A soluble bridge plug failure early warning method and system

Abstract

The application discloses a soluble bridge plug fault early warning method and system, and relates to the technical field of soluble bridge plug fault early warning. The soluble bridge plug fault early warning method obtains the dissolving state time series data and the dissolving environment time series data of the soluble bridge plug based on a set time period; extracts the state evolution feature set and the environment evolution feature set based on the pre-trained time series feature perception model; generates the state evolution feature index and the environment evolution feature index of the soluble bridge plug; judges whether the soluble bridge plug has a fault risk based on the state evolution feature index and the environment evolution feature index, and sends a fault alarm when the fault risk exists. The application combines the dissolving state data and the environment data to construct a complete real-time monitoring and dynamic analysis process, can discover abnormal behaviors in the dissolving process in time, early warns potential faults, avoids the lagging diagnosis after the traditional unblocking failure, and thus improves the operation efficiency and reduces the safety risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of soluble bridge plug failure early warning, in particular to a soluble bridge plug failure early warning method and system. BACKGROUND

[0002] A soluble bridge plug is a commonly used plugging device in oil and gas wells or other underground engineering, usually used to control the flow of fluid in the wellbore or block the pressure of different areas. During the drilling, completion or construction process of the oil and gas well, the bridge plug forms a closed space by contacting the well wall, ensuring that the fluid or gas in the well is effectively isolated. However, as the downhole operation continues, the dissolution process of the bridge plug will be affected by environmental factors such as downhole temperature, pressure, fluid composition, etc., which may cause the failure or failure of the bridge plug.

[0003] During the dissolution process of the bridge plug, due to the combined action of various environmental factors, local cracking, loosening, and even complete failure may occur. If not detected and warned in time, the bridge plug failure will have a serious impact on the subsequent operation process, and even cause damage to downhole equipment, delay the operation, or pose a safety risk to downhole operation personnel. Therefore, timely identification of the dissolution state and potential failure risk of the bridge plug is crucial for improving operation safety, reducing production costs, and optimizing operation efficiency.

[0004] The limitations of the prior art include at least the following problems. In the operation state monitoring of the soluble bridge plug, the prior art generally lacks structured processing and intelligent early warning mechanism for continuous dynamic data in the dissolution process. Most schemes only rely on point parameter reading or downhole intervention feedback in the dissolution result stage, making it difficult to capture and analyze early abnormal evolution trends in the dissolution behavior, thus causing delay in failure identification and missing the best window for failure intervention. Especially in complex well conditions, there is a high coupling between the dissolution state of the bridge plug and environmental disturbances. If the dynamic behavior characteristics cannot be sensed in time, the bridge plug may be stranded, the unblocking may be incomplete, or it may be misjudged as normal, increasing the risk and difficulty of intervention on site. SUMMARY

[0005] To overcome the shortcomings of the prior art, the present application provides a soluble bridge plug failure early warning method and system, which solves the problem that the prior art cannot perform multi-dimensional behavior sensing based on continuous time series data in the dissolution process of the soluble bridge plug, resulting in lag in identifying the failure trend of the bridge plug, lag in the early warning mechanism, and delay in the intervention timing.

[0006] To achieve the above object, the present application is implemented by the following technical solutions: A soluble bridge plug fault early warning method, comprising the following steps: based on a set time period, acquiring soluble bridge plug dissolution state time series data and dissolution environment time series data; based on a pre-trained time series feature perception model, performing dynamic evolution analysis on the soluble bridge plug dissolution state time series data and dissolution environment time series data, and extracting a state evolution feature set and an environment evolution feature set respectively; based on the state evolution feature set and the environment evolution feature set, generating a soluble bridge plug state evolution feature index and an environment evolution feature index; based on the state evolution feature index and the environment evolution feature index, judging whether the soluble bridge plug has a fault risk, and sending a fault alarm when there is a fault risk.

[0007] Further, the time series feature perception model includes a dissolution state evolution perception model and a dissolution environment evolution perception model, the dissolution state evolution perception model includes a state data processing layer, a behavior feature extraction layer, a local change identification layer, a feature result conversion layer and a state feature output layer, and the dissolution environment evolution perception model includes an environment data processing layer, a trend feature extraction layer, an interval fluctuation identification layer, a trend feature conversion layer and an environment feature output layer.

[0008] Further, the dissolution state time series data includes surface resistance values, dissolution rate values and acoustic emission amplitude values at a plurality of time points, and the state evolution feature set includes resistance sudden drop behavior frequency values, maximum resistance sudden drop amplitudes, maximum dissolution rate values, dissolution rate fluctuation amplitude values and high-energy acoustic emission point proportions.

[0009] Further, the specific steps of extracting the state evolution feature set are as follows: in the state data processing layer of the dissolution state evolution perception model, the dissolution state time series data of the soluble bridge plug is standardized to obtain a standardized dissolution state time series input sequence; in the behavior feature extraction layer of the dissolution state evolution perception model, the standardized dissolution state time series input sequence is analyzed by sliding window structure and behavior mode segmentation to extract local behavior segments such as sudden drop segments, rapid rise segments and high-energy excitation segments, and a behavior segment set is obtained; in the local change identification layer of the dissolution state evolution perception model, the behavior segment set is subjected to segment resistance drop amplitude calculation, dissolution rate extreme value search and high-amplitude acoustic emission segment identification to obtain an original state change index set; in the feature result conversion layer of the dissolution state evolution perception model, the original state change index set is subjected to unit time frequency statistics, maximum amplitude extraction, fluctuation amplitude calculation and proportional mapping to generate a structured state evolution feature; and in the state feature output layer of the dissolution state evolution perception model, the structured state evolution feature is sequentially arranged and field encapsulated to finally output the state evolution feature set.

[0010] Further, the specific steps of generating the state evolution characteristic index of the dissolvable bridge plug are as follows: reading the resistance jump behavior frequency value, the maximum resistance jump amplitude, the maximum dissolution rate value, the dissolution rate fluctuation amplitude value, and the high-energy acoustic emission point proportion of the dissolvable bridge plug; and performing weighted analysis on the resistance jump behavior frequency value, the maximum resistance jump amplitude, the maximum dissolution rate value, the dissolution rate fluctuation amplitude value, and the high-energy acoustic emission point proportion of the dissolvable bridge plug to obtain the state evolution characteristic index of the dissolvable bridge plug.

[0011] Further, the dissolution environment time sequence data includes local well temperature values, local well pressure values, dissolution product viscosity values, and fluid disturbance intensities at a plurality of time points, and the environment evolution characteristic set includes a minimum well temperature value, a well pressure fluctuation amplitude value, a viscosity average rising slope, and a disturbance fluctuation proportion.

[0012] Further, the specific steps of extracting the environment evolution characteristic set are as follows: in the environment data processing layer of the dissolution environment trend perception model, performing standardization processing on the dissolution environment time sequence data of the dissolvable bridge plug to obtain a standardized dissolution environment time sequence input sequence; in the trend feature extraction layer, performing trend curve fitting and multi-segment structure analysis on the standardized dissolution environment time sequence input sequence to extract a temperature down-dip segment, a pressure disturbance peak segment, a viscosity rising segment, and a disturbance flicker segment, and obtain a trend structure segment set; in the interval fluctuation identification layer, performing temperature minimum point extraction, pressure fluctuation amplitude calculation, viscosity rising rate regression, and disturbance fluctuation proportion statistics on the trend structure segment set to obtain an original trend change index set; in the trend feature conversion layer, performing normalization mapping and feature fusion on the original trend change index set to construct a standardized environment trend feature vector; and in the environment feature output layer, performing format construction and field output on the standardized environment trend feature vector to finally generate the environment evolution characteristic set.

[0013] Further, the specific steps of generating the environment evolution characteristic index of the dissolvable bridge plug are as follows: reading the minimum well temperature value, the well pressure fluctuation amplitude value, the viscosity average rising slope, and the disturbance fluctuation proportion of the dissolvable bridge plug; and performing weighted analysis on the minimum well temperature value, the well pressure fluctuation amplitude value, the viscosity average rising slope, and the disturbance fluctuation proportion of the dissolvable bridge plug to obtain the environment evolution characteristic index of the dissolvable bridge plug.

[0014] Further, the specific steps of judging whether the dissolvable bridge plug has a fault risk are as follows: performing judgment analysis on the state evolution characteristic index and the environment evolution characteristic index of the dissolvable bridge plug with respect to a preset fault interval set, and the fault interval set includes a state evolution characteristic fault interval and an environment evolution characteristic fault interval; and when the state evolution characteristic index and the environment evolution characteristic index are within the preset fault interval set, the dissolvable bridge plug is considered to have a fault risk.

[0015] The application discloses a soluble bridge plug fault early warning system, which comprises a dissolving data acquisition unit, a data evolution analysis unit, an evolution feature analysis unit and a fault judgment and early warning unit.

[0016] The application has the following beneficial effects:

[0017] (1) The soluble bridge plug fault early warning method constructs a complete process of continuous monitoring, feature extraction and dynamic analysis around the state data and environment data of the bridge plug in the dissolving process. By acquiring the data of resistance, dissolving rate, acoustic emission and the like at a plurality of time points, in combination with environment parameters such as well temperature, well pressure, liquid viscosity and disturbance intensity, whether the dissolving behavior deviates from the normal track can be sensed in time. When the features such as a sudden drop in resistance, an increase in frequency, a sharp fluctuation in rate or a concentration of acoustic emission energy appear, the potential dissolving abnormality can be identified through the model. Compared with the traditional mode of diagnosing after the unblocking fails, the method can give an early warning, avoids the operation delay or even the safety risk caused by the soluble bridge plug due to retention, partial dissolving or unblocking failure, and has significant application effectiveness and reliability guarantee in the high-density downhole operation scene.

[0018] (2) The soluble bridge plug fault early warning method performs domain processing on the state of the bridge plug and the surrounding environment data, respectively extracts key evolution features, dynamically analyzes the dissolving state time series data and the environment time series data through the model, extracts feature indexes such as the frequency of sudden drop in resistance, the fluctuation amplitude of dissolving rate and the disturbance amplitude of well pressure, which can represent the current operation trend, and then constructs the state evolution feature index and the environment evolution feature index based on the features by using the weighted analysis mode, so as to more clearly distinguish the intrinsic fault and the environmental influence. Compared with the single-dimensional judgment mode, the method not only improves the accuracy of fault identification, but also enhances the adaptability to different well conditions.

[0019] (3), the soluble bridge plug fault early warning system, through the modular design, the dissolution data acquisition, data evolution analysis, evolution characteristic analysis and fault judgment early warning function unit are independently divided, so that each unit can work independently while synergistically, ensure the efficient operation of the system, each functional unit can be customized and adjusted according to the actual demand of the field, has strong flexibility and scalability, for example, the dissolution data acquisition unit supports different types of dissolution state and environmental data input, can adapt to the access of various sensors and detection equipment, the data evolution analysis unit and the evolution characteristic analysis unit can continuously improve the early warning effect through model optimization and training, adapt to the change of complex well site environment, the fault judgment early warning unit can judge whether there is a fault risk in the first time through comprehensive analysis of two indexes, timely alarm, provide clear and reliable decision basis for the operator, the modular architecture of the system makes it easy to deploy in different well sites and different application scenarios, and provides rich operation space and efficient support platform for possible future technical upgrade or function expansion, thereby enhancing the sustainability and market competitiveness of the system.

[0020] Of course, implementing any product of the present application does not necessarily require all the advantages described above to be achieved simultaneously. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flow chart of a soluble bridge plug fault early warning method of the present application.

[0022] Figure 2 A flow chart of the specific steps of extracting the environment evolution feature set in the soluble bridge plug fault early warning method of the present application.

[0023] Figure 3 A block diagram of a soluble bridge plug fault early warning system of the present application. DETAILED DESCRIPTION

[0024] Please refer to Figure 1 The embodiment of the present application provides a technical solution: a soluble bridge plug fault early warning method, comprising the following steps: based on a set time period, acquiring the dissolution state time series data and the dissolution environment time series data of the soluble bridge plug; based on the pre-trained time series feature perception model, performing dynamic evolution analysis on the dissolution state time series data and the dissolution environment time series data of the soluble bridge plug, and extracting the state evolution feature set and the environment evolution feature set respectively; based on the state evolution feature set and the environment evolution feature set, generating the state evolution feature index and the environment evolution feature index of the soluble bridge plug; based on the state evolution feature index and the environment evolution feature index, judging whether the soluble bridge plug has a fault risk, and sending a fault alarm to the relevant staff when there is a fault risk.

[0025] The time sequence feature perception model includes a dissolution state evolution perception model and a dissolution environment evolution perception model. The dissolution state evolution perception model includes a state data processing layer, a behavior feature extraction layer, a local change identification layer, a feature result conversion layer, and a state feature output layer. The dissolution environment evolution perception model includes an environment data processing layer, a trend feature extraction layer, an interval fluctuation identification layer, a trend feature conversion layer, and an environment feature output layer.

[0026] The pre-training step of the dissolution state evolution perception model is as follows:

[0027] At the beginning of pre-training, a large amount of dissolution state time sequence data is collected, including time sequence data of physical quantities such as surface resistance value, dissolution rate value, and acoustic emission amplitude value. These data should be time-stamped and consistent within the same time window. Next, discrete smoothing and noise removal techniques are used to process the original data. Specifically, three-point sliding average method is used to smooth each data sequence, effectively removing high-frequency noise. Then, interpolation techniques are used to fill in any missing data, ensuring data continuity. Finally, in order to eliminate the dimensional differences of different physical quantities, z-score standardization processing is needed for each type of data, making the mean value 0 and the standard deviation 1, so that all time sequence data are within the same scale range, facilitating subsequent model training.

[0028] After completing data preprocessing, the next step is feature extraction. In this stage, first, the time sequence data is divided into different time periods using the sliding window method, and each paragraph is analyzed in detail. Specifically, we need to identify and extract behavior patterns such as sudden drop, rapid rise, and high-energy excitation in the dissolution state. These behavior segments correspond to different dissolution state processes, such as resistance drop or rapid change in dissolution rate, which can reflect the key dynamic characteristics of the dissolution process. Then, by calculating the resistance drop amplitude, dissolution rate change, acoustic emission energy value fluctuation, and other local features, these behavior segments are converted into specific physical quantities, providing important input data for subsequent model training.

[0029] After feature extraction is complete, the next step is to input the processed data into a pre-trained neural network model for training. Time series processing models such as LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit) are used to capture long-term dependencies and change patterns in time series data. By setting an appropriate loss function (such as Mean Squared Error (MSE)), the training effect of the model is evaluated, and the weight parameters in the network are optimized. During training, a large amount of labeled time series data is required, and the goal of the model is to automatically extract the corresponding state evolution feature set (such as resistance drop behavior frequency value, dissolution rate fluctuation amplitude value, etc.) from the input dissolution state data. Through cross-validation (K-fold validation) method to evaluate the model, and according to the verification result to adjust and optimize, ensure that the trained model has stable performance on different data sets and has good generalization ability.

[0030] The pre-training steps of the dissolution environment evolution perception model are as follows:

[0031] First, collect enough environmental time series data, including local well temperature, local well pressure, dissolution product viscosity, and fluid disturbance intensity. These data need to ensure time consistency and continuity, and strict data preprocessing is required. First, use smoothing algorithms (such as Gaussian filtering or low-pass filtering) to remove noise from the original data, then use interpolation techniques to fill in any missing values to avoid affecting subsequent analysis. Finally, to ensure that each environmental feature is trained on the same scale, all environmental data needs to be standardized using z-score to ensure that all input data can be learned within the same data distribution range.

[0032] After data preprocessing is complete, enter the feature extraction stage. The main task of this stage is to extract trend structures and fluctuation features from the time series data of the dissolution environment. First, model the environmental data through trend fitting (such as polynomial regression or local regression) to extract trend change segments such as temperature down segment, pressure disturbance peak segment, and viscosity rise segment. Next, calculate the fluctuation amplitude, change rate, and fluctuation proportion of each segment through fluctuation analysis. These fluctuation indicators can effectively reflect the influence of various disturbance factors in the dissolution environment on the dissolution rate, serving as important features for model learning. Especially in the pressure disturbance peak segment and viscosity change segment, the extracted fluctuation information can provide important basis for subsequent fault warning.

[0033] After feature extraction is completed, the processed environmental data is input into an advanced time series learning model such as a time convolutional neural network (TCN) or a Transformer for training. The training goal of the network is to automatically learn environmental evolution features related to the dissolution process (such as minimum well temperature value, well pressure fluctuation amplitude, etc.) from the input environmental time series data. By selecting a suitable loss function (such as mean square error (MSE)) for model optimization, the network can accurately capture long-term trends and local fluctuations in environmental data. A large amount of labeled time series data is used during training, combined with cross-validation methods to ensure the generalization ability of the model, and model hyperparameters (such as learning rate, batch size, network layer number, etc.) are adjusted according to evaluation results, ultimately obtaining a trained environmental evolution perception model with high prediction accuracy.

[0034] Specifically, the dissolution state time series data includes surface resistance values, dissolution rate values, and acoustic emission amplitude values at several time points, and the state evolution feature set includes resistance drop behavior frequency value, maximum resistance drop amplitude, maximum dissolution rate value, dissolution rate fluctuation amplitude value, and high-energy acoustic emission point proportion.

[0035] The surface resistance value refers to the change in the surface conductive ability of the bridge plug material during the dissolution process, which can be measured and obtained by a micro-resistance measurement probe arranged on the surface of the bridge plug.

[0036] The dissolution rate value refers to the change rate of the mass or volume of the bridge plug per unit time, which is used to depict the overall dissolution progress, and can be calculated by combining real-time mass estimation and volume change analysis. The required raw data can be obtained by a downhole ultrasonic ranging device and a differential pressure sensor.

[0037] The acoustic emission amplitude value refers to the intensity of the acoustic wave signal generated by the material rupture, micro-crack propagation, or stress concentration release during the dissolution process of the bridge plug, which can be measured and obtained by a piezoelectric acoustic emission sensor arranged near the well section.

[0038] The specific steps for extracting the state evolution feature set are as follows:

[0039] In the state data processing layer of the dissolution state evolution perception model, the dissolution state time series data of the soluble bridge plug is standardized (missing data completion, outlier elimination, and normalization processing) to obtain a standardized dissolution state time series input sequence. Specifically, by preprocessing the three types of original data of bridge plug surface resistance value, dissolution rate value, and acoustic emission amplitude value, first, linear interpolation is used to continuously complete the time discontinuous data in the resistance and rate curves, then 3σ statistical criterion is used to eliminate outliers in the acoustic emission data, and then each sequence is mapped to the [0, 1] interval to form a normalized format, thereby constructing a standardized dissolution state time series input sequence.

[0040] In the behavior feature extraction layer of the dissolution state evolution perception model, the standardized dissolution state time sequence input sequence is parsed with a sliding window structure and segmented into behavior patterns, local behavior segments such as sudden drop segments, rapid rise segments and high-energy excitation segments are extracted, and a behavior segment set is obtained. Specifically, a fixed-length sliding window of 5 seconds is used to traverse the state input sequence, and in each window, the resistance change curve is analyzed. When the resistance drop in the window exceeds the set threshold (such as the normalized drop amplitude > 0.3), it is marked as a resistance sudden drop segment. When the dissolution rate is continuously positive in the window and the average rise rate is greater than twice the global average, it is marked as a rapid rise segment. If the acoustic emission amplitude continuously appears three amplitude points greater than 0.8 in any window, the segment is identified as a high-energy excitation segment. The above three types of segments constitute the local behavior segment set;

[0041] In the local change recognition layer of the dissolution state evolution perception model, the segment resistance drop amplitude calculation, dissolution rate extreme value search and high-amplitude acoustic emission segment identification are performed on the behavior segment set to obtain the original state change indicator set. Specifically, the maximum initial and final point resistance difference in all sudden drop segments is selected as the "maximum resistance sudden drop amplitude", the single-point maximum rate value in the rate rise segment is extracted as the "maximum dissolution rate value", and the ratio of the number of time points with acoustic emission amplitude greater than 0.8 to the total number of sampling points is calculated as the "high-energy acoustic emission point ratio". In addition, the number of all sudden drop segments is counted to provide a basis for the derivation of subsequent frequency indicators, and finally the original state change indicator set is formed;

[0042] In the feature result conversion layer of the dissolution state evolution perception model, the unit time frequency statistics, maximum amplitude extraction, fluctuation amplitude calculation and proportional mapping are performed on the original state change indicator set to generate structured state evolution features. Specifically, the sudden drop behavior frequency value is obtained by dividing the number of sudden drop behaviors by the total data collection time, which measures the stability of the dissolution process in terms of the number of behaviors per unit time. The difference between the maximum and minimum values in the dissolution rate sequence is defined as the "dissolution rate fluctuation amplitude value", which reflects the unevenness of the rate change. Other indicators such as the maximum sudden drop amplitude and the high-energy acoustic emission point ratio are constructed into a vector field with three significant digits, forming a structured and unitary state evolution feature set;

[0043] In the state feature output layer of the dissolution state evolution perception model, the structured state evolution features are arranged in order and encapsulated into fields, and finally the state evolution feature set is output. Specifically, the "resistance sudden drop behavior frequency value, maximum resistance sudden drop amplitude, maximum dissolution rate value, dissolution rate fluctuation amplitude value" and "high-energy acoustic emission point ratio" are arranged in the predetermined field order, and each item is encapsulated into the feature output structure as a floating-point number, which is the formal output of the state evolution feature set.

[0044] In the embodiment, the differences between data are eliminated by standardization processing, ensuring the consistency of time series data in different dissolution states. This process not only includes data completion and outlier removal, but also ensures the uniformity of data in the same dimension through normalization processing, effectively improving the reliability of subsequent analysis. In the behavior feature extraction stage, sliding window structure is used to accurately analyze the resistance change, dissolution rate and acoustic emission amplitude, which can timely capture the key sudden drop, rapid rise and high-energy excitation behavior in the dissolution process. Through the extraction of these behavior segments, the model can identify potential fault signals and provide early warning of abnormalities in the dissolution process. Further, in the feature conversion and output stage, through the analysis and structured processing of key indicators such as sudden drop behavior frequency and dissolution rate fluctuation amplitude, a set of state evolution features with practical significance is provided. These features can effectively help the system determine whether the dissolution state of the bridge plug enters the fault risk interval, providing a scientific basis for subsequent fault warning and emergency treatment. This method improves the accuracy and response speed of the warning system, can timely and accurately predict the potential failure of the soluble bridge plug, and ensures the safety and efficiency of downhole operations.

[0045] Specifically, the specific steps of generating the state evolution feature index of the soluble bridge plug are as follows: reading the resistance sudden drop behavior frequency value, maximum resistance sudden drop amplitude, maximum dissolution rate value, dissolution rate fluctuation amplitude value, and high-energy acoustic emission point proportion of the soluble bridge plug; performing weighted analysis on the resistance sudden drop behavior frequency value, maximum resistance sudden drop amplitude, maximum dissolution rate value, dissolution rate fluctuation amplitude value, and high-energy acoustic emission point proportion of the soluble bridge plug to obtain the state evolution feature index of the soluble bridge plug.

[0046] In the embodiment, through the weighted analysis of key characteristic values such as resistance sudden drop behavior frequency, maximum resistance sudden drop amplitude, maximum dissolution rate, dissolution rate fluctuation amplitude, and high-energy acoustic emission point proportion, various dynamic behaviors in the bridge plug dissolution process can be comprehensively reflected, thereby accurately evaluating the state. By integrating these features through weighting, the representativeness of the features is enhanced, and the sensitivity and accuracy of fault warning are improved. Ultimately, the state evolution feature index provides a reliable basis for subsequent fault risk judgment, which can help operators identify potential fault risks in time, ensuring the safety and efficiency of downhole operations.

[0047] Specifically, the dissolution environment time series data includes local well temperature values, local well pressure values, dissolution product viscosity values, and fluid disturbance intensities at several time points. The environment evolution feature set includes the minimum well temperature value, well pressure fluctuation amplitude value, viscosity average rising slope, and disturbance fluctuation proportion.

[0048] The local well temperature value refers to the temperature of the liquid in the environment of the well section where the soluble bridge plug is located, is used to analyze the influence of the current thermal environment on the dissolution reaction, and can be measured and obtained by a thermocouple temperature sensor arranged near the bridge plug.

[0049] The local well pressure value refers to the static pressure or dynamic pressure in the well section where the bridge plug is actually in a dissolved state, is used to judge the effect of the environmental pressure on the bridge plug structure, and can be measured and obtained by a piezoelectric pressure sensor arranged in the well.

[0050] The dissolution product viscosity value refers to the degree of viscosity of the released product in the well fluid after the bridge plug is dissolved, is used to evaluate the product flowability and dissolution residue risk, and can be measured and obtained by an online rotary viscometer arranged in the flow path.

[0051] The fluid disturbance intensity value refers to the local disturbance degree of the liquid around the bridge plug, including shear wave fluctuation, micro-vortex or flow rate variation, and can be measured and obtained by a multi-point micro-flow probe and shear stress sensor arranged in the well section.

[0052] As shown in Figure 2 The specific steps of extracting the environment evolution feature set are as follows:

[0053] In the environment data processing layer of the dissolution environment trend perception model, the dissolution environment time series data of the soluble bridge plug is standardized to obtain a standardized dissolution environment time series input sequence, which is specifically: by sequentially preprocessing the four types of environment data of original well temperature, well pressure, viscosity and disturbance intensity, first, the three-point moving average method is used to smooth each time series to suppress high-frequency noise, then the upper and lower quantile point determination rule is used to correct the maximum and minimum values to avoid trend deviation caused by short-time interference, and finally, the z-score standardization processing is performed on each physical quantity to make the mean value 0 and the variance 1, and then the environment input sequence with standardized format and clear data form is obtained;

[0054] In the trend feature extraction layer, the normalized dissolution environment time series input sequence is fitted with a trend curve and analyzed for multi-section structure to extract temperature down-dip sections, pressure disturbance peak sections, viscosity rise sections, and disturbance flicker sections, to obtain a trend structure section set. Specifically, local linear fitting and sliding window derivative analysis methods are used on well temperature, well pressure, viscosity, and disturbance sequences. In the well temperature sequence, regions with negative slope and continuous drop of more than three points are identified as temperature down-dip sections. In the well pressure sequence, sections with peak value mutation points and average value difference of more than twice the standard deviation are detected to form pressure disturbance peak sections. First-order derivative regression analysis is performed on viscosity data, and sections with continuous slope greater than zero and uninterrupted rise of more than 10 seconds are defined as viscosity rise sections. For fluid disturbance sequences, sections with disturbance wave peak number per unit time exceeding a set threshold (e.g., 3 times / 10 seconds) are defined as disturbance flicker sections. All the above marked region sets together form the trend structure section set.

[0055] In the interval fluctuation recognition layer, the trend structure section set is subjected to temperature minimum point extraction, pressure fluctuation amplitude calculation, viscosity rise rate regression, and disturbance fluctuation proportion statistics to obtain an original trend change index set. Specifically, the minimum value points of the temperature curve are extracted from all temperature down-dip sections and recorded as "minimum well temperature values". The amplitudes (maximum and minimum value difference) of each section in the pressure disturbance peak section are calculated, and the maximum amplitude is selected as the "well pressure fluctuation amplitude value". Linear regression curves are fitted for all viscosity rise sections, and their average slopes are calculated as "viscosity average rise slope". The proportion of the time length of high-frequency disturbance behavior to the total time is calculated in the disturbance flicker section and recorded as "disturbance fluctuation proportion".

[0056] In the trend feature conversion layer, the original trend change index set is subjected to normalization mapping and feature fusion to construct a standardized environment trend feature vector. Specifically, the four original trend indexes are linearly normalized to map their values to the [0, 1] interval, ensuring comparability of the indexes on the same scale. The normalized results are then packaged into a unified format environment trend feature vector, including the minimum temperature point value, the maximum pressure amplitude normalized value, the viscosity average rise slope standardized value, and the disturbance fluctuation proportion ratio expression.

[0057] In the environment feature output layer, the standardized environment trend feature vector is subjected to format construction and field output to finally generate an environment evolution feature set. Specifically, the four standardized environment trend features are packaged into field-type structures in the preset order, named "minimum well temperature value", "well pressure fluctuation amplitude value", "viscosity average rise slope", and "disturbance fluctuation proportion". All fields are presented in floating-point number format and stored in a structured data set as the formal output of the environment evolution feature set.

[0058] In this embodiment, by standardizing the local well temperature, well pressure, dissolved product viscosity and fluid disturbance intensity data, the noise and outliers in the original data are eliminated, ensuring the quality and consistency of the data. On this basis, the trend analysis of various environmental data is carried out through the trend feature extraction layer, which can identify the typical patterns of temperature, pressure, viscosity and disturbance changes, providing valuable structured information for in-depth analysis. In addition, the interval fluctuation identification layer further refines the capture of dynamic environmental changes by extracting key fluctuation indicators such as the minimum well temperature, maximum well pressure fluctuation amplitude and viscosity rise rate. These features are normalized through the feature conversion layer to ensure the unified scale of each indicator, facilitating subsequent weighted analysis and fusion. Finally, the generated standardized environmental evolution feature set can accurately reflect the change trend of the downhole environment, providing precise data support for fault risk assessment. Through this process, not only the comparability of environmental data is improved, but also the precision and reliability of soluble bridge plug fault prediction are effectively enhanced, thereby ensuring the safety and efficiency of downhole operations.

[0059] Specifically, the specific steps of generating the environmental evolution feature index of the soluble bridge plug are as follows: reading the minimum well temperature value, well pressure fluctuation amplitude value, viscosity average rise slope and disturbance fluctuation proportion of the soluble bridge plug; performing weighted analysis on the minimum well temperature value, well pressure fluctuation amplitude value, viscosity average rise slope and disturbance fluctuation proportion of the soluble bridge plug to obtain the environmental evolution feature index of the soluble bridge plug.

[0060] In this embodiment, by reading the minimum well temperature, well pressure fluctuation amplitude, viscosity rise slope and disturbance fluctuation proportion, and performing weighted analysis, the influence of environmental changes on the soluble bridge plug dissolution process can be more comprehensively reflected. The weighted analysis of each indicator considers its importance in the overall dissolution process, ensuring that the final environmental evolution feature index can accurately capture the comprehensive effect of environmental changes. Such a feature index provides a more accurate and reliable basis for subsequent fault judgment and warning, helping to improve the monitoring accuracy and response speed of the dissolution process, and ensuring the safety and efficiency of downhole operations.

[0061] Specifically, the specific steps of judging whether the soluble bridge plug has a fault risk are as follows: the state evolution feature index and the environmental evolution feature index of the soluble bridge plug are respectively analyzed and judged by the preset fault interval set, and the fault interval set includes a state evolution feature fault interval and an environmental evolution feature fault interval; when the state evolution feature index and the environmental evolution feature index are within the preset fault interval set, it is considered that the soluble bridge plug has a fault risk.

[0062] In the embodiment, the accurate prediction of the soluble bridge plug failure risk can be realized by comparing the state evolution feature index and the environment evolution feature index with the preset failure interval, the key risk points in the dissolution process can be accurately identified by setting the clear failure interval, when the state and environment feature values are in the failure interval, the system timely sends an alarm, the judgment method based on data driving not only improves the identification accuracy of the failure risk, but also can respond to various environmental changes and operation conditions in real time, ensures the safety in the dissolution process, compared with the limitations of the traditional method relying on single parameter or experience judgment, the multi-dimensional and multi-feature analysis method is more comprehensive, can early warn potential failure, avoids equipment damage and operation delay, and improves the stability and efficiency of the operation.

[0063] Please refer to Figure 3 The embodiment of the present application provides a technical scheme: a soluble bridge plug failure warning system, comprising: a dissolution data acquisition unit, configured to acquire dissolution state time series data and dissolution environment time series data of the soluble bridge plug based on a set time period; a data evolution analysis unit, configured to perform dynamic evolution analysis on the dissolution state time series data and the dissolution environment time series data of the soluble bridge plug based on a pre-trained time series feature perception model, and extract a state evolution feature set and an environment evolution feature set respectively; an evolution feature analysis unit, configured to generate a state evolution feature index and an environment evolution feature index of the soluble bridge plug based on the state evolution feature set and the environment evolution feature set; and a failure judgment and warning unit, configured to judge whether the soluble bridge plug has a failure risk based on the state evolution feature index and the environment evolution feature index, and send a failure alarm to relevant staff when the failure risk exists.

[0064] Although the preferred embodiments of the present application have been described, those skilled in the art can make further changes and modifications to these embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including all the preferred embodiments and all the changes and modifications falling within the scope of the present application.

[0065] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method for failure warning of dissolvable bridge plug, characterized in that, The method comprises the following steps: Based on a set time period, obtain the dissolving state time series data and the dissolving environment time series data of the dissolvable bridge plug, the dissolving state time series data comprising surface resistance values at several time points obtained by a micro-resistance measuring probe arranged on the surface of the bridge plug, dissolving rate values at several time points obtained by a downhole ultrasonic ranging device in cooperation with a differential pressure sensor, and acoustic emission amplitude values at several time points obtained by a piezoelectric acoustic emission sensor arranged near the well section, the dissolving environment time series data comprising local well temperature values, local well pressure values, dissolving product viscosity values and fluid disturbance intensity at several time points; Based on the pre-trained time series feature perception model, the dissolving state time series data and the dissolving environment time series data of the dissolvable bridge plug are dynamically evolved and analyzed to extract a state evolution feature set and an environment evolution feature set, respectively, the state evolution feature set comprising resistance sudden drop behavior frequency value, maximum resistance sudden drop amplitude, maximum dissolving rate value, dissolving rate fluctuation amplitude value and high-energy acoustic emission point proportion, and the environment evolution feature set comprising minimum well temperature value, well pressure fluctuation amplitude value, viscosity average rising slope and disturbance fluctuation proportion; The time series feature perception model comprises a dissolving state evolution perception model and a dissolving environment evolution perception model, the dissolving state evolution perception model comprising a state data processing layer, a behavior feature extraction layer, a local change identification layer, a feature result conversion layer and a state feature output layer, and the dissolving environment evolution perception model comprising an environment data processing layer, a trend feature extraction layer, an interval fluctuation identification layer, a trend feature conversion layer and an environment feature output layer; The specific steps of extracting the state evolution feature set are as follows: In the state data processing layer of the dissolving state evolution perception model, the dissolving state time series data of the dissolvable bridge plug is subjected to standardization processing to obtain a standardized dissolving state time series input sequence; In the behavior feature extraction layer of the dissolving state evolution perception model, a sliding window with a fixed length of 5 seconds is used to analyze the standardized dissolving state time series input sequence, and a local behavior segment is extracted based on a preset judgment rule to form a behavior segment set, wherein the judgment rule is that when the resistance normalized value in the sliding window drops by more than 0.3, the window is marked as a sudden drop segment; when the acoustic emission amplitude value in the sliding window continuously appears three amplitude points greater than 0.8, the window is marked as a high-energy excitation segment; when the dissolving rate derivative value in the sliding window is continuously positive and the average rising rate is greater than twice the global average, the window is marked as a rapid rising segment; In the local change identification layer of the dissolving state evolution perception model, the behavior segment set is subjected to segment resistance drop amplitude calculation, dissolving rate extreme value search and high-amplitude acoustic emission segment identification to obtain an original state change index set; In the feature result conversion layer of the dissolving state evolution perception model, the original state change index set is subjected to unit time frequency statistics, maximum amplitude extraction, fluctuation amplitude calculation and proportion mapping to generate a structured state evolution feature; In the state feature output layer of the dissolving state evolution perception model, the structured state evolution feature is subjected to ordered arrangement and field packaging to finally output the state evolution feature set; The specific steps of extracting the environment evolution feature set are as follows: In the environment data processing layer of the dissolving environment trend perception model, the dissolving environment time series data of the dissolvable bridge plug is standardized to obtain a standardized dissolving environment time series input sequence; In the trend feature extraction layer, the standardized dissolving environment time series input sequence is subjected to trend curve fitting and multi-section structure analysis to extract a temperature down-dip section, a pressure disturbance peak section, a viscosity rise section and a disturbance frequency flashing section to form a trend structure fragment set; wherein, the temperature down-dip section is a region in the well temperature sequence with a negative slope and a continuous drop of more than three sampling points, the pressure disturbance peak section is a section in the well pressure sequence with a peak value mutation point and a difference of more than twice the standard deviation from the average value on both sides, the viscosity rise section is a section in the viscosity data with a first derivative slope continuously greater than zero and an uninterrupted rise of more than 10 seconds, and the disturbance frequency flashing section is a section in the fluid disturbance sequence with a number of disturbance wave peaks per unit time exceeding a set threshold; In the interval fluctuation recognition layer, the trend structure fragment set is subjected to temperature minimum point extraction, pressure fluctuation amplitude calculation, viscosity rise rate regression and disturbance fluctuation proportion statistics to obtain an original trend change index set; In the trend feature conversion layer, the original trend change index set is subjected to normalization mapping and feature fusion to construct a standardized environment trend feature vector; In the environment feature output layer, the standardized environment trend feature vector is subjected to format construction and field output to finally generate an environment evolution feature set; Based on the state evolution feature set and the environment evolution feature set, a state evolution feature index and an environment evolution feature index of the dissolvable bridge plug are generated; Based on the state evolution feature index and the environment evolution feature index, it is determined whether the dissolvable bridge plug has a failure risk, and when there is a failure risk, a failure alarm is sent, and the specific steps are as follows: The state evolution feature index and the environment evolution feature index of the dissolvable bridge plug are respectively subjected to judgment analysis with a preset failure interval set, and the failure interval set includes a state evolution feature failure interval and an environment evolution feature failure interval; When the state evolution feature index and the environment evolution feature index are within the preset failure interval set, it is considered that the dissolvable bridge plug has a failure risk.

2. The method of claim 1, wherein, The specific steps of generating the state evolution feature index of the dissolvable bridge plug are as follows: Read the resistance drop behavior frequency value, the maximum resistance drop amplitude, the maximum dissolution rate value, the dissolution rate fluctuation amplitude value and the high-energy acoustic emission point proportion of the dissolvable bridge plug; The resistance drop behavior frequency value, the maximum resistance drop amplitude, the maximum dissolution rate value, the dissolution rate fluctuation amplitude value and the high-energy acoustic emission point proportion of the dissolvable bridge plug are subjected to weighted analysis to obtain the state evolution feature index of the dissolvable bridge plug.

3. The method of claim 1, wherein, The specific steps of generating the environment evolution feature index of the dissolvable bridge plug are as follows: Read the minimum well temperature value, the well pressure fluctuation amplitude value, the viscosity average rise slope and the disturbance fluctuation proportion of the dissolvable bridge plug; The minimum well temperature value, the well pressure fluctuation amplitude value, the viscosity average rise slope and the disturbance fluctuation proportion of the dissolvable bridge plug are subjected to weighted analysis to obtain the environment evolution feature index of the dissolvable bridge plug.

4. A soluble bridge plug failure warning system applying the soluble bridge plug failure warning method of any one of claims 1-3, characterized in that, Comprise: The dissolution data acquisition unit is configured to acquire the dissolution state time series data and the dissolution environment time series data of the dissolvable bridge plug based on a set time period. The data evolution analysis unit is configured to perform dynamic evolution analysis on the dissolution state time series data and the dissolution environment time series data of the dissolvable bridge plug based on the pre-trained time series feature perception model, and extract a state evolution feature set and an environment evolution feature set, respectively. The evolution feature analysis unit is configured to generate a state evolution feature index and an environment evolution feature index of the dissolvable bridge plug based on the state evolution feature set and the environment evolution feature set. The fault judgment and early warning unit is configured to judge whether the dissolvable bridge plug has a fault risk based on the state evolution feature index and the environment evolution feature index, and send a fault alarm when the fault risk exists.

Images