A control method for a fluid transfer pump

Abstract

This invention discloses a control method for a fluid transfer pump, specifically relating to the field of intelligent control technology. The method involves range constraint and normalization processing of pressure signals and control parameters; constructing a nonlinear mapping matrix and calculating the adjustment coefficient using a bilinear interpolation algorithm; estimating the target output flow rate by performing inverse interpolation based on the current pressure value and the adjustment coefficient; introducing a dynamic compensation function to correct the flow rate according to pressure changes and the adjustment coefficient; further comparing the corrected flow rate with the set target, and if the error exceeds a threshold, adaptively updating the local weights of the mapping matrix based on historical deviations. This invention can achieve sensitive response and real-time correction to minute flow fluctuations, possessing high control accuracy, strong dynamic adaptability, and good engineering scalability, making it suitable for scenarios with extremely high requirements for flow rate stability, such as biopharmaceuticals, chip flow control, and microreactors.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and specifically to a control method for a fluid transfer pump. Background Technology

[0002] In the field of micro-volume fluid control, especially in nanoliter-level high-purity fluid delivery scenarios (such as cell microinjection, lab-on-a-chip fluid control systems, and mass spectrometry front-end liquid supply systems), extremely stringent requirements are placed on flow rate stability, response sensitivity, and system adaptability. In such scenarios, the target flow rate is often as low as μL / min, and any minute pressure disturbance, pipeline expansion effect, or fluid viscosity fluctuation can cause flow rate deviation, leading to distortion of downstream experimental conditions or even failure.

[0003] Traditional fluid transfer pump control systems are mostly designed for medium-to-high flow rate transport at the mL / min level and above. Their control methods do not consider structural inertia, hysteresis response, and nonlinear coupling characteristics under microscale operating conditions. At the same time, in order to ensure control accuracy, it is often necessary to introduce high-precision flow meter closed-loop feedback. However, such sensors are not only expensive, but also have problems such as flow channel contamination, response delay, and difficulty in cleaning, which seriously restricts their deployment in high-cleanliness, small-space, and mobile application scenarios.

[0004] Furthermore, existing flow meter-free control methods generally rely on linear models to approximate the relationship between pump pressure and flow velocity, neglecting dynamic changes in local flow resistance (such as elastic changes caused by fatigue of peristaltic pump hoses, pressure drop disturbances caused by valve vibration, etc.), which can cause the system to become unstable or frequently overshoot in specific small intervals, ultimately affecting the overall reliability and repeatability of flow control. Summary of the Invention

[0005] The purpose of this invention is to provide a control method for a fluid transfer pump to address the shortcomings of the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a control method for a fluid transfer pump, comprising:

[0007] The pressure signal of the target fluid delivery pump and the preset opening control ratio are acquired. The pressure signal is then subjected to nonlinear normalization processing to obtain the processed pressure value. ;

[0008] Based on preset physical modeling parameters and process setting flow rate, a local nonlinear mapping matrix M is established to characterize the response characteristics of different pressure-flow rate combination ranges.

[0009] Based on the pressure value after processing With the current target flow rate Locate the segment region index pair (i,j) in matrix M and calculate its bilinear weight coefficient within the region;

[0010] By combining the index (i,j) with the corresponding bilinear interpolation algorithm, the current adjustment coefficient is calculated. ;

[0011] use and Perform inverse interpolation to estimate the theoretical target output flow rate. ;

[0012] Get the pressure value at the previous moment Calculate the pressure change ΔP, and based on ΔP and... Calculate the dynamic compensation coefficient 'a' and the adjustment coefficient. Obtain the correction coefficient , where f(ΔP) is a nonlinear compensation function;

[0013] Will By inverting the interpolation algorithm, the corrected output flow rate is obtained. ;

[0014] Will With user-defined target flow rate An error assessment is performed. If the error exceeds a preset threshold, the corresponding interval weights of the local mapping matrix M are updated based on the deviation between the historical measured flow velocity and the model output flow velocity.

[0015] Preferably, establishing the local nonlinear mapping matrix M includes:

[0016] The pressure value after processing With the current target flow rate Divide the data into a preset two-dimensional coordinate grid interval and determine the segment index pair (i,j) in which it belongs;

[0017] Read the process mapping parameter set stored in the controller, including the pressure-flow rate response coefficient table for different process media, and select the corresponding parameter set according to the current media type;

[0018] Using the index pair (i,j) as the center, extract the adjustment coefficients of its four adjacent points to construct a local submatrix. ;

[0019] Based on the target flow rate set by the process and The combination, in Interpolation is performed within the cycle to construct a nonlinear mapping matrix M for the current control cycle.

[0020] Preferably, locating the segmented region index pair (i,j) in matrix M and calculating its bilinear weight coefficient within the region includes:

[0021] Pressure value after treatment With target flow rate Normalization was performed separately to obtain normalized parameters. and ;

[0022] Based on the set pressure segment interval ΔP and flow velocity segment interval ΔV, calculate the integer part of the index value i. j = integer part ;

[0023] Calculate the horizontal weight coefficient based on the position of the current input value within the segmented interval. Vertical weighting coefficient .

[0024] Preferably, the calculation of the current adjustment coefficient include:

[0025] Based on the determined index pair (i,j), the adjustment coefficient values ​​of the four adjacent points are extracted from the preset nonlinear mapping matrix, and are denoted as follows: and The four points correspond to (i,j), (i+1,j), (i,j+1), and (i+1,j+1) respectively.

[0026] Using the horizontal weighting coefficient s1 to and , and Perform linear interpolation separately to obtain two intermediate interpolated values. and ;

[0027] Further based on the vertical weighting coefficient s2 and Interpolation is performed to obtain the adjustment coefficient corresponding to the current control cycle. .

[0028] Preferably, the use of and Perform inverse interpolation to estimate the theoretical target output flow rate. ,include:

[0029] The controller reads the current pressure value. The sequence of adjustment coefficients corresponding to all velocity nodes in the pressure range is used to construct the adjustment coefficient array for the target pressure row.

[0030] Iterate through the array and find the two nearest neighbors whose corresponding adjustment coefficient values ​​are less than or equal to 1. and greater than or equal to Determine the index of the reverse interpolation interval;

[0031] Calculate the target adjustment coefficient based on the linear back-calculation algorithm. The relative proportion within this range, combined with the velocity interval ΔV, is used to back-estimate the theoretical target velocity. .

[0032] Preferably, the step based on ΔP and The calculation of the dynamic compensation coefficient 'a' includes:

[0033] The controller reads the current pressure value. and the pressure values ​​stored in the previous control cycle Calculate the pressure change ;

[0034] Determine whether the sign and amplitude of ΔP exceed the set disturbance threshold δP. If they do, initiate the dynamic compensation branch.

[0035] According to the calculation formula Calculate the dynamic compensation coefficient α for the current cycle, where γ is an empirical adjustment factor;

[0036] The dynamic compensation coefficient 'a' and the adjustment coefficient The summations form the corrected adjustment coefficient. .

[0037] Preferably, the will By inverting the interpolation algorithm, the corrected output flow rate is obtained. include:

[0038] The controller obtains the corrected adjustment coefficient. Then, reread the current pressure value. The adjustment coefficient array corresponding to the pressure segment where nn is located is used to construct the target pressure row coefficient sequence, which is used as the input data for inverse interpolation.

[0039] Traverse the coefficient sequence to find the AND. Two adjacent adjustment coefficient nodes have values ​​less than or equal to and greater than or equal to This determines the new inverse interpolation interval;

[0040] calculate The relative position ratio within the interval, the ratio being equal to The difference between the lower boundary adjustment coefficient and the interval is divided by the interval span, and the corrected flow velocity is obtained by combining this with the velocity interval interval ΔV. ;

[0041] Will This is the final output flow rate used for actuator adjustment in this cycle.

[0042] Preferably, the step of updating the corresponding interval weights of the local mapping matrix M based on the deviation between historical measured flow velocity and model output flow velocity includes:

[0043] The controller receives the corrected output flow rate calculated in this cycle. and set the target flow rate with the user. By comparing the differences, the flow velocity error is obtained. ;

[0044] Determine whether the absolute value of the error e exceeds the preset calibration threshold δV. If it does, trigger the local model self-calibration process.

[0045] Read the historical measured flow velocity and model output flow velocity under the most recent M operating conditions, calculate the mean deviation of each, and generate a weight update factor λ based on the mean to reflect the long-term offset trend of the current interval.

[0046] The response coefficients at index interval (i,j) are weighted and updated using an update factor λ, so that the matrix... In the formula, This is the updated matrix.

[0047] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0048] 1. This invention overcomes the limitations of traditional linear control models in dealing with microfluidic instability, pressure disturbances, and environmental drift by introducing a nonlinear mapping matrix, multidimensional interpolation algorithm, dynamic compensation mechanism, and local self-calibration method based on historical deviation. Compared to closed-loop feedback methods relying on flow meters, this control method achieves high-precision, low-hysteresis regulation of the target flow rate without increasing additional hardware costs. It is particularly suitable for microfluidic transport processes at the μL / min to mL / min level, significantly improving the system's dynamic responsiveness and flow control stability.

[0049] 2. The matrix adaptive update mechanism proposed in this invention has long-term memory and error self-recovery capabilities. It can dynamically adjust local control parameters according to the current periodic error and historical deviation trend, enabling the control system to have self-learning capabilities, significantly reducing the dependence on manual calibration and return-to-factory maintenance, extending the equipment life cycle, and improving the level of intelligence. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0051] Figure 1 It is a flowchart of the method of the present invention. Specific embodiments

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, rather than all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0053] Embodiment, please refer to Figure 1 As shown, a control method for a fluid delivery pump in this embodiment includes:

[0054] Collect the pressure signal and the preset opening control ratio of the target fluid delivery pump, perform non-linear normalization processing on the pressure signal, and obtain the processed pressure value .

[0055] In this step, the PLC controller collects the pump outlet pressure signal in real time through the connection with the pressure sensor , and participates in the subsequent regulation algorithm as a feedback variable. At the same time, the system receives and stores the pump opening control ratio K (range 0-100%) set by the user or the upper control logic, and participates in the forward flow rate regulation as a driving instruction.

[0056] In view of the characteristics of the pressure signal in the micro-flow condition, such as non-linearity, small fluctuation amplitude, and susceptibility to disturbance, this embodiment introduces a normalization processing model based on the power-law compression mapping function to improve the signal response sensitivity. Specifically, let P_min and P_max be the lower and upper limits measured by the sensor respectively, and the normalization formula of the original pressure signal is as follows: , where 0 <γ <1 is the non-linear compression exponent, which is used to enhance the resolution of the low-pressure section, and the recommended value range is 0.3-0.7.

[0057] To avoid the influence of abnormal pressure disturbance on the calculation accuracy, a pressure boundary constraint mechanism is introduced before normalization, that is, when <P_min, =P_min; when >P_max, =P_max, thus ensuring that the normalized result is P_max. It always stays within the range [0,1].

[0058] In addition, the preset opening control ratio K is standardized and converted into a continuous control variable between 0 and 1, which participates in the adjustment coefficient interpolation process to construct a complete pump control parameter vector. ,K].

[0059] Based on preset physical modeling parameters and process flow rate settings, a local nonlinear mapping matrix M is established to characterize the response characteristics of different pressure-flow rate combination ranges.

[0060] In the control method described above, to achieve high-precision dynamic matching during flow rate regulation, a local nonlinear mapping model needs to be established based on the current operating conditions to quantify the complex correlation between the pressure signal and the flow rate output. Specifically, this includes:

[0061] First, the control system will process the pressure value. and the target flow rate set by the current process Synchronously mapped to a two-dimensional coordinate system. The horizontal axis represents the normalized pressure value (…). The range is 0 to 1), and the vertical axis represents the normalized flow velocity value. (ranging from 0 to 1), both are discretely divided according to the set equidistant grid intervals ΔP and ΔV respectively.

[0062] The segment index pair (i,j) of its grid is calculated as follows:

[0063] i is The integer part obtained by dividing by ΔP;

[0064] j is The integer part obtained by dividing by ΔV.

[0065] Wherein, ΔP and ΔV are preset equal intervals. For example, when the pressure interval [0,1] is divided into 10 segments, then ΔP=0.1; similarly, if the velocity interval [0,1] is divided into 20 segments, then ΔV=0.05.

[0066] Considering the differences in flow behavior of different fluid media (such as organic solvents, buffer salt solutions, and bioreactors) under the same pressure conditions, the controller pre-stores multiple sets of process mapping parameters. Each set of parameters contains a two-dimensional matrix, corresponding to the adjustment coefficients under different pressure and flow rate combinations, and is called the "pressure-flow rate response coefficient table".

[0067] During system initialization or recipe switching, the controller automatically calls up the matching parameter set based on the media type code selected by the operator in the graphical user interface (e.g., type A is aqueous, type B is organic, etc.). For example, if the current medium is buffer solution, the system will load the process parameter matrix table M_C01, numbered "C01".

[0068] Based on the obtained index pair (i,j), the values ​​of the four adjacent nodes of the coordinate point will be extracted from the process parameter matrix, namely: bottom left: M(i,j); bottom right: M(i+1,j); top left: M(i,j+1); top right: M(i+1,j+1).

[0069] The above four points constitute a basic unit region for two-dimensional interpolation, which is constructed as a local submatrix. This submatrix will be used to represent the current input. and The trend of nonlinear control coefficient changes within the corresponding small interval.

[0070] In actual engineering implementation, the process parameter table is stored in the controller's EEPROM or cache RAM in the form of a two-dimensional array. The table lookup and index matching process is automatically completed by the PLC script or embedded control algorithm.

[0071] In obtaining local submatrices Then, the system further based on and At the relative position within the current interval, perform a bilinear interpolation algorithm to accurately estimate the target adjustment coefficient. The relative position ratios β1 and β2 are defined as follows:

[0072] β1 is The proportion of positions within the ΔP interval, i.e. ;

[0073] β2 is The proportion of positions within the ΔV interval, i.e. .

[0074] Final adjustment coefficient Calculated by the following formula:

[0075]

[0076] The above interpolation algorithm can complete the calculation within 10 milliseconds and is suitable for real-time calls within high-frequency control cycles (typically 50ms~100ms).

[0077] The final adjustment coefficient That is, the output value of the nonlinear mapping matrix M that constitutes the current control cycle.

[0078] Based on the pressure value after processing With the current target flow rate Locate the segment region index pair (i,j) in matrix M and calculate its bilinear weight coefficient within the region.

[0079] To achieve high-precision local interpolation control, this embodiment processes the pressure value... With target flow rate The normalization and indexing operations include the following steps:

[0080] The control system first processes the pressure value. and target flow rate Normalization is performed to map both values ​​to the unit interval [0,1] for standardized interpolation calculations. The normalization process is as follows:

[0081] Normalization pressure equal Divide by the maximum set pressure P_max;

[0082] Normalized flow rate equal Divide by the maximum flow rate V_max.

[0083] Among them, P_max and V_max are the upper limit thresholds preset by the system, which are usually determined according to the pump model or process parameters, for example, P_max=5 bar, V_max=1000 mL / min.

[0084] Normalization can effectively solve the problem of inconsistent interpolation intervals caused by differences in pump types or process ranges, and improve the model's versatility and reusability.

[0085] After normalization, and Mapped to a two-dimensional coordinate grid, it is used to locate the segmented region in which it is located. This grid consists of equidistant pressure intervals ΔP and velocity intervals ΔV. The system calculates the corresponding index values ​​according to the following methods:

[0086] Pressure direction index i is Divide by the integer part of ΔP;

[0087] The flow direction index j is Divide by the integer part of ΔV.

[0088] For example, if ΔP = 0.1 and ΔV = 0.05, then when =0.35、 When = 0.22, we have:

[0089] i = integer part (0.35 / 0.1) = 3;

[0090] j = integer part (0.22 / 0.05) = 4.

[0091] This step locates the local cell containing the target input value in the preset mapping matrix M, which is used for subsequent interpolation region extraction.

[0092] After determining the cell containing the input value, it is necessary to further determine its relative position within that cell to facilitate bilinear interpolation. The system calculates the horizontal and vertical interpolation ratios using the following methods:

[0093] Horizontal weighting coefficient Divide by ΔP;

[0094] Vertical weighting coefficient Divide by ΔV.

[0095] The interpolation ratio ranges from [0,1], representing the degree of offset of the current input within the grid interval. For example, in the example above:

[0096] s1=(0.35-3×0.1) / 0.1=0.05 / 0.1=0.5;

[0097] s2=(0.22-4×0.05) / 0.05=0.02 / 0.05=0.4.

[0098] The calculation results of s1 and s2 will be used as weighting factors in the bilinear interpolation algorithm to fuse the values ​​of the four nodes in the matrix and accurately estimate the adjustment coefficient.

[0099] By combining the index (i,j) with the corresponding bilinear interpolation algorithm, the current adjustment coefficient is calculated. .

[0100] To achieve precise control output, this embodiment uses interpolation to determine the adjustment coefficient for the current control cycle, based on the known index (i,j) of the input parameter region and the bilinear interpolation weighting coefficients s1 and s2. This is used to drive the actuator to regulate the flow rate. The process includes:

[0101] The controller first locates the region where the current index (i,j) is located from the loaded nonlinear mapping matrix M, and extracts the adjustment coefficient values ​​corresponding to the four node positions within that region, as defined below: Let (i,j) be the adjustment coefficient at position (i,j). This is the adjustment coefficient at position (i+1,j); This is the adjustment coefficient at position (i,j+1); This is the adjustment coefficient at position (i+1, j+1).

[0102] The four points mentioned above form a two-dimensional interpolation cell, describing the local control response surface within the current input interval. The regulation coefficient matrix is ​​typically preloaded in the controller memory as a two-dimensional array, with each element generated from historical experimental data, simulation modeling, or a self-calibration module.

[0103] Next, the controller uses the calculated horizontal weighting coefficients s1 to respectively... and , and Perform a linear interpolation operation to obtain the intermediate interpolation value of the current pressure normalized position between the lower and upper boundaries. The calculation method is as follows: , indicating that in the j-th row (i.e., the current flow rate segment) Interpolated value of position; , indicating in the (j+1)th row The interpolated value of the position. The interpolation is implemented using a weighted average method. The programmable logic controller (PLC) performs this calculation through the floating-point arithmetic module, with a calculation accuracy better than 0.01.

[0104] In obtaining and Subsequently, the system further utilizes the longitudinal weighting coefficient s2 to perform a second linear interpolation on these two intermediate values, calculating the adjustment coefficient corresponding to the current flow velocity normalization position. The specific calculation method is as follows: The adjustment coefficient This refers to the control parameters that best match the combination of target flow rate and current pressure within the current control cycle. These parameters will then be used to deduce the theoretical flow rate or directly drive the pump actuator for adjustment.

[0105] use and Perform inverse interpolation to estimate the theoretical target output flow rate. .

[0106] This embodiment obtains the current adjustment coefficient. and the pressure value after treatment Based on this, a reverse interpolation method based on matrix lookup and linear back-calculation algorithms is used to estimate the corresponding theoretical target output velocity. Specifically, it includes:

[0107] The controller first determines the pressure value after processing. For the pressure segment index i, read all the data in the i-th row of the nonlinear mapping matrix M. This row contains the preset adjustment coefficients for all flow velocity ranges (i.e., column j from 0 to n), representing the trend of the adjustment coefficients changing with the flow velocity under a fixed pressure segment i.

[0108] Extract this line into a one-dimensional array K_array[i], denoted as: This array constitutes an approximate monotonic function of the adjustment coefficient as a function of flow velocity, providing a data basis for subsequent back-interpolation.

[0109] The controller iterates through the above K_array[i] array, comparing the adjustment coefficient values ​​one by one, and searches for a pair of adjacent nodes that satisfy the following conditions:

[0110] The adjustment coefficient value of one of the nodes is less than or equal to the current value. ;

[0111] The adjustment coefficient value of adjacent nodes is greater than or equal to the current value. .

[0112] Let the indices of these two nodes be j and j+1, representing the current adjustment coefficient. If the value falls within the index range [j, j+1], and there are identical nodes, the corresponding flow rate can be directly taken; otherwise... If the flow rate exceeds the array boundary, the boundary flow rate is used as the estimation result.

[0113] After determining the adjustment coefficient range, the controller uses a linear back-calculation algorithm to calculate... The relative position ratio s within this interval, where: Combined with the preset flow velocity interval ΔV, the theoretical target output flow velocity is... The calculation formula is: ;

[0114] This formula estimates the target flow velocity corresponding to the adjustment coefficient using a linear weighting method, and is suitable for interpolation scenarios with approximately monotonic changes.

[0115] Get the pressure value at the previous moment Calculate the pressure change ΔP, and based on ΔP and... Calculate the dynamic compensation coefficient 'a' and the adjustment coefficient. Obtain the correction coefficient , where f(ΔP) is a nonlinear compensation function.

[0116] This embodiment introduces a dynamic compensation mechanism based on the pressure change ΔP, which corrects the current adjustment coefficient using a compensation coefficient α, ultimately generating a corrected adjustment coefficient. The process specifically includes:

[0117] The controller records the pressure value after processing in each control cycle. and call the previous cycle Stored pressure values Calculate the pressure change ΔP between two cycles. The formula is: ΔP equals... minus ;in, and All values ​​are normalized pressure values, ranging from 0 to 1. ΔP can be positive (increasing) or negative (decreasing). This step is a prerequisite for determining whether to trigger the compensation logic.

[0118] A disturbance threshold δP is introduced as the trigger criterion. The controller determines whether the absolute value of ΔP exceeds δP: if |ΔP|>δP, it is considered a significant disturbance and enters the dynamic compensation branch; otherwise, it remains unchanged. If unchanged, no compensation operation will be performed. The disturbance threshold δP is a system-defined constant, determined based on the process response sensitivity, with a recommended range of 0.02–0.05. This determination mechanism effectively suppresses interference responses caused by sensor noise or system inertia, improving control stability.

[0119] After determining that the disturbance condition is met, the controller bases its control on ΔP and the current period adjustment coefficient. The dynamic compensation coefficient α is calculated according to the following compensation function: Wherein, γ is an empirical adjustment factor with positive and negative directional adjustment capabilities, and its value range is typically set from 0.3 to 1.5, used to adjust the compensation intensity according to the actual response speed of the system. This compensation function has the following characteristics:

[0120] When the pressure suddenly increases (ΔP is positive), the adjustment coefficient is increased positively to predict the downward trend of the flow rate in advance;

[0121] When the pressure drops suddenly (ΔP is negative), the regulation coefficient is reduced negatively to prevent system overshoot;

[0122] Compensation amount and It is proportional to ensure that the compensation amplitude adapts to different control intensities.

[0123] The controller will compare the compensation coefficient α with the current adjustment coefficient. The superposition generates a correction coefficient for the final back interpolation or flow rate control command in this cycle. The calculation method is as follows: .

[0124] To avoid the control coefficients from exceeding the limit due to compensation, Set a range constraint to force the value to be within the range [0,1], and take the boundary value if it exceeds the range. For example:

[0125] like If less than 0, then set =0;

[0126] like If it is greater than 1, then set =1.

[0127] Corrected adjustment coefficient This will be used for the next step of target flow velocity estimation, enabling the system to achieve adaptive control compensation under disturbance conditions.

[0128] This embodiment, based on the calculated basic compensation coefficient α, introduces a nonlinear compensation function f(ΔP) to adjust the coefficient. A flexible correction is performed to obtain the corrected adjustment coefficient. This is used to drive the next step of flow rate estimation or control execution. The process includes:

[0129] In this embodiment, the nonlinear compensation function f(ΔP) is a monotonically increasing function used to enhance the system's response to large pressure changes and weaken the moderating effect of small disturbances. A typical expression is as follows: Where: ΔP is the current pressure change (unit: normalized pressure, range: -1 to +1); β is the nonlinear compressibility coefficient, used to adjust the suppression amplitude of the function when ΔP is large, with a recommended value range of 3 to 10; the function form ensures that when ΔP is close to 0, f(ΔP) is approximately linear; when ΔP is too large, f(ΔP) gradually saturates, thus avoiding overshoot. In engineering implementation, the f(ΔP) function can be preset to a lookup table form or calculated by embedding a function calculation module in the controller to reduce the computational burden.

[0130] The controller calls the compensation coefficient α calculated in the previous step and multiplies it by the nonlinear function f(ΔP) to obtain the nonlinear correction δK for the current cycle, which is expressed as: Where: f(ΔP) determines the magnitude of the correction and compression characteristics; the value of δK can be positive or negative, depending on the direction of pressure rise or fall.

[0131] The controller will adjust the correction amount δK with the current cycle's base adjustment coefficient. By performing weighted summation, the final correction adjustment coefficient used to control the output is obtained. Its expression is: To ensure the adjustment coefficient Stay within the effective range, for Apply upper and lower boundary constraints, that is: if If less than 0, force it to be set to 0; if If the value is greater than 1, it will be forcibly set to 1.

[0132] Final result It can be used to re-estimate the target output flow rate in the inverse interpolation algorithm, or directly as a control signal to drive pump speed adjustment.

[0133] Will By inverting the interpolation algorithm, the corrected output flow rate is obtained. .

[0134] This embodiment obtains the correction adjustment coefficient. Based on this, the corrected output flow rate is obtained by re-performing the inverse interpolation calculation. The process includes:

[0135] The controller first determines the current post-processing pressure value. The pressure segment index i is used, and all velocity range adjustment coefficients corresponding to that row (the i-th row) in the nonlinear mapping matrix M are reread and constructed into a one-dimensional array K_array[i], that is: This array represents the variation of the regulation coefficient with flow rate under a fixed pressure range i, and serves as a reference sequence for back-interpolation. If the mapping matrix is ​​a two-dimensional array of m×n, the controller extracts only the n regulation coefficient values ​​from the i-th row as subsequent interpolation inputs.

[0136] The controller iterates through the values ​​of each adjacent node in K_array[i] to find a set of values ​​that make... An interval falling within this interval satisfies the following condition: K_array[i][j] is less than or equal to K_array[i][j+1] is greater than or equal to Once an index j that meets the above conditions is found, then it can be determined that... It falls within the index range [j, j+1] and serves as the reference range for inverse interpolation.

[0137] If no completely matching interval is found (e.g.) If the value exceeds the minimum or maximum value, the system will execute the boundary protection strategy and directly... Set it to the corresponding boundary velocity value (such as 0 or V_max) to prevent interpolation overflow.

[0138] After determining the interpolation interval, the controller calculates... The relative position ratio s within this interval is calculated as follows: The proportionality coefficient s∈[0,1] reflects Positional relationship within the interpolation interval.

[0139] Based on the system's preset flow rate segmentation interval ΔV, the final corrected output flow rate is... Calculated using the following formula: Where: j×ΔV represents the initial flow velocity of the current interval; s×ΔV represents... Interpolation offset within this interval.

[0140] The final calculated corrected flow rate The target output value for this control cycle is sent directly to the controller output port to drive the regulating actuator (such as a servo motor or regulating valve) to adjust the opening and achieve pump speed regulation.

[0141] Will With user-defined target flow rate An error assessment is performed. If the error exceeds a preset threshold, the corresponding interval weights of the local mapping matrix M are updated based on the deviation between the historical measured flow velocity and the model output flow velocity.

[0142] The controller outputs the flow rate upon completion of this control cycle. After calculation, immediately match the target flow rate set by the operator. Real-time comparison is performed to calculate the flow velocity error e for the current cycle, which is expressed as: e equals minus This error value reflects the degree of deviation between the model's adjustment accuracy and actual requirements; it can be positive (output too high) or negative (output too low). The error unit can be configured as mL / min, μL / min, or normalized flow rate.

[0143] The controller sets a set of calibration trigger threshold parameters δV (error threshold). The calibration process is triggered when the following condition is met: if |e|>δV, then local model adaptive update is initiated; where δV is a user-defined or experience-based value, with a recommended range of 1%~3% of the target flow rate. For example, when... When the flow rate is 100 mL / min, δV can be set to 2.0 mL / min. If the error does not exceed the threshold, the controller retains the current model and switches to the next cycle of control to avoid overfitting or noise response.

[0144] If the calibration process is initiated, the controller reads the historical measured flow rate V_real and the model output flow rate V_model from the data cache module for the most recent M control cycles, and calculates the deviation for each cycle: Then, the average of these M deviation values ​​from k=1 to M is calculated to obtain the long-term offset trend. : equal Divide by M; then according to Consistent with the direction of the current error e, a local update weight λ is constructed, and its calculation model is as follows: Where: η is an adjustable learning rate, with a recommended range of 0.05~0.2; sign(e) indicates the error direction (positive or negative); λ is used to adjust the magnitude of the local weight correction of the matrix.

[0145] The controller locates the index interval (i,j) into which the current input falls, and adjusts the weights of the corresponding elements in the nonlinear mapping matrix M. The updated local coefficients are denoted as: This update method takes into account both the error direction and historical offset, and is suitable for scenarios requiring localized accuracy enhancement. In particular, it can achieve intelligent self-recovery when fixed parameter models experience nonlinear offsets due to factors such as aging, media changes, and system fatigue.

[0146] After the controller completes the update, it reloads the new matrix DM into the interpolation calculation module and updates the data storage unit simultaneously to ensure the traceability and consistency of the calibration process.

[0147] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A control method for a fluid transfer pump, characterized in that: include: The pressure signal of the target fluid delivery pump and the preset opening control ratio are acquired. The pressure signal is then subjected to nonlinear normalization processing to obtain the processed pressure value. ; Based on preset physical modeling parameters and process setting flow rate, a local nonlinear mapping matrix M is established to characterize the response characteristics of different pressure-flow rate combination ranges. Based on the pressure value after processing With the current target flow rate Locate the segment region index pair (i,j) in matrix M and calculate its bilinear weight coefficient within the region; By combining the index (i,j) with the corresponding bilinear interpolation algorithm, the current adjustment coefficient is calculated. ; use and Perform inverse interpolation to estimate the theoretical target output flow rate. ; Get the pressure value at the previous moment Calculate the pressure change ΔP, and based on ΔP and... Calculate the dynamic compensation coefficient 'a' and the adjustment coefficient. Obtain the correction coefficient , where f(ΔP) is a nonlinear compensation function; Will By inverting the interpolation algorithm, the corrected output flow rate is obtained. ; Will With user-defined target flow rate An error assessment is performed. If the error exceeds a preset threshold, the corresponding interval weights of the local mapping matrix M are updated based on the deviation between the historical measured flow velocity and the model output flow velocity.

2. The control method for a fluid transfer pump according to claim 1, characterized in that: The establishment of the local nonlinear mapping matrix M includes: The pressure value after processing With the current target flow rate Divide the data into a preset two-dimensional coordinate grid interval and determine the segment index pair (i,j) in which it belongs; Read the process mapping parameter set stored in the controller, including the pressure-flow rate response coefficient table for different process media, and select the corresponding parameter set according to the current media type; Using the index pair (i,j) as the center, extract the adjustment coefficients of its four adjacent points to construct a local submatrix. ; Based on the target flow rate set by the process and The combination, in Interpolation is performed within the cycle to construct a nonlinear mapping matrix M for the current control cycle.

3. The control method for a fluid transfer pump according to claim 2, characterized in that: The step of locating the segmented region index pair (i,j) in matrix M and calculating its bilinear weight coefficient within the region includes: Pressure value after treatment With target flow rate Normalization was performed separately to obtain normalized parameters. and ; Based on the set pressure segment interval ΔP and flow velocity segment interval ΔV, calculate the integer part of the index value i. j = integer part ; Calculate the horizontal weight coefficient based on the position of the current input value within the segmented interval. Vertical weighting coefficient .

4. The control method for a fluid transfer pump according to claim 1, characterized in that: The current adjustment coefficient is calculated. include: Based on the determined index pair (i,j), the adjustment coefficient values ​​of the four adjacent points are extracted from the preset nonlinear mapping matrix, and are denoted as follows: and The four points correspond to (i,j), (i+1,j), (i,j+1), and (i+1,j+1) respectively. Using the horizontal weighting coefficient s1 to and , and Perform linear interpolation separately to obtain two intermediate interpolated values. and ; Further based on the vertical weighting coefficient s2 and Interpolation is performed to obtain the adjustment coefficient corresponding to the current control cycle. .

5. The control method for a fluid transfer pump according to claim 1, characterized in that: The use and Perform inverse interpolation to estimate the theoretical target output flow rate. ,include: The controller reads the current pressure value. The sequence of adjustment coefficients corresponding to all velocity nodes in the pressure range is used to construct the adjustment coefficient array for the target pressure row. Iterate through the array and find the two nearest neighbors whose corresponding adjustment coefficient values ​​are less than or equal to 1. and greater than or equal to Determine the index of the reverse interpolation interval; Calculate the target adjustment coefficient based on the linear back-calculation algorithm. The relative proportion within this range, combined with the velocity interval ΔV, is used to back-estimate the theoretical target velocity. .

6. The control method for a fluid transfer pump according to claim 1, characterized in that: The basis of ΔP and Calculate the dynamic compensation coefficient a include: The controller reads the current pressure value. and the pressure values ​​stored in the previous control cycle Calculate the pressure change ; Determine whether the sign and amplitude of ΔP exceed the set disturbance threshold δP. If they do, initiate the dynamic compensation branch. According to the calculation formula Calculate the dynamic compensation coefficient α for the current cycle, where γ is an empirical adjustment factor; The dynamic compensation coefficient 'a' and the adjustment coefficient The summations form the corrected adjustment coefficient. .

7. The control method for a fluid transfer pump according to claim 1, characterized in that: The By inverting the interpolation algorithm, the corrected output flow rate is obtained. include: The controller obtains the corrected adjustment coefficient. Then, reread the current pressure value. The adjustment coefficient array corresponding to the pressure range is used to construct the target pressure row coefficient sequence, which is then used as input data for inverse interpolation. Traverse the coefficient sequence to find the AND. Two adjacent adjustment coefficient nodes have values ​​less than or equal to and greater than or equal to This determines the new inverse interpolation interval; calculate The relative position ratio within the interval, the ratio being equal to The difference between the lower boundary adjustment coefficient and the interval is divided by the interval span, and the corrected flow velocity is obtained by combining this with the velocity interval interval ΔV. ; Will This is the final output flow rate used for actuator adjustment in this cycle.

8. The control method for a fluid transfer pump according to claim 7, characterized in that: The method for updating the corresponding interval weights of the local mapping matrix M based on the deviation between historical measured flow velocity and model output flow velocity includes: The controller receives the corrected output flow rate calculated in this cycle. and set the target flow rate with the user. By comparing the differences, the flow velocity error is obtained. ; Determine whether the absolute value of the error e exceeds the preset calibration threshold δV. If it does, trigger the local model self-calibration process. Read the historical measured flow velocity and model output flow velocity under the most recent M operating conditions, calculate the mean deviation of each, and generate a weight update factor λ based on the mean to reflect the long-term offset trend of the current interval. The response coefficients at index interval (i,j) are weighted and updated using an update factor λ, so that the matrix... In the formula, This is the updated matrix.

Images