A method for intelligent optimization of power outage plans based on a power grid cognitive computing framework
By constructing a power grid cognitive computing framework and rule configuration environment, the problems of manual dependence and data matching in the existing power outage plan preparation are solved, realizing automated power outage plan optimization and risk analysis, and improving verification efficiency and accuracy.
Patent Information
- Application Number
- CN202611618245.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing methods for preparing and verifying power outage plans rely on manual experience and lack automated optimization, resulting in low efficiency in plan optimization and adjustment, insufficient applicability of rules, high data matching failure rate, incomplete risk analysis, and difficulty in ensuring the consistency of verification results.
A power grid cognitive computing framework is constructed, a rule configuration environment is established, and rule items are bound according to plan type, voltage level, dispatch level, equipment type and constraint strength. Multi-dimensional verification and risk analysis are achieved through graph structure data, and power outage plans are automatically adjusted to improve verification efficiency and accuracy.
It has improved the flexibility of rule configuration and the efficiency of verification, enhanced data quality, increased the comprehensiveness and accuracy of verification, and automated and standardized risk analysis, reducing the blindness of manual adjustments and verification omissions.
Smart Images

Figure CN123117484A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automated system scheduling management technology, and in particular to an intelligent optimization method for power outage plans based on a power grid cognitive computing framework. Background Technology
[0002] Power outage planning management is a crucial aspect of power system operation and dispatching, involving multiple aspects such as equipment maintenance, mode arrangement, and risk management. Current outage planning typically relies on the manual experience of dispatchers, lacking sufficient comprehensive analytical capabilities regarding power grid topology, equipment operating status, and load distribution. This leads to delays in identifying scheduling conflicts and low efficiency in optimizing and adjusting outage plans.
[0003] Existing power outage plan verification methods typically employ independent rule verification modules, lacking a unified computational support platform across the various rule items. Ticket rules, scheduling rules, and N-1 risk rules are executed in different systems, with no data sharing and no linkage of calculation results. The same power outage plan must be imported into different systems multiple times for verification, making the operation cumbersome and inconsistent.
[0004] Current power outage risk analysis typically employs offline power flow calculations and manual forecasting, lacking a computable graph structure with equipment as nodes and connections as branches. When the power grid topology is complex and the number of devices is large, manual forecasting is prone to overlooking critical fault scenarios, N-1 risk identification is incomplete, risk rating is highly subjective, and lacks quantitative basis.
[0005] The current conflict optimization of power outage plans usually adopts a manual negotiation method, with dispatchers adjusting outage times based on experience, without establishing an automated optimization solution mechanism. When there are complex coupling relationships between multiple power outage plans, such as equipment mutual exclusion, simultaneous outage constraints, and cross-sectional constraints, manual adjustment is difficult to find the global optimal solution, the adjustment cycle is long, and the adjusted plan is not re-verified, leaving potential risks.
[0006] Existing power outage planning rule management typically uses a fixed set of rules, without differentiated configuration based on plan type, voltage level, dispatch level, or other dimensions. The verification depth and rule scope of annual, monthly, weekly, and daily plans are the same, resulting in overly cumbersome annual plan verification and overly coarse daily plan verification, insufficient rule applicability, and low verification efficiency.
[0007] Current power outage planning data governance typically employs simple string matching without establishing an object-oriented mapping mechanism for equipment. When the equipment names in the power outage plan do not match the naming conventions of the power grid model data, the matching failure rate is high, and unmatched equipment is directly removed or misjudged, impacting the accuracy of verification due to data quality issues. Summary of the Invention
[0008] Therefore, embodiments of the present invention provide a method for intelligent optimization of power outage plans based on a power grid cognitive computing framework. A method for intelligent optimization of power outage plans based on a power grid cognitive computing framework is provided, including the following steps: S1. Construct a power grid cognitive computing framework, which organizes the power grid equipment topology, power grid equipment ledger, equipment measured load, equipment cross-sectional combination relationship and important user power supply ledger, consisting of substations, busbars, lines, main transformers and connection relationships, into a computable graph structure data with equipment as nodes and connection relationships as branches, forming a cognitive layer. In the maintenance mode, planned power outage equipment is superimposed on the graph structure data, and the cognitive layer supports the undervoltage determination, power transfer path search and N-1 prediction analysis. S2. Establish a rule configuration environment, bind the rule items in the ticket rule library, the orchestration rule library and the N-1 risk rule library according to the dimensions composed of plan type, voltage level, scheduling level, equipment type and constraint strength, and configure the start and stop status of each rule item. The constraint strength is divided into two levels: mandatory and suggested. S3. Obtain the power outage plan, load the binding and effective rule items in the ticket rule library, and perform data governance and verification on the power outage plan. The data governance includes matching and objectifying the equipment and power grid model data in the power outage plan, verifying the time rationality and ticket correctness. If the verification fails, the plan data is corrected and re-governed. If the verification passes, proceed to step S4. S4. Bind the effective arrangement rule set according to the plan type of the power outage plan, and perform maintenance verification on the cognitive layer. The maintenance verification includes equipment mutual exclusion and simultaneous outage verification, cross-sectional constraint verification, window period and holiday verification, power supply protection verification, standard construction period verification, repeated power outage verification, important user power supply impact verification and equipment overload verification. S5. Based on the verification under maintenance, driven by the rule items bound and effective in the N-1 risk rule base, the maintenance method is superimposed on the cognitive layer to perform N-1 pre-analysis, and the results of the voltage loss judgment and the transfer path search are aggregated to obtain the load reduction. The power outage plan is then risk-classified according to the power grid safety accident level classification clause. S6. When the power outage plan hits a scheduling conflict or risk rating, and the rule item corresponding to the hit anomaly is a time-adjustable rule item, it is determined that intelligent optimization is possible. Under the constraint of the rule item bound by dimension in step S2, the cross-section constraint and the same-outage constraint are transformed into coupled time constraints on the graph structure data. The optimization scheme is solved with the goal of minimizing the sum of the power outage time interval translation of each power outage plan. After the power outage time interval of any power outage plan is translated, the conflict status of the other power outage plans is updated in linkage on the cognitive layer to generate the time optimization suggestion of the power outage plan. S7. Receive the manual adoption or revision results of the time optimization suggestions, return the adopted or revised power outage plan to step S4 for verification again, and perform verification and optimization repeatedly until the anomaly corresponding to the rule item with mandatory constraint strength is eliminated. S8. When the power outage plan has no scheduling conflicts and no risk rating is hit, or after passing the cyclical verification in step S7, output the verification conclusion, the optimized power outage plan and risk description.
[0009] Beneficial effects The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following: This invention establishes a rule configuration environment, binding rule items in the ticket rule base, orchestration rule base, and N-1 risk rule base according to dimensions consisting of plan type, voltage level, scheduling level, equipment type, and constraint strength. It also configures the start and stop status of each rule item, with constraint strength divided into mandatory and suggested levels. This solves the technical problem of insufficient rule applicability caused by the use of fixed rule sets without differentiated configuration in the prior art. It achieves the technical effect of hierarchical and classified management of rules through multi-dimensional binding, improving the flexibility of rule configuration and the efficiency of verification.
[0010] This invention obtains a power outage plan and loads the bound and effective rule items from the ticket rule library to perform data governance and verification. This includes matching and objectifying the equipment in the power outage plan with the power grid model data, verifying the rationality of the time and the correctness of the ticket. It solves the technical problems of high matching failure rate and direct rejection of unmatched equipment caused by simple string matching in the prior art. It achieves the technical effect of objectifying equipment through plant-level matching and equipment-level matching, and improving data quality through abnormal views and manual completion.
[0011] This invention binds effective arrangement rule sets according to the plan type of the power outage plan, and performs maintenance verification at the cognitive layer. This includes equipment mutual exclusion and simultaneous outage verification, cross-section constraint verification, window period and holiday verification, power supply guarantee verification, standard construction period verification, repeated power outage verification, important user power supply impact verification, and equipment overload verification. This solves the technical problem in the prior art where each rule item is independently verified without being linked to the power grid topology. It achieves the technical effect of realizing multi-dimensional verification linkage and improving the comprehensiveness and accuracy of verification by superimposing maintenance methods at the cognitive layer.
[0012] This invention, based on the verification under maintenance, is driven by the binding and effective rule items in the N-1 risk rule base, and performs N-1 pre-emptive analysis on the cognitive layer by superimposing maintenance methods. It aggregates the results of voltage loss judgment and power transfer path search to obtain the load reduction, and performs risk rating of the power outage plan according to the power grid safety accident level classification clause. This solves the technical problems of easy omission of key fault scenarios and strong subjectivity of risk rating in the prior art by manual prediction. It achieves the technical effect of automating N-1 pre-emptive analysis through graph structure data traversal and standardizing risk rating by quantifying load reduction. Attached Figure Description
[0013] Figure 1 A flowchart provided for an embodiment of this application. Detailed Implementation
[0014] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.
[0015] like Figure 1 As shown, a method for intelligent optimization of power outage plans based on a power grid cognitive computing framework includes the following steps: S1. Construct a power grid cognitive computing framework, which organizes the power grid equipment topology, power grid equipment ledger, equipment measured load, equipment cross-sectional combination relationship and important user power supply ledger, consisting of substations, busbars, lines, main transformers and connection relationships, into a computable graph structure data with equipment as nodes and connection relationships as branches, forming a cognitive layer. In the maintenance mode, planned power outage equipment is superimposed on the graph structure data, and the cognitive layer supports the undervoltage determination, power transfer path search and N-1 prediction analysis. S2. Establish a rule configuration environment, bind the rule items in the ticket rule library, the orchestration rule library and the N-1 risk rule library according to the dimensions composed of plan type, voltage level, scheduling level, equipment type and constraint strength, and configure the start and stop status of each rule item. The constraint strength is divided into two levels: mandatory and suggested. S3. Obtain the power outage plan, load the binding and effective rule items in the ticket rule library, and perform data governance and verification on the power outage plan. The data governance includes matching and objectifying the equipment and power grid model data in the power outage plan, verifying the time rationality and ticket correctness. If the verification fails, the plan data is corrected and re-governed. If the verification passes, proceed to step S4. S4. Bind the effective arrangement rule set according to the plan type of the power outage plan, and perform maintenance verification on the cognitive layer. The maintenance verification includes equipment mutual exclusion and simultaneous outage verification, cross-sectional constraint verification, window period and holiday verification, power supply protection verification, standard construction period verification, repeated power outage verification, important user power supply impact verification and equipment overload verification. S5. Based on the verification under maintenance, driven by the rule items bound and effective in the N-1 risk rule base, the maintenance method is superimposed on the cognitive layer to perform N-1 pre-analysis, and the results of the voltage loss judgment and the transfer path search are aggregated to obtain the load reduction. The power outage plan is then risk-classified according to the power grid safety accident level classification clause. S6. When the power outage plan hits a scheduling conflict or risk rating, and the rule item corresponding to the hit anomaly is a time-adjustable rule item, it is determined that intelligent optimization is possible. Under the constraint of the rule item bound by dimension in step S2, the cross-section constraint and the same-outage constraint are transformed into coupled time constraints on the graph structure data. The optimization scheme is solved with the goal of minimizing the sum of the power outage time interval translation of each power outage plan. After the power outage time interval of any power outage plan is translated, the conflict status of the other power outage plans is updated in linkage on the cognitive layer to generate the time optimization suggestion of the power outage plan. S7. Receive the manual adoption or revision results of the time optimization suggestions, return the adopted or revised power outage plan to step S4 for verification again, and perform verification and optimization repeatedly until the anomaly corresponding to the rule item with mandatory constraint strength is eliminated. S8. When the power outage plan has no scheduling conflicts and no risk rating is hit, or after passing the cyclical verification in step S7, output the verification conclusion, the optimized power outage plan and risk description.
[0016] It should be noted that when implementing the above method, the first... The time interval of the planned power outage is denoted as the start time. and the end time The planned construction period is as follows Calculation. Adjustable time rules refer to rules that can eliminate conflicts by shifting outage time intervals, compressing planned durations, or staggering conflicting plans. When a rule is found to be missing ticket information or inconsistent equipment records, an anomaly conclusion is output and the plan data is returned for correction; the optimization solution is not entered. Minimizing the sum of outage time interval shifts means minimizing the overall change in the adjusted schedule relative to the original schedule. Coupled time constraints refer to the constraints that affect the outage time intervals of multiple outage plans in the graph structure data. After any plan is shifted, the conflict status of plans with coupled time constraints is re-evaluated. Data governance, maintenance verification, N-1 scenario analysis, optimization solution, and re-verification are all performed on the same graph structure data. Each step shares the same computational foundation, solving the problem of unreliable verification conclusions caused by inconsistencies between planning management data and power grid model data. After section constraints and simultaneous outage constraints are transformed into coupled time constraints, changes caused by any plan shift are updated in a linked manner at the cognitive layer. This solves the problem of multiple plan time adjustments being mutually coupled, and the issue of new conflicts arising from manual adjustments without being noticed when changing one point. This allows the optimization solution to satisfy all coupled constraints in a single solution. After manual adoption or revision, the verification is returned to form a closed loop until the anomalies corresponding to the rule items with mandatory constraint strength are eliminated, ensuring that the output power outage plan is executable.
[0017] As an optional embodiment, in step S1, the planned power outage equipment is superimposed onto the graph structure data under maintenance mode, specifically as follows: The operating status of the nodes and branches corresponding to the power outage equipment involved in the power outage plan is marked as maintenance status in the graph structure data; The connectivity of the graph structure data is updated using nodes and branches marked as being under maintenance as boundaries. The updated graph structure data is traversed from each power source point, and buses that cannot maintain connectivity with any power source point are identified as under-voltage buses. Starting from the undervoltage bus, a transfer path is generated by searching for a reachable path to a normal energized power source point through the standby communication equipment. The transferable load of the transfer path is determined according to the minimum rated transmission capacity of each device on the reachable path.
[0018] It should be noted that the updating and traversal of connectivity relationships are performed in depth-first order: after disconnecting the nodes and branches corresponding to the equipment under maintenance from the graph structure data, the system traverses from each power source point. The buses not traversed form the set of undervoltage buses, denoted as... Busbar The load is denoted as The aggregate value of the measured load on this bus is taken from the equipment; the transferable load of the transfer path is denoted as... Its value is constrained by the minimum rated transmission capacity of each device on the transfer path. By traversing the graph to determine the undervoltage bus and constraining the load transfer according to the minimum rated transmission capacity on the transfer path, a quantitative input is provided for load reduction aggregation and risk classification, avoiding overestimation of transfer capacity and underestimation of operational risk, so that the risk classification results reflect the actual power supply capacity under maintenance mode.
[0019] As an optional embodiment, in step S2, the binding is carried out according to the plan type dimension. Specifically, the enabled rule items are bound separately in the same rule base according to the annual plan, monthly plan, weekly plan and daily plan. The annual plan is bound to the ticket rule item and the repeated power outage rule item. The monthly plan and weekly plan are bound to the cross-section constraint rule item, equipment mutual exclusion and simultaneous shutdown rule item and N-1 risk rule item on the basis of the binding result of the annual plan. Each rule item's detailed maintenance includes its name, description, implementation code, and validity flag. Rule items with an invalid validity flag are not included in the binding process. The ticket rule base includes rule items for work repetition verification, time validity verification, plan correlation verification, terminology standardization verification, plant and equipment consistency verification, and hourly user count verification. The orchestration rule base includes rule items for equipment mutual exclusion and simultaneous shutdown, cross-sectional constraints, window periods and holidays, power supply protection, standard construction period, repeated power outages, impact of power supply to important users, and equipment overload; The N-1 risk rule base includes rule items for N-1 contingency planning, pressure loss and power transfer analysis, power reduction load aggregation, and risk rating. The ticket rule base, the arrangement rule base, and the N-1 risk rule base drive the planned data governance in step S3, the maintenance verification in step S4, and the N-1 risk analysis in step S5, respectively. Each step only loads the rule items that are bound and effective in the corresponding rule base.
[0020] It should be noted that the effective rule set for dimension binding is determined as follows: The plan type, voltage level, scheduling level, and equipment type attributes of the power outage plan are read. Rule items with consistent dimension attributes, an enabled start / stop status, and a valid flag are retrieved from the rule configuration environment. These constitute the effective rule set for the power outage plan. The detailed rules for rule items bound to different plan types within the same rule base are independent of each other; modifying the binding of one plan type does not affect the binding of other plan types. Time validity verification checks that the start time is not later than the end time. Plant / station equipment consistency verification checks that the plant / station name and equipment name filled in on the ticket are consistent with the equipment ledger. The calculation process for the number of users per hour is as follows: The... The number of households in the planned power outage is calculated by Calculation, where The area covered by the planned power outage The number of users experiencing power outages in each power supply area is obtained from the equipment ledger and the power supply ledger for important users. For the first The power outage duration for each power supply area is calculated based on the outage time interval minus the time for power restoration. The annual plan only performs checks on ticket-related and repeated outage categories, while monthly and weekly plans are further bound to checks on cross-sections, simultaneous outages, and N-1 classification categories. The strictness of the checks matches the granularity of the plan's time frame, preventing excessive checks on the annual plan that could generate numerous false alarms, while also preventing monthly and weekly plans from missing cross-section and risk-related anomalies. Three rule bases drive three processing stages respectively; adding or modifying rules does not alter the execution flow of verification and classification. The number of users affected by outages is aggregated by the number of outage users and the outage duration, quantifying the impact of ticket-related outages into comparable quantities to support comparisons of the degree of impact between plans.
[0021] As an optional embodiment, in step S3, the equipment in the power outage plan is matched with the power grid model data and objectified, specifically as follows: The power outage plan and the power grid model data are matched at the plant / station level using the plant / station name, and at the device level using the dual device naming, the matching is performed at the plant / station level. Output the abnormal view of unmatched plants and unmatched equipment, and re-execute the matching after receiving the manually supplemented mapping relationship; Devices that have not been objectified are removed from the scope of verification of topology rule items. These devices that have not been objectified only participate in the verification of ticket rule items.
[0022] It should be noted that dual naming of equipment refers to the combined naming of the plant / station name and the equipment name. Before matching, the name text is standardized: spaces are removed, full-width and half-width characters are unified, and voltage level writing is unified. Plant / station level matching is performed by comparing the standardized plant / station name for complete equality, and equipment level matching is performed by comparing the standardized dual naming of the equipment within the successfully matched plant / station. Equipment that fails both levels of matching is entered into the abnormal view, and the mapping relationship is manually completed before rematching. Topology rule items refer to the rules items of equipment mutual exclusion and simultaneous shutdown, cross-sectional constraints, equipment overload, and N-1 pre-planning. The two-level matching at the plant / station level and equipment level places the equipment on the ticket into the specific nodes and branches of the graph structure data, which is the premise of topology verification and pre-planning analysis. Equipment that has not been objectified is removed from the topology verification scope, rather than participating in the calculation as an incorrect mapping, to avoid incorrect undervoltage judgment and incorrect conflict conclusions caused by inconsistent names.
[0023] As an optional embodiment, in step S4: The repeated power outage verification is as follows: for any two power outage plans of the same device in the power outage plan set, the difference between the start time of the later power outage plan and the end time of the previous power outage plan is used as the power outage interval. If the power outage interval is less than the repeated power outage reference interval corresponding to the device, it is determined that a repeated power outage has occurred. The repeated power outage reference interval is read from the rule items bound to the effective rules in the orchestration rule base according to the equipment type and voltage level to which the device belongs. The equipment overload verification is performed by superimposing maintenance methods on the cognitive layer, performing power flow calculation, reading the power flow value of each device in the section and the rated transmission capacity in the power grid equipment ledger, and determining the device with a power flow value greater than the rated transmission capacity as an overloaded device.
[0024] It should be noted that in the repeated power outage verification, for any two power outage plans for the same equipment, the one whose start time is no earlier than the end time of the previous plan is taken as the subsequent plan and recorded as the 1st plan. The plan has its start and end times denoted as follows: and The end time of the previous plan is Power outage intervals according to The calculation determines a hit if the power outage interval is less than the repeating power outage reference interval. The repeating power outage reference interval is maintained by equipment type and voltage level in the rule base maintenance interface, and retrieved during verification by equipment type and voltage level. In equipment overload verification, the overload degree is determined by... Calculation, where The equipment obtained by superimposing maintenance methods after power flow calculation Trend value, For equipment in the power grid equipment ledger Rated transmission capacity, When the value is greater than 1, the device is determined. Overload. The power outage interval is calculated based on the end time of the previous plan, and two plans with overlapping time intervals will not be misjudged as duplicate power outages; the overload check performs power flow calculation on the graph structure data after superimposing maintenance mode, which can detect equipment overloads that do not exist under normal operation mode but only under maintenance mode, thus solving the problem of missed overload detection caused by the disconnect between maintenance schedule and operation mode.
[0025] As an optional embodiment, in step S5: The N-1 anticipatory analysis specifically involves: going through the single component fault anticipation one by one on the graph structure data of the superimposed maintenance method, updating the connectivity relationship, and statistically analyzing the undervoltage bus, the substation with complete shutdown, the main equipment in the tripped station, and the important users with power loss. The determination of important users who have lost power is as follows: the number of power supply lines of each important user is checked according to the power supply ledger of important users. If all power supply lines are lost, it is determined as a complete power outage. If some power supply lines are lost, it is determined as a power supply impact. The reduced load is the sum of the loads carried by the undervoltage busbar minus the loads that can be transferred via the transfer path; The risk classification is as follows: based on the load reduction level, the number and voltage level of substations with complete shutdowns, the tripping of main equipment in the substation and the power outage of important users, the risk level is determined by referring to the fifth, sixth and seventh level clauses of the power grid safety accident classification clauses, and the adjustment direction required to reduce the risk level is output.
[0026] It should be noted that the load reduction is based on... The calculation involves summing the loads carried by the undervoltage busbars and subtracting the loads that can be transferred via the transfer path. The number of power supply lines for important users is taken from the important user power supply ledger, and each power supply line is checked against the superimposed maintenance mode to determine whether it has lost power. When classifying risks, the load reduction level is determined by referring to the classification limits corresponding to Level 5, Level 6, and Level 7 clauses. These classification limits are taken from the State Grid's safety accident level classification clauses and are updated with the current regulations. For substations experiencing complete outages, the number and voltage level are compared against the corresponding clauses; for important user outages, the user level is compared against the corresponding clauses. The adjustment direction is given according to the type of clause that is triggered: when load reduction triggers, the direction for planning or rescheduling is given if the error constitutes a simultaneous outage risk; when an important user outage triggers, the direction for restoring the important user's power supply line is given. The verification of the number of power supply lines distinguishes between complete outages and power supply impacts to avoid overestimating or underestimating the overall risk of important users; the reduced load is deducted from the load that can be transferred via the transfer path to avoid including the transferable portion in the reduction and thus overestimating the risk level, ensuring that the classification result is consistent with the actual power supply capacity under the maintenance mode.
[0027] As an optional embodiment, in step S6, the solution optimization scheme is specifically as follows: For conflicts with time validity, the power outage time interval will be corrected to a valid time interval; For conflicts arising from work resumption during holidays, the power outage time range will be shifted to the nearest working day outside of statutory holidays; For conflicts exceeding the standard construction period, the planned construction period will be compressed to the standard construction period corresponding to the voltage level, equipment type and maintenance nature; For equipment mutual exclusion, simultaneous shutdown, and conflict between the section and important users, the power outage plan for the relocation of conflicting power outages ensures that the corresponding power outage time intervals do not overlap. If the risk level is hit, the power outage plans that pose a risk of simultaneous outage will be staggered or rescheduled. When there are multiple time conflicts in the same power outage plan, the various adjustments will be combined into a single comprehensive time adjustment recommendation; When solving the problem, first handle the conflicts corresponding to the rules with mandatory constraint strength, and then handle the conflicts corresponding to the rules with suggested constraint strength.
[0028] It should be noted that during translation adjustment, the first... The relocation amount of the power outage plan is denoted as The adjusted power outage time interval is the start time. and the end time Compressing the project duration involves adjusting the end date to bring the planned project duration to a shorter timeframe. This equals the standard construction period, which is retrieved from the standard construction period database by voltage level, equipment type, and maintenance nature. The overlap between any two planned time intervals is calculated as follows: calculate, If the value is not greater than zero, the two plans are considered non-overlapping. The set of statutory holidays is denoted as... To avoid the constraints of holiday shifting, the adjusted start time does not belong to... The objective is to minimize the sum of the absolute values of all planned translations, i.e. Under this objective, the conflict elimination corresponding to the rule items with mandatory constraint strength and the conflict elimination corresponding to the rule items with suggested constraint strength are satisfied in sequence. The solution order of handling mandatory conflicts first and then suggested conflicts ensures that anomalies involving power grid safety and compliance are eliminated first; with the goal of minimizing the sum of the absolute values of the shifts, the optimization will minimize the disturbance to the existing planning arrangements and reduce the scope of the chain adjustment of the approved related plans.
[0029] As an optional embodiment, in step S6, the construction of the coupling time constraint specifically involves: The equipment set with the same cross-section is determined according to the equipment cross-section combination relationship, and the power outage time intervals of each power outage plan in the equipment set with the same cross-section are constrained to be non-overlapping. The equipment pairs corresponding to two main transformers, two busbars, two circuits on the same pole and multiple circuits in the same power supply area are constructed as the same stop constraint. The power outage time intervals of the power outage plans corresponding to the same stop constraint do not overlap. After shifting the outage time interval of any outage plan, the conflict status of the outage plans that have coupled time constraints with it is re-determined on the graph structure data.
[0030] It should be noted that the determination of the set of equipment on the same cross-section is as follows: read the equipment cross-section combination relationship, and group the equipment belonging to the same cross-section into one set; the non-overlapping constraint is as described above. The formulaic expression for any two plans within the set. Not greater than zero. Equipment pairs such as dual main transformers, dual busbars, dual-circuit lines on the same pole, and multiple lines supplying power to the same power supply area are identified by the connection relationships in the diagram structure data: main transformers located in the same substation and operating in parallel, two busbar sections connected to the same substation, lines on the same pole with consistent tower erection relationships, and multiple lines supplying power to the same power supply area constitute equipment pairs respectively. The re-determination of conflict states only applies to plans with coupling time constraints to the relocation plan, and does not repeat the verification of all plans. Equipment pairs that are shut down together are automatically identified by the connection relationships in the diagram structure data, eliminating the need for manual maintenance of the equipment list for shutdown together. Constraints are updated accordingly after equipment changes. The re-determination of conflict states only covers plans with coupling time constraints. The computational workload increases with the coupling range rather than with the total number of plans, supporting the optimization and solution of large-scale plans during peak maintenance periods.
[0031] As an optional embodiment, in step S6, for risks that cannot be eliminated by time shifting, a special maintenance method is recommended: Search the graph structure data for devices that are connected to the planned power outage equipment within the same power outage range and whose power outage impact range is covered by the planned power outage equipment, and designate them as accompanying power outage devices; or search for switchable switching paths using the power transfer path search method, and generate special maintenance method suggestions. The maintenance methods corresponding to the special maintenance method suggestions are overlaid onto the graph structure data, and the N-1 anticipatory analysis is re-executed. After verifying that the risk rating has been eliminated, the special maintenance method suggestions are incorporated into the time optimization suggestions.
[0032] It should be noted that the search for accompanying shutdown equipment is performed on the graph structure data according to the connection relationship: starting from the planned power outage equipment, traversing the equipment within the same power outage range, listing the equipment whose power outage impact range is covered by the planned power outage equipment as accompanying shutdown candidates, and confirming that no new undervoltage busbars are added after the accompanying shutdown. Risk mitigation verification is as follows: after overlaying the special maintenance method onto the graph structure data, step S5 is re-executed, and the load reduction is performed as described above. Recalculate the formula. When the risk level drops below the next threshold of the hit clause, the risk is eliminated. If the risk cannot be eliminated in the time dimension, special maintenance methods such as shutdown or switching are given in the operation mode dimension to expand the range of adjustable risk hits. The recommended methods are incorporated into the suggestions only after being superimposed and re-evaluated to eliminate the risk, so as to avoid the adjustment of the method itself introducing new risks of undervoltage or overload.
[0033] As an optional embodiment, in step S7, the manual adoption or revision specifically involves: displaying a comparison between the proposed time and the suggested time of each power outage plan and the suggested conclusions, and receiving the option to adopt a single suggestion, adopt multiple suggestions in batches, revise the suggested time, or temporarily not adopt the suggestion; Return the adopted or revised power outage plan to step S4 for re-verification, and count the number of anomalies that hit the rule items according to the ticket, maintenance, and N-1 dimensions respectively; In each cycle, only the adoption or revision results that have been re-verified and do not add new mandatory anomalies are written into the power outage plan. The cycle exits when the number of anomalies corresponding to the rule item with mandatory constraint strength is zero; otherwise, step S6 is executed. In step S8, the verification conclusions are summarized according to the ticket, maintenance, and N-1 dimensions, including the number of anomalies and handling suggestions for each dimension that hit the rule items. The risk description includes the risk level, the calculation results of the undervoltage bus and reduced load that constitute the risk level, and the adjustment path to reduce the risk level. The optimized power outage plan includes a schedule that has been manually adopted or revised.
[0034] It should be noted that in each iteration, the number of anomalies corresponding to rules with mandatory constraints is recorded as follows: The number of anomalies corresponding to the rule terms with constraint strength equal to the suggested rule term is denoted as The condition for exiting the loop is =Zero. Suggestions adopted manually are added to the power outage plan after further verification to ensure no new mandatory anomalies are added; suggestions not adopted are retained until the next cycle to regenerate time optimization suggestions. Each cycle only includes adjustments that do not add new mandatory anomalies, ensuring monotonous convergence of the closed-loop process and preventing an increase in risk level after optimization; anomaly counts are quantified by dimension to quantify the progress of each round of defect elimination, allowing auditors to understand the degree of closed-loop convergence. Verification conclusions summarize the number of anomalies and handling suggestions by ticket, maintenance, and N-1 dimensions, allowing auditors to pinpoint the source of anomalies by dimension; risk descriptions include calculation results for undervoltage bus and reduced load, as well as degradation paths, ensuring that the rating conclusions are reviewable and the adjustment direction is executable.
[0035] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for intelligent optimization of power outage plans based on a power grid cognitive computing framework, characterized in that, Includes the following steps: S1. Construct a power grid cognitive computing framework, which organizes the power grid equipment topology, power grid equipment ledger, equipment measured load, equipment cross-sectional combination relationship and important user power supply ledger, consisting of substations, busbars, lines, main transformers and connection relationships, into a computable graph structure data with equipment as nodes and connection relationships as branches, forming a cognitive layer. In the maintenance mode, planned power outage equipment is superimposed on the graph structure data, and the cognitive layer supports the undervoltage determination, power transfer path search and N-1 prediction analysis. S2. Establish a rule configuration environment, bind the rule items in the ticket rule library, the orchestration rule library and the N-1 risk rule library according to the dimensions composed of plan type, voltage level, scheduling level, equipment type and constraint strength, and configure the start and stop status of each rule item. The constraint strength is divided into two levels: mandatory and suggested. S3. Obtain the power outage plan, load the binding and effective rule items in the ticket rule library, and perform data governance and verification on the power outage plan. The data governance includes matching and objectifying the equipment and power grid model data in the power outage plan, verifying the time rationality and ticket correctness. If the verification fails, the plan data is corrected and re-governed. If the verification passes, proceed to step S4. S4. Bind the effective arrangement rule set according to the plan type of the power outage plan, and perform maintenance verification on the cognitive layer. The maintenance verification includes equipment mutual exclusion and simultaneous outage verification, cross-section constraint verification, window period and holiday verification, power supply protection verification, standard construction period verification, repeated power outage verification, important user power supply impact verification and equipment overload verification. S5. Based on the verification under maintenance, driven by the rule items bound and effective in the N-1 risk rule base, the maintenance method is superimposed on the cognitive layer to perform N-1 pre-analysis, and the results of the voltage loss judgment and the transfer path search are aggregated to obtain the load reduction. The power outage plan is then risk-classified according to the power grid safety accident level classification clause. S6. When the power outage plan hits a scheduling conflict or risk rating, and the rule item corresponding to the hit anomaly is a time-adjustable rule item, it is determined that intelligent optimization is possible. Under the constraint of the rule item bound by dimension in step S2, the cross-section constraint and the same-outage constraint are transformed into coupled time constraints on the graph structure data. The optimization scheme is solved with the goal of minimizing the sum of the power outage time interval translation of each power outage plan. After the power outage time interval of any power outage plan is translated, the conflict status of the other power outage plans is updated in linkage on the cognitive layer to generate the time optimization suggestion of the power outage plan. S7. Receive the manual adoption or revision results of the time optimization suggestions, return the adopted or revised power outage plan to step S4 for verification again, and perform verification and optimization repeatedly until the anomaly corresponding to the rule item with mandatory constraint strength is eliminated. S8. When the power outage plan has no scheduling conflicts and no risk rating is hit, or after passing the cyclical verification in step S7, output the verification conclusion, the optimized power outage plan and risk description.
2. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 1, characterized in that, In step S1, under maintenance mode, the planned power outage equipment is superimposed onto the graph structure data, specifically as follows: The operating status of the nodes and branches corresponding to the power outage equipment involved in the power outage plan is marked as maintenance status in the graph structure data; The connectivity of the graph structure data is updated using nodes and branches marked as being under maintenance as boundaries. The updated graph structure data is traversed from each power source point, and buses that cannot maintain connectivity with any power source point are identified as under-voltage buses. Starting from the undervoltage bus, a transfer path is generated by searching for a reachable path to a normal energized power source point through the standby communication equipment. The transferable load of the transfer path is determined according to the minimum rated transmission capacity of each device on the reachable path.
3. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 2, characterized in that: In step S2, the binding is performed according to the plan type dimension. Specifically, the enabled rule items are bound separately for annual plans, monthly plans, weekly plans and daily plans in the same rule base. The annual plan is bound to the ticket rule items and the repeated power outage rule items. The monthly plan and the weekly plan are bound to the cross-section constraint rule items, equipment mutual exclusion and simultaneous shutdown rule items and N-1 risk rule items on the basis of the binding result of the annual plan. Each rule item's detailed maintenance includes its name, description, implementation code, and validity flag. Rule items with an invalid validity flag are not included in the binding process. The ticket rule base includes rule items for work repetition verification, time validity verification, plan correlation verification, terminology standardization verification, plant and equipment consistency verification, and hourly user count verification. The orchestration rule base includes rule items for equipment mutual exclusion and simultaneous shutdown, cross-sectional constraints, window periods and holidays, power supply protection, standard construction period, repeated power outages, impact of power supply to important users, and equipment overload; The N-1 risk rule base includes rule items for N-1 contingency planning, pressure loss and power transfer analysis, power reduction load aggregation, and risk rating. The ticket rule base, the arrangement rule base, and the N-1 risk rule base drive the planned data governance in step S3, the maintenance verification in step S4, and the N-1 risk analysis in step S5, respectively. Each step only loads the rule items that are bound and effective in the corresponding rule base.
4. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 3, characterized in that, In step S3, the equipment in the power outage plan is matched with the power grid model data and objectified, specifically as follows: The power outage plan and the power grid model data are matched at the plant / station level using the plant / station name, and at the device level using the dual device naming, the matching is performed at the plant / station where the matching is successful. Output the abnormal view of unmatched plants and stations and unmatched equipment, and re-execute the matching after receiving the manually supplemented mapping relationship; Devices that have not been objectified will be removed from the scope of topology rule item verification. These devices will only participate in ticket rule item verification.
5. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 4, characterized in that, In step S4: The repeated power outage verification is as follows: for any two power outage plans of the same device in the power outage plan set, the difference between the start time of the later power outage plan and the end time of the previous power outage plan is used as the power outage interval. If the power outage interval is less than the repeated power outage reference interval corresponding to the device, it is determined that a repeated power outage has occurred. The repeated power outage reference interval is read from the rule items bound to the effective rules in the orchestration rule base according to the equipment type and voltage level to which the device belongs. The equipment overload verification is performed by superimposing maintenance methods on the cognitive layer, performing power flow calculation, reading the power flow value of each device in the section and the rated transmission capacity in the power grid equipment ledger, and determining the device with a power flow value greater than the rated transmission capacity as an overloaded device.
6. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 5, characterized in that, In step S5: The N-1 anticipatory analysis specifically involves: going through the single component fault anticipation one by one on the graph structure data of the superimposed maintenance method, updating the connectivity relationship, and statistically analyzing the undervoltage bus, the substation with complete shutdown, the main equipment in the tripped station, and the important users with power loss. The determination of important users who have lost power is as follows: the number of power supply lines of each important user is checked according to the power supply ledger of important users. If all power supply lines are lost, it is determined as a complete power outage. If some power supply lines are lost, it is determined as a power supply impact. The reduced load is the sum of the loads carried by the undervoltage busbar minus the loads that can be transferred via the transfer path; The risk classification is as follows: based on the load reduction level, the number and voltage level of substations with complete shutdowns, the tripping of main equipment in the substation and the power outage of important users, the risk level is determined by referring to the fifth, sixth and seventh level clauses of the power grid safety accident classification clauses, and the adjustment direction required to reduce the risk level is output.
7. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 6, characterized in that, In step S6, the solution for optimization is specifically as follows: For conflicts with time validity, the power outage time interval will be corrected to a valid time interval; For conflicts arising from work resumption during holidays, the power outage time range will be shifted to the nearest working day outside of statutory holidays; For conflicts exceeding the standard construction period, the planned construction period will be compressed to the standard construction period corresponding to the voltage level, equipment type and maintenance nature; For equipment mutual exclusion, simultaneous shutdown, and conflict between the section and important users, the power outage plan for the relocation of conflicting power outages ensures that the corresponding power outage time intervals do not overlap. If the risk level is hit, the power outage plans that pose a risk of simultaneous outage will be staggered or rescheduled. When there are multiple time conflicts in the same power outage plan, the various adjustments will be combined into a single comprehensive time adjustment recommendation; When solving the problem, first handle the conflicts corresponding to the rules with mandatory constraint strength, and then handle the conflicts corresponding to the rules with suggested constraint strength.
8. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 7, characterized in that, In step S6, the construction of the coupling time constraint is specifically as follows: The equipment set with the same cross-section is determined according to the equipment cross-section combination relationship, and the power outage time intervals of each power outage plan in the equipment set with the same cross-section are constrained to be non-overlapping. The equipment pairs corresponding to two main transformers, two busbars, two circuits on the same pole and multiple circuits in the same power supply area are constructed as the same stop constraint. The power outage time intervals of the power outage plans corresponding to the same stop constraint do not overlap. After shifting the outage time interval of any outage plan, the conflict status of the outage plans that have coupled time constraints with it is re-determined on the graph structure data.
9. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 8, characterized in that, In step S6, for risks that cannot be eliminated by time shifting, the following special maintenance methods are recommended: Search the graph structure data for devices that are connected to the planned power outage equipment within the same power outage range and whose power outage impact range is covered by the planned power outage equipment, and designate them as accompanying power outage devices; or search for switchable switching paths using the power transfer path search method, and generate special maintenance method suggestions. The maintenance methods corresponding to the special maintenance method suggestions are overlaid onto the graph structure data, and the N-1 anticipatory analysis is re-executed. After verifying that the risk rating has been eliminated, the special maintenance method suggestions are incorporated into the time optimization suggestions.
10. The intelligent optimization method for power outage planning based on a power grid cognitive computing framework according to claim 9, characterized in that: In step S7, the manual adoption or revision specifically involves: displaying a comparison between the proposed time and the suggested time for each power outage plan and the suggested conclusions, and receiving the option to adopt a single suggestion, adopt multiple suggestions in batches, revise the suggested time, or temporarily not adopt the suggestion; Return the adopted or revised power outage plan to step S4 for re-verification, and count the number of anomalies that hit the rule items according to the ticket, maintenance, and N-1 dimensions respectively; In each cycle, only the adoption or revision results that have been re-verified and do not add new mandatory anomalies are written into the power outage plan. The cycle exits when the number of anomalies corresponding to the rule item with mandatory constraint strength is zero; otherwise, step S6 is executed. In step S8, the verification conclusions are summarized according to the ticket, maintenance, and N-1 dimensions, including the number of anomalies and handling suggestions for each dimension that hit the rule items. The risk description includes the risk level, the calculation results of the undervoltage bus and reduced load that constitute the risk level, and the adjustment path to reduce the risk level. The optimized power outage plan includes a schedule that has been manually adopted or revised.
