Heavy oil recovery electric heating power self-adaptive control method based on underground temperature feedback

Abstract

The invention discloses a thickened oil recovery electric heating power self-adaptive control method based on underground temperature feedback, and particularly relates to the field of thickened oil recovery intelligent control. Wellbore heating sections are divided according to the adjustable interval of 0.5-2 m, combined data are synchronously collected through a Pt100 sensor group and a viscosity sensor which are annularly arranged, and the combined data are stored in a database; generating a temperature feature vector and a viscosity sequential sequence; constructing a dynamic weight matrix through fuzzy similarity matching, and generating a temperature control compensation gain in combination with the curvature of the viscosity-temperature fitting curve; adopting a main and auxiliary coil cooperation and neighbor cell compensation strategy to output a segmented power instruction, performing closed-loop correction on a weight matrix through a PI algorithm, and iteratively updating the instruction until the viscosity is stabilized at 1500-3000 mPas; according to the method, accurate sensing of underground temperature and dynamic adaptation of power are achieved, the heating energy efficiency is improved, the problem of overheating or underheating is avoided, and the method adapts to different heavy oil working conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for heavy oil extraction, and more specifically, to an adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback. Background Technology

[0002] Heavy oil, due to its high viscosity and poor fluidity, requires electric heating technology to increase wellbore temperature to reduce viscosity and ensure lift efficiency. Existing electric heating control technology has the following main drawbacks: Temperature monitoring is incomplete: it often uses single-point or local temperature measurement, which cannot capture the three-dimensional temperature gradient changes downhole, and is prone to local overheating and coking or underheating and sticking. The regulation response is hysteretic: it relies on fixed parameter control, has poor adaptability to the nonlinear changes in the viscosity of heavy oil with temperature, and lacks the ability to dynamically adjust the compensation gain; Insufficient energy efficiency: Power distribution relies on single adjustment of the main coil, redundant capacity is not fully utilized, and there is no neighboring area collaborative compensation mechanism, resulting in energy waste.

[0003] For example, the oilfield electric heating vacuum phase change heater disclosed in Chinese patent CN210135689U can monitor the inlet and outlet temperatures and adjust the power according to the temperature difference, but it does not consider the dynamic changes in viscosity and the influence of thermal inertia, resulting in limited control accuracy. The intelligent control system in Chinese patent CN202220172215.1 relies on manually setting power limits and lacks theoretical optimization basis. Therefore, there is an urgent need to develop an adaptive control method based on multi-dimensional temperature feedback to improve the temperature control accuracy and energy efficiency of heavy oil extraction. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A1: Divide the wellbore longitudinally into N independent heating sections at intervals of 0.5-2m. Temperature data for each section is collected by a Pt100 platinum resistance thermometer group with a ring topology layout. Simultaneously acquire the viscosity-temperature combination data of heavy oil in each section to generate a temperature feature vector and viscosity time series. A2: Perform fuzzy similarity matching between the temperature feature vector and the preset heat flow distribution pattern library, calculate the thermal inertia imbalance coefficient of each section, construct a dynamic weight matrix, and the matrix element values ​​represent thermal inertia overload, underload and steady state. A3: Perform curve fitting on the viscosity-temperature combination data, calculate the curvature value of the fitted curve corresponding to the current temperature, and dynamically generate temperature control compensation gain by combining the thermal inertia imbalance coefficient. A4: Based on the dynamic weight matrix and temperature control compensation gain output segmented power regulation command, the redundant power of the main and auxiliary coils is called first. When the viscosity improvement rate does not reach the preset standard, the adjacent section compensation heating is started. A5: Collect the viscosity data of the heated heavy oil, correct the dynamic weight matrix through the proportional-integral algorithm, and iteratively update the segmented power control command until the viscosity enters the preset steady-state range.

[0006] Preferably, in A1, in the longitudinal direction of the heavy oil production wellbore, based on the characteristics of downhole geological stratification, the viscosity distribution law of heavy oil, and the principle of optimal heating energy efficiency, N independent heating sections are uniformly divided at adjustable intervals of 0.5-2m. Among them, the shallow heavy oil low viscosity area is divided at intervals of 1.5-2m, and the deep high viscosity area is divided at intervals of 0.5-1m to ensure accurate coverage of key viscosity-sensitive sections. Each independent heating section is equipped with a ring-shaped Pt100 platinum resistance thermometer sensor group. The sensor group consists of four high-precision temperature probes, which are evenly distributed at equal angles along the circumference of the shaft, with adjacent probes having an angle of 90°-120°, forming a 360° temperature monitoring network. This network can simultaneously capture temperature differences at different locations along the circumference of the section, avoiding local temperature misjudgments caused by single-point temperature measurement. At the same time, a high-frequency response viscosity sensor is deployed in the middle of each section. This sensor uses the same data acquisition sequence as the temperature sensor group and synchronously acquires, stores, and transmits temperature and viscosity data through a mine explosion-proof data acquisition module. During the acquisition process, it is ensured that the temperature-viscosity data at the same time and location correspond one-to-one, forming a complete viscosity-temperature combined dataset.

[0007] Preferably, in A2, the ground control center has a "preset heat flow distribution pattern library" built based on geological models and historical production data; the pattern library contains ideal temperature field distribution patterns under various typical working conditions, such as "uniform heating mode", "high-low gradient mode", "central key heating mode", etc., and each mode is represented by a standardized N-dimensional temperature feature vector template. In real-time control, the system will use the actual temperature feature vectors of the N segments currently collected and constructed ( , ) and the various template vectors in the pattern library ( , The system performs fuzzy similarity matching calculations. The matching process employs a comprehensive similarity metric algorithm combining weighted Euclidean distance and cosine similarity, and introduces a fuzzy membership function to handle the uncertainty and transitional state of the downhole temperature field. Through fuzzy similarity matching, the system identifies which preset mode the current overall temperature field is closest to, and calculates the optimal matching mode for each independent heating segment i. "State deviation" of the corresponding section ; Based on this state deviation and the rate of temperature change in this section The system further calculates the thermal inertia imbalance coefficient. The formula for calculating the thermal inertia imbalance coefficient is: , in, and These are weighting coefficients determined based on the thermal properties of the formation. This is expressed as the expected rate of temperature change for this section under this model.

[0008] Preferably, in A3, the ground control center synchronously acquires viscosity-temperature combined data for each independent heating section i. Real-time processing is performed; firstly, least-squares curve fitting is performed using an Arrhenius-type function to establish a localized viscosity-temperature relationship model for this section under the current operating conditions; a commonly used fitting model is the improved Andreid equation, and the specific calculation method is as follows: , in, Let the viscosity function be expressed as the viscosity function of the i-th segment. Expressed as a proportionality constant, Represented as activation energy type parameters, This is represented as the viscosity-based offset correction term, where T represents temperature; The system calculates at the current real-time temperature The curvature value of the fitted curve is calculated using the following method: , in, Represented as curvature value, Represented as a fitting function The first derivative with respect to temperature Represented as a fitting function The second derivative with respect to temperature; the larger the curvature value, the more likely that temperature changes will cause viscosity changes near the current temperature point, meaning that this section is in the "critical window period" for viscosity reduction.

[0009] Preferably, in A4, based on the dynamic weight matrix constructed above and the real-time generated temperature control compensation gain, the basic power control command calculation for each independent heating section is first completed, that is, the power distribution is precisely matched with the thermal inertia state and viscosity temperature sensitivity characteristics of the section. The element values ​​in the dynamic weight matrix that represent overload, underload or steady state directly determine the adjustment direction of the basic power, while the temperature control compensation gain amplifies or fine-tunes the basic power according to the viscosity's sensitivity to temperature. At the power execution level, a main and auxiliary coil collaborative working mode is adopted, prioritizing the use of coil redundant power to improve energy efficiency. Each heating section is equipped with one main heating coil and two auxiliary heating coils. The main coil undertakes the basic heating load, while the auxiliary coils serve as redundant capacity for backup. Both the main and auxiliary coils have a dynamic adjustment space of 20%-30% of their rated power. When the basic power control command is issued, the system first detects the current load rate of the main coil: if the main coil load rate is below 70%, the power adjustment is completed first through the redundant capacity of the main coil to avoid energy loss caused by frequent start-stop of the auxiliary coils; if the main coil load rate has reached 85% or more, and the basic power still does not meet the control requirements, one auxiliary coil is gradually activated to accurately supplement energy through the power superposition of the main and auxiliary coils. At the same time, the coil operating temperature and insulation status are monitored in real time to ensure operational safety.

[0010] Preferably, in A5, after the main and auxiliary coil heating and the adjacent section compensation heating are executed, the system synchronously collects the heavy oil viscosity data of each heating section according to a preset collection cycle (5-10 minutes / time). The collection process is aligned with the temperature data collection to ensure that each set of viscosity data corresponds to a clear power control condition. The collected viscosity data needs to be validated: outliers caused by fluid disturbance and instantaneous sensor drift are removed (using the moving average filtering method, the average of 3 consecutive samples is taken as valid data), and the data is compared with the initial viscosity value before power adjustment to preliminarily judge the actual effect of power control.

[0011] Based on verified effective viscosity data, a proportional-integral (PI) algorithm is introduced to perform closed-loop correction on the dynamic weight matrix, eliminating the deviation between the current viscosity and the preset steady-state viscosity, and improving the accuracy of the weight matrix in adapting to actual working conditions. The correction logic of the PI algorithm revolves around viscosity deviation: first, the absolute deviation between the real-time viscosity and the preset steady-state viscosity is calculated; the proportional stage (P stage) outputs the correction amount in real time according to the magnitude of the deviation, and quickly adjusts the weight matrix to avoid deviation accumulation; the integral stage (I stage) performs time integration on the deviation to compensate for the static deviation that the proportional stage cannot completely eliminate, improve the stability and accuracy of the correction, and prevent system oscillation; during the correction process, the proportional coefficient and integral coefficient of the PI algorithm need to be dynamically adapted according to the viscosity range of heavy oil. The dynamic weight matrix is ​​updated row by row in segments: For each heating segment, the correction amount output by the PI algorithm is proportionally distributed to each element of the corresponding row of the matrix. The distribution ratio is positively correlated with the original fuzzy similarity value of the element. That is, the element with a higher degree of fit with the current dominant heat flow mode will receive a larger correction weight, ensuring that the matrix update always revolves around the core thermal inertia state. The updated matrix is ​​then normalized. Based on the corrected dynamic weight matrix, the system recalculates the temperature control compensation gain and then iteratively updates the segmented power control commands for each section, forming a closed-loop iterative link of "power adjustment - viscosity feedback - matrix correction - command update". After each iteration, the system continuously monitors the changing trend of viscosity data until the preset steady-state judgment condition is met: within three consecutive acquisition cycles, the fluctuation range of heavy oil viscosity in each section is ≤5%, and the viscosity value is stably within the preset reasonable steady-state range (set to 1500-3000 mPa·s). If abnormal viscosity fluctuation or continuous increase in deviation occurs during the iteration process, the system will automatically trigger the emergency adjustment mechanism, suspend the current iteration and restore to the previous effective power command, and issue an operating condition warning signal to prompt maintenance personnel to check potential influencing factors such as sensor status and downhole fluid characteristics to ensure the reliability and safety of closed-loop control.

[0012] The technical effects and advantages of this invention are as follows: This invention first divides the wellbore longitudinally into independent heating sections, simultaneously collecting temperature and viscosity data. Second, it calculates the thermal inertia imbalance coefficient of each section by matching a preset heat flow mode, constructing a dynamic weight matrix. Then, it fits the viscosity-temperature curve to calculate curvature, dynamically generating temperature control compensation gain based on the imbalance coefficient. Next, based on the aforementioned matrix and gain, it outputs segmented power commands, prioritizing the use of redundant power from the main and auxiliary coils, and initiating compensation heating in adjacent sections when viscosity improvement is insufficient. Finally, it collects viscosity data after heating, uses a proportional-integral algorithm to correct the dynamic weight matrix in a closed loop, iteratively updating the power commands until the heavy oil viscosity stabilizes within a preset range. This invention achieves precise downhole temperature sensing and dynamic power adaptation, improving heating efficiency, avoiding overheating or underheating problems, and through closed-loop iterative calibration, can adapt to the needs of heavy oil extraction with different water cuts and viscosity grades. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0014] Figure 2 This is a schematic diagram of the compensation gain generation of the present invention.

[0015] Figure 3 This is a schematic diagram illustrating the power regulation process of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 As shown, this invention provides an adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback, comprising: A1: Divide the wellbore longitudinally into N independent heating sections at intervals of 0.5-2m. Temperature data for each section is collected by a Pt100 platinum resistance thermometer group with a ring topology layout. Simultaneously acquire the viscosity-temperature combination data of heavy oil in each section to generate a temperature feature vector and viscosity time series. In A1, along the longitudinal direction of the heavy oil production wellbore, based on the characteristics of downhole geological stratification, the viscosity distribution of heavy oil, and the principle of optimal heating efficiency, N independent heating zones are uniformly divided at adjustable intervals of 0.5-2m. Among them, shallow areas with lower heavy oil viscosity are divided at intervals of 1.5-2m, while deep areas with high viscosity are divided at intervals of 0.5-1m to ensure accurate coverage of key viscosity-sensitive sections. Each independent heating section is equipped with a ring-shaped Pt100 platinum resistance thermometer sensor group. The sensor group consists of four high-precision temperature probes, which are evenly distributed at equal angles along the circumference of the shaft. The angle between adjacent probes is 90°-120°, forming a 360° temperature monitoring network. This network can simultaneously capture temperature differences at different locations along the circumference of the section, avoiding local temperature misjudgments caused by single-point temperature measurement. At the same time, a high-frequency response viscosity sensor is deployed in the middle of each section. This sensor and the temperature sensor group use the same data acquisition sequence. Temperature and viscosity data are synchronously acquired, stored, and transmitted through a mine explosion-proof data acquisition module. During the acquisition process, the temperature-viscosity data at the same time and spatial location are guaranteed to correspond one-to-one, forming a complete viscosity-temperature combined dataset. After data acquisition, the temperature data of each segment is preprocessed to remove abnormal fluctuations and perform smoothing filtering. Core parameters such as the average temperature of the segment, the maximum circumferential temperature difference, the rate of temperature rise, and the temperature gradient are extracted to construct a temperature feature vector that characterizes the temperature field distribution of the segment. The viscosity data is sorted according to the time series to form a continuous viscosity time series.

[0018] A2: Perform fuzzy similarity matching between the temperature feature vector and the preset heat flow distribution pattern library, calculate the thermal inertia imbalance coefficient of each section, construct a dynamic weight matrix, and the matrix element values ​​represent thermal inertia overload, underload or steady state. In A2, the ground control center has a "pre-set heat flow distribution pattern library" built based on geological models and historical production data. This pattern library contains ideal temperature field distribution patterns under various typical working conditions, such as "uniform heating mode", "high-low gradient mode", and "central key heating mode". Each mode is represented by a standardized N-dimensional temperature feature vector template. In real-time control, the system will use the actual temperature feature vectors of the N segments currently collected and constructed ( , ) and the various template vectors in the pattern library ( , The system performs fuzzy similarity matching calculations. The matching process employs a comprehensive similarity metric algorithm combining weighted Euclidean distance and cosine similarity, and introduces a fuzzy membership function to handle the uncertainty and transitional state of the downhole temperature field. Through fuzzy similarity matching, the system identifies which preset mode the current overall temperature field is closest to, and calculates the optimal matching mode for each independent heating segment i. "State deviation" of the corresponding section ; Based on this state deviation and the rate of temperature change in this section The system further calculates the thermal inertia imbalance coefficient. The formula for calculating the thermal inertia imbalance coefficient is: , in, and These are weighting coefficients determined based on the thermal properties of the formation, with values ​​ranging from 0 to 1. This is expressed as the expected rate of temperature change in this section under this model; It quantifies the degree of dynamic heat accumulation or heat deficit in this section due to the mismatch between heat input and the formation's heat absorption / fluid heat carrying capacity.

[0019] when This indicates thermal inertia overload (i.e., excessive heat input, posing a risk of overheating or wasting thermal energy). This indicates thermal inertia underload (insufficient heat input, slow temperature rise or failure to meet viscosity reduction requirements). This indicates a steady state of thermal inertia (basic balance between heat supply and demand). in, This is represented as the overload threshold. This is represented as the underload threshold. This is expressed as the steady-state tolerance threshold. Finally, the system constructs an N×N dynamic weight matrix W for differentiated allocation in global control commands; the off-diagonal elements of this matrix are zero (assuming inter-segment thermal coupling has been considered in the model library), and the diagonal element w_ii is the coefficient of thermal inertia imbalance. The weight values ​​obtained from the mapping: The mapping function f(·) is designed as follows: for overloaded sections, a lower control weight (or even a negative feedback weight) is assigned to prioritize suppressing its power and prevent overheating; for underloaded sections, a higher positive control weight is assigned to improve its power distribution and accelerate temperature rise; for steady-state sections, a baseline weight is assigned for maintenance fine-tuning; the real-time construction of this dynamic weight matrix W enables the control system to accurately sense and respond to non-uniform thermal demand in the downhole vertical direction and allocate thermal energy to the most needed sections.

[0020] A3: Perform curve fitting on the viscosity-temperature combination data, calculate the curvature value of the fitted curve corresponding to the current temperature, and dynamically generate temperature control compensation gain by combining the thermal inertia imbalance coefficient. In A3, the ground control center synchronously acquires viscosity-temperature combined data for each independent heating section i. Real-time processing is performed; firstly, least-squares curve fitting is performed using an Arrhenius-type function to establish a localized viscosity-temperature relationship model for this section under the current operating conditions; a commonly used fitting model is the improved Andreid equation, and the specific calculation method is as follows: , in, Let the viscosity function be expressed as the viscosity function of the i-th segment. Expressed as a proportionality constant, Represented as activation energy type parameters, This is represented as the viscosity-based offset correction term, where T represents temperature; The system calculates at the current real-time temperature The curvature value of the fitted curve is calculated using the following method: , in, Represented as curvature value, Represented as a fitting function The first derivative with respect to temperature Represented as a fitting function The second derivative with respect to temperature; the larger the curvature value, the more likely that temperature changes will cause viscosity changes near the current temperature point, meaning that this section is in the "critical window period" for viscosity reduction; The curvature value is fused with the previously obtained thermal inertia imbalance coefficient to dynamically generate the temperature control compensation gain for this section. The specific calculation method is as follows: , in, Represented as the reference control gain, , This is expressed as an adjustment parameter related to the curvature effect. The function maps the effect of curvature to Nonlinear amplification is performed. This is represented as an adjustment parameter. Represented as a symbolic function, This is expressed as the saturation threshold of the thermal inertia imbalance coefficient; Finally, the dynamically generated baseline control gain is incorporated into the proportional or integral term of the control loop for that section to achieve zoned temperature control.

[0021] A4: Based on the dynamic weight matrix and temperature control compensation gain output segmented power regulation command, the redundant power of the main and auxiliary coils is called first. When the viscosity improvement rate does not reach the preset standard, the adjacent section compensation heating is started. In A4, based on the dynamic weight matrix constructed above and the real-time generated temperature control compensation gain, the basic power control command calculation for each independent heating section is first completed, that is, the power distribution is precisely matched with the thermal inertia state and viscosity temperature sensitivity characteristics of the section. The element values ​​in the dynamic weight matrix that represent overload, underload or steady state directly determine the adjustment direction of the basic power, while the temperature control compensation gain amplifies or fine-tunes the basic power according to the viscosity's sensitivity to temperature. At the power execution level, a main and auxiliary coil collaborative working mode is adopted, prioritizing the use of coil redundant power to improve energy efficiency. Each heating section is equipped with one main heating coil and two auxiliary heating coils. The main coil undertakes the basic heating load, while the auxiliary coils serve as redundant capacity for backup. Both the main and auxiliary coils have a dynamic adjustment space of 20%-30% of their rated power. When the basic power control command is issued, the system first detects the current load rate of the main coil: if the main coil load rate is below 70%, the power adjustment is completed first through the redundant capacity of the main coil to avoid energy loss caused by frequent start-stop of the auxiliary coils; if the main coil load rate has reached 85% or more, and the basic power still does not meet the control requirements, one auxiliary coil is gradually activated to accurately supplement energy through the power superposition of the main and auxiliary coils. At the same time, the coil operating temperature and insulation status are monitored in real time to ensure operational safety.

[0022] After power adjustment, the system enters the viscosity improvement effect evaluation stage to determine whether the current power can effectively reduce the viscosity of heavy oil. The evaluation cycle is consistent with the temperature data acquisition cycle. The effect is quantified by calculating the viscosity improvement rate between two adjacent cycles. The viscosity improvement rate is the percentage of the difference between the viscosity of the later cycle and the viscosity of the previous cycle relative to the viscosity of the previous cycle. The preset viscosity improvement rate standards are: ≥8% when the viscosity of heavy oil is ≥5000 mPa·s (high viscosity range); ≥5% when the viscosity is in the 2000-5000 mPa·s medium viscosity range; and ≥3% when the viscosity is in the <2000 mPa·s low viscosity range. If the viscosity improvement rate does not reach the preset standard for the corresponding range in two consecutive evaluation cycles, and the main and auxiliary coils are already operating at more than 90% of the rated power, it indicates that the heating capacity of the current section itself cannot meet the demand, and the collaborative compensation heating mechanism of the adjacent section needs to be activated.

[0023] Compensation heating of adjacent sections must adhere to thermal balance constraints and safety boundaries to avoid temperature imbalances caused by cross-section heating. First, compensation targets should be selected: prioritize adjacent sections with the closest axial distance to the target section and currently in a steady-state thermal inertia to ensure that compensation heating does not affect the temperature control effect of the adjacent section itself. In setting the compensation power, the compensation power of a single adjacent section should not exceed 30% of its rated power, and the total compensation power of multiple adjacent sections should not exceed 40% of the rated power of the target section. Simultaneously, the temperature difference between the target section and adjacent sections should be monitored in real time to ensure that the temperature difference is always controlled within 10℃ to prevent coking of heavy oil or abnormal wellbore stress due to sudden local temperature changes. During the compensation heating process, the viscosity improvement rate should be continuously tracked. Once the viscosity improvement rate reaches the preset standard, or the viscosity of the target section enters a steady-state range, the compensation power of adjacent sections should be gradually reduced until compensation is completely stopped, restoring independent heating to each section.

[0024] A5: Collect the viscosity data of the heated heavy oil, correct the dynamic weight matrix through the proportional-integral algorithm, and iteratively update the segmented power control command until the viscosity enters the preset steady-state range.

[0025] In A5, after the main and auxiliary coil heating and the adjacent section compensation heating are executed, the system synchronously collects the heavy oil viscosity data of each heating section according to the preset collection cycle (5-10 minutes / time). The collection process is aligned with the temperature data collection to ensure that each set of viscosity data corresponds to a clear power control condition. The collected viscosity data needs to be validated: abnormal values ​​caused by fluid disturbance and instantaneous sensor drift are removed (using the moving average filtering method, the average of 3 consecutive samples is taken as valid data), and the data is compared with the initial viscosity value before power adjustment to preliminarily judge the actual effect of power control.

[0026] Based on verified effective viscosity data, a proportional-integral (PI) algorithm is introduced to perform closed-loop correction on the dynamic weight matrix, eliminating the deviation between the current viscosity and the preset steady-state viscosity, and improving the accuracy of the weight matrix in adapting to actual working conditions. The correction logic of the PI algorithm revolves around viscosity deviation: first, the absolute deviation between the real-time viscosity and the preset steady-state viscosity is calculated; the proportional stage (P stage) outputs the correction amount in real time according to the magnitude of the deviation, and quickly adjusts the weight matrix to avoid deviation accumulation; the integral stage (I stage) performs time integration on the deviation to compensate for the static deviation that the proportional stage cannot completely eliminate, improve the stability and accuracy of the correction, and prevent system oscillation; during the correction process, the proportional coefficient and integral coefficient of the PI algorithm need to be dynamically adapted according to the viscosity range of heavy oil. The dynamic weight matrix is ​​updated iteratively row by row in units of segments: For each heating segment, the correction amount output by the PI algorithm is proportionally distributed to each element of the corresponding row of the matrix. The distribution ratio is positively correlated with the original fuzzy similarity value of the element. That is, the element with a higher degree of fit with the current dominant heat flow mode receives a larger correction weight, ensuring that the matrix update always revolves around the core thermal inertia state. The updated matrix is ​​then normalized.

[0027] Based on the corrected dynamic weight matrix, the system recalculates the temperature control compensation gain and then iteratively updates the segmented power control commands for each section, forming a closed-loop iterative link of "power adjustment - viscosity feedback - matrix correction - command update". After each iteration, the system continuously monitors the changing trend of viscosity data until the preset steady-state judgment condition is met: within three consecutive acquisition cycles, the fluctuation range of heavy oil viscosity in each section is ≤5%, and the viscosity value is stably within the preset reasonable steady-state range (set to 1500-3000 mPa·s). If abnormal viscosity fluctuation or continuous increase in deviation occurs during the iteration process, the system will automatically trigger the emergency adjustment mechanism, suspend the current iteration and restore to the previous effective power command, and issue an operating condition warning signal to prompt maintenance personnel to check potential influencing factors such as sensor status and downhole fluid characteristics to ensure the reliability and safety of closed-loop control.

[0028] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback, characterized in that, include: A1: Divide the wellbore longitudinally into N independent heating sections at intervals of 0.5-2m. Temperature data for each section is collected by a Pt100 platinum resistance thermometer group with a ring topology layout. Simultaneously acquire the viscosity-temperature combination data of heavy oil in each section to generate a temperature feature vector and viscosity time series. A2: Perform fuzzy similarity matching between the temperature feature vector and the preset heat flow distribution pattern library, calculate the thermal inertia imbalance coefficient of each section, construct a dynamic weight matrix, and the matrix element values ​​represent thermal inertia overload, underload and steady state. A3: Perform curve fitting on the viscosity-temperature combination data, calculate the curvature value of the fitted curve corresponding to the current temperature, and dynamically generate temperature control compensation gain by combining the thermal inertia imbalance coefficient. A4: Based on the dynamic weight matrix and temperature control compensation gain output segmented power regulation command, the redundant power of the main and auxiliary coils is called first. When the viscosity improvement rate does not reach the preset standard, the adjacent section compensation heating is started. A5: Collect the viscosity data of the heated heavy oil, correct the dynamic weight matrix through the proportional-integral algorithm, and iteratively update the segmented power control command until the viscosity enters the preset steady-state range.

2. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 1, characterized in that: In A1, along the longitudinal direction of the heavy oil production wellbore, based on the characteristics of downhole geological stratification, the viscosity distribution of heavy oil, and the principle of optimal heating efficiency, N independent heating zones are uniformly divided at adjustable intervals of 0.5-2m. Among them, the shallow heavy oil with lower viscosity is divided at intervals of 1.5-2m, and the deep high viscosity zone is divided at intervals of 0.5-1m. Each independent heating section is equipped with a ring-shaped Pt100 platinum resistance thermometer array. After data acquisition, the temperature data of each section is preprocessed to remove abnormal fluctuation values ​​and perform smoothing filtering. The core parameters of the section are extracted, and a temperature feature vector characterizing the temperature field distribution of the section is constructed. The viscosity data is sorted according to the time series to form a continuous viscosity time series.

3. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 1, characterized in that: In A2, the ground control center pre-stores a "preset heat flow distribution pattern library" constructed based on geological models and historical production data. During real-time control, the system performs fuzzy similarity matching calculations between the actual temperature feature vectors of the currently collected and constructed N sections and the template vectors in the pattern library. The matching process adopts a comprehensive similarity measurement algorithm combining weighted Euclidean distance and cosine similarity, and introduces a fuzzy membership function to handle the uncertainty and transition state of the downhole temperature field. Through fuzzy similarity matching, the system identifies which preset pattern is closest to the current overall temperature field and calculates the optimal matching pattern for each independent heating section i. State deviation of corresponding section ; Based on this state deviation and the rate of temperature change in this section The system further calculates the thermal inertia imbalance coefficient. The formula for calculating the thermal inertia imbalance coefficient is: ， in, and These are weighting coefficients determined based on the thermal properties of the formation. This is expressed as the expected rate of temperature change for this section under this model.

4. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 1, characterized in that: In A3, the ground control center processes the viscosity-temperature combination data synchronously acquired for each independent heating section i in real time. First, it uses an Arrhenius-type function to perform least-squares curve fitting to establish a localized viscosity-temperature relationship model for that section under the current operating conditions. The fitting model is an improved Andreid equation, and the specific calculation method is as follows: ， in, Let the viscosity function be expressed as the viscosity function of the i-th segment. Expressed as a proportionality constant, Represented as activation energy type parameters, This is represented as an offset correction term, where T represents temperature; The system calculates at the current real-time temperature The curvature value of the fitted curve is calculated using the following method: ， in, Represented as curvature value, Represented as a fitting function The first derivative with respect to temperature Represented as a fitting function The second derivative with respect to temperature.

5. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 1, characterized in that: In A4, based on the dynamic weight matrix constructed above and the real-time temperature control compensation gain, the basic power control command calculation for each independent heating section is first completed, that is, the power distribution is precisely matched with the thermal inertia state and viscosity-temperature sensitivity characteristics of the section. At the power execution level, a main and auxiliary coil collaborative working mode is adopted. Each heating section is equipped with one main heating coil and two auxiliary heating coils. The main coil undertakes the basic heating load, and the auxiliary coils serve as redundant capacity for backup. When the basic power control command is issued, the system first detects the current load rate of the main coil: if the load rate of the main coil is less than 70%, the power adjustment is completed first through the redundant capacity of the main coil; if the load rate of the main coil has reached more than 85%, and the basic power still does not meet the control requirements, one auxiliary coil is activated to supplement the energy through the superposition of the power of the main and auxiliary coils, while the coil operating temperature and insulation status are monitored in real time.

6. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 5, characterized in that: The compensation heating of adjacent sections must follow the constraints of thermal balance and safety boundaries. First, the compensation target is selected: the adjacent section with the closest axial distance to the target section and whose current thermal inertia is in a steady state is selected first. At the same time, the temperature difference between the target section and the adjacent section is monitored in real time. During the compensation heating process, the change in viscosity improvement rate is tracked. When the viscosity improvement rate reaches the preset standard, or when the viscosity of the target section enters the steady state range, the compensation power of the adjacent section is reduced until the compensation is completely stopped and the independent heating state of each section is restored.

7. The adaptive control method for electric heating power in heavy oil extraction based on downhole temperature feedback according to claim 1, characterized in that: In A5, after the main and auxiliary coil heating and the adjacent section compensation heating are executed, the system synchronously collects the heavy oil viscosity data of each heating section according to the preset collection cycle. The collection process is aligned with the temperature data collection in time. The collected viscosity data needs to be verified for validity: abnormal values ​​caused by fluid disturbance and instantaneous sensor drift are removed. At the same time, the data is compared with the initial viscosity value before power adjustment to preliminarily judge the actual effect of power regulation. Based on the verified effective viscosity data, a proportional-integral (PI) algorithm is introduced to perform closed-loop correction on the dynamic weight matrix. The PI algorithm first calculates the absolute deviation between the real-time viscosity and the preset steady-state viscosity, and the proportional component outputs the correction amount in real time according to the magnitude of the deviation to adjust the weight matrix. The integration stage performs time integration on the deviation; During the correction process, the proportional coefficient and integral coefficient of the PI algorithm need to be dynamically adapted according to the viscosity range of heavy oil.

Images