Oil and gas field data acquisition and processing method, device and system based on a honkong system

Abstract

The application relates to the technical field and particularly relates to an oil and gas field data acquisition processing method, device and system based on a Hongmeng system, wherein in a data processing process, network delay records, bandwidth occupation ratios and data packet size sequences of a wellhead acquisition end are input into a long short-term memory network in a time sequence for recursive calculation, time sequence modeling of a delay trend and abnormality determination are realized, time nodes with an overrun risk are identified, a synchronous comparison method of a delay value and a bandwidth ratio is used, and descending order sorting of data packet sizes is combined, so that preferential selection of high-value nodes and binding management of link identification are effectively realized, node feature aggregation of a graph neural network is introduced when merging oil pipeline monitoring node delays and bandwidth values, and data features and business demands are combined at each link for targeted optimization, so that the acquisition and processing process has good performance improvement in delay control, link resource optimization and task scheduling efficiency.

Description

Technical Field

[0001] This invention relates to the field of technology, and in particular to a method, apparatus and system for oil and gas field data acquisition and processing based on the HarmonyOS system. Background Technology

[0002] The field of data acquisition technology encompasses a series of processes involving the acquisition, transmission, reception, and storage of data from specified sources using computer systems and operating platforms. Specifically, it involves the automated, batch, and periodic acquisition of data through software operating environments and data input interfaces. In oil and gas field applications, this field includes the acquisition of multiple real-time and timed data points such as wellhead production parameters, pressure, temperature, flow rate, and equipment operating status, and the reliable transmission of these data to the processing end via network communication, supporting subsequent storage and computation. This field requires the acquisition system to have compatibility with underlying device protocols, data format parsing capabilities, and fault tolerance for abnormal acquisition.

[0003] The oil and gas field data acquisition and processing method based on the HarmonyOS system refers to a data acquisition and processing program running in the HarmonyOS operating system environment. It is used to acquire multiple production and operation data from oil and gas field production and monitoring equipment, and to perform the data organization, storage and calculation process in the local system and remote server. The purpose is to achieve centralized management and highly consistent recording of oil and gas field production data, and to ensure that the acquired data is not lost or tampered with during transmission and processing. This provides a complete and reliable data foundation for oil and gas field production scheduling, equipment maintenance, safety monitoring and production optimization, which can improve the level of informatization of oil and gas field production, shorten fault response time and improve the accuracy of production operation decision-making.

[0004] While existing data acquisition and processing models can collect and transmit wellhead production parameters, pressure, temperature, flow rate, and equipment operating status in oil and gas field environments, they are mostly driven by fixed cycles and static logic in the data processing chain. They lack proactive prediction and targeted adjustment of acquisition task delay trends, resulting in delay-sensitive data not being transmitted in a timely manner during network fluctuations and high concurrency of multiple tasks. In node selection, existing solutions usually rely on a single parameter to screen node status, failing to fully correlate multiple types of business parameter information. This may result in the inability to prioritize key nodes when link resources are strained. In network path selection, the lack of multi-dimensional integration of node and adjacent node status information can easily lead to problems such as high-load links not being identified in a timely manner and inefficient traffic distribution. In task scheduling, static allocation strategies are mostly used without considering the dynamic balancing of the current workload of core tasks, which can easily lead to overload of some core tasks, causing problems such as delayed reporting of key monitoring data, decreased link resource utilization, and reduced overall operating efficiency. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method, apparatus, and system for oil and gas field data acquisition and processing based on the HarmonyOS system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for oil and gas field data acquisition and processing based on the HarmonyOS system, comprising the following steps:

[0007] Step 1: By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, sort the parameters of each period by time and input them into the Long Short-Term Memory network to recursively generate trend values ​​and calculate adjacent differences. Compare these values ​​with the original sequence synchronously to determine if any exceedances are made and mark them, thereby generating a delay warning node set.

[0008] Step 2: Based on the delay warning node set, read the delay value and bandwidth ratio and compare and filter them synchronously. Extract the nodes with large delay and high bandwidth and rearrange them in descending order according to the data packet size. Record the communication link identifier and obtain the priority link node table.

[0009] Step 3: Based on the priority link node table, the link delay and bandwidth values ​​of the oil pipeline monitoring nodes are added together in the node feature aggregation unit of the graph neural network to form a weight value, high-load links are identified and diversion nodes are filtered to generate a diversion path allocation table.

[0010] Step 4: Based on the aforementioned distribution path allocation table, read the task type of the compressor station monitoring unit, prioritize inserting the safety monitoring task into the execution queue, compare the length of the core running queue, select the valley queue core allocation task, and obtain the core execution task table.

[0011] Step 5: Based on the core execution task table, read each core task set and compare adjacent execution times. Tasks with a time difference less than the synchronization threshold are assigned to the synchronization group and inserted as high-priority tasks. The records are integrated to form a synchronization priority scheduling table.

[0012] As a further aspect of the present invention, the specific steps for generating the delay warning node set are as follows:

[0013] By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, the three types of parameters for each acquisition cycle are arranged in chronological order. The same parameter values ​​in adjacent cycles are subtracted one by one to obtain the difference. The difference is then paired with the time number and stored in an ordered list to generate a cycle difference dataset.

[0014] Based on the aforementioned periodic difference dataset, the differences and corresponding three types of original parameter values ​​are read sequentially according to the time number. The differences are inserted into the end of the parameter group at the same time node to form a complete parameter unit. The parameter units formed at each time node are input into the input layer of the Long Short-Term Memory Network in sequence and the output is calculated recursively according to the time order. The output results are merged with the original parameter units at the same time node, and continuously sorted and stored in the new sequence to generate a composite trend sequence.

[0015] Based on the composite trend sequence, the delay values ​​of each time node are read sequentially and compared with a set threshold. Time nodes with delay values ​​greater than the threshold are marked and written into a mark list. The marked nodes are combined with parameter units to generate a target set and a delay warning node set.

[0016] As a further aspect of the present invention, the execution process of the Long Short-Term Memory Network is as follows: after reading the difference and the corresponding three types of original parameter values ​​in sequence according to the time number, the difference is inserted into the end of the parameter group at the same time node to form a complete parameter unit. The values ​​of all parameter units are arranged in groups according to the time order and input into the input layer of the memory network. The input of each time node is processed sequentially in the hidden storage unit, and the recursive operation is performed in combination with the state value of the previous time node. The current operation result is fused with the parameter unit of the corresponding time node item by item. The time order of the fusion result is maintained, and each time node is output and added to the newly created sequence to form a data set for generating a composite trend sequence.

[0017] As a further aspect of the present invention, the specific steps for generating the priority link node table are as follows:

[0018] Based on the delay warning node set, the delay value and bandwidth usage ratio of each node are read in chronological order. The nodes with high delay value and large bandwidth ratio are selected by comparing the two. The node number is matched with the corresponding data packet size and stored in the list to generate a high-latency, high-bandwidth node set.

[0019] Based on the high-latency, high-bandwidth node set, the nodes are sorted from largest to smallest according to the data packet size. The communication link identifiers of each node in the sorting result are extracted and inserted into the link list in sequence. The link list is stored in correspondence with the node number to generate a priority link node table.

[0020] As a further aspect of the present invention, the specific steps for generating the traffic splitting path allocation table are as follows:

[0021] Based on the priority link node table, the delay and bandwidth values ​​of each link within the oil pipeline monitoring node are read. The two types of values ​​of the same node and the corresponding values ​​of adjacent nodes are added in the node feature aggregation unit of the graph neural network in chronological order to obtain a merged value. The merged value is written into the data column corresponding to the node and stored in a set according to the node time order to generate a set of link weight values.

[0022] Based on the set of link weight values, the weight value of each node is read and compared with a set load threshold. The node numbers with weight values ​​greater than the threshold are written into the high load list to obtain the candidate link data corresponding to the nodes in the list. A table is formed according to the correspondence between the node and the candidate link information to generate a high load candidate link table.

[0023] Based on the high-load candidate link table, the link with the valley delay is selected by comparing the delay values ​​of the candidate links of each node. When the delays are equal, the link with the bandwidth peak is selected by comparing the bandwidth values. The node number and the selected link are combined into a pair record and added to the result set in the order of the records to generate a traffic distribution path allocation table.

[0024] As a further aspect of the present invention, the execution process of the graph neural network is as follows: based on the priority link node table, the link delay values ​​and bandwidth values ​​of the oil pipeline monitoring nodes and adjacent nodes are loaded as graph structure inputs according to the node numbers. The delay values ​​and bandwidth values ​​of each node are combined in chronological order to form feature vectors, which are then passed to the node feature aggregation unit of the network. The feature vectors of the directly connected nodes are paired one by one in the unit and summed. The summing result is written into the feature storage area of ​​the corresponding node. The node features are rearranged according to the time number and stored in the set as the input data source of the link weight value set.

[0025] As a further aspect of the present invention, the specific steps for generating the core execution task table are as follows:

[0026] Based on the diversion path allocation table, read the task type of the compressor station monitoring unit according to the node number, extract the safety monitoring type node and insert it into the head of the execution queue in the order of task number, add the non-safe tasks to the end of the queue in the original number order, and generate a record table of the correspondence between task number and queue position, and generate a priority task queue list.

[0027] Based on the priority task queue list, the number of tasks in the current queue of each running core is counted, the core with the lowest number of tasks is identified as the lowest queue core, the tasks in the priority task queue list are allocated to the cores in the order of node number, and the allocated task number is bound to the core number and recorded in the allocation list to generate the core execution task table.

[0028] As a further aspect of the present invention, the specific steps for generating the synchronization priority scheduling table are as follows:

[0029] Based on the core execution task table, the task sets are read sequentially according to the core number. The planned execution times of the tasks in each set are arranged in order. The time values ​​of adjacent tasks are subtracted and compared with the synchronization threshold. The task numbers with time differences less than the threshold are filtered out. The filtered numbers are grouped according to the core number and added to the group list to generate a synchronization task grouping table.

[0030] Based on the synchronous task grouping table, obtain the core execution queue position corresponding to the task number of each group, move the tasks of each group to the high priority task, maintain the original order of tasks within the group, establish a data record table corresponding to the adjusted queue and the synchronous task group number, and then merge the data record tables to generate a synchronous priority scheduling table.

[0031] An oil and gas field data acquisition and processing device based on the HarmonyOS system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-mentioned oil and gas field data acquisition and processing method based on the HarmonyOS system.

[0032] The oil and gas field data acquisition and processing system based on the HarmonyOS system is used to execute the aforementioned oil and gas field data acquisition and processing method based on the HarmonyOS system, including:

[0033] Trend Analysis Module: Based on wellhead acquisition terminal delay records, bandwidth usage ratio, and data packet size, the data of each period is sequentially input into the Long Short-Term Memory network to calculate and generate trend values. The difference between the trend values ​​and the original data is compared to mark the time nodes that exceed the limit, and a delay warning node set is generated.

[0034] Preferred Node Module: Based on the latency warning node set, read the node latency value and bandwidth ratio, filter nodes with high latency and large bandwidth, sort them in descending order by data packet size, and record the communication link identifier to form a preferred link node table;

[0035] Link allocation module: Based on the priority link node table, the node delay and bandwidth are aggregated into the graph neural network to generate a merged value, and high-load nodes are filtered by comparing the threshold. Qualified paths are selected according to delay and bandwidth to obtain the traffic distribution path allocation table.

[0036] Core scheduling module: Based on the traffic distribution path allocation table, read the task type, put the security monitoring task into the front end of the execution queue, count the number of each core task, select the core with the lowest number of tasks for task allocation, and obtain the core execution task table.

[0037] Synchronization orchestration module: Based on the core execution task table, compare the adjacent time differences of each core task, classify tasks with time differences less than a threshold into a synchronization group, insert high-priority tasks, and generate a synchronization priority scheduling table.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0039] 1. In this invention, by inputting the network delay record, bandwidth occupancy ratio, and data packet size sequence of the wellhead acquisition terminal into the Long Short-Term Memory network in chronological order for recursive calculation during the data processing, the time-series modeling and anomaly judgment of the delay trend are realized, and time nodes with the risk of exceeding limits are identified in advance.

[0040] 2. In this invention, by using a synchronous comparison method of latency value and bandwidth ratio, key nodes with high latency and high bandwidth are screened, and combined with the descending order of data packet size, the priority selection of high-value nodes and the binding management of link identifiers are effectively realized, so that latency-sensitive tasks are connected in an orderly manner in the overall operation link.

[0041] 3. In this invention, by introducing node feature aggregation of graph neural network when merging delay and bandwidth values ​​of oil pipeline monitoring nodes, and by performing targeted optimization in each step in combination with data features and business requirements, the acquisition and processing process has good performance improvement in terms of delay control, link resource optimization and task scheduling efficiency. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0043] Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0045] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0046] Example 1

[0047] Please see Figure 1This invention provides a technical solution: a method for oil and gas field data acquisition and processing based on the HarmonyOS system, comprising the following steps:

[0048] Step 1: By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, sort the parameters of each period by time and input them into the Long Short-Term Memory network to recursively generate trend values ​​and calculate adjacent differences. Compare these values ​​with the original sequence synchronously to determine if any exceedances are made and mark them, thereby generating a delay warning node set.

[0049] Step 2: Based on the delay warning node set, read the delay value and bandwidth ratio and compare and filter them synchronously. Extract the nodes with large delay and high bandwidth, rearrange them in descending order according to the data packet size, record the communication link identifier, and obtain the priority link node table.

[0050] Step 3: Based on the priority link node table, the link delay and bandwidth values ​​of the oil pipeline monitoring nodes are added together in the node feature aggregation unit of the graph neural network to form a weight value, high-load links are identified and diversion nodes are screened to generate a diversion path allocation table.

[0051] Step 4: Based on the distribution path allocation table, read the task type of the compressor station monitoring unit, prioritize inserting the safety monitoring task into the execution queue, compare the length of the core running queue, select the valley queue core allocation task, and obtain the core execution task table.

[0052] Step 5: Based on the core execution task table, read the core task sets and compare the adjacent execution times. Tasks with a time difference less than the synchronization threshold are assigned to the synchronization group and inserted as high-priority tasks. The records are integrated to form a synchronization priority scheduling table.

[0053] The specific steps for generating the delay warning node set are as follows:

[0054] By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, the three types of parameters for each acquisition cycle are arranged in chronological order. The same parameter values ​​in adjacent cycles are subtracted one by one to obtain the difference. The difference is then paired with the time number and stored in an ordered list to generate a cycle difference dataset.

[0055] Based on the periodic difference dataset, the difference and the corresponding three types of original parameter values ​​are read sequentially according to the time number. The difference is inserted into the end of the parameter group at the same time node to form a complete parameter unit. The parameter units formed at each time node are input into the input layer of the long short-term memory network in sequence and the output is calculated recursively according to the time order. The output results are merged with the original parameter units at the same time node, and continuously sorted and stored into a new sequence to generate a composite trend sequence.

[0056] Based on the composite trend sequence, the delay value of each time node is read sequentially and compared with a set threshold. Time nodes with delay values ​​greater than the threshold are marked and written into the mark list. The marked nodes are combined with parameter units to generate a target set and a delay warning node set.

[0057] By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, the three types of parameters for each acquisition cycle are arranged in chronological order. In the data table, the same parameters in adjacent cycles are subtracted one by one, and the difference results are recorded as decimals for the three types of parameters. The results are paired with the corresponding time numbers in an ordered manner. The time numbers are an integer sequence, and the difference data is a set of floating-point values ​​arranged in sequence. The paired items are stored in ascending order of time number to generate a cycle difference dataset.

[0058] Based on the periodic difference dataset, the difference and the corresponding three types of original parameter values ​​are read sequentially according to the time number. The difference is appended to the end of the parameter group at the same time node to form four data units. The four data units of each time node are input into the input layer of the Long Short-Term Memory Network in chronological order. The network is configured with a fixed time step, a fixed number of hidden units and a single output unit. During the recursive processing, the values ​​stored in the previous state of each time node are weighted and superimposed with the relevant values ​​of the current data unit. The output sequence is then fused with the corresponding original parameter units. The fused sequence is filled into a new data set in chronological order to generate a composite trend sequence.

[0059] Based on the composite trend sequence, the delay values ​​of each time node are read sequentially and compared with a set threshold of 200 milliseconds. The time node numbers with delay values ​​greater than the threshold are recorded as target numbers, and the complete data units corresponding to the numbers are read. Then, the time numbers and data units are combined into a two-element structure in the order of time number and data unit. The two-element structure is added to the set in order to generate a delay warning node set.

[0060] The execution process of the Long Short-Term Memory Network is as follows: after reading the difference and the corresponding three types of original parameter values ​​in sequence according to the time number, the difference is inserted into the end of the parameter group at the same time node to form a complete parameter unit. The values ​​of all parameter units are arranged in groups according to the time order and input into the input layer of the memory network. The input of each time node is processed sequentially in the hidden storage unit, and the recursive operation is performed in combination with the state value of the previous time node. The current operation result is fused with the parameter unit of the corresponding time node item by item. The time order of the fusion result is maintained, and each time node is output and added to the newly created sequence to form a data set for generating composite trend sequences.

[0061] Long Short-Term Memory (LSTM) networks, according to the formula:

[0062]

[0063] in: Indicates the current time step The hidden state output vector. Indicates the previous time step The hidden state output vector. This represents the weight matrix from the hidden state to the hidden state. This represents the weight matrix input to the hidden state. Indicates the current time step The input vector, The weight matrix represents the time interval parameter to the hidden state. Indicates the current time step Compared with the previous time step The time interval scalar, The weight matrix representing the link load ratio parameter to the implicit state. Indicates the current time step The link load ratio scalar. Represents the bias vector;

[0064] Execution process: First, the delay difference, bandwidth usage difference, and data packet size difference for the current period are collected from the wellhead acquisition terminal to form the input vector. Obtain the hidden state of the previous time step. The data is stored in a vector, and the time step interval is calculated by subtracting the acquisition time of the previous cycle from the acquisition time of the current cycle. The load ratio is calculated by simultaneously measuring the instantaneous bandwidth usage of the current link in milliseconds and dividing the current bandwidth usage by the maximum bandwidth capacity. ,Will With weight matrix Multiplication yields the hidden layer historical contribution, which will With weight matrix Multiplication yields the contribution of the real-time input in the current cycle, while With weight matrix Multiplying yields the weighted contribution of the time interval signal, and... With weight matrix Multiplying these results yields the weighted contribution of the link load signal. The above results are then compared with the bias vector. The elements are added together to form a composite vector, and the new hidden state is obtained by calculating the element-wise using the hyperbolic tangent function. .

[0065] The specific steps for generating the priority link node table are as follows:

[0066] Based on the delay warning node set, the delay value and bandwidth usage ratio of each node are read in time sequence. The nodes with high delay value and large bandwidth ratio are selected by comparing the two. The node number is matched with the corresponding data packet size and stored in the list to generate a high-latency high-bandwidth node set.

[0067] Based on a high-latency, high-bandwidth node set, the nodes are sorted from largest to smallest according to the data packet size. The communication link identifiers of each node in the sorting result are extracted and inserted into the link list in sequence. The link list is stored in correspondence with the node number to generate a priority link node table.

[0068] Based on the latency warning node set, the latency value and bandwidth usage ratio of each node are read in chronological order. The latency threshold is set to 150 milliseconds and the bandwidth usage ratio threshold is set to 75%. The latency value and latency threshold, as well as the bandwidth ratio and bandwidth threshold, of each node are compared. The logical AND operation of the two conditions is used to determine whether they meet the filtering criteria. The node numbers that meet the criteria are extracted, and the corresponding data packet size is recorded as an integer value. This data packet size is combined with the node number to form a pair. The data packets are written into a list container in chronological order to generate a high-latency, high-bandwidth node set.

[0069] Based on a high-latency, high-bandwidth node set, nodes are sorted from largest to smallest by data packet size, using data packet size as the primary sorting key and node number as the secondary key. Adjacent elements in the initial node array are compared and swapped until the entire array is sorted in descending order. During the sorting process, the original node number order is maintained synchronously. The communication link identifiers of the sorted nodes are read and added to the link list in the sorting order. A mapping relationship is established between the link list and node numbers in a mapping table. The mapping table records the link identifiers corresponding to each node number, generating a priority link node table.

[0070] The specific steps for generating the routing path allocation table are as follows:

[0071] Based on the priority link node table, the delay and bandwidth values ​​of each link within the monitoring node of the oil pipeline are read. The two types of values ​​of the same node and the corresponding values ​​of adjacent nodes are added in the node feature aggregation unit of the graph neural network in chronological order to obtain the merged value. The merged value is written into the data column corresponding to the node and stored in the set according to the node time order to generate the link weight value set.

[0072] Based on the set of link weight values, the weight values ​​of each node are read and compared with the set load threshold. The node numbers with weight values ​​greater than the threshold are written into the high load list to obtain the candidate link data corresponding to the nodes in the list. A table is formed according to the correspondence between the node and candidate link information to generate a high load candidate link table.

[0073] Based on the high-load candidate link table, the link with the valley delay is selected by comparing the delay values ​​of the candidate links of each node. When the delays are equal, the link with the bandwidth peak is selected by comparing the bandwidth values. The node number and the selected link are combined into a pair record and added to the result set in the order of the records to generate a traffic distribution path allocation table.

[0074] Based on the priority link node table, the delay and bandwidth values ​​of each link within the monitoring node of the oil pipeline are read. A graph neural network node feature aggregation method is used to calculate the sum of the delay and bandwidth values ​​of each node with the corresponding delay and bandwidth values ​​of the adjacent nodes in the same time series. The node dimension in the feature aggregation unit is preset to a two-dimensional array structure, with the first dimension being the node number and the second dimension being a two-element arrangement of delay and bandwidth. During aggregation, the superposition value of the corresponding elements is calculated sequentially according to the index order of the adjacent nodes. The calculated merged value is written into the data storage area of ​​the current node. Then, the merged values ​​of the nodes are arranged in ascending order of time number and recorded in the set to generate a set of link weight values.

[0075] Based on the link weight value set, the weight value of each node is read. During the initialization phase, the load threshold value is set to 0.85. The weight value of each node is compared with the threshold. The judgment condition is that when the weight value of a node is greater than the threshold, it is marked as a high load node. The number of the marked node is added to the high load list. The candidate links corresponding to each node are matched in the list. The candidate link data is read with the node number as the index. The nodes and candidate link data are written into a two-dimensional table according to the correspondence relationship to generate a high load candidate link table.

[0076] Based on the high-load candidate link table, the latency values ​​in the candidate link set are first compared. The latency comparison direction is set to ascending, and the link with the minimum latency value is selected as the target link record of the current node. When multiple links have the same latency value, the second-level condition comparison of bandwidth values ​​is performed. The bandwidth comparison direction is set to descending, and the link with the maximum bandwidth value is selected as the target link of the node. The node number and the selected link data are combined into a pairing entry, and added to the new result set in ascending order of node number to generate the traffic distribution path allocation table.

[0077] The execution process of the graph neural network is as follows: Based on the priority link node table, the link delay values ​​and bandwidth values ​​of the monitoring nodes of the oil pipeline and their adjacent nodes are loaded as graph structure input according to the node number. The delay values ​​and bandwidth values ​​of each node are combined in time order to form a feature vector, which is passed to the node feature aggregation unit of the network. The feature vectors of the directly connected nodes are paired one by one in the unit and summed. The summing result is written into the feature storage area of ​​the corresponding node. The node features are rearranged according to the time number and stored in the set as the input data source of the link weight value set.

[0078] Graph neural networks, according to the formula:

[0079]

[0080] in: Indicates the first The updated feature vector of each node in the current aggregation round Indicates the node number currently being processed. Indicates the first The feature vector of each node in the previous aggregation round This indicates the time step corresponding to the previous node. This represents the weight matrix from the node's feature mapping to the current round, with subscripts... This represents the feature mapping of the current node itself. Indicates the first The set of adjacent nodes of a node. This represents the weight matrix that maps the features of adjacent nodes to the current node, with subscripts... This represents the characteristic part of adjacent nodes. Indicates adjacent nodes The previous round of feature vectors, Indicates that the neighboring nodes are in the set The index position in Represents the input feature mapping matrix, with subscripts... Represents external input features. Indicates the first The input feature vector of each node in the current round, Represents the time interval mapping matrix, with subscripts Represents time, element Indicates the first The time weights corresponding to each feature dimension Indicates the first The data collection time interval between each node and the previous round of calculation Represents the bias vector, elements Indicates adding to the first A constant term for each feature dimension. Represents a non-linear activation function;

[0081] Execution process: First, generate the network topology based on the currently constructed priority link node table, and then match each node with its adjacent node set. Establish connections and generate input feature vectors. It includes the real-time latency value, bandwidth usage ratio, and data packet size collected by the node, and reads the feature vector of the node itself from the previous round. and with matrix Multiply to obtain the mapping value of the self-feature of the current round, and traverse the set. Each neighbor node in Take the feature vector from the previous round. , and matrix Multiply the neighbors' contributions, then combine the node's input features. With matrix Multiplying the two yields the contribution of the external input features. The summation of these two features is then performed sequentially over neighboring nodes, and the time interval of the current node is read. , and matrix After multiplication, the temporal feature contribution is obtained. Finally, the self-feature contribution, neighbor feature contribution, input feature contribution, temporal feature contribution, and bias vector are combined. The feature dimensions are added one by one to form a comprehensive vector.

[0082] The specific steps for generating the core execution task table are as follows:

[0083] Based on the flow path allocation table, read the task type of the compressor station monitoring unit according to the node number, extract the safety monitoring type node and insert it into the head of the execution queue in the order of task number, add the non-safe tasks to the end of the queue in the original number order, and generate a record table of the correspondence between task number and queue position, and generate a priority task queue list.

[0084] Based on the priority task queue list, the number of tasks in the current queue of each running core is counted. Cores with a minimum number of tasks are identified as minimum queue cores. Tasks in the priority task queue list are assigned to cores in order of node number. The assigned task number is bound to the core number and recorded in the allocation list to generate a core execution task table.

[0085] Based on the distribution path allocation table, the task type of the compressor station monitoring unit is read according to the node number. The classification condition is set to the task type field value being equal to the safety monitoring class. Nodes that meet the condition are read in ascending order of task number and inserted into the head of the execution queue in sequence. Tasks that do not meet the condition are added to the end of the queue in the original order of their numbers. A list of corresponding node numbers and queue positions is generated. The positions in the list are recorded in integer order starting from zero. A priority task queue list is generated.

[0086] Based on the priority task queue list, the number of tasks in the current queue of each running core is counted. The core number and the current number of tasks are paired to form a key-value pair record, and sorted by the number of tasks from smallest to largest. The core number ranked first is selected as the valley queue core. Tasks in the priority task queue list are assigned to the cores in ascending order of node number. During the assignment, a mapping table between task number and core number is established simultaneously. The information of tasks and corresponding cores in the mapping table is written into the assignment list to generate the core execution task table.

[0087] The specific steps for generating the synchronization priority scheduling table are as follows:

[0088] Based on the core execution task table, the task sets are read sequentially according to the core number. The planned execution times of the tasks in each set are arranged in order. The time values ​​of adjacent tasks are subtracted and compared with the synchronization threshold. The task numbers with time differences less than the threshold are filtered out. The filtered numbers are grouped according to the core number and added to the group list to generate the synchronization task grouping table.

[0089] Based on the synchronous task grouping table, obtain the core execution queue position corresponding to the task number of each group, move the tasks of each group to the high priority task, maintain the original order of tasks within the group, establish a data record table corresponding to the adjusted queue and the synchronous task group number, and then merge the data record tables to generate a synchronous priority scheduling table.

[0090] Based on the core execution task table, the task set is read according to the core number. In each task set, the planned time of the tasks is arranged in order from early to late, and the arrangement is based on the numerical value of the time value from the minimum to the maximum. Then, the time values ​​of two adjacent tasks are directly subtracted to obtain the time difference. The synchronization threshold is set to 200 milliseconds. The time difference is compared with the synchronization threshold. The task numbers with time differences less than the threshold are filtered out. The filtered numbers are grouped according to their core numbers. The grouping results are recorded in the group list in the form of core number as key and task number array as value, generating a synchronization task group table.

[0091] Based on the synchronous task grouping table, the core execution queue position corresponding to the task number of each group is read. The relative positions of tasks in the same group in the queue are moved as a whole after the high-priority task, while keeping the original order of tasks in the group unchanged. When moving, the insertion point is determined according to the next index value of the end position of the high-priority task. Corresponding data records are established according to the task number and synchronization group number of the execution queue after insertion. The synchronization group number is written as the index and the adjusted queue sequence is written as the value in the record table. Finally, the record table is merged into a single set according to the core number order to generate the synchronous priority scheduling table.

[0092] Please see Figure 2 An oil and gas field data acquisition and processing system based on the HarmonyOS operating system is used to execute the aforementioned oil and gas field data acquisition and processing method based on the HarmonyOS operating system. The system includes:

[0093] Trend Analysis Module: Based on wellhead acquisition terminal delay records, bandwidth usage ratio, and data packet size, the data of each period is sequentially input into the Long Short-Term Memory network to calculate and generate trend values. The difference between the trend values ​​and the original data is compared to mark the time nodes that exceed the limit, and a delay warning node set is generated.

[0094] Preferred Node Module: Based on the delay warning node set, read the node delay value and bandwidth ratio, filter nodes with high delay and large bandwidth, sort them in descending order by data packet size, and record the communication link identifier to form a preferred link node table;

[0095] Link allocation module: Based on the priority link node table, the node delay and bandwidth are aggregated into the graph neural network to generate a merged value, which is compared with the threshold to filter high-load nodes, and qualified paths are selected according to delay and bandwidth to obtain the traffic distribution path allocation table;

[0096] Core scheduling module: Based on the traffic distribution path allocation table, read the task type, put the security monitoring task into the front end of the execution queue, count the number of core tasks, select the core with the lowest number of tasks for task allocation, and obtain the core execution task table;

[0097] Synchronization orchestration module: Based on the core execution task table, it compares the time difference between adjacent core tasks, classifies tasks with a time difference less than a threshold into a synchronization group, inserts them into high-priority tasks, and generates a synchronization priority scheduling table.

[0098] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for acquiring and processing oil and gas field data based on the HarmonyOS system, characterized in that, Includes the following steps: Step 1: By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, sort the parameters of each period by time and input them into the Long Short-Term Memory network to recursively generate trend values ​​and calculate adjacent differences. Compare these values ​​with the original sequence synchronously to determine if any exceedances are made and mark them, thereby generating a delay warning node set. Step 2: Based on the delay warning node set, read the delay value and bandwidth ratio and compare and filter them synchronously. Extract the nodes with large delay and high bandwidth and rearrange them in descending order according to the data packet size. Record the communication link identifier and obtain the priority link node table. Step 3: Based on the priority link node table, the link delay and bandwidth values ​​of the oil pipeline monitoring nodes are added together in the node feature aggregation unit of the graph neural network to form a weight value, high-load links are identified and diversion nodes are filtered to generate a diversion path allocation table. Step 4: Based on the aforementioned distribution path allocation table, read the task type of the compressor station monitoring unit according to the node number, extract the safety monitoring type node and insert it into the head of the execution queue in the order of task number, append the non-safety tasks to the end of the queue in the original number order, and generate a record table corresponding to the task number and the queue position. Generate a priority task queue list. Based on the priority task queue list, count the number of tasks in the current queue of each running core, mark the core with the lowest number of tasks as the lowest value queue core, allocate the tasks in the priority task queue list to the cores in the order of node number, and bind the allocated task number with the core number in the allocation list to generate the core execution task table. Step 5: Based on the core execution task table, read each core task set and compare adjacent execution times. Tasks with a time difference less than the synchronization threshold are assigned to the synchronization group and inserted as high-priority tasks. The records are integrated to form a synchronization priority scheduling table. The specific steps for generating the delay warning node set are as follows: By recording network latency, bandwidth usage ratio, and data packet size sequence at the wellhead acquisition terminal, the three types of parameters for each acquisition cycle are arranged in chronological order. The same parameter values ​​in adjacent cycles are subtracted one by one to obtain the difference. The difference is then paired with the time number and stored in an ordered list to generate a cycle difference dataset. Based on the aforementioned periodic difference dataset, the differences and corresponding three types of original parameter values ​​are read sequentially according to the time number. The differences are inserted into the end of the parameter group at the same time node to form a complete parameter unit. The parameter units formed at each time node are input into the input layer of the Long Short-Term Memory Network in sequence and the output is calculated recursively according to the time order. The output results are merged with the original parameter units at the same time node, and continuously sorted and stored in the new sequence to generate a composite trend sequence. Based on the composite trend sequence, the delay values ​​of each time node are read sequentially and compared with a set threshold. Time nodes with delay values ​​greater than the threshold are marked and written into a mark list. The marked nodes are combined with parameter units to generate a target set and a delay warning node set. The specific steps for generating the routing path allocation table are as follows: Based on the priority link node table, the delay and bandwidth values ​​of each link within the oil pipeline monitoring node are read. The two types of values ​​of the same node and the corresponding values ​​of adjacent nodes are added in the node feature aggregation unit of the graph neural network in chronological order to obtain a merged value. The merged value is written into the data column corresponding to the node and stored in a set according to the node time order to generate a set of link weight values. Based on the set of link weight values, the weight value of each node is read and compared with a set load threshold. The node numbers with weight values ​​greater than the threshold are written into the high load list to obtain the candidate link data corresponding to the nodes in the list. A table is formed according to the correspondence between the node and the candidate link information to generate a high load candidate link table. Based on the high-load candidate link table, the link with the valley delay is selected by comparing the delay values ​​of the candidate links of each node. When the delays are equal, the link with the bandwidth peak is selected by comparing the bandwidth values. The node number and the selected link are combined into a pair record and added to the result set in the order of the records to generate a traffic distribution path allocation table.

2. The oil and gas field data acquisition and processing method based on the HarmonyOS system according to claim 1, characterized in that, The execution process of the Long Short-Term Memory Network is as follows: after reading the difference and the corresponding three types of original parameter values ​​in sequence according to the time number, the difference is inserted into the end of the parameter group of the same time node to form a complete parameter unit. The values ​​of all parameter units are arranged in groups according to the time order and input into the input layer of the memory network. The input of each time node is processed in the hidden storage unit in sequence, and the state value of the previous time node is combined to perform recursive calculation. The calculation result of the current step is fused with the parameter unit of the corresponding time node item by item. The time order of the fusion result is maintained, and each time node is output and added to the newly created sequence to form a data set for generating a composite trend sequence.

3. The oil and gas field data acquisition and processing method based on the HarmonyOS system according to claim 1, characterized in that, The specific steps for generating the priority link node table are as follows: Based on the delay warning node set, the delay value and bandwidth usage ratio of each node are read in chronological order. The nodes with high delay value and large bandwidth ratio are selected by comparing the two. The node number is matched with the corresponding data packet size and stored in the list to generate a high-latency, high-bandwidth node set. Based on the high-latency, high-bandwidth node set, the nodes are sorted from largest to smallest according to the data packet size. The communication link identifiers of each node in the sorting result are extracted and inserted into the link list in sequence. The link list is stored in correspondence with the node number to generate a priority link node table.

4. The oil and gas field data acquisition and processing method based on the HarmonyOS system according to claim 1, characterized in that, The execution process of the graph neural network is as follows: based on the priority link node table, the link delay values ​​and bandwidth values ​​of the oil pipeline monitoring nodes and adjacent nodes are loaded as graph structure inputs according to the node number. The delay values ​​and bandwidth values ​​of each node are combined in chronological order to form a feature vector, which is then passed to the node feature aggregation unit of the network. The feature vectors of the directly connected nodes are paired one by one in the unit and summed. The summation result is written into the feature storage area of ​​the corresponding node. The node features are rearranged according to the time number and stored in the set as the input data source of the link weight value set.

5. The oil and gas field data acquisition and processing method based on the HarmonyOS system according to claim 1, characterized in that, The specific steps for generating the synchronization priority scheduling table are as follows: Based on the core execution task table, the task sets are read sequentially according to the core number. The planned execution times of the tasks in each set are arranged in order. The time values ​​of adjacent tasks are subtracted and compared with the synchronization threshold. The task numbers with time differences less than the threshold are filtered out. The filtered numbers are grouped according to the core number and added to the group list to generate a synchronization task grouping table. Based on the synchronous task grouping table, obtain the core execution queue position corresponding to the task number of each group, move the tasks of each group to the high priority task, maintain the original order of tasks within the group, establish a data record table corresponding to the adjusted queue and the synchronous task group number, and then merge the data record tables to generate a synchronous priority scheduling table.

6. An oil and gas field data acquisition and processing device based on the HarmonyOS system, comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the oil and gas field data acquisition and processing method based on the HarmonyOS system as described in any one of claims 1 to 5.

7. An oil and gas field data acquisition and processing system based on the HarmonyOS system, characterized in that, The oil and gas field data acquisition and processing system based on the HarmonyOS system is used to execute the oil and gas field data acquisition and processing method based on the HarmonyOS system according to any one of claims 1-5, the system comprising: Trend Analysis Module: Based on wellhead acquisition terminal delay records, bandwidth usage ratio, and data packet size, the data of each period is sequentially input into the Long Short-Term Memory network to calculate and generate trend values. The difference between the trend values ​​and the original data is compared to mark the time nodes that exceed the limit, and a delay warning node set is generated. Preferred Node Module: Based on the latency warning node set, read the node latency value and bandwidth ratio, filter nodes with high latency and large bandwidth, sort them in descending order by data packet size, and record the communication link identifier to form a preferred link node table; Link allocation module: Based on the priority link node table, the node delay and bandwidth are aggregated into the graph neural network to generate a merged value, and high-load nodes are filtered by comparing the threshold. Qualified paths are selected according to delay and bandwidth to obtain the traffic distribution path allocation table. Core scheduling module: Based on the traffic distribution path allocation table, read the task type, put the security monitoring task into the front end of the execution queue, count the number of each core task, select the core with the lowest number of tasks for task allocation, and obtain the core execution task table. Synchronization orchestration module: Based on the core execution task table, compare the adjacent time differences of each core task, classify tasks with time differences less than a threshold into a synchronization group, insert high-priority tasks, and generate a synchronization priority scheduling table.

Images