A device maintenance plan scheduling optimization method and device, electronic device and medium

Abstract

The application belongs to the technical field of electric power automation, and discloses a device maintenance plan arrangement optimization method and device, electronic equipment and medium; the method comprises the following steps: obtaining the cycle type of a maintenance plan, and creating an evaluation knowledge table according to the cycle type; performing maintenance simulation based on historical power grid data and a set of maintenance devices, obtaining feedback power grid state information, and updating the evaluation knowledge table through the feedback power grid state information; at the same time, updating the evaluation knowledge table based on artificial historical operation experience; obtaining a final evaluation knowledge table; selecting information corresponding to devices included in a preset device list in the final evaluation knowledge table, and forming a new evaluation knowledge table; based on the quantitative value of the future power grid state and the set of devices to be maintained in the next cycle, arranging the best maintenance devices in each period in the next cycle according to the evaluation value of the power grid state and the maintenance devices in the new evaluation knowledge table; according to the future state of the power grid, the application can make a more reasonable maintenance plan from the power grid operation level.

Description

Technical Field

[0001] This invention belongs to the field of power automation technology, and specifically relates to a method, device, electronic equipment and medium for optimizing equipment maintenance planning. Background Technology

[0002] With the construction of a new power system based on new energy sources and the advancement of power market reforms, higher requirements are being placed on power supply reliability. A reasonable scheduling of power equipment maintenance is a crucial measure to improve power supply reliability; therefore, a more rational and refined scheduling of power grid equipment maintenance plans is of great significance.

[0003] Current power equipment maintenance schedules are typically based on the complexity of the lines or equipment, with regular maintenance conducted at fixed intervals. With the development of power monitoring technology, there are also plans to acquire various information and operational statuses of equipment through real-time monitoring, thereby assessing the equipment's health and determining whether maintenance is necessary. In short, existing maintenance schedules, which consider holidays and staffing levels, are tailored to individual pieces of equipment and may affect the stability and security of the power grid system. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, electronic device, and medium for optimizing equipment maintenance planning, in order to solve the technical problem of unreasonable equipment maintenance planning in existing technologies, which affects the stability and security of power grid system operation. This invention, based on the future state of the power grid, enables the formulation of more reasonable maintenance plans from the perspective of power grid operation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] In a first aspect, the present invention provides a method for optimizing equipment maintenance schedules, comprising:

[0007] Obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; acquire historical power grid data, manual historical operation experience, and maintenance equipment set; conduct maintenance simulation based on historical power grid data and maintenance equipment set to obtain feedback power grid status information, and update the assessment knowledge table based on the feedback power grid status information; at the same time, update the assessment knowledge table based on manual historical operation experience; obtain the final assessment knowledge table.

[0008] Obtain the set of equipment expected to be maintained next week and the future power grid operation status data; select information corresponding to the equipment included in the preset equipment list from the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained next week, arrange the best maintenance equipment for each time period in the next cycle according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, and obtain the optimal power grid maintenance equipment arrangement scheme.

[0009] Output the optimal power grid maintenance equipment scheduling scheme.

[0010] A further improvement of the present invention is that the information corresponding to the devices included in the preset device list is specifically the devices listed in the device list in the final evaluation knowledge table.

[0011] A further improvement of this invention is that the step of obtaining the cycle type of the maintenance plan and creating an assessment knowledge table based on the cycle type specifically includes:

[0012] Obtain the cycle type of the maintenance plan, which can be one day, seven days, monthly, quarterly, or annual; determine the corresponding maintenance cycle period T based on the cycle type; and establish an assessment knowledge table with a scheduling length equal to the maintenance cycle period T.

[0013] The created assessment knowledge table lists the grid state similarity quantification value as the row and the maintenance equipment name as the column; the grid state similarity quantification value is calculated from historical grid data.

[0014] A further improvement of this invention is that the step of acquiring historical power grid data, historical manual operation experience, and maintenance equipment sets specifically includes:

[0015] Acquire historical power grid data, historical manual operation experience, and maintenance equipment sets; obtain the data sampling frequency within the maintenance cycle, and determine the number of power grid states N within the maintenance cycle based on the maintenance cycle time period T and the sampling frequency; divide the historical power grid data maintenance cycle time period T into M groups, with each group containing N power grid sections.

[0016] A further improvement of this invention lies in: performing maintenance simulation based on historical power grid data and a set of maintenance equipment to obtain feedback power grid status information, and updating the evaluation knowledge table based on the feedback power grid status information; simultaneously, updating the evaluation knowledge table based on historical manual operation experience; the steps for obtaining the final evaluation knowledge table specifically include:

[0017] Read the unanalyzed power grid section of the i-th group;

[0018] Read the j-th power grid section in the i-th group sequentially according to the time sequence;

[0019] Calculate the power grid state similarity quantification value between the current power grid section state and the reference state, and retrieve the record with the corresponding power grid state quantification value in the evaluation knowledge table; if no record with the corresponding power grid state similarity quantification value is found, create a new record in the evaluation knowledge table according to the power grid state similarity quantification value, and initialize the equipment evaluation value of the corresponding state to 0;

[0020] If the historical operation experience is analyzed, and it is determined that the current power grid status requires equipment maintenance, a reward value is calculated and added to the corresponding power grid status and equipment evaluation knowledge table value, and the next section is read. If the current status does not exist in the historical operation experience, a power flow environment simulation is performed to obtain feedback power grid status information, and the evaluation knowledge table is updated based on the feedback power grid status information.

[0021] After all power grid sections in Group M have been processed, the final evaluation knowledge table is obtained.

[0022] A further improvement of this invention lies in the step of simulating the power flow environment, which specifically includes:

[0023] A random variable r is generated between 0 and 1. When r < ε, the device name with the highest evaluation value corresponding to the current power grid state similarity quantification value is selected; otherwise, a device name is randomly selected. The selected device name and the current power grid state are forwarded to the power flow environment simulator via message routing. The power flow environment simulator Psenvi, which has the longest idle time, obtains the current power grid state and the selected device name, performs maintenance simulation on the received current power grid state and selected device, obtains power grid state information, and then feeds back the simulated power grid state information via message routing Router. The security quantification evaluation of the fed-back power grid state information is performed, and a reward value is calculated. The reward value is added to the evaluation knowledge table value of the corresponding power grid state and device. ε is a random threshold value. If r is within ε, the optimal value is found according to the previous learning results to strengthen the previous learning results. If r is greater than ε, a maintenance device is randomly selected from the records and explored with a certain probability.

[0024] A further improvement of this invention is that the step of acquiring future power grid operating status data specifically includes:

[0025] Based on the maintenance cycle period T and the power grid data sampling period, determine the number of power grid states N within the maintenance cycle, and obtain N future power grid operation status data.

[0026] A further improvement of this invention lies in the following steps: based on the quantified value of the future power grid state and the set of equipment expected to be maintained in the next week, and according to the evaluation values ​​of the power grid state and the maintenance equipment in the new evaluation knowledge table, the optimal maintenance equipment for each time period in the next cycle is arranged to obtain the optimal power grid maintenance equipment arrangement scheme; specifically including:

[0027] Generate a list L of N power grid state similarity quantification values, and simultaneously create an empty list S;

[0028] Read the power grid state similarity quantization value from list L one by one;

[0029] Retrieve the evaluation table record with the closest read power grid state similarity quantification value from the new evaluation knowledge table, and select the device name with the largest value from the record; each row of the new evaluation knowledge table corresponds to a power grid state similarity quantification value. For each future state, form a power grid state similarity quantification value, find the corresponding row with the closest power grid state similarity quantification value, and each column in each row corresponds to a device. Find the device with the largest cumulative evaluation value.

[0030] Add the selected device name to list S;

[0031] If it is not the last value in L, read the next grid state similarity quantization value in list L; after processing one list L, process the next list L, until all lists L are processed to obtain the optimal grid maintenance equipment scheduling scheme; the optimal grid maintenance equipment scheduling scheme includes maintenance equipment lists S corresponding to the next N time periods.

[0032] Secondly, the present invention provides an equipment maintenance plan scheduling and optimization device, comprising:

[0033] The assessment knowledge table creation module is used to obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; acquire historical power grid data, manual historical operation experience, and maintenance equipment set; perform maintenance simulation based on historical power grid data and maintenance equipment set to obtain feedback power grid status information, and update the assessment knowledge table based on the feedback power grid status information; at the same time, update the assessment knowledge table based on manual historical operation experience; and obtain the final assessment knowledge table.

[0034] The scheme optimization module is used to obtain the set of equipment expected to be maintained in the next week and the future power grid operation status data; select information corresponding to the equipment included in the preset equipment list from the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained in the next week, and according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, arrange the best maintenance equipment for each time period in the next cycle to obtain the optimal power grid maintenance equipment arrangement scheme.

[0035] The output module is used to output the optimal power grid maintenance equipment scheduling scheme.

[0036] Thirdly, the present invention provides an electronic device including a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the equipment maintenance plan scheduling optimization method.

[0037] Fourthly, the present invention provides a computer-readable storage medium storing at least one instruction, which, when executed by a processor, implements the equipment maintenance plan scheduling optimization method.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] This invention provides a method, apparatus, electronic device, and medium for optimizing equipment maintenance plan scheduling. The method involves: acquiring the cycle type of the maintenance plan and creating an evaluation knowledge table; acquiring historical power grid data, historical manual operation experience, and a set of maintenance equipment; performing maintenance simulations based on the historical power grid data and the set of maintenance equipment to obtain feedback power grid status information, and updating the evaluation knowledge table based on the feedback power grid status information; updating the evaluation knowledge table based on historical manual operation experience; obtaining a final evaluation knowledge table; acquiring load forecast and generation plan data to determine future power grid operating status data; determining the set of equipment to be maintained in the next week; and, based on the quantified value of the future power grid status and the set of equipment to be maintained in the next week, scheduling the optimal maintenance equipment for each time period in the next cycle according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, thus obtaining the optimal power grid maintenance equipment scheduling scheme; and outputting the optimal power grid maintenance equipment scheduling scheme. This invention utilizes intelligent auxiliary scheduling technology for equipment maintenance plans based on power grid status similarity, relying on historical data and manual experience knowledge to intelligently schedule equipment maintenance plans based on safety and feasibility, given a set of maintenance equipment. This invention mainly consists of two parts: First, it involves forming a safety assessment knowledge table. Specifically, based on historical operational data, various equipment maintenance scenarios are simulated, and then the power grid safety of each scenario is quantitatively assessed or evaluated using human experience, resulting in an assessment knowledge table corresponding to each piece of equipment under different power grid conditions. Second, it involves intelligent assisted scheduling. Based on the assessment knowledge table and the future state of the power grid, specific maintenance times are scheduled for the equipment to be maintained. This invention, by considering the future state of the power grid and from the perspective of power grid operation, enables the formulation of more reasonable maintenance plans, solving the technical problem of unreasonable equipment maintenance plan scheduling in existing technologies, which affects the stability and safety of the power grid system.

[0040] To improve the safety and rationality of maintenance planning, this invention roughly schedules the equipment to be maintained, while considering holidays and personnel and material allocation. Based on this, it uses a knowledge table of maintenance strategy safety assessments to retrieve and sort specific equipment maintenance times, thereby improving both the safety and ease of scheduling.

[0041] When forming a knowledge table for maintenance strategy safety assessment based on historical data, this invention constructs a power grid baseline state and then quantifies the similarity of the power grid state with the baseline state to form a similarity value representing the power grid state at the corresponding moment. This allows the system to retrieve the closest power grid state to the current state at a speed of milliseconds.

[0042] In the process of generating the evaluation table, this invention can perform a quantitative safety assessment of the power grid state after environmental simulation, and can also refer to human experience to adjust the quantitative evaluation value of the equipment under maintenance in the corresponding power grid state, thereby directly introducing human experience into the learning of historical data.

[0043] When interacting with the power flow environment, the intelligent agent, message routing, and power flow environment satisfy the basic input and output rules. As long as the agreed communication protocol is met, the software functions and hardware configuration can be dynamically expanded, which satisfies the simulation of distributed parallel environment.

[0044] Given the independence of knowledge table generation and maintenance plan scheduling, this invention allows the knowledge table to be continuously improved based on power grid data and maintenance simulations. The maintenance plan scheduling only reads the latest knowledge table during scheduling, thereby optimizing the allocation of overall computing resources.

[0045] This invention, based on historical power grid operation data, simulates different equipment maintenance environments under various power grid conditions or references human experience to form an equipment maintenance assessment knowledge table for each power grid operation condition. Finally, equipment maintenance is scheduled based on this assessment knowledge table. This invention overcomes at least one or more of the following technical problems: 1) The massive scale of power grid data significantly impacts the efficiency of power grid condition identification using traditional methods; how to quickly retrieve the required power grid condition or identify differences between different power grid conditions; 2) The diverse range of power grid maintenance scheduling; how to flexibly interact with the power grid simulation environment to simulate numerous strategies and provide feedback on simulation results; 3) How to integrate environmental simulation feedback information and human experience evaluation to generate a reasonable assessment knowledge table for power grid condition strategies; 4) How to schedule future equipment maintenance plans using the final assessment knowledge table.

[0046] This invention characterizes the power grid state using power grid state similarity metrics, thereby constructing a millisecond-level power grid state retrieval system. By introducing message routing, this invention enables rapid and flexible interaction between the power grid agent and the power flow environment simulation, thus constructing a maintenance plan learning framework under power flow simulation. This invention integrates environmental simulation feedback information evaluation and human experience evaluation through a maintenance strategy safety assessment knowledge table. This invention optimizes the allocation of overall computing resources by decoupling the generation of the knowledge table and the arrangement of maintenance plans. Attached Figure Description

[0047] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0048] Figure 1 This is a schematic diagram of the quantification value curve of the state similarity throughout the day;

[0049] Figure 2 A structural relationship diagram between the maintenance agent, the message router, and the power flow calculator Psenv;

[0050] Figure 3 A schematic diagram illustrating the implementation process of the safety assessment knowledge table;

[0051] Figure 4 This is a schematic diagram of the intelligent assisted arrangement process;

[0052] Figure 5 This is a flowchart illustrating an equipment maintenance plan scheduling optimization method according to the present invention.

[0053] Figure 6 This is a schematic diagram of the structure of an equipment maintenance plan scheduling optimization device according to the present invention;

[0054] Figure 7 This is a structural block diagram of an electronic device according to the present invention. Detailed Implementation

[0055] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0056] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this invention is for describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0057] Explanation of relevant terms:

[0058] Maintenance refers to all procedures and activities performed to keep a product in use and operational condition, as well as to eliminate malfunctions and defects.

[0059] A maintenance plan is a planned schedule for maintenance and repair work on power grid equipment.

[0060] This invention provides a method for optimizing equipment maintenance planning, which intelligently assists in the planning of equipment maintenance based on the similarity of power grid conditions. The main task is to intelligently plan equipment maintenance schemes based on safety and feasibility, relying on historical data and human experience and knowledge, given a set of equipment to be maintained.

[0061] This invention is mainly divided into two parts: The first part is to form a maintenance assessment knowledge table, which is based on historical operating data, simulates various equipment maintenance scenarios, and then conducts quantitative assessment or manual evaluation of power grid safety to form a safety assessment knowledge table for each piece of equipment under various power grid conditions; The second part is intelligent auxiliary scheduling, which is based on the assessment knowledge table and the future power grid conditions to schedule specific maintenance times for the equipment to be maintained.

[0062] This invention is primarily applied to equipment maintenance scheduling in power grid dispatching operations. Based on historical power grid operating status data, it can simulate different equipment maintenance environments under various power grid conditions or refer to human experience during maintenance periods, according to future power grid conditions and pre-maintenance equipment. This generates a knowledge table for equipment outage maintenance assessment under various power grid operating conditions, and finally, equipment maintenance is scheduled based on this knowledge table. This invention differs from traditional maintenance scheduling techniques in that it can retrieve similar power grid states within a millisecond-level accuracy range for any power grid condition. Furthermore, for any equipment maintenance under any power grid condition, a safety assessment is performed after environmental simulation, forming a cumulative assessment knowledge table. To achieve these technical objectives, the following problems need to be addressed:

[0063] 1) The power grid data is massive, and traditional data comparison seriously affects the efficiency of power grid status identification. How to quickly retrieve the required power grid status or identify the differences between various power grid statuses?

[0064] 2) There are many different types of power grid maintenance schedules. How can we flexibly interact with the power grid simulation environment to simulate a large number of strategies and provide feedback on the simulation results?

[0065] 3) How to integrate environmental simulation feedback information evaluation and human experience evaluation to generate a reasonable evaluation knowledge table for power grid state strategies;

[0066] 4) How to plan future equipment maintenance based on the final assessment knowledge sheet.

[0067] Example 1

[0068] 1. Quantification of power grid status

[0069] For each moment of power grid operation data, the active power and active power limit values ​​of all lines and high-voltage side branches of transformers in the power grid are obtained, forming a state list L. p and L bNext, refer to L. p Form a vector V representing the current power grid state. p Refer to L b Form a reference vector V that characterizes the state of this power grid. b Finally, following a power system state similarity analysis method disclosed in patent CN106816871B, a quantitative similarity value SIM of the current power grid state relative to the baseline state is generated. p This value characterizes the magnitude of the difference between the current grid state and the grid reference state.

[0070] Among them, L p ={P l1 ,P l2 ,…P li ,P x1 ,P x2 ,…P xj} is a list of active power on the high-voltage side of the line and transformer, P li Let P be the active power of the i-th line. xj V represents the active power on the high-voltage side of the j-th transformer. p =[P l1 ,P l2 ,…P li ,P x1 ,P x2 ,…P xj [L] is the vector formed by the active power on the high-voltage side of the line and transformer. b ={P l1max ,P l2max ,…P limax ,P x1max ,P x2max ,…P xjmax} is a list of active power limits for the high-voltage side of lines and transformers, P limax Let P be the active power limit for the i-th line. xjmax V represents the active power limit on the high-voltage side of the j-th transformer. b =[P l1max ,P l2max ,…P limax ,P x1max ,P x2max ,…P xjmax ] is a vector formed by the active power limits on the high-voltage side of the lines and transformers, where there are i lines and j transformers.

[0071] For any given moment, the state vector of the power grid is used to determine its distance from the reference vector using a cosine similarity function. The quantified state similarity values ​​at each moment are as follows:

[0072]

[0073] Where x = [x1, x2, ..., x n ] = V p y = (x1, x2, ..., x n ) = V b ,

[0074] <x,y> = x1y1 + x2y2 + ... + x n y n ,

[0075]

[0076]

[0077] Taking a provincial power grid as an example, the state similarity quantification values ​​(d(x,y)) for each time point throughout the year are generated and stored. The storage structure is shown in Table 1. `index` represents the corresponding sequence number, `sim` represents the state similarity quantification value (the highest precision can be 13 decimal places; for simplicity, it is truncated to 4 decimal places here), and `timepos` represents the provincial power grid data for the corresponding time point. In actual orchestration and analysis, the corresponding `sim` is used as the corresponding power grid state, rather than the power grid data file for the corresponding time point.

[0078] Table 1. Similarity Value Data Table

[0079] 0 0.4492 20220801_0000 1 0.4506 20220801_0005 2 0.4495 20220801_0010 3 0.4479 20220801_0015 4 0.4488 20220801_0020 5 0.4504 20220801_0025 6 0.4465 20220801_0030 7 0.4446 20220801_0035 8 0.4433 20220801_0040 9 0.4405 20220801_0045 10 0.4405 20220801_0050

[0080] Using data from a specific day, a curve is plotted to generate a quantitative value curve of the similarity of states throughout the day, such as... Figure 1 As shown.

[0081] 2. Interacting with the power grid simulation environment

[0082] The maintenance strategy safety assessment knowledge table is primarily based on maintenance simulations of a large number of historical power grid states. After determining the maintenance cycle T, the maintenance agent determines the number N of power grid states within the maintenance cycle according to the sampling period of the power grid sections. Finally, M sets of data are formed, with each group containing N sections. The evaluation of each group is the evaluation of a set of maintenance plans. In a specific implementation, for example, the maintenance cycle is weekly maintenance, and the maintenance duration is arranged for the next 7 days. If the actual historical data sampling period is 5 minutes (288 power grid sections per day), then the weekly maintenance involves 7*288 sections. The historical data is then grouped into M sets of data, each containing 7*288 sections. Thus, T = 7, N = 7*288, and M N represents the total number of historical sections. The entire simulation process is divided into three parts: the maintenance agent, the message routing router, and the power flow environment simulator Psenv. Please refer to [link to relevant documentation]. Figure 2As shown, the maintenance agent forms a corresponding evaluation knowledge table based on the grid state similarity value and the list of equipment to be inspected. The message routing router is only responsible for optimally allocating simulation tasks and feeding back information after the maintenance simulation. The power flow environment simulator Psenv is responsible for simulating the maintenance operations of specified equipment under specified power flow states and feeding back grid state information.

[0083] The interaction rules for the three types of program entities are as follows:

[0084] 1) The message routing router creates network sockets based on the server IP and port of its entity as data transmission and reception relay nodes, and allocates power flow simulation tasks.

[0085] 2) When outputting messages, the maintenance agent sends the power grid status and maintenance command as the message body to the message router. After receiving the message, it parses and saves the power grid status after simulating the maintenance operation.

[0086] 3) The power flow simulator Psenv contains multiple program entities with the same functionality. When there are no computational tasks, it periodically sends empty messages to the message routing router. Upon receiving the grid state and maintenance command assigned for forwarding, it simulates maintenance operations based on the corresponding grid state to perform power flow simulation calculations. After completion, it sends the simulated grid state as a feedback message to the message routing router.

[0087] 3. Evaluation based on feedback information from tidal environment simulation and evaluation based on human experience.

[0088] The safety assessment of power grid state information based on power flow simulation feedback is divided into two aspects: voltage and branch power. Additionally, under certain power grid operating conditions, based on operator experience, there is a preference to designate specific equipment for maintenance. All these assessments and preferences are ultimately quantified and reflected in the knowledge assessment table.

[0089] Quantitative assessment of voltage safety:

[0090]

[0091] Among them, Estimate v is the voltage offset, with a value range of [0,1], where n is the number of bus voltages and v is the per-unit value of the corresponding node voltage.

[0092] Quantitative assessment of branch power flow safety:

[0093]

[0094] Among them, Estimate b e represents the branch offset, with a value ranging from [0,1], where m is the number of branches.i This represents the safety offset coefficient for the corresponding branch i; when e i Equal to 1; when e i equal If the value is less than or equal to 1, there will be no overload, and the branch is safe. If the value is greater than 1, there will be an overload, and the larger the value, the less safe it is. i For the power of the corresponding branch i, p lmti This is the power limit for the corresponding branch i.

[0095] Quantitative assessment of human experience:

[0096]

[0097] Among them, Estimate m Let (s, a) be a state-action pair ∈ {0,1}, representing the maintenance of equipment a under a given power grid state s. Experience is a set of human experience records state-action pairs that tend to perform maintenance on equipment a under a given power grid state. Human experience represents past judgments based on human experience regarding the most suitable maintenance for equipment a under a given power grid state s; this can be obtained from a big data platform or manually added. Historical data refers to historical power grid operation data, typically using state estimation results generated every 5 minutes.

[0098] Reward value:

[0099] For the evaluation of the maintenance action pair (s,a) of the equipment under a specified power grid condition, a corresponding reward value will be generated on the agent side.

[0100] Estimate = αEstimate v +βEstimate b

[0101]

[0102] Here, Estimate is a comprehensive evaluation value generated by the power grid state data fed back after receiving the corresponding (s,a) in the power flow environment simulation. The larger the value, the further it deviates from the safe range. The threshold value γ ranges from (0,1). When the comprehensive evaluation value Estimate is greater than the threshold value γ, the reward value is -1; when the comprehensive evaluation value is less than or equal to the threshold value γ, the reward value is 0.5; when the corresponding (s,a) belongs to human experience, the reward value is 1.

[0103] This invention provides a method for optimizing equipment maintenance plan scheduling, mainly comprising two parts: forming a safety assessment knowledge table and intelligent assisted scheduling. The former is the foundation for the latter. The implementation process of each part is described below:

[0104] Please see Figure 3 As shown, the implementation process of the security assessment knowledge table is as follows:

[0105] 1) Determine the cycle type of the maintenance plan, determine the maintenance cycle period T based on the cycle type, and determine the number of grid states N included in the maintenance cycle based on the grid data sampling period. In one specific implementation, the maintenance plan may be a daily, monthly, quarterly, or annual maintenance plan. Different cycle types of maintenance plans have different types of assessment knowledge tables. The cycle type determines the maintenance cycle period T and the arrangement length of the maintenance plan. In one specific implementation, the maintenance cycle period T can be one day, seven days, monthly, quarterly, or annual; the sampling period is the historical data interval, such as one section every 5 minutes, 15 minutes, or one section every hour. If it is a daily maintenance, and the cycle is 5 minutes, then N = 24 * 60 / 5 = 288; if it is one section every hour, then N = 24 * 1 = 24.

[0106] 2) Obtain historical power grid data and maintenance equipment sets, and divide the maintenance cycle time period T of the historical power grid data into M groups, with each group containing N power grid sections (historical data is divided into M groups according to the number of power grid states N contained in the weekly and monthly plans);

[0107] 3) Create an evaluation knowledge table that quantifies the similarity of behavior to the power grid state and lists the device names; where the set of quantified power grid state similarity values ​​is represented by S, the set of device names is represented by A, and a random threshold value ε is set; in one specific implementation, ε is equal to 0.7;

[0108] 4) Read the unanalyzed power grid cross-section of the i-th group;

[0109] 5) Read the j-th cross-section in the i-th group sequentially according to the time sequence;

[0110] 6) Refer to the quantification method of power grid status (refer to d for specific calculation formula)<x,y> This process generates a quantized value of the current grid section state relative to the reference state. The value is then truncated to the required precision to obtain the grid state similarity quantization value (SIM). p The system then retrieves records with corresponding grid state quantification values ​​from the assessment knowledge table. (If no record with corresponding grid state similarity quantification values ​​is found, the system searches the assessment knowledge table by grid state similarity quantification value SIM.) p Create a new record, and initialize the equipment evaluation value (the specific values ​​of s (row) and a (column) in the evaluation knowledge table) to 0 for the corresponding state; where the subscript p represents the active power data of the branch; if found, take the record of this row as the maintenance evaluation of each equipment under the current power grid section state, and then accumulate and modify the evaluation value in this record based on this (the record of this row finally represents the maintenance evaluation value of different equipment under this power grid state).

[0111] 7) Obtain historical manual operation experience. If the historical manual operation experience determines that the current power grid status requires equipment maintenance, a positive reward value will be set and added to the corresponding power grid status and equipment evaluation knowledge table value (the table elements of the evaluation knowledge table are quantitative values ​​that are accumulated). Proceed to step 5. If the current status does not exist in the historical manual operation experience, proceed to step 8.

[0112] 8) Generate a random variable r between 0 and 1. When r < ε, select the device name with the highest evaluation value corresponding to the current power grid state similarity quantization value. Otherwise, randomly select a device name and forward the selected device name and the current power grid state to the power flow environment simulator through message routing.

[0113] 9) After obtaining the current power grid status and the selected device name, the Psenvi power flow environment simulator with the longest idle time performs maintenance simulation on the received current power grid status and selected device to obtain power grid status information. After completion, it feeds back the simulated power grid status information through the message routing Router.

[0114] 10) The maintenance agent performs a security quantification assessment of the fed-in power grid status information and calculates the reward. (s,a) (A positive reward is given when security is high, and a negative reward is given when security is low);

[0115] 11) Add the reward value to the corresponding power grid status and equipment assessment knowledge table value;

[0116] 12) Check if this is the final section in the current group. If not, proceed to step 5; if yes, proceed to step 13.

[0117] 13) Check if it is the last group. If it is not the last group, go to step 4; if it is, go to step 14.

[0118] 14) Obtain the final assessment knowledge sheet.

[0119] Please see Figure 4 As shown, the intelligent assisted orchestration process includes:

[0120] 1) Obtain the list of equipment to be inspected;

[0121] 2) In the final evaluation knowledge table, only select the devices listed in the equipment list to form a new evaluation knowledge table qtable;

[0122] 3) Based on the maintenance cycle period T and the power grid data sampling period, determine the number of power grid states N within the maintenance cycle, and obtain N future power grid operating status data; in one specific implementation, the future power grid power flow status data can be determined by power flow calculation based on the power generation plan and load forecast; these data can be provided by the platform;

[0123] 4) Generate a list L of N power grid state similarity quantification values, and simultaneously create an empty list S;

[0124] 5) Read the power grid state similarity quantization value from list L one by one;

[0125] 6) Retrieve the evaluation table record with the closest grid state similarity quantization value from the qtable table, and select the device name with the largest value from the record; each row of the evaluation knowledge table qtable corresponds to a grid state similarity quantization value, and for each future state, form a quantization value SIM. p Find SIM p The closest corresponding row is selected, with each column in each row corresponding to a device. The device with the largest cumulative evaluation value is then identified. The values ​​in the evaluation knowledge table qtable are the cumulative reward values ​​obtained from historical data simulations. Therefore, for each grid state and device combination, there is an evaluation value that assesses its quality.

[0126] 7) Add the selected device name to list S;

[0127] 8) If it is not the last value in L, go to step 5; after processing one list L, go to step 5 to process the next list L, until all lists L have been processed, then go to step 9.

[0128] 9) Obtain the list S of maintenance equipment for the next N time periods.

[0129] Example 2

[0130] Please see Figure 5 As shown, the present invention provides a method for optimizing equipment maintenance schedules, comprising:

[0131] S1. Obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; obtain historical power grid data, manual historical operation experience, and maintenance equipment set; perform maintenance simulation based on historical power grid data and maintenance equipment set to obtain feedback power grid status information, and update the assessment knowledge table based on the feedback power grid status information; at the same time, update the assessment knowledge table based on manual historical operation experience; obtain the final assessment knowledge table.

[0132] S2. Obtain the set of equipment to be inspected next week and the future power grid operation status data; select only the equipment listed in the equipment list in the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment to be inspected next week, arrange the best maintenance equipment for each period in the next cycle according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, and obtain the optimal power grid maintenance equipment arrangement scheme.

[0133] S3, Output the optimal power grid maintenance equipment scheduling scheme.

[0134] In one specific implementation: the step of obtaining the cycle type of the maintenance plan and creating an assessment knowledge table based on the cycle type specifically includes:

[0135] Obtain the cycle type of the maintenance plan, which can be one day, seven days, monthly, quarterly, or annual; determine the corresponding maintenance cycle period T based on the cycle type; and establish an assessment knowledge table with a scheduling length equal to the maintenance cycle period T.

[0136] The created assessment knowledge table lists the grid state similarity quantification value as the row and the maintenance equipment name as the column; the grid state similarity quantification value is calculated from historical grid data.

[0137] In one specific implementation: the formula for calculating the power grid state similarity quantization value is:

[0138]

[0139] Where x = [x1, x2, ..., x n ] = V p y = (x1, x2, ..., x n ) = V b ,

[0140] <x,y> = x1y1 + x2y2 + ... + x n y n ,

[0141]

[0142]

[0143] Specifically, for each moment of power grid operation data, the active power and active power limit values ​​of all lines and high-voltage side branches of transformers in the power grid are obtained, forming a state list L. p and L b Next, refer to L. p Form a vector V representing the current state of the power grid. p Refer to L bForm a reference vector V that characterizes the state of this power grid. b Quantization of Grid State Similarity (SIM) p =d(x,y), representing the magnitude of the difference between the current grid state and the grid reference state. Preferably, the grid state similarity quantization value SIM... p It can be calculated according to a power system state similarity analysis method disclosed in patent CN106816871B.

[0144] Among them, L p ={P l1 ,P l2 ,…P li ,P x1 ,P x2 ,…P xj} is a list of active power on the high-voltage side of the line and transformer, P li Let P be the active power of the i-th line. xj V represents the active power on the high-voltage side of the j-th transformer. p =[P l1 ,P l2 ,…P li ,P x1 ,P x2 ,…P xj [L] is the vector formed by the active power on the high-voltage side of the line and transformer. b ={P l1max ,P l2max ,…P limax ,P x1max ,P x2max ,…P xjmax} is a list of active power limits for the high-voltage side of lines and transformers, P limax Let P be the active power limit for the i-th line. xjmax V represents the active power limit on the high-voltage side of the j-th transformer. b =[P l1max ,P l2max ,…P limax ,P x1max ,P x2max ,…P xjmax ] is a vector formed by the active power limits on the high-voltage side of the lines and transformers, where there are i lines and j transformers.

[0145] In one specific implementation: the step of acquiring historical power grid data, manual historical operation experience, and maintenance equipment set specifically includes:

[0146] Acquire historical power grid data, historical manual operation experience, and a set of maintenance equipment; obtain the data sampling frequency within the maintenance cycle; determine the number of power grid states N within the maintenance cycle based on the maintenance cycle period T and the sampling frequency; divide the historical power grid data maintenance cycle period T into M groups, with each group containing N power grid sections. For example, the maintenance cycle is weekly, and the maintenance duration is arranged for the next 7 days. If the actual historical data sampling period is 5 minutes (288 power grid sections per day), then the weekly maintenance involves 7*288 sections. The historical data is then grouped into M groups of 7*288 sections each. Thus, T = 7, N = 7*288, and M N represent the total number of historical sections.

[0147] In one specific implementation: maintenance simulation is performed based on historical power grid data and a set of maintenance equipment to obtain feedback power grid status information, and the evaluation knowledge table is updated based on the feedback power grid status information; simultaneously, the evaluation knowledge table is updated based on historical manual operation experience; the steps to obtain the final evaluation knowledge table specifically include:

[0148] Read the unanalyzed power grid section of the i-th group;

[0149] Read the j-th power grid section in the i-th group sequentially according to the time sequence;

[0150] Calculate the power grid state similarity quantification value between the current power grid section state and the reference state, and retrieve the record with the corresponding power grid state quantification value in the evaluation knowledge table; if no record with the corresponding power grid state similarity quantification value is found, create a new record in the evaluation knowledge table according to the power grid state similarity quantification value, and initialize the equipment evaluation value of the corresponding state to 0;

[0151] If the historical operation experience is analyzed, and it is determined that the current power grid status requires equipment maintenance, a reward value is calculated and added to the corresponding power grid status and equipment evaluation knowledge table value, and the next section is read. If the current status does not exist in the historical operation experience, a power flow environment simulation is performed to obtain feedback power grid status information, and the evaluation knowledge table is updated based on the feedback power grid status information.

[0152] After all power grid sections in Group M have been processed, the final evaluation knowledge table is obtained.

[0153] In one specific implementation, the reward value for evaluating the maintenance action pair (s,a) of a specified power grid condition equipment is as follows:

[0154]

[0155] Where Estimate = αEstimate v +βEstimate bEstimate is a comprehensive evaluation value generated by the power grid state data fed back after receiving the corresponding (s,a) in the power flow environment simulation. The larger the value, the more it deviates from the safe range. When the comprehensive evaluation value Estimate is greater than the threshold value γ, the reward value is -1. When the comprehensive evaluation value Estimate is less than or equal to the threshold value γ, the reward value is 0.5. When the corresponding (s,a) belongs to human experience, the reward value is 1.

[0156]

[0157] Among them, Estimate v is the voltage offset, with a value range of [0,1], where n is the number of bus voltages and v is the per-unit value of the corresponding node voltage.

[0158]

[0159] Among them, Estimate b e represents the branch offset, with a value ranging from [0,1], where m is the number of branches. i This represents the safety offset coefficient for the corresponding branch i; when e i Equal to 1; when e i equal If the value is less than or equal to 1, there will be no overload, and the branch is safe. If the value is greater than 1, there will be an overload, and the larger the value, the less safe it is. i For the power of the corresponding branch i, p lmti The power limit for the corresponding branch i;

[0160] Quantitative assessment of human experience:

[0161]

[0162] Among them, Estimate m Let (s, a) be a state-action pair ∈ {0,1}, representing the maintenance of equipment a under a given power grid state s. Experience is a set of human experience records state-action pairs that tend to perform maintenance on equipment a under a given power grid state. Human experience represents past judgments based on human experience regarding the most suitable maintenance for equipment a under a given power grid state s; this can be obtained from a big data platform or manually added. Historical data refers to historical power grid operation data, typically using state estimation results generated every 5 minutes.

[0163] In one specific implementation: the step of performing power flow environment simulation specifically includes:

[0164] A random variable r is generated between 0 and 1. When r < ε, the device name with the highest evaluation value corresponding to the current power grid state similarity quantification value is selected; otherwise, a device name is randomly selected. The selected device name and the current power grid state are forwarded to the power flow environment simulator via message routing. The power flow environment simulator Psenvi, which has the longest idle time, obtains the current power grid state and the selected device name, performs maintenance simulation on the received current power grid state and selected device, obtains power grid state information, and then feeds back the simulated power grid state information via message routing Router. The security quantification evaluation of the fed-back power grid state information is performed, and a reward value is calculated. The reward value is added to the evaluation knowledge table value of the corresponding power grid state and device.

[0165] In one specific implementation: the step of obtaining future power grid operating status data specifically includes:

[0166] Based on the maintenance cycle period T and the power grid data sampling period, determine the number of power grid states N within the maintenance cycle, and obtain N future power grid operation status data.

[0167] In one specific implementation: Based on the quantified value of the future power grid state and the set of equipment expected to be maintained in the next week, and according to the evaluation values ​​of the power grid state and the maintenance equipment in the new evaluation knowledge table, the steps of arranging the optimal maintenance equipment for each time period in the next cycle to obtain the optimal power grid maintenance equipment arrangement scheme specifically include:

[0168] Generate a list L of N power grid state similarity quantification values, and simultaneously create an empty list S;

[0169] Read the power grid state similarity quantization value from list L one by one;

[0170] Retrieve the evaluation table record with the closest read power grid state similarity quantification value from the new evaluation knowledge table, and select the device name with the largest value from the record; each row of the new evaluation knowledge table corresponds to a power grid state similarity quantification value. For each future state, form a power grid state similarity quantification value, find the corresponding row with the closest power grid state similarity quantification value, and each column in each row corresponds to a device. Find the device with the largest cumulative evaluation value.

[0171] Add the selected device name to list S;

[0172] If it is not the last value in L, read the next grid state similarity quantization value in list L; after processing one list L, process the next list L, until all lists L are processed to obtain the optimal grid maintenance equipment scheduling scheme; the optimal grid maintenance equipment scheduling scheme includes maintenance equipment lists S corresponding to the next N time periods.

[0173] Example 3

[0174] Please see Figure 6 As shown, the present invention provides an equipment maintenance plan scheduling and optimization device, comprising:

[0175] The assessment knowledge table creation module is used to obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; acquire historical power grid data, manual historical operation experience, and maintenance equipment set; perform maintenance simulation based on historical power grid data and maintenance equipment set to obtain feedback power grid status information, and update the assessment knowledge table based on the feedback power grid status information; at the same time, update the assessment knowledge table based on manual historical operation experience; and obtain the final assessment knowledge table.

[0176] The scheme optimization module is used to obtain the set of equipment expected to be maintained in the next week and the future power grid operation status data; select only the equipment listed in the equipment list in the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained in the next week, and according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, arrange the best maintenance equipment for each time period in the next cycle to obtain the optimal power grid maintenance equipment arrangement scheme.

[0177] The output module is used to output the optimal power grid maintenance equipment scheduling scheme.

[0178] In one specific implementation: the step of the evaluation knowledge table creation module obtaining the cycle type of the maintenance plan and creating an evaluation knowledge table based on the cycle type specifically includes:

[0179] Obtain the cycle type of the maintenance plan, which can be one day, seven days, monthly, quarterly, or annual; determine the corresponding maintenance cycle period T based on the cycle type; and establish an assessment knowledge table with a scheduling length equal to the maintenance cycle period T.

[0180] The created assessment knowledge table lists the grid state similarity quantification value as the row and the maintenance equipment name as the column; the grid state similarity quantification value is calculated from historical grid data.

[0181] In one specific implementation: the steps of the evaluation knowledge table building module to acquire historical power grid data, manual historical operation experience, and maintenance equipment sets specifically include:

[0182] Acquire historical power grid data, historical manual operation experience, and maintenance equipment sets; obtain the data sampling frequency within the maintenance cycle, and determine the number of power grid states N within the maintenance cycle based on the maintenance cycle time period T and the sampling frequency; divide the historical power grid data maintenance cycle time period T into M groups, with each group containing N power grid sections.

[0183] In one specific implementation: the evaluation knowledge table establishment module performs maintenance simulations based on historical power grid data and a set of maintenance equipment to obtain feedback power grid status information, and updates the evaluation knowledge table based on the feedback power grid status information; simultaneously, it updates the evaluation knowledge table based on historical manual operation experience; the steps to obtain the final evaluation knowledge table specifically include:

[0184] Read the unanalyzed power grid section of the i-th group;

[0185] Read the j-th power grid section in the i-th group sequentially according to the time sequence;

[0186] Calculate the power grid state similarity quantification value between the current power grid section state and the reference state, and retrieve the record with the corresponding power grid state quantification value in the evaluation knowledge table; if no record with the corresponding power grid state similarity quantification value is found, create a new record in the evaluation knowledge table according to the power grid state similarity quantification value, and initialize the equipment evaluation value of the corresponding state to 0;

[0187] If the historical operation experience is analyzed, and it is determined that the current power grid status requires equipment maintenance, a reward value is calculated and added to the corresponding power grid status and equipment evaluation knowledge table value, and the next section is read. If the current status does not exist in the historical operation experience, a power flow environment simulation is performed to obtain feedback power grid status information, and the evaluation knowledge table is updated based on the feedback power grid status information.

[0188] After all power grid sections in Group M have been processed, the final evaluation knowledge table is obtained.

[0189] In one specific implementation: the step of evaluating the knowledge table building module to simulate the power flow environment specifically includes:

[0190] A random variable r is generated between 0 and 1. When r < ε, the device name with the highest evaluation value corresponding to the current power grid state similarity quantification value is selected; otherwise, a device name is randomly selected. The selected device name and the current power grid state are forwarded to the power flow environment simulator via message routing. The power flow environment simulator Psenvi, which has the longest idle time, obtains the current power grid state and the selected device name, performs maintenance simulation on the received current power grid state and selected device, obtains power grid state information, and then feeds back the simulated power grid state information via message routing Router. The security quantification evaluation of the fed-back power grid state information is performed, and a reward value is calculated. The reward value is added to the evaluation knowledge table value of the corresponding power grid state and device.

[0191] In one specific implementation: the step of the scheme optimization module acquiring future power grid operation status data specifically includes:

[0192] Based on the maintenance cycle period T and the power grid data sampling period, determine the number of power grid states N within the maintenance cycle, and obtain N future power grid operation status data.

[0193] In one specific implementation: the scheme optimization module, based on the quantified value of the future power grid state and the set of equipment expected to be maintained in the next week, arranges the optimal maintenance equipment for each time period in the next cycle according to the power grid state and the evaluation value of the maintenance equipment in the new evaluation knowledge table, to obtain the optimal power grid maintenance equipment arrangement scheme; the steps specifically include:

[0194] Generate a list L of N power grid state similarity quantification values, and simultaneously create an empty list S;

[0195] Read the power grid state similarity quantization value from list L one by one;

[0196] Retrieve the evaluation table record with the closest read power grid state similarity quantification value from the new evaluation knowledge table, and select the device name with the largest value from the record; each row of the new evaluation knowledge table corresponds to a power grid state similarity quantification value. For each future state, form a power grid state similarity quantification value, find the corresponding row with the closest power grid state similarity quantification value, and each column in each row corresponds to a device. Find the device with the largest cumulative evaluation value.

[0197] Add the selected device name to list S;

[0198] If it is not the last value in L, read the next grid state similarity quantization value in list L; after processing one list L, process the next list L, until all lists L are processed to obtain the optimal grid maintenance equipment scheduling scheme; the optimal grid maintenance equipment scheduling scheme includes maintenance equipment lists S corresponding to the next N time periods.

[0199] Example 4

[0200] Please see Figure 7 As shown, the present invention also provides an electronic device 100 for implementing an equipment maintenance plan scheduling optimization method; the electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0201] The memory 101 can be used to store the computer program 103. The processor 102 implements the steps of the equipment maintenance plan scheduling optimization method described in Embodiment 1 or 2 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0202] The at least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor. The processor 102 is the control center of the electronic device 100, connecting various parts of the electronic device 100 via various interfaces and lines.

[0203] The memory 101 in the electronic device 100 stores multiple instructions to implement a device maintenance plan scheduling optimization method, and the processor 102 can execute the multiple instructions to achieve the following:

[0204] Obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; acquire historical power grid data, manual historical operation experience, and maintenance equipment set; conduct maintenance simulation based on historical power grid data and maintenance equipment set to obtain feedback power grid status information, and update the assessment knowledge table based on the feedback power grid status information; at the same time, update the assessment knowledge table based on manual historical operation experience; obtain the final assessment knowledge table.

[0205] Obtain the set of equipment expected to be maintained next week and the future power grid operation status data; select only the equipment listed in the equipment list in the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained next week, arrange the best maintenance equipment for each period in the next cycle according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, and obtain the optimal power grid maintenance equipment arrangement scheme.

[0206] Output the optimal power grid maintenance equipment scheduling scheme.

[0207] Example 5

[0208] If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0209] This invention provides intelligent auxiliary scheduling of equipment maintenance plans based on the similarity of power grid conditions. First, a knowledge table is formed based on equipment maintenance simulation and human experience using historical power grid operation data. Second, the most suitable equipment is retrieved for each future power grid condition based on the knowledge table, and finally, a list of maintenance equipment, i.e., the optimal maintenance plan, is obtained.

[0210] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0211] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0212] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0213] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0214] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing equipment maintenance schedules, characterized in that, include: Obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type; Historical power grid data, manual historical operation experience, and a set of maintenance equipment are acquired. The created assessment knowledge table contains rows representing the power grid state similarity quantification value and columns representing the equipment names in the maintenance equipment set. The power grid state similarity quantification value is obtained from historical power grid data in the following way: For each moment of power grid operation data, the active power and active power limit values ​​of all lines and high-voltage side branches of transformers in the power grid are acquired, forming a vector Vp representing the current power grid state and a reference vector Vb representing the current power grid state, respectively. A cosine similarity function is used to generate the power grid state similarity quantification value between the current power grid state and the reference state. The process involves three parts: a maintenance agent, a message router, and a power flow simulator. The maintenance agent simulates maintenance operations on specified equipment under a given power flow state and updates the evaluation knowledge table based on this information. Simultaneously, it updates the evaluation knowledge table based on historical manual operation experience, resulting in the final evaluation knowledge table. The simulation process consists of three main components: a maintenance agent, a message router, and a power flow simulator (Psenv). The maintenance agent forms the corresponding evaluation knowledge table based on the power flow state similarity value and the list of equipment to be inspected. The message router is responsible for optimizing the allocation of simulation tasks and providing feedback on the post-simulation information. The power flow simulator (Psenv) simulates maintenance operations on specified equipment under a given power flow state and provides feedback on the power flow state. The message router uses the network socket created by its server IP and port as a data relay node and allocates power flow simulation tasks. When outputting messages, the maintenance agent sends the power flow state and maintenance command as the message body to the message router. After receiving the message, it parses and saves the power flow state after the simulated maintenance operation. The simulator Psenv has multiple program entities with the same function. When there is no computational task, it periodically sends empty messages to the message routing Router. After receiving the grid status and maintenance command assigned for forwarding, it simulates maintenance operations based on the corresponding grid status to perform power flow simulation calculations. After completion, it sends the simulated grid status as a message feedback to the message routing Router. During the power flow environment simulation, a random variable r between 0 and 1 is generated. When r < ε, the device name with the highest evaluation value corresponding to the current grid status similarity quantization value is selected; otherwise, a device name is randomly selected. The selected device name and the current grid status are forwarded to the power flow environment simulator through the message routing. The power flow environment simulator Psenvi, which has the longest idle time, obtains the current grid status and the selected device name, performs maintenance simulation on the received current grid status and selected device, obtains grid status information, and then feeds back the simulated grid status information through the message routing Router. The security quantification evaluation of the fed-back grid status information is performed, and a reward value is calculated. The reward value is added to the evaluation knowledge table value of the corresponding grid status and device. Here, ε is a random threshold value. The security evaluation based on the power grid state information fed back by power flow simulation is divided into two aspects: voltage and branch power. Quantitative assessment of voltage safety: in, is the voltage offset, with a value range of [0,1], where n is the number of bus voltages and v is the per-unit value of the corresponding node voltage; Quantitative assessment of branch power flow safety: ， in, is the branch offset, with a value range of [0,1], and m is the number of branches; Indicates the corresponding branch i The safety offset coefficient; when , Equal to 1; when , equal If the value is less than or equal to 1, there will be no overload and the branch is safe; if it is greater than 1, there will be an overload, and the larger the value, the less safe it is. For the corresponding branch i power, For the corresponding branch i Power limits; Quantitative assessment of human experience: in, (s,a) represents the state-action pair for performing maintenance on equipment a under a certain power grid state s; Experience is a set of human experience records the state-action pairs that tend to perform maintenance on equipment under a certain power grid state. Reward value: For the evaluation of the maintenance action pair (s,a) of the equipment under a specified power grid condition, a corresponding reward value will be generated on the agent side; in, This is a comprehensive evaluation value generated from the power grid state data fed back after receiving the corresponding (s,a) in the power flow environment simulation. The larger the value, the further it deviates from the safe range. (Threshold value) The value range is (0,1); when the comprehensive evaluation value Greater than the threshold value If the total evaluation value is less than or equal to the threshold value, the reward value is -1; If the value is 0.5, the reward is 1; if the corresponding (s,a) belongs to human experience, the reward is 1. Obtain the set of equipment expected to be maintained next week and the future power grid operation status data; select information corresponding to the equipment included in the preset equipment list from the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained next week, arrange the best maintenance equipment for each time period in the next cycle according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, and obtain the optimal power grid maintenance equipment arrangement scheme; among them, retrieve the evaluation table record with the closest power grid status similarity quantitative value from the new evaluation knowledge table, select the equipment name with the largest value in the record, add the selected equipment name to list S, until all the quantitative values ​​of the future power grid status are processed, and obtain the optimal power grid maintenance equipment arrangement scheme including the maintenance equipment list S corresponding to the next N time periods; Output the optimal power grid maintenance equipment scheduling scheme.

2. The equipment maintenance plan scheduling optimization method according to claim 1, characterized in that, The steps for obtaining the cycle type of the maintenance plan and creating an assessment knowledge table based on the cycle type include: Obtain the cycle type of the maintenance plan, which can be one day, seven days, monthly, quarterly, or annual; determine the corresponding maintenance cycle period T based on the cycle type; and establish an assessment knowledge table with a scheduling length equal to the maintenance cycle period T. The power grid state similarity quantification value is obtained by calculating historical power grid data.

3. The equipment maintenance plan scheduling optimization method according to claim 1, characterized in that, The steps of acquiring historical power grid data, historical manual operation experience, and maintenance equipment sets specifically include: Acquire historical power grid data, manual historical operation experience, and maintenance equipment set; obtain the data sampling frequency within the maintenance cycle, and determine the number of power grid states N within the maintenance cycle based on the maintenance cycle time period T and the sampling frequency; divide the historical power grid data maintenance cycle time period T into M groups, with each group containing N power grid sections.

4. The equipment maintenance plan scheduling optimization method according to claim 3, characterized in that, Maintenance simulations are performed based on historical power grid data and a set of maintenance equipment to obtain feedback on power grid status information, and the evaluation knowledge table is updated based on the feedback on power grid status information. At the same time, the evaluation knowledge table is updated based on historical operational experience. The steps to obtain the final assessment knowledge sheet include: Read the unanalyzed power grid section of the i-th group; Read the j-th power grid section in the i-th group sequentially according to the time sequence; Calculate the quantized value of the power grid state similarity between the current power grid section state and the reference state. The system then retrieves records with corresponding grid state quantification values ​​from the assessment knowledge table. If no records with corresponding grid state similarity quantification values ​​are found, the system searches the assessment knowledge table based on grid state similarity quantification values. Create a new record, and initialize the device evaluation value for the corresponding status to 0; If the historical operation experience is analyzed, and it is determined that the current power grid status requires equipment maintenance, a reward value is calculated and added to the corresponding power grid status and equipment evaluation knowledge table value, and the next section is read. If the current status does not exist in the historical operation experience, a power flow environment simulation is performed to obtain feedback power grid status information, and the evaluation knowledge table is updated based on the feedback power grid status information. After all power grid sections in Group M have been processed, the final evaluation knowledge table is obtained.

5. The equipment maintenance plan scheduling optimization method according to claim 1, characterized in that, The steps for obtaining future power grid operation status data specifically include: Based on the maintenance cycle period T and the power grid data sampling period, determine the number of power grid states N within the maintenance cycle, and obtain N future power grid operation status data.

6. The equipment maintenance plan scheduling optimization method according to claim 5, characterized in that, Based on the quantitative values ​​of the future power grid state and the set of equipment expected to be maintained in the next week, and according to the evaluation values ​​of the power grid state and the maintenance equipment in the new assessment knowledge table, the steps to arrange the optimal maintenance equipment for each time period in the next cycle and obtain the optimal power grid maintenance equipment arrangement scheme include: Generate a list L of N power grid state similarity quantification values, and simultaneously create an empty list S; Read the power grid state similarity quantization value from list L one by one; The evaluation table record with the closest grid state similarity quantification value is retrieved from the new evaluation knowledge table, and the device name with the largest value is selected from the record. Each row of the new evaluation knowledge table corresponds to a grid state similarity quantification value. For each future state, a grid state similarity quantification value is formed. Find the quantification value of similarity with the power grid state. Find the closest corresponding row, with each column in each row corresponding to a device, and then find the device with the largest cumulative evaluation value. Add the selected device name to list S; If it is not the last value in L, read the next grid state similarity quantization value in list L; after processing one list L, process the next list L, until all lists L are processed to obtain the optimal grid maintenance equipment scheduling scheme; the optimal grid maintenance equipment scheduling scheme includes maintenance equipment lists S corresponding to the next N time periods.

7. A device for optimizing equipment maintenance planning, characterized in that, include: The assessment knowledge table creation module is used to obtain the cycle type of the maintenance plan and create an assessment knowledge table based on the cycle type. Historical power grid data, manual historical operation experience, and a set of maintenance equipment are acquired. The created assessment knowledge table contains rows representing the power grid state similarity quantification value and columns representing the equipment names in the maintenance equipment set. The power grid state similarity quantification value is obtained from historical power grid data in the following way: For each moment of power grid operation data, the active power and active power limit values ​​of all lines and high-voltage side branches of transformers in the power grid are acquired, forming a vector Vp representing the current power grid state and a reference vector Vb representing the current power grid state, respectively. A cosine similarity function is used to generate the power grid state similarity quantification value between the current power grid state and the reference state. The system simulates maintenance based on historical power grid data and a set of equipment under maintenance, obtaining feedback on power grid status information and updating the evaluation knowledge table using this information. Simultaneously, it updates the evaluation knowledge table based on historical manual operation experience, resulting in the final evaluation knowledge table. The simulation process consists of three parts: a maintenance agent, a message router, and a power flow simulator (Psenv). The maintenance agent forms the corresponding evaluation knowledge table based on power grid status similarity values ​​and the list of equipment to be inspected. The message router is only responsible for optimizing the allocation of simulation tasks and providing feedback on the information after the maintenance simulation. The power flow simulator (Psenv) simulates maintenance operations on specified equipment under specified power flow conditions and provides feedback on power grid status information. The message router uses network sockets created based on the server IP and port of its entity as data relay nodes and allocates power flow simulation tasks. When outputting messages, the maintenance agent sends the power grid status and maintenance command as the message body to the message router, parses and saves the power grid status after the simulated maintenance operation upon receiving the message. The power flow simulator (Psenv) contains multiple program entities with the same function. In the absence of computational tasks, the system periodically sends empty messages to the message routing router. Upon receiving the grid status and maintenance command assigned for forwarding, it simulates maintenance operations based on the corresponding grid status to perform power flow simulation calculations. After completion, it sends the simulated grid status as a message feedback to the message routing router. During power flow environment simulation, a random variable r between 0 and 1 is generated. When r < ε, the device name with the highest evaluation value corresponding to the current grid status similarity quantization value is selected; otherwise, a device name is randomly selected. The selected device name and the current grid status are forwarded to the power flow environment simulator via the message routing router. The power flow environment simulator Psenvi, with the longest idle time, obtains the current grid status and the selected device name, performs maintenance simulation on the received current grid status and selected device, obtains grid status information, and then feeds back the simulated grid status information via the message routing router. A security quantification assessment is performed on the fed-back grid status information, and a reward value is calculated. The reward value is added to the evaluation knowledge table value of the corresponding grid status and device. Here, ε is a random threshold value. The security evaluation of the grid status information fed back from the power flow simulation is divided into two aspects: voltage and branch power. Quantitative assessment of voltage safety: in, is the voltage offset, with a value range of [0,1], where n is the number of bus voltages and v is the per-unit value of the corresponding node voltage; Quantitative assessment of branch power flow safety: ， in, is the branch offset, with a value range of [0,1], and m is the number of branches; Indicates the corresponding branch i The safety offset coefficient; when , Equal to 1; when , equal If the value is less than or equal to 1, there will be no overload and the branch is safe; if it is greater than 1, there will be an overload, and the larger the value, the less safe it is. For the corresponding branch i power, For the corresponding branch i Power limits; Quantitative assessment of human experience: in, (s,a) represents the state-action pair for performing maintenance on equipment a under a certain power grid state s; Experience is a set of human experience records the state-action pairs that tend to perform maintenance on equipment under a certain power grid state. Reward value: For the evaluation of the maintenance action pair (s,a) of the equipment under a specified power grid condition, a corresponding reward value will be generated on the agent side; in, This is a comprehensive evaluation value generated from the power grid state data fed back after receiving the corresponding (s,a) in the power flow environment simulation. The larger the value, the further it deviates from the safe range. (Threshold value) The value range is (0,1); when the comprehensive evaluation value Greater than the threshold value If the total evaluation value is less than or equal to the threshold value, the reward value is -1; If the value is 0.5, the reward is 1; if the corresponding (s,a) belongs to human experience, the reward is 1. The scheme optimization module is used to obtain the set of equipment expected to be maintained in the next week and the future power grid operation status data; select information corresponding to the equipment included in the preset equipment list from the final evaluation knowledge table to form a new evaluation knowledge table; obtain the quantitative value of the future power grid status based on the future power grid operation status data; based on the quantitative value of the future power grid status and the set of equipment expected to be maintained in the next week, according to the power grid status and the evaluation value of the maintenance equipment in the new evaluation knowledge table, arrange the best maintenance equipment for each time period in the next cycle to obtain the optimal power grid maintenance equipment arrangement scheme; among them, the evaluation table record with the closest power grid status similarity quantitative value is retrieved from the new evaluation knowledge table, the equipment name with the largest value is selected from the record, and the selected equipment name is added to list S, until all the quantitative values ​​of the future power grid status are processed, and the optimal power grid maintenance equipment arrangement scheme including the maintenance equipment list S corresponding to the next N time periods is obtained; The output module is used to output the optimal power grid maintenance equipment scheduling scheme.

8. An electronic device, characterized in that, It includes a processor and a memory, the processor being used to execute a computer program stored in the memory to implement a device maintenance plan scheduling optimization method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, which, when executed by a processor, implements a method for optimizing equipment maintenance schedules as described in any one of claims 1 to 6.

Images