiments of this application to implement a machine learning-based substation safe and economical operation control method. The processor 101 can be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor 101 can also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0047] Bus 103 may include a pathway for transmitting information between the aforementioned components. Bus 103 may be a PCI bus or an EISA bus, etc. Bus 103 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The bus 103 is represented by only one thick line, but this does not mean that there is only one bus 103 or one type of bus 103.

[0048] The memory 102 may be a ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0049] The memory 102 stores the application code that executes the scheme of this application, and its execution is controlled by the processor 101. The processor 101 executes the application code stored in the memory 102 to implement... Figure 1 The illustrated embodiment provides the action of a machine learning-based substation safe and economical operation control method.

[0050] The third aspect of the present invention discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the machine learning-based substation safe and economical operation control method described in the first aspect of the present invention.

[0051] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made under the concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A substation safety and economic operation control method based on machine learning, characterized in that: Includes the following steps: Obtain the date nature of the target date and the hourly meteorological data corresponding to the target date, wherein the date nature is whether it is a holiday; Input the date characteristics of the target date and the hourly meteorological data corresponding to the target date into the trained substation transformer load rate prediction model to predict the hourly load rate of the substation transformer and obtain the predicted load rate of the substation transformer for each time period. Input the date nature of the target date and the hourly meteorological data corresponding to the target date into the trained substation circuit power supply load prediction model to predict the hourly power supply load of each circuit, and obtain the predicted power supply load of each circuit in each time period; For each time period, the predicted load rate of the substation transformer is determined to see if it exceeds the rated load rate of the substation transformer. If it does, then some circuits are selected for power outage planning based on the predicted power supply load of each circuit during that time period. If not, then the optimal transformer operation combination is selected as the operation plan for that time period.

2. The substation safety and economic operation control method based on machine learning according to claim 1, characterized in that: The process of selecting some circuits for power outage planning based on the predicted power supply load of each circuit during this time period includes the following steps: The predicted power supply load of each circuit is traversed and combined to obtain multiple power supply combinations; Summing up the loads of each power supply combination yields the sum of the power supply loads for that power supply combination. Select power supply combinations whose sum of power supply loads is less than the substation's rated total power supply load to obtain a candidate set; Select the power supply combination with the largest sum of power supply loads from the candidate set, and select the power supply circuit other than the power supply combination as the proposed power outage circuit.

3. The substation safety and economic operation control method based on machine learning according to claim 1, characterized in that: Selecting the optimal transformer operating combination as the operating plan for this period includes the following steps: The total load demand is obtained based on the predicted load rate of the substation transformers. By iterating through the combined transformers, a transformer combination is obtained, wherein the total rated capacity of the transformer combination is not less than the total load requirement; In each transformer assembly, the total load demand is allocated to each transformer with the goal of minimizing total losses. Based on the allocation results and the efficiency curves of each transformer, the total loss value of each transformer combination is calculated. The transformer combination with the minimum total loss is selected as the optimal transformer operating combination.

4. The substation safety and economic operation control method based on machine learning according to claim 1, characterized in that: The hourly meteorological data includes temperature parameters, relative humidity parameters, solar radiation parameters, and wind speed parameters.

5. The substation safety and economic operation control method based on machine learning according to claim 1, characterized in that: This includes establishing a substation transformer load rate prediction model, which comprises the following steps: Acquire historical hourly meteorological data and hourly load rate data of substation transformers based on historical operation; Local historical hourly meteorological data and substation transformer historical hourly load rate data are divided into holidays and working days based on time characteristics to obtain the first holiday data and the first working day data. Data sets were created for the data from the first holiday and the data from the first working day, and the data were cleaned and divided into a first training set and a first test set. Based on the first training set and the first test set, a load rate prediction model for substation transformers under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the substation transformer load rate prediction model.

6. The substation safety and economic operation control method based on machine learning according to claim 5, characterized in that: This includes establishing a power supply load prediction model for each circuit of the substation, which includes the following steps: Obtain local historical hourly meteorological data and hourly power supply load data of each circuit of the substation during historical operation; Local historical hourly meteorological data and the historical hourly power supply load data of each circuit of the substation are divided into holidays and working days based on time characteristics, to obtain the second holiday data and the second working day data of each circuit of the substation. Data sets were created for the second holiday data and the second working day data of each circuit in the substation, and the data were cleaned. A second training set and a second test set were also created for each circuit in the substation. Based on the second training set and the second test set of each circuit in the substation, the power supply load prediction model of each circuit in the substation under different time characteristics during holidays and weekdays is obtained through machine learning, thus obtaining the power supply load prediction model of each circuit in the substation.

7. In the substation safe and economical operation control method based on machine learning according to claim 5, the characteristic is: The data cleaning process includes missing value processing, which is used to process the hourly load rate data of the substation transformer's historical operation. The missing value processing includes the following steps: Determine the proportion of missing values ​​within a set time period. If the proportion of missing values ​​is less than the first set threshold, then use the nearest neighbor values ​​or interpolation to fill in the missing values. If the proportion of missing values ​​is greater than the first set threshold but less than the second set threshold, then use data from dates with similar meteorological and temporal characteristics to fill in the missing values. If the proportion of missing values ​​is greater than the second set threshold, then discard the data for the current time period.

8. In the substation safe and economical operation control method based on machine learning according to claim 7, the characteristic is: The data cleaning process also includes outlier processing, which is used to process the hourly load rate data of the substation transformer's historical operation. The outlier handling method is to take the change in value at each sampling time as the amplitude at that sampling time, and determine whether a sudden change has occurred based on the magnitude of the amplitude at that time. If a sudden change has occurred, the value at the previous sampling time is taken as the value at the current sampling time.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images