Method and system for generating transformer substation full-stop risk plan based on multi-time scale model

Abstract

The invention provides a multi-time-scale model-based transformer substation full-stop risk plan generation method and a multi-time-scale model-based transformer substation full-stop risk plan generation system. In line load prediction, a multi-time-scale collaborative optimization model with two time scales of medium and long term and short term is introduced, and prediction quantities of in-station line load during maintenance and on-demand line load during maintenance are solved; establishing a power loss probability optimization model by taking the minimum difference value between the short-term scale predictive quantity and the historical data in the past one year as an objective function, and solving to obtain the predictive quantity of the power loss probability of the line fault in the maintenance period and the power loss probability of the important users and the sensitive users in the maintenance period; and by taking the obtained pre-measured quantity as input, taking an optimization variable as an optimal transfer path, and taking an objective function as a minimum line load and a minimum power loss probability, establishing an optimization model of a load transfer scheme, and solving to obtain an optimal load transfer scheme. According to the method, the workload of designing the line transfer scheme is reduced, the accuracy of the load transfer scheme is improved, and the method has reliability, reasonability and practicability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of substation risk preplan, and particularly relates to a substation full outage risk preplan generation method and system based on a multi-time scale model. BACKGROUND

[0002] In the maintenance work of the power grid substation, special operation modes of the substation are often involved, such as "one line with two substations" and "single main transformer operation". Under the special operation mode, the outage risk of the substation is greatly increased, and a substation full outage risk preplan needs to be formulated to improve the emergency handling capability of the substation full outage risk. The substation full outage risk preplan mainly includes a main grid risk preplan and a distribution network risk preplan. In the short and medium term, the main grid topology changes are not obvious, while the distribution network topology changes are obvious, seasonal load is large, and the output of new energy such as wind power and photovoltaic is random, which leads to great difficulty in formulating a load transfer scheme of the distribution network risk preplan and low reliability. At present, when formulating the load transfer scheme, only the historical load characteristics and weather conditions can be used to roughly estimate the line load, and a load transfer scheme is formulated. However, with the large access of new energy and the rapid transformation of the power grid network, the randomness of the line load is greatly improved, and the reliability of the risk preplan formulated in advance is greatly limited.

[0003] At present, in the actual work application, the process of formulating the load transfer scheme of the substation full outage risk preplan is as shown in Figure 1 : first, obtaining prediction data: weather conditions during maintenance, load conditions in the past two weeks, load conditions in the same period last year, monthly / seasonal load conditions, adverse weather or large wind power and photovoltaic weather, and closing current value of the transfer path; second, estimating the line load conditions and the transfer line load conditions during the maintenance period according to the weather conditions during the maintenance, the load conditions in the past two weeks, the load conditions in the same period last year, and the monthly / seasonal load conditions; estimating the outage risk and the power protection conditions of important users / sensitive users when the line fails during the maintenance period according to the adverse weather or large wind power and photovoltaic conditions; then, combining the load transfer contact path of the power grid in the future, calculating the load of the load transfer line after the transfer through each transfer path, and estimating the outage risk of the load transfer line; finally, combining the calculation model to select the optimal load transfer scheme.

[0004] The existing transfer scheme formulation process is to obtain the prediction model and the optimal model according to the human estimation, which has no reliability, and when obtaining the objective function of the optimal scheme, there is no uniform rule, and the influence of each prediction data cannot be reflected in the optimal model. SUMMARY

[0005] The purpose of this invention is to provide a method and system for generating substation total outage risk contingency plans based on a multi-time-scale model. By introducing a multi-time-scale collaborative model, the reliability and accuracy of the prediction model are improved. At the same time, by combining the optimal model, the human evaluation model is transformed into the calculation of the objective function, thereby improving the accuracy of the load transfer scheme and avoiding random errors caused by human estimation.

[0006] The technical solution to achieve the purpose of this invention is: a method for generating a substation total outage risk contingency plan based on a multi-time-scale model, comprising the following steps:

[0007] Step 1: In line load forecasting, a multi-timescale collaborative optimization model with both medium- and long-term and short-term timescales is introduced, as detailed below:

[0008] The medium- and long-term forecasts use a long short-term memory network as the forecast model. The inputs are the load conditions of the same period last year, the weather conditions during the maintenance period, severe weather or large-scale photovoltaic and wind power generation, and the annual load growth of the substation. The forecast model outputs medium- and long-term forecasts, including the load of the substation's lines during the maintenance period and the load of the lines supplied on behalf of the substation during the maintenance period.

[0009] Short-term forecasts use a long short-term memory network as the forecasting model, with the load conditions of the past two weeks, the weather conditions during the maintenance period, and severe weather or large-scale photovoltaic and wind power generation as inputs. The forecasting model outputs short-term forecasts, including the load of the station's lines during the maintenance period and the load of the lines supplied by others during the maintenance period.

[0010] Using medium- and long-term forecasts and short-term forecasts as inputs, the optimization variables are the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period. The objective function is to minimize the difference between the medium- and long-term forecasts and the short-term forecasts. A multi-time-scale collaborative optimization model is established and solved to obtain the forecasts of the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period.

[0011] Step 2: Predict the probability of power outage, as follows:

[0012] Using short-term severe weather or large-scale photovoltaic and wind power generation and the loop current of the transfer path as input, and using a long short-term memory network as the prediction model, the prediction model outputs short-term predicted quantities, including the probability of power loss due to line faults during maintenance and the probability of power loss for important and sensitive users during maintenance.

[0013] Using short-term forecasts and historical data from the past year as inputs, the optimization variables are the probability of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance. The objective function is to minimize the difference between the short-term forecasts and historical data from the past year. A power outage probability optimization model is established and solved to obtain the predicted probabilities of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance.

[0014] Step 3: Using the predicted load of the station's lines during the maintenance period and the predicted load of the lines supplied during the maintenance period obtained in Step 1, as well as the predicted probability of power outage due to line faults during the maintenance period and the probability of power outage for important and sensitive users during the maintenance period obtained in Step 2, as input variables, the optimal transfer path is the optimization variable, and the objective function is to minimize the line load and the probability of power outage. An optimization model for the load transfer scheme is established and solved to obtain the optimal load transfer scheme.

[0015] Further, in step 1, the objective function of the multi-timescale collaborative optimization model is:

[0016]

[0017] The constraints are:

[0018]

[0019]

[0020]

[0021] in, This indicates the medium- to long-term forecast of the line load within the station during the maintenance period. This indicates the medium- to long-term forecast of the load on the lines supplied during the maintenance period. This indicates a short-term forecast of the line load within the station during the maintenance period. This indicates a short-term forecast of the load on the lines supplied during the maintenance period. The objective function is to minimize the total difference between the medium- and long-term forecasts and the short-term forecasts of the line load during the maintenance period.

[0022] The first constraint states that the combined load of a line within a station and its connected external lines at any given time shall not exceed the line's rated capacity. Rated current, Rated voltage, This refers to the rated capacity of the line.

[0023] The second constraint states that the total load of all lines within the station at any given time does not exceed [a certain limit]. ; The capacity of the substation under maintenance. This refers to the load factor.

[0024] The third constraint states that after the load of a certain line within the station is transferred to an external line, the load of the substation where the external connecting line is located shall not exceed [the specified limit]. ; This indicates the load of the substation where the external connection line is located at a certain moment. This indicates the capacity of the substation where the external connecting line is located. This indicates the load of the substation where the external connecting line is located after the line load is transferred to the external line.

[0025] Furthermore, in both medium- and long-term forecasts and short-term forecasts, the long short-term memory network used in the forecasting model is directly trained and used in MATLAB's long short-term memory network model.

[0026] Furthermore, the specific input quantities for the medium- and long-term forecasts are as follows:

[0027] Load conditions for the same period last year: During the substation maintenance period, go back one year and take the same time period to obtain the maximum load value of each line in the substation. ;

[0028] Weather conditions during maintenance: operating temperature T and weather H, H is 1 for sunny days and 0 for cloudy days, T is in degrees Celsius;

[0029] Severe weather or large-scale solar and wind power generation: This indicates the number of days with severe weather in the past year. This indicates the probability of severe weather occurring in the past year. Full-load operation of photovoltaic and wind power is considered a major success. This indicates the number of days in the past year when wind and solar power generation was at its peak. This indicates the probability of a major wind and solar power generation event occurring in the past year. ;

[0030] Annual load growth of substations: Load growth includes air conditioning load, industrial load and natural load growth.

[0031] Furthermore, the specific inputs for short-term scale prediction in step 2 are as follows:

[0032] Severe weather may lead to a surge in solar and wind power generation: This indicates the number of days with severe weather in the past two weeks. This indicates the probability of severe weather occurring in the past two weeks. Full-load operation of photovoltaic and wind power is considered a major success. This indicates the number of days with high wind and solar power generation in the past two weeks. This indicates the probability of a major wind and solar power generation event occurring in the past two weeks. ;

[0033] Supply path loop current: , This indicates the current amplitude at the instant the power supply loop closes. , These are the current phasors for the station's internal lines and the lines supplied on behalf of others, respectively. The current at the moment of switching the supply loop does not exceed the protection setting of the tie switch.

[0034] Furthermore, in step 2, an optimization model for the probability of power loss is established, with the objective function being:

[0035]

[0036] in, This represents the difference between short-term forecasts and historical data from the past year. Short-term prediction of the probability of power outage due to line faults during maintenance. Short-term prediction of the probability of power outage for important and sensitive users during maintenance; , These represent the probability of power outages due to line faults over the past year, as well as the probability of power outages for important and sensitive users.

[0037] Furthermore, the optimization model for the load transfer scheme in step 3 is as follows:

[0038] The input quantity is: the load of line i within the station during the maintenance period. During the maintenance period, the load of the substitute power supply line j connected to line i will be reduced. The probability of power failure during maintenance of the temporary power supply line. Probability of power outage for important and sensitive users during maintenance ;

[0039] The objective function is:

[0040]

[0041]

[0042] Where i represents the station line number, and j represents the line number that is connected to line i;

[0043] The constraints are:

[0044]

[0045]

[0046]

[0047]

[0048]

[0049] Take the time when the above objective function is most satisfied Sequence, each There is only one optimal route. By combining these methods, the optimal load transfer scheme can be obtained.

[0050] Furthermore, the multi-timescale collaborative optimization model, the power outage probability optimization model, and the load transfer scheme optimization model are all solved using MATLAB CPLEX.

[0051] A substation outage risk contingency plan generation system based on a multi-time-scale model is provided. This system implements the aforementioned substation outage risk contingency plan generation method. The system includes a line load prediction module, a power outage probability prediction module, and a load transfer scheme generation module.

[0052] The line load forecasting module incorporates a multi-timescale collaborative optimization model with both medium- and long-term and short-term timescales, as detailed below:

[0053] The medium- and long-term forecasts use a long short-term memory network as the forecast model. The inputs are the load conditions of the same period last year, the weather conditions during the maintenance period, severe weather or large-scale photovoltaic and wind power generation, and the annual load growth of the substation. The forecast model outputs medium- and long-term forecasts, including the load of the substation's lines during the maintenance period and the load of the lines supplied on behalf of the substation during the maintenance period.

[0054] Short-term forecasts use a long short-term memory network as the forecasting model, with the load conditions of the past two weeks, the weather conditions during the maintenance period, and severe weather or large-scale photovoltaic and wind power generation as inputs. The forecasting model outputs short-term forecasts, including the load of the station's lines during the maintenance period and the load of the lines supplied by others during the maintenance period.

[0055] Using medium- and long-term forecasts and short-term forecasts as inputs, the optimization variables are the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period. The objective function is to minimize the difference between the medium- and long-term forecasts and the short-term forecasts. A multi-time-scale collaborative optimization model is established and solved to obtain the forecasts of the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period.

[0056] The power failure probability prediction module is as follows:

[0057] Using short-term severe weather or large-scale photovoltaic and wind power generation and the loop current of the transfer path as input, and using a long short-term memory network as the prediction model, the prediction model outputs short-term predicted quantities, including the probability of power loss due to line faults during maintenance and the probability of power loss for important and sensitive users during maintenance.

[0058] Using short-term forecasts and historical data from the past year as inputs, the optimization variables are the probability of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance. The objective function is to minimize the difference between the short-term forecasts and historical data from the past year. A power outage probability optimization model is established and solved to obtain the predicted probabilities of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance.

[0059] The load transfer scheme generation module takes the predicted load of the substation lines during the maintenance period and the predicted load of the lines supplied during the maintenance period obtained by the line load prediction module, as well as the predicted probability of line faults and the probability of power outages for important and sensitive users during the maintenance period obtained by the power outage probability prediction module, as inputs. The optimization variable is the optimal transfer path, and the objective function is to minimize the line load and the probability of power outage. An optimization model for the load transfer scheme is established and solved to obtain the optimal load transfer scheme.

[0060] A mobile terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the method for generating a substation outage risk contingency plan.

[0061] Compared with the prior art, the significant advantages of this invention are:

[0062] (1) Multi-timescale prediction from medium- and long-term and short-term, and then optimization model solution, is more scientific and reasonable than manual planning, greatly reduces the workload of manually designing line transfer schemes, and reasonably considers various influencing factors in the model, thus improving the accuracy of load transfer schemes.

[0063] (2) In the process of predicting the risk of power outage of the line, the probability of power outage of the line in the past year is introduced. The power supply reliability of the line can be mined from historical data, which greatly reduces the risk of power outage of the line after the transfer of power supply.

[0064] (3) The prediction model and the optimization solution model combine multiple dimensions of influencing factors such as medium and long-term data, short-term data, and historical data, and are covered in the solution process, making the risk plan more reliable, reasonable and feasible. Attached Figure Description

[0065] Figure 1 This is a flowchart for developing load transfer schemes in existing substation outage risk contingency plans.

[0066] Figure 2 This is a flowchart of the method for generating a substation total outage risk contingency plan based on a multi-timescale model, according to the present invention. Detailed Implementation

[0067] It is readily understood that, based on the technical solution of this invention, those skilled in the art can conceive of various embodiments of this invention without altering its essential spirit. Therefore, the following specific embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of this invention or as limitations or restrictions on its technical solution.

[0068] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0069] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0070] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0071] Combination Figure 2 This invention provides a method for generating a substation total outage risk contingency plan based on a multi-time-scale model, comprising the following steps:

[0072] Step 1: In line load forecasting, a multi-timescale collaborative optimization model with both medium- and long-term and short-term timescales is introduced, as detailed below:

[0073] The medium- and long-term forecasts use a long short-term memory network as the forecast model. The inputs are the load conditions of the same period last year, the weather conditions during the maintenance period, severe weather or large-scale photovoltaic and wind power generation, and the annual load growth of the substation. The forecast model outputs medium- and long-term forecasts, including the load of the substation's lines during the maintenance period and the load of the lines supplied on behalf of the substation during the maintenance period.

[0074] Short-term forecasts use a long short-term memory network as the forecasting model, with the load conditions of the past two weeks, the weather conditions during the maintenance period, and severe weather or large-scale photovoltaic and wind power generation as inputs. The forecasting model outputs short-term forecasts, including the load of the station's lines during the maintenance period and the load of the lines supplied by others during the maintenance period.

[0075] Using medium- and long-term forecasts and short-term forecasts as inputs, the optimization variables are the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period. The objective function is to minimize the difference between the medium- and long-term forecasts and the short-term forecasts. A multi-time-scale collaborative optimization model is established and solved to obtain the forecasts of the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period.

[0076] Step 2: Predict the probability of power outage, as follows:

[0077] Using short-term severe weather or large-scale photovoltaic and wind power generation and the loop current of the transfer path as input, and using a long short-term memory network as the prediction model, the prediction model outputs short-term predicted quantities, including the probability of power loss due to line faults during maintenance and the probability of power loss for important and sensitive users during maintenance.

[0078] Using short-term forecasts and historical data from the past year as inputs, the optimization variables are the probability of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance. The objective function is to minimize the difference between the short-term forecasts and historical data from the past year. A power outage probability optimization model is established and solved to obtain the predicted probabilities of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance.

[0079] Step 3: Using the predicted load of the station's lines during the maintenance period and the predicted load of the lines supplied during the maintenance period obtained in Step 1, as well as the predicted probability of power outage due to line faults during the maintenance period and the probability of power outage for important and sensitive users during the maintenance period obtained in Step 2, as input variables, the optimal transfer path is the optimization variable, and the objective function is to minimize the line load and the probability of power outage. An optimization model for the load transfer scheme is established and solved to obtain the optimal load transfer scheme.

[0080] As a specific example, in step 1, the objective function of the multi-timescale collaborative optimization model is:

[0081]

[0082] The constraints are:

[0083]

[0084]

[0085]

[0086] in, This indicates the medium- to long-term forecast of the line load within the station during the maintenance period. This indicates the medium- to long-term forecast of the load on the lines supplied during the maintenance period. This indicates a short-term forecast of the line load within the station during the maintenance period. This indicates a short-term forecast of the load on the lines supplied during the maintenance period. The objective function is to minimize the total difference between the medium- and long-term forecasts and the short-term forecasts of the line load during the maintenance period.

[0087] The first constraint states that the combined load of a line within a station and its connected external lines at any given time shall not exceed the line's rated capacity. The rated current is typically 480A. This is the rated voltage, typically taken as 10kV; The rated capacity of the line is typically taken as 8400kW;

[0088] The second constraint states that the total load of all lines within the station at any given time does not exceed [a certain limit]. ; The capacity of the substation under maintenance. This is the load factor, typically taken as 0.8;

[0089] The third constraint states that after the load of a certain line within the station is transferred to an external line, the load of the substation where the external connecting line is located shall not exceed [the specified limit]. ; This indicates the load of the substation where the external connection line is located at a certain moment. This indicates the capacity of the substation where the external connecting line is located. This indicates the load of the substation where the external connecting line is located after the line load is transferred to the external line. and The value can be obtained from D5000.

[0090] As a specific example, in both medium- and long-term forecasting and short-term forecasting, the prediction model uses a long short-term memory network, which is directly trained and used in MATLAB's long short-term memory network model.

[0091] As a specific example, the model training and sample data, specifically the training data, are as follows:

[0092] The specific inputs for the medium- and long-term forecasts are as follows:

[0093] Load conditions for the same period last year: During the substation maintenance period, go back one year and take the same time period to obtain the maximum load value of each line in the substation. This data can be obtained from the D5000;

[0094] Weather conditions during maintenance: operating temperature T and weather H, H is 1 for sunny days and 0 for cloudy days, T is in degrees Celsius;

[0095] Severe weather or large-scale solar and wind power generation: This indicates the number of days with severe weather in the past year. This indicates the probability of severe weather occurring in the past year. Full-load operation of photovoltaic and wind power is considered a major success. This indicates the number of days in the past year when wind and solar power generation was at its peak. This indicates the probability of a major wind and solar power generation event occurring in the past year. ;

[0096] Substation annual load growth: Load growth includes air conditioning load, industrial load, and natural load growth. Air conditioning load is seasonal; the difference between summer / winter load and autumn / winter line load can be used to roughly estimate air conditioning load (excluding industrial load on the lines). Natural load growth is calculated at 5%, and industrial load growth can be obtained from the details of newly registered users in the power grid marketing system.

[0097] Training data: Load conditions, weather conditions, severe weather or peak solar and wind power generation on a typical day during spring, summer, autumn, and winter of the past year, along with load growth from last year to this year, were selected as training input data. The actual load for this year was used as the training output data. The Long Short-Term Memory (LSTM) network module in MTLab was used for training to obtain the training model.

[0098] Using the same period last year's load, weather conditions during the maintenance period, severe weather or large-scale photovoltaic and wind power generation, and the annual load growth of the substation as the input data, the forecast data for the substation maintenance period on a medium to long term can be obtained.

[0099] The short-term forecasting model is the same as the medium- and long-term forecasting model. The short-term forecasting model uses the load data from the past two weeks.

[0100] As a specific example, the input for short-term scale prediction in step 2 is as follows:

[0101] Severe weather may lead to a surge in solar and wind power generation: This indicates the number of days with severe weather in the past two weeks. This indicates the probability of severe weather occurring in the past two weeks. Full-load operation of photovoltaic and wind power is considered a major success. This indicates the number of days with high wind and solar power generation in the past two weeks. This indicates the probability of a major wind and solar power generation event occurring in the past two weeks. ;

[0102] Supply path loop current: , This indicates the current amplitude at the instant the power supply loop closes. , These are the current phasors for the station's internal lines and the lines supplied by others. The instantaneous current when the loop is closed is too large, but it must meet the protection setting of the tie switch not exceeding 600A. The protection setting of 600A is the requirement of the State Grid distribution network line switch overcurrent protection setting.

[0103] As a specific example, the objective function for establishing the power loss probability optimization model in step 2 is:

[0104]

[0105] in, This represents the difference between short-term forecasts and historical data from the past year. Short-term prediction of the probability of power outage due to line faults during maintenance. Short-term prediction of the probability of power outage for important and sensitive users during maintenance; , These figures represent the probability of power outages due to line faults over the past year, as well as the probability of power outages for important and sensitive users. This data can be obtained from the power grid OMS system.

[0106] As a specific example, the optimization model for the load transfer scheme in step 5 is as follows:

[0107] The input quantity is: the load of line i within the station during the maintenance period. During the maintenance period, the load of the substitute power supply line j connected to line i will be reduced. The probability of power failure during maintenance of the temporary power supply line. Probability of power outage for important and sensitive users during maintenance ;

[0108] The objective function is:

[0109]

[0110]

[0111] Where i represents the station line number, and j represents the line number that is connected to line i;

[0112] The constraints are:

[0113]

[0114]

[0115]

[0116]

[0117]

[0118] This optimization model is a multi-objective optimization model. MATLAB CPLEX is used to solve this model. The two objective functions are equally important, meaning they have the same weight. The optimal solution is chosen when both objective functions are satisfied. Sequence, each There is only one optimal route. By combining these methods, the optimal load transfer scheme can be obtained.

[0119] As a specific example, the multi-timescale collaborative optimization model, the power outage probability optimization model, and the load transfer scheme optimization model are all solved using MATLAB CPLEX.

[0120] This invention also provides a substation outage risk contingency plan generation system based on a multi-timescale model. This system is used to implement the aforementioned substation outage risk contingency plan generation method. The system includes a line load prediction module, a power outage probability prediction module, and a load transfer scheme generation module.

[0121] The line load forecasting module incorporates a multi-timescale collaborative optimization model with both medium- and long-term and short-term timescales, as detailed below:

[0122] The medium- and long-term forecasts use a long short-term memory network as the forecast model. The inputs are the load conditions of the same period last year, the weather conditions during the maintenance period, severe weather or large-scale photovoltaic and wind power generation, and the annual load growth of the substation. The forecast model outputs medium- and long-term forecasts, including the load of the substation's lines during the maintenance period and the load of the lines supplied on behalf of the substation during the maintenance period.

[0123] Short-term forecasts use a long short-term memory network as the forecasting model, with the load conditions of the past two weeks, the weather conditions during the maintenance period, and severe weather or large-scale photovoltaic and wind power generation as inputs. The forecasting model outputs short-term forecasts, including the load of the station's lines during the maintenance period and the load of the lines supplied by others during the maintenance period.

[0124] Using medium- and long-term forecasts and short-term forecasts as inputs, the optimization variables are the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period. The objective function is to minimize the difference between the medium- and long-term forecasts and the short-term forecasts. A multi-time-scale collaborative optimization model is established and solved to obtain the forecasts of the load on the substation lines during the maintenance period and the load on the lines supplied on behalf of others during the maintenance period.

[0125] The power failure probability prediction module is as follows:

[0126] Using short-term severe weather or large-scale photovoltaic and wind power generation and the loop current of the transfer path as input, and using a long short-term memory network as the prediction model, the prediction model outputs short-term predicted quantities, including the probability of power loss due to line faults during maintenance and the probability of power loss for important and sensitive users during maintenance.

[0127] Using short-term forecasts and historical data from the past year as inputs, the optimization variables are the probability of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance. The objective function is to minimize the difference between the short-term forecasts and historical data from the past year. A power outage probability optimization model is established and solved to obtain the predicted probabilities of power outage due to line faults during maintenance and the probability of power outage for important and sensitive users during maintenance.

[0128] The load transfer scheme generation module takes the predicted load of the substation lines during the maintenance period and the predicted load of the lines supplied during the maintenance period obtained by the line load prediction module, as well as the predicted probability of line faults and the probability of power outages for important and sensitive users during the maintenance period obtained by the power outage probability prediction module, as inputs. The optimization variable is the optimal transfer path, and the objective function is to minimize the line load and the probability of power outage. An optimization model for the load transfer scheme is established and solved to obtain the optimal load transfer scheme.

[0129] The present invention also provides a mobile terminal, 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 substation total outage risk contingency plan generation method.

[0130] In summary, this invention employs multi-timescale predictions across medium- to long-term and short-term periods, followed by optimization model solving. This approach is more scientific and rational than manual planning, significantly reducing the workload of manually designing power transfer schemes. Furthermore, it incorporates various influencing factors into the model, improving the accuracy of the load transfer scheme. In addition, the invention introduces the power outage probability of the line over the past year during the prediction of line power loss risk, allowing for the extraction of power supply reliability from historical data and greatly reducing the risk of power loss after transfer. The prediction and optimization models combine influencing factors from multiple dimensions, including medium- to long-term data, short-term data, and historical data, and these factors are integrated into the solution process, making the risk mitigation plan more reliable, reasonable, and feasible.

[0131] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for generating a substation total outage risk preplan based on a multi-time scale model, characterized in that, The method comprises the following steps: Step 1, in line load prediction, a multi-time scale collaborative optimization model of medium and long term and short term time scales is introduced, and the specific steps are as follows: Medium and long term prediction, taking the long short term memory network as a prediction model, taking the load situation of the same period last year, the weather condition during the maintenance period, the adverse weather or large generation of photovoltaic and wind power and the annual load growth of the substation as input quantities, outputting the medium and long term prediction quantity through the prediction model, including the line load in the station during the maintenance period and the line load of the replacement supply during the maintenance period; Short term prediction, taking the long short term memory network as a prediction model, taking the load situation in the past two weeks, the weather condition during the maintenance period and the adverse weather or large generation of photovoltaic and wind power as input quantities, outputting the short term prediction quantity through the prediction model, including the line load in the station during the maintenance period and the line load of the replacement supply during the maintenance period; Taking the medium and long term prediction quantity and the short term prediction quantity as input quantities, taking the line load in the station during the maintenance period and the line load of the replacement supply during the maintenance period as optimization variables, and taking the minimum difference between the medium and long term prediction and the short term prediction as a target function, a multi-time scale collaborative optimization model is established, and the line load in the station during the maintenance period and the line load of the replacement supply during the maintenance period are obtained through solving; Step 2, loss of power probability prediction, and the specific steps are as follows: Taking the adverse weather or large generation of photovoltaic and wind power of the short term scale and the loop current of the replacement supply path as input quantities, taking the long short term memory network as a prediction model, outputting the short term scale prediction quantity through the prediction model, including the line fault loss of power probability during the maintenance period and the loss of power probability of important users and sensitive users during the maintenance period; Taking the short term scale prediction quantity and the historical data of the past year as input quantities, taking the line fault loss of power probability during the maintenance period and the loss of power probability of important users and sensitive users during the maintenance period as optimization variables, and taking the minimum difference between the short term scale prediction and the historical data of the past year as a target function, a loss of power probability optimization model is established, and the prediction quantity of the line fault loss of power probability during the maintenance period and the loss of power probability of important users and sensitive users during the maintenance period is obtained through solving; Step 3, taking the prediction quantity of the line load in the station during the maintenance period and the line load of the replacement supply during the maintenance period obtained in step 1 and the prediction quantity of the line fault loss of power probability during the maintenance period and the loss of power probability of important users and sensitive users during the maintenance period obtained in step 2 as input quantities, taking the optimal replacement supply path as an optimization variable, and taking the minimum line load and the minimum loss of power probability as a target function, an optimization model of the load replacement supply scheme is established, and the optimal load replacement supply scheme is obtained through solving.

2. The method of claim 1, wherein, In step 1, the target function of the multi-time scale collaborative optimization model is: ； The constraint condition is: ； ； ； wherein, represents a medium and long term prediction of the in-station line load during the overhaul period, represents a medium and long term prediction of the substitute line load during the overhaul period; represents a short term prediction of the in-station line load during the overhaul period, represents a short term prediction of the substitute line load during the overhaul period; is the total difference value of the medium and long term prediction and the short term prediction of the line load during the overhaul period, and the objective function is the minimum total difference value. The first constraint: the load of a certain station line and its associated off-station line at any time does not exceed the rated capacity of the line, for rated current, for rated voltage, for line rated capacity; 2nd constraint: the total load of all lines in the station at any time does not exceed ; the capacity of the substation under repair, is the load factor; The third constraint: represents the load of a certain line in the station transferred to the outside line after the load of the outside line does not exceed ; represents the load of the substation where the outside line is located at a certain time, represents the capacity of the substation where the outside line is located, represents the load of the substation where the outside line is located after the load of the line is transferred to the outside line. 3.The substation total outage risk preplan generation method based on a multi-time scale model according to claim 1, characterized in that, In the medium and long term prediction and the short term prediction, the long short term memory network used in the prediction model directly calls the long short term memory network model of MATLAB for training and prediction.

4. The method of claim 1, wherein, The input quantity of the medium and long term prediction is specifically as follows: Last year's load conditions: during the substation maintenance work, one year ago, take the same time period, get the maximum load value of each line in the station ; The weather condition during the maintenance period: temperature T and weather H are used, the sunny day H is 1, the non-sunny day H is 0, and the unit of T is Celsius; Bad weather or high production of photovoltaic and wind power: Number of days with bad weather in the past year, Probability of bad weather in the past year, High production of photovoltaic and wind power is considered high when the production is above the average production of the past year, Number of days with high production of wind and photovoltaic power in the past year, Probability of high production of wind and photovoltaic power in the past year, ; The annual load growth of the substation: the load growth includes the air conditioning load, the industrial load and the natural growth load.

5. The method of claim 1, wherein, The input quantity of the short term scale prediction in step 2 is specifically as follows: Bad weather or high wind power production: Number of days with bad weather in the past 2 weeks, Probability of bad weather in the past 2 weeks, High wind power production is considered as high if the wind power production is above the average wind power production of the past 2 weeks, Number of days with high wind power production in the past 2 weeks, Probability of high wind power production in the past 2 weeks, ; The transfer path loop current: , The current amplitude at the transfer loop closing moment is represented by I, , The current phasor of the station line and the substitute line respectively, and the current at the transfer loop closing moment does not exceed the protection setting value of the tie switch.

6. The method of claim 1, wherein, The target function of the loss of power probability optimization model in step 2 is: ； wherein, represents the short-term scale prediction of the difference between the short-term scale prediction and the past one-year historical data; the short-term scale prediction of the line outage probability during the maintenance period, the short-term scale prediction of the important user and sensitive user outage probability during the maintenance period; , respectively, the line outage probability, the important user and sensitive user outage probability in the past one year.

7. The method of claim 1, wherein, The optimization model of the load replacement supply scheme in step 3 is specifically as follows: Input quantities are: load of i line in station during overhaul load of j line which is connected with i line during overhaul failure probability of j line during overhaul failure probability of important and sensitive users during overhaul ; The target function is: ； ； Wherein, i represents the station line number, j represents the contact line number of i line; The constraint condition is: ； ； ； ； ； Take the time when the above objective function is most satisfied Sequence, each There is only one optimal route. By combining these methods, the optimal load transfer scheme can be obtained. 8.The substation total outage risk scenario generation method based on a multi-time scale model according to claim 1, wherein, The multi-time scale collaborative optimization model, the power failure probability optimization model and the load transfer scheme optimization model are solved by using MATLAB CPLEX.

9. A substation total stop risk scenario generation system based on a multi-time scale model, characterized by, The system is used for realizing the substation full stop risk plan generation method in any one of claims 1-8, and the system comprises a line load prediction module, a power failure probability prediction module and a load transfer scheme generation module. The line load prediction module introduces a multi-time scale collaborative optimization model of a medium and long term and a short term, and specifically as follows: The medium and long term prediction takes a long short-term memory network as a prediction model, takes a same period load situation of last year, a weather situation during maintenance, a severe weather or a large generation of photovoltaic and wind power and a substation annual load growth situation as input quantities, and outputs a medium and long term prediction quantity through the prediction model, including a station line load during maintenance and a contact line load during maintenance; The short term prediction takes a long short-term memory network as a prediction model, takes a load situation of the past two weeks, a weather situation during maintenance and a severe weather or a large generation of photovoltaic and wind power as input quantities, and outputs a short term prediction quantity through the prediction model, including a station line load during maintenance and a contact line load during maintenance; The medium and long term prediction quantity and the short term prediction quantity are taken as input quantities, a station line load during maintenance and a contact line load during maintenance are taken as optimization variables, a difference between the medium and long term prediction and the short term prediction is taken as a target function, a multi-time scale collaborative optimization model is established, and the station line load during maintenance and the contact line load during maintenance are obtained through solution. The power failure probability prediction module is as follows: The severe weather or the large generation of photovoltaic and wind power of the short term scale and the transfer path loop current are taken as input quantities, a long short-term memory network is taken as a prediction model, a short term scale prediction quantity is output through the prediction model, including a line fault power failure probability during maintenance, an important user and a sensitive user power failure probability during maintenance; The short term scale prediction quantity and the past one year historical data are taken as input quantities, the line fault power failure probability during maintenance and the important user and the sensitive user power failure probability during maintenance are taken as optimization variables, a difference between the short term scale prediction and the past one year historical data is taken as a target function, a power failure probability optimization model is established, and the line fault power failure probability during maintenance and the important user and the sensitive user power failure probability during maintenance are obtained through solution. The load transfer scheme generation module takes the station line load during maintenance and the contact line load during maintenance obtained by the line load prediction module and the line fault power failure probability during maintenance and the important user and the sensitive user power failure probability during maintenance obtained by the power failure probability prediction module as input quantities, takes an optimal transfer path as an optimization variable, takes a line load and a power failure probability as target functions, establishes a load transfer scheme optimization model, and obtains an optimal load transfer scheme through solution.

10. A mobile terminal comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor realizes the substation full stop risk plan generation method in any one of claims 1-8 when executing the program.

Images