A downhole data acquisition and predictive analysis system for coiled tubing operations optimization

Abstract

The present application relates to the technical field of coiled tubing operation optimization, in particular to a downhole data acquisition and prediction analysis system for coiled tubing operation optimization; containing geological engineering baseline, operation risk prediction, dynamic model correction, optimization decision and closed-loop operation control unit; the system constructs the reference baseline by fusing static geology and planned engineering, and predicts the sand plug risk in real time; the core is that when the geological mutation is monitored, the geological index is calculated inversely and dynamically updated by using real-time engineering data to correct the reference baseline; and the optimal injection rate is solved by combining the forward-looking operation value function; the system executes hierarchical closed-loop control in response to different risk signals; the present application realizes the deep fusion and dynamic synchronization of geological-engineering data, and solves the problem of monitoring benchmark failure of static geological model under dynamic mutation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coiled tubing operation optimization, in particular to a downhole data acquisition and predictive analysis system for coiled tubing operation optimization. BACKGROUND

[0002] In the coiled tubing operation environment, the downhole geological conditions have high uncertainty, and real-time engineering parameters are generated; in order to ensure operation safety and efficiency, the existing technology generally adopts a mode of separating geological information from engineering operation, mainly relying on reactive monitoring of absolute values of engineering parameters; this scheme has significant defects: a static geological model cannot reflect dynamic geological mutations, resulting in invalid monitoring benchmarks; risk identification relies on threshold triggering, lacking early warning ability of early signs of risks such as sand plugging; operation decision relies on artificial experience, which is lagging and difficult to achieve optimal balance between efficiency and risk;

[0003] This mode leads to a sharp contradiction between operation efficiency and operation risk, making it difficult to achieve proactive risk avoidance; therefore, how to realize the deep integration of geological-engineering data, from reactive monitoring to predictive analysis, and closed-loop execution of proactive optimal economic decisions, has become a technical problem to be solved; SUMMARY

[0004] To solve the above technical problems, the present application provides a downhole data acquisition and predictive analysis system for coiled tubing operation optimization, in particular, the technical scheme of the present application comprises:

[0005] The geological engineering baseline unit is used to: acquire geological data to construct a static geomechanics index; and combine the index with the planned injection rate obtained to solve a reference pressure gradient baseline;

[0006] The operation risk prediction unit is used to: acquire engineering data to obtain real-time measured pressure gradient and real-time injection rate; calculate real-time pressure deviation and real-time pressure volatility; calculate the sand plugging risk probability through a preset operation risk prediction model; and identify the probability to output a safe operation area signal, a first warning area signal and a second risk area signal;

[0007] The dynamic model correction unit is used to: when it is determined that a geological mutation is encountered, the real-time geomechanics index is calculated; and the real-time geomechanics index is smoothed and filtered to generate a dynamically updated geological index, so as to correct the reference pressure gradient baseline;

[0008] The optimization decision unit is used to: obtain the economic benefit and the maximum risk cost rate of the unit injection rate; construct a proactive operation value function; and solve the function to obtain the optimal injection rate;

[0009] The closed-loop operation control unit is configured to perform a hierarchical closed-loop correction strategy on the optimal injection rate in response to the safe operation zone signal, the first pre-warning zone signal, and the second risk zone signal.

[0010] Preferably, the specific process of the geological engineering baseline unit is as follows:

[0011] Collecting geosteering logging geological data in the process of rotary steerable drilling;

[0012] Extracting key indicators related to the fracturability of the formation, and constructing a static geomechanics index;

[0013] Obtaining a planned injection rate, a fluid dynamics coefficient, and a geological influence weight;

[0014] And coupling the static geomechanics index, the planned injection rate, the fluid dynamics coefficient, and the geological influence weight, a reference pressure gradient baseline is calculated.

[0015] Preferably, the fluid dynamics coefficient is a physical constant determined by fluid mechanics formula according to the downhole fluid properties and the annular geometry; and the geological influence weight is an empirical coefficient calibrated by regression analysis on historical operation data.

[0016] Preferably, the process of the operation risk prediction unit for calculating the real-time pressure deviation and the real-time pressure fluctuation rate is as follows:

[0017] Subtracting the reference pressure gradient baseline from the real-time measured pressure gradient to obtain the real-time pressure deviation;

[0018] Calculating the rate of change of the real-time measured pressure gradient over time to obtain the real-time pressure fluctuation rate.

[0019] Preferably, the process of the discrimination probability for outputting the safe operation zone signal, the first pre-warning zone signal, and the second risk zone signal is as follows:

[0020] Comparing and analyzing the sand plug risk probability with the first preset risk threshold and the second preset risk threshold;

[0021] When the sand plug risk probability is less than or equal to the first preset risk threshold, a safe operation zone signal is generated;

[0022] When the sand plug risk probability is greater than the first preset risk threshold and less than or equal to the second preset risk threshold, a first pre-warning zone signal is generated;

[0023] When the sand plug risk probability is greater than the second preset risk threshold, a second risk zone signal is generated.

[0024] Preferably, the specific process of the dynamic model correction unit is as follows:

[0025] Monitoring real-time pressure deviation, when the real-time pressure deviation is continuously positive and the real-time injection rate is unchanged, it is determined that a geological mutation is encountered;

[0026] Starting data assimilation logic, calculating the real-time geomechanical index by inversely calculating the reference pressure gradient baseline model, using the real-time measured pressure gradient and the real-time injection rate;

[0027] Obtaining the geological index and the data assimilation coefficient at the last time;

[0028] And using the preset exponential moving average algorithm, combining the geological index at the last time, the real-time geomechanical index and the data assimilation coefficient, the dynamically updated geological index is calculated.

[0029] Preferably, the specific process of the optimization decision unit is as follows:

[0030] Obtaining the economic benefit and the maximum risk cost rate per unit injection rate;

[0031] Multiplying the economic benefit per unit injection rate by the injection rate to be decided to obtain the expected income rate;

[0032] Multiplying the maximum risk cost rate by the sand plug risk probability to obtain the expected loss rate;

[0033] Subtracting the expected loss rate from the expected income rate to construct a forward-looking operation value function;

[0034] And using a numerical optimization method to solve the forward-looking operation value function to obtain the optimal injection rate that maximizes the operation value rate.

[0035] Preferably, the hierarchical closed-loop correction strategy comprises:

[0036] In response to the safe operation area signal, the real-time injection rate is adjusted to the optimal injection rate to maximize the operation efficiency;

[0037] In response to the first warning area signal, it is determined that the risk is rising, the real-time injection rate is smoothly reduced to the optimal injection rate, and forward-looking avoidance is performed;

[0038] In response to the second risk area signal, it is determined that sand plug is about to occur, and the injection is immediately stopped to perform operation braking.

[0039] Compared with the prior art, the present application has the following beneficial effects:

[0040] 1、The system solves the problem that the geological information and the engineering operation are separated, the static geological model cannot reflect the dynamic geological mutation, and the monitoring benchmark is invalid, realizes the deep fusion and dynamic synchronization of geological-engineering data, by constructing a reference baseline integrating static geological data and planned engineering parameters, and dynamically updating the geological index by using real-time engineering data when a geological mutation is monitored.

[0041] 2、The system captures early warning features of risks by calculating the deviation of real-time pressure gradient relative to the dynamic reference baseline, and the rate of change of the gradient over time; based on these early warning features, the system calculates the probability value of sand plug risk through a preset model, realizes the transition from reactive monitoring relying on absolute value overrun of engineering parameters to predictive risk analysis based on early warning features, and improves early warning capability.

[0042] 3、The system solves the technical defects that the operation decision relies on artificial experience, has hysteresis, and is difficult to achieve optimal balance between efficiency and risk, realizes the unified optimization of economic benefits and operation risk, by constructing a forward-looking operation value function integrating the economic benefits of unit injection rate, the maximum risk cost rate, and the predicted sand plug risk probability, and automatically obtaining the optimal injection rate that maximizes the operation value by using a numerical optimization method.

[0043] 4、The system realizes a complete technical closed loop from risk prediction to optimal decision to automatic control, solves the problems of artificial decision lag and excessive conservatism, and realizes the automatic balance of risk avoidance and operation efficiency, by mapping the continuous sand plug risk probability into three discrete signals of safe operation area, first warning area, and second risk area, and automatically executing the hierarchical correction strategies including maximizing efficiency, smooth avoidance, and operation braking on the calculated optimal injection rate by the closed-loop operation control unit in response to these hierarchical signals. BRIEF DESCRIPTION OF DRAWINGS

[0044] The application will be further explained below in conjunction with the drawings and embodiments:

[0045] Figure 1 is a structural diagram of the system of the application. DETAILED DESCRIPTION

[0046] To make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below in conjunction with specific embodiments.

[0047] Example 1:

[0048] Please refer to Figure 1 A downhole data acquisition and predictive analysis system for coiled tubing operation optimization, comprising:

[0049] The geology-engineering baseline unit is configured to collect geology data to construct a static geomechanics index, and to calculate a reference pressure gradient baseline based on the index and a planned injection rate;

[0050] The operation risk prediction unit is configured to collect engineering data to obtain a real-time measured pressure gradient and a real-time injection rate, to calculate a real-time pressure deviation and a real-time pressure fluctuation rate, to calculate a sand plug risk probability based on a preset operation risk prediction model, and to determine the probability to output a safe operation area signal, a first-level early warning area signal, and a second-level risk area signal;

[0051] The dynamic model correction unit is configured to, when a geological mutation is determined, inversely calculate a real-time geomechanics index, and to smooth and filter the real-time geomechanics index to generate a dynamically updated geology index, so as to correct the reference pressure gradient baseline;

[0052] The optimization decision unit is configured to obtain an economic benefit and a maximum risk cost rate per unit injection rate, to construct a forward-looking operation value function, and to solve the function to obtain an optimal injection rate;

[0053] The closed-loop operation control unit is configured to, in response to the safe operation area signal, the first-level early warning area signal, and the second-level risk area signal, execute a hierarchical closed-loop correction strategy on the optimal injection rate.

[0054] The embodiment provides a downhole data collection and prediction analysis system for coiled tubing operation optimization, which aims to solve the technical defects of separation of geology information and engineering operation, dependence on reactive monitoring, and inability to avoid risks in a forward-looking manner in the prior art. The system realizes unified optimization of operation efficiency and operation risk by constructing a geology-engineering fusion baseline model, predicting operation risk in real time, dynamically correcting model parameters, and making forward-looking optimization decisions and closed-loop control.

[0055] The system comprises:

[0056] The geology-engineering baseline unit is configured to fuse static geology data and planned engineering parameters to establish an operation benchmark with physical meaning. In the embodiment, the unit is configured to collect geology data such as logging while drilling (LWD) data to construct a static geomechanics index representing the fracturability of a formation , and to calculate a nonlinear reference pressure gradient baseline based on the index and a planned injection rate obtained. The baseline is a zero-point reference for deviation detection and risk prediction of all subsequent modules, so that the system has forward-looking capability.

[0057] An operating risk prediction unit, which aims to convert downhole physical monitoring values into risk probability required for operation management; in this embodiment, the unit is used to: collect engineering data to obtain real-time measured pressure gradient and real-time injection rate ; calculate real-time pressure deviation by comparing ; calculate rate of change over time to obtain real-time pressure fluctuation rate ; calculate sand plug risk probability by a preset operating risk prediction model such as logistic function using and two precursor features; and determine the relationship between the probability and the preset threshold to output safe operation area signal, first-level early warning area signal and second-level risk area signal;

[0058] A dynamic model correction unit, which aims to solve the time scale conflict between static geological model and dynamic operation reality; when the system monitors that continuously positive while unchanged, it is determined that a geological mutation is encountered, indicating that the static has failed; at this time, the unit starts data assimilation logic to calculate real-time geomechanics index by back-calculation of baseline model using and ; and adopts exponential moving average EMA algorithm to smooth filter , generating a dynamically updated geological index to correct the reference pressure gradient baseline, ensuring that the baseline always reflects the real downhole geological conditions;

[0059] An optimization decision unit, which aims to actively solve the optimal economic decision under the current risk level; in this embodiment, the unit is used to: obtain operation parameters, including economic benefit of unit injection rate and maximum risk cost rate ; construct a forward-looking operation value function , which quantifies the operation value by expected income rate minus expected loss rate ; and adopts numerical optimization method such as gradient descent to solve the function to obtain the optimal injection rate that maximizes ;

[0060] ​A closed-loop operation control unit aims to automatically close the optimization decision to the downhole engineering execution; in this embodiment, the unit is used to: in response to the safe operation area signal, the first early warning area signal and the second risk area signal output by the operation risk prediction unit, the optimal injection rate calculated by the optimization decision unit The hierarchical closed-loop correction strategy is executed to realize the automatic balance of risk avoidance and efficiency optimization.

[0061] Through the cooperative work of the above five units, this embodiment constructs a complete technical closed loop of closed-loop operation control from geological-engineering baseline modeling to real-time risk prediction, to dynamic model correction and forward-looking optimization decision; the system deeply integrates geological data and engineering data, solves the sharp contradiction between operation efficiency and operation risk in the prior art, realizes the predictive operation optimization of geological-engineering integration, and significantly improves the safety and economy of coiled tubing operation.

[0062] Embodiment 2:

[0063] The specific process of the geological engineering baseline unit is as follows:

[0064] Acquire the logging-while-drilling geological data in the process of rotary steering drilling;

[0065] Extract the key indicators related to the fracturing property of the formation, and construct a static geomechanics index;

[0066] Obtain the planned injection rate, the fluid dynamics coefficient and the geological influence weight;

[0067] And couple the static geomechanics index, the planned injection rate, the fluid dynamics coefficient and the geological influence weight to calculate the reference pressure gradient baseline.

[0068] This embodiment is a specific embodiment of the geological engineering baseline unit in embodiment 1; the specific process of the geological engineering baseline unit is as follows:

[0069] The unit acquires the logging-while-drilling LWD geological data in the process of rotary steering drilling, wherein the logging-while-drilling geological data refers to the data reflecting the physical properties of the formation rock such as gamma, resistivity, density, porosity, etc. which are acquired in real time during drilling, and its function is to provide original input for geological modeling, and its source is the downhole LWD sensor;

[0070] Extract the key indicators related to the fracturing property of the formation, and construct a static geomechanics index, wherein the static geomechanics index Refers to a dimensionless comprehensive index for coiled tubing operation, which characterizes the fracturing property or receiving capacity of the formation, wherein is the fracture stage number; its function is to reduce the dimension of multiple LWD geological data parent features in different dimensions to a single lower feature that can be used for physical modeling; its source is, for example, constructed by processing LWD data through principal component analysis PCA or weighted average algorithm, etc.

[0071] The unit obtains the planned injection rate, the fluid dynamics coefficient, and the geological influence weight, wherein the planned injection rate refers to the fluid injection rate of the first stage set by the operator according to the construction design; the fluid dynamics coefficient and the geological influence weight are key physical parameters and empirical parameters for constructing the baseline model;

[0072] The unit couples the static geomechanical index, the planned injection rate, the fluid dynamics coefficient, and the geological influence weight to calculate the reference pressure gradient baseline; in this embodiment, the coupling process is realized through a reference pressure gradient baseline model, and the purpose is to fuse static geological data and planned engineering parameters before the operation to establish a nonlinear physical benchmark ; the specific form of the model is as follows:

[0073] ;

[0074] wherein, is the reference pressure gradient baseline, and its physical meaning is the normal pressure gradient corresponding to the injection at the first stage under the first rate, the dimension is Pa / m, and the source is calculated by the formula;

[0075] is the planned injection rate, and its physical meaning is the first stage planned injection rate, the dimension is m³ / s, and the source is set by the operation plan in advance;

[0076] is the static geomechanical index, and its physical meaning is the first stage static geomechanical index, the dimension is dimensionless, and the source is constructed from the LWD data in the previous step;

[0077] is the fluid dynamics coefficient, and its physical meaning is to quantify the contribution of fluid dynamics to the pressure, the dimension is Pa·s² / m 7 ;

[0078] Geology Impact Weight, which quantifies the contribution of geology index to pressure, dimension of Pa / m;

[0079] This embodiment explicitly constructs a reference pressure gradient baseline model, which mathematically couples traditionally separated static geology data and planned engineering parameters together; this resolves the defects of prior art which relies on experience or single engineering parameter for monitoring, and establishes an operation benchmark model with explicit physical dimension and geological basis, providing data basis for realizing the transition from reactive monitoring to predictive analysis.

[0080] Embodiment 3:

[0081] The fluid dynamics coefficient is a physical constant determined by the known formula of fluid mechanics according to the downhole fluid properties and annulus geometry; the geology impact weight is an empirical coefficient calibrated by regression analysis on historical operation data.

[0082] This embodiment is a further limitation of the two parameter sources in embodiment 2 and to ensure the implementability and accuracy of the baseline model;

[0083] The fluid dynamics coefficient is a physical constant determined by the known formula of fluid mechanics according to the downhole fluid properties such as density and viscosity and annulus geometry; here the physical constant refers to its value determined by the inherent physical properties of the system, not obtained by data fitting; this design ensures that the part representing fluid dynamics in the model has universal physical interpretability;

[0084] And the geology impact weight is an empirical coefficient calibrated by regression analysis on historical operation data such as adjacent wells or completed stages of the well; here the empirical coefficient refers to its value learned from historical data by statistical methods, which is used to quantify the actual contribution of dimensionless geology index to pressure gradient Pa / m; to further illustrate the calibration process, the static geomechanics index of the th completed stage in the historical data is denoted as , the actual measured pressure gradient corresponding to this stage is denoted as , and the actual injection rate of this stage is denoted as ; the calibration process is to apply least squares regression analysis to , on the premise that the value of has been determined according to the known formula of fluid mechanics, to The relationship is fitted to obtain the solution. This design allows the model to absorb operational experience from specific work areas, giving it adaptability.

[0085] This embodiment clearly distinguishes physical constants and The empirical coefficients represent the source of an organic combination of physical and empirical models; this allows the reference pressure gradient baseline model to strictly adhere to the laws of fluid dynamics. Guaranteeing, while also being able to flexibly adapt to the geological experience of different blocks. The calibration method significantly improves the accuracy and robustness of the baseline model compared to purely physical or purely data-driven models.

[0086] Example 4:

[0087] The operational risk prediction unit calculates real-time pressure deviation and real-time pressure volatility as follows:

[0088] The real-time pressure deviation is obtained by subtracting the reference pressure gradient baseline from the real-time measured pressure gradient.

[0089] The rate of change of the real-time measured pressure gradient over time is calculated to obtain the real-time pressure fluctuation rate.

[0090] This embodiment is a concretization of how the operational risk prediction unit in Embodiment 1 calculates key risk indicators; the calculation process is as follows:

[0091] The real-time pressure deviation is obtained by subtracting the reference pressure gradient baseline from the real-time measured pressure gradient; this real-time pressure deviation... The purpose is to quantify the degree of deviation of the current downhole state from the expected baseline, which is a core precursor to sand blockage risk; its calculation formula is:

[0092] ;

[0093] in, This is the real-time pressure deviation, measured in Pa / m.

[0094] To measure the pressure gradient in real time, its physical meaning is the current moment. The pressure gradient values ​​collected by downhole sensors are measured in Pa / m and are sourced from real-time downhole engineering data.

[0095] The dynamic reference pressure gradient baseline, whose physical meaning is the current time... The expected pressure benchmark, in Pa / m, is derived from the geological engineering baseline unit. It is calculated in real time and updated in real time by the dynamic model correction unit;

[0096] The rate of change of the real-time measured pressure gradient with respect to time is calculated to obtain a real-time pressure fluctuation rate; the real-time pressure fluctuation rate is to quantify the severity of the downhole state change, which is another key precursor of sand plug risk; in a discrete sampling system, the calculation formula is approximately:

[0097]

[0098] wherein, the real-time pressure fluctuation rate is (Pa / m) / s;

[0099] the real-time measured pressure gradient at the previous time is Pa / m;

[0100] the data sampling time interval is s;

[0101] The embodiment calculates the deviation state deviation and the fluctuation rate state change severity at the same time, capturing the precursor characteristics of sand plug risk from two orthogonal dimensions; compared with the traditional method of only monitoring whether the absolute value exceeds the limit, the present scheme can identify abnormal trends earlier and more sensitively, providing more abundant and reliable input features for subsequent probability prediction, and significantly improving the early warning ability of risk prediction.

[0102] Embodiment 5:

[0103] The process of determining the probability to output the safe operation area signal, the first-level warning area signal and the second-level risk area signal is as follows:

[0104] The sand plug risk probability is compared and analyzed with the first preset risk threshold and the second preset risk threshold;

[0105] When the sand plug risk probability is less than or equal to the first preset risk threshold, a safe operation area signal is generated;

[0106] When the sand plug risk probability is greater than the first preset risk threshold and less than or equal to the second preset risk threshold, a first-level warning area signal is generated;

[0107] When the sand plug risk probability is greater than the second preset risk threshold, a second-level risk area signal is generated.

[0108] The embodiment is a specific embodiment of how the operation risk prediction unit in embodiment 1 determines the risk level; by using an operation risk prediction model, for example, a logistic regression model using and as inputs ​​Calculate the probability of sand blockage risk Then, among the parameters , and The model coefficients are obtained by training a large historical database of operations containing both successful and unsuccessful sand blockage cases using machine learning methods such as logistic regression. The intercept term is dimensionless, and its physical meaning is the logarithmic probability of the base risk at zero deviation and zero volatility. The physical meaning of is to quantify the contribution of pressure deviation to the logarithmic probability of risk, and its dimension is . ; The physical meaning of is to quantify the contribution of pressure volatility to the logarithmic odds of risk, and its dimensions are... To clarify this calibration process, the first [item] in the historical database will be [used]. The pressure deviation of each sample is denoted as Pressure volatility is denoted as And record whether sand blockage has occurred as a label. The training objective is to find the optimal [training method]. and This makes the model output The prediction accuracy is highest before historical sand blockage events occur; the judgment process is as follows:

[0109] Risk of sand blockage Compare and analyze the results with the first and second preset risk thresholds;

[0110] The first and second preset risk thresholds here are decision parameters used to classify risk levels; in this embodiment, the first preset risk threshold can be set to 0.4 and the second preset risk threshold can be set to 0.8.

[0111] To clarify the logic, these thresholds of 0.4 and 0.8 are not arbitrarily assigned, but scientifically determined based on the machine learning training results of the operational risk prediction model and by analyzing the receiver operating characteristics ROC curve of the model. 0.4 is chosen as the warning point, i.e., the first threshold, to balance the sensitivity and false alarm rate of early detection. 0.8 is chosen as the high-risk point, i.e., the second threshold, to ensure a high degree of confidence when triggering avoidance actions such as operational braking.

[0112] The comparison logic is as follows:

[0113] When the risk of sand blockage Less than or equal to the first preset risk threshold At that time, a safe operating zone signal is generated;

[0114] When the risk of sand blockage Greater than the first preset risk threshold and less than or equal to the second preset risk threshold, i.e. At that time, a Level 1 warning zone signal is generated;

[0115] When the risk of sand blockage Greater than the second preset risk threshold At that time, a level-two risk zone signal is generated;

[0116] This embodiment introduces two risk thresholds scientifically calibrated based on ROC curves, dividing the continuous range between 0 and 1... The probability value is clearly mapped to three discrete risk levels: safety, early warning, and risk, each with a clear operational meaning. This hierarchical strategy enables refined risk management, allowing subsequent closed-loop operation control units to execute differentiated hierarchical correction strategies based on different signals. This avoids the traditional one-size-fits-all approach of alarm or shutdown, achieving a balance between safety and efficiency.

[0117] Example 6:

[0118] The specific process of the dynamic model correction unit is as follows:

[0119] Monitor real-time pressure deviation. When the real-time pressure deviation remains positive and the real-time injection rate does not change, it is determined that a geological change has occurred.

[0120] The data assimilation logic is initiated, and the real-time geomechanical index is calculated by back-calculating the reference pressure gradient baseline model and using real-time measured pressure gradient and real-time injection rate.

[0121] Obtain the geological index and data assimilation coefficient of the previous moment;

[0122] The system employs a pre-defined exponential moving average algorithm, combining the geological index from the previous moment, the real-time geomechanical index, and the data assimilation coefficient, to calculate dynamically updated geological indices.

[0123] This embodiment is a concretization of the dynamic model correction unit in Embodiment 1; the specific process of this unit is as follows:

[0124] Unit monitors real-time pressure deviation Calculated from the implementation method of Example 4; when the real-time pressure deviation For example, a duration exceeding 60 seconds and a real-time injection rate that is consistently positive. When nothing changes, the system determines that a geological upheaval has occurred; the basis for this logic is: with the injection rate remaining constant, a sustained and unexpected increase in pressure occurs. The most reasonable attribution is that the geological conditions underwent an unexpected change, i.e., a static change. Failure;

[0125] After triggering the geological abrupt change determination, the unit initiates the data assimilation logic; it then calculates the reference pressure gradient baseline model, as shown in Example 2. , using real-time measured pressure gradient and real-time injection rate , real-time geomechanical index is calculated ;

[0126] The real-time geomechanical index here refers to the actual geological index that can explain the current pressure response, which is back calculated using real-time measured values; its back calculation formula is as follows:

[0127] ;

[0128] wherein, is the real-time back calculated geological index, dimensionless;

[0129] is the real-time measured pressure gradient, from real-time acquisition of the sensor;

[0130] is the real-time injection rate, from real-time acquisition of engineering data;

[0131] is the calibration coefficient, from the same embodiment as in embodiment 3; is the preset minimum geological impact weight, a minimum value greater than 0, used to prevent model calculation overflow when the regression result approaches 0, to enhance the robustness of the model; The unit obtains the geological index at the previous time

[0132] and the data assimilation coefficient ;

[0133] The data assimilation coefficient here is a dimensionless adjustable parameter between and ; its value is based on: the setting of the signal-to-noise ratio SNR of the downhole sensor and the expected severity of geological changes; for example, when the signal-to-noise ratio is high or the expected geological change is gentle, a smaller such as 0.1-0.3, is taken to smooth the noise; in known geological mutation zones, the such as 0.6-0.8, can be temporarily increased to quickly track the geological changes;

[0134] The preset exponential moving average EMA algorithm is adopted, combined with the geological index at the previous time , the real-time geomechanical index and the data assimilation coefficient , to calculate the dynamically updated geological index ; the purpose of this algorithm is to smooth the back calculated​​ Smooth filtering is performed to avoid over-oscillation of the model caused by sensor noise; the formula is as follows:

[0135] ;

[0136] wherein, is the dynamically updated geological index;

[0137] The embodiment realizes dynamic correction of the geological model parameters through the logic of mutation identification, back calculation and smoothing assimilation; the system uses to replace the static to update the reference pressure gradient baseline, i.e. ; this solves the time scale conflict between static geological data and dynamic engineering data, ensures that the zero point baseline relied on by the risk prediction model always reflects the real downhole geological conditions, and greatly improves the prediction accuracy and adaptability of the system under complex geological conditions.

[0138] Embodiment 7:

[0139] The specific process of the optimization decision unit is as follows:

[0140] The economic benefit of the unit injection rate and the maximum risk cost rate are obtained;

[0141] The economic benefit of the unit injection rate is multiplied by the injection rate to be decided to obtain the expected income rate;

[0142] The maximum risk cost rate is multiplied by the sand plug risk probability to obtain the expected loss rate;

[0143] The expected income rate is subtracted from the expected loss rate to construct a forward-looking operation value function;

[0144] And a numerical optimization method is used to solve the forward-looking operation value function to obtain the optimal injection rate that maximizes the operation value rate.

[0145] The embodiment is a specific embodiment of the optimization decision unit in embodiment 1; the specific process of the unit is as follows:

[0146] The unit obtains the economic benefit of the unit injection rate and the maximum risk cost rate ;

[0147] The economic benefit of the unit injection rate here is an operation parameter, and its physical meaning is the economic benefit in monetary units / s brought by each increase in unit injection rate m³ / s, and its dimension is monetary units / m³; its value is based on: The operation data such as daily rate, construction period and material cost are input by the operator;

[0148] The maximum risk cost rate here is also an operating parameter, whose physical meaning is the equivalent operating loss rate when the sand plug risk probability is determined to occur, with the dimension of monetary unit / s; its value is based on: which can be estimated by the total loss of a single sand plug in historical data which is the historical calibration data, with the unit of monetary unit and the characteristic time leading to sand plug which is the historical calibration data, with the unit of s estimated, for example , or calibrated as a risk preference parameter of the operating strategy;

[0149] Unit construction forward-looking operating value function ; this function aims to unify the economic target income and risk cost loss in an optimization function;

[0150] The economic benefit of the unit injection rate is multiplied by the injection rate to be decided , to get the expected income rate, that is ;

[0151] The maximum risk cost rate is multiplied by the sand plug risk probability , to get the expected loss rate, that is ;

[0152] The expected income rate is subtracted from the expected loss rate to construct the forward-looking operating value function;

[0153] The complete form of this function is as follows:

[0154] ;

[0155] Among them, is the expected operating value rate, with the dimension of monetary unit / s;

[0156] is the injection rate to be decided, with the dimension of m³ / s;

[0157] as defined above;

[0158] is the forward-looking sand plug risk probability, with the dimension of dimensionless; this function is the connecting core of the entire system, which depends on the injection rate to be decided and the current geology from example 6 ;

[0159] Its calculation process is: the system passes through a forward-looking pressure response model To predict in the current geological Under the condition, the rate of execution of decisions The expected pressure gradient will be generated at that time In this embodiment, the forward-looking pressure response model The model uses the same structure as the dynamically updated reference pressure gradient baseline model in Example 6. This model integrates fluid dynamics and dynamic geological features, and its specific form is defined as follows:

[0160] ;

[0161] in, For fluid dynamics coefficients, The geological influence weight is derived from the same implementation method as in Example 3; The geological indices are currently dynamically updated from Example 6; The injection rate to be decided; this definition ensures consistency between the physical model used in the forward-looking prediction and the dynamically corrected system baseline.

[0162] The system utilizes this Calculate forward-looking risk characteristics:

[0163] Forward-looking pressure deviation ;

[0164] Forward-looking pressure volatility ;

[0165] These two forward-looking characteristics and Substituting the logistic function from Example 5, we obtain Thus, decision variables Risk Probability Mathematical coupling was performed;

[0166] Numerical optimization methods, such as gradient descent, are used to solve the forward-looking operational value function, resulting in the operational value rate. Maximize the optimal injection rate ;Right now ;

[0167] This embodiment constructs and solves a forward-looking operational value function to address downhole geological risks. Dynamic geological model Economic goals of ground operations Mathematical unification was performed; the optimal injection rate output by the system was [not specified]. It is a forward-looking, dynamic, and economically optimal decision-making instruction; this move resolves the conflict between the two goals of ensuring safety and improving efficiency in traditional operations, and realizes the transformation from passively weighing risks to actively seeking the optimal economic decision.

[0168] Example 8:

[0169] The hierarchical closed-loop correction strategy includes:

[0170] In response to a signal from the safe operating zone, the command adjusts the real-time injection rate to the optimal injection rate to maximize operational efficiency.

[0171] In response to the Level 1 warning zone signal, the risk is determined to be increased, and the real-time injection rate is smoothly reduced to the optimal injection rate to perform proactive avoidance.

[0172] In response to the Level 2 risk zone signal, it was determined that sand blockage was about to occur, and an instruction was given to immediately stop the injection and execute operational braking.

[0173] This embodiment is a concretization of the hierarchical closed-loop correction strategy executed by the closed-loop operation control unit in Embodiment 1; this strategy responds to the three signals generated in Embodiment 5 and executes the calculations in Embodiment 7. ;

[0174] This strategy includes:

[0175] In response to the signal in the safe operating area At this point, the system determines the risk to be extremely small, and the operational value function... Risk loss item Minimal; its decision-making logic lies in It will mainly consist of expected returns. The decision, calculated by the system This will tend towards the physical limits of the equipment or formation; therefore, the system issues instructions to increase the injection rate in real time. Adjust to typically increase to the optimal injection rate. To maximize operational efficiency;

[0176] In response to the Level 1 warning zone signal At this point, the system determines that the risk has increased, and the operational value function... Risk loss item It becomes significant; its decision-making logic lies in Optimal solution of the function It will automatically calculate a value lower than the current value. The value is determined to balance returns and risks; therefore, the system issues instructions to inject the rate in real time. Smoothly reduce to a new, lower optimal injection rate Implement proactive risk avoidance;

[0177] In response to the Level 2 risk zone signal At this point, the system determines that sand blockage is about to occur, and the operational value function... Risk loss item Predominates; its decision logic is in Optimized solution of function A value approaching 0 will be automatically calculated; therefore the system issues an instruction to stop injection immediately , execute operational braking to avoid sand plugging accidents;

[0178] In this embodiment, the risk level from embodiment 5 and the optimized decision from embodiment 7 are effectively linked through the above hierarchical closed-loop correction strategy; this strategy realizes differentiated, automated and closed-loop responses for different risk levels: automatically pursuing efficiency when safe, automatically anticipating and avoiding in advance when on alert, and automatically executing operational braking when in danger; this effectively solves the problem of manual decision lag, excessive conservatism or excessive aggressiveness, and maximizes operational efficiency under the premise of ensuring job safety.

[0179] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A downhole data acquisition and predictive analysis system for coiled tubing job optimization, characterized by, The geological engineering baseline unit is configured to collect geological data to construct a static geomechanics index, and to calculate a reference pressure gradient baseline in combination with the index and a planned injection rate. The operating risk prediction unit is configured to collect engineering data to obtain a real-time measured pressure gradient and a real-time injection rate, to calculate a real-time pressure deviation and a real-time pressure fluctuation rate, and to calculate a sand plug risk probability by using a preset operating risk prediction model. The dynamic model correction unit is configured to recalculate a real-time geomechanics index when a geological mutation is determined, to smooth and filter the real-time geomechanics index, and to generate a dynamically updated geomechanics index to correct the reference pressure gradient baseline. The optimization decision unit is configured to obtain an economic benefit per unit injection rate and a maximum risk cost rate, to construct a forward-looking operating value function, and to solve the function to obtain an optimal injection rate. The specific process of the optimization decision unit is as follows: The preset operation risk prediction model is a logistic regression model, and a formula is: wherein, is a sand plug risk probability, is a real-time pressure deviation, is a real-time pressure fluctuation rate, , , is a model coefficient; and the discrimination probability outputs a safe operation area signal, a first-level early warning area signal and a second-level risk area signal; An economic benefit per unit injection rate and a maximum risk cost rate are obtained. The economic benefit per unit injection rate is multiplied by a to-be-decided injection rate to obtain an expected income rate. The maximum risk cost rate is multiplied by the sand plug risk probability to obtain an expected loss rate. The expected income rate is subtracted from the expected loss rate to construct a forward-looking operating value function. A numerical optimization method is used to solve the forward-looking operating value function to obtain an optimal injection rate that maximizes the operating value rate. The closed-loop operation control unit is configured to respond to a safe operation area signal, a first-level early warning area signal, and a second-level risk area signal to perform a hierarchical closed-loop correction strategy on the optimal injection rate. The specific process of the geological engineering baseline unit is as follows: Geological data during a while-drilling process of a rotary steerable drilling is collected.

2. A downhole data acquisition and predictive analysis system for coiled tubing job optimization according to claim 1, characterized in that, Key indicators related to formation fracturability are extracted to construct a static geomechanics index. A planned injection rate, a fluid dynamics coefficient, and a geology impact weight are obtained. The static geomechanics index, the planned injection rate, the fluid dynamics coefficient, and the geology impact weight are coupled to calculate a reference pressure gradient baseline. The fluid dynamics coefficient is a physical constant determined by fluid mechanics formulas according to downhole fluid properties and annular geometry, and the geology impact weight is an empirical coefficient calibrated by regression analysis on historical operation data. The specific process of the operating risk prediction unit to calculate a real-time pressure deviation and a real-time pressure fluctuation rate is as follows:

3. A downhole data acquisition and predictive analysis system for coiled tubing job optimization according to claim 2, characterized in that, A real-time pressure deviation is obtained by subtracting the reference pressure gradient baseline from the real-time measured pressure gradient.

4. The downhole data acquisition and predictive analysis system for coiled tubing job optimization of claim 1, wherein, A real-time pressure fluctuation rate is obtained by calculating a rate of change of the real-time measured pressure gradient over time. The process of distinguishing the probability to output a safe operation area signal, a first-level early warning area signal, and a second-level risk area signal is as follows: The sand plug risk probability is compared with a first preset risk threshold and a second preset risk threshold.

5. The downhole data acquisition and predictive analysis system for coiled tubing job optimization of claim 1, wherein, When the sand plug risk probability is less than or equal to the first preset risk threshold, a safe operation area signal is generated. When the sand plug risk probability is greater than the first preset risk threshold and less than or equal to the second preset risk threshold, a first-level early warning area signal is generated. When the sand plug risk probability is greater than the second preset risk threshold, a second-level risk area signal is generated. The specific process of the dynamic model correction unit is as follows: ​ 6. The downhole data acquisition and predictive analysis system for coiled tubing job optimization of claim 1, wherein, ​ Monitoring real-time pressure deviation, when the real-time pressure deviation is continuously positive and the real-time injection rate is unchanged, it is determined that a geological mutation is encountered; Starting data assimilation logic, by inversely calculating the reference pressure gradient baseline model, using the real-time measured pressure gradient and the real-time injection rate, the real-time geomechanical index is calculated; Obtaining the geological index and the data assimilation coefficient at the last time; And using the preset exponential moving average algorithm, combining the geological index at the last time, the real-time geomechanical index and the data assimilation coefficient, the dynamically updated geological index is calculated.

7. The downhole data acquisition and predictive analysis system for coiled tubing job optimization of claim 1 or 5, wherein, The hierarchical closed-loop correction strategy includes: In response to the safe operation area signal, the real-time injection rate is adjusted to the optimal injection rate to maximize operational efficiency; In response to the first-level early warning area signal, it is determined that the risk is rising, and the real-time injection rate is smoothly reduced to the optimal injection rate to perform forward-looking avoidance; In response to the second-level risk area signal, it is determined that sand plugging is about to occur, and the injection is immediately stopped to perform operational braking.

Images