Hierarchical management and control method and system for power transmission and transformation engineering risk operation

Abstract

The application discloses a hierarchical management and control method and system for power transmission and transformation engineering risk operation, belongs to the field of computer data processing and engineering project risk supervision, and comprises the following steps: extracting engineering data and constructing correlation graph data based on time sequence and logs; performing topological weighted accumulation calculation to generate a time sequence risk pressure prediction sequence; projecting to generate a two-dimensional management and control pressure potential field matrix and extracting provincial and local two-level management and control boundary coordinate sets according to quantile numbers; generating a management and control token set with a check slot based on boundary positioning nodes; issuing tokens to identify empty violations to freeze permissions and update the matrix; and combining backhaul data to update the graph data by weighting iteration. The application adopts graph calculation and space-time data processing technology, can realize quantitative deduction of risk evolution and dynamic allocation of management and control resources, and improves the data objectivity of cross-level engineering supervision and the automatic closed-loop processing efficiency.

Description

Technical Field

[0001] This invention relates to the fields of computer data processing and engineering project risk monitoring, and in particular to a hierarchical control method and system for risk management in power transmission and transformation engineering. Background Technology

[0002] The construction and operation of power transmission and transformation projects involve numerous high-risk operations, with risk factors exhibiting complexity and dynamic changes. Within enterprise and project management systems, accurately assessing and classifying on-site operational risks and controlling resource allocation through computer data analysis is a core business data processing step to ensure the safe progress of projects.

[0003] Among related technologies, Chinese invention patent application CN121810034A discloses a method and system for dynamic risk classification and assessment of transmission line inspection areas based on the Conformer model. The method includes: spatiotemporal perception fusion, constructing a multi-scale spatiotemporal grid based on the line topology and risk propagation model and performing data fusion; collaborative optimization assessment, inputting a dynamic risk analysis model and outputting the comprehensive risk quantification value of each unit; quality assessment and reliable traceability; resource constraint decision-making, constructing a multi-objective optimization function to obtain the dynamic risk level and corresponding operation and maintenance strategy; and closed-loop evolution, jointly optimizing model parameters based on strategy effectiveness.

[0004] The aforementioned technologies primarily focus on unidirectionally outputting risk levels and operational strategy recommendations through intelligent assessment models. However, they lack mandatory interactive verification mechanisms in the flow of control tasks and on-site execution. Furthermore, the rigid collaborative boundaries between management entities at all levels result in a disconnect between the system's risk scheduling instructions and actual on-site operations. Consequently, risk control measures are difficult to effectively close the loop, leading to low overall efficiency in risk supervision and collaborative management of engineering projects. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a hierarchical management and control method and system for risk-prone operations in power transmission and transformation projects. Employing graph calculation and spatiotemporal data processing technologies, it enables quantitative projection of risk evolution and dynamic allocation of management and control resources, thereby improving the data objectivity and automated closed-loop processing efficiency of cross-level project supervision.

[0006] The above objectives can be achieved through the following approach: A hierarchical control method for risky operations in power transmission and transformation projects includes: extracting task lists, task time series, geographical coordinates, and historical inspection logs from an engineering management database; constructing a directed graph based on the task time series; extracting the frequency of violations and the frequency of temporal overlap from the historical inspection logs and writing them into the node weights and edge weights of the directed graph, respectively, to generate a correlation graph data; extracting the edge weights as transmission factors based on the correlation graph data; performing a weighted cumulative calculation along the topology of the directed graph on the node weights; fusing the task time series to generate a temporal risk pressure prediction sequence for the task list; performing matrix projection on the temporal risk pressure prediction sequence based on the geographical coordinates and the task time series to generate a two-dimensional control pressure potential field matrix; calculating the quantiles of the cell values ​​in the two-dimensional control pressure potential field matrix and extracting the superquantiles. The system counts cells and outputs a set of provincial and municipal level control boundary coordinates with hierarchical labels. Based on the provincial and municipal level control boundary coordinates, it locates the target operation node of the associated graph data and generates a control token. It compares the hierarchical labels and writes the provincial verification permit slot or municipal verification permit slot to the control token, outputting a set of control tokens to be allocated. It sends the set of control tokens to be allocated to the control terminal and receives the returned slot filling data. It extracts violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, writes an operation permission freeze flag to the violation token, and uses the violation token to update the two-dimensional control pressure potential field matrix. It extracts the actual filling timestamp and hierarchical identity parameter of the slot filling data, extracts the node weight as the calculation base, performs weighted iterative calculation based on the actual filling timestamp and hierarchical identity parameter, and updates the associated graph data.

[0007] Optionally, generating the associated graph data includes: extracting independent work items from the task list and mapping them as graph nodes; comparing the planned start and end times of the task time series; inserting directed connections between adjacent graph nodes in chronological order to construct a directed graph; parsing the historical verification logs; counting the total number of abnormal records for the independent work items as the violation frequency; using the number of days of intersection of the nodes of the directed graph at the planned start and end times as the temporal overlap frequency; and writing the violation frequency and the temporal overlap frequency into the node weights and edge weights of the directed graph, respectively, to generate the associated graph data.

[0008] Optionally, the step of fusing the task time series to generate the time-series risk and pressure prediction sequence for the job task list includes: extracting the edge weights as transmission factors based on the association graph data; traversing the upstream neighboring nodes of the target node along the topology of the directed graph; performing a product operation between the node weights and the transmission factors; adding the product to the node weights of the target node; and outputting the node aggregated risk base; extracting continuous time windows of the task time series; mapping the node aggregated risk base according to the continuous time windows; and concatenating the mapped values ​​under each slice index to generate the time-series risk and pressure prediction sequence for the job task list.

[0009] Optionally, the output of the provincial and municipal level control boundary coordinate set with hierarchical labels includes: extracting the extreme values ​​of latitude and longitude of the geographical coordinates and the start and end times of the task time series, constructing a spatiotemporal two-dimensional projection coordinate system, filling the values ​​in the time series risk pressure prediction sequence into the spatiotemporal two-dimensional projection coordinate system according to the coordinate matching relationship, and generating a two-dimensional control pressure potential field matrix; extracting the cell values ​​in the two-dimensional control pressure potential field matrix to construct a numerical distribution histogram, calculating the cumulative probability distribution curve of the numerical distribution histogram and extracting the inflection point of the curve as a quantile, comparing and extracting superquantile cells with values ​​greater than the quantile, extracting the vertex coordinate data of the superquantile cells and writing them into the control level code as a hierarchical label, and outputting the provincial and municipal level control boundary coordinate set with hierarchical labels.

[0010] Optionally, the output set of control tokens to be allocated includes: extracting the spatial coordinates of the graph nodes in the associated graph data, performing spatial inclusion relationship calculation between the spatial coordinates of the graph nodes and the provincial and municipal level control boundary coordinate sets, extracting them as target operation nodes, creating an empty state data object containing the target operation nodes as a control token; extracting the hierarchical marker of the target operation node in the provincial and municipal level control boundary coordinate sets, performing string matching and parsing of the hierarchical marker, and writing the provincial verification permit slot or municipal verification permit slot into the memory address range of the control token, and outputting the set of control tokens to be allocated.

[0011] Optionally, the method further includes: extracting the local numerical matrix block of the target operation node in the two-dimensional control pressure potential field matrix, performing a tensor inner product operation on the local numerical matrix block and the time-series risk pressure prediction sequence, and outputting a spatiotemporal risk evolution trend tensor.

[0012] Optionally, updating the two-dimensional control pressure potential field matrix using the violation token includes: monitoring the data feedback port of the control terminal to receive the slot filling data; comparing the receiving timestamp of the slot filling data with the planned start time of the task time series; extracting control tokens whose timestamps are out of bounds and whose filling data has not been parsed as violation tokens; changing the control field of the violation token to write a job permission freeze flag; extracting the matrix trace of the spatiotemporal risk evolution trend tensor as a dynamic penalty base; performing scalar addition calculation on the dynamic penalty base and the pressure value of the two-dimensional control pressure potential field matrix; and updating the two-dimensional control pressure potential field matrix.

[0013] Optionally, updating the association graph data includes: parsing the slot filling data, comparing the actual filling timestamp with the planned start time of the task time series to generate a time offset scalar, performing hash mapping encoding on the hierarchical identity parameters to output identity weight coefficients; extracting the node weights as the calculation base, performing continuous multiplication of the calculation base, the time offset scalar, and the identity weight coefficients to generate new node weights, and using the new node weights to update the association graph data.

[0014] Based on the same inventive concept, this invention also provides a hierarchical management and control system for risky operations in power transmission and transformation projects. The system includes: a correlation graph construction module, used to extract task lists, task time series, geographical coordinates, and historical verification logs from an engineering management database; construct a directed graph according to the task time series; extract the frequency of violations and the frequency of time-series overlap from the historical verification logs and write them into the node weights and edge weights of the directed graph, respectively, to generate correlation graph data; a time-series risk prediction module, used to extract the edge weights as transmission factors based on the correlation graph data; perform weighted cumulative calculations on the node weights along the topology of the directed graph; and fuse the task time series to generate a time-series risk pressure prediction sequence for the task lists; and a control boundary delineation module, used to perform matrix projection on the time-series risk pressure prediction sequence based on the geographical coordinates and the task time series to generate a two-dimensional control pressure potential field matrix; and calculate the number of cells within the two-dimensional control pressure potential field matrix. The system calculates the quantiles of the values ​​and extracts the superquantile cells, outputting a set of provincial and municipal level control boundary coordinates with hierarchical labels; a control token generation module is used to locate the target operation node of the associated graph data based on the provincial and municipal level control boundary coordinate set and generate a control token, compares the hierarchical labels and writes the provincial verification permit slot or municipal verification permit slot to the control token, and outputs a set of control tokens to be allocated; a violation interception feedback module is used to send the set of control tokens to be allocated to the control terminal and receive the returned slot filling data, extract violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, write an operation permission freeze flag to the violation token, and use the violation token to update the two-dimensional control pressure potential field matrix; a graph weight update module is used to extract the actual filling timestamp and hierarchical identity parameter of the slot filling data, extract the node weight as the calculation base, perform weighted iterative calculation based on the actual filling timestamp and hierarchical identity parameter, and update the associated graph data.

[0015] Compared with the prior art, the present invention has the following advantages: 1. By constructing a correlation graph that integrates historical verification logs and time-series overlap frequencies, and performing weighted cumulative calculations along a directed topology, this method changes the traditional static risk assessment model that relies on manual experience. This approach can capture the hidden risk transmission effects between complex construction procedures, generate time-series risk pressure prediction sequences, and improve the objectivity and scientific rigor of risk assessment. 2. By extracting and calculating the quantiles of matrix values, a purely data-driven division of provincial and prefectural control boundaries was achieved. This spatial grid-based scheduling mechanism breaks down the traditional, rigid administrative jurisdiction barriers, enabling high-level security control resources to automatically tilt towards high-pressure risk red zones, maximizing the efficiency of cross-regional regulatory resource allocation; 3. A state machine interception mechanism was built using a control token with verification permission slots. When a slot is left unused due to violations, permission freezing and weighted iterative updates of the underlying graph and matrix are automatically triggered. This mechanism not only eliminates management gaps and "starting work with defects" from the underlying algorithm, but also creates a self-renewing chain of "prediction-execution-feedback-correction," giving it the ability to continuously adapt to complex environments.

[0016] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a hierarchical control method for risky operations in power transmission and transformation projects, according to an embodiment of the present invention. Figure 2 This is a numerical mapping and extreme value envelope evolution diagram of the time-series risk pressure prediction sequence according to an embodiment of the present invention; Figure 3 This is a two-dimensional control pressure potential field matrix thermodynamic distribution diagram according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the hierarchical management and control system for risky operations in power transmission and transformation projects according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 One embodiment of the present invention proposes a hierarchical management and control method for risk operations in power transmission and transformation projects. By employing graph calculation and spatiotemporal data processing technology, it is possible to realize the quantitative deduction of risk evolution and the dynamic allocation of management and control resources, thereby improving the data objectivity and automated closed-loop processing efficiency of cross-level engineering supervision.

[0021] The method described in this embodiment specifically includes: Extract the task list, task time series, geographic coordinates and historical verification logs from the engineering management database, construct a directed graph according to the task time series, extract the violation frequency and time series overlap frequency from the historical verification logs and write them into the node weight and edge weight of the directed graph respectively, and generate the associated graph data. Optionally, the generation of association map data includes: Extract the independent task items from the task list and map them as graph nodes. Compare the planned start and end times of the task time series and insert directed connections between adjacent graph nodes in chronological order to construct a directed graph. The association graph construction module reads the task list from the project management database, extracts fields that uniquely identify independent construction tasks, and instantiates each independent construction task as a discrete graph node in memory. The module then reads the task time series bound to each graph node and extracts the planned start and end timestamps. Next, it performs an ascending sort algorithm on all graph nodes based on their planned start timestamps. In the sorted sequence, the module inserts unidirectional directed connections between adjacent preceding and succeeding graph nodes. These directed connections represent the sequential progression of construction work on the physical timeline and the direction of potential risk transmission. The module uses this to construct a directed graph reflecting the temporal topological relationships.

[0022] For example, the association graph construction module extracts three independent task items: foundation pit excavation, tower erection, and conductor stringing. The module maps the foundation pit excavation to node A, the tower erection to node B, and the conductor stringing to node C. By comparing the task time series, the module finds that the planned start timestamp for node A is May 1st, for node B it is May 5th, and for node C it is May 8th. The module then sorts these tasks in ascending order chronologically, inserts a directed connection from node A to node B between nodes A and B, and inserts a directed connection from node B to node C between nodes B and C, thus completing the underlying topology construction of the directed graph.

[0023] The historical verification logs are analyzed, and the total number of abnormal records for the independent work items is counted as the frequency of violations. The number of days of intersection of the nodes of the directed graph at the start and end times of the plan is taken as the temporal overlap frequency. The frequency of violations and the temporal overlap frequency are written into the node weights and edge weights of the directed graph, respectively, to generate associated graph data.

[0024] The association graph construction module extracts historical inspection logs, performs text matching based on the identifier information of graph nodes, and accumulates the total number of times a corresponding graph node is judged as a violation in the historical supervision records, which is taken as the violation frequency. For any two adjacent graph nodes in a directed graph with a directed connection, the module extracts the planned start and end times of each node, calculates the number of calendar days that overlap between the two planned start and end times, and generates the intersection days. To eliminate dimensional differences in subsequent matrix operations, the module introduces a time normalization coefficient to map the intersection days, generating a dimensionless temporal overlap frequency. The module writes the violation frequency into the memory attribute of the corresponding graph node as the node weight, and writes the temporal overlap frequency into the memory attribute of the corresponding directed connection as the edge weight. The calculation relationship between node weight and edge weight satisfies the following formula: , , Among them, letters This represents the node weight of the i-th graph node. (Letter) This represents the frequency of violations recorded in the historical verification log for the independent task corresponding to the i-th graph node. (Letter) Represents the edge weight of the directed connection from the i-th graph node to the j-th graph node. (Letter) This represents the number of days that the i-th graph node and the j-th graph node intersect at the planned start and end times. (Letter) This represents the time normalization coefficient. The association graph construction module iterates through the task list for this assignment, extracts the latest planned end timestamp and the earliest planned start timestamp from all independent task items, calculates the absolute time span in days between the two, and then calculates the time normalization coefficient. The value of is equal to the reciprocal of the absolute time span in days. The physical meaning of this is to transform the absolute number of overlapping days into a relative overlap probability with respect to the total construction period, thereby ensuring that the edge weights of projects of different sizes are all limited to the range of zero to one.

[0025] For example, the association graph construction module parses the historical verification logs and counts that the total number of abnormal records for graph node A is four, and the total number of abnormal records for graph node B is one. The association graph construction module directly assigns the value four to the node weight of graph node A and the value one to the node weight of graph node B. The association graph construction module extracts the planned start and end time of graph node A as May 1st to May 10th, and extracts the planned start and end time of graph node B as May 5th to May 15th. After calculation, the overlapping interval of the two on the time axis is May 5th to May 10th, with an intersection of five days. At the same time, the association graph construction module traverses all tasks and determines that the earliest start time of this project is May 1st, the latest end time is May 20th, and the absolute time span is twenty days. The association graph construction module calculates the time normalization coefficient to be the reciprocal of twenty, which is 0.05. Subsequently, the association graph construction module substitutes the intersection of five days and the time normalization coefficient of 0.05 into the formula. Performing a multiplication operation, five multiplied by 0.05 equals 0.25. The association graph construction module writes the calculated temporal overlap frequency of 0.25 into the directed connection from graph node A to graph node B as an edge weight, completing the full generation of the association graph data.

[0026] Based on the associated graph data, the edge weights are extracted as transmission factors, and the node weights are subjected to weighted cumulative calculation along the topological structure of the directed graph. The task time series is then fused to generate the time-series risk and pressure prediction sequence of the job task list. Optionally, the time-series risk and pressure prediction sequence for generating the job task list by fusing the task time series includes: Based on the associated graph data, the edge weights are extracted as transmission factors. The upstream neighboring nodes of the target node are traversed along the topology of the directed graph. The product operation of the node weights and the transmission factors is performed and added to the node weights of the target node. The node aggregation risk cardinality is then output. The temporal risk prediction module reads the generated association graph data, extracts the edge weights carried by the directed connections in the association graph data, and directly defines and assigns the edge weights to the transmission factor. Following the underlying network topology of the directed graph, the module iterates through each graph node, setting it as a target node. For the current target node, the module searches for its preceding input connections through topological relationships, thereby locking all graph nodes that send these preceding input connections and identifying them as upstream adjacent nodes. The module reads the node weights of all locked upstream adjacent nodes, multiplies each extracted node weight by the transmission factor on its corresponding connection, and obtains the risk component scalar. The module performs an algebraic summation operation on all risk component scalars calculated from multiple upstream adjacent nodes and adds the sum to the target node's initial node weight. Through this calculation, the module mathematically fuses historical static risk with topological transmission risk, ultimately outputting the node aggregate risk cardinality of the target node. The calculation relationship of the node aggregation risk base satisfies the following formula: , Among them, letters Represents the current target node The node aggregation risk cardinality output after topological accumulation. (Letter) Represents the target node Its own original node weight, i.e., the frequency of independent violations. (Letter) Represents the target node The set of all upstream adjacent nodes. Letters Representative set A specific upstream neighboring node in the array. (Letter) Represents upstream adjacent nodes The original node weights. (Letters) Represents upstream adjacent nodes Point to target node The conduction factor corresponding to the directed connection. The physical meaning of this formula lies in simulating the danger amplification effect when high-risk operations overlap in power engineering. Pre-existing abnormal operations. Not only is it dangerous on its own, but it also moves along overlapping timelines. The potential hazards are delayed and passed on to subsequent operations. The product of the two still has the dimension of frequency. The addition operation ensures that the node's own security foundation and the received external interference can be uniformly quantified in linear space.

[0027] For example, the temporal risk prediction module begins traversing the associated graph data. The module sets graph node B as the target node. Searching the directed graph reveals that graph node B has only one preceding input connection from graph node A. The module then identifies graph node A as its unique upstream neighbor. The module extracts the node weight of graph node A as a value of four, and the edge weight connecting graph node A and graph node B as 0.25. The module multiplies the node weight value of four with the transmission factor of 0.25 to obtain the risk component scalar value of one. The module further extracts the node weight of graph node B itself as a value of one. The module performs an addition calculation, adding the risk component scalar value of one to its own node weight of one, resulting in a value of two. The module outputs the final result, value two, as the node aggregation risk cardinality of graph node B. Using the same logic, when the time-series risk prediction module traverses node A in the graph, since it has no upstream adjacent nodes, it directly outputs its aggregate risk base as its own weight value of four.

[0028] Extract continuous time windows from the task time series, map the node aggregation risk base according to the continuous time windows, and concatenate the mapped values ​​under each slice index to generate the time-series risk pressure prediction sequence of the task list.

[0029] The time-series risk prediction module extracts the earliest planned start timestamp of all independent work items in the original task time series as the zero point of the time axis and the latest planned end timestamp as the end point of the time axis. Between these two extremes, the module discretizes the continuous calendar time in the time dimension, using a 24-hour physical segmentation period, constructing a continuous time window composed of multiple discrete periods. The module assigns an incremental slice index to each segmentation period within the continuous time window. Based on the task time series, the module locates the specific date covered by each target node in the real calendar and maps it to a specific set of slice indices. The module assigns the node aggregate risk base calculated for the target node to all the slice index positions it covers as mapping values. When the module detects that a slice index is simultaneously covered by multiple target nodes under parallel construction, it executes a numerical comparison logic, forcibly extracting and retaining the largest scalar value among the aggregate risk bases of these target nodes as the final mapping value for that slice index. The time-series risk prediction module, based on the natural ascending order of the slice indices, concatenates all mapped values ​​into an array to generate the final product, a one-dimensional array structure, which is the time-series risk stress prediction sequence. The data construction logic of the time-series risk stress prediction sequence satisfies the following piecewise function formula: When graph nodes exist Covering slice index hour, When no graph node covers the slice index hour, Among them, letters The generated time-series risk stress prediction sequence is in the first... Predicted values ​​at slice index positions. Letters Represents the target node The node aggregation risk base. Letters Represents the target node The set of slice indices actually covered on the calendar. The logic for selecting the extreme values ​​in this formula follows the "barrel principle" of safety management, which states that the overall risk pressure at a power construction site at the same point in time is not determined by the average state, but directly by the most dangerous task currently in progress.

[0030] For example, the time-series risk prediction module extracts the earliest timestamp of the task time series as May 1st and the latest timestamp as May 20th, and constructs a continuous time window of length of 20 days by dividing it by day, with slice indices labeled 0 to 19. The time-series risk prediction module identifies the planned coverage dates of graph node A as May 1st to May 10th, corresponding to slice indices 0 to 9; and the planned coverage dates of graph node B as May 5th to May 15th, corresponding to slice indices 4 to 14. The time-series risk prediction module fills the node aggregation risk base value of graph node A (4) into indices 0 to 9, and fills the node aggregation risk base value of graph node B (2) into indices 4 to 14. The time-series risk prediction module detects that graph node A and graph node B overlap between slice indices 4 to 9. The time-series risk prediction module compares value 4 with value 2, extracting the maximum value 4 as the final mapping value for slice indices 4 to 9. The time-series risk prediction module then concatenates all values ​​from indices 0 to 19 sequentially. Ultimately, the time-series risk prediction module generated a set of time-series risk pressure prediction sequences. The first ten values ​​were all four, the next five were all two, and the last five indices were automatically filled with zeros due to lack of operational coverage. This sequence directly reflects the temporal evolution of the construction site in May, with the first half characterized by high risk and full load, and the second half by a gradual release of risk. Figure 2 As shown, the overlapping state of the underlying parallel construction operations on the time slice is mapped, and the maximum value envelope algorithm is used to extract the aggregated risk peak value under each cycle, thereby constructing a global continuous risk pressure pattern.

[0031] Based on the geographical coordinates and the task time series, matrix projection is performed on the time series risk pressure prediction sequence to generate a two-dimensional control pressure potential field matrix. The quantiles of the cell values ​​in the two-dimensional control pressure potential field matrix are calculated and the superquantile cells are extracted. The coordinate set of the provincial and local two-level control boundary with hierarchical labels is output. Optionally, the output set of provincial and municipal two-level control boundary coordinates with hierarchical markings includes: Extract the extreme values ​​of latitude and longitude of the geographical location coordinates and the start and end times of the task time series to construct a spatiotemporal two-dimensional projection coordinate system. Fill the values ​​in the time series risk pressure prediction sequence into the spatiotemporal two-dimensional projection coordinate system according to the coordinate matching relationship to generate a two-dimensional control pressure potential field matrix. The control boundary delineation module extracts the minimum and maximum longitude parameters from all geographic coordinates mapped by the associated map data. It then calculates the difference between the maximum and minimum longitude parameters to obtain the absolute spatial longitude span of the entire project task set. Simultaneously, the module extracts the earliest planned start timestamp and the latest planned end timestamp from the task time series, calculating the total number of natural days between them as the absolute time span. The module constructs a spatiotemporal two-dimensional projected coordinate system with the absolute time span as the horizontal axis and an integer containing the total number of all map nodes as the vertical axis (total grid count). The module iterates through each target task node, converting its geographic coordinate longitude parameters into discrete spatial indices in the spatiotemporal two-dimensional projected coordinate system. The mapping logic of these spatial indices satisfies the following formula: , Among them, letters This represents the discrete space index generated by the mapping, i.e., the row number of the coordinate system's vertical axis. (Function) This represents the mathematical operation of rounding down. (Letter) The total number of graph nodes representing the extracted independent task items is used as the resolution cardinality for spatial tiling, ensuring that each node, in principle, has an independent projection row. (Letter) This represents the longitude coordinates of the current target task node. (Letter) This represents the longitude parameter with the smallest extracted value. (Letter) This represents the absolute span of spatial longitude. The physical meaning of this formula is to linearly and dimensionlessly compress continuous absolute geographical longitude into a discrete matrix row index interval according to the spatial span ratio. After obtaining the spatial index and time slice index, the control boundary delineation module directly writes the values ​​in the time-series risk pressure prediction sequence into the cells where the spatial index and time slice index intersect, according to their corresponding time slices. If there is no job coverage, the cell value defaults to zero. After completing the full mapping, a two-dimensional control pressure potential field matrix is ​​output. For example... Figure 3 As shown, through grayscale step changes, the superposition and evolution trend of risk pressure of multiple engineering operations under the interweaving of continuous time slices and discrete spatial indices are presented, as well as the high-pressure control boundary characteristics formed at extreme value inflection points.

[0032] For example, the control boundary delineation module extracts fifty map nodes covering the project, with a minimum longitude of 108 degrees and a maximum of 110 degrees, and a spatial longitude absolute span of two degrees. The absolute time span is from May 1st to May 20th, a total of twenty days. The control boundary delineation module constructs a spatiotemporal two-dimensional projected coordinate system of fifty rows and twenty columns. Map node A is extracted, with a longitude of 109 degrees, and substituted into the discrete spatial index calculation formula. Fifty multiplied by the longitude difference of one degree and then divided by the span of two degrees, the spatial index of map node A is finally calculated to be row twenty-five. The operation coverage period of map node A is from May 1st to May 10th, corresponding to time index columns zero to nine. The control boundary delineation module reads the time series risk pressure prediction sequence value four corresponding to this node and fills it into the cells of columns zero to nine of row twenty-five. In this way, the control boundary delineation module completes the grid filling of all nodes and generates a two-dimensional control pressure potential field matrix.

[0033] Extract the cell values ​​within the two-dimensional control pressure potential field matrix to construct a numerical distribution histogram, calculate the cumulative probability distribution curve of the numerical distribution histogram and extract the inflection point of the curve as a quantile, compare and extract the superquantile cells whose values ​​are greater than the quantile, extract the vertex coordinate data of the superquantile cells and write them into the control level code as a level marker, and output the provincial and municipal two-level control boundary coordinate set with level markers.

[0034] The control boundary delineation module traverses all cells of the two-dimensional control pressure potential field matrix, extracting all non-zero values. It then sorts the values ​​in ascending order and counts the total frequency of each independent value in the matrix. The module divides the total frequency of each value by the total number of cells in the matrix to calculate the discrete frequency, thus constructing a histogram of the numerical distribution. Subsequently, the module performs continuous addition operations on the discrete frequencies sequentially along the ascending numerical order to generate a cumulative probability distribution curve. To find the physical threshold for risk mutation, the module performs discrete second-order difference calculations on the cumulative probability distribution curve, the calculation logic of which satisfies the following formula: , Among them, letters The cumulative probability distribution curve represents the curve at the th The second-order difference values ​​at each numerical node. Letters Representing the The cumulative probability corresponding to each numerical node. Letters and These represent the cumulative probabilities corresponding to their preceding and following adjacent nodes, respectively. The control boundary delineation module extracts a specific coordinate point from all calculation results where the sign of the second-order difference numerical value reverses and the absolute value is the largest. This point represents the location where the cumulative probability growth rate decreases sharply, physically representing the critical point of a sudden change from a large area of ​​low risk to a very small number of high-clustered risks. The control boundary delineation module extracts the value corresponding to this specific coordinate point and establishes it as a quantile. The control boundary delineation module compares the cells in the matrix that are greater than this quantile and establishes them as superquantile cells. Based on the row and column index of this cell, it multiplies it in reverse by the spatial and temporal deviation lengths to reconstruct the latitude and longitude boundaries and calendar time boundaries in the real environment, which are used as vertex coordinate data. Furthermore, the control boundary delineation module determines whether the value inside the superquantile cell is greater than or equal to twice the quantile. If it meets the condition of twice or more, the control boundary delineation module writes a string sequence of provincial verification permission to this cell as a level marker; if it is only greater than the quantile but does not meet the condition of twice, it writes a string sequence of municipal verification permission as a level marker. The control boundary delineation module combines the above data to output a set of control boundary coordinates for both provincial and municipal levels with hierarchical labels.

[0035] For example, the control boundary delineation module extracts one thousand cell values ​​and statistically finds that the value of zero accounts for 80%, the value of two accounts for 10%, and the value of four accounts for 10%. The cumulative probability distribution curve shows that the cumulative probabilities for values ​​of zero, two, and four are 80%, 90%, and 100%, respectively. The control boundary delineation module substitutes these values ​​into the second-order difference calculation formula. The calculation process for value two is: 100% plus 80% minus twice 90%, resulting in a second-order difference of zero, at which point the sign changes. The control boundary delineation module establishes value two as a quantile. The control boundary delineation module iterates through and filters out cells in the matrix with values ​​greater than two, i.e., all cells with a value of four, totaling one hundred, establishing them as superquantile cells and retrieving their actual vertex coordinates. Simultaneously, the control boundary delineation module identifies that the internal value of four in these cells is exactly equal to twice the quantile two, triggering a high-order alarm threshold. The control boundary delineation module writes a hierarchical marker representing provincial verification permission to these one hundred cells. Finally, the control boundary delineation module combines the coordinate data with provincial markers to output a set of control boundary coordinates for both provincial and municipal levels with hierarchical markers.

[0036] Based on the coordinate set of the provincial and municipal control boundaries, the target operation node of the associated map data is located and a control token is generated. The hierarchical mark is compared and written into the provincial verification permit slot or municipal verification permit slot of the control token. The set of control tokens to be allocated is output. Optionally, the set of output control tokens to be assigned includes: Extract the spatial coordinates of the graph nodes in the associated graph data, perform spatial inclusion relationship calculation between the spatial coordinates of the graph nodes and the coordinate sets of the provincial and municipal control boundaries, extract them as target operation nodes, and create an empty state data object containing the target operation node as a control token; The control token generation module extracts the real geographical location parameters, i.e., the spatial coordinates of each graph node, from the associated map data. The spatial coordinates of each graph node include longitude and latitude values. The control token generation module reads all vertex coordinate data from the provincial and municipal control boundary coordinate sets in parallel. The vertex coordinate data defines multiple closed rectangular control polygons in a two-dimensional coordinate system. Each rectangular control polygon consists of four extreme values: minimum longitude, maximum longitude, minimum latitude, and maximum latitude. The control token generation module performs a spatial inclusion relationship calculation between the graph node spatial coordinates and the rectangular control polygons. This spatial inclusion relationship calculation logic satisfies the following double-judgment inequality: and , Among them, letters With letters Representing graph nodes respectively The longitude and latitude values. Letters With letters These represent the minimum and maximum longitude extremes of a specific rectangular control polygon within the provincial and municipal level control boundary coordinate set. (Letter) With letters This represents the minimum and maximum latitude extrema of the specific rectangular control polygon. (If the graph nodes...) The coordinates of the node simultaneously satisfy the two boundary constraint inequalities mentioned above. The control token generation module determines that the node physically falls entirely within the high-pressure control zone, extracts it, and specifically marks it as a target job node. Subsequently, the control token generation module instantiates an empty state data object in the underlying memory using an object-oriented programming paradigm, injects the unique identification code of the target job node into the attribute field of this empty state data object, thereby materializing the empty state data object into a control token. At this point, the control token and the target job node have completed a one-to-one hard binding, but its verification permission field remains in an empty state with memory set to zero.

[0037] For example, the control token generation module extracts the spatial coordinates of graph node A as 109 degrees east longitude and 34 degrees north latitude. The module then reads a rectangular control polygon from the provincial and regional control boundary coordinate set, with a minimum longitude of 108 degrees, a maximum longitude of 110 degrees, a minimum latitude of 33 degrees, and a maximum latitude of 35 degrees. Substituting these coordinates into a inequality check, the module identifies that 109 degrees strictly lies between 108 and 110 degrees, and 34 degrees strictly lies between 33 and 35 degrees. Having passed the spatial inclusion relationship check, the module extracts graph node A as the target job node. Next, the module allocates a new address in the memory heap, creates an empty state data object, writes the graph node A's identity code 001 into the object's header, and generates a unique control token bound to graph node A.

[0038] Extract the hierarchical marker of the target operation node in the coordinate set of the provincial and municipal control boundaries, perform string matching and parsing of the hierarchical marker, write the provincial verification permission slot or municipal verification permission slot into the memory address range of the control token, and output the set of control tokens to be allocated.

[0039] The control token generation module extracts the hierarchical markers that are centrally mounted on the polygon based on the mapping relationship between the target operation node and the rectangular control polygon. The module then calls underlying standard library functions to perform string matching and parsing of the hierarchical markers. The module pre-writes two standard judgment constants into the underlying registers: a fixed-length string constant representing provincial-level authority and a fixed-length string constant representing municipal-level authority. The module compares the hierarchical markers byte-by-byte with the standard judgment constants. If the extracted hierarchical markers are completely identical to the string constants representing provincial-level authority in their underlying bytecode, the module addresses and locks the corresponding memory address segment of the control token, overwriting a high-order binary permission feature code into that segment. This feature code is then physically instantiated as a provincial-level verification permission slot. Conversely, if the hierarchical markers are completely identical to the string constants representing municipal-level authority, the module writes a low-order permission feature code to instantiate a municipal-level verification permission slot. After the control token generation module completes the logical matching and slot writing operations for all target job nodes, it encapsulates all processed control tokens into a serialized array and packages and outputs the set of control tokens to be allocated.

[0040] For example, the control token generation module extracts the hierarchical marker "PROVINCE" bound to the rectangular control polygon where graph node A is located. The module performs string matching and parsing, comparing this string with the standard decision constants in the underlying registers using bytecode exhaustive comparison. The module accurately identifies a perfect match with the preset provincial-level permission string constant. The module then obtains the pointer to the control token corresponding to graph node A, addresses its memory address segment, executes the underlying write instruction, and writes the hexadecimal signature representing the provincial-level permission into that address segment, thus officially solidifying it as a provincial-level verification permission slot. Finally, the module pushes this control token and other assembled tokens onto the set stack, outputting a set of one hundred unassigned control tokens.

[0041] Optionally, the method further includes: Extract the local numerical matrix block of the target operation node in the two-dimensional control pressure potential field matrix, perform tensor inner product operation on the local numerical matrix block and the time series risk pressure prediction sequence, and output the spatiotemporal risk evolution trend tensor.

[0042] To support dynamic penalty feedback, the control token generation module needs to extract an evolutionary tensor. The module reads the spatial slice index and the covered continuous time slice interval of the target job node in the two-dimensional control pressure potential field matrix. To quantify the risk squeezing and diffusion effect of the surrounding environment on the central node, the module uses this spatial slice index as the absolute center and extends it upwards and downwards by a fixed grid physical step along the vertical axis, extracting a neighborhood space set composed of three consecutive spatial slice indices. Based on the two-dimensional cutting boundary formed by the neighborhood space set and the continuous time slice interval, the module cuts and extracts a sub-matrix composed of multiple rows and columns of continuous values ​​from the two-dimensional control pressure potential field matrix, defining it as a local numerical matrix block. Simultaneously, the module precisely extracts numerical segments from the time-series risk pressure prediction sequence that are perfectly aligned with the aforementioned continuous time slice interval, reducing their dimensionality to a one-dimensional time vector. The module performs a tensor inner product operation on the local numerical matrix block and the one-dimensional time vector. This tensor inner product operation is essentially a matrix shrinking calculation performed in the time series dimension, and its underlying mathematical logic satisfies the following formula: , Among them, letters The tensor representing the spatiotemporal risk evolution trend of the final output in the spatial dimension is... The projected component values ​​on each neighboring slice. Letters A set of indices representing the continuous time slice intervals covered by the target job node. (Letters) Representative set The specific loop time index. Letters Represents the local numerical matrix block in the th The spatial slice and the first The original matrix element values ​​at the intersection of each time slice. (Letter) Represents a one-dimensional time vector at the th The values ​​are located on a time slice. The elements of the local numerical matrix block and the elements of the one-dimensional time vector retain the dimension of "violation frequency" in their respective upstream calculations. The physical meaning of this formula is that it transforms isolated spatial distribution matrices... Utilizing global time evolution trends Partial multiplication and weighted summation are performed. The product operation elevates the dimension to the "square of the frequency," quantifying the intensity of the violent oscillations of risk energy at this spatiotemporal intersection point. The control token generation module eliminates the time dimension through this summation operation, reducing the dimensionality of the two-dimensional matrix to a first-order spatiotemporal risk evolution trend tensor.

[0043] Send the set of control tokens to be assigned to the control terminal and receive the returned slot filling data, extract the violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, write the job permission freeze flag to the violation tokens, and use the violation tokens to update the two-dimensional control pressure potential field matrix. Optionally, updating the two-dimensional control pressure potential field matrix using the violation token includes: The monitoring terminal's data feedback port is used to receive the slot filling data. The receiving timestamp of the slot filling data is compared with the planned start time of the task time sequence. The control token with an out-of-bounds timestamp and no filling data is parsed is extracted as a violation token. The violation interception and feedback module opens an asynchronous non-blocking Transmission Control Protocol (TCP) network socket at the server level and continuously listens to the data return port of the designated management terminal. The module receives slot filling data returned by the management terminal via the network link, extracts the absolute time of arrival of the slot filling data at the server's physical network card, and uses this as the received timestamp. Based on the identity key carried in the received slot filling data, the module reverse-addresses the memory heap and matches it with the previously issued management token, then reads the planned start timestamp of the target job node bound to the management token in the task time sequence. The module performs a logical comparison operation along the time dimension. If the received timestamp is greater than the planned start timestamp in absolute calendar time, the module determines that a timestamp out-of-bounds state has been triggered. If the timestamp out-of-bounds state is confirmed, the module further addresses and reads the provincial or municipal verification permit slot within the management token. If the violation interception and feedback module detects that the value in the memory segment of the slot is an initialized sequence of zero values, meaning that no substantial data entered manually has been parsed, the module determines from its underlying logic that the target job node constitutes a "failure to verify upon expiration" security management violation. The module then specially marks and extracts the data structure of the control token in memory, registers it as a violation token, and thus achieves automated violation interception without human intervention.

[0044] The control field of the violation token is changed to write the job permission freeze flag. The matrix trace of the spatiotemporal risk evolution trend tensor is extracted as the dynamic penalty base. The dynamic penalty base is added to the pressure value of the two-dimensional control pressure potential field matrix, and the two-dimensional control pressure potential field matrix is ​​updated.

[0045] For the extracted violation token, the violation interception and feedback module addresses and locates the control field within the underlying data structure of the violation token. The module overwrites this control field with a specific machine code representing the highest blocking level as a job permission freeze marker, completely cutting off the subsequent flow permission of the construction task corresponding to the violation token at the resource scheduling level. To transform local violations into a risk warning for the global environment, the violation interception and feedback module extracts the spatiotemporal risk evolution trend tensor generated for the target job node bound to the violation token in the preceding calculation. Since the spatiotemporal risk evolution trend tensor is represented as a first-order linear array containing multiple spatiotemporal evolution intensity values, the module performs a generalized matrix trace operation on it, that is, performs a pure algebraic summation operation on all numerical elements within the first-order linear array. To prevent direct superposition from causing a dimensional explosion in the two-dimensional control pressure potential field matrix, the module introduces a tensor dimension constant as the denominator to perform dimensional meanization calculation, ultimately outputting a dynamic penalty base. The calculation logic of the dynamic penalty base satisfies the following formula: , Among them, letters Represents the dynamic penalty base of the calculated output. (Letter) The total number of numerical elements contained in the tensor representing the spatiotemporal risk evolution trend, i.e., the tensor dimension constant defined here. (Letter) This represents a cyclically incrementing index variable, strictly increasing from the number one to the tensor dimension constant. (Letter) The tensor representing the spatiotemporal risk evolution trend is the first The formula extracts the average risk and damage equivalent accumulated by the illegal operation node within a local spatiotemporal space, serving as the mathematical benchmark for global punishment. The source of these numerical elements is entirely based on objective historical deduction, eliminating the subjective arbitrariness of manually preset punishment scores. Subsequently, the violation interception and feedback module traverses each cell in the two-dimensional control pressure potential field matrix, extracts its original pressure value, and performs a constant-level scalar addition calculation with the dynamic punishment base. The violation interception and feedback module then rewrites the newly summed value back into the corresponding cell, completing the global evolution and update of the two-dimensional control pressure potential field matrix.

[0046] For example, the violation interception and feedback module locates the underlying control field of violation token 001, forcibly overwriting the normal permission bit, which originally had a status value of zero, with a value of one, thereby injecting a job permission freeze flag, putting the job in a stopped and locked state. The violation interception and feedback module extracts the spatiotemporal risk evolution trend tensor corresponding to violation token 001. This first-order linear array contains three numerical elements: thirty, sixty, and fifteen. The violation interception and feedback module substitutes these into the dynamic penalty base calculation formula. First, the module performs a chain addition operation on the trace of the generalized matrix: thirty plus sixty plus fifteen, the sum equals one hundred and five. Since this tensor contains three discrete elements, the tensor dimension constant is three. The module divides one hundred and five by three, mathematically obtaining a dynamic penalty base of thirty-five. Next, the module reads the two-dimensional control pressure potential field matrix of the global dimension. The original pressure value of a specific cell in this matrix is ​​four. The module performs scalar addition, adding the original value four to the dynamic penalty base of thirty-five, resulting in a new pressure value of thirty-nine. The violation interception and feedback module safely overwrites the value 39 back to that specific cell. Using the same additive iteration logic, the violation interception and feedback module updates all one thousand cells in the two-dimensional control pressure potential field matrix one by one, raising the overall numerical baseline of the entire matrix by 35. This operation, through purely low-level numerical calculations, maps the phenomenon of a sudden drop in safety management pressure across all time and space dimensions caused by a single construction point's "overdue inspection" violation.

[0047] Extract the actual filling timestamp and hierarchical identity parameters of the slot filling data, extract the node weight as the calculation base, and perform weighted iterative calculation by combining the actual filling timestamp and hierarchical identity parameters to update the association graph data.

[0048] Optionally, updating the association map data includes: The slot filling data is parsed, the actual filling timestamp is compared with the planned start time of the task time series to generate a time offset scalar, and the hierarchical identity parameters are hash-mapped to output identity weight coefficients. The graph weight update module reads the actual filling timestamp and hierarchical identity parameters from the slot filling data returned by the control terminal. The actual filling timestamp represents the absolute time when the on-site verification command was physically triggered by the terminal. The graph weight update module traces back to the task time series through the identifier associated with this filling data, extracting the planned start time and planned end time of the corresponding target operation node. The graph weight update module performs a difference calculation between the actual filling timestamp and the planned start time to obtain the time difference value. To ensure uniformity of units and quantify the attenuation effect of time delay on risk characteristics, the graph weight update module uses the time difference value to calculate a time offset scalar. The calculation of the time offset scalar satisfies the following formula: , Among them, letters This represents the time offset scalar of the calculated output, which is a dimensionless decay ratio parameter. (Letter) This represents the time difference calculated by subtracting the planned start time from the actual timestamp, expressed in hours. The duration of the planned operation span is represented by the absolute value of the difference between the planned end time and the planned start time of the same target operation node in the task time series, expressed in hours. The physical meaning of this formula is that, using the planned duration of the construction operation itself as a reference benchmark for attenuation, the larger the time difference of the verification operation delay, the larger the denominator, and the smaller the calculated time offset scalar, thus objectively quantifying the decrease in the reliability of the safety status caused by verification lag. Simultaneously, the graph weight update module performs hash mapping encoding on the hierarchical identity parameters. The graph weight update module maintains a hash mapping table in memory. The key value of this hash mapping table is the binary feature code of the hierarchical identity parameter, and its corresponding hash value is generated by normalization deduction based on the reciprocal of the legal administrative management coverage radius of different verification entities, thereby eliminating the arbitrariness of subjective settings. The graph weight update module inputs the hierarchical identity parameters into the hash function and looks up the corresponding dimensionless identity weight coefficient in the table.

[0049] For example, the graph weight update module receives slot filling data, extracts the actual filling timestamp as May 3rd at 12:00, and the hierarchical identity parameter as the provincial-level authority feature code. The graph weight update module queries the task time series, obtaining the planned start time of the task as May 1st at 12:00 and the planned end time as May 5th at 12:00. The graph weight update module performs a difference calculation, subtracting the planned start time from the actual filling timestamp, resulting in a time difference of 48 hours. Simultaneously, the graph weight update module calculates the planned end time minus the planned start time, resulting in a planned duration of 96 hours. The graph weight update module substitutes these values ​​into the time offset scalar formula: the numerator is 96, and the denominator is 96 plus 48, equaling 144. Dividing 96 by 144 yields a time offset scalar of 0.67. Furthermore, the graph weight update module inputs the provincial-level authority feature code into a hash function. According to the underlying definition of the hash mapping table, the reciprocal normalized value of the provincial administrative coverage radius is 1.0; therefore, the graph weight update module's hash mapping output identity weight coefficient is 1.0.

[0050] The node weights are extracted as the base for calculation. A new node weight is generated by performing a continuous multiplication operation of the base, the time offset scalar, and the identity weight coefficient. The association graph data is then updated using the new node weights.

[0051] The graph weight update module locates the associated graph data and extracts the node weight of the current target job node from its memory attributes. This node weight represents the frequency of violations recorded for that job item in historical periods, and the graph weight update module extracts it into a register as the calculation base. The graph weight update module calls the underlying arithmetic logic unit to perform a continuous multiplication operation of the calculation base, the time offset scalar, and the identity weight coefficient to generate a new node weight. The logic of the continuous multiplication operation satisfies the following formula: , Among them, letters Represents the weight of the new node generated by consecutive multiplication operations. (Letter) Represents the computational base extracted from the correlation graph data. (Letter) Represents the calculated time offset scalar. (Letter) This represents the identity weight coefficient output by the hash mapping encoding. In this formula, the calculation base retains the original dimension of the violation frequency. Both the time offset scalar and the identity weight coefficient are dimensionless convergence factors. After the multiplication operation, the output new node weight still strictly maintains the dimension of the violation frequency, ensuring the consistency of the calculation dimension throughout the entire chain. The physical meaning of the new node weight is that it dynamically discounts and converges the historical violation frequency, which was originally in a static state, using the timeliness of the verification and the authority level of the verification subject in the actual execution process. The graph weight update module directly overwrites the calculated new node weight into the original memory attribute address of the graph node in the associated graph data, completing the data structure iteration of the associated graph data and providing a brand-new base data that has undergone on-site feedback correction for the risk simulation of the next time period.

[0052] For example, the graph weight update module reads that the initial node weight of the target job node in the associated graph data is a value of six. The graph weight update module extracts the value six as the base for calculation into a register. Based on the pre-calculated parameters, the time offset scalar is 0.67, and the identity weight coefficient is 1.0. The graph weight update module performs a continuous multiplication operation. First, the value six is ​​multiplied by 0.67, resulting in 4.02. Then, 4.02 is multiplied by 1.0, and the final calculation result is 4.02. Since the underlying physical meaning of the node weight represents the absolute abnormal frequency base, the graph weight update module performs the conventional rounding logic on the calculation result to determine the new node weight as a value of four. The graph weight update module addresses the memory block corresponding to the associated graph data, erases the original value six, and safely writes the new value four. This operation completes the evolution and update of the graph at the physical level, objectively reflecting that due to the timely intervention of the provincial authoritative verification, the historically accumulated high-risk weight of this node has been reasonably reduced and recalibrated.

[0053] Based on the same inventive concept, this invention also provides a hierarchical management and control system for risky operations in power transmission and transformation projects, such as... Figure 4 As shown, the system includes: The association graph construction module is used to extract the work task list, task time series, geographical coordinates and historical verification logs from the engineering management database, construct a directed graph according to the task time series, extract the violation occurrence frequency and time series overlap frequency from the historical verification logs and write them into the node weights and edge weights of the directed graph respectively, and generate association graph data. The time-series risk prediction module is used to extract the edge weights as transmission factors based on the correlation graph data, perform weighted cumulative calculations on the node weights along the topological structure of the directed graph, and fuse the task time series to generate the time-series risk pressure prediction sequence of the job task list. The control boundary delineation module is used to perform matrix projection on the time-series risk pressure prediction sequence based on the geographical coordinates and the task time series, generate a two-dimensional control pressure potential field matrix, calculate the quantiles of the cell values ​​in the two-dimensional control pressure potential field matrix and extract the superquantile cells, and output a set of provincial and local control boundary coordinates with hierarchical labels. The control token generation module is used to locate the target operation node of the associated map data based on the coordinate set of the provincial and municipal control boundaries and generate a control token, compare the hierarchical mark and write the control token into the provincial verification permission slot or the municipal verification permission slot, and output the control token set to be allocated. The violation interception and feedback module is used to send the set of control tokens to be allocated to the control terminal and receive the returned slot filling data, extract violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, write a job permission freeze flag to the violation tokens, and use the violation tokens to update the two-dimensional control pressure potential field matrix. The graph weight update module is used to extract the actual filling timestamp and hierarchical identity parameters of the slot filling data, extract the node weight as the calculation base, and perform weighted iterative calculation by combining the actual filling timestamp and hierarchical identity parameters to update the associated graph data.

[0054] It should be noted that the functional division and information interaction between the various modules described above are logical, but in terms of physical implementation, they can be integrated on the same software platform or deployed in a distributed manner. The connections between them represent data flow and control flow, aiming to collaboratively achieve the objectives of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of protection of this invention.

Claims

1. A graded control method for risky operations in power transmission and transformation projects, characterized in that, The method includes: Extract the task list, task time series, geographic coordinates and historical verification logs from the engineering management database, construct a directed graph according to the task time series, extract the violation frequency and time series overlap frequency from the historical verification logs and write them into the node weight and edge weight of the directed graph respectively, and generate the associated graph data. Based on the associated graph data, the edge weights are extracted as transmission factors, and the node weights are subjected to weighted cumulative calculation along the topological structure of the directed graph. The task time series is then fused to generate the time-series risk and pressure prediction sequence of the job task list. Based on the geographical coordinates and the task time series, matrix projection is performed on the time series risk pressure prediction sequence to generate a two-dimensional control pressure potential field matrix. The quantiles of the cell values ​​in the two-dimensional control pressure potential field matrix are calculated and the superquantile cells are extracted. The coordinate set of the provincial and local two-level control boundary with hierarchical labels is output. Based on the coordinate set of the provincial and municipal control boundaries, the target operation node of the associated map data is located and a control token is generated. The hierarchical mark is compared and written into the provincial verification permit slot or municipal verification permit slot of the control token. The set of control tokens to be allocated is output. Send the set of control tokens to be assigned to the control terminal and receive the returned slot filling data, extract the violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, write the job permission freeze flag to the violation tokens, and use the violation tokens to update the two-dimensional control pressure potential field matrix. Extract the actual filling timestamp and hierarchical identity parameters of the slot filling data, extract the node weight as the calculation base, and perform weighted iterative calculation by combining the actual filling timestamp and hierarchical identity parameters to update the association graph data.

2. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The generated association map data includes: Extract the independent task items from the task list and map them as graph nodes. Compare the planned start and end times of the task time series and insert directed connections between adjacent graph nodes in chronological order to construct a directed graph. The historical verification logs are analyzed, and the total number of abnormal records for the independent work items is counted as the frequency of violations. The number of days of intersection of the nodes of the directed graph at the start and end times of the plan is taken as the temporal overlap frequency. The frequency of violations and the temporal overlap frequency are written into the node weights and edge weights of the directed graph, respectively, to generate associated graph data.

3. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The time-series risk and pressure prediction sequence for generating the job task list by fusing the task time series includes: Based on the associated graph data, the edge weights are extracted as transmission factors. The upstream neighboring nodes of the target node are traversed along the topology of the directed graph. The product operation of the node weights and the transmission factors is performed and added to the node weights of the target node. The node aggregation risk cardinality is then output. Extract continuous time windows from the task time series, map the node aggregation risk base according to the continuous time windows, and concatenate the mapped values ​​under each slice index to generate the time-series risk pressure prediction sequence of the task list.

4. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The output set of provincial and municipal level control boundary coordinates with hierarchical markings includes: Extract the extreme values ​​of latitude and longitude of the geographical location coordinates and the start and end times of the task time series to construct a spatiotemporal two-dimensional projection coordinate system. Fill the values ​​in the time series risk pressure prediction sequence into the spatiotemporal two-dimensional projection coordinate system according to the coordinate matching relationship to generate a two-dimensional control pressure potential field matrix. Extract the cell values ​​within the two-dimensional control pressure potential field matrix to construct a numerical distribution histogram, calculate the cumulative probability distribution curve of the numerical distribution histogram and extract the inflection point of the curve as a quantile, compare and extract the superquantile cells whose values ​​are greater than the quantile, extract the vertex coordinate data of the superquantile cells and write them into the control level code as a level marker, and output the provincial and municipal two-level control boundary coordinate set with level markers.

5. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The set of output control tokens to be assigned includes: Extract the spatial coordinates of the graph nodes in the associated graph data, perform spatial inclusion relationship calculation between the spatial coordinates of the graph nodes and the coordinate sets of the provincial and municipal control boundaries, extract them as target operation nodes, and create an empty state data object containing the target operation node as a control token; Extract the hierarchical marker of the target operation node in the coordinate set of the provincial and municipal control boundaries, perform string matching and parsing of the hierarchical marker, write the provincial verification permission slot or municipal verification permission slot into the memory address range of the control token, and output the set of control tokens to be allocated.

6. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The method further includes: Extract the local numerical matrix block of the target operation node in the two-dimensional control pressure potential field matrix, perform tensor inner product operation on the local numerical matrix block and the time series risk pressure prediction sequence, and output the spatiotemporal risk evolution trend tensor.

7. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 6, characterized in that, The step of updating the two-dimensional control pressure potential field matrix using the violation token includes: The monitoring terminal's data feedback port is used to receive the slot filling data. The receiving timestamp of the slot filling data is compared with the planned start time of the task time sequence. The control token with an out-of-bounds timestamp and no filling data is parsed is extracted as a violation token. The control field of the violation token is changed to write the job permission freeze flag. The matrix trace of the spatiotemporal risk evolution trend tensor is extracted as the dynamic penalty base. The dynamic penalty base is added to the pressure value of the two-dimensional control pressure potential field matrix, and the two-dimensional control pressure potential field matrix is ​​updated.

8. The hierarchical control method for risky operations in power transmission and transformation projects according to claim 1, characterized in that, The updating of the associated map data includes: The slot filling data is parsed, the actual filling timestamp is compared with the planned start time of the task time series to generate a time offset scalar, and the hierarchical identity parameters are hash-mapped to output identity weight coefficients. The node weights are extracted as the base for calculation. A new node weight is generated by performing a continuous multiplication operation of the base, the time offset scalar, and the identity weight coefficient. The association graph data is then updated using the new node weights.

9. A hierarchical control system for risky operations in power transmission and transformation projects, applied to the hierarchical control method for risky operations in power transmission and transformation projects as described in any one of claims 1-8, characterized in that, The system includes: The association graph construction module is used to extract the work task list, task time series, geographical coordinates and historical verification logs from the engineering management database, construct a directed graph according to the task time series, extract the violation occurrence frequency and time series overlap frequency from the historical verification logs and write them into the node weights and edge weights of the directed graph respectively, and generate association graph data. The time-series risk prediction module is used to extract the edge weights as transmission factors based on the correlation graph data, perform weighted cumulative calculations on the node weights along the topological structure of the directed graph, and fuse the task time series to generate the time-series risk pressure prediction sequence of the job task list. The control boundary delineation module is used to perform matrix projection on the time-series risk pressure prediction sequence based on the geographical coordinates and the task time series, generate a two-dimensional control pressure potential field matrix, calculate the quantiles of the cell values ​​in the two-dimensional control pressure potential field matrix and extract the superquantile cells, and output a set of provincial and local control boundary coordinates with hierarchical labels. The control token generation module is used to locate the target operation node of the associated map data based on the coordinate set of the provincial and municipal control boundaries and generate a control token, compare the hierarchical mark and write the control token into the provincial verification permission slot or the municipal verification permission slot, and output the control token set to be allocated. The violation interception and feedback module is used to send the set of control tokens to be allocated to the control terminal and receive the returned slot filling data, extract violation tokens that have exceeded the planned start time in the task time series and whose slots are empty, write a job permission freeze flag to the violation tokens, and use the violation tokens to update the two-dimensional control pressure potential field matrix. The graph weight update module is used to extract the actual filling timestamp and hierarchical identity parameters of the slot filling data, extract the node weight as the calculation base, and perform weighted iterative calculation by combining the actual filling timestamp and hierarchical identity parameters to update the associated graph data.

Images