A machine learning based power equipment outage window period correction method and system
By using machine learning-based methods to predict power outage durations and adjust outage windows, the problem of insufficient precision in existing outage planning has been solved, resulting in more intelligent and reliable outage planning and ensuring the safety and stability of the power grid.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD
- Filing Date
- 2020-06-29
- Publication Date
- 2026-08-04
AI Technical Summary
Existing technologies fail to adequately consider power flow conditions when formulating power outage windows for electrical equipment, resulting in insufficiently precise outage plans that affect the safe and stable operation of the power grid.
A machine learning-based approach is used to predict power outage durations by utilizing the characteristic information of power equipment. The outage duration prediction model is then trained using gradient boosting decision tree method to correct the outage window period and ensure the rationality and reliability of the outage plan.
It improves the rationality of power outage window formulation, makes the power outage planning system more intelligent and reliable, provides a basis for refined equipment maintenance, and ensures the safe and stable operation of the power grid.
Smart Images
Figure CN111799782B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system dispatch and control, and specifically to a method and system for correcting power equipment outage windows based on machine learning. Background Technology
[0002] Power equipment maintenance is a crucial task in power grid operation. The rationality of equipment maintenance planning is closely related to the safe and stable operation of the power grid and directly affects the economic interests of power companies and social users.
[0003] In recent years, with the large-scale construction of ultra-high voltage AC / DC interconnected power grids, the coupling between power grids at all levels has become closer, and the scope of equipment and the scale of data have increased significantly. Power equipment maintenance needs to be arranged according to the equipment life cycle, as well as temporary maintenance to be arranged in case of infrastructure construction, technical transformation, equipment failure and other situations.
[0004] The power outage window is the optimal time period for maintenance of power generation, transmission and transformation equipment across the entire network. Currently, the power outage window is formulated based on the power flow situation of equipment in the past period, and the selection of the window length is not precise enough. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide a machine learning-based method for correcting power outage windows. This method utilizes the predicted duration of each power outage device in the outage window file to correct the outage window period of each device, thereby improving the rationality of the outage window period setting, making the outage planning system more intelligent and reliable, providing a basis for refined equipment maintenance, and providing effective protection for the safe and stable operation of the power grid.
[0006] The objective of this invention is achieved through the following technical solution:
[0007] This invention provides a machine learning-based method for correcting power outage windows, the improvement of which is that the method includes:
[0008] Predict the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file.
[0009] The outage window period of each power equipment to be shut down is corrected in the outage window file using the outage duration of each power equipment to be shut down.
[0010] Preferably, the prediction of the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file includes:
[0011] The feature information of each power device to be shut down in the power outage window file is used as the input data of the trained power outage duration prediction model, and the power outage duration of each power device to be shut down is obtained from the output of the power outage duration prediction model.
[0012] Furthermore, the training process of the trained power equipment outage duration prediction model includes:
[0013] The feature information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model.
[0014] The actual power outage time for each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage.
[0015] Furthermore, the process of determining the actual start time of power outage for each electrical device in the power outage plan execution table includes:
[0016] Traverse the power outage plan execution table to extract the measurement values of the i-th power device at the start of the power outage [s] time period. ms,i , t ms,i At each time point within the range, if there exists a time t... s1 satisfy And there exists a time t s Satisfy f(t) s ) = 0 and t s ∈[t s1 ,t s2 ], then at time t s1 Let s be the actual start time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... ms,i =s ms,i -Δt,t ms,i =t ms,i +Δt, and re-execute the traversal operation until the actual start time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0017] Among them, t s t s1 t s2 ∈[s ms,i , t ms,i ], s ms,i Let t be the starting time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. ms,is represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. ms,i =s i -Δt,t ms,i =t ms,i +Δt s i Let Δt be the planned start time of the power outage for the i-th electrical device in the power outage plan execution table, where Δt is the time interval. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t s1 () represents the i-th power device in the power outage plan execution table at time t. s1 The measured value, θ s,i This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage begins. The time interval for collecting measurements from power equipment;
[0018] The process of determining the actual end time of the power outage for the i-th power device in the power outage plan execution table includes:
[0019] Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the end of the power outage [s] time period. es,i , t es,i At each time point within the range, if there exists a time t... e2 satisfy And there exists a time t e Satisfy f(t) e ) = 0 and t e ∈[t e1 ,t e2 ], then at time t e2 Let s be the actual end time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... es,i =s es,i -Δt,t es,i =t es,i +Δt, and re-execute the traversal operation until the actual end time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0020] Among them, t e t e1 t e2 ∈[s me,i , t me,i ], s me,i Let t be the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. me,i s represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends.me,i =e i -Δt,t me,i =e i +Δt, e i This refers to the planned end time of the power outage for the i-th electrical device in the power outage plan execution table. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t e2 () represents the i-th power device in the power outage plan execution table at time t. e2 The measured value, θ e,i Let be the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends, i∈(1~N), where N is the number of power devices in the power outage plan execution table.
[0021] Furthermore, the threshold θ for the change in the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. s,i The process of determining includes:
[0022] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device in the power outage plan execution table during the sampling period at the start of the power outage, are used as input data for the trained first regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the start of the power outage is then obtained from the output of the first regression model. s,i ;
[0023] The threshold θ for the change in the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. e,i The process of determining includes:
[0024] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device at the end of the power outage within the sampling period, are used as input data for the trained second regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the end of the power outage is then obtained from the output of the second regression model. e,i .
[0025] Furthermore, the training process of the pre-trained first regression model includes:
[0026] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device during the sampling period at the start of the power outage are used as the input data of the initial first regression model. The actual jump threshold of the measurement values of each power device at the start of the power outage in the historical data is used as the output data of the initial first regression model. The initial first regression model is trained using the random forest method to obtain the trained first regression model.
[0027] The training process of the trained second regression model includes:
[0028] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device at the end of the power outage within the sampling period are used as the input data of the initial second regression model. The actual jump threshold of the measurement values of each power device at the end of the power outage in the historical data is used as the output data of the initial second regression model. The initial second regression model is trained using the random forest method to obtain the trained second regression model.
[0029] Furthermore, the power equipment includes: generators, AC lines, busbars, transformers, and DC transmission systems;
[0030] Among them, the measurement value corresponding to the generator is the generator active power measurement value, the measurement value corresponding to the AC line is the AC line active power measurement value, the measurement value corresponding to the bus is the bus line voltage measurement value, the measurement value corresponding to the transformer is the transformer active power measurement value, and the measurement value corresponding to the DC transmission system is the DC transmission system active power measurement value.
[0031] The generator's characteristic information includes: generator outage nature information, generator outage type information, generator region information, generator rated capacity information, and generator voltage level information;
[0032] The characteristic information of AC lines includes: the nature of the power outage, the type of power outage, the area to which the AC line belongs, the line type, the line length, and the voltage level.
[0033] The characteristic information of the busbar includes: the nature of the busbar outage, the type of the busbar outage, the area to which the busbar belongs, and the voltage level of the busbar.
[0034] The characteristic information of a transformer includes: the nature of the power outage, the type of power outage, the region to which the transformer is located, the rated capacity, the winding type, and the voltage level.
[0035] The characteristic information of a DC transmission system includes the nature of the power outage, the type of power outage, the region to which the DC transmission system belongs, the rated capacity, the transmission distance, and the voltage level.
[0036] Preferably, the step of adjusting the outage window period of each power equipment to be shut down in the outage window file using the outage duration of each power equipment to be shut down includes:
[0037] Select the maximum value t for the power outage duration of each power equipment to be shut down. w ;
[0038] Iterate through the outage window file, searching for time w' within each outage window period of the power equipment to be de-energized. s To satisfy the time period [w' s ,w' s +t w In the power flow simulation analysis of the internal power system, the sum of the load power of each power device to be de-energized is minimized;
[0039] t w As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, time w' s This serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.
[0040] This invention provides a machine learning-based power equipment outage window correction system, the improvement of which is that the system includes:
[0041] The prediction module is used to predict the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file.
[0042] The correction module is used to correct the outage window period of each power device to be shut down in the outage window file by using the outage duration of each power device to be shut down.
[0043] Preferably, the prediction module is specifically used for:
[0044] The feature information of each power device to be shut down in the power outage window file is used as the input data of the trained power outage duration prediction model, and the power outage duration of each power device to be shut down is obtained from the output of the power outage duration prediction model.
[0045] Furthermore, the training process of the trained power equipment outage duration prediction model includes:
[0046] The characteristic information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The initial regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model.
[0047] The actual power outage time for each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage.
[0048] Preferably, the correction module includes:
[0049] The selection unit is used to select the maximum value t from the outage duration of each power equipment to be shut down. w ;
[0050] The search unit is used to iterate through the outage window file and find the time w' within each time period of the outage window for each power device to be de-energized. s To satisfy the time period [w' s ,w' s +t w In the power flow simulation analysis of the internal power system, the sum of the load power of each power device to be de-energized is minimized;
[0051] Correction unit, used to adjust the t w As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, time w' s This serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.
[0052] Compared with the closest existing technology, the present invention has the following advantages:
[0053] The technical solution provided by this invention predicts the outage duration of each power device to be shut down based on its characteristic information in an outage window file; and then corrects the outage window period of each power device to be shut down in the outage window file using the predicted outage duration. This solution uses the predicted duration of each power device to be shut down in the outage window file to correct the outage window period, improving the rationality of the outage window period setting, making the outage planning system more intelligent and reliable, providing a basis for refined equipment maintenance, and providing effective protection for the safe and stable operation of the power grid.
[0054] The technical solution provided by this invention is based on the gradient boosting decision tree method. It predicts the power outage duration of the power equipment to be shut down from multiple dimensions, such as the equipment parameters, outage nature, outage type, and the region to which the power equipment to be shut down belongs. This improves the accuracy of predicting the power outage duration of the power equipment to be shut down and provides a basis for correcting the outage window period of the power equipment to be shut down.
[0055] The technical solution provided by this invention proposes a method for determining the actual start time and actual end time of power outages for each power device in the power outage plan execution table. This method can fill in the information gaps caused by the failure to upload the outage time in a timely manner after the power outage plan has been executed. Attached Figure Description
[0056] Figure 1 This is a flowchart of a machine learning-based method for correcting power outage windows for electrical equipment.
[0057] Figure 2 This is a structural diagram of a power equipment outage window correction system based on machine learning. Detailed Implementation
[0058] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.
[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0060] This invention provides a method for correcting power outage windows for electrical equipment based on machine learning, such as... Figure 1 As shown, the method includes:
[0061] Step 101 is used to predict the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file.
[0062] Step 102 is used to modify the power outage window period of each power equipment to be shut down in the power outage window file using the power outage duration of each power equipment to be shut down.
[0063] Specifically, step 101 includes:
[0064] The feature information of each power device to be shut down in the power outage window file is used as the input data of the trained power outage duration prediction model, and the power outage duration of each power device to be shut down is obtained from the output of the power outage duration prediction model.
[0065] Furthermore, the training process of the trained power equipment outage duration prediction model includes:
[0066] The characteristic information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The initial regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model.
[0067] The actual power outage time for each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage.
[0068] In practical engineering applications, the actual end time and start time of power outages for each electrical device in the power outage plan execution table are updated by on-site personnel. Due to the low efficiency and untimely nature of manual updates, this information may become unavailable, necessitating calculation for determination. The process of determining the actual start time of power outages for each electrical device in the power outage plan execution table includes:
[0069] Traverse the power outage plan execution table to extract the measurement values of the i-th power device at the start of the power outage [s] time period. ms,i , t ms,i At each time point within the range, if there exists a time t... s1 satisfy And there exists a time t s Satisfy f(t) s ) = 0 and t s ∈[t s1 ,t s2 ], then at time t s1 Let s be the actual start time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... ms,i =s ms,i -Δt,t ms,i =t ms,i +Δt, and re-execute the traversal operation until the actual start time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0070] Among them, t s t s1 t s2 ∈[s ms,i , t ms,i], s ms,i Let t be the starting time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. ms,i s represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. ms,i =s i -Δt,t ms,i =t ms,i +Δt s i Let Δt be the planned start time of the power outage for the i-th electrical device in the power outage plan execution table, where Δt is the time interval. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t s1 () represents the i-th power device in the power outage plan execution table at time t. s1 The measured value, θ s,i This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage begins. The time interval for collecting measurements from power equipment;
[0071] The process of determining the actual end time of the power outage for the i-th power device in the power outage plan execution table includes:
[0072] Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the end of the power outage [s] time period. es,i , t es,i At each time point within the range, if there exists a time t... e2 satisfy And there exists a time t e Satisfy f(t) e ) = 0 and t e ∈[t e1 ,t e2 ], then at time t e2 Let s be the actual end time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... es,i =s es,i -Δt,t es,i =t es,i +Δt, and re-execute the traversal operation until the actual end time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0073] Among them, t e t e1 t e2 ∈[s me,i , t me,i ], s me,iLet t be the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. me,i s represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. me,i =e i -Δt,t me,i =e i +Δt, e i This refers to the planned end time of the power outage for the i-th electrical device in the power outage plan execution table. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t e2 () represents the i-th power device in the power outage plan execution table at time t. e2 The measured value, θ e,i Let be the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends, i∈(1~N), where N is the number of power devices in the power outage plan execution table.
[0074] In determining the actual start and end times of power outages for each power device in the power outage plan execution table, it is necessary to use the jump threshold of the corresponding measurement values for each power device in the power outage plan execution table at the start and end times of the power outage.
[0075] The threshold for the jump of the measurement value corresponding to each power device in the power outage plan execution table at the start of the power outage is equal to the absolute value of the difference between the measurement value corresponding to the actual start of the power outage and the measurement value corresponding to the previous measurement time of the actual start of the power outage in the power outage plan execution table.
[0076] Similarly, the threshold value for the change of the measurement value corresponding to each power device at the end of the power outage in the power outage plan execution table is equal to the absolute value of the difference between the measurement value corresponding to the next measurement time after the actual end of the power outage in the power outage plan execution table and the measurement value corresponding to the actual end of the power outage in the power outage plan execution table.
[0077] Since the actual start and end times of power outages for each electrical device in the power outage plan execution table are unknown, this value is also unknown.
[0078] Here, we use historical data with complete power outage information to predict the jump threshold of the corresponding measurement values of each power device in the power outage plan execution table at the start and end of the power outage based on the random forest method. This method can ensure the accuracy of the prediction as much as possible.
[0079] Wherein, in the power outage plan execution table, the threshold θ for the change in the measurement value corresponding to the i-th power device at the start of the power outage. s,i The process of determining includes:
[0080] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device in the power outage plan execution table during the sampling period at the start of the power outage, are used as input data for the trained first regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the start of the power outage is then obtained from the output of the first regression model. s,i ;
[0081] The threshold θ for the change in the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. e,i The process of determining includes:
[0082] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device at the end of the power outage within the sampling period, are used as input data for the trained second regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the end of the power outage is then obtained from the output of the second regression model. e,i .
[0083] Furthermore, the training process of the pre-trained first regression model includes:
[0084] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device during the sampling period at the start of the power outage are used as the input data of the initial first regression model. The actual jump threshold of the measurement values of each power device at the start of the power outage in the historical data is used as the output data of the initial first regression model. The initial first regression model is trained using the random forest method to obtain the trained first regression model.
[0085] The training process of the trained second regression model includes:
[0086] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device at the end of the power outage within the sampling period are used as the input data of the initial second regression model. The actual jump threshold of the measurement values of each power device at the end of the power outage in the historical data is used as the output data of the initial second regression model. The initial second regression model is trained using the random forest method to obtain the trained second regression model.
[0087] Specifically, the power equipment includes: generators, AC lines, busbars, transformers, and DC transmission systems;
[0088] Among them, the measurement value corresponding to the generator is the generator active power measurement value, the measurement value corresponding to the AC line is the AC line active power measurement value, the measurement value corresponding to the bus is the bus line voltage measurement value, the measurement value corresponding to the transformer is the transformer active power measurement value, and the measurement value corresponding to the DC transmission system is the DC transmission system active power measurement value.
[0089] The generator's characteristic information includes: generator outage nature information, generator outage type information, generator region information, generator rated capacity information, and generator voltage level information;
[0090] The characteristic information of AC lines includes: the nature of the power outage, the type of power outage, the area to which the AC line belongs, the line type, the line length, and the voltage level.
[0091] The characteristic information of the busbar includes: the nature of the busbar outage, the type of the busbar outage, the area to which the busbar belongs, and the voltage level of the busbar.
[0092] The characteristic information of a transformer includes: the nature of the power outage, the type of power outage, the region to which the transformer is located, the rated capacity, the winding type, and the voltage level.
[0093] The characteristic information of a DC transmission system includes the nature of the power outage, the type of power outage, the region to which the DC transmission system belongs, the rated capacity, the transmission distance, and the voltage level.
[0094] In a specific embodiment of the present invention, the power outage nature information, power outage type information, and area information of the power equipment are information after one-hot encoding, while the equipment rated capacity information, equipment voltage level information, equipment line type information, equipment line length information, equipment winding type information, and equipment transmission distance information are information after normalization processing.
[0095] Specifically, step 102 includes:
[0096] Step 102.1 is used to select the maximum value t from the power outage duration of each power equipment to be shut down. w ;
[0097] Step 102.2 is used to iterate through the outage window file and find the time w' within each time period of the outage window for each power device to be de-energized. s To satisfy the time period [w' s ,w' s +t w In the power flow simulation analysis of the internal power system, the sum of the load power of each power device to be de-energized is minimized;
[0098] Step 102.3, used to transfer the t w As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, time w' s This serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.
[0099] This invention provides a machine learning-based power equipment outage window correction system, such as... Figure 2 As shown, the system includes:
[0100] The prediction module is used to predict the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file.
[0101] The correction module is used to correct the outage window period of each power device to be shut down in the outage window file by using the outage duration of each power device to be shut down.
[0102] Specifically, the prediction module is used for:
[0103] The feature information of each power device to be shut down in the power outage window file is used as the input data of the trained power outage duration prediction model, and the power outage duration of each power device to be shut down is obtained from the output of the power outage duration prediction model.
[0104] Furthermore, the training process of the trained power equipment outage duration prediction model includes:
[0105] The characteristic information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The initial regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model.
[0106] The actual power outage time for each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage.
[0107] Specifically, the process of determining the actual start time of power outage for each electrical device in the power outage plan execution table includes:
[0108] Traverse the power outage plan execution table to extract the measurement values of the i-th power device at the start of the power outage [s] time period. ms,i , t ms,i At each time point within the range, if there exists a time t... s1 satisfy And there exists a time t s Satisfy f(t) s ) = 0 and t s ∈[t s1 ,t s2 ], then at time t s1 Let s be the actual start time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... ms,i =s ms,i -Δt,t ms,i =t ms,i +Δt, and re-execute the traversal operation until the actual start time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0109] Among them, t s t s1 t s2 ∈[s ms,i , t ms,i ], s ms,i Let t be the starting time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. ms,i s represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. ms,i =s i -Δt,t ms,i =t ms,i +Δt s i Let Δt be the planned start time of the power outage for the i-th electrical device in the power outage plan execution table, where Δt is the time interval. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t s1 () represents the i-th power device in the power outage plan execution table at time t. s1 The measured value, θ s,i This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage begins. The time interval for collecting measurements from power equipment;
[0110] The process of determining the actual end time of the power outage for the i-th power device in the power outage plan execution table includes:
[0111] Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the end of the power outage [s] time period. es,i , t es,i At each time point within the range, if there exists a time t... e2 satisfy And there exists a time t e Satisfy f(t) e ) = 0 and t e ∈[t e1 ,t e2 ], then at time t e2 Let s be the actual end time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let s be... es,i =s es,i -Δt,t es,i =t es,i +Δt, and re-execute the traversal operation until the actual end time of the power outage of the i-th power device in the power outage plan execution table is obtained;
[0112] Among them, t e t e1 t e2 ∈[s me,i , t me,i ], s me,i Let t be the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. me,i s represents the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. me,i =e i -Δt,t me,i =e i +Δt, e i This refers to the planned end time of the power outage for the i-th electrical device in the power outage plan execution table. For the i-th power device in the power outage plan execution table at time... The measured value, f i (t e2 () represents the i-th power device in the power outage plan execution table at time t. e2 The measured value, θ e,i Let be the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends, i∈(1~N), where N is the number of power devices in the power outage plan execution table.
[0113] Specifically, the threshold θ for the change in the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. s,i The process of determining includes:
[0114] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device in the power outage plan execution table during the sampling period at the start of the power outage, are used as input data for the trained first regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the start of the power outage is then obtained from the output of the first regression model. s,i ;
[0115] The threshold θ for the change in the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. e,i The process of determining includes:
[0116] The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device at the end of the power outage within the sampling period, are used as input data for the trained second regression model. The jump threshold θ of the measured value of the i-th power device in the power outage plan execution table at the end of the power outage is then obtained from the output of the second regression model. e,i .
[0117] Specifically, the training process of the trained first regression model includes:
[0118] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device during the sampling period at the start of the power outage are used as the input data of the initial first regression model. The actual jump threshold of the measurement values of each power device at the start of the power outage in the historical data is used as the output data of the initial first regression model. The initial first regression model is trained using the random forest method to obtain the trained first regression model.
[0119] The training process of the trained second regression model includes:
[0120] The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device at the end of the power outage within the sampling period are used as the input data of the initial second regression model. The actual jump threshold of the measurement values of each power device at the end of the power outage in the historical data is used as the output data of the initial second regression model. The initial second regression model is trained using the random forest method to obtain the trained second regression model.
[0121] Specifically, the power equipment includes: generators, AC lines, busbars, transformers, and DC transmission systems;
[0122] Among them, the measurement value corresponding to the generator is the generator active power measurement value, the measurement value corresponding to the AC line is the AC line active power measurement value, the measurement value corresponding to the bus is the bus line voltage measurement value, the measurement value corresponding to the transformer is the transformer active power measurement value, and the measurement value corresponding to the DC transmission system is the DC transmission system active power measurement value.
[0123] The generator's characteristic information includes: generator outage nature information, generator outage type information, generator region information, generator rated capacity information, and generator voltage level information;
[0124] The characteristic information of AC lines includes: the nature of the power outage, the type of power outage, the area to which the AC line belongs, the line type, the line length, and the voltage level.
[0125] The characteristic information of the busbar includes: the nature of the busbar outage, the type of the busbar outage, the area to which the busbar belongs, and the voltage level of the busbar.
[0126] The characteristic information of a transformer includes: the nature of the power outage, the type of power outage, the region to which the transformer is located, the rated capacity, the winding type, and the voltage level.
[0127] The characteristic information of a DC transmission system includes the nature of the power outage, the type of power outage, the region to which the DC transmission system belongs, the rated capacity, the transmission distance, and the voltage level.
[0128] Specifically, the correction module includes:
[0129] The selection unit is used to select the maximum value t from the outage duration of each power equipment to be shut down. w ;
[0130] The search unit is used to iterate through the outage window file and find the time w' within each time period of the outage window for each power device to be de-energized. s To satisfy the time period [w' s ,w' s +t w In the power flow simulation analysis of the internal power system, the sum of the load power of each power device to be de-energized is minimized;
[0131] Correction unit, used to adjust the t w As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, time w' sThis serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.
[0132] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0133] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0134] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0135] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A machine learning based power equipment outage window period correction method, characterized by, The method includes: Based on the feature information of each power device to be shut down in the power outage window file, the power outage duration of each power device to be shut down is predicted by a trained power device outage duration prediction model. The outage window period of each power equipment to be shut down in the outage window file is corrected using the outage duration of each power equipment to be shut down. The training process of the trained power equipment outage duration prediction model includes: The characteristic information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The initial regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model. Wherein, the actual power outage time of each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage of each power device; The process of determining the actual start time of power outage for each power device in the power outage plan execution table includes: Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the start of the power outage. At each time point within the time frame, if there exists a time frame... satisfy And it exists at any time. satisfy and Then at time Let be the actual start time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let , Then, the traversal operation is re-executed until the actual start time of the power outage for the i-th power device in the power outage plan execution table is obtained; in, , This refers to the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. This refers to the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. , This refers to the planned start time of the power outage for the i-th electrical device in the power outage plan execution table. For time intervals, For the i-th power device in the power outage plan execution table at time... The measured value, For the i-th power device in the power outage plan execution table at time... The measured value, This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage begins. The time interval for collecting measurements from power equipment; The process of determining the actual end time of the power outage for the i-th power device in the power outage plan execution table includes: Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the end of the power outage. At each time point within the time frame, if there exists a time frame... satisfy And it exists at any time. satisfy and Then at time Let be the actual end time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let , Then, the traversal operation is re-executed until the actual end time of the power outage for the i-th power device in the power outage plan execution table is obtained; in, , This refers to the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. This refers to the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. , , This refers to the planned end time of the power outage for the i-th electrical device in the power outage plan execution table. For the i-th power device in the power outage plan execution table at time... The measured value, For the i-th power device in the power outage plan execution table at time... The measured value, This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. , This represents the number of electrical equipment listed in the power outage plan execution table; The threshold value change of the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. The process of determining includes: The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device in the power outage plan execution table during the sampling period at the start of the power outage, are used as input data for the trained first regression model. The jump threshold of the measured value of the i-th power device in the power outage plan execution table at the start of the power outage is then obtained from the output of the first regression model. ; The threshold value of the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. The process of determining includes: The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device at the end of the power outage within the sampling period, are used as input data for the trained second regression model. The jump threshold of the measured value of the i-th power device in the power outage plan execution table at the end of the power outage is then obtained from the output of the second regression model. .
2. The method as described in claim 1, characterized in that, The training process of the pre-trained first regression model includes: The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device during the sampling period at the start of the power outage are used as the input data of the initial first regression model. The actual jump threshold of the measurement values of each power device at the start of the power outage in the historical data is used as the output data of the initial first regression model. The initial first regression model is trained using the random forest method to obtain the trained first regression model. The training process of the trained second regression model includes: The voltage values of each power device in the historical data and the maximum, minimum and average measurement values of each power device at the end of the power outage within the sampling period are used as the input data of the initial second regression model. The actual jump threshold of the measurement values of each power device at the end of the power outage in the historical data is used as the output data of the initial second regression model. The initial second regression model is trained using the random forest method to obtain the trained second regression model.
3. The method as described in claim 1, characterized in that, The power equipment includes: generators, AC lines, busbars, transformers, and DC transmission systems; Among them, the measurement value corresponding to the generator is the generator active power measurement value, the measurement value corresponding to the AC line is the AC line active power measurement value, the measurement value corresponding to the bus is the bus line voltage measurement value, the measurement value corresponding to the transformer is the transformer active power measurement value, and the measurement value corresponding to the DC transmission system is the DC transmission system active power measurement value. The generator's characteristic information includes: generator outage nature information, generator outage type information, generator region information, generator rated capacity information, and generator voltage level information; The characteristic information of AC lines includes: the nature of the power outage, the type of power outage, the area to which the AC line belongs, the line type, the line length, and the voltage level. The characteristic information of the busbar includes: the nature of the busbar outage, the type of the busbar outage, the area to which the busbar belongs, and the voltage level of the busbar. The characteristic information of a transformer includes: the nature of the power outage, the type of power outage, the region to which the transformer is located, the rated capacity, the winding type, and the voltage level. The characteristic information of a DC transmission system includes the nature of the power outage, the type of power outage, the region to which the DC transmission system belongs, the rated capacity, the transmission distance, and the voltage level.
4. The method as described in claim 1, characterized in that, The step of correcting the outage window period of each power equipment to be shut down in the outage window file using the outage duration of each power equipment to be shut down includes: Select the maximum value for the power outage duration of each power equipment to be shut down. ; Iterate through the outage window file, searching for the time within each outage window period of the power equipment to be de-energized. To make it meet the requirements of the time period In the internal power flow simulation analysis, the sum of the load power of each power device to be de-energized is minimized; The As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, it will be used to determine the time period. This serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.
5. A power equipment outage window correction system based on machine learning, characterized in that, The system includes: The prediction module is used to predict the power outage duration of each power device to be shut down based on the feature information of each power device to be shut down in the power outage window file and through a trained power device power outage duration prediction model. The correction module is used to correct the outage window period of each power equipment to be shut down in the outage window file by using the outage duration of each power equipment to be shut down; The training process of the trained power equipment outage duration prediction model includes: The characteristic information of each power device in the power outage plan execution table is used as the input data of the initial regression tree model, and the actual power outage time of each power device in the power outage plan execution table is used as the output data of the initial regression tree model. The initial regression tree model is trained using the gradient boosting decision tree method to obtain the trained power outage duration prediction model. Wherein, the actual power outage time of each power device in the power outage plan execution table is the time difference between the actual end time of the power outage and the actual start time of the power outage of each power device; The process of determining the actual start time of power outage for each power device in the power outage plan execution table includes: Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the start of the power outage. At each time point within the time frame, if there exists a time frame... satisfy And it exists at any time. satisfy and Then at time Let be the actual start time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let , Then, the traversal operation is re-executed until the actual start time of the power outage for the i-th power device in the power outage plan execution table is obtained; in, , This refers to the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. This refers to the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table at the start of the power outage. , This refers to the planned start time of the power outage for the i-th electrical device in the power outage plan execution table. For time intervals, For the i-th power device in the power outage plan execution table at time... The measured value, For the i-th power device in the power outage plan execution table at time... The measured value, This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage begins. The time interval for collecting measurements from power equipment; The process of determining the actual end time of the power outage for the i-th power device in the power outage plan execution table includes: Traverse the power outage plan execution table to extract the measurement values corresponding to the i-th power device at the end of the power outage. At each time point within the time frame, if there exists a time frame... satisfy And it exists at any time. satisfy and Then at time Let be the actual end time of the power outage for the i-th electrical device in the power outage plan execution table; otherwise, let , Then, the traversal operation is re-executed until the actual end time of the power outage for the i-th power device in the power outage plan execution table is obtained; in, , This refers to the start time of the sampling period for the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. This refers to the end time of the sampling period for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. , , This refers to the planned end time of the power outage for the i-th electrical device in the power outage plan execution table. For the i-th power device in the power outage plan execution table at time... The measured value, For the i-th power device in the power outage plan execution table at time... The measured value, This is the threshold value for the measurement value of the i-th power device in the power outage plan execution table when the power outage ends. , This represents the number of electrical equipment listed in the power outage plan execution table; The threshold value change of the measurement value corresponding to the i-th power device in the power outage plan execution table at the start of the power outage. The process of determining includes: The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device in the power outage plan execution table during the sampling period at the start of the power outage, are used as input data for the trained first regression model. The jump threshold of the measured value of the i-th power device in the power outage plan execution table at the start of the power outage is then obtained from the output of the first regression model. ; The threshold value of the measurement value corresponding to the i-th power device in the power outage plan execution table when the power outage ends. The process of determining includes: The voltage value of the i-th power device in the power outage plan execution table, along with the maximum, minimum, and average measured values of the i-th power device at the end of the power outage within the sampling period, are used as input data for the trained second regression model. The jump threshold of the measured value of the i-th power device in the power outage plan execution table at the end of the power outage is then obtained from the output of the second regression model. .
6. The system as described in claim 5, characterized in that, The correction module includes: The selection unit is used to select the maximum value among the power outage durations of each power equipment to be shut down. ; The search unit is used to iterate through the outage window file and find the time within each outage window period of each power equipment to be de-energized. To make it meet the requirements of the time period In the internal power flow simulation analysis, the sum of the load power of each power device to be de-energized is minimized; Correction unit, used to correct the As the duration of the power outage window for each piece of electrical equipment to be shut down in the power outage window file, it will be used to determine the time period. This serves as the start time of the power outage window period for each power equipment to be shut down in the power outage window file.