A scalp pulsing dose closed loop control system and method

Abstract

This invention discloses a closed-loop control system and method for scalp pulsed drug delivery. A metering pump and pulse valve output micro-droplet pulses through a nozzle. The section from the nozzle outlet to the delivery head outlet is a straight-through measurement section without valves, filters, or replaceable connectors. A through-beam extinction sensor and an outlet-adjacent pressure sensor collect light intensity and gauge pressure sequences under the same hardware trigger. The controller calculates the single-pulse liquid mass M_i using the baseline-corrected extinction integral and the pressure impulse J_i using the outlet equivalent area and gauge pressure integral. Paired pulse records are formed only when both responses are within the numerical event window, the working distance is 15–20 mm, and the baseline is restored. A partitioned ledger accumulates the liquid mass S_M,z and the pressure impulse S_J,z respectively; the partition is locked only when both reach the target lower limit, and stops when either reaches the upper limit. The parameters for the next pulse are updated by two remaining quantities, the single-pulse error, and the two-dimensional response matrix. The above-mentioned quantities are measurable physical quantities of the device.

Description

Technical Field

[0001] This invention relates to the fields of microdroplet pulse delivery devices, optical and pressure measurement, and automatic control technology. Specifically, it relates to a system and method for performing paired measurements of the liquid mass and pressure impulse of the same discrete pulse within the direct measurement section of a scalp-operated head, and updating the parameters of the next pulse with partitioned dual ledgers and dual threshold states. Background Technology

[0002] CN116456860A discloses a method for determining the spatial conditions of the scalp or hair using proximity, contact, distance, or position sensors, and accordingly altering the output of liquid or atomized formulations. This type of spatial interlock can reduce erroneous output positions; however, even if the actuator has received a spray command or detected the actuating head approaching the scalp, it cannot directly prove the actual liquid mass and mechanical impulse of a discrete pulse leaving the output end.

[0003] CN103889488B discloses a method for estimating aerosol output rate and feeding back nebulizer power using optical signals; WO2020254313A1 discloses a method for determining instantaneous dose rate and accumulating dose using aerosol concentration and pressure or flow information. Therefore, detecting actual aerosol output and using it for feedback, as well as integrating over time to form a cumulative dose, are both methods known in the art.

[0004] WO2021022118A1 discloses the association of event identifiers, timestamps, and detection time windows with inhaler actions; EP3215219B1 discloses recording the processing status according to skin location and reducing or stopping processing after the target is reached. CN103399487B also discloses matrix decoupling, singularity switching, and saturation handling for multi-input multi-output systems. Therefore, event identifiers, partition ledgers, and two-dimensional control matrices cannot constitute inventive step independently of specific measurement paths, calculation relationships, and state transitions.

[0005] The existing solution still does not solve the following specific problems: under the condition that the scalp action head moves continuously and adjacent pulses may be close, how to ensure that the liquid quality and pressure impulse come from the same physical output event, and avoid inferring one physical quantity from the other; and how to make the two measured physical quantities enter the same area for completion judgment and next pulse limit respectively, so as to avoid prematurely judging the completion of the area or missing the impulse limit by accumulating a single liquid volume. Summary of the Invention

[0006] The purpose of this invention is to provide a closed-loop control system and method for scalp pulsed dosing, which specifically defines the measurement position after the nozzle, the conversion relationship between light intensity and pressure raw signals, the boundary of the same pulse event, the effective distance condition, the accumulation of dual physical quantities in the partition, and the dual variable control of the next pulse, thereby forming a complete closed-loop chain that can be verified by the original record.

[0007] To achieve the above objectives, the present invention provides a closed-loop control system 100, including a pulse application component 110, an output measurement component 120, an application head 130, a target coupling detection component 140, a pulse event processor 150, an application dose estimator 160, a zone dose ledger 170, a position / zone module 180, and a closed-loop controller 190. Each module is not an arbitrary device summarized by its functional results, but is defined by the specific structure, numerical fields, and calculation steps described below.

[0008] The pulse action assembly 110 uses a metering pump as a liquid actuator 111 and a gas storage chamber and a pulse valve as an impulse actuator 112. Liquid and gas flow through a nozzle in a common output path 113 to form discrete micro-droplet pulses. A straight-through measurement section is formed between the nozzle outlet and the actuating head outlet 131. No valves, filters, replaceable connectors, or atomizing components are installed in this section. Therefore, there are no variable flow path elements between the measurement section and the actuating head outlet that would change the output again. The straight-through measurement section is defined by observable component positions and exclusion structures.

[0009] The liquid quality detection channel 121 of the output measurement component 120 is a through-beam extinction sensor that spans the straight-through measurement section. The dark level I_d is obtained when the transmitter is off; the light intensity baseline I_0,i of the i-th pulse is the average value within the baseline window W_B,i before triggering. The normalized extinction is integrated over the liquid volume response window W_M,i with the sampling interval Δt to obtain X_i=Σ[(I_0,i-I_i,k) / (I_0,i-I_d)]Δt, and then the single-pulse liquid quality is calculated according to M_i=a_0+a_1X_i. a_0 and a_1 are obtained by least-squares fitting of the reference mass collected by weighing and X_i, and the applicable liquid, temperature, nozzle type, upper and lower limits of X_i, and upper limit of residual and coefficient are saved together.

[0010] The pressure impulse detection channel 122 employs an outlet-adjacent pressure sensor with its pressure tap flush with the inner wall of the straight-through measurement section. Using the average pressure within W_B,i as p_0,i and the effective flow cross-sectional area at the pressure tap as A_eff, the pressure impulse of the i-th pulse is calculated according to J_i=A_effΣ(p_i,k-p_0,i)Δt, accumulating only the positive gauge pressure sampling term within W_J,i. The target coupling detection component 140 employs a time-of-flight ranging sensor 141 with its measuring and output axes fixedly calibrated; the effective distance sampling values ​​within the distance confirmation window W_C,i must all be 15–20 mm to meet the spatial action conditions.

[0011] The pulse event processor 150 generates the pulse identification i and the trigger time t_i with the same hardware trigger line, and timestamps the I_i,k, p_i,k and d_i,k with the common sampling clock. The event window is not an abstract "corresponding window", but is explicitly defined by the sampling interval Δt and the integer boundaries N_B, N_M1, N_M2, N_J1, N_J2, N_C1, N_C2, etc. The integer boundaries are determined by the response of the signal crossing the baseline mean plus / minus three times the standard deviation in no less than 30 calibration pulses and are written into the calibration record.

[0012] The valid decision module 153 sets V_i=1 only when both W_M,i and W_J,i fall within the i-th event boundary and do not overlap with the adjacent response window; the distance within W_C,i meets 15-20 mm and the minimum number of valid samples; I_0,i-I_d is greater than the minimum dynamic range; M_i and J_i are within the upper and lower limits of the respective calibration record and single pulse; the light intensity and pressure after the pulse re-enter the W_B,i mean plus / minus three times the standard deviation range within the pre-stored recovery time. V_i=0 when any condition is not met, and a distinguishable invalid reason code is saved.

[0013] The position / partition module 180 determines the unique target region z before triggering by the pre-set scanning trajectory and mechanical encoder. The partition dose account 170 saves the complete paired pulse record (M_i, J_i) with the pulse identification i and the target region z as the combined index; only for the records with V_i=1 and not repeated, S_M,z=ΣV_iM_i and S_J,z=ΣV_iJ_i are calculated respectively. The "paired pulse record" referred to in this paper refers to M_i and J_i from the same hardware trigger identification, passing through the time window and distance valid decision at the same time, not derived from one type of sensing quantity to another physical quantity.

[0014] Each target region stores L_M,z, U_M,z, L_J,z and U_J,z respectively. Only when S_M,z≥L_M,z and S_J,z≥L_J,z, the region is completed and locked; S_M,z≥U_M,z or S_J,z≥U_J,z immediately prohibits further output. When not locked, the closed-loop controller 190 calculates the single pulse error limited by U_M,z-S_M,z and U_J,z-S_J,z respectively. A two-dimensional response data set is established by combining at least 5x5 metering pump strokes and impulse parameters, no less than 20 valid pulses for each combination, and the local finite difference or least squares method is used to obtain [ΔM, ΔJ]^T=G[Δu_M, Δu_J]^T. Only when the current input is within the calibration input convex hull and G meets the pre-stored minimum singular value and condition number criterion, the inverse is calculated, otherwise the nearest input combination in the data set that does not cause any physical quantity to exceed the limit is selected.

[0015] The application also provides a corresponding device control method, including loading three types of calibration records and double target ranges, determining areas and distances, assigning hardware trigger identifiers, collecting baseline and discrete pulse raw signals, calculating M_i and J_i respectively according to specific formulas, generating V_i, only writing effective paired records into a double ledger, executing double lower limit lock zones and any upper limit stop, and updating next pulse parameters using two remaining amounts and a two-dimensional response matrix. The method controls device output physical quantities and internal states, and does not take disease diagnosis or human efficacy as feedback input.

[0016] The verifiable effects of the application are that the straight-through measurement section excludes variable flow path elements after the measurement section, M_i and J_i obtained by weighing calibration and pressure integration can be compared with external mass reference and force impulse reference respectively, hardware trigger identifiers and numerical windows enable whether two records belong to the same pulse to be reviewed by original time stamps, double ledgers and double thresholds enable areas to be completed by simultaneously satisfying two physical quantities, and avoid missing J_i by only containing M_i, and the two-dimensional response matrix is only used within the measured input domain and reversible criteria, so that the source, limiting and degradation path of each correction can be traced back. Each effect can be quantified by single pulse mass error, pressure impulse error, pulse correct pairing rate, number of times of early lock zone and number of times of upper limit triggering. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a structural schematic diagram of the scalp pulse action dose closed-loop control system of the application.

[0018] Figure 2 It is a layout schematic diagram of the nozzle, straight-through measurement section, opposite shooting extinction sensor, outlet adjacent pressure sensor and time-of-flight ranging sensor of the application.

[0019] Figure 3 It is a sampling point boundary schematic diagram of the baseline window, liquid volume response window, impulse response window and distance confirmation window of the single pulse of the application.

[0020] Figure 4 It is a determination schematic diagram of the liquid mass M_i, pressure impulse J_i and effective identifier V_i forming paired pulse records of the application.

[0021] Figure 5 It is a scanning track, partition double ledger and double threshold lock zone schematic diagram of the scalp target area of the application.

[0022] Figure 6 It is a two-dimensional response matrix control schematic diagram of the liquid mass feedback channel and pressure impulse feedback channel of the application.

[0023] Figure 7 It is a flow chart of the scalp pulse action dose closed-loop control method of the application.

[0024] Reference signs

[0025] 100—Closed-loop control system; 110—Pulse action component; 111—Liquid volume actuator; 112—Impulse actuator; 113—Common output path; 120—Output measurement component; 121—Through-beam extinction liquid quality detection channel; 122—Outlet proximity pressure impulse detection channel; 130—Actuating head; 131—Actuating head outlet; 132—Solid dispensing component; 133—Nozzle and replaceable wet end; 140—Target coupling detection component; 141—Time-of-flight ranging sensor; 142—Distance Valid determination unit; 143—Baseline recovery determination unit; 150—Pulse event processor; 151—Time base / synchronization module; 152—Event window generation module; 153—Valid determination module; 160—Action dose estimator; 170—Zone dose ledger; 180—Location / zone module; 190—Closed-loop controller; 191—Liquid mass controller; 192—Pressure impulse controller; 193—Cross-coupling compensator; 194—Limiting and degradation module; 200—Target scalp area. Detailed Implementation

[0026] The following describes a feasible implementation scheme employing a metering pump, a gas storage chamber, a pulse gas valve, a through-beam extinction sensor, an outlet proximity pressure sensor, and a time-of-flight ranging sensor, with reference to the accompanying drawings. To avoid exceeding the scope of protection of the disclosed content, this embodiment does not list sensors with significant differences in scattering imaging, flow velocity, acoustics, strain, or other mechanisms in the same conversion channel; when replacing sensors, the corresponding structure, original signal, and calibration relationship should be disclosed separately.

[0027] like Figure 1 As shown, the closed-loop control system 100 includes a pulse action component 110, an output measurement component 120, an action head 130, a target coupling detection component 140, a pulse event processor 150, an action dose estimator 160, a zone dose ledger 170, a position / zone module 180, and a closed-loop controller 190. The closed-loop controller 190 includes a liquid mass controller 191, a pressure impulse controller 192, a cross-coupling compensator 193, and a limiting and degradation module 194.

[0028] like Figure 2 As shown, the metering pump 111 supplies liquid to the atomizing chamber with a stroke u_M; the gas storage chamber and the pulse gas valve 112 provide airflow with gas storage pressure or valve opening time u_J, and the two form discrete micro-droplet pulses through the nozzle 133. The nozzle 133 is fixed inside the actuating head 130, and the section from the nozzle outlet to the actuating head outlet 131 is a straight-through measuring section with a constant inner diameter; there are no valves, filters, replaceable connectors, branch lines, or secondary atomizing components in this section. Solid distributors 132 with hairpins are provided on both sides of the actuating head outlet 131. The solid distributors 132 do not carry liquid or gas.

[0029] The emission and receiving ends of the transmission extinction sensor 121 are placed on both sides of the straight-through measurement section, with the optical axis crossing the center of the channel. The pressure sensor 122 has its pressure tapping hole located downstream of the optical axis and flush with the inner wall to reduce the hysteresis and liquid accumulation in the pressure tapping cavity. The time-of-flight distance sensor 141 is fixed on the action head 130, and the translation of its measurement axis relative to the output axis and the angular offset are recorded in the assembly calibration; the controller converts the sensor readings into the axial distance from the action head outlet end face to the scalp surface accordingly.

[0030] The calibration of the liquid mass channel is performed in sequence for the dark level, baseline, and mass calibration. Close the emission end to collect not less than N_D samples to get I_d; in the absence of pulses and when the straight-through measurement section is clean, collect not less than N_B samples within W_B to get I_0. Select not less than 5 pump strokes, each outputting not less than 20 mutually separated pulses, collect with pre-weighed collection sheets or filter membranes, and divide the mass increment after collection by the number of pulses to get M_ref; calculate X_i for each pulse, and perform least squares fitting with M_ref = a_0 + a_1X_i. The calibration file also records the liquid batch, temperature range, nozzle model, a_0, a_1, X_i range, maximum absolute residual, root mean square residual, and original weighing record. Do not extrapolate outside the X_i range during operation.

[0031] The pressure sensor 122 uses a calibrated pressure source to check the zero point and sensitivity, and the sampling frequency is at least ten times the highest effective frequency of the pressure pulse to be measured. Take the effective flow area at the pressure tapping hole as A_eff, and the average pressure within W_B,i as p_0,i. Discretely integrate the positive gauge pressure within W_J,i: J_i = A_effΣ(p_i,k - p_0,i)Δt. A calibrated single-axis force sensor or impulse target can be placed in front of the action head outlet to compare the force-time integral J_ref measured by it with J_i, and the resulting deviation is used for verification but not to replace the operational calculation of the pressure channel.

[0032] In this embodiment, the time-of-flight distance sensor 141 only bears the output permission and event validity determination. All valid distance sampling values within W_C,i should be 15-20 mm, and the effective sample number and range should also meet the minimum sample number and upper limit of the range in the calibration file, respectively. The median particle size Dv50 of the droplet volume can be in the range of 5-30 microns as a consistency check value for consumables and nozzles, but Dv50 does not participate in the conversion of M_i or J_i, nor does it serve as an index of human absorption or therapeutic effect.

[0033] As Figure 3As shown, the hardware trigger line generates a trigger edge at t_i, and simultaneously latches the pulse identifier i. W_B,i = [t_i - N_BΔt, t_i - Δt], W_M,i = [t_i + N_M1Δt, t_i + N_M2Δt], W_J,i = [t_i + N_J1Δt, t_i + N_J2Δt], W_C,i = [t_i - N_C1Δt, t_i + N_C2Δt]. The calibration outputs no less than 30 non-overlapping pulses each other, the point where the signal first exceeds the baseline mean three times the standard deviation as the response starting point, the first point where the subsequent N_R samples return to the three times standard deviation range as the response end point, and the corresponding N value is written in the minimum integer interval covering all calibration pulses.

[0034] The effective determination module 153 checks item by item: whether I_0,i - I_d is greater than the minimum dynamic range; whether W_M,i and W_J,i belong to the i-th trigger and do not overlap with adjacent windows; whether M_i and J_i are respectively within the calibration domain and the upper and lower limits of the single pulse; whether the distances of W_C,i are all within 15-20 mm; whether the pulse light intensity and pressure return to the baseline mean plus or minus three times the standard deviation within the recovery time limit. Only all pass V_i = 1. Each item that does not pass has an independent reason code, and numerical comparison cannot be replaced by relative expressions such as "basically consistent", "close to the target" or "within the safety envelope".

[0035] As shown in Figure 4 After the action dose estimator 160 outputs M_i and J_i, the pulse event processor 150 combines them with the same pulse identifier, trigger time, region identifier, distance minimum and maximum, actuator input, calibration version, and V_i into a non-repetitive paired pulse record. The record with V_i = 0 retains the original light intensity sequence, pressure sequence, distance sequence, and reason code, but is not used as a normal control error sample and does not enter the region accumulation.

[0036] This embodiment does not fit continuous conversion coefficients by distance, inclination, target type, and shielding degree, nor does it combine liquid mass and pressure impulse into a single accumulation. Distance is only formed as a Boolean effective condition in a clear interval of 15-20 mm; liquid mass and pressure impulse are saved and accumulated as two physical quantities with different dimensions and independent measurements, thereby avoiding compressing both into a single scalar when there is no common calibration basis.

[0037] As shown in Figure 5 The position / partition module 180 drives the action head with a preset scanning trajectory, and the mechanical encoder feeds back the current position. The region boundary table maps the encoder coordinates to a unique target region z in Z1 to Zn; the region identifier is frozen before the trigger edge is generated, and is not rewritten according to the subsequent position of the action head after the trigger ends. If the coordinates cannot be uniquely attributed within the region boundary tolerance band, this trigger is prohibited.

[0038] The partitioned dose account 170 performs S_M,z =∑V_iM_i and S_J,z =∑V_iJ_i for each zone separately. Each zone stores L_M,z, U_M,z, L_J,z and U_J,z; a zone is locked only if S_M,z ≥ L_M,z and S_J,z ≥ L_J,z, and any cumulative value reaches the corresponding upper limit; the system stops output and alarms. When a zone is not locked, R_M,z = U_M,z - S_M,z and R_J,z = U_J,z - S_J,z limit the maximum allowed mass and impulse of the next pulse respectively, not replaced by the residual amount of either.

[0039] As shown in Figure 6 The two-dimensional response dataset is established by not less than 5x5 input combinations. u_M takes the metering pump stroke or the liquid valve opening time, and u_J takes the gas storage pressure or the pulse gas valve opening time; not less than 20 pulses of V_i = 1 are collected for each combination to calculate the average value and covariance of M_i and J_i. Within the current input neighborhood, G is obtained by finite difference or least squares. The calibration file saves the input combination, output average, sample number, residual covariance, G, minimum singular value σ_min(G), condition number κ_2(G) and applicable input convex hull.

[0040] When controlling, the single-pulse target mass is first limited to not more than R_M,z, and the single-pulse target impulse is limited to not more than R_J,z, to obtain errors e_M and e_J. Only when the current input is located within the calibration convex hull, σ_min(G) is not less than the numerical field σ_min,lim, and κ_2(G) is not greater than the numerical field κ_max, calculate [Δu_M, Δu_J]^T = G^(-1)[e_M, e_J]^T. The correction amount is then limited by the actuator physical range, single-step change rate and two-zone residual amount; when the inverse condition is not met, select the nearest input combination from the measured data set that will not cause any cumulative value to exceed the upper limit, and find none when stopping output.

[0041] For pulses of V_i = 0, the system executes waiting for baseline recovery, repositioning, cleaning the straight-through measurement section, pressure relief or stopping according to the reason code. Only when the distance from re-entry is 15-20 mm, the zone is not locked, I_0-I_d recovers the dynamic range, the pressure baseline recovers and the number of consecutive invalidations is less than the upper limit, retry with a new pulse identifier; the original invalid record cannot be overwritten by a new record. The upper limit of the number of consecutive invalidations is written as an integer to the current calibration file.

[0042] As shown in Figure 7As shown, step S101 loads the mass, impulse, window, and control calibration records, as well as the dual target ranges for each region; S102 determines the region based on the encoder and verifies the 15-20 mm distance; S103 generates a pulse identifier and hardware trigger edge and acquires W_B; S104 outputs a discrete pulse and acquires three raw sequences; S105 calculates M_i and J_i according to the explicit formula and determines V_i; S106 writes the valid paired records into S_M,z and S_J,z respectively; S107 executes dual lower limit locking and stops any upper limit; S108 calculates the correction amount of G or selects a conservative input combination when the region is not locked and the limit is not exceeded; S109 records the anomaly and handles it according to the cause code.

[0043] Each replaceable liquid container and nozzle combination has a unique calibration version. The quality calibration record must include at least the liquid batch, temperature range, nozzle type, I_d, lower limit of dynamic range, a_0, a_1, upper and lower limits of X_i, and residual; the impulse calibration record must include at least the pressure sensor number, zero point, sensitivity, A_eff, Δt, upper and lower limits of J_i, and recovery time; the window record must include at least all N values; the control record must include at least 25 input combinations, the number of samples for each combination, the output mean, G, input convex hull, σ_min, lim, and κ_max. Output is disabled if any required numerical field is missing or the version does not match.

[0044] Repeatable verification uses the same data caliber as the operation records. Quality verification collects at least 20 pulses over at least 5 pump strokes, comparing the absolute error, relative error, and root mean square error of M_i with the weighing reference M_ref. Impulse verification compares J_i with the integral J_ref of the calibrated force sensor under the same input combination. Event attribution verification changes the pulse interval, counting the number of rejected overlapping windows and correctly matched known pulse identifiers. Zonal control verification compares S_M,z, S_J,z, the number of early zonal locking times, and the number of upper limit triggers under four states: liquid volume zonal locking only, impulse zonal locking only, matrix compensation not enabled, and the dual-ledger closed-loop of this scheme, on an artificial scalp model. Each state maintains consistency in liquid, nozzle, working distance, area trajectory, and target range, and retains the original pulse sequence, weighing value, force curve, calibration coefficients, and calculation process to verify the effects of quality conversion, impulse conversion, same-pulse binding, dual-threshold zonal locking, and cross-coupling compensation.

Claims

1. A closed-loop control system for scalp pulsed drug delivery, characterized in that, The system includes a pulse action assembly, an action head, an output measurement assembly, a time-of-flight ranging sensor, a pulse event processor, a position / zoning module, a zoned dose ledger, and a closed-loop controller. The pulse action assembly includes a metering pump as a liquid volume actuator and a gas storage chamber and a pulse valve as an impulse actuator. The liquid supplied by the metering pump and the gas flow output by the pulse valve form discrete pulses containing liquid droplets through a nozzle in a common output path. The action head has a straight-through measurement section between the nozzle outlet and the action head outlet. No valves, filters, replaceable connectors, or atomizing components are installed in the straight-through measurement section. The output measurement assembly includes components that cross the straight-through measurement section. The system includes a through-beam extinction sensor and a pressure sensor positioned near the outlet of the actuating head, with the pressure-sensing surface flush with the inner wall of the through-beam measurement section. The time-of-flight ranging sensor has a pre-fixed spatial relationship with the output axis of the actuating head outlet, used to measure the distance from the actuating head outlet to the target scalp surface pulse by pulse. The pulse event processor assigns a pulse identifier i and a trigger time t_i to each discrete pulse using the same hardware trigger signal, and collects the light intensity sampling value I_i,k of the through-beam extinction sensor, the pressure sampling value p_i,k of the pressure sensor, and the distance measurement value of the time-of-flight ranging sensor using a common sampling clock. The closed-loop controller stores the dark level I_d, light intensity baseline I_0,i, pressure baseline p_0,i, sampling interval Δt, outlet equivalent area A_eff, mass calibration coefficients a_0 and a_1, and baseline windows W_B,i, liquid volume response windows W_M,i, impulse response windows W_J,i, and distance confirmation windows W_C,i defined by the boundaries of integer sampling points. It calculates the liquid mass of the i-th pulse according to X_i=Σ_(k∈W_M,i)[(I_0,i-I_i,k) / (I_0,i-I_d)]Δt and M_i=a_0+a_1X_i, and calculates the liquid mass according to J_i=A_eff. The pressure impulse of the i-th pulse is calculated using Σ_(k∈W_J,i)(p_i,k-p_0,i)Δt. The closed-loop controller sets the valid flag V_i to 1 only when W_M,i and W_J,i are both within the event boundary of the i-th pulse and do not overlap with the response windows of adjacent pulses, all effective distance samples within W_C,i are 15-20 mm, M_i and J_i are within the pre-stored single-pulse value range, and the light intensity and pressure after the pulse have recovered to the baseline range obtained by W_B,i. Otherwise, V_i is set to 0. The position / partition module binds the pulse identifier i to a target area z before triggering. The partitioned dose ledger only accumulates paired pulse records (M_i, J_i) where V_i=1 and neither the pulse identifier nor the target region is repeated, to obtain the cumulative liquid mass value S_M,z and the cumulative pressure impulse value S_J,z, and calculates the remaining amount of the region relative to the target liquid mass range and the target pressure impulse range, respectively. The closed-loop controller outputs a zone lock signal only when S_M,z reaches the lower limit of the liquid mass target and S_J,z reaches the lower limit of the pressure impulse target. When S_M,z reaches the upper limit of the liquid mass target or S_J,z reaches the upper limit of the pressure impulse target, it prohibits further output and sets different metering pump stroke parameters and impulse. The M_i and J_i measured simultaneously under the actuator parameter combination are established as a two-dimensional response dataset. When the current parameter combination is located within the input convex hull of the two-dimensional response dataset and the minimum singular value and condition number of the local 2×2 response matrix G satisfy the pre-stored numerical criteria, the inverse matrix of G is applied to the liquid quality error and pressure impulse error limited by the two regional residual quantities to update the metering pump stroke parameters and gas storage pressure or pulse valve opening time parameters of the next pulse. When the numerical criteria are not satisfied, the nearest parameter combination in the two-dimensional response dataset that does not cause S_M,z or S_J,z to exceed the corresponding upper limit is selected.

2. The system according to claim 1, characterized in that, The straight-through measuring section is a circular or rectangular channel with a constant inner diameter. Its upstream end is fixedly connected to the nozzle outlet, and its downstream end constitutes the outlet of the action head. The emission optical axis of the through-beam extinction sensor intersects with the center line of the straight-through measuring section. The pressure tap of the pressure sensor is located between the emission optical axis and the outlet of the action head, and there is no branching path from the cross section where the emission optical axis is located to the outlet of the action head.

3. The system according to claim 1, characterized in that, The mass calibration coefficients a_0 and a_1 are obtained as follows: Under no less than 5 metering pump stroke settings, each setting outputs no less than 20 mutually separate pulses. The single-pulse reference mass M_ref is obtained by dividing the weighing collection value by the corresponding number of pulses. X_i of each pulse is calculated, and the least squares method is used to fit M_ref=a_0+a_1X_i. The minimum and maximum values ​​of liquid type, temperature range, nozzle type, a_0, a_1, upper limit of fitting residual, and X_i are saved as the same mass calibration record.

4. The system according to claim 1, characterized in that, The equivalent area A_eff at the outlet is determined by the effective flow cross-sectional area of ​​the straight-through measurement section at the pressure sensor tapping hole; the pressure baseline p_0,i is the arithmetic mean of the pressure sampling values ​​within W_B,i; the pressure impulse J_i is accumulated only for sampling terms where p_i,k-p_0,i is positive and belongs to W_J,i; the zero point, range, sampling frequency, and A_eff of the pressure sensor are all saved as numerical fields in the impulse calibration record.

5. The system according to claim 1, characterized in that, W_B,i=[t_i-N_BΔt,t_i-Δt],W_M,i=[t_i+N_M1Δt,t_i+N_M2Δt],W_J,i=[t_i+N_J1Δt,t_i+N_J2Δt],W_C,i=[t_i-N_C1Δt,t_i+N_C2Δt];where N_B, N_M1, N_M2, N_J1, N_J2, N_C1 and N_C2 are non-negative integers read from the calibration record by the closed-loop controller. N_M1, N_M2, N_J1 and N_J2 are determined by the first sampling point in at least 30 independent calibration pulses where the corresponding signal exceeds three standard deviations of the baseline mean before triggering and the first sampling point that recovers to within three standard deviations.

6. The system according to claim 1, characterized in that, The time-of-flight ranging sensor acquires distance sampling values ​​within W_C,i at a sampling frequency not less than ten times the pulse repetition frequency; when the effective distance sampling value is less than the pre-stored minimum number of samples, any effective distance sampling value is less than 15 mm or greater than 20 mm, or the range of the distance sampling values ​​is greater than the pre-stored upper limit of the range, V_i is set to 0.

7. The system according to claim 1, characterized in that, The position / partition module includes a motion mechanism that drives the actuator head to move according to a preset scanning trajectory and a mechanical encoder that detects the position of the motion mechanism; before pulse triggering, the target region z is obtained by querying a pre-stored region boundary table based on the position of the mechanical encoder, and triggering is prohibited when no unique target region is found.

8. The system according to claim 1, characterized in that, Each pair of pulse records includes at least the pulse identifier i, trigger time t_i, target area z, minimum and maximum distances within V_i, X_i, M_i, J_i, and W_C,i, metering pump stroke, gas storage pressure, pulse valve opening time, calibration record version, and invalid reason code; records with V_i=0 are only saved without adding S_M,z and S_J,z.

9. The system according to claim 1, characterized in that, Each target region z stores the lower limit of liquid mass target L_M,z, the upper limit of liquid mass U_M,z, the lower limit of pressure impulse target L_J,z, and the upper limit of pressure impulse U_J,z. When S_M,z≥L_M,z and S_J,z≥L_J,z, the target region z is locked. When S_M,z≥U_M,z or S_J,z≥U_J,z, the output stops and an over-limit reason code is generated. When not locked, the allowed liquid mass of the next pulse is no greater than U_M,z-S_M,z, and the allowed pressure impulse of the next pulse is no greater than U_J,z-S_J,z.

10. The system according to claim 1, characterized in that, The closed-loop controller outputs no less than 20 effective pulses on at least 5×5 input combinations consisting of the metering pump stroke parameter u_M and the impulse parameter u_J. A two-dimensional response dataset is established using the average values ​​of M_i and J_i of each combination. Within the neighborhood of the current input combination, a 2×2 response matrix G satisfying [ΔM, ΔJ]^T=G[Δu_M, Δu_J]^T is determined using finite difference or least squares method. G, the input combination boundary, and the residual covariance are stored in the same control calibration record.

11. The system according to claim 10, characterized in that, The control calibration record also includes a minimum singular value lower limit σ_min,lim and a condition number upper limit κ_max; only when the current input is located within the convex hull of the two-dimensional response dataset, the minimum singular value of G is not less than σ_min,lim and the L2 condition number is not greater than κ_max, the correction is calculated according to [Δu_M, Δu_J]^T=G^(-1)[e_M, e_J]^T, where e_M is the difference between the single-pulse target mass limited by the remaining liquid mass in the region and M_i, and e_J is the difference between the single-pulse target impulse limited by the remaining pressure impulse in the region and J_i.

12. The system according to claim 11, characterized in that, The correction amount is subject to the metering pump stroke range, gas storage pressure range, pulse valve opening time range, single-step change rate, and U_M,z-S_M,z and U_J,z-S_J,z limits, respectively. When the matrix inversion condition is not met or any actuator reaches the limit boundary, the integral update is stopped and the most recent input combination of the two-dimensional response data set and the two regions' remaining amounts are not exceeded is adopted.

13. The system according to claim 1, characterized in that, When I_0,i-I_d is not greater than the pre-stored minimum dynamic range, the light intensity or pressure after the pulse does not return to the mean of W_B,i plus or minus three standard deviations within the pre-stored recovery time, adjacent pulse response windows overlap, or M_i or J_i exceeds the numerical range of their respective calibration records, V_i is set to 0 and cause codes for optical path abnormality, pressure baseline abnormality, event conflict, quality out of bounds, or impulse out of bounds are generated respectively.

14. The system according to claim 13, characterized in that, For pulses with V_i=0, the closed-loop controller maintains the original paired pulse records and does not overwrite their pulse identifiers; only when the distance is again satisfied by 15-20 mm, the target area is not locked, both baselines are restored, and the number of consecutive invalid attempts is less than the pre-stored upper limit, a new pulse identifier is assigned for a retry; when the number of consecutive invalid attempts reaches the pre-stored upper limit, the metering pump and pulse gas valve are shut down and a cleaning or maintenance prompt is output.

15. A closed-loop control method for scalp pulsed drug delivery executed by the system according to any one of claims 1 to 14, characterized in that, include: Read the mass calibration record, impulse calibration record, event window boundary, two-dimensional response dataset, and liquid mass target range and pressure impulse target range of each target area corresponding to the current liquid, nozzle, and actuator head; determine the unique target area z before triggering based on the mechanical encoder position, and assign a pulse identifier i and hardware trigger time t_i to the pulse to be output when the distance from the actuator head outlet to the target scalp surface measured by the time-of-flight ranging sensor is 15-20 mm; Within W_B,i, obtain the light intensity baseline I_0,i and pressure baseline p_0,i, drive the metering pump, gas storage chamber, and pulse valve to output a discrete pulse containing liquid droplets, and obtain I_i,k, p_i,k, and d_i,k using a common sampling clock; calculate the liquid mass according to X_i=Σ_(k∈W_M,i)[(I_0,i-I_i,k) / (I_0,i-I_d)]Δt and M_i=a_0+a_1X_i, and calculate the pressure impulse according to J_i=A_effΣ_(k∈W_J,i)(p_i,k-p_0,i)Δt; within W_M,i and W_J,i, both belong to the i-th pulse and do not overlap with adjacent response windows, the distance from the sampled value within W_C,i satisfies 15-20 mm, and M_i and J_i are both within their respective single pulse value ranges. When both baselines are restored, V_i is set to 1 and a pair of pulse records (M_i, J_i) bound to the target region z are formed; otherwise, V_i is set to 0 and the reason for invalidity is recorded. Only the pair of pulse records with V_i=1 and no repetition are accumulated to obtain S_M,z and S_J,z. When S_M,z reaches the lower limit of the liquid quality target and S_J,z reaches the lower limit of the pressure impulse target, the target region z is locked. When S_M,z reaches the upper limit of the liquid quality target or S_J,z reaches the upper limit of the pressure impulse target, the output is prohibited. When the target region z is not locked and has not exceeded the limit, the metering pump stroke parameters and gas storage pressure or pulse valve opening time parameters of the next pulse are updated after decoupling and limiting by the two-dimensional response matrix G based on the remaining quantities of the two regions and the errors of M_i and J_i relative to the corresponding single pulse targets.

16. The method according to claim 15, characterized in that, Establishing a quality calibration record includes: outputting no less than 20 pulses for no less than 5 metering pump stroke settings, obtaining the reference mass of each set single pulse using the weighing collection method, calculating X_i and obtaining a_0 and a_1 using the least squares method; establishing an impulse calibration record includes: verifying the zero point and sensitivity of the pressure sensor using a calibrated pressure source, measuring the effective flow cross-sectional area of ​​the straight-through measurement section at the pressure tap as A_eff, and recording the sampling interval Δt.

17. The method according to claim 15, characterized in that, Establishing the event window boundary includes: outputting no less than 30 calibration pulses with non-overlapping adjacent pulse responses; taking the sampling point where the corresponding signal first exceeds three times the standard deviation of the baseline mean before triggering as the response start point in each calibration pulse; taking the first sampling point within the range of the mean plus or minus three times the standard deviation after the first consecutive sampling point of no less than N_R as the response end point; and writing the minimum integer sampling point interval covering the entire calibration pulse response start point and response end point into the boundary fields of W_M and W_J, respectively.

18. The method according to claim 15, characterized in that, The partitioned dose ledger performs S_M,z=ΣV_iM_i and S_J,z=ΣV_iJ_i on each target region, with the summation range being records bound to the target region z and with unique pulse identifiers. When determining the completion of a region, S_M,z is compared with L_M,z and S_J,z is compared with L_J,z, and M_i and J_i are not combined into a single settlement quantity that omits either of the physical quantities.

19. The method according to claim 15, characterized in that, For at least 5×5 input combinations of metering pump stroke parameters and impulse parameters, at least 20 effective pulses of M_i and J_i are collected for each combination, and G is obtained by finite difference or least squares method. The inverse matrix of G is used only when the current input is within the convex hull of the input combination, the minimum singular value of G is not less than the pre-stored lower limit, and the number of conditions is not greater than the pre-stored upper limit. Otherwise, the nearest input combination in the two-dimensional response dataset that does not cause the cumulative value of any partition to exceed the corresponding upper limit is selected.

20. The method according to claim 15, characterized in that, For pulses with V_i=0, S_M,z, S_J,z and the two-dimensional response matrix G are not updated; when the distance, light intensity baseline and pressure baseline all meet the corresponding numerical conditions again and the target area is not locked, a new pulse identifier is assigned for retry; when the number of consecutive invalid attempts reaches the pre-stored upper limit, the output is stopped and the original sampling sequence, control parameters and invalid reason code are retained.

Images