A method and system for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells

Abstract

The present invention relates to the technical field of data processing, and in particular to a method and system for monitoring and early warning of the flow conductivity of a simulated perforation channel of an oil and gas well. The method comprises: collecting data of various dimensions in a channel during a simulated perforation operation of an oil and gas well to obtain a data sequence of each dimension, obtaining a mutation point in the data sequence of each dimension through data analysis, wherein adjacent mutation points constitute a subsequence, calculating a correction value of an exponential smoothing parameter between subsequences according to a difference between a median value and two end values ​​of the subsequence, correcting a preset exponential smoothing parameter using the correction value to dynamically adjust the smoothing parameter, performing an exponential smoothing operation on all subsequences using the corrected smoothing parameter, judging whether the subsequence has an abnormality according to an exponential smoothing result, solving the problem that a traditional fixed smoothing coefficient cannot capture rapid changes in data in a timely manner, improving the accuracy of exponential smoothing operation prediction, and being able to accurately monitor the flow conductivity of the perforation channel.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for monitoring and early warning the flow conductivity of simulated perforation channels in oil and gas wells. Background Art

[0002] With the continuous growth of global energy demand, the effective development and utilization of oil and gas resources has become particularly important. The simulated perforation operation of oil and gas wells is a key technical link in the development of oil and gas fields. Its purpose is to enhance the conductivity of oil and gas wells through perforation technology, thereby increasing oil and gas production. During the perforation operation, the impact effect of the perforation moment, the dynamic entry of fluids, rock fracture and debris migration and other factors work together to cause changes in key parameters such as pressure, temperature, and fluid flow in the channel. By analyzing and processing these parameter data, the conductivity of the perforation channel can be evaluated.

[0003] In the process of analyzing and processing the data of simulated perforating operations in oil and gas wells to realize the monitoring and early warning of flow conductivity, it is usually analyzed and processed based on historical data, and abnormal early warning of future perforation flow conductivity is carried out. For example, a Chinese patent application document with the publication number CN117386350A discloses a method for intelligent monitoring of oil and gas well perforation. The method deploys a sensor network inside the oil and gas well, collects physical parameters related to perforation, records the vibration signal of seismic waves near the wellbore through the vibration sensor and analyzes it to determine whether the perforation of the oil and gas well is abnormal. In addition, through data partitioning and model training, a recurrent neural network model is constructed to calculate the prediction results of key parameters of the perforation process, fuse multi-source data, analyze a large amount of perforation data, find the correlation pattern between perforation quality and production capacity, and improve the perforation effect and production capacity. The method also uses a clustering algorithm to judge abnormalities. By using different colors on the scatter plot to represent abnormal samples, the distribution and trend of abnormal data are intuitively displayed, and the abnormal alarm mechanism is triggered for the abnormal situation detected.

[0004] Exponential smoothing is a traditional data prediction method that is often used to monitor data anomalies. For example, a Chinese patent application document with publication number CN106872663A discloses a gas outburst prediction and early warning method based on a big data platform, which includes using linear exponential smoothing to preprocess real-time monitoring data; determining the cycle of real-time monitoring data and detection and prevention data; taking the real-time monitoring data within a detection cycle as a data set; extracting the characteristic parameters of the monitoring data in each data set; combining the initial velocity of gas outburst and the maximum amount of drill cuttings in the borehole with the characteristic parameters of the monitoring data to form a gas outburst hazard sample; and finally using the BP neural network to predict gas outbursts and determine whether to issue an early warning.

[0005] However, the traditional exponential smoothing method usually predicts recent data by using fixed exponential smoothing parameters. When using the traditional exponential smoothing method to predict various parameter data in the process of simulating perforating operations in oil and gas wells, due to factors such as the sharp fluctuation of pressure in the pores, the sudden destruction of rock structure, the rapid change of fluid properties, and the interaction between fluid and pore wall, the collected data may change significantly in a short period of time, resulting in rapid fluctuations in data trends and directions. In this case, the fixed smoothing parameters may not be able to capture and accurately reflect the changing trend of the data in a timely manner, resulting in a prediction deviation that does not match the actual trend, thereby affecting the accuracy of evaluating the conductivity of the perforating pores. Summary of the invention

[0006] In order to solve the problem that in the data monitoring of simulated perforation operations in oil and gas wells, fixed smoothing parameters produce prediction deviations, thereby affecting the accuracy of evaluating the conductivity of perforation channels, the present invention proposes a monitoring and early warning method and system for the conductivity of simulated perforation channels in oil and gas wells, which specifically adopts the following scheme:

[0007] In one aspect, the present invention provides a method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells, comprising:

[0008] Collect data of various dimensions in the hole during the simulated perforation operation of oil and gas wells to obtain data sequences of various dimensions;

[0009] By data analysis, mutation points in the data sequence of each dimension are obtained, and adjacent mutation points form a subsequence, and the product of the median value of the subsequence and the difference between the two end values ​​is used as the numerator, and the difference between the two end values ​​is used as the denominator, and the value after normalization of the fraction is used as the correction value of the exponential smoothing parameter between the subsequences;

[0010] The preset exponential smoothing parameter is corrected by using the correction value, the corrected exponential smoothing parameter is positively correlated with the correction value, the exponential smoothing operation is performed on all subsequences by using the corrected smoothing parameter to obtain the predicted value of each data in the subsequence, and whether the subsequence is abnormal is judged according to the difference between the predicted value and the actual value, and if there is an abnormality, an early warning is issued.

[0011] The above technical solution obtains the mutation points in the data sequence of each dimension through data analysis, and then divides the data sequence into multiple subsequences. This method can better identify the key points of data change, lays the foundation for the subsequent adjustment of smoothing parameters, and forms a subsequence between adjacent mutation points. This division method helps to capture the rapid changes of data, so that the data change trend in each subsequence is relatively consistent, thereby improving the accuracy of the prediction model, and by dynamically adjusting the smoothing parameters, it solves the problem that the traditional fixed smoothing coefficient cannot capture the rapid changes of data in time. The smoothing parameters can be adjusted according to the amplitude of data changes, which improves the accuracy of exponential smoothing. After the exponential smoothing operation, by calculating the difference between the predicted value and the actual value, the abnormal data in the data sequence can be effectively identified, which helps to timely discover problems in perforating operations and improve the accuracy of evaluating the drainage capacity of the perforating channel. The early warning mechanism can help operators to discover and solve problems in time and avoid potential risks.

[0012] Furthermore, the mutation points in the data sequences of each dimension are obtained through data analysis, including:

[0013] Each data in the data sequence of each dimension and several adjacent data before and after it are taken as the local sequence of the data, and the number of data contained in the local sequence of each data is the same;

[0014] Evaluate the correlation between the local sequence of each data in each dimension and the local sequence of data at the same collection time in other dimensions by using statistical methods;

[0015] Calculate the mutation degree of each data in the data sequence of each dimension according to the slope difference between each data before and after the adjacent data and the correlation, wherein the mutation degree is positively correlated with the correlation and the slope difference;

[0016] The mutation points in the data sequence of each dimension are screened out according to the magnitude of the mutation degree.

[0017] The above technical solution takes each data and several adjacent data before and after it as a local sequence, and evaluates the correlation between the local sequence and the data at the same collection time in other dimensional data sequences. By analyzing the mutation degree of each data point and the correlation of the corresponding sequences of the data point in different dimensions, it can effectively deal with the problems of sharp fluctuations in data and rapid changes in trends. This method enhances the adaptability of the algorithm to data changes and improves the stability of prediction.

[0018] Furthermore, the use of the correction value to correct the preset exponential smoothing parameter is to multiply the correction value and the preset exponential smoothing parameter to obtain a value as the corrected exponential smoothing parameter.

[0019] Further, judging whether the subsequence is abnormal according to the difference between the predicted value and the actual value includes:

[0020] The sum of the mean values ​​of the prediction deviations of the data in the data sequence of each dimension within the time period corresponding to the subsequence is taken as the comprehensive abnormal performance degree of the subsequence; the prediction deviation is the absolute value of the difference between the predicted value and the actual value of the data;

[0021] Preset the warning value of the comprehensive abnormal performance degree;

[0022] If the comprehensive abnormal performance degree of the subsequence is greater than the warning value, the subsequence is abnormal; otherwise, the subsequence is not abnormal.

[0023] The above technical solution calculates the sum of the mean values ​​of the prediction deviations of the data in each dimension in the subsequence as the comprehensive abnormal performance degree. This method can evaluate the abnormality of the subsequence more comprehensively as a whole, thereby improving the accuracy of anomaly detection.

[0024] Furthermore, the correlation between the local sequences satisfies the following relationship:

[0025]

[0026] In the formula, For the In the data series of dimensions The local sequence of the data is In the data series of dimensions The correlation between the local sequences of data, The function is used for normalization, is the number of data contained in each of the two local sequences, , Respectively , In the data series of dimensions In the local sequence of data The value of the data, , Respectively , In the data series of dimensions The mean of all data in the local sequence of data, For the In the data series of dimensions All data in the local sequence of data and the In the data series of dimensions The standard deviation of the differences between all data in the local series of data.

[0027] The above technical solution helps to identify the synchronous change trend between data of different dimensions by calculating the correlation between local sequences, thereby enhancing the ability to detect abnormal data.

[0028] Furthermore, a method for screening out mutation points in the data sequence of each dimension according to the magnitude of the mutation degree is as follows: normalizing the mutation degrees of all data in the data sequence of each dimension, and taking data whose normalized mutation degree is greater than a preset mutation degree threshold as mutation points.

[0029] Furthermore, the mutation degree of the data is calculated based on the slope difference between each data and its adjacent data before and after it in the data sequence of each dimension, and the correlation:

[0030]

[0031] In the formula, For the In the data series of dimensions The degree of mutation of the data, , Respectively In the data series of dimensions The slope of a data point and the previous and next adjacent data points, For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The mean of the correlation between the local series of data, For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The standard deviation of the correlation between the local series of data, To take the absolute value sign.

[0032] In the above technical solution, the difference in slope reflects the The calculation of correlation reflects the degree of synchronous change between individual data and the previous and subsequent adjacent data. The combination of the two can enhance the sensitivity to changes in data trends.

[0033] Furthermore, the slope satisfies the following relationship:

[0034]

[0035] In the formula, For the In the data series of dimensions The slope of a data point and the previous adjacent data point, For the In the data series of dimensions The slope of a data point and the next adjacent data point, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The time when the data is collected.

[0036] The above technical solution calculates the slope and can sensitively capture the amplitude of data changes.

[0037] Furthermore, the dimensions include pressure, temperature, and fluid flow.

[0038] On the other hand, the present invention also provides a monitoring and early warning system for the conductivity of simulated perforation channels in oil and gas wells, the monitoring and early warning system comprising a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of any of the monitoring and early warning methods described.

[0039] The present invention has the following effects:

[0040] The present invention obtains mutation points in data sequences of various dimensions through data analysis, and forms a subsequence between adjacent mutation points. This division method helps to capture rapid changes in data and improves the accuracy of the prediction model. Then, the smoothing parameters are dynamically adjusted to solve the problem that traditional fixed smoothing coefficients cannot capture rapid changes in data in a timely manner, improve the accuracy of prediction, improve the accuracy of evaluating the conductivity of perforation channels, and avoid potential risks through an early warning mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0042] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0044] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0045] Reference Figure 1 A method for monitoring and early warning the conductivity of simulated perforation channels in oil and gas wells, comprising steps S1 to S4, which are specifically as follows:

[0046] S1: Collect data sequences of various dimensions in the hole during the simulated perforation operation of oil and gas wells.

[0047] The monitoring and early warning of the conductivity of the simulated perforation channel in oil and gas wells mainly include the following steps:

[0048] Data collection: A sensor network is deployed inside the oil and gas well to collect physical parameters related to perforation to monitor abnormal conditions of the perforation operation.

[0049] Data preprocessing: Clean and preprocess the collected raw data to remove noise and outliers to ensure data quality. Specifically, the real-time monitoring data is preprocessed using methods such as exponential smoothing for subsequent analysis. This embodiment dynamically adjusts the smoothing parameters of the exponential smoothing method, and uses the corrected smoothing parameters for prediction to improve the accuracy of the prediction.

[0050] Conductivity assessment: Based on the difference between the predicted value and the actual value, the conductivity of the perforation channel is assessed, that is, whether the data is abnormal. If the data is not abnormal, the conductivity of the perforation channel is strong. If the data is abnormal, the conductivity of the perforation channel is weak, indicating that there is a problem with the current perforation operation and adjustment or optimization is needed.

[0051] This step is mainly for data collection, through real-time collection of different dimensional data in the channel, including pressure, temperature and fluid flow, and the different dimensional data collection methods in the channel are as follows:

[0052] Pressure data collection: By installing a voltage transformer in the perforator for data collection and monitoring, the voltage transformer can convert the pressure changes in the hole into electrical signals, and then into measurable pressure values.

[0053] Temperature data collection: By installing a temperature transformer in the perforator for collection and monitoring, the temperature transformer can sense the temperature changes in the hole and convert them into electrical signals for temperature data collection.

[0054] Fluid flow data collection: By installing a flow transformer in the perforation hole for data collection and monitoring, the flow transformer can measure the fluid flow based on the signal generated by the fluid flow.

[0055] In this step, it is preset to collect pressure, temperature and fluid flow data once a second. Data of these three dimensions will be collected every second. The implementers can adjust it according to the actual situation to ensure that the frequency of data collection can meet the monitoring needs. The collected data of different dimensions will then be transmitted to the database for storage and subsequent analysis.

[0056] This step ensures the integrity of the data by real-time monitoring of data of different dimensions such as pressure, temperature and fluid flow in the channel during the simulated perforating operation of oil and gas wells, and adopting efficient collection equipment. The real-time collection of these data provides a solid foundation for subsequent data analysis, model building and decision support.

[0057] S2: Obtain mutation points in data sequences of various dimensions through data analysis.

[0058] Specifically include:

[0059] S21: Set the local sequence of each data in the data sequence of each dimension.

[0060] Each data in the data sequence of each dimension and the 60 data before and after it are taken as the local sequence of the data, that is, each data and the data 60 seconds before and after its collection time together constitute the local sequence of the data. For any data in the pressure data sequence, temperature data sequence, and fluid flow data sequence collected in step S1, the local sequence of each data is obtained according to this method. No matter which dimension the data is in, the number of data contained in its local sequence is the same, that is, 121 data.

[0061] S22: Evaluate the correlation between the local sequence of each data in the data sequence of each dimension and the local sequence of data at the same collection time in the data sequence of other dimensions by statistical methods.

[0062] The correlation between the above local sequences satisfies the following relationship:

[0063]

[0064] In the formula, For the In the data series of dimensions The local sequence of the data is In the data series of dimensions The correlation between the local sequences of data, is the hyperbolic tangent function, which is used for normalization. The final result is The value of is between (0, 1). is the number of data contained in each of the two local sequences, that is, both local sequences contain data, , Respectively , In the data series of dimensions In the local sequence of data The value of the data, , Respectively , In the data series of dimensions The mean of all data in the local sequence of data, For the In the data series of dimensions All data in the local sequence of data and the In the data series of dimensions The standard deviation of the differences between all data in the local series of data.

[0065] In this formula, Reflects the In the data series of dimensions The data change trend in the local series of data and the In the data series of dimensions The consistency of the data change trend of the local series of the data. If the data in the two local series increases or decreases at the same time compared with the mean, it means that the change trends of the two series are consistent. If the data in any series increases while the corresponding data in the other series decreases, it means that the change trends of the two series are inconsistent. Specifically, if If the value of is greater than 0, it means that the change trends of the data of the two local series are consistent. If it is less than 0, the change trends of the data of the two local series are inconsistent.

[0066] In this formula, , the smaller the value, the more stable the difference change of the data at the same acquisition time in the two local sequences. If the consistency of the data change trend of the two local sequences is better, that is, The larger the value of , the more correlated the local sequences are. If the value is larger, it means that the difference in the data at the same collection time in the two local sequences fluctuates greatly. If the consistency of the data change trend of the two local sequences is worse, that is, The smaller the value of , the less correlated the local sequences are.

[0067] In this formula, the use of the hyperbolic tangent function can ensure that the normalized correlation is within the range of 0-1, which is convenient for comparison and analysis.

[0068] In this way, we can calculate the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The correlation between local sequences of data.

[0069] S23: Calculate the mutation degree of each data in the data sequence of each dimension according to the slope difference between each data and its adjacent data before and after it, and the correlation, wherein the mutation degree is positively correlated with the correlation and the slope difference.

[0070] In actual scenarios, due to the impact effect at the moment of perforation, dynamic entry of fluid, rock fracture and debris migration, the pressure, temperature and fluid flow in the channel will change dramatically, resulting in rapid fluctuations in data trends and directions. When the pressure in the channel increases, the temperature and fluid flow will also show an increasing trend.

[0071] In order to more accurately capture these rapidly changing data points and perform data segmentation processing accordingly, although the correlation of the data series corresponding to any data point in different dimensions can reveal the changing trends of the three-dimensional series at these data points, this method is not sufficient to accurately identify which data points constitute mutation points. Therefore, further analysis is needed.

[0072] Here, the slope difference between each data and its adjacent data is calculated to capture the severity of the data change. At the same time, the correlation between any dimension and the corresponding data series of the data in all other dimensions is combined to more accurately obtain the mutation point of any dimension.

[0073] Among them, the correlation calculates the mutation degree of the data:

[0074]

[0075] In the formula, For the In the data series of dimensions The degree of mutation of the data, , Respectively In the data series of dimensions The slope of a data point and the previous and next adjacent data points, To take the absolute value sign, For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. For example, the correlation between the local sequence of the 20th data in the pressure data sequence and the local sequence of the 20th data in the temperature data sequence is 0.6, and the correlation between the local sequence of the 20th data in the pressure data sequence and the local sequence of the 20th data in the fluid flow data sequence is 0.4. Then, the average of the correlations between the local sequence of the 20th data in the pressure data sequence and the local sequences of the 20th data in the data sequences of all other dimensions is 0.5. For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The standard deviation of the correlation between the local sequences of data. For example, the standard deviation of the correlation between the local sequence of the 20th data in the pressure data sequence and the local sequence of the 20th data in the data sequences of all other dimensions is:

[0076]

[0077] In this formula (the calculation formula for the degree of mutation), Indicates The difference in slope between the first and last data. The greater the difference in slope, the The greater the difference between the first data and the previous and next adjacent data, the The more likely the data is a mutation point, the smaller the slope difference is, indicating that the The smaller the difference between a data point and the previous and next adjacent data, the less likely it is to be a mutation point. It can be seen that the slope difference is positively correlated with the degree of mutation.

[0078] In this formula and The ratio is the normalized value, The range reflects the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The comprehensive correlation of the local series of data, if the The dimension is the first in the data series of all other dimensions The smaller the standard deviation of the correlation of the local series of data (the smaller the denominator), the larger the mean of these correlations (the larger the numerator), which means that the first The local sequences of data are correlated in multiple dimensions. The higher the comprehensive correlation, the The more the data changes in all dimensions in the time period corresponding to the local sequence of the data are in line with the actual change trend, the The more credible the mutation degree of the data is, the more positively correlated the mutation degree is with the ratio.

[0079] In summary, if The larger the value of and The larger the ratio is, the In the data series of dimensions The greater the difference between the first data and the previous and next adjacent data, the The closer the data change trend of the local sequence of a data is to other dimensions, the more likely the data is a mutation point in the actual change of the data, and the greater the degree of mutation, otherwise, the less likely the data is to be a mutation point.

[0080] The slope of each data and its adjacent data before and after it satisfies the following relationship:

[0081]

[0082] In the formula, For the In the data series of dimensions The slope of a data point and the previous adjacent data point, For the In the data series of dimensions The slope of a data point and the next adjacent data point, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The time when the data is collected.

[0083] S24: Filter out mutation points in the data sequence of each dimension according to the degree of mutation.

[0084] The mutation degree of all data in the data sequence of each dimension is normalized, and the preset mutation degree threshold is 0.6. For a certain data in the data sequence of any dimension, if its normalized mutation degree is greater than the mutation degree threshold, it means that the data has a larger change compared with the previous and next adjacent data, and the greater the comprehensive correlation of the local sequence of the data in different dimensions, then the data is considered to be a mutation point. On the contrary, if its mutation degree is less than or equal to the mutation degree threshold, it is considered that the data has a smaller change compared with the previous and next adjacent data, and the smaller the comprehensive correlation of the local sequence of the data in different dimensions (the lower the credibility of the mutation degree), the data is not a mutation point.

[0085] In this way, all mutation points in the data sequence of each dimension are obtained.

[0086] S3: forming a subsequence between adjacent mutation points, calculating a correction value of the exponential smoothing parameter between the subsequences, and using the correction value to correct the preset exponential smoothing parameter.

[0087] For the data sequence of each dimension, after obtaining all mutation points, all the data between two adjacent mutation points are combined into a subsequence (note that the undivided data at both ends are two subsequences), thereby completing the division of the subsequences.

[0088] In order to more accurately capture the changing trend of data and divide the data into subsequences, it is necessary to consider the different stages of data change and analyze the data characteristics of each stage. For example, when the data shows an increasing trend, recent data points can better reflect the current direction and speed of change, so their importance should exceed historical data points. On the contrary, in the segmentation of a decreasing trend, if recent data is overemphasized and historical data is ignored, it may cause the prediction model to overly pessimistically estimate the overall downward trend. Therefore, it is necessary to give historical data a higher weight.

[0089] Based on the above ideas, the fixed smoothing parameters in the traditional algorithm can be modified to improve the prediction accuracy. Specifically, the product of the median value of the subsequence and the difference between the two end values ​​is used as the numerator, the difference between the two end values ​​is used as the denominator, and the value after normalization of the fraction is used as the correction value of the exponential smoothing parameter between the subsequences, which is expressed as follows:

[0090]

[0091] In the formula, Indicates In the data series of dimensions The corrected value of the dimensional smoothing parameter for the subsequences, represents the median of the subsequence, Indicates the value of the first data of the subsequence. Indicates the value of the last data in the subsequence.

[0092] In this formula, if The larger the value is, the greater the difference between the initial value, median and final value is, which means that the data in the subsequence has a large change range, and it is considered that a larger correction weight is needed. If the value is greater than 0, it means that the data is increasing from the initial value to the median and then to the final value. At this time, the recent data points can better reflect the direction and speed of the current change, and the value of the smoothing parameter should be increased. If the value is less than 0, it indicates that it is decreasing. At this time, if the recent data is overemphasized and the historical data is ignored, it may cause the prediction model to over-pessimistically estimate the overall downward trend. Therefore, it is necessary to give a higher weight to the historical data, that is, to reduce the value of the smoothing parameter.

[0093] In this formula, The inverse tangent function is used to normalize the modified value of the exponential smoothing parameter to a reasonable value. Within the range.

[0094] According to the above steps, the modified value of the exponential smoothing parameter of each subsequence in a data sequence of any dimension can be obtained.

[0095] The method of using the correction value to correct the preset exponential smoothing parameter is as follows:

[0096] The preset exponential smoothing parameter is 0.5. The product of the revised value of the exponential smoothing parameter of each subsequence and 0.5 is used as the revised smoothing parameter of the subsequence. Through the above method, the smoothing parameter can be more accurately adjusted dynamically according to the changing trend of the data, thereby improving the accuracy of the prediction model.

[0097] S4: performing exponential smoothing operations on all subsequences using the modified smoothing parameters, and judging abnormalities and issuing warnings based on the exponential smoothing results.

[0098] Performing exponential smoothing operations on all subsequences using the corrected smoothing parameters of all subsequences to obtain predicted values ​​of each data in all subsequences, and judging whether the subsequences are abnormal based on the difference between the predicted values ​​and the actual values, including:

[0099] Set the forecast deviation to the absolute value of the difference between the predicted value and the actual value of the data.

[0100] The sum of the mean values ​​of the prediction deviations of the data in the data sequence of each dimension within the time period corresponding to the subsequence is taken as the comprehensive abnormal performance degree of the subsequence. The larger the value is, the higher the mean values ​​of the abnormal performance degrees of the data in each dimension within the time period corresponding to the subsequence are, which means that the data in the subsequence has significantly deviated from the normal operating range.

[0101] The preset warning value of the comprehensive abnormal performance degree is 0.75. If the comprehensive abnormal performance degree of the subsequence is greater than 0.75, the subsequence contains abnormal data, and these abnormal data will have an adverse effect on the flow conductivity of the pores, resulting in a decrease in the flow conductivity. It is necessary to issue a warning in time to notify the management personnel to handle it. Otherwise, there is no abnormality in the subsequence and no warning notification is required.

[0102] The present invention also provides a monitoring and early warning system for the conductivity of simulated perforation channels in oil and gas wells. The monitoring and early warning system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of any one of the detection methods.

[0103] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0104] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for monitoring and early warning the flow conductivity of simulated perforation channels in oil and gas wells, characterized in that: include: Collect data of various dimensions in the hole during the simulated perforation operation of oil and gas wells to obtain data sequences of various dimensions; Dimensions include pressure, temperature, and fluid flow; Obtaining mutation points in data sequences of each dimension through data analysis includes: taking each data and several adjacent data before and after each data sequence of each dimension as the local sequence of the data, and the number of data contained in the local sequence of each data is the same; evaluating the correlation between the local sequence of each data in the data sequence of each dimension and the local sequence of data at the same collection time in the data sequence of other dimensions through statistical methods includes: ; In the formula, For the In the data series of dimensions The local sequence of the data is In the data series of dimensions The correlation between the local sequences of data, The function is used for normalization. is the number of data contained in each of the two local sequences, , Respectively , In the data series of dimensions In the local sequence of data The value of the data, , Respectively , In the data series of dimensions The mean of all data in the local sequence of data, For the In the data series of dimensions All data in the local sequence of data and the In the data series of dimensions The standard deviation of the difference of all data in the local sequence of each data; the mutation degree of the data is calculated based on the slope difference and correlation of each data in the data sequence of each dimension and its adjacent data before and after, and the mutation degree is positively correlated with the correlation and the slope difference; the mutation points in the data sequence of each dimension are screened out according to the size of the mutation degree, and the adjacent mutation points form a subsequence, with the product of the median value of the subsequence and the difference between the two end values ​​as the numerator, and the difference between the two end values ​​as the denominator, and the value after normalization of the fraction is used as the correction value of the exponential smoothing parameter between the subsequences; The preset exponential smoothing parameter is corrected by using the correction value. The corrected exponential smoothing parameter is positively correlated with the correction value. The corrected smoothing parameter is used to perform exponential smoothing operations on all subsequences to obtain the predicted value of each data in the subsequence. The difference between the predicted value and the actual value is used to determine whether the subsequence has an abnormality. If an abnormality exists, an early warning is issued.

2. The method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells according to claim 1 is characterized in that: The use of the correction value to correct the preset exponential smoothing parameter is to multiply the correction value and the preset exponential smoothing parameter to obtain a value as the corrected exponential smoothing parameter.

3. The method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells according to claim 1 is characterized in that: Judging whether the subsequence is abnormal according to the difference between the predicted value and the actual value includes: The sum of the mean values ​​of the prediction deviations of the data in the data sequence of each dimension within the time period corresponding to the subsequence is taken as the comprehensive abnormal performance degree of the subsequence; the prediction deviation is the absolute value of the difference between the predicted value and the actual value of the data; Preset the warning value of the comprehensive abnormal performance degree; If the comprehensive abnormal performance degree of the subsequence is greater than the warning value, the subsequence is abnormal; otherwise, the subsequence is not abnormal.

4. The method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells according to claim 1 is characterized in that: The method for screening out mutation points in the data sequence of each dimension according to the magnitude of the mutation degree is as follows: normalizing the mutation degrees of all data in the data sequence of each dimension, and taking data whose normalized mutation degrees are greater than a preset mutation degree threshold as mutation points.

5. The method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells according to claim 1 is characterized in that: The mutation degree of the data is calculated based on the slope difference between each data and its adjacent data before and after it in the data sequence of each dimension, and the correlation: ； In the formula, For the In the data series of dimensions The degree of mutation of the data, , Respectively In the data series of dimensions The slope of a data point and the previous and next adjacent data points, For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The mean of the correlation between the local series of data, For the In the data series of dimensions The local sequence of data is the same as the data sequence of all other dimensions. The standard deviation of the correlation between the local series of data, To take the absolute value sign.

6. The method for monitoring and early warning of the conductivity of simulated perforation channels in oil and gas wells according to claim 5 is characterized in that: The slope satisfies the following relationship: ； In the formula, For the In the data series of dimensions The slope of a data point and the previous adjacent data point, For the In the data series of dimensions The slope of a data point and the next adjacent data point, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The value of the data, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The data collection time, For the In the data series of dimensions The time when the data is collected.

7. A monitoring and early warning system for the conductivity of simulated perforation channels in oil and gas wells, characterized in that: The monitoring and early warning system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the monitoring and early warning method according to any one of claims 1 to 6.

Images