Environment-friendly fracturing fluid proportioning homogeneous self-adaptive adjustment method based on advanced control

Abstract

The present application relates to the field of oil and gas field fracturing fluid preparation and process control, specifically to an environmentally-friendly fracturing fluid proportioning homogeneous self-adaptive adjustment method based on advanced control, which discloses a proportioning homogeneous self-adaptive adjustment method: synchronously collecting the flow, pressure, concentration, temperature, density, and mixed channel and outlet quality parameters of each branch, aligning the time sequence according to a unified sampling period, calculating the effective component mass flow, delay, and proportioning deviation, determining the control priority in combination with the outlet quality deviation and the original liquid disturbance, generating a preselected increment and obtaining a compensation control amount through an outlet proportioning / mass prediction model; when the present application does not obtain a feasible solution, a time limit amplitude output is outputted, and the delay and model parameters are updated online according to the prediction residual.

Description

Technical Field

[0001] This invention relates to the field of fracturing fluid preparation and process control in oil and gas fields, specifically to an environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control. Background Technology

[0002] In the continuous preparation of environmentally friendly fracturing fluid, it is usually necessary to transport multiple feedstocks containing effective components to the mixing channel according to the target ratio, and to maintain the outlet quality parameters to meet the process requirements while ensuring the stability of the total flow rate. Due to the differences in the physical properties of the feedstocks in each branch, the pipeline transportation and mixing process, and the certain lag between the outlet detection results and the control actions, it is easy for ratio deviations, uneven mixing and outlet quality fluctuations to occur during operation, which affects the stability of continuous preparation.

[0003] Traditional methods rely heavily on fixed formulas and experience-based adjustments, which do not adequately consider the time-series relationship between branch flow, mixing energy, and outlet quality. They also make it difficult to identify the impact of differences in actuator response and disturbances in the raw liquid in a timely manner. When switching formulas or changing operating conditions, the outlet ratio may be misaligned with the target ratio due to asynchronous adjustment of control quantities, reducing adjustment accuracy and system stability. Summary of the Invention

[0004] The purpose of this invention is to provide an environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control, which avoids misalignment between the outlet ratio and the target ratio due to asynchronous adjustment of control quantities when switching formulas or changing operating conditions, prevents reduction in adjustment accuracy and system stability, and makes it easier to control the time alignment relationship between multi-branch flow, mixing energy and outlet quality, and timely identify and compensate for the impact of actuator response differences and raw fluid disturbances.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] An environmentally friendly fracturing fluid ratio homogenization adaptive adjustment method based on advanced control includes: each branch separately delivers raw fluid containing effective components to the mixing channel, and collects the flow rate, pressure, control input of the corresponding branch actuator, concentration, temperature and density of the effective components of the raw fluid, as well as the mixing energy, outlet quality parameters, target total flow rate, target ratio based on the quality of effective components and corresponding target quality parameters of the mixing channel, and processes them synchronously according to a unified sampling cycle;

[0007] Based on the synchronous data, the effective component mass flow rate, inlet instantaneous ratio, equivalent total delay and control response prediction delay of each branch are determined. The mass flow rate of each branch is mapped to the same outlet time according to the equivalent total delay, and the outlet alignment ratio and outlet target ratio are determined.

[0008] The inlet and outlet ratio deviations are determined based on the difference between the inlet instant ratio, the outlet aligned ratio, and the corresponding target ratio. The outlet quality deviation is determined based on the difference between the outlet and the target quality parameters. The raw liquid disturbance index is determined based on the changes in raw liquid concentration, temperature, and density.

[0009] Based on the inlet and outlet ratio deviation, outlet quality deviation and raw liquid disturbance index, the control priority is determined and an instantaneous control pre-selection increment is generated; using the prediction model and the instantaneous control pre-selection increment as the optimization reference, the predicted compensation increment of the branch and the hybrid actuator is solved, and the final control quantity is synthesized and output. If there is no feasible solution, the instantaneous control pre-selection increment after amplitude limiting is output.

[0010] The quality model prediction residual is generated based on the difference between the measured export quality parameters and the predicted values. The delay parameters, quality prediction model parameters, and quality sensitivity matrix are then updated by combining the control variables and changes in the export quality parameters.

[0011] Preferably, the export quality parameters include online export density, online export apparent viscosity, online export electrical conductivity, and solid phase homogeneity.

[0012] A uniform sampling period is required to satisfy the requirement that it be less than or equal to the smaller of the response time and the minimum equivalent lag time of detection, thereby ensuring that both the dynamic changes of the branch and the quality response of the outlet can be effectively identified.

[0013] Preferably, the density of the stock solution is corrected according to the pressure, temperature and effective component concentration of the stock solution in each branch by using a preset density correction model, and the mass flow rate of the effective component is calculated accordingly to obtain the inlet instant ratio.

[0014] Simultaneously, a target ratio is generated based on the target formula trajectory, and the target branch flow rate is determined by combining the target total flow rate, the density of the original solution, and the concentration of the effective components, so as to ensure that the ratio control and the total flow rate control are coordinated and consistent.

[0015] Preferably, the corresponding historical data is called according to the delay steps of each branch, so that the data of different branches are aligned to the same exit time, and the exit alignment ratio and the exit target ratio are calculated accordingly to obtain the exit ratio deviation.

[0016] At the same time, the export quality parameters are normalized to construct a comprehensive quality index and its changing trend. When the quality improves, the control increment of the first branch actuator and the control increment of the hybrid actuator are appropriately scaled to avoid over-control.

[0017] Preferably, the inlet ratio deviation index, outlet ratio deviation index, mixing homogenization deviation index, and stock solution disturbance index are determined based on the inlet ratio deviation, outlet ratio deviation, comprehensive quality index, and normalized changes in stock solution concentration, temperature, and density.

[0018] When multiple indicators exceed limits simultaneously, control requirements are generated in descending order of the degree of exceedance, and corresponding to branch flow adjustment, predictive compensation, mixed energy adjustment, and target branch flow update, thereby improving the pertinence and stability of control allocation.

[0019] Preferably, an instantaneous control pre-selection increment is generated based on the inlet ratio deviation and its rate of change, and the control increment of the branch actuator and the hybrid actuator is continuously optimized in the prediction time domain and the control time domain. At the same time, upper and lower limits of control quantity, increment limit, rate of change limit and upper and lower limits of hybrid total flow are applied.

[0020] When no feasible solution is found, the incremental limit, rate limit, and control quantity limit are applied sequentially to the pre-selected increment of the real-time control before outputting the control command to ensure that the control command can be executed.

[0021] Preferably, when the sensor diagnosis is normal, the actuator is not in a limited state, no formula switching is performed, and the sliding data window meets the preset identification conditions, the residual is updated online based on the quality model prediction.

[0022] The target delay step number is determined by the correlation between the shifted control increment sequence and the export quality parameter change sequence, and the delay parameter is corrected by a single-step small-amount update method.

[0023] By combining the changes in the control quantity of the actuator and the changes in the export quality parameters, the parameters of the quality prediction model and the quality sensitivity matrix are updated using a recursive identification method, and upper and lower limits are set for each matrix element to suppress model drift.

[0024] Preferably, after receiving the formula switching instruction, an outlet target ratio trajectory is generated to smoothly transition from the original formula target ratio to the new formula target ratio, and the target branch flow transition trajectory that satisfies the actuator constraints and total flow constraints is determined by combining the delay of each branch, the density of the original liquid, the concentration of the effective component and the target total flow.

[0025] During formula switching, parameter updates are stopped. Parameter updates will resume when the weighted norm and rate of change of the export ratio deviation and the quality model prediction residual are both less than the corresponding threshold for consecutive preset sampling periods.

[0026] The beneficial effects of this invention are:

[0027] 1. This invention corrects the density of the raw liquid, comprehensively evaluates the online density, apparent viscosity, conductivity and solid phase uniformity at the outlet, and scales the control increment using comprehensive quality indicators and quality change trends. It can automatically reduce the control intensity when the outlet quality has improved, reduce the frequent action and repetitive adjustment of the actuator, and enhance the stability and continuity of the system.

[0028] 2. This invention classifies the inlet ratio deviation, outlet ratio deviation, mixing homogenization deviation, and raw liquid disturbance, and generates control requirements according to the degree of exceeding the limit. These requirements are then synthesized into real-time control pre-selection increments. This enables different physical causes to correspond to different adjustment objects, reduces mutual interference caused by multiple deviations entering the same adjustment channel at the same time, and improves the rationality of control priority allocation.

[0029] 3. This invention solves the predicted compensation increment of branch actuators and hybrid actuators by rolling optimization, and combines the online update of delay parameters, quality prediction model parameters and quality sensitivity matrix of the predicted residual of the quality model, so that the control strategy can be adaptively corrected as the operating conditions change; when switching recipes, it can also generate a smooth transition trajectory and set the advance action time to reduce switching shock and outlet fluctuation, thereby improving the system's stability adjustment capability and recipe switching accuracy. Attached Figure Description

[0030] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0031] Figure 1 This is a flowchart of the environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control of the present invention. Detailed Implementation

[0032] 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.

[0033] Please see Figure 1 An environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control includes: each branch separately delivers raw fluid containing effective components to the mixing channel, and collects the flow rate, pressure, control input of the corresponding branch actuator, concentration, temperature and density of the effective components of the raw fluid, as well as the mixing energy, outlet quality parameters, target total flow rate, target ratio based on the quality of effective components and corresponding target quality parameters of the mixing channel, and processes them synchronously according to a unified sampling cycle;

[0034] Based on the synchronous data, the effective component mass flow rate, inlet instantaneous ratio, equivalent total delay and control response prediction delay of each branch are determined. The mass flow rate of each branch is mapped to the same outlet time according to the equivalent total delay, and the outlet alignment ratio and outlet target ratio are determined.

[0035] The inlet and outlet ratio deviations are determined based on the difference between the inlet instant ratio, the outlet aligned ratio, and the corresponding target ratio. The outlet quality deviation is determined based on the difference between the outlet and the target quality parameters. The raw liquid disturbance index is determined based on the changes in raw liquid concentration, temperature, and density.

[0036] Based on the inlet and outlet ratio deviation, outlet quality deviation and raw liquid disturbance index, the control priority is determined and an instantaneous control pre-selection increment is generated; using the prediction model and the instantaneous control pre-selection increment as the optimization reference, the predicted compensation increment of the branch and the hybrid actuator is solved, and the final control quantity is synthesized and output. If there is no feasible solution, the instantaneous control pre-selection increment after amplitude limiting is output.

[0037] The quality model prediction residual is formed based on the difference between the measured export quality parameters and the predicted values. The delay parameters, quality prediction model parameters and quality sensitivity matrix are updated by combining the control quantity and the changes in export quality parameters.

[0038] As a preferred embodiment, the export quality parameters include online export density, online export apparent viscosity, online export conductivity, and solid phase homogeneity; a unified sampling period is also included. satisfy ,in, , For the first The response time from a change in the control input of a branch to the point that the branch flow reaches a preset stable range. The minimum equivalent lag time required for a change in export quality parameters to produce an identifiable response to the export quality test value;

[0039] Within the preset candidate hysteresis range and sliding data window, according to the first... The correlation between the control input increment sequence of each branch actuator and the measured volumetric flow rate increment sequence of that branch determines the number of actuator lag steps. ;

[0040] According to the The conveying delay is determined by the effective conveying volume and measured total mixing flow rate of each branch to the mixing channel. The mixing delay is determined based on the mixing volume, measured total mixing flow rate, outlet apparent viscosity, and mixing energy. This mixed delay is then combined with the outlet quality detection link delay to form the equivalent total delay. Control response prediction delay is .

[0041] In practice, a unified discrete time sequence is formed based on each synchronous sampling moment, and the flow rate, pressure, concentration, temperature, density of the raw fluid in each branch circuit, as well as the mixing energy and outlet quality parameters of the mixing channel, are collected accordingly. The outlet quality parameters are used to reflect whether the outlet fracturing fluid meets the requirements for continuous preparation in terms of physical properties and uniformity.

[0042] The sampling time sequence number is... This indicates that the branch number is... It indicates that the total number of branch roads is based on This indicates that when summing multiple branches is involved, Indicates the branch number participating in the summation;

[0043] The unified sampling period setting is used to ensure that the three types of information, namely actuator action, pipeline delivery and outlet detection, can be time-aligned on the same period scale; when the sampling period is lower than the first preset threshold, the process change reflected by adjacent sampling data is less than the effective identification threshold; when the sampling period is higher than the second preset threshold, it is impossible to identify the ratio deviation and control response within the preset time window.

[0044] Therefore, the sampling scale is determined based on the minimum of the control response time of each branch and the identifiable response time of the exit detection, and then... Set a stability margin;

[0045] Specifically, Before the system is put into operation, a step test is conducted to determine that, under the condition that the increment does not exceed the allowable increment of the actuator, a known control input change is applied to each branch, and the time it takes for the branch flow to enter and remain within a stable range is recorded.

[0046] Next, a known mixing energy change is applied to the mixing channel, and the time it takes for the outlet quality detection value to produce a stable and identifiable change is recorded; the smaller of the minimum response time obtained from the experiment and the minimum detection response time is used as the basis for selecting the sampling period. A preset stability coefficient of no more than 1 is selected and remains unchanged in subsequent runs unless recalibration is performed.

[0047] Furthermore, in response to the timing difference between the current detection value and the historical control action during continuous mixing, this embodiment calculates the effective component mass flow rate of each branch and maps the historical mass flow rate to the same outlet time according to the equivalent total delay of each branch, thereby establishing the correspondence between the outlet aligned mixing ratio and the outlet target mixing ratio.

[0048] Therefore, changes in export detection are interpreted according to their corresponding historical feeding times, and corresponding control compensation is determined in combination with conveying delay and mixing hysteresis, avoiding direct comparison of data from different time slices;

[0049] To determine the time delay of each branch, this implementation method comprehensively considers the time contribution of the actuator response, material conveying, mixing and homogenization process, and detection link; among which, the actuator response contribution reflects the time taken for the actuators such as pumps and valves to take control actions and for the branch flow to produce a stable change;

[0050] The contribution of the conveying process reflects the volume transfer of material from the branch injection point to the outlet detection point; the contribution of the mixing process reflects the time effect of mixing energy, apparent viscosity and local backmixing on the material reaching a homogeneous state; and the contribution of the detection link reflects the time required for the outlet quality detection device to generate a recognizable response.

[0051] The number of lag steps of the actuator is determined within the sliding window by comparing the correlation strength and stability of the candidate lags: the candidate lag is only accepted when the correlation corresponding to the candidate lag exceeds the preset correlation threshold and the lag difference obtained in the two sub-intervals before and after the window does not exceed one sampling step.

[0052] Otherwise, the previous effective hysteresis value is retained; the delivery delay is determined by the correspondence between the effective volume of the pipeline and the measured total mixed flow at the same time, while the mixing delay is obtained by obtaining the detection response time under different apparent viscosity ranges through a mixing energy step test, and is obtained by interpolation by looking up the table according to the current apparent viscosity and the mixing energy range during operation;

[0053] The export quality inspection link delay is obtained by the step response calibration of the inspection device; all delays are converted to non-negative times, and the corresponding export alignment calculation is not performed when historical data is insufficient.

[0054] This implementation establishes a time correspondence between actuator control input, branch flow, mixing process and outlet detection results through cross-channel time delay mapping, and synthesizes the time contributions of each part into an equivalent total delay for control compensation.

[0055] In a preferred embodiment, a preset density correction model is used, based on the first... The pressure, temperature, and effective component concentration of the stock solution in each branch are used to correct the stock solution density. and in accordance with Determine the effective component mass flow rate, according to Determine the immediate intake ratio;

[0056] in, For the measured volumetric flow rate of the branch, This represents the mass fraction of the effective component in the original solution. Number of branch roads; target allocation ratio To meet The effective component mass ratio, the target branch flow rate is ;

[0057] according to Determine the first The delay steps for each branch are calculated, and each branch is called separately. Historical data at each moment, so that each historical data point corresponds to the same exit moment;

[0058] The export alignment ratio is:

[0059]

[0060] The export target ratio is:

[0061]

[0062] Export ratio deviation is:

[0063]

[0064] in, The target effective component mass flow rate, which varies with the target formulation trajectory, remains at the value corresponding to the current target formulation when no formulation switch is performed;

[0065] according to Determine the first The normalized deviation of each export quality parameter, among which This is the allowable deviation scale for this parameter;

[0066] And in accordance with Determine the comprehensive quality indicators, among which, For export quality parameters and quantities, Let be the normalized weight of the q-th export quality parameter, and and ;according to Determine the trend of quality change, when season ,in The minimum control percentage that can be retained. This is the scaling factor for quality improvement; otherwise, let... And utilize before finally controlling the incremental output. Scaling the first branch actuator control increment and the hybrid actuator control increment.

[0067] Furthermore, the density of the stock solution is not directly adopted as a fixed value, but is corrected by pressure, temperature and effective component concentration to reflect the quality attributes of the stock solution under the current working conditions;

[0068] For example, the density and flow characteristics of the same concentration stock solution may change after the temperature changes. If the calculation is still based on a fixed density, it will cause the calculation deviation of the effective component mass flow rate.

[0069] Target total flow It indicates that the superscript This represents the target value, where the target branch flow is allocated according to the mass ratio of the effective components of the target in each branch under the constraint of the total flow.

[0070] The preset density correction model is established through calibration before the system is put into operation. Specifically, multiple calibration points are selected within the working range of pressure, temperature and effective component concentration of the original liquid in each branch. The corresponding density is measured using a calibrated densitometer, and a three-dimensional calibration table is established using pressure, temperature and concentration as indexes.

[0071] During operation, piecewise linear interpolation is performed on adjacent calibration points. When the value exceeds the calibration range, the nearest boundary value is used and marked as an out-of-range condition. The calibration table is updated only when the sensor diagnosis is normal and the raw liquid is in a stable delivery state, so as to avoid writing abnormal measurement values ​​into the density correction model.

[0072] The effective component mass flow rate is used to characterize the effective component mass intensity currently fed into the mixing channel of each branch, and the inlet instant ratio is used to characterize the instantaneous proportional relationship of each branch at the inlet. Both are based on the effective component mass as a unified benchmark; the inlet target ratio is the effective component mass ratio of each branch determined according to the current target formula.

[0073] The target branch flow rate is used to ensure that each branch enters the mixing process according to the target ratio under the premise that the total flow rate is constrained; the target effective component mass flow rate changes with the target formula trajectory, and transitions synchronously with the outlet target ratio during formula switching, and maintains the stable value corresponding to the current target formula when not switching;

[0074] When calculating the inlet instantaneous proportion, the sum of the effective component mass flow rates of each branch is used as the normalization benchmark; the proportion calculation is performed under normal conveying conditions where the sum of the effective component mass flow rates is greater than zero.

[0075] When the total effective component mass flow rate is lower than the preset effective calculation lower limit, the ratio result of the previous effective sampling time is maintained, and this state is passed to the subsequent control logic as a low flow state, so as to avoid abnormal amplification of the ratio deviation caused by the normalization benchmark being lower than the preset effective calculation lower limit.

[0076] The effective calculation lower limit is determined based on the minimum reliable measured flow rate of each branch flow meter and the minimum calibration value of the effective component concentration, and is taken as the lower limit of mass flow rate that can make the proportioning calculation results not dominated by measurement noise; the proportioning deviation and mass model identification data are not updated under low flow conditions;

[0077] The delay step is rounded up according to the ratio of the equivalent total delay to the uniform sampling period. It is used to align the historical data of different branches to the same exit time, so that the exit alignment ratio and the exit target ratio have the same time base.

[0078] During the startup phase or when historical data has not yet covered the corresponding delay steps, the unformed historical data is not called. Instead, the calculation of the export ratio deviation is temporarily suspended, and the alignment calculation is started after the historical data reaches the required length.

[0079] During normal operation, historical valid data is retrieved according to the delay steps corresponding to each branch. If the target formula has not been switched, the target mass flow rate of each branch is calculated using the same current target formula. If the target formula is in transition, the target mass flow rate trajectory is retrieved according to the delay corresponding to each branch, so that the mass flow rate and the target mass flow rate have the same time reference at the exit time.

[0080] The comprehensive quality index is used to measure the overall deviation of multiple quality parameters such as online density, apparent viscosity, electrical conductivity, and solid phase homogeneity at the outlet.

[0081] In the formula Indicates the serial number of the export quality parameter. This indicates the total number of export quality parameters. Indicates the first Normalized weights of each quality parameter; Indicates the first The measured values ​​of each export quality parameter at the current sampling time. This indicates its corresponding target value. This indicates the allowable deviation scale for this quality parameter;

[0082] This represents the number of historical sampling points used to calculate the short-window average. This indicates the change in the current overall quality index relative to the average value of the short window. This represents the control scaling factor used for quality improvement. This indicates the minimum control percentage that is allowed to be retained. This indicates the scaling ratio of the final control increment;

[0083] Permissible deviation scales The allowable range of the corresponding quality parameters is determined based on the process, and the allowable absolute deviation of the parameter under the normal export quality standard is taken as a positive value.

[0084] Weights The importance of quality parameters to the continuous preparation quality of fracturing fluid is determined, and the sum of the weights of all effective parameters is normalized to 1.

[0085] The permissible deviations of online density, apparent viscosity, and conductivity are determined by the quality acceptance range, while the permissible deviation of solid phase homogeneity is determined by the calibration repeatability of the homogeneity testing device and the process homogeneity requirements. Take the integer number of sampling points that cover at least one quality response period and do not exceed the length of the sliding window;

[0086] Take a value greater than zero and do not make the control ratio lower than zero. calibration coefficients, The minimum control ratio at which the actuator can maintain stable adjustment is determined; the above parameters are calibrated jointly using historical stability data and step tests before commissioning, and do not change with a single abnormal sampling value during operation;

[0087] When the allowable deviation scale of a certain export quality parameter is zero or the corresponding measured data is invalid, normalization calculation is not performed directly. Instead, the update of the comprehensive quality index of the parameter is suspended according to the preset invalid data processing rules, and the sensor diagnosis results determine whether to maintain the previous valid index value or trigger fault processing.

[0088] For normal and valid export quality parameters, they are normalized according to their allowable deviation scale and then used in the calculation of comprehensive quality indicators; the normalized deviation is only used for dimensionless comparison, and its normal and valid values ​​are not pre-limited to a fixed range.

[0089] When the absolute normalization deviation exceeds the preset abnormal upper limit, the parameter is marked as abnormal to prevent it from causing the control increment to increase indefinitely. Instead, the logic is switched to sensor verification or fault handling.

[0090] This is used to appropriately reduce the control intensity when the quality indicators are declining, to prevent the execution end from continuing to generate adjustment actions that exceed the preset incremental limit when the quality is improving.

[0091] The first branch actuator control increment refers to the branch control increment output according to control priority in the current rolling control cycle; Only the control action of the first branch and the control action of the hybrid actuator are synchronously scaled, without changing the already determined constraints and control increment order of other branches;

[0092] This implementation method uses the judgment of quality change trends and the control increment scaling to gradually weaken the control action when the quality index improves, and maintain the original control ratio when the quality index does not improve, thereby reducing repeated adjustments and frequent actions of the actuator caused by detection lag.

[0093] As a preferred embodiment, based on the weighted norm of the inlet ratio deviation, the weighted norm of the outlet ratio deviation, the comprehensive quality index, and the normalized maximum values ​​of the concentration, temperature, and density of the stock solution relative to the previous sampling time or calibration reference value, the inlet ratio deviation index, the outlet ratio deviation index, the mixing homogenization deviation index, and the stock solution disturbance index are determined respectively; for any index according to Determine the degree of exceeding the limit, among which This is the preset allowable threshold corresponding to this indicator;

[0094] When multiple indicators exceed the limit simultaneously, according to The control requirements are generated from large to small and synthesized into real-time control pre-selected increments. Among them, the inlet ratio deviation corresponds to the branch flow adjustment requirement, the outlet ratio deviation corresponds to the prediction compensation requirement, the mixing homogeneity deviation corresponds to the mixing energy adjustment requirement, and the raw liquid disturbance corresponds to the target branch flow update requirement.

[0095] Real-time control pre-selection increments are generated based on the inlet ratio deviation, the rate of change of the inlet ratio deviation relative to the previous sampling time, and the target branch flow change; according to Determine the number of control response prediction steps and set the prediction time domain. and control time domain ;

[0096] Rolling optimization is performed using a quadratic cost function with a weight matrix to construct the cost function. :

[0097]

[0098] Where P represents the prediction time domain and C represents the control time domain. and For forecasting and target export ratios, To predict quality parameters With target quality parameters The dimensionless normalized deviation vector obtained by combining the allowable deviation scale conversion, To control the increment, To enable real-time control of pre-selected increments; All are diagonal positive definite pre-defined weight matrices; therefore Minimize the control increment of the branch actuators and hybrid actuators for the target rolling optimization, and impose upper and lower limits on the control quantity, increment limit, rate of change limit and upper and lower limits on the total hybrid flow for each actuator;

[0099] When no feasible solution is found, the incremental limit, rate limit, and control quantity limit are applied sequentially to the pre-selected increment of the instantaneous control before output.

[0100] When the sensor diagnosis is normal, the actuator is not in the limiting state, no formula switching has been performed, and the sliding data window meets the preset identification trigger conditions, the online update is performed based on the quality model prediction residual; if any of the above conditions are not met, the current delay parameters, quality prediction model parameters, and quality sensitivity matrix remain unchanged.

[0101] The control increment sequences of branch actuators and hybrid actuators are shifted using the current quality model delay step number and its adjacent candidate step number. The target delay step number is determined based on the correlation between the shift sequence and the change sequence of each outlet quality parameter, and updated with an amplitude not exceeding one sampling step at a time. Based on the changes in the control quantity of each actuator and the changes in each outlet quality parameter, the quality prediction model parameters and quality sensitivity matrix are updated using a recursive identification method, and amplitude is limited according to the upper and lower limits set for each matrix element.

[0102] Obtain the formula switching instruction, which includes the original formula, the new formula, and the expected export switching time, and generate a smooth export target ratio transition trajectory that starts with the original formula target ratio and ends with the new formula target ratio.

[0103] The sampling time corresponding to the expected exit switching time is denoted as . Using the control response prediction steps determined before the formula changeover begins, according to Determine the first The advance action time of each branch is determined, and the target branch flow transition trajectory that satisfies the actuator constraints and total flow constraints is determined based on the smooth outlet target ratio transition trajectory, the delay of each branch, the original liquid density, the effective component concentration and the target total flow.

[0104] During formula switching, parameter updates are stopped. Parameter updates resume when the weighted norm of the export ratio deviation and the weighted norm and rate of change of the quality model prediction residual are less than their respective thresholds for consecutive preset sampling periods.

[0105] In practical implementation, the index number is... express, Indicates the first Item control evaluation indicators, This indicates the preset allowable threshold corresponding to this indicator. This indicates the degree of normalized exceedance of the indicator after it exceeds the corresponding threshold;

[0106] The inlet ratio deviation index, outlet ratio deviation index, mixing homogeneity deviation index, and raw liquid disturbance index correspond to the control requirements of branch ratio, outlet ratio, mixing homogeneity, and raw liquid property changes, respectively. Multiple control requirements are determined in order of action from largest to smallest degree of exceeding limits, and are combined into an instantaneous control pre-selected increment under the premise of satisfying the constraints of the actuator and total flow.

[0107] When a certain indicator does not exceed its corresponding threshold, no control requirement is generated solely based on that indicator; thresholds for each indicator Determined based on historical data from the continuous and stable operation phase, the scope of process acceptance, and the adjustability of the actuator:

[0108] First, obtain the fluctuation range of each indicator using normal data, and then select a threshold that is higher than the upper limit of normal fluctuation but does not exceed the upper limit of process allowable deviation, in combination with the corresponding quality or proportion allowable deviation.

[0109] The normalized change of the indicator is divided by the denominator of the allowable range of the corresponding parameter and is limited to a non-negative absolute change. If the calibrated range is zero or the sensor data is invalid, the indicator is suspended from participating in the priority ranking and is taken over by the fault handling logic.

[0110] For the original liquid disturbance index, during normal operation, the change between two adjacent effective sampling times is uniformly used as the disturbance benchmark, that is, the absolute value of the difference between the current sample value and the previous effective sample value is normalized.

[0111] The reference value obtained before commissioning is used as the comparison reference only when the system is started, continuous sampling is interrupted, or the previous sampled value is diagnosed as invalid.

[0112] Concentration, temperature and density are normalized using their respective positive allowable change scales, and the maximum value of the normalized absolute change among the three is taken as the original solution disturbance index.

[0113] The baseline value and allowable variation scale remain unchanged when the sensor diagnosis is normal and the concentrate is in a stable delivery state, and are not updated immediately due to a single abnormal sampling value.

[0114] When the corresponding allowable change scale is not greater than zero, the calculation of the original liquid disturbance index is suspended; for the calculation of the degree of exceeding the limit... Pre-set to positive values; when a certain indicator data is invalid, the calculation of the degree of exceeding the limit of that indicator will not be performed, nor will it be included in the control requirement ranking;

[0115] This implementation method classifies different deviation indicators, so that branch ratio deviation, outlet ratio deviation, mixing homogenization deviation and raw liquid property disturbance correspond to the corresponding adjustment objects. Then, the order of control requirements is determined according to the degree of exceeding the limit, and they are synthesized into real-time control pre-selection increments under constraints, thereby reducing the mutual interference caused by deviations from different physical causes entering the same adjustment channel at the same time.

[0116] To avoid the comprehensive indicators masking serious local deviations, in addition to participating in the weighted norm, the inlet and outlet ratio deviations are also subject to upper limits for deviations in individual branches. When any branch exceeds the upper limit, the flow correction requirement for that branch is generated first, and the correction requirement is not offset by other branches having deviations below the corresponding preset lower limit.

[0117] When quality parameters such as solid phase uniformity become invalid or undergo abnormal abrupt changes, the corresponding abnormal sampling point is removed and the previous valid quality index is maintained until the sensor diagnosis returns to normal.

[0118] The real-time control pre-selected increment is generated according to a pre-fixed priority synthesis rule: the inlet ratio deviation is converted into the difference between the target flow and the measured flow of each branch, and the increment direction of the corresponding branch actuator is determined according to the deviation sign;

[0119] The export ratio deviation is shifted according to the predicted delay of the control response of each branch to form the predicted compensation demand of each branch; the incremental demand for mixed energy is calculated again, as follows:

[0120]

[0121] In the formula, To meet the incremental demand for hybrid energy; This is the preset energy regulation ratio coefficient; This is a deviation from the homogeneous mixture;

[0122] At the same time, the original liquid disturbance index is superimposed on the calibration benchmark value to serve as the updated benchmark for the target branch flow.

[0123] All demands are first pre-limited according to the incremental limits of the corresponding implementing agency, and then... Write the pre-selected increments in descending order; when a subsequent demand and an already written demand cause a violation of the mixed total flow or any actuator constraint, reduce the subsequent demand by the remaining available margin; if the remaining margin is zero, skip the demand.

[0124] Therefore, under the same input data and the same constraints, a unique instantaneous control preselection increment can be obtained in a determined order, without relying on manual selection or an undefined optimization process.

[0125] The control process is executed cyclically according to a unified sampling cycle, and sequentially goes through the states of data confirmation, demand generation, prediction solution, amplitude limiting output and result confirmation. After entering a sampling cycle, the flow rate of each branch, the physical properties of the raw liquid, the outlet quality parameters and the feedback of the actuator are checked to see if they are within the effective time range.

[0126] When any key data is missing, the timestamp is not aligned with the current sampling period, or the number of consecutive anomalies reaches the fault threshold, the period is marked as an unidentifiable period, the previous effective control state is maintained, and only the safety correction after the limit is performed based on the still effective ratio deviation, without updating the model parameters.

[0127] After the data is confirmed, a pre-selected increment is generated according to the degree of exceedance of the indicators; if multiple indicators have the same degree of exceedance, they are processed in a fixed order of inlet ratio deviation, outlet ratio deviation, mixing homogenization deviation and original liquid disturbance.

[0128] After the prediction is completed, the upper and lower limits of the control quantity, the single-cycle increment limit, the rate of change between adjacent cycles and the range of mixed total flow are checked one by one for the candidate control quantities. If any constraint is not satisfied, the constraint check step is immediately returned to reduce the corresponding increment.

[0129] If no feasible solution is found after reduction, the output will be limited. After the control quantity is output, the actuator feedback will be read in the next sampling cycle. Only when the feedback reaches the allowable error range of the corresponding control quantity will the current cycle be marked as executed. Otherwise, the unconfirmed state will be maintained and parameter identification based on the control action will be prohibited.

[0130] Furthermore, in the predictive control phase, the inlet ratio deviation, deviation change rate, and target branch flow change are used to generate real-time control pre-selection increments, which are then used as optimization references. The control response prediction steps are obtained by superimposing the actuator lag steps and the equivalent total delay steps, and are used to determine the sampling position where the control action is expected to act on the outlet.

[0131] Here, the number of steps the implementing agency lags behind. And equivalent total delay steps All are represented using discrete step counts under a uniform sampling period. It is obtained by rounding up the equivalent total delay. Determined by superimposing the discrete step counts of the two;

[0132] The continuous-time control response prediction delay is used only for physical timing calibration and parameter recording, and is no longer repeatedly added to the determined discrete step count; the prediction model places the control increment after the current sampling time. The sampling locations are calculated based on the identified response duration steps at subsequent sampling locations;

[0133] Prediction Time Domain Not less than the maximum control response prediction steps, control time domain This limits the search to the prediction time domain, thus defining the scope of the optimization search.

[0134] The export ratio and export quality prediction model takes the aligned mass flow rate, target ratio, mixed energy, branch actuator control increment, and mixed actuator control increment of each branch outlet as inputs, and places the control action at the corresponding future sampling position based on the predicted control response steps of each branch, outputting the predicted export ratio and export quality parameters. In specific implementation, a discrete incremental model based on the current operating point is adopted.

[0135] Using the current export ratio and quality parameters as the baseline output, the increments of each input in the prediction time domain are placed at the future sampling positions according to the corresponding delays, and the future output changes are calculated through the identified branch response coefficients, mixed energy response coefficients and their historical response terms.

[0136] Among them, the branch response coefficient and the mixed energy response coefficient together constitute the mass sensitivity matrix, and the historical response term is used to characterize the existing control effect that has not yet decayed.

[0137] The initial parameters of the model were obtained by conducting small-amplitude step tests on each branch and the hybrid actuator before commissioning. During the step test, other actuators were kept unchanged, and the stable response of the outlet ratio and quality parameters was used as the basis for determining the corresponding response coefficients and the effective response duration steps.

[0138] All inputs and outputs are standardized in terms of their calibrated working range before entering the model. If they exceed the calibrated range, online identification is stopped and the most recently effective model parameters are used for prediction.

[0139] To clarify the correspondence between the data structures in the prediction model, the output at any future sampling time is arranged into an output vector composed of the export proportion component and the export quality parameter component, where the export proportion has... Each component and export quality parameter are available. One component;

[0140] Will The control increments of the branch actuators and one hybrid actuator are arranged as input vectors; therefore, after normalization using the calibration range, the rows of the quality sensitivity matrix correspond to... Each output component corresponds to a column. Each actuator input component has the following dimensions:

[0141]

[0142] This matrix represents the local incremental sensitivity at the current operating point and the corresponding delay position, and does not require that the output components be independent of each other or that the matrix be full rank.

[0143] When there are normalization constraints among the components of the export ratio, identifiable independent components or constrained recursive updates are used to avoid duplicate identification of the same normalization constraint.

[0144] Unnormalized physical quantities are not directly multiplied by the matrix. Instead, they are first converted into dimensionless increments according to their respective calibration scales. The predicted dimensionless output increments are then restored to the changes in proportion or mass parameters according to the corresponding output scale. This avoids the direct mixing of dimensions for different flow rates, energy, proportions, and mass parameters.

[0145] The prediction model operates according to the following fixed steps: first, read the output ratio, output quality parameters, and historical response items at the current sampling time; then, write the candidate control increments of each branch actuator and hybrid actuator in the prediction time domain into their corresponding control response prediction step positions.

[0146] The increments of each outlet ratio and outlet quality parameter are calculated by multiplying the corresponding actuator column in the quality sensitivity matrix with the control increment, and then added to the current benchmark value and historical response terms to obtain the prediction results for each future sampling time.

[0147] When the same export quality parameter is affected by multiple actuators, the increments corresponding to each actuator are added together; for control increments that have not yet reached the corresponding delay position, their predictive effect is set to zero.

[0148] For historical control actions that exceed the effective response duration steps, the response term decays to zero; thus, the input, delay placement rules, output, and synthesis method of each actuator action of the prediction model are all determined by the current sampled data, candidate control increments, and the identified matrix.

[0149] In each solution of the rolling optimization, the instantaneous control pre-selected increment is first used as the initial candidate solution, and the candidate increment sequence of branch actuators and hybrid actuators is generated step by step according to the control time domain.

[0150] For each candidate increment generated at a sampling location, it is immediately written into the prediction output sequence according to the corresponding control response prediction step, and the ratio, quality parameters and total flow constraints of that location and subsequent response locations are checked.

[0151] If a candidate increment causes any constraint to go out of bounds, then back off in the opposite direction of the candidate increment, and back off until the constraint that went out of bounds first reaches the boundary.

[0152] If other constraints cannot be satisfied after rollback, the increments that have already satisfied the constraints are fixed, and the remaining increments of the actuator in the current control time domain are skipped; after all candidate solutions have completed constraint checks, the cost is calculated according to the prediction deviation, the degree of deviation relative to the pre-selected increment, and the control magnitude, and candidate solutions whose costs satisfy the preset minimization condition and the constraints are satisfied are retained;

[0153] The current cycle ends when the preset number of iterations is reached, the cost change is less than the convergence criterion twice consecutively, or the number of candidate solutions is empty; when the number of candidate solutions is empty, the process switches to the predetermined real-time control pre-selected incremental limiting output path.

[0154] The export ratio prediction results are used to calculate the difference between the predicted export ratio and the target export ratio, and the export quality prediction results are used to calculate the difference between the predicted export quality parameters and the target quality parameters. After the export quality parameters are reached, the measured values ​​and predicted values ​​at the same export time are compared to form the quality model prediction residuals.

[0155] Among them, the cost term takes into account the difference between the predicted export ratio and the target ratio, the difference between the predicted export quality parameters and the target quality parameters, the deviation of the control increment from the real-time control pre-selected increment, and the amplitude of the control quantity itself, in order to limit the action range of the execution end while meeting the export ratio and export quality requirements.

[0156] The weights of each cost item are determined based on the allowable deviation of the proportion, the allowable deviation of the quality, and the cost of the actuator's actions, and are adjusted using historical data before commissioning;

[0157] When a certain quality parameter is invalid, its corresponding cost item is not used as a valid constraint item; the upper and lower limits of the control quantity, the increment limit, the change rate limit, and the upper and lower limits of the total mixing flow rate together constitute the execution constraints, which are used to ensure that the actuators such as pumps, valves, and mixers operate within the allowable operating range;

[0158] This implementation incorporates the exit status within several future sampling periods into the control calculation through rolling optimization, and determines the compensation action based on the time differences between the execution end response, the mixing process, and the exit detection.

[0159] When no feasible solution is found, the instantaneous control pre-selected increment is used as the output after sequentially passing through incremental limiting, rate limiting and control quantity limiting to ensure that the control process continues to operate under constraints; the limiting is executed sequentially according to the current actuator control quantity, the control quantity output in the previous sampling period and the allowable rate of change.

[0160] If the increment after the previous level of limiting still causes the control quantity to exceed the upper and lower limits, then the remaining available increment corresponding to the upper and lower limits of the control quantity is used to replace it, thereby ensuring that the final output simultaneously meets the increment, rate and control quantity constraints.

[0161] Furthermore, when the sensor diagnosis is normal, the actuator is not limited, and no formula switching is performed, the model is used to predict the residual and its changes within the sliding window to determine whether the model parameters and time delay need to be updated.

[0162] The quality model prediction residual is a residual vector obtained by comparing the measured value and the predicted value at the same exit time after the exit quality parameters are predicted using the current delay parameter and the current quality prediction model.

[0163] Updates are only initiated when the residuals reach the preset identification trigger conditions within the sliding window. The identification trigger conditions include at least the following: the number of valid data points within the window reaches the preset number, the absolute value of the residuals continuously exceeds the normal error band of the model, and the control input has effective changes sufficient to distinguish the actions of each actuator.

[0164] If the control input remains basically unchanged, the residual is caused by a single abnormal sampling point, or the actuator is in a dead zone or limited state, then no update will be triggered; the normal error band of the model is determined based on the residual distribution of the stable data before commissioning, and the preset number is the number of effective sampling points that can cover at least one maximum response cycle;

[0165] Specifically, during online updates, the quality model parameters are kept unchanged. The control increment sequences of each branch actuator and the hybrid actuator are shifted only within the current delay step and its adjacent candidate step range. The shifted control changes are then time-corresponded to the changes in the outlet quality parameters within the same window.

[0166] For each candidate delay, calculate the correlation and the corresponding change in prediction residuals, and prioritize the candidate delays that can form a clear correlation and reduce the prediction residuals;

[0167] When the evaluation difference between candidate delays does not meet the preset discrimination condition, the original delay remains unchanged; after determining the target delay, the delay is adjusted only in the target direction for no more than one sampling step at a time, and the model parameters and quality sensitivity matrix are updated recursively based on the changes in control quantity and the changes in outlet quality parameters.

[0168] The evaluation of candidate delays considers the correlation strength, the stability of the correlation within the window sub-interval, and the amount of residual change; a single-step adjustment is only performed when the target candidate delay has a stable correlation with the current delay and the residual decreases beyond the identification threshold.

[0169] Historical samples used for online identification are entered into a sliding window of finite length according to a uniform sampling period. The window length must at least cover the current maximum control response prediction steps and its effective response duration.

[0170] Each sampling cycle first performs a time continuity check and a sensor validity check on the sample, and only samples with normal time intervals, confirmed control quantity feedback, and valid outlet quality parameters are sent to the identification window.

[0171] Invalid samples are not included in the correlation calculation, and the time interval between the previous valid sample and the next valid sample is retained; when the window capacity reaches the upper limit, the earliest sample is removed in a first-in-first-out manner to avoid the identification calculation load from increasing with the running time.

[0172] If there is a lack of effective stimulus for several consecutive sampling periods, the window is only used to retain the historical state and does not update the parameters; the identification state will only be re-entered when the effective stimulus is restored and continues for a preset period.

[0173] The recursive identification adopts a recursive least squares process with a forgetting factor: the control increment, historical response term and bias term of each actuator at the current sampling time and its determined delay position are composed of a regression vector, and the measured increment of the corresponding export ratio or export quality parameter is used as the regression target.

[0174] For each valid sampling point obtained, the identification residual is calculated. The gain vector is updated sequentially according to the following set of matrix equations. Parameter vector Covariance Matrix :

[0175]

[0176]

[0177]

[0178]

[0179] in, This is the current parameter vector; For regression vectors; This represents the measured increment between adjacent periods; The preset value range is within The forgetting factor within;

[0180] After the update, the corresponding branch actuator response coefficient and hybrid actuator response coefficient are obtained for each output parameter.

[0181] When the effective change of the regression vector is lower than the preset activation threshold, only the original parameters and covariance are retained, and no parameter correction is performed.

[0182] Therefore, the parameters of the quality prediction model and the quality sensitivity matrix are obtained recursively from the explicit changes in control quantity, changes in outlet parameters and their time shift relationships;

[0183] Among them, the quality sensitivity matrix is ​​used to characterize the direction and intensity of the effect of changes in the control quantity of each actuator on changes in the outlet quality parameters, and the recursive identification method is used to superimpose new data in the current cycle onto the existing model, thereby maintaining the consistency between the model and the current operating conditions.

[0184] Matrix elements are set with upper and lower limits to restrict individual identification results from exceeding the preset range due to short-term fluctuations; when a matrix element reaches the limit value, only that element is kept within the corresponding limit range, without affecting the normal recursive update of other elements;

[0185] The upper and lower limits of matrix elements are determined based on the maximum and minimum stable responses obtained from the step test, the physical action direction of the actuator, and the process allowable variation range of the outlet quality parameters. For elements with known action directions, when the update result is reversed in the direction of action, the original value is maintained or the value is limited to the zero side to avoid sudden changes in action direction caused by noise.

[0186] Furthermore, when switching formulas, the system acquires the original formula, the new formula, and the expected outlet switching time, generates a smooth outlet target ratio transition trajectory, and acts in advance according to the predicted steps of each branch control response.

[0187] The advance action time is determined separately based on the total predicted delay of each branch to ensure that the formula changes at the exit of different branches can reach the predetermined trajectory;

[0188] The transition trajectory consists of a monotonically segmented linear trajectory from the original formula to the new formula or a trapezoidal trajectory with a continuous slope. Its starting and ending points are the original formula and the new formula, respectively. The transition slope does not exceed the allowable rate of change of each branch actuator and the total mixing flow.

[0189] When the expected switching time is shorter than the shortest time required for any branch to reach the constraint, the shortest feasible time is used to replace the expected switching time.

[0190] When the advance action time calculated by a branch according to its control response prediction steps is earlier than the current executable time, the control command that failed due to hysteresis is not executed. Instead, the action start point of the branch is set to the current executable time, and the target flow transition trajectory of the branch is shifted backward and widened according to the current control quantity, increment limit, change rate limit and upper and lower limits of the mixed total flow of the actuator.

[0191] In cases where the switch cannot be completed at the originally expected time, the start and end times of the export target ratio trajectory will be postponed to the shortest feasible time that satisfies all branch constraints, and the trajectory endpoint will remain the new formula target ratio.

[0192] Other branches recalculate the advance action time according to the same switching trajectory, thereby avoiding situations where the controller output has missed the action time or different branches use inconsistent output time bases.

[0193] During the switching period, parameter updates are stopped to avoid identifying normal changes during the formula switching process as changes in model parameters. When the weighted norm of the export ratio deviation, the weighted norm of the quality model prediction residual vector, and the rate of change of the above two quantities are all less than the corresponding threshold within a continuous preset sampling period, parameter updates are resumed.

[0194] The recovery threshold is determined based on the residual fluctuation range of the corresponding quantity during the stable operation phase before the switchover and the allowable deviation of the ratio control. The continuous judgment period shall at least cover the prediction delay of the maximum branch control response.

[0195] If any indicator exceeds the threshold again within the cycle, the continuous stability determination will restart. This implementation method coordinates the early action of the branch path with the transition trajectory to reduce the formula misalignment and outlet ratio fluctuation caused by different delays of multiple branches.

Claims

1. An environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control, characterized in that, include: Each branch delivers a stock solution containing effective components to the mixing channel. The flow rate, pressure, control input of the corresponding branch actuator, concentration, temperature and density of the effective components of the stock solution, as well as the mixing energy, outlet quality parameters, target total flow rate, target ratio based on the mass of the effective components and corresponding target quality parameters of the mixing channel are collected and processed synchronously according to a unified sampling cycle. Based on the synchronous data, the effective component mass flow rate, inlet instantaneous ratio, equivalent total delay and control response prediction delay of each branch are determined. The mass flow rate of each branch is mapped to the same outlet time according to the equivalent total delay, and the outlet alignment ratio and outlet target ratio are determined. The inlet and outlet ratio deviations are determined based on the difference between the inlet instant ratio, the outlet aligned ratio, and the corresponding target ratio. The outlet quality deviation is determined based on the difference between the outlet and the target quality parameters. The raw liquid disturbance index is determined based on the changes in raw liquid concentration, temperature, and density. Based on the inlet and outlet ratio deviation, outlet quality deviation and raw liquid disturbance index, the control priority is determined and an instantaneous control pre-selection increment is generated; using the prediction model, with the instantaneous control pre-selection increment as the optimization reference, the predicted compensation increment of the branch and the hybrid actuator is solved, and the final control quantity is synthesized and output. If there is no feasible solution, the instantaneous control pre-selection increment with amplitude limit is output. The quality model prediction residual is generated based on the difference between the measured export quality parameters and the predicted values. The delay parameters, quality prediction model parameters, and quality sensitivity matrix are then updated by combining the control variables and changes in the export quality parameters.

2. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 1, characterized in that, The export quality parameters include online density, online apparent viscosity, online electrical conductivity, and solid phase homogeneity. The unified sampling period satisfy ,in, As a preset stability coefficient, and , For the first The response time from a change in the control input of a branch to the point that the branch flow reaches a preset stable range. The minimum equivalent lag time from a change in export quality parameters to the generation of an identifiable response in the export quality detection value.

3. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 2, characterized in that, Within the preset candidate hysteresis range and sliding data window, according to the first... The correlation between the control input increment sequence of each branch actuator and the measured volumetric flow rate increment sequence of that branch determines the number of actuator lag steps. ; According to the The conveying delay is determined by the effective conveying volume and measured total mixing flow rate of each branch to the mixing channel. The mixing delay is determined based on the mixing volume, measured total mixing flow rate, outlet apparent viscosity, and mixing energy. This mixed delay is then combined with the outlet quality detection link delay to form the equivalent total delay. The control response prediction delay is .

4. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 3, characterized in that, By using a preset density correction model, according to the first The pressure, temperature, and effective component concentration of the stock solution in each branch are used to correct the stock solution density. and in accordance with Determine the effective component mass flow rate, according to Determine the immediate intake ratio; in, For the measured volumetric flow rate of the branch, This represents the mass fraction of the effective component in the original solution. Number of branch roads; Target ratio To meet The effective component mass ratio, the target branch flow rate is .

5. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 4, characterized in that, according to Determine the first The delay steps for each branch are calculated, and each branch is called separately. Historical data at each moment, so that each historical data point corresponds to the same exit moment; The export alignment ratio is: The export target ratio is: Export ratio deviation is: in, The target effective component mass flow rate varies with the target formulation trajectory and remains at the value corresponding to the current target formulation when no formulation switch is performed.

6. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 5, characterized in that, according to Determine the first The normalized deviation of each export quality parameter, among which This is the allowable deviation scale for this parameter; And in accordance with Determine the comprehensive quality indicators, among which, For export quality parameters and quantities, Let be the normalized weight of the q-th export quality parameter, and and ; according to Determine the trend of quality change, when season ,in The minimum control percentage that can be retained. This is the scaling factor for quality improvement; otherwise, let... And utilize before finally controlling the incremental output. Scaling the first branch actuator control increment and the hybrid actuator control increment.

7. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 6, characterized in that, Based on the weighted norm of the inlet ratio deviation, the weighted norm of the outlet ratio deviation, the comprehensive quality index, and the maximum normalized change of the concentration, temperature, and density of the stock solution relative to the previous sampling time or calibration reference value, the inlet ratio deviation index, the outlet ratio deviation index, the mixing homogenization deviation index, and the stock solution disturbance index are determined respectively. For any indicator according to Determine the degree of exceeding the limit, among which This is the preset allowable threshold corresponding to this indicator; When multiple indicators exceed the limit simultaneously, according to The control requirements are generated from large to small and synthesized into the real-time control pre-selected increment, wherein the inlet ratio deviation corresponds to the branch flow adjustment requirement, the outlet ratio deviation corresponds to the prediction compensation requirement, the mixing homogenization deviation corresponds to the mixing energy adjustment requirement, and the raw liquid disturbance corresponds to the target branch flow update requirement.

8. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 7, characterized in that, Real-time control pre-selection increments are generated based on the inlet ratio deviation, the rate of change of the inlet ratio deviation relative to the previous sampling time, and the target branch flow change. according to Determine the number of control response prediction steps and set the prediction time domain. and control time domain ; Using the difference between the predicted export ratio and the target export ratio, the difference between the predicted export quality parameters and the target quality parameters, the difference between the control increment and the real-time control pre-selected increment, and the magnitude of the control increment as costs, the control increment of the branch actuators and the hybrid actuators is optimized in a rolling manner according to preset weights, and upper and lower limits of control quantity, increment limit, change rate limit, and upper and lower limits of total mixed flow are imposed on each actuator. When no feasible solution is found, the incremental limit, rate limit, and control quantity limit are applied sequentially to the pre-selected increment of the instantaneous control before output.

9. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 8, characterized in that, When the sensor diagnosis is normal, the actuator is not in the limiting state, no formula switching has been performed, and the sliding data window meets the preset identification trigger conditions, the online update is performed based on the quality model prediction residual; if any of the above conditions are not met, the current delay parameters, quality prediction model parameters, and quality sensitivity matrix remain unchanged. The control increment sequence of the branch actuator and the hybrid actuator is shifted using the current quality model delay step number and its adjacent candidate step number. The target delay step number is determined based on the correlation between the shift sequence and the change sequence of each outlet quality parameter, and updated with an amplitude of no more than one sampling step at a time. Based on the changes in the control quantities of each actuator and the changes in the quality parameters of each outlet, the parameters of the quality prediction model and the quality sensitivity matrix are updated using a recursive identification method, and the amplitude is limited according to the upper and lower limits set for each matrix element.

10. The environmentally friendly fracturing fluid ratio homogeneous adaptive adjustment method based on advanced control according to claim 9, characterized in that, Obtain the formula switching instruction, which includes the original formula, the new formula, and the expected export switching time, and generate a smooth export target ratio transition trajectory that starts with the original formula target ratio and ends with the new formula target ratio. The sampling time corresponding to the expected exit switching time is denoted as . Using the control response prediction steps determined before the formula changeover begins, according to Determine the first The advance action time of each branch is determined, and the target branch flow transition trajectory that satisfies the actuator constraints and total flow constraints is determined based on the smooth outlet target ratio transition trajectory, the delay of each branch, the original liquid density, the effective component concentration and the target total flow. During formula switching, parameter updates are stopped. Parameter updates resume when the weighted norm of the export ratio deviation and the weighted norm and rate of change of the quality model prediction residual are less than their respective thresholds for consecutive preset sampling periods.

Images