A future-state dynamic generation and optimal scheduling system for power grid maintenance

Abstract

The present invention discloses a future-state dynamic generation and optimal scheduling system for power grid maintenance in the technical field of power grid maintenance, including a multi-time plan input module, a maintenance mode adjustment module, which dynamically adjusts the maintenance plan of the power grid based on the base-state data; a safety analysis module, which is used to verify the stability of the power grid, conduct static safety analysis, and calculate the power flow distribution of the power grid; an iteration module, which is used to dynamically correct the maintenance plan according to the risk result data and generate the final plan; a preview verification module, which is used to simulate the operation state of the power grid under complex conditions of new equipment commissioning and maintenance; by integrating multi-source data such as commissioning plans, maintenance plans, and load forecasts in different time dimensions, the base-state maintenance mode, future power grid change sections, and measurement data time series sections are generated, and a detailed base-state data is constructed, providing comprehensive, accurate, and dynamically updated data support for subsequent links such as power grid risk assessment, maintenance strategy formulation, and safety analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid maintenance, and more specifically, to a future state dynamic generation and optimization scheduling system for power grid maintenance. Background Art

[0002] With the development of society, the scale of power grids continues to expand, and the types of power equipment are becoming more and more diverse and widely distributed. The safe and stable operation of power grids is becoming more and more important to social production and life. However, power equipment is prone to failures due to manufacturing defects, installation problems, insulation aging, etc. in long-term operation. Traditional power grid maintenance is mostly based on fixed cycles or relies on manual inspections to passively respond to failures, which is difficult to adapt to the high standards and strict requirements of modern power grids for safe and reliable power supply. In recent years, although there have been technical assistance such as sensors and information systems, the amount of data has exploded. Traditional methods have many shortcomings in data fusion, analysis and risk assessment, and it is difficult to meet the needs of refined and intelligent operation and maintenance of power grids.

[0003] The risk assessment is not accurate and comprehensive enough. Existing methods are mostly based on single-dimensional indicators, such as the current status of equipment or simple load forecasts, ignoring the comprehensive consideration of multi-source data, resulting in inaccurate risk assessment results and unable to fully reflect the actual risk status of the power grid. On the other hand, the maintenance plan lacks flexibility. The maintenance plan based on a fixed time window or empirical rules is difficult to adjust dynamically according to real-time risks. In addition, the multi-objective optimization capability is insufficient. Existing technologies focus on single resource optimization, such as only focusing on labor costs or spare parts inventory, without comprehensively considering multiple factors such as maintenance time and risk level, resulting in unreasonable resource allocation and inefficiency. Furthermore, the rehearsal verification function is weak, and the simulation of complex working conditions is insufficient, making it difficult to identify potential hidden dangers in advance, especially in extreme scenarios, which can easily lead to more serious consequences due to the spread of faults. Finally, data fusion and analysis are difficult. Different business systems are independently developed and constructed, which makes data fusion difficult. In addition, the quality of data collected by early sensors is uneven, and there are measurement errors, data anomalies, loss or duplication, which affects the effective use of data and analysis results. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a future state dynamic generation and optimization scheduling system for power grid maintenance.

[0005] The present invention provides a future state dynamic generation and optimization scheduling system for power grid maintenance, including a multi-time plan input module for integrating production plans, maintenance plans and load forecasts in different time dimensions, generating base state maintenance methods, future power grid change sections and measurement data time series sections, and obtaining base state data;

[0006] The maintenance mode adjustment module dynamically adjusts the maintenance plan of the power grid based on the baseline data;

[0007] A safety analysis module, which is used to verify the grid stability, conduct static safety analysis, and calculate the grid power flow distribution;

[0008] A risk discrimination module, which is used to generate the future state under the single maintenance work schedule, generate the future state grid section, evaluate the maintenance schedule risk, discriminate the time-series grid risk brought by the maintenance power outage, and obtain the risk result data;

[0009] An iteration module, which is used to dynamically modify the maintenance plan according to the risk result data and generate the final plan;

[0010] A preview verification module, which is used to simulate the grid operation state under the complex working conditions of new equipment commissioning and maintenance.

[0011] Preferably, the multi-time plan input module includes a commissioning plan unit, a maintenance plan unit, and a load forecasting unit;

[0012] The commissioning plan unit is used to formulate the commissioning plan for different stages, and then form the future grid change section according to the formulated commissioning plan;

[0013] The maintenance plan unit is used to formulate the maintenance plan for different stages, and then conduct the analysis of the base-state maintenance mode to form the planned base-state section, specifically including the parameter data in the case of grid equipment maintenance;

[0014] The load forecasting unit is used to formulate the load threshold for different stages, and then conduct power generation forecasting and low-voltage bus forecasting to form the future measurement data time-series section.

[0015] Preferably, the specific working steps of the maintenance mode adjustment module are as follows:

[0016] Obtain the base-state data, construct a spatio-temporal sensitive maintenance index to determine the priority maintenance order of equipment;

[0017] Substitute the collected data into the spatio-temporal sensitive maintenance index to calculate the spatio-temporal sensitive maintenance index value of each equipment.

[0018] Sort the equipment according to the magnitude of the spatio-temporal sensitive maintenance index value. The larger the spatio-temporal sensitive maintenance index value, the higher the priority maintenance level of the equipment, and the maintenance should be arranged preferentially to determine the priority maintenance order of the equipment.

[0019] Preferably, the specific working steps of the maintenance mode adjustment module further include the following:

[0020] For spatio-temporal collaborative conflict resolution, solve the maintenance conflict, specifically including two constraint conditions: spatial coupling constraint and time elasticity constraint. The spatial coupling constraint identifies conflicts by considering the geographical proximity coefficient between equipment and the electrical coupling degree and geographical topology; the time elasticity constraint determines the time range of the maintenance window based on the elasticity coefficient generated by meteorological forecasting.

[0021] Preferably, the specific working steps of the maintenance method adjustment module further include the following;

[0022] Elastic resource dynamic adaptation, constructing a multi-objective resource optimization model, maximizing the sum of the maintenance time multiplied by the resource reduction coefficient, and at the same time minimizing the sum of the product of the labor cost and the square of the distance between the equipment geographical coordinates and the team station location as the objective function.

[0023] Preferably, the specific steps for the safety analysis module to perform static safety analysis are as follows:

[0024] Static safety analysis aims to verify the stability of the power grid after the maintenance method adjustment under a single equipment fault (N-1). First, perform N-1 fault enumeration, that is, traverse all maintenance equipment and associated lines, simulate the power grid topology after their outage, and generate a new adjacency matrix by removing the corresponding equipment;

[0025] Then carry out stability verification, including voltage stability verification and line overload test. Finally, perform probabilistic risk assessment, calculate the system instability probability in combination with the equipment failure rate. Due to the problem of combinatorial explosion in large-scale power grid scenarios, Monte Carlo sampling or heuristic pruning optimization can be used for calculation, and output the risk equipment list and corresponding over-limit indicators.

[0026] Preferably, the safety analysis module is also used to perform power grid power flow calculation, and the specific steps are as follows:

[0027] Generating the future state under a single maintenance operation sequence aims to construct a time-tagged power grid future state sequence to support dynamic risk analysis;

[0028] First, perform time-section slicing, decompose the maintenance plan according to time slices, generate a topology change sequence, considering the outage / resumption operations of the maintenance equipment and the time-sequence changes of load forecasting;

[0029] Then use the state prediction model to predict the future state parameters;

[0030] Finally, use the time-series database to store these section snapshots for quick backtracking and comparison, and output a set of future state power grid models with time stamps.

[0031] Preferably, the specific steps of the risk discrimination module are as follows:

[0032] The goal of the time-series power grid risk discrimination brought by maintenance power outage is to quantify the risk levels at different time points during the maintenance process and identify the cascading fault chain;

[0033] First, the temporal risk chain modeling is carried out and a Bayesian network is constructed to characterize the causal relationship between equipment failure, load transfer, and protection action. Node A is the initial maintenance power outage event, and node B and C are the subsequent affected equipment or load changes.

[0034] Then, the risk indicators, such as the probability of load loss, are calculated by judging the relationship between the load and the available power in each period;

[0035] Then, a risk heat map is generated, which integrates overload rate, voltage deviation, and new energy fluctuation factors, and adds them according to the weights to obtain the risk value of each location and time;

[0036] Then, based on the graph theory algorithm, the critical path is identified and the maintenance path with the largest impact range is determined.

[0037] Preferably, the specific steps of the iteration module dynamically revising the maintenance plan according to the risk result data and generating the final plan are as follows:

[0038] Starting from the initial maintenance plan, safety analysis and risk identification are carried out in sequence. If the risk identification results meet the standards, the maintenance plan is executed. If the risk identification results do not meet the standards, the dynamic adjustment phase is entered to revise the maintenance plan, and then safety analysis and risk identification are carried out again until the results meet the standards.

[0039] Preferably, the specific working steps of the preview verification module are as follows:

[0040] Generate a variety of extreme scenarios, including a sudden drop in renewable energy output and a typhoon passing through, and calculate the weight of each scenario, taking into account the scenario deviation and correlation coefficient;

[0041] The device activity queue is divided based on the grouping matrix, corresponding to the set of operable devices in a specific time window.

[0042] Beneficial effects: By integrating multi-source data such as commissioning plans, maintenance plans and load forecasts in different time dimensions, the base state maintenance method, future power grid change sections and measurement data time series sections are generated, and detailed base state data is constructed, providing comprehensive, accurate and dynamically updated data support for subsequent power grid risk assessment, maintenance strategy formulation and safety analysis, ensuring that each analysis module can work based on a unified and complete data foundation, thereby improving the scientificity and accuracy of the entire power grid operation and maintenance decision-making;

[0043] Existing technologies in power grid maintenance resource allocation mostly focus on optimizing a single objective, such as focusing only on optimizing labor costs or spare parts inventory, and lack effective methods for comprehensive optimization of multiple objectives. This step, by establishing a multi-objective resource optimization model, takes into account multiple factors such as maintenance time, labor costs, spare parts demand index, and risk level, and can achieve dynamic adaptation and comprehensive optimization of resources, which is more comprehensive and advanced.

[0044] Traditional power grid maintenance resource allocation methods often struggle to achieve multi-objective optimization of resources and cannot well balance multiple factors such as maintenance time, labor costs, and spare part requirements, leading to problems such as unreasonable resource allocation and low efficiency. This step solves the problem of insufficient optimization of maintenance resources in the existing technology and realizes the dynamic adaptation and comprehensive optimization of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] As Figure 1 shown: A future-state dynamic generation and optimal scheduling system for power grid maintenance includes a multi-time plan input module for integrating production plans, maintenance plans, and load forecasts in different time dimensions to generate a base-state maintenance mode, future power grid change sections, and measurement data time series sections, obtaining base-state data;

[0047] It should be noted that through base-state analysis (such as equipment status and load forecasting), the initial topology and operating state of the future power grid are formed, providing a data basis for subsequent adjustments;

[0048] A maintenance mode adjustment module for dynamically adjusting the power grid maintenance plan based on the base-state data;

[0049] It should be noted that an adjusted maintenance plan is generated by combining the "base-section of the maintenance plan" and the "future measurement data time series section";

[0050] A safety analysis module for verifying the stability of the power grid, performing static safety analysis, and calculating the power grid power flow distribution;

[0051] It should be noted that ensure that the power grid after the adjustment of the maintenance mode meets the safety operation constraints;

[0052] A risk discrimination module for generating the future state under the single-maintenance work time series, generating the future-state power grid section, evaluating the maintenance time series risk, and discriminating the time series power grid risk brought by the maintenance power outage, obtaining risk result data;

[0053] It should be noted that a future-state section with time series tags is generated to quantify the risk level;

[0054] An iteration module for dynamically correcting the maintenance plan according to the risk result data to generate the final plan;

[0055] It should be noted that dynamic optimization is achieved through multiple "iterations", and finally medium / long-term / short-term / ultra-short-term maintenance plans are output;

[0056] The rehearsal verification module is used to simulate the operation status of the power grid under complex working conditions such as the commissioning and maintenance of new equipment.

[0057] It should be noted that the grid operation preview based on future state sections covers extreme scenarios;

[0058] It should also be noted that the risk identification result triggers the adjustment of the maintenance mode, and the future state generation and risk identification module to the maintenance mode adjustment module;

[0059] After the preview verification finds problems, return to iterative optimization, from the preview verification module to the iterative optimization module;

[0060] The "future measurement data time series section" generated by the data preparation module provides input for the adjustment of maintenance methods;

[0061] The power flow calculation results of the safety analysis module are directly used for future state generation and risk identification;

[0062] During the high-risk periods identified by the risk identification module, maintenance resources are reallocated through the iterative optimization module;

[0063] The simulation results of the preview verification module are fed back to the data preparation module to update the base state section;

[0064] Medium- and long-term plans provide framework constraints for short-term / ultra-short-term, and ultra-short-term real-time data correct medium- and long-term forecasts.

[0065] As an optional embodiment: the multi-time plan input module includes a production plan unit, a maintenance plan unit and a load forecasting unit;

[0066] It should be noted that the following different stages include medium-term, short-term and ultra-short-term. In this embodiment, medium-term is 6 to 18 months, short-term is 1 to 6 months, and ultra-short-term is hourly real-time data.

[0067] Establish an incremental model library containing commissioning / decommissioning plans, and store the commissioning time, topological connection relationship and parameter data of power grid equipment in the medium term (6-18 months), short term (1-6 months) and ultra-short term (hourly level);

[0068] Adopting incremental storage, each version only stores the changed data relative to the previous version (such as new substations and line decommissioning), reducing storage redundancy;

[0069] Extract model data of mid-term / short-term production equipment (such as node number and connection relationship of new substation), and automatically splice it with the current real-time model through the graph model node number (such as type+co_no+fac_node) to form the full model data;

[0070] Mark the maintenance equipment as out of service and set the corresponding switch disconnection logic in the model;

[0071] Convert the medium-term / short-term load thresholds and power generation prediction data into a bus load prediction file (such as an E-format file) and associate it with the corresponding nodes in the full model;

[0072] The commissioning plan unit is used to formulate commissioning plans for different stages, and then form a future power grid change section according to the formulated commissioning plan;

[0073] It should be noted that for newly added equipment such as wind farms and transformers, the shortest grid connection path is determined through in-depth topological search:

[0074] Search for the connection switch path to the energized bus, and close the switch with the shortest path to achieve automatic grid connection;

[0075] The newly added wind farm automatically closes the tie switch and connects to the power grid by searching for the 220 kV bus path of the adjacent substation;

[0076] Use the MapReduce framework to decompose the future section generation task into multi-period parallel calculations:

[0077] Each period independently executes plan data decomposition (such as power generation plan allocation), power flow calculation, outputs section results, summarizes the section data of each period, and generates a time-series section file;

[0078] The maintenance plan unit is used to formulate maintenance plans for different stages, and then conduct a basic state maintenance mode analysis to form a planned basic state section, which specifically includes parameter data under the condition of power grid equipment maintenance;

[0079] It should be noted that obtain the current status information of the equipment from the equipment status monitoring system or historical maintenance records, such as the operation duration, fault history, performance indicators, etc. of the equipment, to determine whether the equipment is in a normal, abnormal or serious state;

[0080] Collect the real-time operation data of the power grid from the power grid real-time monitoring system (such as the SCADA system), including key parameters such as bus voltage, unit output, and line power flow;

[0081] Inject the collected real-time operation data of the power grid into the whole network model as the initial operation mode to ensure that the model can accurately reflect the current actual operation state of the power grid;

[0082] After determining the maintenance strategy and equipment outage status, call a professional power flow calculation engine (such as BPA / PSASP, etc.) to perform a basic state power flow calculation on the whole network model injected with real-time data;

[0083] Obtain the key parameters of voltage distribution and power distribution of the power grid under maintenance through power flow calculation, and form a planned basic state section;

[0084] The load forecasting unit is used to formulate load thresholds for different stages, and then perform power generation forecasting and low-voltage bus forecasting to form a time-series section of future measurement data.

[0085] It should be noted that the ARIMA or LSTM model is used to predict annual / monthly loads, and a meteorological correction factor (such as the temperature cumulative effect coefficient) is superimposed to improve the accuracy.

[0086] Real-time SCADA measurement data is collected, and the predicted values are dynamically corrected through Kalman filtering to generate a minute-level load time-series curve.

[0087] The deviation between the predicted data and the actual measurement is compared. If it exceeds the threshold (such as the error > 5%), an early warning is triggered and the parameters of the prediction model are adjusted.

[0088] A standardized time-series section file (such as CIM / E format) is output, which contains time-series data such as bus loads and new energy output.

[0089] As an optional embodiment: The specific working steps of the maintenance method adjustment module are as follows:

[0090] The base state data is obtained, and a spatio-temporal sensitive maintenance index is constructed to determine the priority maintenance order of equipment.

[0091] The collected data is substituted into the spatio-temporal sensitive maintenance index to calculate the spatio-temporal sensitive maintenance index value of each device.

[0092] The equipment is sorted according to the magnitude of the spatio-temporal sensitive maintenance index value. The larger the spatio-temporal sensitive maintenance index value, the higher the priority maintenance level of the equipment, and maintenance should be arranged first to determine the priority maintenance order of the equipment.

[0093] Considering multiple dimensions of status, and at the same time combining the relationship between the continuous operation duration of the equipment and the average life threshold of similar equipment, the equipment with the highest risk and located on the critical path is processed first to avoid cascading failures; specifically, it includes multiple dimensions of status such as the real-time health of the equipment, the power flow sensitivity factor, and the fault propagation probability.

[0094] It should be noted that in this embodiment, the specific calculation formula of the spatio-temporal sensitive maintenance index is as follows:

[0095] ;

[0096] Among them is the real-time health of the equipment, which is obtained by averaging and weighting the comparison between the power grid equipment parameter data and the corresponding standard value, and comprehensively reflects the current health status of the equipment;

[0097] is the power flow sensitivity factor, and the calculation formula is , the rated capacity divided by the number of overloads reflects the overload tolerance of the device, and the node betweenness reflects the importance of the node in the network, comprehensively measuring the sensitivity of the device in the power flow distribution;

[0098] is the fault propagation probability, generated by simulating the N-2 cascading fault scenario based on the Bayesian network, that is, considering the probability of fault propagation when two devices fail simultaneously;

[0099] is the continuous operation duration of the device, is the average life threshold of similar devices, used to measure the relative relationship between the device operation time and the average life;

[0100] is the preset time decay coefficient, with a value of 0.876 in this embodiment;

[0101] It should be noted that the prior art usually adopts a static weight assignment mode to evaluate the device status and determine the maintenance priority. However, this step breaks through this traditional mode, introduces dynamic weights, and combines factors such as the continuous operation duration of the device and the average life threshold of similar devices to more scientifically reflect the spatio-temporal sensitivity risk of the device, making the determination of the maintenance priority more reasonable and dynamic;

[0102] By constructing a spatio-temporal sensitive maintenance index (TSMI), comprehensively considering the multi-dimensional status of the device, realizing the spatio-temporal correlation modeling of the device risk, and dynamically determining the maintenance priority of the device; this helps to more accurately identify high-risk devices, reasonably arrange maintenance resources, improve the efficiency and pertinence of power grid maintenance, and ensure the safe and stable operation of the power grid;

[0103] The traditional method for determining the maintenance priority often based on a single-dimensional status index is difficult to comprehensively and accurately reflect the actual risk status of the device. This step solves the problems of incomplete and inaccurate device risk assessment and inability to dynamically adapt to spatio-temporal changes in the prior art.

[0104] As an optional embodiment: The specific working steps of the maintenance method adjustment module further include the following:

[0105] is spatio-temporal coordination conflict resolution, solving maintenance conflicts, specifically including two constraint conditions: spatial coupling constraint and time elasticity constraint. The spatial coupling constraint identifies conflicts by considering the geographical proximity coefficient between devices and the electrical coupling degree and geographical topology; the time elasticity constraint determines the time range of the maintenance window based on the elasticity coefficient generated by meteorological prediction. It should be noted that the parameters will be adjusted to compress the window during high-risk periods to achieve multi-dimensional collaborative optimization of the power grid physical constraints and external environmental factors;

[0106] It should also be noted that the specific expression of the spatial coupling constraint is:

[0107] i, j ∈ Ecouple);

[0108] Among them is the geographical proximity coefficient of device and which is composed of the passing time of the maintenance vehicle and the safety isolation duration;

[0109] couple is a set of devices identified jointly through electrical coupling degree and geographical topology, ensuring that devices adjacent in space and closely electrically connected do not conflict in the maintenance time arrangement;

[0110] The time flexibility constraint expression is:

[0111] ;

[0112] Among them represents the maintenance time window of device i;

[0113] Among them is the maintenance window flexibility coefficient generated based on meteorological predictions (typhoons, rain, snow, etc.), used to measure the impact of meteorological conditions on the maintenance time window; is a dynamic adjustment parameter, which decreases by 30% during high-risk periods to compress the window and ensure that the maintenance is carried out within a relatively safe time range;

[0114] According to the expression, by adjusting and values, the maintenance time window of each device can be dynamically determined;

[0115] When considering the maintenance time window in the prior art, the dynamic impact of external environmental factors such as meteorology is often insufficiently considered. Usually, a fixed time window or only simple empirical rules are used for adjustment. After introducing the time flexibility constraint expression, the maintenance time window can be more accurately adjusted dynamically according to meteorological conditions, enhancing the flexibility and adaptability of the maintenance plan, and effectively reducing the maintenance risk and delay risk caused by external factors such as meteorology;

[0116] In the prior art, simple heuristic algorithms or empirical rules are usually adopted in resolving power grid maintenance conflicts, and the physical constraints of the power grid and external environmental factors are not considered comprehensively and systematically enough. However, this step uses a mixed integer spatio-temporal planning model, takes various complex constraint conditions into account, and performs precise solution through mathematical programming methods, and can obtain a better maintenance plan scheme, avoiding the limitations and one-sidedness of traditional methods;

[0117] When dealing with the power grid maintenance plan, the existing technology often has difficulty in fully considering the spatial coupling relationship between devices and the impact of external environmental factors on the maintenance time, resulting in the ineffective resolution of maintenance conflicts. This step specifically addresses these issues and realizes more accurate maintenance conflict resolution by comprehensively considering multiple factors such as the geographical proximity coefficient, electrical coupling degree, and meteorological prediction.

[0118] As an optional embodiment: The specific working steps of the maintenance method adjustment module further include the following;

[0119] Dynamically adapt elastic resources, construct a multi-objective resource optimization model, maximize the sum of the maintenance time multiplied by the resource reduction coefficient, and at the same time minimize the sum of the product of the labor cost and the square of the distance between the device geographical coordinates and the team's stationed location as the objective function. It should be noted that in the constraint conditions, factors such as the spare part demand index, dynamic inventory, and risk level are considered to support the real-time linkage optimization of human resource scheduling, spare part supply chain, and risk level;

[0120] It should also be noted that the objective function of the multi-objective resource optimization model is:

[0121] Determine the maintenance time of each device, which is negatively correlated with the skill level of the personnel, that is, the higher the skill level of the personnel (from LV1 to LV3, the skill weights are 1.0 to 0.6), the shorter the maintenance time. Then, multiply the T maintenance of each device by the corresponding resource reduction coefficient (R reduction), and finally add up the products of all devices to obtain a total. The resource reduction coefficient reflects the degree of resource consumption that can be reduced through optimized resource allocation. The larger this value, the higher the resource utilization efficiency;

[0122] Minimize the sum of the product of the labor cost and the square of the distance between the device geographical coordinates and the team's stationed location: Determine the geographical coordinates of each device and the stationed location of each maintenance team, calculate the square of the distance between device i and team j, and then multiply it by the corresponding labor cost. Finally, add up the products of all device-team combinations to obtain another total. The labor cost represents the labor cost required to arrange the team to perform maintenance at the device. The farther the distance, the higher the labor cost usually is;

[0123] Combining the above two parts, the objective function is to make the first total as large as possible and the second total as small as possible under the premise of meeting the constraint conditions, so as to achieve efficient resource utilization and effective cost control;

[0124] Calculate the sum of the products of the spare part demand indexes of all devices and the corresponding variables. This sum must be less than or equal to the dynamic inventory multiplied by the risk level. The spare part demand index incorporates factors such as the supplier delivery cycle and the urgency of the fault, reflecting the urgency and importance of the spare parts required by the device; the dynamic inventory is the current available inventory of spare parts; the risk level is determined based on factors such as the operating status of the power grid and the external environment, and is used to measure the risk level faced by the current maintenance work. This constraint ensures that while meeting the maintenance requirements, the inventory of spare parts is within a reasonable range, avoiding inventory backlogs or shortages, and also considering the impact of risk factors on the inventory;

[0125] Most of the existing technologies focus on the optimization of a single objective in the allocation of power grid maintenance resources, such as only focusing on the optimization of labor costs or spare part inventory, lacking an effective method for multi-objective comprehensive optimization. This step, by establishing a multi-objective resource optimization model and considering multiple factors such as maintenance time, labor costs, spare part demand index, and risk level, can achieve the dynamic adaptation and comprehensive optimization of resources, and is more comprehensive and advanced compared with others;

[0126] Traditional power grid maintenance resource allocation methods often have difficulty in achieving multi-objective optimization of resources and cannot well balance multiple factors such as maintenance time, labor costs, and spare part demand, resulting in problems such as unreasonable resource allocation and low efficiency. This step solves the problem of insufficient optimization of maintenance resources in the existing technology and realizes the dynamic adaptation and comprehensive optimization of resources.

[0127] As an optional embodiment: The specific steps for the safety analysis module to perform static safety analysis are as follows:

[0128] The purpose of static safety analysis is to verify the stability of the power grid under a single device fault (N-1) after the adjustment of the maintenance mode. First, perform N-1 fault enumeration, that is, traverse all maintenance devices and associated lines, simulate the power grid topology after their outage, and generate a new adjacency matrix by removing the corresponding devices;

[0129] Then carry out stability verification, including voltage stability verification and line overload test. Finally, perform probabilistic risk assessment, calculate the system instability probability in combination with the device failure rate. Due to the problem of combinatorial explosion in large-scale power grid scenarios, Monte Carlo sampling or heuristic pruning optimization can be used for calculation, and output a list of risk devices and the corresponding over-limit indicators.

[0130] It should be noted that the specific steps of the adjacency matrix generation formula are as follows:

[0131] ;

[0132] Among them, represents the adjacency matrix after simulating the outage of the device, the adjacency matrix of the original power grid, Represents the removed faulty device. The specific steps for calculating the system instability probability by this formula for describing the topological structure change of the power grid after the outage of the device due to a fault are as follows:

[0133] ;

[0134] Where Is the threshold for calculating the thermal stability limit;

[0135] For large-scale power grid scenario combination explosion, Monte Carlo sampling or heuristic pruning optimization is used for calculation;

[0136] Among them, Is the system instability probability, Is the failure rate of device , Is an indicator function. When the power flow Of line Exceeds its thermal stability limit , Take 1, otherwise 0. This formula is used to evaluate the possibility of system instability caused by device failure.

[0137] As an optional embodiment: The safety analysis module is also used for power grid power flow calculation, and the specific steps are as follows:

[0138] Generate the future state under a single maintenance work time sequence, aiming to construct a time-tagged power grid future state sequence to support dynamic risk analysis;

[0139] First, perform time-section slicing, decompose the maintenance plan according to time slices, generate a topological change sequence, considering the outage / resumption operations of the maintenance equipment and the time-sequence changes of load forecasting;

[0140] Then, use the state prediction model to predict the future state parameters; it should be noted that, for example, the fluctuations of new energy output, these models are trained based on historical data and relevant influencing factors to predict the power parameters at future moments;

[0141] Finally, use the time-sequence database to store these section snapshots for quick backtracking and comparison, and output a set of future state power grid models with time stamps.

[0142] It should be noted that the topological change sequence formula is as follows:

[0143] ;

[0144] Where Is the power grid topological structure at future time t, which synthesizes the basic state and various changes;

[0145] It is the basic topology of the power grid, which is the initial state during the normal operation of the power grid and is derived from the original planning, design, and operation data of the power grid;

[0146] It is the change caused by the outage or restoration operation of the equipment to be repaired at time t and is determined according to the maintenance plan;

[0147] It is the change in load forecasting at time t. In this embodiment, it can be obtained by a load forecasting model based on historical load data and relevant influencing factors;

[0148] The state prediction model can be a Markov prediction model, and the following specific formula can be adopted to meet the requirements in this embodiment:

[0149] ;

[0150] It is the predicted wind power output at the future time t, which is the result output by the model and reflects the future state estimation of wind power;

[0151] It is a prediction model, such as LSTM or Transformer, which makes predictions by learning patterns in historical data;

[0152] It is the actual value or predicted value of the wind power output at the previous time t−1, providing continuous information in the time series for the model;

[0153] Meteorological data are meteorological factors affecting wind power output, such as wind speed and wind direction. These data are from meteorological observations or forecasts and are important external variables affecting new energy output.

[0154] As an alternative embodiment: The specific steps of the risk discrimination module are as follows:

[0155] The goal of time-sequence power grid risk discrimination caused by maintenance power outages is to quantify the risk levels at different time points during the maintenance process and identify the chain of cascading failures;

[0156] First, perform time-sequence risk chain modeling and construct a Bayesian network to depict the causal relationships among equipment failures, load transfers, and protection actions. Among them, node A is the initial maintenance power outage event, and B and C are subsequent affected equipment or load changes;

[0157] Then calculate risk indicators, such as the probability of load loss, by judging the relationship between the load and available power in each time period;

[0158] Then generate a risk heat map, comprehensively considering factors such as overload rate, voltage deviation, and new energy fluctuations, and adding them according to weights to obtain the risk values at each location and time;

[0159] Based on the graph theory algorithm, identify the critical path and determine the maintenance path with the largest influence range.

[0160] It should be noted that a Bayesian network is constructed to depict the causal relationship among equipment failures, load transfer, and protection actions in the power grid, forming a causal inference network of equipment failure - load transfer - protection action.

[0161] Determine the nodes, where node A is the initial maintenance power outage event. According to the maintenance plan arrangement and historical statistical data such as the average failure rate during equipment maintenance, analyze and obtain the probability P(A) of the initial maintenance power outage event occurring.

[0162] Determine the subsequent nodes B, C, etc. Analyze the conditional probability P(B|A) of the subsequent affected equipment or load mutation event B occurring under the condition that the initial maintenance power outage event A occurs. This needs to be determined based on the electrical connection relationship between equipment, the action logic of protection devices, and the load transfer path and other power grid operation characteristics and historical fault data statistical rules.

[0163] Similarly, under the condition that event B occurs, determine the conditional probability P(C|B) of event C occurring, and so on. The determination of each subsequent conditional probability depends on the power grid structure, operation mode, and reliability parameters of equipment, etc. Calculate the probability of the entire cascading failure occurring, so as to quantify the likelihood of a series of subsequent failures caused by the initial maintenance power outage event;

[0164] Calculate the loss of load probability (LOLP). First, determine the total number of time periods T divided within the entire maintenance time period, for example, divided at intervals such as hours or 15 minutes.

[0165] For each time period t, obtain the load power, which is obtained from the load prediction model combined with the load change trend during the maintenance period.

[0166] At the same time, when determining the available generation power, factors such as the impact of the maintenance arrangement on the output of the generating units and the transmission limitations of the power grid need to be considered.

[0167] Using the indicator function, in the t-th time period, if the load power exceeds the available generation power, the function value is 1, indicating that a loss of load situation occurs in this time period; otherwise, it is 0. Sum the indicator function values of each time period and then divide by the total number of time periods T to obtain the loss of load probability LOLP, that is, it is used to measure the probability that the power grid cannot meet the load demand during the maintenance period, reflecting the degree of insufficiency of the power grid power supply reliability.

[0168] As an optional embodiment: The specific steps for the iterative module to dynamically modify the maintenance plan according to the risk result data and generate the final plan are as follows:

[0169] Starting from the initial maintenance plan, safety analysis and risk discrimination are carried out in sequence. If the risk discrimination result meets the standard, the maintenance plan is executed. If the risk discrimination result does not meet the standard, it enters the dynamic adjustment link to revise the maintenance plan, and then safety analysis and risk discrimination are carried out again until the result meets the standard.

[0170] As an optional embodiment: The specific working steps of the rehearsal verification module are as follows:

[0171] Generate a variety of extreme scenarios, specifically including a sharp drop in new energy output and a typhoon passing through, calculate the weights of each scenario, considering the scenario deviation degree and the correlation coefficient;

[0172] Based on the grouping matrix, divide the equipment activity queue corresponding to the set of operable equipment for a specific time window.

[0173] The above are only the preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements should also be regarded as within the protection scope of this template.

Claims

1. A future-state dynamic generation and optimal scheduling system for power grid maintenance, characterized in that, It includes a multi-time plan input module, which is used to integrate the production plan, maintenance plan, and load forecast in different time dimensions, generate the base maintenance mode, future power grid change section, and measurement data time series section, and obtain the base data; A maintenance mode adjustment module, which dynamically adjusts the power grid maintenance plan based on the base data; A safety analysis module, which is used to verify the power grid stability, conduct static safety analysis, and calculate the power grid power flow distribution; A risk discrimination module, which is used to generate the future state under the single maintenance work time series, generate the future state power grid section, evaluate the maintenance time series risk, discriminate the time series power grid risk brought by the maintenance power outage, and obtain the risk result data; An iteration module, which is used to dynamically correct the maintenance plan according to the risk result data and generate the final plan; A preview verification module, which is used to simulate the power grid operation state under the complex working conditions of new equipment commissioning and maintenance; The multi-time plan input module includes a production plan unit, a maintenance plan unit, and a load forecast unit; The production plan unit is used to formulate the production plan in different stages, and then form the future power grid change section according to the formulated production plan; The maintenance plan unit is used to formulate the maintenance plan in different stages, and then conduct the base maintenance mode analysis to form the planned base section, specifically including the parameter data in the case of power grid equipment maintenance; The load forecast unit is used to formulate the load threshold in different stages, and then conduct power generation forecast and low-voltage bus forecast to form the future measurement data time series section; The specific working steps of the maintenance mode adjustment module are as follows: Obtain the base data, and construct a spatio-temporal sensitive maintenance index to determine the priority maintenance order of equipment; The specific calculation formula of the spatio-temporal sensitive maintenance index is as follows: ； wherein is the real-time health degree of the device, which is obtained by the average weighted processing after comparing the power grid device parameter data with the corresponding standard values, and comprehensively reflects the current health status of the device; is the trend sensitivity factor, and the calculation formula is , where the rated capacity divided by the number of overloads reflects the overload tolerance of the equipment; is the failure propagation probability, generated based on the simulation of the simultaneous failure scenario of two devices by a Bayesian network, that is, the probability of failure propagation when two devices fail simultaneously is considered; is the continuous operation duration of the device, is the average life threshold of similar devices; is a preset time decay coefficient; Substitute the collected data into the spatio-temporal sensitive maintenance index, and calculate the spatio-temporal sensitive maintenance index value of each equipment; Sort the equipment according to the size of the spatio-temporal sensitive maintenance index value. The larger the spatio-temporal sensitive maintenance index value, the higher the priority maintenance level of the equipment, and the maintenance should be arranged first to determine the priority maintenance order of the equipment.

2. The future-state dynamic generation and optimized scheduling system for power grid maintenance according to claim 1, wherein The specific working steps of the maintenance mode adjustment module also include the following: For spatio-temporal collaborative conflict resolution, resolve the maintenance conflict, specifically including two constraint conditions: spatial coupling constraint and time elasticity constraint. The spatial coupling constraint identifies conflicts by considering the geographical proximity coefficient between equipment and the electrical coupling degree and geographical topology; the time elasticity constraint determines the time range of the maintenance window based on the elasticity coefficient generated by meteorological prediction.

3. The future-state dynamic generation and optimal scheduling system for power grid maintenance according to claim 1, wherein The specific working steps of the maintenance mode adjustment module also include the following; Dynamic adaptation of elastic resources, construct a multi-objective resource optimization model, with the sum of the maintenance time multiplied by the resource reduction coefficient maximized, and the sum of the product of the labor cost and the square of the distance between the equipment geographical coordinates and the team station location minimized as the objective function.

4. A future-state dynamic generation and optimized scheduling system for power grid maintenance according to claim 1, characterized in that, For the safety analysis module, the specific steps for conducting static safety analysis are as follows: The static safety analysis aims to verify the stability of the power grid after the maintenance mode adjustment under a single equipment failure. First, conduct a single equipment failure enumeration, that is, traverse all maintenance equipment and associated lines, simulate the power grid topology after their outage, and generate a new adjacency matrix by removing the corresponding equipment; Then, stability verification is carried out, including voltage stability verification and line overload test. Finally, probabilistic risk assessment is performed, and the system instability probability is calculated by combining the equipment failure rate. Due to the problem of combinatorial explosion in large-scale power grid scenarios, Monte Carlo sampling or heuristic pruning optimization can be used for calculation, and a list of risk equipment and corresponding over-limit indicators are output.

5. A future-state dynamic generation and optimal scheduling system for power grid maintenance according to claim 4, characterized in that, The safety analysis module is also used to perform power grid power flow calculation, and the specific steps are as follows: Generating a future state under a single maintenance work time sequence aims to construct a sequence of future power grid states with time tags to support dynamic risk analysis; First, perform time-section slicing, decompose the maintenance plan according to time slices, generate a topological change sequence, considering the outage / resumption operations of the maintained equipment and the time-sequence changes of load forecasting; Then, use the state prediction model to predict the future state parameters; Finally, use the time-sequence database to store these section snapshots for quick backtracking and comparison, and output a set of future state power grid models with time stamps.

6. The future-state dynamic generation and optimal scheduling system for power grid maintenance according to claim 5, characterized in that, The specific steps of the risk discrimination module are as follows: The goal of time-sequence power grid risk discrimination brought by maintenance power outage is to quantify the risk levels at different time points during the maintenance process and identify the chain of cascading failures; First, perform time-sequence risk chain modeling, construct a Bayesian network to depict the causal relationship between equipment failures, load transfer, and protection actions; Then, calculate risk indicators, such as the probability of load loss, by judging the relationship between the load and available power in each time period for statistics; Next, generate a risk heat map, comprehensively considering factors such as the overload rate, voltage deviation, and new energy fluctuations, and add them according to weights to obtain the risk values at each location and time; Then, based on graph theory algorithms, identify the critical path and determine the maintenance path with the largest impact range.

7. A future-state dynamic generation and optimized scheduling system for power grid maintenance according to claim 6, characterized in that The specific steps for the iteration module to dynamically correct the maintenance plan according to the risk result data and generate the final plan are as follows: Starting from the initial maintenance plan, perform safety analysis and risk discrimination in sequence. If the risk discrimination result meets the standard, execute the maintenance plan. If the risk discrimination result does not meet the standard, enter the dynamic adjustment link to correct the maintenance plan, and then perform safety analysis and risk discrimination again until the result meets the standard.

8. A future-state dynamic generation and optimized scheduling system for power grid maintenance according to claim 7, characterized in that, The specific working steps of the preview verification module are as follows: Generate multiple extreme scenarios, specifically including a sharp drop in new energy output and a typhoon passing through, calculate the weights of each scenario, considering the scenario deviation degree and correlation coefficient; Based on the grouping matrix, divide the equipment activity queue corresponding to the set of operable equipment in a specific time window.

Images