Oil and gas field management platform and method supporting edge calculation and intelligent early warning

Abstract

The invention provides an oil and gas field management platform and method supporting edge calculation and intelligent early warning, relates to the technical field of environment-friendly oil and gas field digital management, and aims to solve the problems of data lag, untimely early warning and low collaboration efficiency in traditional management. The platform adopts an edge end and cloud end distributed architecture: an edge end automatically collects pollution data of waste solids, waste gas and the like through an Internet of Things module, and after edge calculation processing, an AI diagnosis early warning edge sub-module performs real-time feature extraction and anomaly recognition; after the cloud receives the data, the AI cloud sub-module carries out deep analysis and model optimization, and the correction coefficient module dynamically improves the parameter accuracy through a multi-factor formula. The cloud four-dimensional cooperation module realizes task distribution and information synchronization of owners, construction parties, processing parties and supervision parties. Real-time monitoring, intelligent early warning and compliance management of oil and gas field pollution are realized, and environmental protection management efficiency and precision are improved through cooperation of the edge and the cloud.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital management of environmentally friendly oil and gas fields, in particular to an oil and gas field management platform and method supporting edge computing and intelligent early warning, for improving the intelligent, real-time and compliance level of environmental protection work. BACKGROUND

[0002] When monitoring and managing such pollutants, massive data needs to be processed. However, traditional environmental protection management systems generally use centralized data processing architecture, which has the following core problems: Data processing lag: Traditional systems rely on uploading all data collected on site to the central cloud for processing. Due to factors such as network bandwidth and transmission delay, real-time analysis of pollution data cannot be achieved, resulting in slow response to environmental protection abnormal events.

[0003] High degree of manual intervention: Traditional environmental protection management relies on manual inspection, data recording and abnormal judgment, which is not only inefficient, but also prone to data accuracy due to human operation errors, making it difficult to meet the needs of fine environmental protection management.

[0004] Lack of intelligent analysis capability: Existing systems lack effective artificial intelligence technology support, cannot automatically extract pollution data features and predict pollution trends, and cannot achieve intelligent diagnosis and early warning of environmental protection conditions.

[0005] Insufficient collaborative management: Oil and gas field environmental protection management involves multiple parties such as the owner, the contractor, the treatment party and the supervisor. In the traditional management mode, the information of each party is not shared, the responsibilities are not clear, the collaborative efficiency is low, and the management transparency is poor.

[0006] Weak compliance management: Traditional systems cannot automatically benchmark against laws, regulations, standards and enterprise systems, and the compliance judgment is highly subjective, making it impossible to achieve quantitative evaluation and continuous optimization of environmental protection management.

[0007] In the prior art, environmental management systems mostly focus on single-link monitoring or processing, and have not yet formed a complete solution of "data collection-local processing-intelligent analysis-early warning response-collaborative management-compliance evaluation". At the same time, there is a lack of edge computing and cloud collaboration architecture design, which cannot meet the needs of real-time response on site and centralized management. Therefore, there is an urgent need for a digital management platform integrating edge computing, artificial intelligence, big data, Internet of Things and other new generation information technologies to solve the above-mentioned pain points in oil and gas field environmental management, and to realize the intelligent, real-time and compliance of environmental management. SUMMARY

[0008] The present application aims to provide an oil and gas field management platform and method supporting edge computing and intelligent early warning to make up for the defects of the prior art. By integrating edge computing, artificial intelligence, big data and Internet of Things and other new generation information technologies, a digital solution capable of coping with the complex needs of oil and gas field environmental management is constructed.

[0009] To achieve the above-mentioned purpose, the present application first discloses an oil and gas field management platform supporting edge computing and intelligent early warning, which is characterized by: comprising a lightweight distributed architecture covering edge and cloud; the edge integrates an Internet of Things data acquisition module, an edge computing processing module and an AI intelligent diagnosis and early warning edge submodule; the Internet of Things data acquisition module is used to collect waste solid, waste gas, waste liquid and noise pollution data of the oil and gas field; the edge computing processing module is used to perform localized real-time processing on the collected pollution data; the AI intelligent diagnosis and early warning edge submodule performs real-time feature extraction, preliminary trend prediction and abnormality identification and early warning based on the localized processing data; the cloud integrates an AI intelligent diagnosis and early warning cloud submodule, a four-dimensional collaborative management module, a compliance management module and a correction coefficient adjustment module; the AI intelligent diagnosis and early warning cloud submodule is used to receive data and early warning information uploaded by the edge, perform deep feature analysis, long-term trend prediction and model training optimization; the four-dimensional collaborative management module is used to realize the collaborative work of the owner party, the construction party, the treatment party and the supervision party; the compliance management module is used to perform environmental protection compliance evaluation and quantitative scoring according to laws and regulations and enterprise systems; the correction coefficient adjustment module dynamically corrects the environmental protection score, the pollution trend prediction value, the image recognition accuracy, the feature extraction efficiency, the edge device computing power utilization rate and the task execution completion rate based on a multi-factor correction formula to improve accuracy; the multi-factor correction formula is , wherein is a corrected parameter value, is an original parameter value, , , is each correction factor.

[0010] Further, the correction coefficient adjustment module comprises an environmental protection score correction unit, which adopts an optimization formula , taking the environmental complexity correction coefficient as a correction factor, and correcting the environmental protection score based on the multi-factor correction formula; wherein, is the corrected environmental protection score, is the total number of environmental protection evaluation indexes, is the weight of the i-th evaluation index, is the original score of the i-th evaluation index, is used to adapt to the environmental differences of different regions, climates and pollution types.

[0011] Further, the correction coefficient adjustment module comprises a pollution trend prediction correction unit, which adopts an optimization formula , taking a model bias correction coefficient as a correction factor to correct the pollution trend prediction value based on the multi-factor correction formula; wherein, is the corrected pollution trend prediction value at time t, is the original pollution trend prediction value output by the LSTM model at time t, and satisfies , is a hidden layer weight matrix, is the hidden layer state at time t-1, is an input layer weight matrix, is the input data at time t, is a bias term; is used to compensate for the prediction error of the AI model under a specific pollution type.

[0012] Further, the correction coefficient adjustment module comprises an image recognition accuracy correction unit, which adopts an optimization formula , taking an image quality correction coefficient as a correction factor to correct the image recognition accuracy based on the multi-factor correction formula; wherein, is the corrected image recognition accuracy, is the original image recognition accuracy, and satisfies , is the number of true positive samples, is the number of true negative samples, is the number of false positive samples, is the number of false negative samples; is used to consider the influence of image-related factors on image recognition, which includes but is not limited to light, occlusion, and fog.

[0013] Further, the correction coefficient adjustment module comprises a feature extraction efficiency correction unit, which adopts an optimization formula , taking a data noise correction coefficient as a correction factor to correct the feature extraction efficiency based on the multi-factor correction formula; wherein, is the corrected feature extraction efficiency, is the original feature extraction efficiency, and satisfies , is the number of extracted features, is the feature extraction time consumption; is used to eliminate the interference of redundant or invalid features on extraction efficiency.

[0014] Further, the correction coefficient adjustment module comprises an edge device computing power utilization rate correction unit, the edge device computing power utilization rate correction unit adopts an optimization formula , which takes the device performance correction coefficient as a correction factor to correct the edge device computing power utilization rate based on the multi-factor correction formula; wherein, is the corrected edge device computing power utilization rate, is the original edge device computing power utilization rate, and satisfies , is the actual computing power usage time, is the total available computing power time; is used to compensate for performance differences caused by performance-related factors, including but not limited to device aging, temperature, and load.

[0015] Further, the correction coefficient adjustment module comprises a task execution completion rate correction unit, the task execution completion rate correction unit adopts an optimization formula , which takes the task attribute correction coefficient as a correction factor to correct the task execution completion rate based on the multi-factor correction formula; wherein, is the corrected task execution completion rate, is the original task execution completion rate, and satisfies , is the number of completed tasks, is the number of allocated tasks; is used to consider the impact of task-related factors, including but not limited to task priority, urgency, and execution difficulty.

[0016] Further, the edge computing processing module is deployed on a low-power edge device, including but not limited to an NVIDIA Jetson device, a Raspberry Pi AI module, and an edge intelligent gateway; the edge end deploys multiple edge computing nodes, each edge computing node connects multiple Internet of Things sensors, including gas concentration sensors, noise sensors, and image acquisition devices.

[0017] Further, the edge-end AI intelligent diagnosis and early warning edge sub-module deploys a light AI model, including MobileNet, SqueezeNet and LSTM, for real-time image recognition, short-term trend prediction and threshold type abnormality early warning on the pollution data output by the edge computing processing module; the cloud-end AI intelligent diagnosis and early warning cloud sub-module includes a model training unit for parameter optimization of the AI model by using the pollution data, early warning records and processing result data uploaded by the edge end, and the optimized model is issued to the AI intelligent diagnosis and early warning edge sub-module of the edge end through automatic updating, gray publishing and rollback mechanism.

[0018] Based on the foregoing description, the application further provides a method applied to the oil and gas field management platform supporting edge computing and intelligent early warning, characterized in that the method comprises the following steps: S1: The owner party sets an environmental protection target through the four-dimensional collaborative management module of the cloud end; S2: The construction party executes environmental protection measures based on the environmental protection target, and uploads execution data and on-site photos to the edge end; at the same time, the Internet of Things data acquisition module of the edge end automatically acquires waste solid, waste gas, waste liquid and noise pollution data of the oil and gas field; S3: The edge computing processing module of the edge end performs local processing on the automatically acquired pollution data and the execution data uploaded by the construction party, and the AI intelligent diagnosis and early warning edge sub-module analyzes and identifies environmental protection abnormalities in real time, and triggers local early warning if there is an abnormality; S4: The edge end uploads the abnormal information and the automatically acquired key pollution data and the data uploaded by the construction party to the cloud end, the AI intelligent diagnosis and early warning cloud sub-module of the cloud end carries out deep analysis, and the four-dimensional collaborative management module notifies the processing party of the abnormality processing task; S5: The processing party uploads the processing result and evidence to the cloud end after executing the abnormality processing; S6: The supervision party adjusts the related parameters based on a multi-factor correction formula after correction by the compliance management module and the correction coefficient adjustment module of the cloud end, and carries out system scoring and feedback to the owner party to adjust the subsequent environmental protection target.

[0019] The application effectively solves many problems existing in the traditional environmental protection management system by constructing an oil and gas field management platform with light distributed architecture and applying the corresponding method, and brings significant advantages: (1) In terms of data processing, the edge computing processing module performs localized real-time processing on data, reducing data transmission delay, achieving rapid analysis of pollution data, and greatly improving the response speed to environmental abnormal events, avoiding problems caused by data processing lag in traditional systems. Through real-time image recognition, short-term trend prediction and threshold type abnormal early warning of lightweight AI model at the edge, environmental abnormal conditions can be discovered in time, and time is gained for subsequent processing.

[0020] (2) In terms of intelligent analysis capability, the AI intelligent diagnosis and early warning cloud module in the cloud uses the data uploaded by the edge to perform deep feature analysis, long-term trend prediction and model training optimization. This enables the system to automatically extract pollution data features, accurately predict pollution trends, and realize intelligent diagnosis and early warning of environmental conditions. At the same time, the use of multi-factor correction formula dynamically corrects multiple parameters such as environmental score and pollution trend prediction value, improving the accuracy of data and the reliability of analysis results.

[0021] (3) In terms of collaborative management, the four-dimensional collaborative management module realizes the collaborative work of the owner, the construction party, the treatment party and the supervision party. The information of each party can be exchanged, the responsibilities are clear, and the collaborative efficiency and management transparency are improved. When environmental abnormalities occur, the treatment party can be quickly notified of the abnormal treatment task, and the treatment party can upload the results and evidence in time after executing the abnormal treatment, forming an efficient closed-loop management process.

[0022] (4) In terms of compliance management, the compliance management module conducts environmental compliance evaluation and quantitative scoring according to laws and regulations and enterprise systems. After correction of related parameters by the multi-factor correction formula, the system score is obtained, avoiding the subjective problem of compliance judgment in traditional systems, and realizing quantitative evaluation and continuous optimization of environmental management. The owner can adjust the subsequent environmental goals in time according to the scoring results fed back by the supervision party, so that the environmental management work is more scientific and reasonable.

[0023] In summary, the oil and gas field management platform and method supporting edge computing and intelligent early warning of the present application combines edge computing, artificial intelligence, big data, Internet of Things and other new generation information technologies, realizes a full-process solution of "data collection-local processing-intelligent analysis-early warning response-collaborative management-compliance evaluation", and meets the needs of real-time response and centralized management on site, effectively solves the pain points in oil and gas field environmental management, and improves the intelligent, real-time and compliance level of environmental management. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0025] Figure 1 Interaction flowchart of the lightweight distributed architecture in embodiment one; Figure 2 Data processing flowchart of the edge computing processing module in embodiment one; Figure 3 Model updating and iteration flowchart in embodiment one; Figure 4 Simplified flowchart of the management method in embodiment one. DETAILED DESCRIPTION

[0026] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0027] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0028] Figures 1 to 3The first embodiment of the present application is shown: an oil and gas field management platform supporting edge computing and intelligent early warning, comprising a lightweight distributed architecture covering edge and cloud; the edge end integrates an Internet of Things data acquisition module, an edge computing processing module and an AI intelligent diagnosis and early warning edge submodule; the Internet of Things data acquisition module is used to collect solid waste, waste gas, waste liquid and noise pollution data of the oil and gas field; the edge computing processing module is used to perform localized real-time processing on the collected pollution data; the AI intelligent diagnosis and early warning edge submodule performs real-time feature extraction, preliminary trend prediction and abnormality identification and early warning based on the localized processing data; the cloud end integrates an AI intelligent diagnosis and early warning cloud end submodule, a four-dimensional collaborative management module, a compliance management module and a correction coefficient adjustment module; the AI intelligent diagnosis and early warning cloud end submodule is used to receive data and early warning information uploaded by the edge end, perform deep feature analysis, long-term trend prediction and model training optimization; the four-dimensional collaborative management module is used to realize the collaborative work of the owner party, the construction party, the treatment party and the supervision party; the compliance management module is used to perform environmental protection compliance evaluation and quantitative scoring according to laws and regulations and enterprise systems; the correction coefficient adjustment module dynamically corrects the environmental protection score, the pollution trend prediction value, the image recognition accuracy, the feature extraction efficiency, the edge device computing power utilization rate and the task execution completion rate based on a multi-factor correction formula to improve accuracy; the multi-factor correction formula is wherein is a corrected parameter value, is an original parameter value, , , is each correction factor.

[0029] In this embodiment, the edge end and the various modules of the cloud end cooperate with each other to build an efficient and intelligent oil and gas field management system. The Internet of Things data acquisition module of the edge end relies on the Internet of Things sensors deployed on the oil and gas field site, such as gas concentration sensors (used to detect the concentration of waste gas such as hydrogen sulfide and methane), noise sensors (used to monitor equipment / construction noise values), image acquisition equipment (used to take pictures of waste solid accumulation, waste liquid discharge, etc. on-site images), real-time collection of four types of pollution data, and transmission of data to the edge computing processing module (local preliminary processing) or the cloud end (centralized in-depth analysis). The edge computing processing module performs local real-time processing on these data, greatly reducing the burden on the cloud end and improving the response speed of the system. The AI intelligent diagnosis and early warning edge sub-module can quickly perform real-time feature extraction, preliminary trend prediction and abnormality identification and early warning based on the local processing data, so that the on-site personnel can timely detect potential problems. The various modules of the cloud end are managed and coordinated from a higher level. After receiving the data and early warning information uploaded by the edge end, the AI intelligent diagnosis and early warning cloud end sub-module uses its powerful computing power to perform deep feature analysis, long-term trend prediction and model training optimization, providing a scientific basis for long-term environmental protection management of oil and gas fields. The four-dimensional collaborative management module enables the owner, the construction party, the treatment party and the supervision party to work efficiently and ensure the smooth progress of environmental protection work. The compliance management module evaluates and quantifies the environmental protection compliance according to laws, regulations and enterprise systems, and points out a clear standard and direction for the environmental protection work of the oil and gas field. The correction coefficient adjustment module dynamically corrects various key parameters through a multi-factor correction formula, further enhancing the accuracy and reliability of the system.

[0030] For example, when the Internet of Things data acquisition module collects abnormal waste gas pollution data in a certain area, the edge computing processing module will immediately perform local real-time processing on the data, the AI intelligent diagnosis and early warning edge sub-module will quickly perform real-time feature extraction and preliminary trend prediction, and issue an abnormality identification and early warning. After the edge end uploads these data and early warning information to the cloud end, the AI intelligent diagnosis and early warning cloud end sub-module will perform deep feature analysis and long-term trend prediction to determine whether the abnormality will continue to develop and its impact on the overall environmental protection of the oil and gas field. The four-dimensional collaborative management module will promptly notify the relevant treatment party and supervision party, and the treatment party will execute abnormal treatment according to the actual situation and upload the treatment results and evidence to the cloud end after the treatment is completed. The compliance management module will evaluate and quantize the treatment, and the correction coefficient adjustment module will dynamically correct the relevant parameters according to the actual situation to ensure the accuracy and reliability of the entire system, thereby better protecting the environmental protection work of the oil and gas field.

[0031] In specific implementation, the correction coefficient adjustment module includes an environmental protection score correction unit, which adopts an optimization formula , the formula is modified by the environmental complexity correction coefficient As a correction factor, the environmental protection score is corrected based on the multi-factor correction formula; wherein, to correct the environmental protection score, is the total number of environmental protection evaluation indexes, is the weight of the i-th evaluation index, is the original score of the i-th evaluation index, to adapt to the environmental differences of different regions, climates and pollution types.

[0032] environmental complexity correction coefficient will be dynamically adjusted according to different environmental scenarios. For example, in areas with dry climate and large wind sand, the value of the atmospheric pollution related index will be increased accordingly to consider the influence of wind sand on air quality monitoring data; while in rainy and humid areas, the value of the water pollution related index may be adjusted according to the dilution or scouring of pollutants in water by rainfall. At the same time, for different pollution types, such as industrial pollution and domestic pollution, the value will also be different. When the pollution type is more complex, the value will increase to ensure that the environmental protection score can more accurately reflect the actual environmental conditions. In addition, the environmental protection score correction unit will collect and analyze various environmental data in real time, including meteorological data, geographic information data, etc., to dynamically update the value. And it will interact with the compliance management module to re-standardize the corrected environmental protection score with the regulatory indicators and enterprise system indicators. If the corrected environmental protection score is still abnormal, the system will timely issue a warning information to remind the relevant personnel to take corresponding measures. At the same time, the environmental protection score correction unit will also record and analyze the correction process, summarize the adjustment rules of in different environmental scenarios, so as to be more rapid and accurate in the future environmental protection management.

[0033] In this embodiment, the correction coefficient adjustment module includes a pollution trend prediction correction unit, which uses an optimization formula , the formula is modified by the model bias correction coefficient As a correction factor, the pollution trend prediction value is corrected based on the multi-factor correction formula; wherein, is the corrected pollution trend prediction value at time t, is the original pollution trend prediction value output based on the LSTM model at time t, and satisfies , is the hidden layer weight matrix, is the hidden layer state at time t-1, is the input layer weight matrix, is the input data at time t, is the bias term; to compensate for the prediction error of the AI model under a specific pollution type.

[0034] To determine the model bias correction coefficient , the historical data backtracking analysis method is adopted in the embodiment. Specifically, the actual pollution data of the specific pollution type in the past period of time and the original pollution trend prediction value output by the LSTM model at the same period are collected. The actual pollution data and the original pollution trend prediction value are compared, and the prediction error at each time point is calculated. By statistically analyzing these errors, such as calculating the average error, mean square error, etc., a suitable value is determined, so that the corrected pollution trend prediction value is closer to the actual situation. In addition, in order to ensure the accuracy and timeliness of the correction, the pollution trend prediction correction unit also sets a dynamic adjustment mechanism. As time goes on and new data is continuously input, the system will periodically reevaluate and adjust When it is found that the deviation between the actual pollution situation and the corrected prediction value exceeds the preset threshold, the system will automatically trigger the program to recalculate to ensure that the accuracy of the pollution trend prediction is always at a high level. In actual application, the pollution trend prediction correction unit will receive the results of execution data and on-site photo analysis in real time, and dynamically correct the pollution trend combined with these information. For example, when the on-site photos show that a new pollution source appears in a certain area or the pollution degree suddenly increases, the system will quickly adjust the pollution trend prediction value to provide more reliable basis for subsequent environmental protection decision. At the same time, the corrected pollution trend prediction value is fed back to other modules of the digital management platform, so that the whole platform can work cooperatively and more efficiently handle environmental protection abnormal situations.

[0035] Specifically, the correction coefficient adjustment module includes an image recognition accuracy correction unit, which adopts an optimization formula , taking the image quality correction coefficient as the correction factor, and correcting the image recognition accuracy based on the multi-factor correction formula; wherein, is the corrected image recognition accuracy, is the original image recognition accuracy, and satisfies , is the number of true positive samples, is the number of true negative samples, is the number of false positive samples, is the number of false negative samples; For considering the influence of image-related factors on image recognition, the image-related factors include but are not limited to light, occlusion and fog.

[0036] In order to ensure the accuracy and effectiveness of the image quality correction coefficient , the system will perform multi-dimensional feature extraction and analysis on the image. For the light factor, the system will analyze the brightness, contrast and color distribution of the image to determine whether the light is uniform, too strong or too weak. If the light is not uniform, the system will adjust the local brightness and contrast of the image to improve the clarity of the image. At the same time, according to the influence degree of light, the value of is dynamically adjusted, so that the corrected image recognition accuracy is closer to the true situation. For the occlusion factor, the system will use target detection and segmentation algorithm to identify the occluded area in the image. By analyzing the size, shape and position of the occluded area, the influence degree of the occluded area on image recognition is evaluated. If the occluded area is small and does not affect the recognition of key features, the adjustment range of is relatively small; if the occluded area is large and seriously affects the recognition of the target, the system will appropriately increase the adjustment range of to reduce the influence of occlusion on image recognition accuracy. For the fog factor, the system will use image de-fogging algorithms such as physical model-based de-fogging algorithm or deep learning-based de-fogging algorithm to de-fog the image with fog. During the de-fogging process, the system will adjust the value of according to the concentration and distribution of the fog. After de-fogging, the image is recognized and analyzed again to verify whether the corrected image recognition accuracy has been effectively improved. In addition, the image recognition accuracy correction unit also establishes a historical image database to record the image recognition results and correction coefficients under different image-related factors. Through analysis and mining of historical data, the calculation method and adjustment strategy of are continuously optimized. When encountering new images, the system can refer to historical data to quickly and accurately determine the value of , further improving the efficiency and precision of image recognition accuracy correction. At the same time, the unit also interacts with other modules of the digital management platform in real time, feeds back the corrected image recognition accuracy to related modules such as pollution trend prediction correction unit, and provides more reliable support for environmental abnormality identification and processing of the entire platform.

[0037] In actual application, the correction coefficient adjustment module includes a feature extraction efficiency correction unit, which uses an optimization formula to correct the feature extraction efficiency based on the multi-factor correction formula, with the data noise correction coefficient as the correction factor; wherein, is the corrected feature extraction efficiency, For the original feature extraction efficiency, and meet , For the number of extracted features, For feature extraction time-consuming; For removing redundant or invalid features to interfere with extraction efficiency.

[0038] The feature extraction efficiency correction unit will monitor the noise in the data in real time in actual work. When the noise level in the data is low, The value will be close to 1, and the corrected feature extraction efficiency Close to the original feature extraction efficiency , indicating that the data quality is good, and the feature extraction process is less disturbed by noise. When the noise level in the data is high, The value will be less than 1, which will make the corrected feature extraction efficiency Lower than the original feature extraction efficiency , indicating that redundant or invalid features have a greater interference on the extraction efficiency. Through this correction, the actual efficiency of feature extraction can be more truly reflected. At the same time, the feature extraction efficiency correction unit also has the ability of dynamic adjustment. It will continuously collect and analyze new feature extraction data, and adjust the value of In real time according to the change of data noise. For example, if it is found that the proportion of redundant or invalid features in the extracted features gradually increases over a period of time, the unit will accordingly reduce the value of To more effectively eliminate these interference factors and improve the accuracy and efficiency of feature extraction. In addition, the feature extraction efficiency correction unit will also work with other modules of the digital management platform. It will timely deliver the corrected feature extraction efficiency information to the image recognition accuracy correction unit and other related modules. The image recognition accuracy correction unit can further optimize the correction strategy for image recognition results according to this information, such as considering the influence of feature extraction efficiency when determining , so as to more comprehensively improve the identification and processing capacity of the entire platform for environmental protection abnormalities. In the face of complex environmental protection data and on-site photos, through the accurate correction and coordination of the feature extraction efficiency correction unit, the digital management platform can more efficiently and accurately automatically identify environmental protection abnormalities, providing strong support for the environmental protection management of oil and gas fields.

[0039] In specific application scenarios, the correction coefficient adjustment module includes an edge device computing power utilization rate correction unit, which uses an optimization formula , which takes the device performance correction coefficient As a correction factor, the edge device computing power utilization rate is corrected based on the multi-factor correction formula; wherein, For correcting the edge device computing power utilization rate, For the original edge device computing power utilization rate, and meet , For the actual use time of computing power, For the total available time of computing power; For compensating for performance differences caused by performance-related factors, including but not limited to device aging, temperature, and load.

[0040] In actual operation, the performance of edge devices will fluctuate with time and environmental changes. Device aging will cause hardware performance to gradually decline, so that the same task will consume more computing power. At this time, the device performance correction coefficient will increase accordingly to accurately reflect the actual computing power utilization rate of the edge device. Temperature is also an important factor affecting device performance. Both excessively high and excessively low temperatures can affect device operating efficiency. When the temperature exceeds the optimal working range of the device, adjustment will be made according to the degree of temperature impact on performance. Load conditions cannot be ignored either. When an edge device is processing multiple complex tasks simultaneously, the load increases, which may cause performance bottlenecks, The edge device computing power utilization rate will be corrected according to the size and complexity of the load. The introduction of this edge device computing power utilization rate correction unit can improve the accuracy of the digital management platform in assessing the performance of edge devices. Through real-time correction of the edge device computing power utilization rate, the platform can more accurately grasp the running state of the device and promptly identify potential performance problems. For example, when the corrected edge device computing power utilization rate is consistently high, it may mean that the device is about to reach its performance limit and needs to be maintained or upgraded in a timely manner. This helps to avoid problems such as delayed environmental abnormality identification due to insufficient device performance, thereby ensuring the stable operation of the entire oil and gas field management platform and the effective realization of environmental protection goals. At the same time, based on accurate correction of computing power utilization rate, the platform can also allocate resources more reasonably, improve resource utilization efficiency, and reduce operating costs.

[0041] In specific implementation, the correction coefficient adjustment module includes a task execution completion rate correction unit, which uses an optimization formula , which uses a task attribute correction coefficient as a correction factor to correct the task execution completion rate based on the multi-factor correction formula; wherein, is the corrected task execution completion rate, is the original task execution completion rate, and meets , is the number of completed tasks, is the number of allocated tasks; for considering the impact of task-related factors, including but not limited to task priority, urgency, and execution difficulty.

[0042] Task priority determines the importance of a task within the overall task system. For high-priority tasks, A higher value will be assigned to reflect its importance in assessing task completion rate. Urgency, on the other hand, focuses on the time limit for task completion; urgent tasks... The value will also increase accordingly, prompting the platform to pay more attention to the execution of such tasks. Execution difficulty reflects the resources, skills, and time required to complete the task; the more difficult the task, the higher the value. A higher value allows for a more accurate measurement of the actual performance of edge devices when faced with tasks of varying difficulty. This correction method provides a more objective reflection of the task execution status of edge devices. For example, if an edge device has been assigned a large number of tasks, but these include many low-priority, low-difficulty, and non-urgent tasks, even if the original task completion rate appears high, the corrected task completion rate may decrease after correction by the task completion rate correction unit. This more accurately reflects the device's ability to handle critical tasks. Simultaneously, this correction mechanism provides a more scientific basis for resource allocation and scheduling on the digital management platform. When the corrected task completion rate is low, the platform can analyze specific task-related factors to determine whether the low rate is due to the task's inherent difficulty or insufficient edge device resources. This allows for reasonable adjustments to resource allocation, such as increasing the computing power of edge devices or reallocating tasks to other more suitable devices, thereby improving overall task execution efficiency, better enabling automatic identification of environmental anomalies, and ensuring the achievement of environmental goals for the oil and gas field management platform.

[0043] In this embodiment, the edge computing processing module is deployed on a low-power edge device, which includes, but is not limited to, NVIDIA Jetson devices, Raspberry Pi AI modules, and edge smart gateways; multiple edge computing nodes are deployed at the edge, and each edge computing node is connected to multiple IoT sensors, including gas concentration sensors, noise sensors, and image acquisition devices.

[0044] These IoT sensors continuously collect environmental data. Gas concentration sensors meticulously monitor the real-time concentration of various harmful gases in the oil and gas field area, and once the concentration of a certain gas exceeds the normal range, the data will be quickly transmitted to the edge computing node. Noise sensors accurately capture the noise generated during oil and gas field operations to determine whether there are abnormal conditions such as noise pollution. Image acquisition devices play an important role, as they can capture real-time on-site images of the oil and gas field, capturing visual environmental anomalies such as equipment leaks, illegal emissions, etc. After receiving data from IoT sensors, the edge computing node performs preliminary processing and analysis. It cleans and filters the data, removing interference information and invalid data, and extracts valuable features. Then, the processed data is sent to the edge computing processing module. Based on powerful computing capabilities and optimized models, the edge computing processing module conducts in-depth analysis of the data. It can quickly identify hidden environmental anomalies in the data, such as determining whether there is a leak by analyzing gas concentration data, or identifying whether there are illegal waste dumps based on image data. At the same time, the edge computing processing module also feeds back the analysis results to the digital management platform. After receiving the feedback, the digital management platform conducts comprehensive evaluation and decision-making. If it confirms the existence of environmental anomalies, the platform will quickly issue a warning message to notify relevant personnel to take timely measures to handle it. Moreover, the edge computing processing module also regularly checks and optimizes its own operation to ensure its continuous and stable operation, efficiently completing the task of environmental anomaly identification.

[0045] Specifically, the AI intelligent diagnosis and early warning edge module of the edge end deploys a lightweight AI model, including MobileNet, SqueezeNet, and LSTM, for real-time image recognition, short-term trend prediction, and threshold-type anomaly warning of pollution data output by the edge computing processing module; the AI intelligent diagnosis and early warning cloud module of the cloud end includes a model training unit for parameter optimization of AI models using pollution data, early warning records, and processing result data uploaded by the edge end, and automatically updating, gray publishing, and rollback mechanisms to distribute the optimized models to the AI intelligent diagnosis and early warning edge module of the edge end.

[0046] This edge-cloud collaboration mode greatly improves the efficiency and accuracy of environmental abnormality identification and processing. At the edge, lightweight AI models can perform preliminary and rapid processing of data locally, reducing data transmission delay and bandwidth requirements. For example, MobileNet can efficiently identify real-time images and quickly determine whether there are environmental abnormalities such as illegal emissions, garbage accumulation, etc. SqueezeNet can extract and analyze pollution data features on resource-limited edge devices with relatively low computational cost. LSTM is good at predicting short-term trends of pollution data, discovering potential environmental problems in advance, and implementing threshold-based anomaly warning. The model training unit of the AI intelligent diagnosis and warning cloud module in the cloud continuously learns various data uploaded from the edge, continuously optimizing the parameters of the AI model. Training with actual data generated makes the model better adapt to different environments and actual situations, improving its identification and warning accuracy. Automatic updating, gray release, and rollback mechanisms ensure that the optimized model can be safely and stably deployed to the edge. Automatic updating ensures that the edge model is always the latest and optimal; gray release allows testing on some edge devices first to observe the model's performance, avoiding potential risks caused by model updates; rollback mechanism can restore the model to the previous stable version in case of problems, ensuring the normal operation of the entire system. In addition, this architecture has good scalability and flexibility.

[0047] In specific applications, the four-dimensional collaborative management module of the cloud receives automatically collected pollution data uploaded from the edge, execution data and abnormal warning information uploaded from the construction party, automatically assigns tasks to the construction party and the treatment party, and synchronizes the treatment results and evidence uploaded from the treatment party to the supervision party, supporting the supervision party to carry out compliance evaluation. The edge and the cloud transmit data through MQTT or CoAP communication protocol, and the edge only uploads key pollution data, abnormal warning information and model optimization data to the cloud, reducing data transmission volume.

[0048] Please refer to Figure 4 Based on the foregoing description, embodiment one also discloses a method applied to the oil and gas field management platform supporting edge computing and intelligent warning, comprising the following steps: S1: The owner party sets environmental protection targets through the four-dimensional collaborative management module of the cloud; S2: The construction party executes environmental protection measures based on the environmental protection targets, and uploads execution data and on-site photos to the edge; at the same time, the Internet of Things data acquisition module of the edge automatically collects solid waste, waste gas, waste liquid and noise pollution data of the oil and gas field; S3: The edge computing processing module at the edge end performs localized processing on the automatically collected pollution data and the execution data uploaded by the construction party, and then the AI intelligent diagnosis and early warning edge sub-module analyzes and identifies environmental protection abnormalities in real time. If there is an abnormality, a local warning is triggered; S4: The edge end uploads the abnormal information, automatically collected key pollution data, and construction party uploaded data to the cloud end, and the AI intelligent diagnosis and early warning cloud end sub-module of the cloud end conducts in-depth analysis, while the four-dimensional collaborative management module notifies the processing party of the abnormal handling task; S5: After the processing party performs abnormal handling, the processing result and evidence are uploaded to the cloud end; S6: The supervision party corrects the related parameters based on the multi-factor correction formula through the compliance management module and the correction coefficient adjustment module of the cloud end, and then performs system scoring and feedback to the owner party to adjust the subsequent environmental protection target.

[0049] In the entire process, each step is closely connected to form a complete environmental protection management closed loop. For the owner party, the environmental protection target set through the four-dimensional collaborative management module of the cloud end is the starting point of the entire process, which provides a clear direction for the subsequent actions of the construction party. The construction party strictly executes environmental protection measures according to the environmental protection target, and timely uploads the execution data and on-site photos to ensure that the environmental protection situation during the construction process can be timely feedback. The Internet of Things data collection module at the edge end automatically collects various pollution data of the oil and gas field, providing comprehensive and accurate information for subsequent analysis. The edge computing processing module at the edge end performs localized processing on the data, which can timely discover environmental protection abnormalities and trigger early warning at the local end, greatly improving the timeliness of abnormal handling. When the edge end uploads the abnormal information to the cloud end, the AI intelligent diagnosis and early warning cloud end sub-module of the cloud end conducts in-depth analysis to further confirm the abnormal situation, and the four-dimensional collaborative management module notifies the processing party to handle. The processing party uploads the processing result and evidence after performing abnormal handling, which provides an important basis for the scoring of the supervision party. The supervision party corrects the related parameters based on the multi-factor correction formula through the compliance management module and the correction coefficient adjustment module of the cloud end, and then performs system scoring. This scoring result not only reflects the effectiveness of the current environmental protection management, but also provides a strong reference for the owner party to adjust the subsequent environmental protection target. By continuously adjusting the environmental protection target, the environmental protection management level of the entire oil and gas field can be continuously improved, realizing more scientific and efficient environmental protection management.

[0050] In summary, the oil and gas field management platform and method supporting edge computing and intelligent early warning can fully utilize the advantages of edge computing, reduce data transmission volume, improve data processing efficiency, and reduce system response time. The application of the AI intelligent diagnosis and early warning module can timely discover environmental abnormalities, take measures in advance, and avoid further deterioration of environmental pollution. Moreover, the collaborative management mechanism of the entire system enables the owner, the construction party, the treatment party, and the supervision party to closely collaborate and jointly promote the environmental protection work of the oil and gas field.

[0051] The above only discloses a preferred embodiment of the present application, and of course cannot limit the scope of the rights of the present application. Those skilled in the art can understand that all or part of the processes of the above-mentioned embodiments are implemented, and equivalent changes made according to the claims of the present application still belong to the scope covered by the present application.

Claims

1. An oil and gas field management platform supporting edge computing and intelligent early warning, characterized in that: Lightweight distributed architecture covering edge end and cloud end; the edge end integrates Internet of Things data acquisition module, edge computing processing module and AI intelligent diagnosis and early warning edge submodule; the Internet of Things data acquisition module is used for collecting waste solid, waste gas, waste liquid and noise pollution data of oil and gas field; the edge computing processing module is used for local real-time processing of collected pollution data; the AI intelligent diagnosis and early warning edge submodule carries out real-time feature extraction, preliminary trend prediction and abnormal identification and early warning based on local processing data; the cloud end integrates AI intelligent diagnosis and early warning cloud end submodule, four-dimensional collaborative management module, compliance management module and correction coefficient adjustment module; the AI intelligent diagnosis and early warning cloud end submodule is used for receiving data and early warning information uploaded by the edge end, carrying out deep feature analysis, long-term trend prediction and model training optimization; the four-dimensional collaborative management module is used for realizing the collaborative work of the owner party, the construction party, the treatment party and the supervision party; the compliance management module is used for environmental protection compliance evaluation and quantitative scoring according to laws and regulations and enterprise system; the correction coefficient adjustment module dynamically corrects environmental protection score, pollution trend prediction value, image recognition accuracy, feature extraction efficiency, edge device computing power utilization rate and task execution completion rate based on multi-factor correction formula to improve accuracy; the multi-factor correction formula is wherein is the corrected parameter value, is the original parameter value, , , is each correction factor. 2.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, wherein: The correction coefficient adjustment module comprises an environmental protection score correction unit The formula takes environmental complexity correction coefficient as a correction factor, and corrects the environmental protection score based on the multi-factor correction formula; wherein In order to correct the environmental protection score, is the total number of environmental protection evaluation indexes, is the weight of the i th evaluation index, is the original score of the i th evaluation index, for adapting to the environmental differences of different regions, climates and pollution types. 3.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, characterized in that: The correction coefficient adjustment module comprises a pollution trend prediction correction unit The formula uses a model bias correction coefficient As a correction factor, the pollution trend prediction value is corrected based on the multi-factor correction formula; wherein is the corrected pollution trend prediction value at time t, is the original pollution trend prediction value output by the LSTM model at time t, and satisfies , is the hidden layer weight matrix, is the hidden layer state at time t-1, is the input layer weight matrix, is the input data at time t, is the bias term; for compensating the prediction error of the AI model under a specific pollution type. 4.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, characterized in that: The correction coefficient adjustment module comprises an image recognition accuracy correction unit, the image recognition accuracy correction unit adopts an optimization formula , the formula takes an image quality correction coefficient as a correction factor, and corrects the image recognition accuracy based on the multi-factor correction formula; wherein, is the corrected image recognition accuracy, is the original image recognition accuracy, and satisfies , is the number of true positive samples, is the number of true negative samples, is the number of false positive samples, is the number of false negative samples; The image recognition accuracy is corrected by considering the influence of image-related factors on image recognition, the image-related factors including but not limited to light, shielding and fog. 5.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, characterized in that: The correction coefficient adjustment module comprises a feature extraction efficiency correction unit, the feature extraction efficiency correction unit adopts an optimization formula , the formula takes data noise correction coefficient as a correction factor, and corrects the feature extraction efficiency based on the multi-factor correction formula; wherein, , in order to correct the feature extraction efficiency, , is the original feature extraction efficiency, and satisfies , , is the number of extracted features, , is the feature extraction time consumption; , for eliminating the interference of redundant or invalid features on the extraction efficiency. 6.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, characterized in that: The correction coefficient adjustment module comprises an edge device computing power utilization rate correction unit The formula takes device performance correction coefficient as a correction factor to correct the edge device computing power utilization rate based on the multi-factor correction formula; wherein To correct the edge device computing power utilization rate, is the original edge device computing power utilization rate, and satisfies , is the actual computing power usage time, is the total available computing power time; to compensate for the performance difference caused by performance-related factors, including but not limited to device aging, temperature and load. 7.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 1, characterized in that: The correction coefficient adjustment module comprises a task execution completion rate correction unit The formula takes the task attribute correction coefficient as a correction factor, corrects the task execution completion rate based on the multi-factor correction formula; wherein In order to correct the task execution completion rate, is the original task execution completion rate, and satisfies , is the number of completed tasks, is the number of allocated tasks; for considering the influence of task-related factors, which include but are not limited to task priority, urgency and execution difficulty. 8.The oil and gas field management platform supporting edge computing and intelligent early warning according to any one of claims 1-7, characterized in that: The edge computing processing module is deployed on a low-power edge device, which includes but is not limited to an NVIDIA Jetson device, a Raspberry AI module and an edge intelligent gateway; a plurality of edge computing nodes are deployed on the edge end, and each edge computing node is connected with a plurality of Internet of Things sensors, including a gas concentration sensor, a noise sensor and an image acquisition device. 9.The oil and gas field management platform supporting edge computing and intelligent early warning according to claim 8, characterized in that: The AI intelligent diagnosis and early warning edge submodule of the edge end deploys a lightweight AI model, including MobileNet, SqueezeNet and LSTM, for real-time image recognition, short-term trend prediction and threshold abnormality early warning on the pollution data output by the edge computing processing module; the AI intelligent diagnosis and early warning cloud submodule of the cloud end includes a model training unit, which is used to optimize the parameters of the AI model by using the pollution data, early warning records and processing result data uploaded by the edge end, and to issue the optimized model to the AI intelligent diagnosis and early warning edge submodule of the edge end through automatic updating, gray publishing and rollback mechanism.

10. The method applied to the oil and gas field management platform supporting edge computing and intelligent early warning according to any one of claims 1-9, characterized in that, The method comprises the following steps: S1: the owner party sets an environmental protection target through the four-dimensional collaborative management module of the cloud end; S2: the construction party executes environmental protection measures based on the environmental protection target, and uploads execution data and on-site photos to the edge end; at the same time, the Internet of Things data acquisition module of the edge end automatically collects waste solid, waste gas, waste liquid and noise pollution data of the oil and gas field; S3: the edge computing processing module of the edge end performs localized processing on the automatically collected pollution data and the execution data uploaded by the construction party, and then the AI intelligent diagnosis and early warning edge submodule analyzes and identifies environmental protection abnormalities in real time, and triggers local early warning if there is an abnormality; S4: the edge end uploads the abnormal information and the automatically collected key pollution data and the data uploaded by the construction party to the cloud end, the AI intelligent diagnosis and early warning cloud submodule of the cloud end carries out deep analysis, and the four-dimensional collaborative management module of the cloud end notifies the processing party of the abnormal treatment task; S5: the processing party executes abnormal treatment, and uploads the treatment result and evidence to the cloud end; S6: the supervision party adjusts the related parameters based on the multi-factor correction formula through the compliance management module and the correction coefficient adjustment module of the cloud end, and then carries out system scoring, and feeds back to the owner party to adjust the subsequent environmental protection target.

Images