Injection pump automatic calibration method and system based on flow velocity stability

Abstract

The invention relates to the technical field of injection pump calibration, in particular to an injection pump automatic calibration method and system based on flow velocity stability, and the method comprises the following steps: S1, collecting flow velocity data in the operation process of an injection pump, and recognizing a boundary section of stable-to-unstable conversion based on a dynamic window; constructing a boundary drift indication map reflecting response delay, amplitude fluctuation and a rate inflection point; s2, identifying a consistency offset error mode from the atlas in combination with a multi-cycle drift behavior, marking a start window, a transfer window and a peak window, and generating a historical drift memory sequence; and S3, constructing a cross-cycle inheritance type calibration parameter template based on the memory sequence, and adaptively adjusting key parameters in input control logic to realize pre-calibration response before the injection pump is unstable. According to the method, by constructing the boundary drift map, the inheritance type calibration template and the dynamic parameter fusion mechanism, accurate pre-adjustment calibration and self-adaptive compensation optimization of the injection pump before the injection pump is in the unstable state are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of injection pump calibration, and particularly relates to an injection pump automatic calibration method and system based on flow rate stability. BACKGROUND

[0002] As a key device in the clinical drug delivery process, the running accuracy and stability of the injection pump directly affect the efficacy and safety of drug infusion. In particular, in the continuous infusion process, the injection pump needs to maintain a stable flow rate output to avoid abnormal flow rate caused by device performance fluctuations, changes in pipeline resistance or external interference, resulting in fluctuations in drug concentration or delays in drug delivery. Therefore, how to effectively perceive and adjust the flow rate stability during the operation of the injection pump is an important research direction to ensure the quality of clinical treatment.

[0003] Existing injection pump calibration methods mostly rely on periodic static calibration or passive adjustment after sudden abnormalities, and cannot perceive the flow rate boundary fluctuation characteristics in the running process in real time, especially for the transition process between stable and unstable states. The recognition ability is weak, it is difficult to judge the trend of the flow rate deviation that will occur in time, in addition, the existing scheme lacks a historical inheritance mechanism for cross-cycle error behavior, and cannot use the repeated deviation patterns in the past running data for forward-looking response, resulting in the device triggering compensation measures only when it is close to the unstable state, which has the problems of response lag and control delay. SUMMARY

[0004] The present application provides an injection pump automatic calibration method and system based on flow rate stability, which aims to construct a boundary drift indication atlas through a dynamic window mechanism, mine consistent deviation error patterns in multiple cycle operations, and generate a calibration parameter template that can be inherited across cycles, and apply it to the control logic to realize adaptive pre-adjustment response. By identifying non-stable signs in advance and dynamically adjusting the calibration parameters, the flow rate control accuracy and response efficiency of the injection pump in the continuous infusion process are improved, thereby significantly enhancing the infusion stability and safety and reliability of clinical use.

[0005] An injection pump automatic calibration method based on flow rate stability, comprising the following steps:

[0006] S1, collecting flow rate data in the continuous operation process of the injection pump, and identifying a boundary transition section in the transition process from a stable state to a non-stable state based on a dynamic window mechanism, constructing a boundary drift indication atlas, the boundary drift indication atlas reflecting the response delay, amplitude fluctuation and rate inflection point distribution of the injection pump in the stable critical region in different operation cycles;

[0007] S2, based on the constructed boundary drift indication atlas, combined with multi-cycle drift behavior comparison, identify the error mode set showing consistent offset characteristics, and mark the starting window, transition window and peak window associated with each error mode to generate a historical drift memory sequence;

[0008] S3, based on the historical drift memory sequence, generate a cross-cycle inheritance calibration parameter template, input the cross-cycle inheritance calibration parameter template into the injection pump control logic for adaptive adjustment, and dynamically update the definition of the stable state trigger threshold, response delay compensation and rate return curve in the cross-cycle inheritance calibration parameter template to realize the early pre-calibration response of the injection pump before the induction of the unstable boundary.

[0009] Optionally, the S1 comprises:

[0010] S11, during the continuous operation of the injection pump, real-time acquisition of flow rate data sequence, and construction of sliding window group according to preset time interval, statistical analysis of flow rate data sequence in each sliding window, extraction of mean, range and slope change rate to capture the initial signs of flow rate fluctuation, and adaptive adjustment of sliding window according to flow rate change frequency by setting dynamic sliding window width;

[0011] S12, in the obtained sliding window sequence, analyze the slope mutation point, fluctuation amplitude improvement area and response delay phenomenon between adjacent windows, identify the boundary transition section representing the gradual transition from stable state to disturbed state, and construct a boundary drift indication atlas for representing the rate change mode, amplitude expansion structure and delay response amplitude of the injection pump near the instability point in multiple operation cycles.

[0012] Optionally, the S11 comprises:

[0013] S111, real-time acquisition of flow rate value at the current time from the injection pump flow monitoring unit to form a time sequence flow rate data time sequence

[0014] S112, divide the flow rate data sequence into a plurality of adjacent sliding windows, each window size is w0, sliding step is ε, form a window sequence, denoted as:

[0015]

[0016] Wherein, W k is the kth sliding window, N is the total number of sliding windows, w k is the current width of the kth sliding window, t k is the starting point index of the kth sliding window;

[0017] S113, for each sliding window W kCalculate the characteristic index, including mean, range, slope change rate;

[0018] S114, introduce window fluctuation coefficient ξ k Perform normalization judgment to determine whether the window needs to be contracted or expanded according to ξ k Dynamically adjust the width of the next window;

[0019]

[0020] Where, δ w is the window width step size of each adjustment, w k+1 is the adaptive width of the k+1 sliding window, τ1, τ2 are the upper and lower thresholds of the fluctuation coefficient respectively.

[0021] Optionally, the S12 comprises:

[0022] S121, for any two adjacent sliding windows W k and W k+1 , measure the degree of mutation with their respective average slope change rate, define the slope mutation quantity Δθ k ;

[0023] S122, define the amplitude index of each window as the range of the sliding window, then the fluctuation lifting quantity is Δρ k ;

[0024] S123, construct the delay response index δ k , which is used to measure the tailing effect of gradual change in multiple windows when the disturbance signal enters, when δ k >τ δ , it is considered that there is a response delay section, wherein τ δ is the delay threshold;

[0025] S124, define the boundary awareness score function B k , which is used to comprehensively judge whether the window W k is in a stable to disturbance boundary in multiple dimensions, when B k >B B , mark W k as a boundary transition section, wherein τ B is the boundary score threshold;

[0026] S125, mark all windows judged as boundary transition sections according to the time axis to form a graph, wherein the X axis is the time window number k, and the Y axis is the index B k , Δθ k , Δρ k , δ kThe superimposed curves are highlighted, and the perturbation boundary window intervals and their perturbation types (slope type, amplitude type, delay type) are marked. The final output is a boundary drift indicator map.

[0027] Optionally, S2 includes:

[0028] S21, Suppose the operation process is divided into C historical operation cycles, and extract the corresponding boundary score map in each operation cycle c. And construct trajectory matrix B cyc ;

[0029] S22, Construct a set of window-aligned indexes with consistent drift behavior across cycles. Select window points that appear in similar positions and have similar scoring patterns from multiple cycles;

[0030] S23, using density clustering DBSCAN to align the window-aligned index set Clustering is performed on the window points in the data, and each cluster forms an error pattern unit ε. i For each error mode unit ε i Extract key windows, including the start window. (Window where the score first exceeds the threshold), transition window (Location with the highest rate of change in rating), peak window (The location where the score reaches a local maximum);

[0031] S24, structuring each error pattern into a historical drift memory sequence item.

[0032] Optionally, S3 includes:

[0033] S31. Based on the constructed historical drift memory sequence, the key position windows (start, transition, peak) of different perturbation type error modes are summarized, and their corresponding calibration behavior characteristics are statistically analyzed, including stable trigger time, average score offset, recovery rate, and cross-cycle inherited calibration parameter template is constructed.

[0034] S32, the constructed cross-cycle inherited calibration parameter template is input to the injection pump control logic unit, and the current boundary scoring trajectory is monitored in real time. If the window behavior is identified to match the error mode in the cross-cycle inherited calibration parameter template, the corresponding calibration compensation operation is triggered, including correcting the target flow rate setting, switching the feedback response strategy in advance, or injecting slope positivity factor, so as to complete the pre-adjustment calibration before the unstable state officially occurs.

[0035] S33 continuously receives new boundary identification results during the operation of the syringe pump and matches and merges them with the cross-cycle inherited calibration parameter template to dynamically adjust key calibration parameters, including trigger threshold, delay compensation amount and homing strategy.

[0036] Optionally, S31 includes:

[0037] S311, For error modes τ of different disturbance types, statistically analyze their corresponding calibration behavior characteristics, including the stable trigger point. Average rating offset ΔB τ Recovery rate

[0038] S312 categorizes and summarizes the error mode calibration behavior characteristics corresponding to different disturbance types, and constructs a cross-cycle inherited calibration parameter structure template including stable trigger time, average score offset, and recovery rate.

[0039] Optionally, S32 includes:

[0040] S321, During the continuous operation of the infusion pump, the control logic unit collects the boundary score values ​​of the most recent L sliding windows in real time, forming the current score trajectory S. cur ;

[0041] S322, from the cross-cycle inherited calibration parameter structure template In the process, the template reference scoring trajectory corresponding to each type of perturbation τ is extracted, and the matching degree κ between the current scoring trajectory and the template reference scoring trajectory is calculated using weighted cosine similarity. τ When the disturbance type τ * satisfy If so, it is determined that the current operating state of the injection pump has exhibited a disturbance type τ. * Similar behavior to the corresponding error mode should trigger the next calibration response phase, where θ match The matching threshold;

[0042] S323, when the matching degree κ τ When the matching threshold is exceeded, the injection pump control logic triggers calibration compensation operations according to the corresponding disturbance type, including correcting the target flow rate setting, switching the feedback response strategy in advance, or the injection slope positivity factor, to achieve advance pre-adjustment calibration and dynamic correction of unstable states.

[0043] Optionally, S33 includes:

[0044] S331: After each sliding window scoring cycle is completed, a new scoring trajectory segment is recorded in real time, and its scoring mean offset, scoring slope change value and delayed response feature value are extracted to form candidate update samples.

[0045] S332, calculate the weighted cosine similarity between the new scoring trajectory segment and the current template reference trajectory, and determine whether it belongs to the same error type as the template. If the similarity κ... update Above the conservative fusion threshold θ update Then it will be included in the update and integration;

[0046] S333 performs a sliding weighted fusion update on the average score offset, average delay time compensation value, and score slope recovery factor.

[0047] An automatic calibration system for an infusion pump based on flow rate stability, used to implement the aforementioned automatic calibration method for an infusion pump based on flow rate stability, includes the following modules:

[0048] Flow rate acquisition module: Acquires flow rate data during continuous operation of the syringe pump, identifies boundary transition sections based on a dynamic window mechanism, and constructs a boundary drift indicator map that reflects the characteristics of stable critical response;

[0049] Error pattern recognition module: Based on the boundary drift indicator map, it compares the drift behavior over multiple cycles, extracts the consistency shift features, and generates a historical drift memory sequence marked with start window, transition window and peak window.

[0050] The calibration strategy generation module generates a cross-cycle inherited calibration parameter template based on the historical drift memory sequence, and inputs the cross-cycle inherited calibration parameter template into the injection pump control logic to perform adaptive dynamic updates of the steady-state trigger threshold, response delay compensation amount, and rate homing curve, so as to achieve early pre-calibration response.

[0051] The beneficial effects of this invention are:

[0052] This invention, by introducing a dynamic sliding window mechanism and a multi-dimensional flow velocity feature extraction method, can adaptively identify the boundary transition segment from a stable state to an unstable state during the operation of an injection pump, and construct a boundary drift indication map in real time. This comprehensively reflects the response delay, amplitude fluctuation, and rate mutation characteristics of the injection pump in the critical operating range, providing a precise intervention reference for subsequent calibration strategies.

[0053] This invention proposes a cross-cycle inherited calibration parameter template generation and dynamic calling mechanism by constructing historical drift memory sequences and extracting multi-cycle consistency error patterns. This enables the classification modeling and behavioral feature extraction of different types of error patterns, allowing for real-time identification of matching patterns during operation. Based on the error characteristics, it automatically triggers target flow rate correction, feedback strategy switching, and slope homing factor injection, intervening in unstable states in advance and significantly improving the robustness and response speed of the injection pump.

[0054] This invention introduces a fusion threshold criterion and a sliding weighted update mechanism to perform template fusion correction on the new scoring trajectories continuously generated during the operation of the injection pump, thereby achieving continuous optimization and self-evolution of calibration parameters. It maintains the timeliness and adaptability of the calibration template under different operating cycles and conditions, solving the problems of response lag and compensation deviation caused by calibration template aging and static parameter fixation in traditional methods. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram of the calibration method flow according to an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the system functional modules according to an embodiment of the present invention. Detailed Implementation

[0058] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art may employ other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0059] like Figure 1 As shown, an automatic calibration method for an injection pump based on flow rate stability includes the following steps:

[0060] S1. Collect flow rate data during continuous operation of the syringe pump, and identify the boundary transition segment during the transition from a stable state to an unstable state based on the dynamic window mechanism. Construct a boundary drift indicator map, which reflects the distribution of response delay, amplitude fluctuation and rate inflection point of the syringe pump in the stable critical region within different operating cycles.

[0061] S2, based on the constructed boundary drift indicator map, combined with the comparison of multi-cycle drift behavior, identifies a set of error patterns that exhibit consistent shift characteristics, and marks the start window, transition window and peak window associated with each error pattern to generate a historical drift memory sequence.

[0062] S3 generates a cross-cycle inherited calibration parameter template based on the historical drift memory sequence. The cross-cycle inherited calibration parameter template is input into the injection pump control logic for adaptive adjustment. The definitions of steady-state trigger threshold, response delay compensation amount and rate homing curve in the cross-cycle inherited calibration parameter template are dynamically updated to realize the pre-calibration response of the injection pump before the instability boundary is induced.

[0063] S1 includes:

[0064] S11. During the continuous operation of the injection pump, the flow rate data sequence is collected in real time, and a sliding window group is constructed according to a preset time interval. The flow rate data sequence in each sliding window is statistically analyzed to extract the mean, range, and slope change rate to capture the initial signs of flow rate fluctuations. By setting the dynamic sliding window width, the sliding window is adaptively adjusted according to the frequency of flow rate changes to ensure that the transition process between stability and disturbance is effectively covered in different operating cycles.

[0065] S12. In the acquired sliding window sequence, analyze the slope abrupt change point, fluctuation amplitude increase region and response delay phenomenon between adjacent windows, identify the boundary transition segment representing the gradual transition from a stable state to a disturbed state, and construct a boundary drift indicator map to characterize the rate change pattern, amplitude expansion structure and delayed response amplitude of the injection pump near the instability point in multiple operating cycles.

[0066] S11 includes:

[0067] S111, collects the current flow rate value from the syringe pump flow monitoring unit in real time, forming a time series flow rate data. Among them, f t Let T be the instantaneous flow velocity value collected at time t, where t = 1, 2, 3, ..., T, and T is the number of sampling points corresponding to the total duration of the current data collection.

[0068] S112, the flow velocity data sequence is divided into multiple adjacent sliding windows, each with a size of w0 and a sliding step size of s, forming a window sequence, represented as:

[0069]

[0070] Among them, W k Let w be the k-th sliding window, N be the total number of sliding windows, and w k Let w be the current width of the k-th sliding window, initially set to w0, t k =(k-1)·s+1 is the starting point index of the k-th sliding window;

[0071] S113, for each sliding window W k Calculate characteristic indicators, including mean, range, and rate of change of slope, specifically including:

[0072] Mean:

[0073] Range (the difference between the maximum and minimum values): ρ k =max(W k )-min(W k );

[0074] Rate of change of slope (average of absolute values ​​of increments at any adjacent points within the window):

[0075] Where, μ k For sliding window W k The average flow velocity, ρ k For sliding window W k The range within the range measures the amplitude, θ k This represents the average absolute rate of change of the local slope within the window.

[0076] S114, Introducing the window volatility coefficient ξ k Perform a normalization judgment to determine whether the window needs to shrink or expand, based on ξ. k Dynamically adjusting the width of the next window is represented as:

[0077]

[0078] Wherein, λ is the slope influence factor, representing the weighting coefficients of fluctuation amplitude and slope in the total fluctuation;

[0079]

[0080] Where, δ w w is the window width increment for each adjustment. k+1 τ1 and τ2 are the adaptive width of the (k+1)th sliding window, respectively, and the upper and lower thresholds of the fluctuation coefficient are the upper and lower thresholds of the fluctuation coefficient.

[0081] The upper and lower thresholds for the volatility coefficient are set as follows:

[0082]

[0083] Among them, Percentile q The qth quantile, For the fluctuation coefficient samples under the preset sampling stage or historical multi-window, q1 and q2 are the quantiles used for the low threshold and high threshold, respectively. q1 = 25% represents the lower limit of low fluctuation, q2 = 75% represents the lower limit of high fluctuation, and M is the number of reference sliding windows.

[0084] S12 includes:

[0085] S121, for any two adjacent sliding windows W k With W k+1 The degree of abrupt change is measured by the average rate of change of the slope of each individual, and the slope abrupt change Δθ is defined as such. k , is represented as:

[0086]

[0087] in, Window W k With W k+1 The average rate of change of slope;

[0088] S122, defining the amplitude index of each window as the range of the sliding window, then the fluctuation increase is Δρ. k , is represented as:

[0089] Δρ k =ρ k+1 -ρ k ;

[0090] Where, ρ k+1 For sliding window W k+1 The range within;

[0091] S123, Construct the latency response index δ k It is used to measure the trailing effect that gradually changes within multiple windows after a disturbance signal enters, when δ k >τ δ When τ is such that a response delay segment exists, it is assumed that there is such a segment. δ The delay threshold is expressed as:

[0092]

[0093] Where H is the number of backward observation windows, Let be the average flow velocity within the (k+h)th window. The average flow velocity within the (k+h-1)th window;

[0094] Delayed discrimination threshold τ δ The statistical method is used to define it, and it is expressed as follows:

[0095] τ δ =μ δ +γ·σ δ ;

[0096] Where, μ δ Historical lag index δ k The average value, σ δ γ is the standard deviation of the historical lag index, and γ is the sensitivity factor;

[0097] S124, Define the boundary-aware scoring function B k , used to make a judgment window W based on multiple dimensions of features k Is it at the stable-to-perturbation boundary, when B k >τ B When, mark W k This is the boundary transition segment, where τ B The boundary scoring threshold is expressed as:

[0098] B k =ω1·Norm(Δθ) k )+ω2·Norm(Δρ k )+ω3·Norm(δ k );

[0099] Where Norm is the normalization function, and ω1, ω2, and ω3 are the corresponding weights;

[0100] Boundary scoring threshold τ B Represented as:

[0101] τ B =μ B +η·σ B ;

[0102] Where, μ B Score historical boundaries B k The mean, σ B η is the standard deviation of the historical boundary score, and η is the sensitivity parameter;

[0103] S125, mark all windows identified as boundary transition segments according to the time axis to form a graph, where the X-axis represents the time window number k, and the Y-axis represents the index B. k , Δθ k , Δρ k δ k The superimposed curves are highlighted, and the perturbation boundary window intervals and their perturbation types (slope type, amplitude type, delay type) are marked. The final output is a boundary drift indicator map. Represented as:

[0104]

[0105] Among them, type k The labels are for boundary perturbation types: slope for slope-type perturbation, amplitude for amplitude-type perturbation, and delay for delay-type perturbation.

[0106] S2 includes:

[0107] S21, Suppose the operation process is divided into C historical operation cycles, and extract the corresponding boundary score map in each operation cycle c. And construct trajectory matrix B cyc , is represented as:

[0108]

[0109] Among them, L c Let c be the total number of boundary windows in the c-th period, where c = 1, 2, ..., C. This is the score value of the i-th boundary window in the c-th period;

[0110] S22, To identify window points with consistent offset behavior, construct a set of candidate window alignment indices with consistent drift behavior across periods. Window points that appear in similar positions and have similar scoring patterns across multiple cycles are represented as follows:

[0111]

[0112] Where (c,j) is the j-th window in period c, δ t =2 is the time-position alignment tolerance threshold, δ b =0.05 is the tolerance threshold for score similarity, c ′ To compare the period numbering, j ′ To compare window indices in the comparison period, This is the window score value for the current period. To compare window score values ​​within the comparison period;

[0113] S23, using density clustering DBSCAN to align the window-aligned index set Clustering is performed on the window points in the data, and each cluster forms an error pattern unit ε. i For each error mode unit ε i Extract key windows, including the start window. (Window where the score first exceeds the threshold), transition window (Location with the highest rate of change in rating), peak window (The location where the score reaches a local maximum) is represented as:

[0114]

[0115] Where j1 and j2 are the corresponding window indices, These are the corresponding boundary scores, λ b The impact factor of score difference;

[0116] S24, structuring each error pattern into a historical drift memory sequence item. Represented as:

[0117]

[0118] in, This represents the average score value of the window corresponding to the error mode.

[0119] S3 includes:

[0120] S31. Based on the constructed historical drift memory sequence, the key position windows (start, transition, peak) of different perturbation type error modes are summarized, and their corresponding calibration behavior characteristics are statistically analyzed, including stable trigger time, average score offset, recovery rate, and cross-cycle inherited calibration parameter template is constructed.

[0121] S32, the constructed cross-cycle inherited calibration parameter template is input to the injection pump control logic unit, and the current boundary scoring trajectory is monitored in real time. If the window behavior is identified to match the error mode in the cross-cycle inherited calibration parameter template, the corresponding calibration compensation operation is triggered, including correcting the target flow rate setting, switching the feedback response strategy in advance, or injecting slope positivity factor, so as to complete the pre-adjustment calibration before the unstable state officially occurs.

[0122] S33 continuously receives new boundary identification results during the operation of the syringe pump and matches and merges them with the cross-cycle inherited calibration parameter template to dynamically adjust key calibration parameters, including trigger threshold, delay compensation amount and homing strategy.

[0123] S31 includes:

[0124] S311, For error modes τ of different disturbance types, statistically analyze their corresponding calibration behavior characteristics, including the stable trigger point. Average rating offset ΔB τ Recovery rate Represented as:

[0125]

[0126] in, To stabilize the trigger point distribution, ΔB represents the average position at which the steady state begins to shift in this type of error mode. τ N represents the average score offset, indicating the overall increase in scores within this type of error pattern, reflecting the required intensity of calibration. τ Let τ be the number of error modes of the disturbance type τ. Let be the score value of the peak window in the i-th mode. Rate the starting window. Let i be the index of the window where the score falls back to a stable value in the i-th error pattern. For window The rating;

[0127] S312 categorizes and summarizes the error mode calibration behavior characteristics corresponding to different disturbance types, and constructs a cross-cycle inherited calibration parameter structure template including stable trigger time, average score offset, and recovery rate. Represented as:

[0128]

[0129] S32 includes:

[0130] S321, During the continuous operation of the infusion pump, the control logic unit collects the boundary score values ​​of the most recent L sliding windows in real time, forming the current score trajectory S. cur , is represented as:

[0131] S cur =[β t-L+1 ,β t-L+2 ,...,β t ];

[0132] Where t is the current window number, L is the length of the scoring trajectory, and β t-L+1 β t-L+2 β t These are the boundary score values ​​for the (t-L+1), (t-L+2), and (t)th windows, respectively.

[0133] S322, from the cross-cycle inherited calibration parameter structure template T, extract the template reference scoring trajectory corresponding to each perturbation type τ, and calculate the matching degree k between the current scoring trajectory and the template reference scoring trajectory using weighted cosine similarity. τ When the disturbance type τ * satisfy If so, it is determined that the current operating state of the injection pump has exhibited a disturbance type τ. * Similar behavior to the corresponding error mode should trigger the next calibration response phase, where θ match The matching threshold is expressed as:

[0134]

[0135] Where, ω i β is the time-location weighting coefficient. t-L+i Let be the boundary score value at the i-th position of the current scoring trajectory. Let τ be the score value at the i-th position of the template reference score trajectory for the perturbation type τ;

[0136]

[0137] in, Let N be the score value at the l-th position of the template reference score trajectory for the disturbance type τ. τLet τ be the number of error pattern samples. Let be the boundary score value of the i-th scoring trajectory at the l-th position;

[0138] θ match =λ μ μ pos +(1-λ μ )·μ neg ;

[0139] Where, λ μ μ is the weighting coefficient. pos μ is the mean of the matching scores that triggered calibration in history. neg The average match score in history that did not trigger calibration;

[0140] S323, when the matching degree κ τ When the matching threshold is exceeded, the injection pump control logic triggers a calibration compensation operation based on the corresponding disturbance type. This includes correcting the target flow rate setting, switching the feedback response strategy in advance, or adjusting the injection slope homing factor. This enables pre-adjustment calibration and dynamic correction for unstable states, specifically including:

[0141] Correct target flow rate setting: based on the average score offset in error mode. The calculated correction value is expressed as:

[0142]

[0143] Among them, Q adj Q is the corrected target flow rate value. set γ is the target flow rate set for the current syringe pump. q Weighting factor for flow rate correction;

[0144] Early switching of feedback response strategy: Average delay time compensation value extracted from cross-cycle inherited calibration parameter structure template The action of giving early feedback is represented as:

[0145]

[0146] in, The average response delay compensation time under type τ, This refers to the original feedback execution time point;

[0147] Injection slope recovery factor: Obtain the scoring slope recovery factor from the cross-cycle inherited calibration parameter structure template. Used for controller output correction, denoted as:

[0148]

[0149] in, The positive factor for the slope of the score under type τ. γ represents the slope or output velocity term in the current control strategy. s These are the weight parameters.

[0150] S33 includes:

[0151] S331, after each sliding window scoring cycle, records the new scoring trajectory segment in real time, and extracts its scoring mean offset, scoring slope change value, and delayed response feature value to form candidate update samples, represented as:

[0152]

[0153]

[0154] in, This represents the deviation from the mean score. This represents the change in the slope of the rating. L is the characteristic value of the delayed response. s The length of the current rating window. Let j be the score value of the current boundary scoring trajectory at position j. Let be the average score of the template trajectory of category τ at the j-th position, where j is the position index within the local time window. This represents the time position corresponding to the point with the highest score in the current scoring trajectory. This refers to the time point when the current score first continuously exceeds the stable threshold.

[0155] S332, calculate the weighted cosine similarity between the new scoring trajectory segment and the current template reference trajectory, and determine whether it belongs to the same error type as the template. If the similarity k... update Above the conservative fusion threshold θ update Then it will be included in the update and fusion, as shown below:

[0156]

[0157] in, λ represents the representative score of the current scoring segment. s λ represents the weight of the scoring bias factor. ζ λ represents the weight of the factor influencing the score slope. d For the weights of the delayed response influence factors, μ γ σ represents the historical average of representative scores. γ The historical standard deviation of the representative score This is the adjustment coefficient;

[0158] S333 performs a sliding weighted fusion update on the average score offset, average delay time compensation value, and score slope recovery factor, as follows:

[0159]

[0160] in, These are the updated average rating offset, average delay time compensation value, and rating slope recovery factor, α. u To update the fusion coefficients.

[0161] like Figure 2 As shown, an automatic calibration system for an infusion pump based on flow rate stability is used to implement the aforementioned automatic calibration method for an infusion pump based on flow rate stability, and includes the following modules:

[0162] Flow rate acquisition module: Acquires flow rate data during continuous operation of the syringe pump, identifies boundary transition sections based on a dynamic window mechanism, and constructs a boundary drift indicator map that reflects the characteristics of stable critical response;

[0163] Error pattern recognition module: Based on the boundary drift indicator map, it compares the drift behavior of multiple cycles, extracts the consistency shift feature, and generates a historical drift memory sequence with start window, transition window and peak window markers;

[0164] Calibration strategy generation module: Generates cross-cycle inherited calibration parameter templates based on historical drift memory sequences, and inputs the cross-cycle inherited calibration parameter templates into the injection pump control logic to perform adaptive dynamic updates of steady-state trigger thresholds, response delay compensation amounts, and rate homing curves, so as to achieve advance pre-calibration response.

[0165] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0166] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An automatic calibration method for an injection pump based on flow rate stability, characterized in that, Includes the following steps: S1. Collect flow rate data during the continuous operation of the syringe pump, and identify the boundary transition segment during the transition from a stable state to an unstable state based on the dynamic window mechanism, and construct a boundary drift indicator map. The boundary drift indicator map reflects the distribution of response delay, amplitude fluctuation and rate inflection point of the syringe pump in the stable critical region within different operating cycles. S2, based on the constructed boundary drift indicator map, combined with the comparison of multi-cycle drift behavior, identifies a set of error patterns that exhibit consistent shift characteristics, and marks the start window, transition window and peak window associated with each error pattern to generate a historical drift memory sequence. S3. Generate a cross-cycle inherited calibration parameter template based on the historical drift memory sequence. Input the cross-cycle inherited calibration parameter template into the injection pump control logic for adaptive adjustment. Dynamically update the definitions of steady-state trigger threshold, response delay compensation amount and rate homing curve in the cross-cycle inherited calibration parameter template to realize the pre-calibration response of the injection pump before the instability boundary is induced.

2. The automatic calibration method for an injection pump based on flow rate stability according to claim 1, characterized in that, S1 includes: S11. During the continuous operation of the injection pump, the flow rate data sequence is collected in real time, and a sliding window group is constructed according to a preset time interval. The flow rate data sequence in each sliding window is statistically analyzed to extract the mean, range, and slope change rate in order to capture the initial signs of flow rate fluctuation. The sliding window is also dynamically adjusted according to the frequency of flow rate change by setting the dynamic sliding window width. S12. In the acquired sliding window sequence, analyze the slope abrupt change point, fluctuation amplitude increase region and response delay phenomenon between adjacent windows, identify the boundary transition segment representing the gradual transition from a stable state to a disturbed state, and construct a boundary drift indicator map to characterize the rate change pattern, amplitude expansion structure and delayed response amplitude of the injection pump near the instability point in multiple operating cycles.

3. The automatic calibration method for an injection pump based on flow rate stability according to claim 2, characterized in that, S11 includes: S111, collects the current flow rate value from the syringe pump flow monitoring unit in real time, forming a time series flow rate data. S112, the flow velocity data sequence is divided into multiple adjacent sliding windows, each with a size of w0 and a sliding step size of s, forming a window sequence, represented as: Among them, W k Let w be the k-th sliding window, N be the total number of sliding windows, and w k Let t be the current width of the k-th sliding window. k This is the index of the starting point of the k-th sliding window; S113, for each sliding window W k Calculate characteristic indicators, including mean, range, and rate of change of slope; S114, Introducing the window volatility coefficient ξ k Perform a normalization judgment to determine whether the window needs to shrink or expand, based on ξ. k Dynamically adjust the width of the next window; Where, δ w w is the window width increment for each adjustment. k+1 τ1 and τ2 are the adaptive width of the (k+1)th sliding window, and the upper and lower thresholds of the fluctuation coefficient are respectively.

4. The automatic calibration method for an injection pump based on flow rate stability according to claim 3, characterized in that, S12 includes: S121, for any two adjacent sliding windows W k With W k+1 The degree of abrupt change is measured by the average rate of change of the slope of each individual, and the slope abrupt change Δθ is defined as such. k ; S122, defining the amplitude index of each window as the range of the sliding window, then the fluctuation increase is Δρ. k ; S123, Construct the latency response index δ k It is used to measure the trailing effect that gradually changes within multiple windows after a disturbance signal enters, when δ k >τ δ When τ is such that a response delay segment exists, it is assumed that there is such a segment. δ The threshold for delayed discrimination; S124, Define the boundary-aware scoring function b k , used to make a judgment window W based on multiple dimensions of features k Is it at the stable-to-perturbation boundary, when B k >τ B When, mark W k This is the boundary transition segment, where τ B The boundary scoring threshold; S125, mark all windows identified as boundary transition segments according to the time axis to form a graph, where the X-axis represents the time window number k, and the Y-axis represents the index B. k , Δθ k , Δρ k δ k The superimposed curves are highlighted to indicate the perturbation boundary window range and its perturbation type, and the final output is a boundary drift indicator map.

5. The automatic calibration method for an injection pump based on flow rate stability according to claim 4, characterized in that, S2 includes: S21, Suppose the operation process is divided into C historical operation cycles, and extract the corresponding boundary score map in each operation cycle c. And construct trajectory matrix B cyc ; S22, Construct a set of window-aligned indexes with consistent drift behavior across cycles. Select window points that appear in similar positions and have similar scoring patterns from multiple cycles; S23, using density clustering DBSCAN to align the window-aligned index set Clustering is performed on the window points in the data, and each cluster forms an error pattern unit ε. i For each error mode unit ε i Extract key windows, including the start window. Transfer window Peak window S24, structuring each error pattern into a historical drift memory sequence item.

6. The automatic calibration method for an injection pump based on flow rate stability according to claim 5, characterized in that, S3 includes: S31. Based on the constructed historical drift memory sequence, the key position windows of different perturbation types error modes are summarized, and their corresponding calibration behavior characteristics are statistically analyzed, including stable trigger time, average score offset, recovery rate, and cross-cycle inherited calibration parameter template is constructed. S32, the constructed cross-cycle inherited calibration parameter template is input to the injection pump control logic unit, and the current boundary scoring trajectory is monitored in real time. If the window behavior is identified to match the error mode in the cross-cycle inherited calibration parameter template, the corresponding calibration compensation operation is triggered, including correcting the target flow rate setting, switching the feedback response strategy in advance, or injecting slope positivity factor, so as to complete the pre-adjustment calibration before the unstable state officially occurs. S33 continuously receives new boundary identification results during the operation of the syringe pump and matches and merges them with the cross-cycle inherited calibration parameter template to dynamically adjust key calibration parameters, including trigger threshold, delay compensation amount and homing strategy.

7. The automatic calibration method for an injection pump based on flow rate stability according to claim 6, characterized in that, S31 includes: S311, For error modes τ of different disturbance types, statistically analyze their corresponding calibration behavior characteristics, including the stable trigger point. Average rating offset ΔB τ Recovery rate S312 categorizes and summarizes the error mode calibration behavior characteristics corresponding to different disturbance types, and constructs a cross-cycle inherited calibration parameter structure template including stable trigger time, average score offset, and recovery rate.

8. The automatic calibration method for an injection pump based on flow rate stability according to claim 7, characterized in that, S32 includes: S321, During the continuous operation of the infusion pump, the control logic unit collects the boundary score values ​​of the most recent L sliding windows in real time, forming the current score trajectory S. cur ; S322, from the cross-cycle inherited calibration parameter structure template In the process, the template reference scoring trajectory corresponding to each type of perturbation τ is extracted, and the matching degree κ between the current scoring trajectory and the template reference scoring trajectory is calculated using weighted cosine similarity. τ When the disturbance type τ * satisfy If so, it is determined that the current operating state of the injection pump has exhibited a disturbance type τ. * Similar behavior to the corresponding error mode should trigger the next calibration response phase, where θ match The matching threshold; S323, when the matching degree κ τ When the matching threshold is exceeded, the injection pump control logic triggers calibration compensation operations according to the corresponding disturbance type, including correcting the target flow rate setting, switching the feedback response strategy in advance, or the injection slope positivity factor, to achieve advance pre-adjustment calibration and dynamic correction of unstable states.

9. The automatic calibration method for an injection pump based on flow rate stability according to claim 8, characterized in that, S33 includes: S331: After each sliding window scoring cycle is completed, a new scoring trajectory segment is recorded in real time, and its scoring mean offset, scoring slope change value and delayed response feature value are extracted to form candidate update samples. S332, calculate the weighted cosine similarity between the new scoring trajectory segment and the current template reference trajectory, and determine whether it belongs to the same error type as the template. If the similarity κ... update Above the conservative fusion threshold θ update Then it will be included in the update and integration; S333 performs a sliding weighted fusion update on the average score offset, average delay time compensation value, and score slope recovery factor.

10. An automatic calibration system for an infusion pump based on flow rate stability, used to implement the automatic calibration method for an infusion pump based on flow rate stability as described in any one of claims 1-9, characterized in that, Includes the following modules: Flow rate acquisition module: Acquires flow rate data during continuous operation of the syringe pump, identifies boundary transition sections based on a dynamic window mechanism, and constructs a boundary drift indicator map that reflects the characteristics of stable critical response; Error pattern recognition module: Based on the boundary drift indicator map, it compares the drift behavior over multiple cycles, extracts the consistency shift features, and generates a historical drift memory sequence marked with start window, transition window and peak window. The calibration strategy generation module generates a cross-cycle inherited calibration parameter template based on the historical drift memory sequence, and inputs the cross-cycle inherited calibration parameter template into the injection pump control logic to perform adaptive dynamic updates of the steady-state trigger threshold, response delay compensation amount, and rate homing curve, so as to achieve early pre-calibration response.

Images