Long-distance pipeline risk event distinguishing monitoring system and method based on artificial intelligence

Abstract

The invention relates to the technical field of pipeline safety monitoring, in particular to a long-distance pipeline risk event distinguishing monitoring system and method based on artificial intelligence. The artificial intelligence-based long-distance pipeline risk event distinguishing monitoring method comprises the following steps of S10, acquiring humidity time sequence data and anode output current time sequence data of soil in an area where a target pipe section is located, and respectively drawing a humidity time curve and a current time curve; acquiring a vibration signal acquired by the optical fiber sensing network; and S20, periodically fluctuating according to the current-time curve. According to the method, the relation between the regional self-adaptive humidity saturation interval and the corrosion rate is established, the huge cost for customizing a vibration analysis model for each region is avoided, the event recognition accuracy is improved under high-risk scenes such as coating damage by utilizing deep coupling of the corrosion state and vibration, and the event recognition efficiency is improved. And passive alarm is converted into active and accurate early warning based on humidity prediction.

Description

Technical Field

[0001] This invention relates to the field of pipeline safety monitoring technology, and in particular to an artificial intelligence-based risk event differentiation monitoring system and method for long-distance pipelines. Background Technology

[0002] Long-distance pipelines are a crucial component of the national energy transportation system, responsible for the stable transport of oil and gas resources from production areas to consumption regions. Their operational safety directly impacts the continuity of energy supply and socio-economic development. Because the soil environment where long-distance pipelines are buried is consistently characterized by high humidity, they are constantly exposed to risks such as corrosion, geological disasters, and third-party construction. Leaks or ruptures can cause severe environmental pollution and economic losses. Therefore, comprehensive, real-time, and accurate risk monitoring and early warning systems for long-distance pipelines are critical to ensuring their safe operation.

[0003] Currently, corrosion risk monitoring of long-distance pipelines typically involves acquiring time-series data such as corrosion rate, cathodic protection potential, and soil resistivity through online monitoring, and then using data analysis models to predict future corrosion trends. For example, the method disclosed in Chinese patent application CN121073443A integrates multi-source monitoring data and uses a spatiotemporal graph attention network to extract correlation features across spatial nodes and time windows, generating future corrosion event probabilities, corrosion rate estimates, and uncertainties, thus achieving dynamic and forward-looking corrosion risk prediction. For the identification of third-party construction, vibration signals are typically collected in real time through fiber optic sensor networks deployed along the pipeline. Vibration intensity and frequency are used to distinguish between construction activities and heavy object compaction scenarios, enabling timely intervention and risk reduction.

[0004] However, the aforementioned methods are largely based on the assumption that the geological conditions of the monitoring area are relatively uniform. When there are significant differences in soil water-holding capacity and moisture content, the impact of soils with different moisture contents on the corrosion of the pipeline's external metals varies considerably, especially during the alternation of rainy seasons. This leads to strong spatial heterogeneity in the soil environment around the pipeline, with areas of high moisture content being more prone to electrochemical corrosion. Since long-distance pipelines often traverse multiple geological regions, and the rainfall conditions and soil water-holding capacity vary significantly in different regions, if the aforementioned spatiotemporal attention network is used for corrosion rate prediction, it is often necessary to construct multiple prediction models for different regions to adapt to their respective geological conditions. This not only results in a significant waste of resources but also ignores the spatial correlation between regions (i.e., the mutual influence between the prediction models constructed by the networks), leading to inaccurate corrosion risk prediction results. Furthermore, in cases of severe corrosion, the differences in vibration intensity and frequency caused by construction or heavy object rolling are small, making it difficult to effectively distinguish between them based on the vibration intensity and frequency of the vibration signals, further reducing the accuracy of risk monitoring. Summary of the Invention

[0005] This invention provides an artificial intelligence-based risk event differentiation and monitoring system and method for long-distance pipelines, in order to solve the problem of low accuracy in monitoring risk events of long-distance pipelines in areas with uneven soil moisture distribution.

[0006] To solve the above-mentioned technical problems, this application provides the following technical solution: An artificial intelligence-based method for distinguishing and monitoring risk events in long-distance pipelines includes the following steps: S10: Acquire the soil moisture time-series data and anode output current time-series data in the area where the target pipe section is located, and plot the moisture time curve and current time curve respectively; acquire the vibration signal collected by the fiber optic sensor network. S20: Based on the periodic fluctuations of the current-time curve, select two adjacent maxima with a difference less than the preset current difference as the humidity reference maxima, obtain the humidity-time curve segment corresponding to the two humidity reference maxima as the humidity saturation curve, and take the humidity value range of the humidity saturation curve as the humidity saturation range; extract the continuous current curve segment with humidity within the humidity saturation range from the current-time curve, perform integration on the continuous current curve segment and convert the result into corrosion amount, combine the integration time span to obtain the corrosion rate change relationship with time, as the corrosion rate relationship; associate and store the humidity saturation range with the corresponding corrosion rate relationship. S30: Combining the corrosion amount at different times, the vibration frequency and intensity of the vibration signal, and the results of distinguishing risk events of the vibration signal, a vibration analysis model is constructed based on a neural network model; S40: Obtain the environmental humidity time series data within the prediction time and process it into an environmental humidity curve. Combine the humidity saturation interval to divide the environmental humidity curve into multiple environmental humidity sub-curves. Obtain the corrosion rate relationship corresponding to the humidity saturation interval where the environmental humidity sub-curve is located. Then combine the time nodes on the environmental humidity sub-curve with the corresponding corrosion rate to calculate the corrosion amount corresponding to different time nodes within the prediction time. S50: Input the risk event differentiation results and corrosion amount corresponding to each time point within the prediction time into the vibration analysis model, output the vibration frequency and vibration intensity corresponding to different risk events at each time point, and then compare them with the real-time collected vibration signals and corresponding collection time to generate warning information.

[0007] The basic principle and beneficial effects of this invention are as follows: First, by intelligently identifying the periodic characteristics of the current curve under humidity saturation, this invention extracts and establishes a database of the correspondence between humidity saturation intervals and corrosion rates specific to the region, thus completing adaptive modeling of complex soil heterogeneity environments without the need to set up prediction models for different geological regions. Subsequently, this invention uses the cumulative corrosion amount as a key state variable, fusing it with the characteristics of vibration signals to train a vibration analysis model capable of understanding how corrosion states modulate pipeline vibration responses. Based on the dynamic matching of future humidity changes with historical corrosion evolution patterns under saturation intervals, this invention calculates the expected vibration characteristics accompanying the corrosion process, and then identifies and distinguishes risk events by comparing them with measured vibration signals. In this way, this invention can utilize the electrochemical and physicomechanical responses exhibited by the pipeline itself under extreme humidity conditions as a reference, thereby bypassing direct measurement of specific soil physicochemical properties and achieving universal monitoring across different geological regions.

[0008] Meanwhile, through the above methods, this invention avoids the enormous engineering burden and resource consumption associated with developing and maintaining separate prediction models for each section with distinct soil characteristics when dealing with long-distance pipelines traversing varied geological environments. This invention only needs to rely on the pipeline's own current and vibration responses to autonomously generate a corrosion behavior profile adapted to local conditions, significantly reducing the complexity and cost of monitoring deployment.

[0009] This invention, through coupled analysis of corrosion amount and vibration signals, transforms the hidden physical process of corrosion-induced degradation of pipeline structural stiffness or changes in interfacial friction characteristics into explicit features that can be captured by vibration signals. This allows for the discrepancy between the mechanical disturbances exerted on the pipeline structure by the two types of events (external construction and natural heavy load) in stages of severe coating damage and active corrosion, even when the original vibration signals generated by external construction and natural heavy load are highly similar in spectrum. This significantly improves the accuracy of early warning in high-risk environments.

[0010] This invention not only qualitatively identifies ongoing vibration events but also, by combining humidity prediction data, predicts the evolution of pipeline corrosion and its potential impact on vibration signal characteristics over a future period. This allows managers to deduce which pipeline sections will be more sensitive to third-party construction activities based on future soil moisture data, enabling them to proactively deploy inspection resources or issue preventative control notices, thus elevating risk management from a reactive response to an active intervention.

[0011] This invention utilizes the output vibration signal (including vibration frequency and vibration amplitude) to indirectly interpret environmental input and internal state. It does not rely on precise soil parameters that are difficult to obtain comprehensively; instead, it infers the situation through the pipeline's own behavior. This approach provides technicians with an easy-to-implement monitoring method when facing complex underground uncertainties.

[0012] In summary, this invention eliminates the huge cost of customizing vibration analysis models for each region by establishing a region-adaptive relationship between humidity saturation range and corrosion rate. By utilizing the deep coupling between corrosion state and vibration, it improves the accuracy of event recognition in high-risk scenarios such as coating damage and realizes the transition from passive alarm to proactive and accurate early warning based on humidity prediction.

[0013] Furthermore, a reference electrode is electrically connected to the target pipe section via a wire, and the potential timing data between the reference electrode and the target pipe section is monitored. The trend segment of the current value continuously decreasing in the current-time curve is analyzed, and the trend segment of the potential value continuously moving in the positive direction in the potential timing data is also analyzed. When the continuous decreasing trend of the current and the continuous positive shift of the potential occur synchronously in time, and the potential value exceeds a preset protection potential threshold, the time node is taken as the corrosion critical time. The corrosion critical time characterizes the time when the anolyte material is consumed to the point of failure and the pipe body begins to corrode.

[0014] This invention determines the critical corrosion time by simultaneously monitoring the coordinated changes in potential and current, solving the problem of misjudgment of anodic failure caused by environmental interference in areas with uneven soil resistivity distribution. Methods relying solely on current decay trends are prone to false alarms when soil suddenly dries, increasing resistance and causing a current drop. This invention requires that the current drop be accompanied by a positive potential shift exceeding a threshold, thus accurately distinguishing between environmental interference and actual failure. In scenarios where pipelines traverse areas alternating between dry sand and moist clay, this invention avoids false alarms caused by sudden current drops due to localized soil drying, triggering an alarm only when the anodic potential is truly depleted and the pipeline potential fails, significantly improving alarm accuracy and achieving a leap from symptom-based alarm to fundamental judgment.

[0015] Furthermore, the corrosion amount corresponding to the corrosion critical time is calculated as the protective corrosion amount, and the humidity-time curve after the corrosion critical time is obtained as the humidity trend curve; the protective corrosion amount and the corresponding corrosion critical time are obtained based on the potential time series data and current time series data, and a mapping relationship between the protective corrosion amount, the preset anode material quality and the corrosion critical time is established as the critical mapping relationship; based on the critical mapping relationship and the anode material quality, the relationship between the protective corrosion amount and the corrosion critical time is obtained; the corrosion amount at each time node corresponding to the environmental humidity time series data is obtained, and when the relationship between the time node and the corrosion amount conforms to the relationship between the protective corrosion amount and the corrosion critical time, the time node is used as the predicted corrosion critical time, and the predicted corrosion critical time is used to generate recommended information for replacing the anode material.

[0016] By establishing a mapping relationship between corrosion amount, anode quality, and critical time, personalized life prediction based on the pipeline's own historical data is achieved, solving the resource waste problem of modeling separately for different geological regions. This invention utilizes data from past corrosion critical points to self-learn the corrosion kinetics specific to the region, thereby accurately predicting future anode replacement times. In areas where pipelines traverse significantly different corrosiveness, such as farmland and swamps, establishing independent prediction models for different sections requires substantial manual and computational resources. This invention, however, only needs to record the data of the first occurrence of a critical point in each pipeline section to automatically generate a life prediction curve adapted to local soil characteristics, achieving accurate adaptation of the prediction model and significantly reducing modeling and maintenance costs.

[0017] Furthermore, the location of the target pipe section within the long-distance pipeline is used as the monitoring location, and current time-series data, potential time-series data, and humidity time-series data are acquired at each monitoring location. A potential-time curve is plotted using the potential time-series data, and the potential-time curve, current-time curve, and humidity-time curve are time-aligned. Local humidity curves within the same humidity saturation range are extracted from the humidity-time curves at each monitoring location, and corresponding local current curves and local potential curves are obtained. Instantaneous current and instantaneous potential values ​​at the same time are extracted from the local current and local potential curves, and the instantaneous current values ​​are converted into local instantaneous corrosion rates based on electrochemical equivalence relationships and a preset potential correction factor. The local instantaneous corrosion rate at each monitoring location is integrated over the corresponding time span to obtain the local corrosion amount, and the local corrosion amounts within a preset historical monitoring time are accumulated to obtain the cumulative corrosion amount corresponding to each monitoring location. A spatial distribution map of pipeline corrosion is generated by combining the monitoring locations.

[0018] This invention aligns and fuses multi-source monitoring data such as current, potential, and humidity across time and space, and converts current signals into corrosion quantities based on electrochemical principles, ultimately generating a spatial distribution map of pipeline corrosion. This enables a shift from overall assessment of corrosion status in long-distance pipelines to visualized local lesion-level location. This invention overcomes the limitation of treating long-distance pipe sections as homogeneous corrosion bodies, accurately identifying and visually presenting hidden high-risk corrosion points caused by differences in soil microenvironment, such as localized water accumulation, uneven backfill, or stray current interference.

[0019] Furthermore, based on the cumulative corrosion amount in the spatial distribution map, a median interval is divided. Monitoring locations with cumulative corrosion amounts higher than the upper limit of the median interval are designated as corrosion hotspots, and monitoring locations with corrosion amounts lower than the lower limit of the median interval are designated as corrosion colds. Maintenance prompts are generated to indicate whether to strengthen protection at corrosion hotspots or optimize the placement of anode materials at corrosion colds.

[0020] The methods of uniformly distributing anodes or conducting full-line inspections at fixed intervals are costly and inefficient. This invention, when traversing long distances through saline soil areas, can identify corrosion hotspots caused by stray underground currents or damage to old coatings and prompt for denser anode deployment. Simultaneously, it marks corrosion cold spots caused by soil dryness and prompts for reduced inspection frequency. This data-driven, precise scheduling significantly improves maintenance efficiency and reduces overall operational risk.

[0021] Furthermore, the soil temperature time-series data of the target pipe section area is acquired synchronously, and the temperature time-series data is plotted; the section in the temperature time-series data where the difference between the maximum and minimum temperature values ​​is less than the preset stable difference value is identified as the temperature saturation curve; the local current curve and local potential curve aligned with the temperature saturation curve are analyzed to obtain the corresponding local instantaneous corrosion rate, and the correlation between the local instantaneous corrosion rate and the slope of the temperature saturation curve is analyzed as the temperature-corrosion relationship; when converting the instantaneous current value into the local instantaneous corrosion rate, the local instantaneous corrosion rate is dynamically corrected based on the real-time temperature data through the temperature-corrosion relationship.

[0022] By introducing temperature saturation ranges and temperature-corrosion relationships, the explicit impact of temperature fluctuations on electrochemical reactions is quantified and incorporated into corrosion rate calculations, solving the problem of inaccurate predictions in areas with significant temperature differences. In pipelines in desert regions with large diurnal temperature variations, models based solely on humidity changes cannot explain abnormal fluctuations in nighttime currents. This invention, through temperature correction, accurately isolates the contribution of temperature-induced changes in ion activity to the current signal, yielding a more realistic corrosion rate and significantly improving the accuracy of condition assessment under drastic temperature variations.

[0023] Furthermore, the current-time curve segment after the occurrence of the corrosion critical time is extracted from the current-time curve, and consecutive current maxima and adjacent current minima are identified. The slope of the current-time curve between each pair of adjacent maxima and minima is calculated to form a slope sequence. The changing trend of the slope sequence is analyzed, and the time node corresponding to the slope changing from negative to positive is marked as the time point when the corrosion of the pipeline body enters the stable acceleration stage. The relationship between the time point of the stable acceleration stage and the corrosion critical time is analyzed. Combined with the predicted corrosion critical time corresponding to the environmental humidity curve, the time point of the stable acceleration stage corresponding to the environmental humidity curve is obtained. Based on the time point of the stable acceleration stage, early warning information for replacing the anode material is generated.

[0024] By analyzing the sign change of the current curve slope after the corrosion critical point, this invention can identify the turning point from slow initiation to stable acceleration of pipeline corrosion, enabling early prediction of corrosion development and providing a more forward-looking early warning signal than the critical point itself. When the slope changes from negative to positive, even if the pipeline potential has not yet exceeded the protection potential threshold, it indicates that the corrosion current density has begun to increase continuously, and the cathodic polarization of the pipeline surface is rapidly decreasing. Based on this, this invention can issue an early warning of the corrosion acceleration trend, buying valuable time for personnel to intervene before the pipeline becomes completely unprotected (i.e., before the anodic material loses its function and long-distance pipelines begin to be extensively corroded), thus achieving further advance risk warning.

[0025] Furthermore, during the process of obtaining the predicted critical corrosion time, the ambient temperature time series data within the predicted time period is obtained simultaneously; based on the temperature-corrosion relationship, the corrosion amount at the time node corresponding to the ambient humidity time series data is corrected; the relationship between the corrected corrosion amount and the critical corrosion time is matched, and when the corrected relationship conforms to the relationship between the protective corrosion amount and the critical corrosion time, the corresponding time node is taken as the predicted critical corrosion time under the influence of temperature, and recommendation information for replacing the anode material is generated accordingly.

[0026] When predicting future corrosion criticality time, temperature prediction data is simultaneously incorporated to correct corrosion rate, achieving more accurate lifetime prediction based on a dual environmental driver of humidity and temperature. This addresses the problem of unreliable anode lifetime assessment results caused by the interplay of multiple factors in complex climate change regions. For example, during the spring snowmelt season, increased soil moisture typically accelerates corrosion while low temperatures inhibit the reaction; predictions based solely on humidity changes would produce significant errors. This invention, by combining humidity and temperature change trends to jointly correct corrosion rates, effectively recreates the actual corrosion accumulation process in complex environments and accurately determines anode failure time, making predictions more consistent with the actual environment and improving the reliability of recommended information.

[0027] Furthermore, in the process of generating the spatial distribution map of pipeline corrosion, the calculated local instantaneous corrosion rate at each monitoring location is corrected using the temperature-corrosion relationship; based on the corrected local instantaneous corrosion rate, the cumulative corrosion at each monitoring location is recalculated; and the spatial distribution map is adjusted according to the corrected cumulative corrosion.

[0028] This invention applies temperature correction to the generation of corrosion spatial distribution maps, resulting in a more realistic distribution map reflecting the coupling relationship between the temperature field and the corrosion field. It can reveal spatial distribution characteristics that are difficult to identify based solely on humidity and current data. For long-distance pipelines near industrial hot drainage channels, corrosion maps based solely on humidity and current may only show uniform corrosion, while the temperature-corrected maps of this invention clearly show higher corrosion levels closer to the heat source, explicitly indicating the existence of thermal corrosion effects and their spatial range of influence. This provides a more refined basis for assessing the risks of special thermal environments and designing asymmetric protection schemes, enabling a spatial deconstruction of complex corrosion-causing factors. Attached Figure Description

[0029] Figure 1 This is a flowchart of the AI-based long-distance pipeline risk event differentiation and monitoring method in Example 1; Figure 2 This is a structural diagram of the long-distance pipeline and anode material in Example 1; The attached figures include: 1. Soil; 2. Current acquisition device; 3. Wire; 4. Long-distance pipeline; 5. Anode material; 6. Humidity acquisition device. Detailed Implementation

[0030] The following will describe the concept and technical effects of the present invention clearly and completely with reference to embodiments, so as to fully understand the purpose, features and effects of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention. Example 1 like Figure 2As shown, the long-distance pipeline and anode material are buried in the soil and electrically connected by a conductor. The horizontal distance between the anode material and the pipeline is typically 0.5-3m, but in this embodiment, it is 1m by default. In this embodiment, the anode material is either a magnesium alloy anode or a zinc alloy anode. A current acquisition device is installed on the conductor to collect the current. The current acquisition device combines the collected current and the collection time to process it into current time-series data. The anode material and the pipeline are on the same horizontal plane, and a humidity acquisition device is installed in the soil near the anode material and the pipeline. The humidity acquisition device combines the collected humidity and the collection time to process it into humidity time-series data. In this embodiment, the humidity acquisition device is defaulted to being located on the same side as the anode material, and the straight-line distance between the humidity acquisition device and the pipeline does not exceed the straight-line distance between the pipeline and the anode material, and the straight-line distance between the humidity acquisition device and the anode material does not exceed 1m.

[0031] like Figure 1 As shown, the AI-based method for distinguishing and monitoring risk events in long-distance pipelines includes the following steps: S10: Acquire the time-series data of soil moisture and anode material output current in the target pipe section area, and plot the moisture-time curve and current-time curve respectively; acquire the vibration signal collected by the fiber optic sensor network. The fiber optic sensor network is laid along the axial direction of the long-distance pipeline. In this embodiment, it is buried directly above or to the side of the pipeline by default, with a horizontal distance of 0.5-1m from the pipeline.

[0032] With soil moisture remaining relatively constant, the periodic fluctuations in the anode output current are primarily caused by the periodic changes in the concentration of metal ions in the soil. At lower ion concentrations, the displacement reaction is limited by metal ion migration, resulting in a smaller current. As the ion concentration increases, the reaction rate accelerates, and the current rises. When the metal ion concentration approaches saturation, the reaction rate increases more slowly, and the current tends to stabilize. This periodic rise and fall in metal ion concentration causes the current-time curve to exhibit periodic fluctuations, thus reflecting the dynamic changes in soil electrochemical activity.

[0033] S20: Based on the periodic fluctuations of the current-time curve, select two adjacent maxima with a difference less than the preset current difference (set by the administrator according to the calculation accuracy) as humidity reference maxima. Obtain the humidity-time curve segment corresponding to the two humidity reference maxima as the humidity saturation curve (the humidity change curve corresponding to the current being in a stable high value range). The humidity value range of the humidity saturation curve is taken as the humidity saturation range. Within the humidity saturation range, the corrosion reaction rate has a good linear relationship with the current, so the corrosion amount can be accurately calculated by integrating the current. Extract the continuous current curve segment within the humidity saturation range from the current-time curve, perform integration on the continuous current curve segment, and convert the result into the corrosion amount. Combine the integration time span to obtain the corrosion rate change relationship with time, which is taken as the corrosion rate relationship. Store the humidity saturation range and the corresponding corrosion rate relationship together.

[0034] Different humidity saturation ranges correspond to different soil electrochemical environments, and corrosion rates vary significantly under different environments. By establishing a one-to-one correspondence between humidity saturation ranges and corrosion rates, a corrosion rate more suitable for the current humidity environment can be matched based on real-time humidity, thereby improving the accuracy of corrosion prediction.

[0035] S30: Combine the corrosion amount at different times, the vibration frequency and intensity of the vibration signal, and the results of distinguishing risk events of the vibration signal, and construct a vibration analysis model based on a neural network model.

[0036] The vibration analysis model employs convolutional neural networks or long short-term memory networks to effectively handle the temporal characteristics of vibration signals. Its architecture includes an input layer, convolutional layers (or recurrent layers), pooling layers, fully connected layers, and an output layer. The input consists of corrosion amount, vibration frequency, vibration intensity, and risk event differentiation results (i.e., construction or compaction). The output is the predicted vibration frequency and intensity values ​​corresponding to different risk events. Administrators set up long-distance pipelines with varying corrosion levels as experimental materials and buried them at the depths specified by the administrators. Vibration signals were collected and manually labeled under simulated construction intensities and vehicle compaction conditions. The labeling included the event type, vibration intensity level, and whether it constituted an absolute risk. Vibration signals with insufficient construction intensity or insufficient compaction weight had low intensity and frequency and would not pose an actual risk to the pipeline. Labeling allowed the vibration analysis model to learn the truly dangerous vibration characteristics, thereby improving the accuracy of event differentiation.

[0037] S40: Obtain environmental humidity time-series data (which can be obtained from meteorological data) within the predicted time (a future period set by the administrator), process it into an environmental humidity curve, and divide the environmental humidity curve into multiple environmental humidity sub-curves (corresponding to different corrosion rates in different humidity ranges) by combining the humidity saturation range. Obtain the corrosion rate relationship corresponding to the humidity saturation range where the environmental humidity sub-curve is located, and then combine the time nodes on the environmental humidity sub-curve with the corresponding corrosion rates to calculate the corrosion amount corresponding to different time nodes within the predicted time.

[0038] First, the humidity value range of the environmental humidity sub-curve is matched with the corresponding stored humidity saturation interval, and the change pattern of corrosion rate over time within the corresponding humidity saturation interval is retrieved. Using the time axis as a reference, each time node on the environmental humidity sub-curve is mapped to the corrosion rate at the same time position in the corrosion rate relationship, establishing a correspondence between time nodes and corrosion rates. The calculation process is based on Faraday's law of electrolysis. The corrosion rate at each time node is multiplied by the time interval between adjacent nodes to obtain the corrosion increment for that period. The corrosion increments for each period are then summed sequentially to obtain the cumulative corrosion amount corresponding to each time node within the predicted time period.

[0039] S50: Differentiate between the results and corrosion of risk events corresponding to each time point within the predicted time period. The vibration analysis model takes a sample as input and outputs the vibration frequency and intensity corresponding to different risk events at each time point. It then compares these values ​​with the real-time collected vibration signals and the corresponding collection time to generate warning information.

[0040] Based on the labeled data input from the vibration analysis model, risk levels are set for different vibration frequencies, intensities, and event types. Risk levels are categorized as low, medium, high, and extremely high. Warning criteria are composed of the risk level, vibration frequency range, and vibration intensity threshold. When a real-time acquired vibration signal meets a certain warning criterion, the corresponding pipe section location is determined based on the signal acquisition location, and a warning message is generated in conjunction with the risk level. Administrators can then verify the warning information on-site.

[0041] A reference electrode is electrically connected to the target pipe section via a wire, and the potential timing data between the reference electrode and the target pipe section is monitored. The reference electrode provides a stable potential reference for accurately measuring the polarization potential of the pipe; in this embodiment, the material is assumed to be copper sulfate or silver chloride. The reference electrode reflects changes in the electrochemical state of the pipe surface and is a key basis for determining whether the anode has failed.

[0042] After the anolyte is consumed, the cathodic polarization of the pipeline decreases, and the potential gradually approaches the natural corrosion potential. The trend segment of the continuously decreasing current value in the current-time curve is analyzed, as well as the trend segment of the continuously positively shifting potential value in the potential time series data. When the continuously decreasing current trend and the continuously positively shifting potential occur synchronously in time, and the potential value exceeds the preset protection potential threshold (set according to the minimum potential requirement for cathodic protection of the pipeline, representing the critical potential at which the pipeline surface achieves sufficient cathodic polarization and corrosion is effectively suppressed; when the potential value exceeds the preset protection potential threshold, it indicates insufficient pipeline polarization and a significant increase in corrosion risk), this time point is taken as the critical corrosion time. The critical corrosion time characterizes the time from the consumption of the anolyte to its failure and the beginning of corrosion in the pipeline body.

[0043] The segment where the potential continuously moves in the positive direction reflects the gradual weakening of the cathode polarization of the pipeline, which is achieved by fitting a sliding window to the potential time series data. A linear fit is performed on the potential data using a fixed time window (set by the administrator). When the fitting slope is positive and continuously exceeds a first set threshold (set by the administrator, usually determined based on the stability of the reference electrode, soil noise level, and historical potential fluctuation range to ensure the ability to distinguish between actual polarization changes and measurement noise), it is identified as a segment of positive potential shift.

[0044] Linear fitting of current time series data is performed by sliding window. When the fitting slope is negative and continuously exceeds the second set threshold (determined by the administrator based on the stability of anode output, the magnitude of soil ion concentration change and current measurement accuracy, to avoid misjudging anode decay due to short-term ion concentration fluctuations), it is determined to be a current decline trend segment.

[0045] The corrosion amount corresponding to the critical corrosion time is calculated as the protective corrosion amount (referring to the accumulated corrosion amount in the pipeline when the anode fails, representing the maximum corrosion amount the pipeline can withstand under anode protection). The humidity-time curve after the critical corrosion time is obtained as the humidity trend curve. Based on the potential time series data and current time series data, the protective corrosion amount and the corresponding critical corrosion time are obtained. A mapping relationship (a mathematical relationship) between the protective corrosion amount, the preset anode material mass, and the critical corrosion time is established as the critical mapping relationship. For example, the larger the anode mass, the greater the protective corrosion amount and the later the critical corrosion time.

[0046] Based on the critical mapping relationship and the quality of the anode material, the relationship between the protective corrosion amount and the critical corrosion time is obtained. The curve of the predicted corrosion amount changing with time is aligned with the typical curve determined by the critical mapping relationship, with the alignment method based on the consistency of the ratio of the time axis and the corrosion amount axis. Subsequently, point-by-point error calculation is performed on the corresponding points of the two curves, and the error includes absolute error and relative error, which is used to measure the degree of deviation between the predicted curve and the typical curve.

[0047] When the predicted curve and the critical mapping curve maintain a consistent main trend, and the overall error is controlled within a preset range of 5%-10%, they are considered to be in agreement. This 5%-10% range is set based on a comprehensive consideration of corrosion calculation accuracy, current acquisition error, humidity fluctuations, and the stability of historical data, effectively avoiding misjudgments caused by environmental noise or measurement deviations. Furthermore, the rate of change of the curves is compared, including the slope of the rising phase, the inflection point position, and the consistency of the time taken to reach the protective corrosion level. If the predicted curve is highly consistent with the critical mapping curve at key feature points, even with minor local deviations, it can still be considered a match. High consistency is characterized by a small overall deviation between the predicted curve and the critical mapping curve (the specific deviation judgment criteria are set by the administrator based on accuracy), and the trend and rate of change are basically the same. Minor deviations refer to limited local differences that do not affect the overall trend, have a short duration, and do not change the key features and inflection point positions of the curve.

[0048] The corrosion amount at each time point corresponding to the environmental humidity time series data is obtained. When the relationship between the time point and the corrosion amount conforms to the relationship between the protective corrosion amount and the corrosion critical time, the time point is used as the predicted corrosion critical time. The predicted corrosion critical time is then used to generate recommended information for replacing the anode material. The recommended information includes the predicted corrosion critical time and the location of the anode that needs to be replaced.

[0049] Extract the current-time curve segment after the occurrence of the corrosion critical time from the current-time curve, and identify consecutive current maxima and adjacent current minima. Calculate the slope of the current-time curve between each pair of adjacent maxima and minima (reflecting the rate of current change, representing the activity of the corrosion reaction), forming a slope sequence (containing the rate of current change between adjacent maxima and minima). Analyze the changing trend of the slope sequence, and mark the time point corresponding to the slope turning from negative to positive as the time point when the pipeline corrosion enters the stable accelerated phase. For example, when the slope turns from negative to positive, it indicates that the current decrease trend has stopped and has begun to rise, indicating that the pipeline corrosion has intensified. The time point when the slope turns from negative to positive is marked as the beginning of the stable accelerated phase.

[0050] The relationship between the time point of the stable acceleration phase and the critical corrosion time is analyzed. Combined with the predicted critical corrosion time corresponding to the ambient humidity curve, the time point of the stable acceleration phase corresponding to the ambient humidity curve is obtained. Based on the time point of the stable acceleration phase, early warning information for replacing anode materials is generated.

[0051] Early warning messages arrive earlier, indicating that the anode is about to fail but has not yet completely failed. Recommendation messages, on the other hand, issue explicit replacement instructions when the predicted critical corrosion time is reached.

[0052] The current-time curve, humidity-time curve, potential-time curve, corrosion rate curve, and vibration signal curve are all plotted with time as the X-axis and current (amperes), humidity (percentage of volumetric water content), potential (volts), corrosion rate (millimeters), vibration intensity (meters per second squared), or vibration frequency (hertz) as the Y-axis, respectively.

[0053] In practice, the fiber optic sensor network is continuously deployed along the long-distance pipeline, synchronously collecting vibration intensity and frequency data around the pipeline. After the vibration wave is transmitted through the soil to the optical fiber, it causes a phase change in the optical signal within the fiber. This phase change is demodulated and converted into continuous monitoring data of vibration intensity and frequency, enabling real-time capture of vibrations around the pipeline.

[0054] Based on vibration data collected under various simulated working conditions, the data is classified and labeled according to event type to distinguish between vibrations that may endanger the pipeline and background vibrations that pose no risk. The labeled vibration intensity and frequency data are combined with the corresponding corrosion amount and risk classification results and input into the vibration analysis model to complete the training, so that the vibration analysis model can master the characteristic laws of dangerous vibrations.

[0055] During actual monitoring, the vibration analysis model combines the current corrosion level with historical vibration characteristics to output reference ranges for vibration intensity and frequency corresponding to different risk events. The two parameters collected in real time are compared with the reference ranges to determine whether the vibration poses a risk.

[0056] When real-time parameters meet the characteristics of a certain risk level, the associated signal acquisition location is linked to generate a warning message containing the risk level and pipe section information. At the same time, the remote sensing image acquisition device is triggered to take an image of the warning pipe section, which is then used by the administrator to verify and process the vibration data and the image.

[0057] This embodiment also includes an AI-based long-distance pipeline risk event differentiation and monitoring system that uses an AI-based long-distance pipeline risk event differentiation and monitoring method.

[0058] Example 2 The only difference between this embodiment and Embodiment 1 is that the location of the target pipe section within the long-distance pipeline is used as the monitoring location, and current time-series data, potential time-series data, and humidity time-series data are acquired at each monitoring location. The monitoring locations are evenly distributed along the pipeline axis, and the administrator can flexibly adjust the interval between monitoring locations based on preliminary site survey results, pipeline operation experience, and risk assessment reports. Each monitoring location is equipped with a current acquisition device, a reference electrode, and a humidity sensor, as in Embodiment 1, to synchronously acquire current time-series data, potential time-series data, and humidity time-series data. A potential-time curve is plotted using the potential time-series data, and the potential-time curve, current-time curve, and humidity-time curve are aligned in time.

[0059] Local humidity curves within the same humidity saturation range are extracted from the humidity-time curves at each monitoring location, and the corresponding local current and local potential curves are obtained. Instantaneous current and instantaneous potential values ​​at the same moment are extracted from the local current and local potential curves, and the instantaneous current values ​​are converted into local instantaneous corrosion rates according to the electrochemical equivalence relationship and a preset potential correction factor.

[0060] The electrochemical equivalence relationship refers to the quantitative relationship between the amount of material dissolved or deposited and the amount of electricity passing through the system during the electrolysis or corrosion process of a metal. It indicates that the amount of material consumed by metal corrosion is directly proportional to the magnitude of the corrosion current, and the proportionality coefficient is determined by the electrochemical equivalence of the metal. The electrochemical equivalence reflects the change in metal mass caused by a unit amount of electricity, and its magnitude depends on the molar mass of the metal, the number of electrons participating in the reaction, and the Faraday constant.

[0061] The potential correction factor is determined based on the polarization characteristics of pipeline steel, the influence of soil environment on electrode response, reference protection potential, and historical monitoring data. This is achieved through experimental testing of the polarization curves of pipeline steel at different potentials, correction of electrode response patterns using soil environmental parameters, and statistical calibration using the reference protection potential and historical data. Statistical calibration is used to analyze the relationship between potential and actual corrosion rate based on extensive field monitoring and experimental data. Regression analysis and error minimization methods are employed to optimize the potential correction factor, enabling it to more accurately reflect the impact of potential changes on corrosion rate.

[0062] Specifically, according to the electrochemical equivalence relationship, the corrosion rate is directly proportional to the instantaneous current value, and the proportionality coefficient is determined by the electrochemical equivalence of the metal and the Faraday constant. A potential correction factor is introduced to correct the effect of potential deviation from the reference protection potential on the corrosion rate. Multiplying the instantaneous current value by the electrochemical equivalence coefficient and the potential correction factor yields the local instantaneous corrosion rate. The local instantaneous corrosion rate is equal to the electrochemical equivalence coefficient multiplied by the instantaneous current value, and then multiplied by the potential correction factor.

[0063] The local instantaneous corrosion rate at each monitoring location is integrated over the corresponding time span to obtain the local corrosion amount. The local corrosion amounts within a preset historical monitoring time are accumulated to obtain the cumulative corrosion amount corresponding to each monitoring location. A spatial distribution map of pipeline corrosion is generated by combining the monitoring locations.

[0064] From the spatial distribution map, median intervals are divided according to the cumulative corrosion amount. Monitoring locations with cumulative corrosion amounts higher than the upper limit of the median interval are designated as corrosion hotspots, and monitoring locations with corrosion amounts lower than the lower limit of the median interval are designated as corrosion colds. Maintenance prompts are generated to indicate whether to strengthen protection at corrosion hotspots or optimize the placement of anode materials at corrosion colds.

[0065] The median interval is determined based on the statistical distribution characteristics of cumulative corrosion at each monitoring location. Centered on the median, the corrosion distribution characteristics are statistically analyzed, taking into account environmental factors such as soil moisture and resistivity, referencing historical and similar pipeline data, and adjusting the position and width of the median interval based on engineering experience. This ensures that the median interval covers the main normal corrosion range of the pipeline while maintaining the sensitivity and reliability of identifying corrosion hotspots and cold spots.

[0066] The system synchronously acquires and plots time-series soil temperature data for the target pipe section's area. Administrators determine preset stability difference values ​​based on natural temperature fluctuations, the stability requirements of temperature's impact on corrosion, and calculation accuracy, ensuring the identified temperature saturation curve reflects the true stable state. The temperature saturation curve is defined as the segment in the time-series temperature data where the difference between the maximum and minimum temperature values ​​is less than the preset stability difference value. Local current and potential curves aligned with the temperature saturation curve are analyzed to obtain the corresponding local instantaneous corrosion rates. The correlation between the local instantaneous corrosion rate and the slope of the temperature saturation curve is analyzed as the temperature-corrosion relationship. When converting instantaneous current values ​​to local instantaneous corrosion rates, the local instantaneous corrosion rates are dynamically corrected based on real-time temperature data using the temperature-corrosion relationship.

[0067] In this process, after acquiring real-time temperature data, the rate of temperature change is calculated, the established temperature-corrosion relationship is queried, and the corresponding temperature influence coefficient is obtained. The temperature influence coefficient is then multiplied by the local instantaneous corrosion rate calculated based on current and potential to obtain the corrected local instantaneous corrosion rate, so that the corrosion rate is adjusted in real time with temperature changes.

[0068] In practice, multiple monitoring locations are first set up along the axial direction of the target pipe section. At each location, a current acquisition device, a reference electrode, and a humidity sensor are deployed as described in Example 1. Current time-series data, potential time-series data, and humidity time-series data are collected simultaneously, and the three types of data are aligned by time. Subsequently, local humidity curves within the same humidity saturation range are identified and extracted from the humidity-time curves. At the same time, the corresponding local current curves and local potential curves are obtained, and the instantaneous current and instantaneous potential values ​​at the same moment are extracted from them.

[0069] According to the electrochemical equivalence relationship, the corrosion rate is directly proportional to the instantaneous current value, and the proportionality coefficient is determined by the electrochemical equivalence of the metal and the Faraday constant. Combining the potential correction factor determined through experiments and statistical calibration based on the polarization characteristics of the pipeline steel, the influence of the soil environment, the reference protection potential, and historical data, the instantaneous current value is multiplied by the electrochemical equivalence coefficient and the potential correction factor to obtain the local instantaneous corrosion rate.

[0070] The local instantaneous corrosion rate at each monitoring location is integrated over the corresponding time span to obtain the local corrosion amount. The local corrosion amounts over historical monitoring periods are then accumulated to obtain the cumulative corrosion amount at each monitoring location. This, combined with location information, generates a spatial distribution map of the corrosion amount. Based on the statistical distribution characteristics of corrosion, and considering environmental factors such as soil moisture and resistivity, as well as historical data, the system divides the data into median intervals. Locations above the upper limit of the interval are marked as corrosion hotspots, and those below the lower limit are marked as corrosion colds, generating corresponding maintenance prompts.

[0071] Simultaneously, soil temperature time-series data is collected. Preset stable difference values ​​are set based on the natural temperature fluctuation range and stability requirements, and stable temperature change segments are identified as temperature saturation curves. Local current and potential curves aligned with the temperature saturation curves are analyzed to establish a correlation between the local instantaneous corrosion rate and the slope of temperature change, serving as the temperature-corrosion relationship. In subsequent corrosion rate calculations, the temperature change rate is calculated based on real-time temperature data, and the temperature influence coefficient is obtained by querying the temperature-corrosion relationship. This coefficient is then multiplied by the corrosion rate obtained based on current and potential, achieving dynamic correction of the local instantaneous corrosion rate and improving the accuracy of corrosion calculation.

[0072] Example 3 The only difference between this embodiment and embodiments 1-2 is that, during the process of obtaining the predicted corrosion critical time, environmental temperature time-series data within the prediction time period is acquired simultaneously; and the corrosion amount at the time node corresponding to the environmental humidity time-series data is corrected based on the temperature-corrosion relationship. Specifically, during the prediction of the corrosion critical time, environmental temperature time-series data within the prediction time range is acquired simultaneously, ensuring that the time nodes of the temperature data completely correspond to the humidity time-series data. Subsequently, for each time node, the temperature change rate at that moment is calculated based on the temperature time-series data, and then, based on the established temperature-corrosion relationship, the temperature influence coefficient corresponding to the temperature change rate is obtained. The temperature influence coefficient reflects the degree to which the current temperature conditions promote or inhibit the corrosion rate. The temperature influence coefficient is multiplied by the predicted corrosion amount at that time node to obtain the corrected corrosion amount.

[0073] The relationship between the corrected corrosion amount and the critical corrosion time is matched. When the corrected relationship matches the relationship between the protective corrosion amount and the critical corrosion time, the corresponding time node is taken as the predicted critical corrosion time under the influence of temperature, and recommendation information for replacing the anode material is generated accordingly.

[0074] During the acquisition of the predicted corrosion critical time, ambient temperature time-series data within the prediction period are acquired simultaneously. Based on the temperature-corrosion relationship, the corrosion amount at the corresponding time nodes of the ambient humidity time-series data is corrected. Specifically, it is first ensured that the time nodes of the ambient temperature time-series data and the ambient humidity time-series data within the prediction period correspond one-to-one, and each humidity data point is matched with a unique temperature data point to ensure the synchronization of the correction. For each corresponding time node, the temperature change rate (such as the temperature rise or fall per unit time) is calculated based on the temperature time-series data to quantify the dynamic change state of the current temperature. The established temperature-corrosion relationship (reflecting the correlation between the temperature change rate and the corrosion rate) is invoked, and the corresponding temperature influence coefficient is obtained based on the calculated temperature change rate. The original predicted corrosion amount at this time node, which does not consider the temperature influence, is multiplied by the obtained temperature influence coefficient to obtain the corrected corrosion amount, completing the temperature correction of the corrosion amount at this time node. The above steps are repeated to correct the corrosion amount at each time node corresponding to all humidity time-series data, forming complete corrosion amount time-series data corrected for temperature influence, providing an accurate basis for subsequent corrosion critical time matching.

[0075] The relationship between the corrected corrosion amount and the critical corrosion time is matched. When the corrected relationship matches the relationship between the protective corrosion amount and the critical corrosion time, the corresponding time node is taken as the predicted critical corrosion time under the influence of temperature, and recommendation information for replacing the anode material is generated accordingly.

[0076] In practice, during the prediction of the critical corrosion time, ambient temperature time-series data within the prediction time range is acquired simultaneously, ensuring a one-to-one correspondence between the temperature data and humidity time-series data. For each time point, the temperature change rate is calculated based on the temperature time-series data, and then the corresponding temperature influence coefficient is looked up based on the established temperature-corrosion relationship. The temperature influence coefficient reflects the degree to which temperature promotes or inhibits the corrosion rate. The temperature influence coefficient is multiplied by the original predicted corrosion amount for that time point to obtain the corrected corrosion amount. The above steps are repeated for all time points to form complete corrected corrosion amount time-series data.

[0077] Subsequently, the relationship between the corrected corrosion amount and the critical corrosion time is matched. When the corrected corrosion amount reaches the critical condition corresponding to the protective corrosion amount, the time node is determined as the predicted critical corrosion time under the influence of temperature. Based on this, anode material replacement recommendation information is generated, including the predicted critical time, the remaining safe operating time, the recommended replacement time window, and the key areas of concern.

[0078] The above are merely embodiments of the present invention. The invention is not limited to the fields covered by these embodiments. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are able to access all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for distinguishing and monitoring risk events in long-distance pipelines based on artificial intelligence, characterized in that, Includes the following steps: S10: Acquire the time-series data of soil moisture and anode material output current in the target pipe section area, and plot the moisture-time curve and current-time curve respectively; acquire the vibration signal collected by the fiber optic sensor network. S20: Based on the periodic fluctuations of the current-time curve, select two adjacent maxima with a difference less than the preset current difference as the humidity reference maxima, obtain the humidity-time curve segment corresponding to the two humidity reference maxima as the humidity saturation curve, and take the humidity value range of the humidity saturation curve as the humidity saturation range; extract the continuous current curve segment with humidity within the humidity saturation range from the current-time curve, perform integration on the continuous current curve segment and convert the result into corrosion amount, combine the integration time span to obtain the corrosion rate change relationship with time, as the corrosion rate relationship; associate and store the humidity saturation range with the corresponding corrosion rate relationship. S30: Combining the corrosion amount at different times, the vibration frequency and intensity of the vibration signal, and the results of distinguishing risk events of the vibration signal, a vibration analysis model is constructed based on a neural network model; S40: Obtain the environmental humidity time series data within the prediction time and process it into an environmental humidity curve. Combine the humidity saturation interval to divide the environmental humidity curve into multiple environmental humidity sub-curves. Obtain the corrosion rate relationship corresponding to the humidity saturation interval where the environmental humidity sub-curve is located. Then combine the time nodes on the environmental humidity sub-curve with the corresponding corrosion rate to calculate the corrosion amount corresponding to different time nodes within the prediction time. S50: Input the risk event differentiation results and corrosion amount corresponding to each time point within the prediction time into the vibration analysis model, output the vibration frequency and vibration intensity corresponding to different risk events at each time point, and then compare them with the real-time collected vibration signals and corresponding collection time to generate warning information.

2. The method for distinguishing and monitoring risk events in long-distance pipelines based on artificial intelligence according to claim 1, characterized in that: A reference electrode is electrically connected to the target pipe segment via a wire, and the potential timing data between the reference electrode and the target pipe segment is monitored; the trend segment in the current-time curve where the current value continuously decreases is analyzed, and the trend segment in the potential timing data where the potential value continuously moves in the positive direction is also analyzed. When the continuous downward trend of the current and the continuous positive shift of the potential occur synchronously in time, and the potential value exceeds the preset protection potential threshold, the time node is taken as the corrosion critical time; the corrosion critical time characterizes the time when the anode material is consumed to the point of failure and the pipeline body begins to corrode.

3. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 2, characterized in that: The corrosion amount corresponding to the corrosion critical time is calculated as the protective corrosion amount, and the humidity-time curve after the corrosion critical time is obtained as the humidity trend curve. Based on potential time series data and current time series data, the protective corrosion amount and the corresponding corrosion critical time are obtained, and a mapping relationship between the protective corrosion amount, the preset anode material mass and the corrosion critical time is established as the critical mapping relationship; Based on the critical mapping relationship and the quality of the anode material, the relationship between the protective corrosion amount and the critical corrosion time is obtained; the corrosion amount at each time node corresponding to the environmental humidity time series data is obtained. When the relationship between the time node and the corrosion amount conforms to the relationship between the protective corrosion amount and the critical corrosion time, the time node is used as the predicted critical corrosion time, and the predicted critical corrosion time is used to generate recommended information for replacing the anode material.

4. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 2, characterized in that: The location of the target pipe section in the long-distance pipeline is used as the monitoring location. Current time series data, potential time series data and humidity time series data are collected at each monitoring location. Potential time curves are plotted using the potential time series data, and the potential time curves, current time curves and humidity time curves are aligned in time. Local humidity curves within the same humidity saturation range are extracted from the humidity-time curves at each monitoring location, and corresponding local current and local potential curves are obtained. Instantaneous current and instantaneous potential values ​​at the same time are extracted from the local current and local potential curves. The instantaneous current values ​​are converted into local instantaneous corrosion rates according to the electrochemical equivalence relationship and a preset potential correction factor. The local instantaneous corrosion rate at each monitoring location is integrated over the corresponding time span to obtain the local corrosion amount. The local corrosion amounts over a preset historical monitoring time are accumulated to obtain the cumulative corrosion amount corresponding to each monitoring location. A spatial distribution map of pipeline corrosion is generated by combining the monitoring locations.

5. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 4, characterized in that: From the spatial distribution map, median intervals are divided according to the cumulative corrosion amount. Monitoring locations with cumulative corrosion amounts higher than the upper limit of the median interval are designated as corrosion hotspots, and monitoring locations with corrosion amounts lower than the lower limit of the median interval are designated as corrosion colds. Maintenance prompts are generated to indicate whether to strengthen protection at corrosion hotspots or optimize the placement of anode materials at corrosion colds.

6. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 4, characterized in that: The soil temperature time-series data of the target pipe section area is acquired synchronously, and the temperature time-series data is plotted. The section in the temperature time-series data where the difference between the maximum and minimum temperature values ​​is less than the preset stable difference value is identified as the temperature saturation curve. The local current curve and local potential curve that are time-aligned with the temperature saturation curve are analyzed to obtain the corresponding local instantaneous corrosion rate. The correlation between the local instantaneous corrosion rate and the slope of the temperature saturation curve is analyzed as the temperature-corrosion relationship. When converting the instantaneous current value into the local instantaneous corrosion rate, the local instantaneous corrosion rate is dynamically corrected based on the real-time temperature data and the temperature-corrosion relationship.

7. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 2, characterized in that: Extract the current-time curve segment after the occurrence of the corrosion critical time from the current-time curve, identify consecutive current maxima and adjacent current minima, calculate the slope of the current-time curve between each pair of adjacent maxima and minima to form a slope sequence, analyze the changing trend of the slope sequence, and mark the time node corresponding to the slope turning from negative to positive as the time point when the corrosion of the pipeline body enters the stable acceleration stage. The relationship between the time point of the stable acceleration phase and the critical corrosion time is analyzed. Combined with the predicted critical corrosion time corresponding to the ambient humidity curve, the time point of the stable acceleration phase corresponding to the ambient humidity curve is obtained. Based on the time point of the stable acceleration phase, early warning information for replacing anode materials is generated.

8. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 3 or 6, characterized in that: During the process of obtaining the predicted critical corrosion time, the ambient temperature time series data within the predicted time period is acquired simultaneously. Based on the temperature-corrosion relationship, the corrosion amount at the time node corresponding to the ambient humidity time series data is corrected. The relationship between the corrected corrosion amount and the critical corrosion time is matched. When the corrected relationship conforms to the relationship between the protective corrosion amount and the critical corrosion time, the corresponding time node is taken as the predicted critical corrosion time under the influence of temperature, and recommendation information for replacing the anode material is generated accordingly.

9. The artificial intelligence-based risk event differentiation and monitoring method for long-distance pipelines according to claim 6, characterized in that: In the process of generating the spatial distribution map of pipeline corrosion, the local instantaneous corrosion rate of each monitoring location is corrected by the temperature-corrosion relationship; based on the corrected local instantaneous corrosion rate, the cumulative corrosion at each monitoring location is recalculated; and the spatial distribution map is adjusted according to the corrected cumulative corrosion.

10. A long-distance pipeline risk event differentiation and monitoring system based on artificial intelligence, characterized in that, The method for distinguishing and monitoring risk events in long-distance pipelines based on any one of claims 1-9 was used.

Images