Continuous coiled tubing hot washing wax removal and plugging removal technology for oil and gas wells

Abstract

This application relates to the field of petroleum engineering and provides a process for hot washing paraffin removal and plugging removal of coiled tubing in oil and gas wells. First, a hot oil truck is used to heat the well washing fluid, and then a pump truck is used to inject the heated well washing fluid into the coiled tubing operation machine. Subsequently, these well washing fluids are sent into the well through the coiled tubing and circulate in the circulation channel in the well. During the circulation of the well washing fluid, the water horsepower impact, heat exchange, and mechanical crushing effects generated by the well washing fluid are used to remove or break up the blockages in the well, and finally the decomposed blockages return to the ground together with the well washing fluid. In this way, the effect of paraffin removal and plugging removal in the oil and gas well can be improved, which is beneficial to better realizing the effective maintenance of the oil and gas well.

Description

Technical Field

[0001] This application relates to the field of petroleum engineering, and more specifically, to a continuous coiled tubing hot washing wax removal and plugging removal process for oil and gas wells. Background Art

[0002] Wax removal and plugging removal are important parts of the management work of oil and gas wells. During the crude oil production process, due to the decrease in temperature and the escape of light hydrocarbons, wax precipitates and adheres to oil production equipment such as the tubing wall and casing wall, resulting in a decrease in production or even blockage of the production string, which is one of the prominent problems affecting the high and stable production of oil and gas wells.

[0003] However, the traditional mechanical wax removal, chemical wax melting, and hot oil injection wax melting operations for oil and gas wells have high risks, incomplete wax removal, and cannot solve the problem of the production string being blocked by wax and other blockages. Specifically, although mechanical wax removal can directly remove some wax deposits, it has limited effect on severely blocked strings, and the operation is complex and the workload is large. Chemical wax melting dissolves wax by injecting chemical solvents, but the cost of chemical reagents is high, it has a certain impact on the environment, and it cannot completely solve the blockage problem. Hot oil injection wax melting has a high construction pressure, high risks, low success rate of plugging removal, large workload, incomplete wax removal, and poor effect.

[0004] Therefore, an optimized continuous coiled tubing hot washing wax removal and plugging removal scheme for oil and gas wells is desired. Summary of the Invention

[0005] This application aims at the disadvantages in the prior art and provides a continuous coiled tubing hot washing wax removal and plugging removal process for oil and gas wells.

[0006] According to an aspect of the present application, a continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells is provided, which includes: heating the well washing fluid by a hot oil truck to obtain the heated well washing fluid; injecting the heated well washing fluid into a coiled tubing operation machine by a pump truck; the coiled tubing operation machine lowering the well washing fluid into the well through a coiled tubing, and there is a well washing fluid circulation channel in the well, and the well washing fluid can circulate in the well through the well washing fluid circulation channel to utilize the hydraulic horsepower impact, heat energy exchange and mechanical crushing generated during the circulation of the well washing fluid to remove or break up the blockage, including: real-time monitoring of the state parameters of the well washing fluid during the in-well circulation in the well washing fluid circulation channel to obtain a time queue of the state parameters, and the state parameters include temperature value, pressure value and flow rate value; sorting the time queue of the state parameters to obtain a time queue of temperature values, a time queue of pressure values and a time queue of flow rate values; performing sequence coding on the time queue of temperature values and the time queue of flow rate values to obtain temperature time series features and flow rate time series features; performing temperature-flow rate time series dual-perspective response correlation analysis on the temperature time series features and the flow rate time series features to obtain temperature-flow rate time series interaction response coding features; performing pressure inference on the temperature-flow rate time series interaction response coding features to obtain a time queue of inferred pressure values; based on the time queue of pressure values and the time queue of inferred pressure values, determining whether there is a circulation anomaly; the decomposed blockage returns to the ground together with the well washing fluid.

[0007] Due to the adoption of the above technical solutions, the present application has remarkable technical effects: for the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells provided by the present application, it first heats the well washing fluid by a hot oil truck, then injects the heated well washing fluid into the coiled tubing operation machine by a pump truck, and then these well washing fluids are sent into the well through the coiled tubing and circulate in the circulation channel in the well. During the circulation of the well washing fluid, the hydraulic horsepower impact, heat exchange and mechanical crushing generated by the well washing fluid are used to remove or break up the blockage in the well, and finally the decomposed blockage returns to the ground together with the well washing fluid. In this way, the effect of paraffin removal and plugging removal in the oil and gas well can be improved, which is beneficial to better realizing the effective maintenance of the oil and gas well. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 It is a flowchart of the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to the embodiments of the present application.

[0010] Figure 2 It is a flowchart of step S3 in the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to an embodiment of the present application.

[0011] Figure 3 It is a flowchart of step S34 in the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to an embodiment of the present application.

[0012] Figure 4 It is a flowchart of step S36 in the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to an embodiment of the present application. Detailed implementation manners

[0013] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0014] Paraffin removal and plugging removal is a key task in oil and gas well management. During the crude oil production process, due to the decrease in temperature and the volatilization of light hydrocarbons, wax will gradually precipitate and adhere to oil production equipment such as the tubing wall and casing wall, thereby causing a decrease in production, and even leading to the blockage of the production string. This phenomenon has become one of the main problems restricting the efficient and stable production of oil and gas wells.

[0015] Existing paraffin removal technologies mainly include mechanical paraffin removal, chemical wax melting, and hot oil squeezing wax melting. However, these traditional methods have great limitations in practical applications and are difficult to completely solve the problem of string blockage. Specifically, although mechanical paraffin removal can directly remove some wax deposits, for severely blocked strings, its effect is limited, and at the same time, the operation is complex and the workload is large. Chemical wax melting dissolves wax by injecting chemical reagents. However, this method has a high cost, may have an adverse impact on the environment, and the effect is not ideal when dealing with stubborn blockages. Although hot oil squeezing wax melting can melt wax through high temperature, the pressure is high during the construction process, there are certain safety risks, and the plugging removal success rate is low, and the cleaning effect is still not thorough enough.

[0016] Based on this, the present application proposes a continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells, which significantly improves the effect and safety of paraffin removal and plugging removal, reduces construction risks and maintenance costs by introducing advanced technologies and optimizing operation processes. Specifically, Figure 1 It is a flowchart of the continuous coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to an embodiment of the present application. As Figure 1As shown in the figure, the continuous coiled tubing hot washing and paraffin removal and plugging removal process according to the embodiment of the present application includes: S1, heating the washing fluid by a hot oil truck to obtain the heated washing fluid; S2, injecting the heated washing fluid into a coiled tubing operation machine by a pump truck; S3, the coiled tubing operation machine lowering the washing fluid into the well through the coiled tubing. There is a washing fluid circulation channel in the well, and the washing fluid can circulate in the well through the washing fluid circulation channel to utilize the hydraulic horsepower impact, heat energy exchange, and mechanical crushing generated during the circulation process of the washing fluid to remove or break up the blockages; S4, the decomposed blockages return to the ground together with the washing fluid.

[0017] In step S1, the washing fluid is heated by a hot oil truck to obtain the heated washing fluid. It should be understood that wax will precipitate from crude oil in a low-temperature environment and adhere to oil production equipment such as the tubing wall and casing wall, and may block the production string seriously in severe cases. By heating the washing fluid, the temperature of the washing fluid can be increased, and the heat energy is used to melt the wax blocks adhering to the surface of the equipment, so as to achieve the purpose of paraffin removal. Compared with the traditional operation of squeezing and injecting hot oil to melt wax, this method reduces the construction pressure, reduces the operation risk, and improves the paraffin removal efficiency and thoroughness at the same time. It is worth noting that according to different operation conditions and blockage situations, a suitable type of washing fluid (such as water or crude oil) can be selected to optimize the operation efficiency and effect.

[0018] In step S2, the heated washing fluid is injected into the coiled tubing operation machine by a pump truck. It should be understood that the washing fluid cannot circulate in the well by itself and play the role of paraffin removal and plugging removal only after being heated. The pump truck can provide powerful power to inject the heated washing fluid into the coiled tubing operation machine, and push the washing fluid into the coiled tubing in the well, so as to establish a washing fluid circulation channel in the well and make the washing fluid circulate continuously. Simply put, only when the washing fluid circulates continuously can it continuously contact and clean the wax and blockages on the tubing wall and casing wall to achieve paraffin removal and plugging removal. Moreover, the conditions of different oil and gas wells vary, and the requirements for the flow rate and pressure of the washing fluid are different. The pump truck can accurately adjust the flow rate and pressure of the washing fluid injected into the coiled tubing operation machine according to actual needs. When dealing with severely blocked well sections, the flow rate and pressure can be increased to enhance the impact force of the washing fluid and better break up and remove the blockages; for relatively slightly blocked or more complex well sections, the flow rate and pressure can be appropriately reduced to avoid damaging the internal structure of the well.

[0019] In step S3, the coiled tubing operation machine lowers the well-washing fluid into the well through the coiled tubing. The well has a well-washing fluid circulation channel. The well-washing fluid can circulate in the well through the well-washing fluid circulation channel to remove or break the blockage by utilizing the water horsepower impact, heat energy exchange and mechanical crushing generated by the well-washing fluid during the circulation process. It should be understood that the well-washing fluid has a high temperature after being heated, and exchanges heat energy with the wax attached to the oil pipe wall, casing wall and other oil production equipment during the circulation process. The heat transfer increases the temperature of the wax, and after reaching the melting point, it changes from solid to liquid, thereby weakening the adhesion to the pipe wall. At the same time, the water horsepower impact continuously acts on the melted wax, peeling it off the pipe wall, so that the originally attached wax gradually detaches from the equipment surface, creating conditions for the subsequent well-washing fluid to be taken out of the wellhead. Since the coiled tubing can penetrate into different depths and positions in the well, the well-washing fluid can flow in the entire well-washing fluid circulation channel after entering the well through the coiled tubing. This allows the well-washing fluid to reach every corner of the well, including some parts that are difficult to reach with conventional wax removal methods, ensuring that every part of the pipe string can be cleaned, achieving comprehensive wax removal, and avoiding wax residue in local areas, effectively solving the problem of incomplete wax removal by traditional wax removal methods. For blockages in the well, such as sand blockages, combustible ice blockages, etc., the water horsepower impact and mechanical crushing effect generated by the circulation of the well-washing fluid play a key role. The water horsepower impact can exert a strong external force on the blockage to loosen and break it. When the strength of the blockage is large, a combination of continuous tubing + screw drill + crushing tool can also be used to further enhance the mechanical crushing ability and break large pieces of blockage into small pieces. These crushed blockages will return to the ground with the circulation of the well-washing fluid, thereby removing the blockage of the pipe string and restoring the normal production of the oil and gas well. That is, the present application can more thoroughly remove or crush the blockage by combining the three mechanisms of water horsepower impact, heat energy exchange and mechanical crushing, significantly improve the wax removal effect, and effectively restore the production capacity of the oil well. It should be noted that the entire process does not rely on chemical reagents, reduces the impact on the environment, and is fully in line with the concept of modern green mining.

[0020] It should be particularly noted that in the step of circulating the well flushing fluid in the well through the well flushing fluid circulation channel to utilize the hydraulic horsepower impact, heat energy exchange and mechanical crushing to remove or break up blockages, it is crucial to detect abnormal circulation. This can ensure that the well flushing fluid effectively exerts the functions of hydraulic horsepower impact, heat energy exchange and mechanical crushing, avoid problems such as ineffective paraffin removal and plugging removal, equipment damage and environmental pollution, and ensure the efficient, safe and environmentally friendly operation of the well flushing operation for oil and gas wells. However, traditional abnormal circulation detection methods have limitations. They only focus on a few parameters, such as pressure thresholds, and it is difficult to comprehensively analyze multi-parameter data such as temperature, pressure and flow rate. For example, when the pressure briefly exceeds the threshold and then returns to normal, it is easily misjudged as normal, ignoring the relationship between pressure fluctuations and temperature and flow rate. In addition, traditional methods rely on fixed thresholds and simple rules, and cannot automatically adjust the judgment criteria according to complex working conditions, making it difficult to adapt to changes under different formation temperatures and prone to missed or misjudgments.

[0021] Based on this, in the process where the well flushing fluid can circulate in the well through the well flushing fluid circulation channel to utilize the hydraulic horsepower impact, heat energy exchange and mechanical crushing generated during the circulation process of the well flushing fluid to remove or break up blockages, the technical concept of this application is to real-time monitor the state parameters of the well flushing fluid during the in-well circulation in the well flushing fluid circulation channel to obtain a time queue of state parameters (temperature values, pressure values and flow rate values), use data analysis and processing techniques based on deep learning to organize the data of temperature values, pressure values and flow rate values to obtain a time queue of temperature values, a time queue of pressure values and a time queue of flow rate values. Then, perform time series feature encoding on the time queues of temperature values and flow rate values respectively, and perform dual-perspective response correlation on the encoded temperature time series features and flow rate time series features, so as to intelligently judge whether there is abnormal circulation based on the deviation between the time queue of the inferred pressure value obtained by pressure inference according to the temperature-flow rate time series interaction representation and the time queue of pressure values. This application not only focuses on pressure thresholds, but also monitors multi-parameter data such as temperature, pressure and flow rate at the same time to comprehensively reflect the dynamic changes during the well flushing fluid circulation process. Moreover, it can dynamically adjust the judgment criteria according to the actual situation, which is beneficial to improving the accuracy of abnormal detection and ensuring the effective detection of abnormal situations under various conditions.

[0022] Figure 2 It is a flowchart of step S3 in the coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to the embodiment of this application. As Figure 2As shown, step S3 includes: S31, continuously monitoring the state parameters of the well flushing fluid during in-well circulation in the well flushing fluid circulation channel to obtain a time queue of state parameters, where the state parameters include temperature values, pressure values, and flow values; S32, organizing the data of the time queue of state parameters to obtain a time queue of temperature values, a time queue of pressure values, and a time queue of flow values; S33, performing sequence encoding on the time queue of temperature values and the time queue of flow values to obtain temperature time series features and flow time series features; S34, performing temperature-flow time series dual-perspective response correlation analysis on the temperature time series features and the flow time series features to obtain temperature-flow time series interaction response coding features; S35, performing pressure inference on the temperature-flow time series interaction response coding features to obtain a time queue of inferred pressure values; S36, based on the time queue of pressure values and the time queue of inferred pressure values, determining whether there is a circulation anomaly.

[0023] In step S31, continuously monitoring the state parameters of the well flushing fluid during in-well circulation in the well flushing fluid circulation channel to obtain a time queue of state parameters, where the state parameters include temperature values, pressure values, and flow values. It should be understood that the temperature value can intuitively show the heat carried by the well flushing fluid. One of the main tasks of the well flushing fluid is to melt blockages such as wax through heat energy exchange. If the temperature value decreases abnormally, it may mean that there is a malfunction in the heating system, resulting in the inability of the well flushing fluid to effectively melt the blockages, and the effect of wax removal and blockage removal will be greatly reduced; the pressure value reflects the power of the well flushing fluid during circulation. Sufficient pressure is the key to achieving hydraulic horsepower impact and mechanical crushing. Insufficient pressure will weaken the impact force of the well flushing fluid and make it difficult to break the blockages; the flow value indicates the circulation speed of the well flushing fluid. An appropriate flow rate can ensure that the well flushing fluid fully contacts the pipe string within a unit time, taking away the melted wax and broken blockages. Abnormal flow may lead to the accumulation of blockages and affect the effect of wax removal and blockage removal. Relying solely on any one parameter (such as only focusing on the pressure threshold) may lead to misjudgment of the circulation state. For example, a short-term pressure fluctuation may not immediately cause problems, but if combined with a temperature drop and a decrease in flow, it may be due to some blockage that causes an abnormality in the entire circulation. That is, through continuous monitoring and in-depth correlation analysis of these key parameters of temperature, pressure, and flow in this application, abnormal patterns of the well flushing fluid circulation can be more accurately identified, thereby effectively improving the reliability of circulation anomaly detection, which is conducive to ensuring the smooth progress of the well flushing operation.

[0024] The following is a detailed elaboration of a specific implementation process for "real-time monitoring of the state parameters of the well flushing fluid during its in-well circulation in the well flushing fluid circulation channel to obtain a time queue of state parameters, where the state parameters include temperature values, pressure values, and flow values": First is the selection and installation of monitoring equipment. In terms of temperature monitoring, a platinum resistance temperature sensor can be selected, which has the characteristics of high temperature resistance, corrosion resistance, and high precision. Installing the temperature sensor near the wellhead of the coiled tubing can enable real-time grasping of the initial temperature of the well flushing fluid before it enters the well. At the same time, according to the specific structure of the well and the areas where there may be a risk of blockage, temperature sensors are reasonably arranged at key nodes at different depths in the well to ensure that the temperature changes of the well flushing fluid at different positions can be comprehensively and accurately measured. During installation, it is necessary to ensure that the probe of the sensor is in full contact with the well flushing fluid to guarantee accurate temperature values are obtained. In the pressure monitoring section, strain gauge pressure sensors are widely used because they can adapt to the high-pressure downhole environment and have good stability. On the coiled tubing, the installation position needs to be carefully selected to ensure that the sensor is directly connected to the well flushing fluid, and at the same time, full consideration should be given to avoiding the influence of mechanical collision and blockage on the sensor. Pressure sensors are set at the wellhead and key parts in the well where pressure changes easily, such as near the pipe section that may be blocked. In this way, the pressure fluctuations during the well flushing fluid circulation can be monitored in real time and accurately, providing an important basis for judging the circulation state. Flow monitoring is also indispensable for well flushing operations. According to the properties and flow range of the well flushing fluid, an electromagnetic flowmeter or an ultrasonic flowmeter is reasonably selected. If the well flushing fluid is conductive, the electromagnetic flowmeter can accurately measure the flow based on its principle; for non-conductive well flushing fluid, the ultrasonic flowmeter can play its advantages. When installing the flow sensor, a straight pipe section of the coiled tubing should be selected because in the straight pipe section, the flow state of the well flushing fluid is relatively stable, which can effectively avoid the adverse effects on the measurement accuracy caused by pipe bending or diameter change, thus ensuring the accuracy of flow measurement.

[0025] Next, it is necessary to determine an appropriate data acquisition frequency. At the initial stage of starting the well flushing operation, since the well flushing fluid circulation state is in an unstable stage, various parameters change rapidly and the trends are not obvious. In order to comprehensively and carefully understand the start-up situation of the circulation and the change trends of the parameters, a relatively high data acquisition frequency needs to be set, such as collecting data once per second. This can obtain rich initial data and provide sufficient information for subsequent analysis. As the well flushing operation gradually stabilizes, if the changes in various parameters are relatively stable, the acquisition frequency can be appropriately reduced and adjusted to collect data once every 5 - 10 seconds to balance the data volume and data processing cost. However, once abnormal situations occur, such as sudden pressure fluctuations or sharp temperature changes, the system should have the function of automatically increasing the acquisition frequency to more timely and accurately capture the abnormal change information of the parameters, providing strong support for timely discovery and solution of problems.

[0026] The transmission and storage of data are important links in the entire monitoring process. In terms of data transmission, wired or wireless transmission methods can be selected according to the actual situation. When using wired transmission, armored cables are commonly chosen due to their good protection performance. By connecting sensors to the ground data acquisition station with armored cables, signal interference can be effectively avoided, ensuring the stability and reliability of data transmission. However, in some special wellsite environments, such as areas with complex terrain, difficult cable laying, or high costs, wireless transmission technology shows its advantages. Wireless modules based on low-power, long-distance transmission protocols such as ZigBee and LoRa can send the data collected by sensors to ground receiving devices, realizing wireless data transmission.

[0027] In terms of data storage, a dedicated database needs to be established in the ground data processing system to store the collected status parameter data. According to the characteristics of the data and subsequent processing requirements, relational databases (such as MySQL) or time series databases (such as InfluxDB) can be selected. These databases can store the collected temperature values, pressure values, and flow values in an orderly manner in the form of a time queue according to the chronological order of collection time, which can provide reliable raw data support for subsequent data sorting, analysis, and model processing.

[0028] To ensure the accuracy and reliability of data, data verification and preprocessing are essential steps. During data collection and transmission, due to the influence of various factors, data errors or outliers may occur. Therefore, strict verification of the collected data is required. By setting reasonable threshold ranges, such as determining the normal range of temperature values, the safe interval of pressure values, and the reasonable fluctuation range of flow values based on the actual experience and theoretical calculations of well flushing operations, the rationality of the data can be judged. For data beyond the threshold, it should be clearly marked and its accuracy further verified to ensure the authenticity and reliability of the data.

[0029] In the data preprocessing stage, mainly missing values and outliers are processed. For a small number of missing values, interpolation methods can be used for supplementation. Linear interpolation estimates the missing values through the linear relationship between known data points, while Lagrange interpolation uses polynomial functions to fit the data to obtain an approximate value of the missing values. For outliers, they should be processed according to the characteristics and actual situation of the data. If the outliers are caused by measurement errors, they can be corrected; if their rationality cannot be determined and they have a significant impact on subsequent analysis, they can be considered for elimination. At the same time, for the convenience of subsequent data analysis and model processing, the data also needs to be normalized, converting data with different physical dimensions to a unified numerical range to eliminate the influence of dimension differences on the analysis results.

[0030] In step S32, data sorting is performed on the time queue of the state parameters to obtain a time queue of temperature values, a time queue of pressure values, and a time queue of flow values. Correspondingly, considering that temperature, pressure, and flow are parameters of different physical properties, the states of the well-washing fluid reflected by them and the influence mechanisms on the paraffin removal and plugging removal processes are different. Temperature mainly involves heat energy transfer and wax dissolution, pressure is related to the integrity of the circulation system and water horsepower impact, and flow focuses on the circulation efficiency of the well-washing fluid and the ability to carry blockages. Therefore, in order to be able to analyze each parameter specifically to capture and extract the time-series characteristics of each parameter in the time dimension, in the technical solution of this application, it is necessary to perform data sorting on the time queue of the state parameters to obtain a time queue of temperature values, a time queue of pressure values, and a time queue of flow values.

[0031] In step S33, the time queue of the temperature value and the time queue of the flow value are sequence-encoded to obtain the temperature time series feature and the flow time series feature. Specifically, in the embodiment of the present application, the step S33 includes: using a sequence encoder to sequence-encode the time queue of the temperature value and the time queue of the flow value respectively to obtain the temperature time series feature implicit encoding vector as the temperature time series feature and the flow time series feature implicit encoding vector as the flow time series feature. In particular, the sequence encoder is a sequence encoder based on a bidirectional gated cyclic unit. It should be understood that the sequence data of the temperature value and the flow value changing over time contains rich information about the well washing process, which is crucial to understanding the dynamic changes of the well washing fluid during the circulation process. Based on this, the present application performs sequence encoding on the time queue of the temperature value and the time queue of the flow value to mine the hidden features of the data in the time dimension, such as trend, periodicity, seasonality, etc., to obtain the temperature time series feature implicit encoding vector and the flow time series feature implicit encoding vector, so as to better understand the dynamic changes of the well washing process. In particular, in a specific embodiment of the present application, the time queue of the temperature value and the time queue of the flow value are respectively sequence-encoded by using a sequence encoder based on a bidirectional gated cyclic unit to simultaneously process the forward and backward time series information, thereby obtaining an implicit encoding vector of the temperature time series feature and an implicit encoding vector of the flow time series feature. A person skilled in the art should know that the bidirectional gated cyclic unit is an improved recurrent neural network (RNN) structure that combines the advantages of the gated cyclic unit and bidirectional processing. It enhances the ability to understand time dependencies by simultaneously processing the forward and backward information of time series data. The change of temperature and flow over time is a complex dynamic process. Not only is the current value affected by the past value, but it may also be related to the future change trend. For example, in the process of hot washing and paraffin removal of oil and gas wells, the temperature rise at a certain moment may be because more high-temperature well washing fluid is about to be injected (future information) later, and it is also related to the temperature change of the previous well washing fluid (past information). The bidirectional gated cyclic unit can process the time series data of temperature and flow from both the forward and reverse directions, and dig out these nonlinear relationships hidden in the data, thereby fully characterizing this complex time-dependent feature.

[0032] In step S34, a temperature-flow time series dual-view response correlation analysis is performed on the temperature time series characteristics and the flow time series characteristics to obtain a temperature-flow time series interactive response coding feature. Specifically, Figure 3 This is a flow chart of step S34 in the hot-washing, wax-removing and plugging-removing process of coiled tubing in oil and gas wells according to an embodiment of the present application. Figure 3As shown in the figure, step S34 includes: S341, performing a homography projection transformation on the temperature time-series feature implicit encoding vector and the flow rate time-series feature implicit encoding vector to obtain a temperature time-series feature homography projection encoding vector and a flow rate time-series feature homography projection encoding vector; S342, calculating a dual-view temperature-flow rate time-series feature attention score field between the temperature time-series feature homography projection encoding vector and the flow rate time-series feature homography projection encoding vector; S343, modulating the temperature time-series feature homography projection encoding vector and the flow rate time-series feature homography projection encoding vector based on the dual-view temperature-flow rate time-series feature attention score field to obtain a temperature time-series feature homography projection attention modulation encoding vector and a flow rate time-series feature homography projection attention modulation encoding vector; S344, calculating a temperature-flow rate time-series interaction response encoding vector by dividing the temperature time-series feature homography projection attention modulation encoding vector and the flow rate time-series feature homography projection attention modulation encoding vector at the position points as the temperature-flow rate time-series interaction response encoding feature.

[0033] It should be understood that during the hot washing paraffin removal and plugging removal process, temperature and flow rate do not exist independently, but rather influence each other and are closely related. For example, a change in temperature will affect the viscosity of the well-washing fluid, thereby affecting its fluidity and ultimately resulting in a change in flow rate; conversely, the magnitude of the flow rate will also affect the transfer and distribution of heat in the wellbore, thus acting on the temperature. Therefore, in order to capture and excavate this complex internal relationship to comprehensively understand the interaction mechanism between parameters, this application introduces a temperature-flow rate time-series dual-view response correlation analysis method to perform an interaction response on the temperature time-series feature and the flow rate time-series feature to obtain a temperature-flow rate time-series interaction response encoding feature. In particular, this method can fuse the information from two perspectives, enhance the model's understanding ability of multi-dimensional time-series data, provide a richer semantic representation, so that the obtained representation not only contains the change laws of their respective parameters, but also reflects the new information generated by the interaction between the two, thus more comprehensively describing the physical phenomena during the well-washing process.

[0034] Specifically, in the embodiment of this application, step S341 includes: mapping the temperature time-series feature implicit encoding vector through a temperature time-series feature mapping homography matrix to obtain the temperature time-series feature homography projection encoding vector, and this process can be expressed as: ; where is the temperature time-series feature implicit encoding vector, is the temperature time-series feature mapping homography matrix, is the temperature time-series feature homography projection encoding vector.

[0035] Mapping the implicit encoding vector of the flow time-series features through the homography matrix of the flow time-series features to obtain the homography projection encoding vector of the flow time-series features, which can be expressed as: ; where is the implicit encoding vector of the flow time-series features, is the homography matrix of the flow time-series feature mapping, is the homography projection encoding vector of the flow time-series features.

[0036] It should be understood that performing a homography projection transformation on the implicit encoding vector of the temperature time-series features and the implicit encoding vector of the flow time-series features can map the implicit encoding vector of the temperature time-series features and the implicit encoding vector of the flow time-series features into a new feature space. In the new feature space, the temperature and flow features can be observed from different perspectives, which enables the model to discover some information that is difficult to detect in the original space. This helps to better understand the essence of the physical phenomena during the well flushing process. Moreover, the homography projection transformation can ensure that in the new feature space, the relative positional relationship between the temperature and flow time-series features is maintained. This means that in the original feature space, key information such as the mutual correlation, sequence, and relative importance between the temperature and flow features will not be lost. For example, during the hot flushing and paraffin removal and plugging removal process, if the temperature increase is always accompanied by a certain change in the flow rate, this correlation hidden in the original vector can still be reflected after the projection transformation. This provides a stable basis for further analyzing the relationship between the two and ensures the reliability of the analysis results.

[0037] Specifically, in the embodiment of the present application, the step S342 includes: calculating the forward temperature-flow time-series feature attention score field of the homography projection encoding vector of the temperature time-series features relative to the homography projection encoding vector of the flow time-series features, which can be expressed as: ; where is the homography projection encoding vector of the temperature time-series features, is the homography projection encoding vector of the flow time-series features is the transposed vector of, is matrix multiplication, is the length of, is the forward temperature-flow time-series feature attention score field.

[0038] Calculating the reverse temperature-flow time-series feature attention score field of the homography projection encoding vector of the flow time-series features relative to the homography projection encoding vector of the temperature time-series features, which can be expressed as: ; where is the homography projection encoding vector of the temperature time-series features, is the homography projection coding vector of the flow time series features, is the transposed vector of, is matrix multiplication, is the length of, is the reverse temperature-flow time series feature attention score field.

[0039] Based on the forward temperature-flow time series feature attention score field and the reverse temperature-flow time series feature attention score field, the dual-view temperature-flow time series feature attention score field is obtained, and this process can be expressed as: ; where, is the forward temperature-flow time series feature attention score field, is the reverse temperature-flow time series feature attention score field, is the feature concatenation operation, is the convolutional coding with a 3×3 convolutional kernel, is the dual-view temperature-flow time series feature attention score field.

[0040] Correspondingly, considering that during the hot washing and paraffin removal and plugging removal process, temperature and flow are key factors that affect each other, but the relationship between them is complex and non-linear. Calculating the forward temperature-flow time series feature attention score field can accurately quantify the degree of association between the temperature time series feature homography projection coding vector and the flow time series feature homography projection coding vector. Through this quantification method, it is possible to clearly understand which parts of the temperature features have a more critical impact on the entire well washing process under the background of flow features, thereby uncovering the deep internal connection between the two. That is, by obtaining the forward temperature-flow time series feature attention score field, the model can more deeply understand the complex interaction relationship between temperature and flow. This understanding is not limited to the surface change trend, but can also uncover the causal relationship and synergy effect hidden behind the data. When analyzing the well washing fluid circulation process, the model can clearly distinguish which temperature changes are caused by flow changes based on the forward attention score field, and how these changes affect the paraffin removal and plugging removal effect, thereby improving the understanding level of the entire well washing process. Moreover, in the actual monitoring and data processing process, there will inevitably be some noise information, which may interfere with the judgment of the true relationship between temperature and flow. Calculating the forward attention score field is equivalent to setting a filter, which can filter out those temperature feature parts that have a weak correlation with the flow feature vector, that is, irrelevant noise information, thereby improving the data quality and analysis accuracy.

[0041] It should be understood that the forward temperature-flow time-series feature attention score field mainly focuses on the importance distribution of temperature time-series features against the background of flow time-series features. However, this may not comprehensively reflect the relationship between the two feature vectors. During the hot washing and paraffin removal and plugging removal process, the influence of flow changes on temperature may be different from that of temperature changes on flow, and their interaction is relatively complex. By calculating the reverse temperature-flow time-series feature attention score field, information that cannot be covered by the forward attention score field can be supplemented from the perspective of the importance distribution of flow features against the background of temperature features, so as to more comprehensively understand the interaction between temperature and flow. That is, the reverse temperature-flow time-series feature attention score field provides an opportunity to understand the relationship between temperature and flow features from the opposite direction. Through reverse analysis, interaction details that may be overlooked in forward analysis can be mined, and the internal connection between the two features can be more deeply understood. For example, during the well washing process, the change in flow may affect the distribution and change of temperature in a special way, and this influence may not be obvious in forward analysis, but can be clearly presented through the calculation of the reverse attention score field.

[0042] Accordingly, considering that when analyzing the relationship between temperature and flow, the mutual influence between the two feature vectors needs to be comprehensively considered to prevent overemphasizing the role of a certain feature vector. If only relying on the forward temperature-flow time-series feature attention score field or the reverse temperature-flow time-series feature attention score field, the influence of one feature on the other may be overemphasized in the final analysis result, while ignoring the role of the other direction. The construction of the dual-view temperature-flow time-series feature attention score field can balance this influence, ensure that the contributions of the two feature vectors are fully reflected in the interaction response encoding, and thus more accurately describe the essential connection behind the physical phenomena during the well washing process.

[0043] Specifically, in the embodiment of the present application, the step S343 includes: multiplying the dual-view temperature-flow time-series feature attention score field by the temperature time-series feature homography projection coding vector and the flow time-series feature homography projection coding vector respectively to obtain the temperature time-series feature homography projection attention modulation coding vector and the flow time-series feature homography projection attention modulation coding vector. The above process can be expressed as: ; where is the temperature time-series feature homography projection coding vector, is the flow time-series feature homography projection coding vector, is the dual-view temperature-flow time-series feature attention score field, is matrix multiplication, is the temperature time-series feature homography projection attention modulation coding vector, is the flow time-series feature homography projection attention modulation coding vector.

[0044] It should be understood that modulating the homographic projection encoding vectors of the temperature time-series features and the flow rate time-series features based on the dual-view temperature-flow rate time-series feature attention score field can enable these two vectors to no longer be limited to their original feature information alone, but rather incorporate the influence from the features of the other party. This makes the newly generated homographic projection attention modulation encoding vectors of the temperature time-series features and the flow rate time-series features contain richer and more comprehensive information. That is, when describing the process of well flushing fluid circulation, the new encoding vectors can not only reflect the change trends of the temperature or flow rate itself, but also the influence of their interaction on each other's changes. Moreover, under different oil and gas well conditions, the interaction patterns between temperature and flow rate are different. This modulation method enables the model to flexibly adjust its understanding and representation of features according to specific situations. For example, in the special oil and gas well environment of high temperature and high pressure, the changes in temperature and flow rate are more complex. Through modulation, the model can better adapt to this special working condition, accurately capture the unique interaction relationship between temperature and flow rate, improve the analysis and processing ability of the well flushing process under different working conditions, and thus enhance the adaptability and generalization ability of the model.

[0045] Finally, calculate the temperature-flow rate time-series interaction response encoding vector by dividing the homographic projection attention modulation encoding vector of the temperature time-series features and the homographic projection attention modulation encoding vector of the flow rate time-series features at each position point as the temperature-flow rate time-series interaction response encoding feature. The above process can be expressed as: ; where is the homographic projection attention modulation encoding vector of the temperature time-series features, is the homographic projection attention modulation encoding vector of the flow rate time-series features, is the temperature-flow rate time-series interaction response encoding vector.

[0046] It should be understood that by dividing the homographic projection attention modulation encoding vector of the temperature time-series features and the homographic projection attention modulation encoding vector of the flow rate time-series features at each position point, new relationships presented after modulation can be discovered. These new relationships are derived from the encoding vectors that have incorporated mutual influence information and are more complex and in-depth than the relationships of the original feature vectors. When analyzing the well flushing process, this division operation can clearly show the specific degree of association between the temperature and flow rate features after interaction at each time point. This can provide more comprehensive and targeted information for subsequent data analysis tasks, and thus help to more accurately grasp the circulation state of the well flushing fluid during the well flushing process.

[0047] In step S35, pressure inference is performed on the temperature-flow time-series interaction response coding feature to obtain a time queue of inferred pressure values. Specifically, in the embodiment of the present application, step S35 includes: inputting the temperature-flow time-series interaction response coding vector into a pressure simulator based on an RNN model to obtain the time queue of the inferred pressure values. It should be understood that during the well flushing process, pressure is one of the important indicators for judging whether the circulation is normal. Moreover, the changes in temperature and flow will directly affect the well pressure, and these changes are often dynamic and complex. Based on this, in order to infer the actual situation of the ideal pressure based on the interaction representation between temperature and flow, in the technical solution of the present application, the temperature-flow time-series interaction response coding vector is input into a pressure simulator based on an RNN model to achieve intelligent prediction of pressure and obtain a time queue of inferred pressure values. Those of ordinary skill in the art should know that the RNN has memory and can handle the time-dependent relationships in sequence data. The temperature-flow time-series interaction response coding vector is a data containing time-series information, which records the interaction characteristics of temperature and flow at different times. The RNN model can capture the potential law between the dynamic changes of temperature-flow and pressure by learning these sequence information. For example, in different well flushing stages, the change patterns of temperature and flow are different, and the RNN can predict the corresponding pressure changes according to these historical patterns. That is, the RNN model can establish a mapping relationship between temperature-flow and pressure, so as to predict the corresponding pressure value according to the current temperature-flow interaction characteristics, providing data support for more accurate judgment of whether there is an abnormality in the well flushing cycle in the future.

[0048] The following is a detailed elaboration of a specific implementation process of "inputting the temperature-flow time-series interaction response coding vector into a pressure simulator based on an RNN model to obtain the time queue of the inferred pressure values": First, it is necessary to construct a pressure simulator based on the RNN model. Determining the model hyperparameters is a key step in constructing the model, which includes the number of hidden layers, the number of neurons in each hidden layer, the learning rate, and the number of training epochs, etc. The values of these hyperparameters have a profound impact on the model performance, and usually, experiments and adjustments are needed to find the optimal values. Methods such as grid search or random search can be used to try different hyperparameter combinations within the set range and select the optimal model configuration according to the performance on the validation set. For example, by training the model under different combinations of the number of hidden layers and the learning rate, observing the change of the mean square error (MSE) on the validation set, so as to determine the best hyperparameter settings.

[0049] Next, model training is required. Model training is the core process for the pressure simulator to learn the relationship between temperature-flow data and pressure. Before conducting model training, a high-quality dataset must be prepared. This includes historical data of temperature, flow rate, and pressure collected from the oil and gas well site. These data should cover various situations under different working conditions so as to train a model that can adapt to various environmental changes. The collected data needs to be preprocessed, and the preprocessed data should be divided into a training set, a validation set, and a test set according to a certain proportion. The training set is used to train the model to let the model learn the inherent laws in the data; the validation set is used to evaluate the model performance during training, adjust the hyperparameters, and prevent overfitting; the test set is used to evaluate the generalization ability of the final model. Input the data of the training set into the constructed RNN model, and adjust the weights and parameters of the model with the help of the backpropagation algorithm to minimize the error (such as the mean square error MSE) between the predicted value of the model and the actual pressure value. During the training process, in order to accelerate model convergence and improve training efficiency, a suitable optimizer can be selected, such as Adam, SGD, etc. During the training process, use the data of the validation set to evaluate the performance metrics of the model (such as MSE, RMSE, etc.). If the performance of the model on the validation set no longer improves or even shows a downward trend, it indicates that overfitting may have occurred. At this time, training should be stopped and the current optimal model parameters should be saved.

[0050] Model validation is an important step to test the reliability of the model. Use the data of the test set to validate the trained model and evaluate the performance of the model on unseen data. By calculating the error metrics between the inferred pressure value predicted by the model and the actual pressure value in the test set, such as MSE, RMSE, MAPE (mean absolute percentage error), etc., the prediction accuracy of the model can be intuitively reflected. If the error metrics of the model on the test set meet the requirements of practical applications, it means that the model has good generalization ability and can be used for actual pressure inference; if the error is large, it is necessary to further adjust the model structure, hyperparameters or re-preprocess the data, and then re-train and validate the model.

[0051] After the above series of strict steps, when both the model training and validation achieve the expected results, the inference and prediction stage can be entered. The temperature-flow time-series interaction response coding vector obtained from the actual analysis is input into the trained and validated model. Based on the learned relationship between temperature-flow and pressure, the model infers and predicts the pressure value at each time point, thereby obtaining a time queue of inferred pressure values. In the actual application scenario, as new temperature-flow data continuously emerges, the updated temperature-flow time-series interaction response coding vector is continuously input into the model to update the time queue of inferred pressure values in real time. In this way, timely and accurate data support can be provided for judging whether there are abnormalities in the wash fluid circulation, and further, the smooth progress of the coiled tubing hot washing and paraffin and plug removal process for oil and gas wells can be ensured, which is beneficial to improving the production efficiency and safety of oil and gas wells.

[0052] Preferably, inputting the temperature-flow time-series interaction response coding vector into a pressure simulator based on an RNN model to obtain a time queue of inferred pressure values includes: calculating the low-order proximity and high-order similarity between the th eigenvalue and the th eigenvalue of the temperature-flow time-series interaction response coding vector, thereby obtaining a temperature-flow time-series interaction low-order proximity matrix and a temperature-flow time-series interaction high-order similarity matrix: ; where and respectively represent the th eigenvalue and the th eigenvalue of the temperature-flow time-series interaction response coding vector, represents the eigenvalue at the position of the temperature-flow time-series interaction low-order proximity matrix, represents the eigenvalue at the position of the temperature-flow time-series interaction high-order similarity matrix.

[0053] Multiply the temperature-flow time-series interaction response coding vector by the temperature-flow time-series interaction low-order proximity matrix and the temperature-flow time-series interaction high-order similarity matrix respectively to obtain a first temperature-flow time-series interaction response coding recurrence vector: ; where represents the temperature-flow time-series interaction low-order proximity matrix, represents the temperature-flow time-series interaction high-order similarity matrix, represents matrix multiplication, represents the temperature-flow time-series interaction response coding vector, represents the first temperature-flow time-series interaction response coding recurrence vector.

[0054] Multiply the temperature-flow time series interaction response encoding vector with the temperature-flow time series interaction response high-order similarity matrix and the temperature-flow time series interaction response low-order proximity matrix respectively to obtain a second temperature-flow time series interaction response encoding recursive vector: ; where represents the second temperature-flow time series interaction response encoding recursive vector.

[0055] Multiply the temperature-flow time series interaction response low-order proximity matrix with the self-correlation matrix of the first temperature-flow time series interaction response encoding recursive vector and the second temperature-flow time series interaction response encoding recursive vector to obtain a first temperature-flow time series interaction response sparse expansion matrix , where represents the transpose of a vector.

[0056] Multiply the temperature-flow time series interaction response high-order similarity matrix with the self-correlation matrix of the first temperature-flow time series interaction response encoding recursive vector and the second temperature-flow time series interaction response encoding recursive vector to obtain a second temperature-flow time series interaction response sparse expansion matrix .

[0057] After performing element-wise addition on the first temperature-flow time series interaction response sparse expansion matrix and the second temperature-flow time series interaction response sparse expansion matrix, multiply the result with the transpose vector of the temperature-flow time series interaction response encoding vector to obtain an optimized temperature-flow time series interaction response encoding vector ; where represents element-wise addition.

[0058] Input the optimized temperature-flow time series interaction response encoding vector into a pressure simulator based on an RNN model to obtain a time queue of inferred pressure values.

[0059] Here, when the temperature time series feature implicit encoding vector and the flow time series feature implicit encoding vector represent the time series implicit encoding features of temperature and pressure respectively, during the interaction response based on the feature forward and reverse attention fields, the insufficient reinforcement of the corresponding relationship of the forward and reverse attention fields caused by the source time series distribution difference will lead to the sparse correlation interaction response of the temperature-flow time series interaction response encoding vector. As a result, due to the lack of time series cyclic inference degree, the time series distribution accuracy of the time queue of the inferred pressure values obtained by inputting into the pressure simulator based on the RNN model is reduced, affecting the accuracy of the final determination result.

[0060] Therefore, for the temperature-flow time-series interaction response coding vector, its low-order proximity and high-order similarity matrix representation are used as multi-scale feature correlation quantization characterization. Through recursive optimization, the coupling of the time-delay auto-encoding characterization of the temperature-flow time-series interaction response coding vector among different correlation modes is simulated respectively, so as to simulate the sparse activation mechanism of the feature distribution neuron group, and the distributed sub-dimensions in the feature component space are used for prospective sparse constraint expansion to achieve the multi-granularity predictable expression of the temperature-flow time-series interaction response coding vector, effectively avoiding the time-series loop attenuation problem caused by the failure of joint inference, improving the time-series distribution accuracy of the inference pressure value obtained by inputting the temperature-flow time-series interaction response coding vector into the pressure simulator based on the RNN model, and improving the accuracy of the final determined result.

[0061] In step S36, based on the time queue of the pressure values and the time queue of the inference pressure values, it is determined whether there is a cyclic anomaly. Specifically, Figure 4 FIG. is a flowchart of step S36 in the coiled tubing hot washing paraffin removal and plugging removal process for oil and gas wells according to an embodiment of the present application. As Figure 4 shown, the step S36 includes: S361, calculating the difference between the pressure values and the inference pressure values at each time point in the time queue of the pressure values and the time queue of the inference pressure values to obtain a time queue of pressure deviation values; S362, determining whether there is a cyclic anomaly based on the mean and variance of the time queue of the pressure deviation values.

[0062] In step S361, the differences between the pressure values and the inferred pressure values at each time point in the time queue of the pressure values and the time queue of the inferred pressure values are calculated to obtain the time queue of the pressure deviation values. Accordingly, considering that the actually measured pressure value represents the real pressure condition during the well flushing process, while the inferred pressure value is a predicted value obtained through model simulation based on other parameters such as temperature and flow rate. Therefore, in order to more clearly and intuitively observe whether there are abnormal fluctuations in the pressure value, which may be signals of cyclic anomalies, in the technical solution of the present application, the differences between the pressure values and the inferred pressure values at each time point in the time queue of the pressure values and the time queue of the inferred pressure values are calculated to obtain the time queue of the pressure deviation values. And the obtained time queue of the pressure deviation values can be used as an important basis for detecting whether the well flushing cycle process is abnormal. Under normal circumstances, since the model is trained based on the physical laws of the process and a large amount of historical data, the inferred pressure value should be relatively close to the actual pressure value, and the pressure deviation value is within a small range and fluctuates smoothly. If abnormal situations such as local blockage of the well flushing pipeline or leakage of the well flushing fluid occur, the actual pressure value will change significantly, resulting in the pressure deviation value exceeding the normal range or showing abnormal fluctuations. By real-time monitoring the time queue of the pressure deviation values, these abnormal signals can be detected in time, providing early warnings for operators to take corresponding measures.

[0063] In step S362, based on the mean and variance of the time queue of the pressure deviation values, it is determined whether there is a cyclic anomaly. It should be understood that in order to quantify the degree of these differences and provide an objective basis for judging whether there is a cyclic anomaly, this application determines whether there is a cyclic anomaly based on the mean and variance of the time queue of the pressure deviation values. Specifically, the mean of the time queue of the pressure deviation values represents the average level of the sequence data and reflects the average deviation degree between the inferred pressure value and the actual pressure value. During a normal well flushing cycle, due to the stability of the system and the relative consistency of the process parameters, the pressure deviation values should fluctuate around a relatively small mean. If the mean deviates from the normal range, it indicates that there is a systematic deviation between the inferred pressure and the actual pressure, which may imply potential problems in the well flushing cycle system, such as systematic errors of measuring instruments, offsets of model parameters, etc. The variance measures the degree of dispersion of the pressure deviation values around the mean. A smaller variance indicates that the pressure deviation values are relatively stably distributed around the mean, indicating that the pressure changes during the well flushing process are relatively regular and the system operates stably. A larger variance, on the other hand, indicates that the pressure deviation values fluctuate greatly, and there may be some random or sudden factors affecting the well flushing cycle, such as the sudden movement of blockages in the wellbore, local instability of the flow of the well flushing fluid, etc. By setting reasonable threshold ranges for the mean and variance, and comparing the actually calculated mean and variance of the time queue of the pressure deviation values with them, it is possible to accurately judge whether there is an anomaly in the well flushing cycle. When the mean exceeds the normal range or the variance is too large, it can be determined that there may be a cyclic anomaly. This method of judgment based on the statistical characteristics of data has high accuracy and reliability, can avoid misjudgment caused by the fluctuation of a single data point, and provides clear anomaly signals for operators so that they can take measures in time for investigation and handling.

[0064] In summary, step S3 is clearly described. It uses data analysis and processing techniques based on deep learning to organize the data of temperature values, pressure values, and flow values to obtain the time queue of temperature values, the time queue of pressure values, and the time queue of flow values. Then, the time queues of temperature values and flow values are respectively encoded with time series features, and the encoded temperature time series features and flow time series features are correlated with a dual-perspective response. Based on this, the deviation between the time queue of the inferred pressure value obtained by pressure inference according to the temperature-flow time series interaction representation and the time queue of the pressure value is used to intelligently judge whether there is a cyclic anomaly. In this way, the accuracy of cyclic anomaly detection can be effectively improved.

[0065] In step S4, the decomposed blockage returns to the ground together with the well flushing fluid. It should be understood that through the circulation of the well flushing fluid, blockages such as wax blocks are melted, broken or loosened and can be carried back to the ground together with the well flushing fluid. This ensures that the blockage is effectively removed rather than just broken up and redeposited in other parts. Moreover, since the blockage is completely taken out of the well, the risk of secondary blockage caused by the redeposition of residues is reduced, which is beneficial to ensuring the long-term smoothness of the production string. That is to say, as the blockage is removed and broken, the circulation channel of the well flushing fluid gradually becomes smooth. The well flushing fluid that was originally blocked can flow more smoothly, forming a stable circulation. This is not only beneficial to continuously carrying out wax removal and blockage removal operations, but also ensures the normal flow of oil and gas in the pipe string, thereby improving the oil and gas production efficiency.

[0066] In summary, the continuous coiled tubing hot washing wax removal and blockage removal process for oil and gas wells based on the embodiments of the present application is elucidated. First, the well flushing fluid is heated by a hot oil truck, and then the heated well flushing fluid is injected into the coiled tubing workover rig by a pump truck. Subsequently, these well flushing fluids are sent into the well through the coiled tubing and circulate in the circulation channel in the well. During the circulation of the well flushing fluid, the water horsepower impact, heat exchange and mechanical crushing effects generated by the well flushing fluid are used to remove or break the blockage in the well. Finally, the decomposed blockage returns to the ground together with the well flushing fluid. In this way, the effect of wax removal and blockage removal in the oil and gas well can be improved, which is beneficial to better realizing the effective maintenance of the oil and gas well.

Claims

1. A process for hot washing, wax removal and plugging removal of coiled tubing in oil and gas wells, characterized in that: include: The well washing fluid is heated by a hot oil truck to obtain heated well washing fluid; The heated well-washing fluid is injected into the coiled tubing operation machine by a pump truck; The coiled tubing operation machine lowers the well-washing fluid into the well through the coiled tubing, and the well has a well-washing fluid circulation channel. The well-washing fluid can circulate in the well through the well-washing fluid circulation channel to remove or break the blockage by utilizing the water horsepower impact, heat energy exchange and mechanical crushing generated by the well-washing fluid during the circulation process, including: real-time monitoring of the state parameters of the well-washing fluid circulating in the well-washing fluid circulation channel to obtain a time queue of the state parameters, the state parameters including temperature value, pressure value and flow value; data sorting of the time queue of the state parameters to obtain a time queue of temperature value, a time queue of pressure value and a time queue of flow rate; time queue and time queue of flow value; sequence encoding the time queue of temperature value and the time queue of flow value to obtain temperature time series characteristics and flow time series characteristics; perform temperature-flow time series dual-view response correlation analysis on the temperature time series characteristics and the flow time series characteristics to obtain temperature-flow time series interactive response coding characteristics; perform pressure reasoning on the temperature-flow time series interactive response coding characteristics to obtain a time queue of inferred pressure value; determine whether there is a circulation abnormality based on the time queue of pressure value and the time queue of inferred pressure value; the decomposed blockage is returned to the ground together with the well washing fluid; Among them, pressure inference is performed on the temperature-flow timing interaction response coding feature to obtain a time queue of inferred pressure values, including: inputting the temperature-flow timing interaction response coding vector into a pressure simulator based on an RNN model to obtain a time queue of inferred pressure values.

2. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 1, characterized in that: The time queue of the temperature values ​​and the time queue of the flow values ​​are sequence-encoded to obtain temperature time series characteristics and flow time series characteristics, including: using a sequence encoder to sequence-encode the time queue of the temperature values ​​and the time queue of the flow values ​​respectively to obtain an implicit encoding vector of temperature time series characteristics as the temperature time series characteristics and an implicit encoding vector of flow time series characteristics as the flow time series characteristics.

3. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 2, characterized in that: The sequence encoder is a sequence encoder based on a bidirectional gated recurrent unit.

4. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 3, characterized in that: The temperature-flow time series dual-perspective response correlation analysis is performed on the temperature time series feature and the flow time series feature to obtain the temperature-flow time series interactive response coding feature, including: performing homography projection transformation on the temperature time series feature implicit coding vector and the flow time series feature implicit coding vector to obtain the temperature time series feature homography projection coding vector and the flow time series feature homography projection coding vector; calculating the dual-perspective temperature-flow time series feature attention score field between the temperature time series feature homography projection coding vector and the flow time series feature homography projection coding vector; based on the dual-perspective temperature-flow time series feature attention score field Viewing angle temperature-flow timing feature attention score field, modulating the temperature timing feature homography projection coding vector and the flow timing feature homography projection coding vector to obtain the temperature timing feature homography projection attention modulation coding vector and the flow timing feature homography projection attention modulation coding vector; calculating the temperature timing feature homography projection attention modulation coding vector and the flow timing feature homography projection attention modulation coding vector divided by the position point to obtain the temperature-flow timing interaction response coding vector as the temperature-flow timing interaction response coding feature.

5. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 4, characterized in that: The temperature time series feature implicit coding vector and the flow time series feature implicit coding vector are subjected to homography projection transformation to obtain the temperature time series feature homography projection coding vector and the flow time series feature homography projection coding vector, including: mapping the temperature time series feature implicit coding vector through a temperature time series feature mapping homography matrix to obtain the temperature time series feature homography projection coding vector; mapping the flow time series feature implicit coding vector through a flow time series feature mapping homography matrix to obtain the flow time series feature homography projection coding vector.

6. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 5, characterized in that: Calculate the dual-view temperature-flow timing feature attention score field between the temperature timing feature homography projection coding vector and the flow timing feature homography projection coding vector, including: calculating the forward temperature-flow timing feature attention score field of the temperature timing feature homography projection coding vector relative to the flow timing feature homography projection coding vector; calculating the reverse temperature-flow timing feature attention score field of the flow timing feature homography projection coding vector relative to the temperature timing feature homography projection coding vector; based on the forward temperature-flow timing feature attention score field and the reverse temperature-flow timing feature attention score field, obtain the dual-view temperature-flow timing feature attention score field.

7. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 6, characterized in that: Based on the dual-view temperature-flow timing feature attention score field, the temperature timing feature homography projection coding vector and the flow timing feature homography projection coding vector are modulated to obtain the temperature timing feature homography projection attention modulation coding vector and the flow timing feature homography projection attention modulation coding vector, including: multiplying the dual-view temperature-flow timing feature attention score field with the temperature timing feature homography projection coding vector and the flow timing feature homography projection coding vector respectively to obtain the temperature timing feature homography projection attention modulation coding vector and the flow timing feature homography projection attention modulation coding vector.

8. The process for hot-washing, wax-removing and plugging-removing of coiled tubing in oil and gas wells according to claim 7, characterized in that: Based on the time queue of the pressure values ​​and the time queue of the inferred pressure values, determining whether there is a circulation abnormality includes: calculating the difference between the pressure value and the inferred pressure value at each time point in the time queue of the pressure values ​​and the time queue of the inferred pressure values ​​to obtain a time queue of pressure deviation values; and determining whether there is a circulation abnormality based on the mean and variance of the time queue of the pressure deviation values.

Images