A Method and System for Speaker Receiver Production Control Based on Parameter Adaptation
By injecting AC excitation current into the voice coil to obtain the mechanical impedance sequence, and using the polynomial least squares algorithm and dynamic switching of the Jacobian matrix, the adaptive adjustment of the loudspeaker receiver production control system is realized, which solves the problem of insufficient material phase state perception in the existing technology and improves acoustic performance and production stability.
Patent Information
- Application Number
- CN202610722658.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-25
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-05-25
AI Technical Summary
Existing loudspeaker receiver production control systems cannot detect the rheological phase evolution of adhesive materials in real time when faced with differences in the size of individual materials and fluctuations in workshop temperature and humidity. This leads to equipment control lag and accumulated errors, affecting the consistency of acoustic characteristics and total harmonic distortion.
By injecting AC excitation current into the voice coil, a mechanical impedance sequence is obtained. Then, using a polynomial least squares algorithm and dynamic switching of the Jacobian matrix, adaptive control commands are generated to adjust the energy input and holding voltage parameters in real time, thereby achieving sensitivity matching to the phase state of the adhesive material.
It improves the consistency of acoustic performance of finished loudspeakers, reduces internal stress shift caused by curing shrinkage, corrects the feed baseline drift of upstream fluid equipment, and improves the stability of the production process and product quality.
Smart Images

Figure CN122632761B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automated control technology in electroacoustic product manufacturing, specifically to a method and system for controlling the production of loudspeaker receivers based on parameter adaptation. Background Technology
[0002] In automated production and assembly, miniature loudspeakers and receivers typically require the bonding, curing, and pressing of voice coils and diaphragms. Existing production control systems often employ pre-set fixed parameters for open-loop control. This approach fails to detect the internal rheological phase evolution of the adhesive material when faced with variations in individual material dimensions or fluctuations in workshop temperature and humidity. Consequently, the equipment cannot adjust the holding pressure and energy input parameters based on the actual curing shrinkage stress changes of the material. Furthermore, conventional industrial bus communication, due to its periodic ladder diagram scanning mechanism, causes lag in the control actions of the underlying physical actuators, making it difficult for the pressing equipment to issue accurate control commands in a timely manner when the material crosses the liquid-solid phase critical point. In addition, upstream fluid dispensing equipment may experience baseline drift due to mechanical wear or air pressure fluctuations during continuous operation. Without a closed-loop correction mechanism, minor errors in the individual processing can accumulate across batches, leading to poor acoustic consistency of the miniature loudspeaker components output from the production line and an increased defect rate due to excessive total harmonic distortion. Summary of the Invention
[0003] To address the shortcomings of existing technologies, this invention provides a parameter-adaptive loudspeaker receiver production control method and system, which solves the problems of lack of fluid phase perception mechanism in the existing loudspeaker bonding and curing process, resulting in fixed parameter control being unable to cope with curing stress deviation caused by material differences and cumulative errors caused by equipment control lag.
[0004] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a method for controlling the production of a loudspeaker receiver based on parameter adaptation, comprising the following steps: The system receives a hardware interrupt trigger signal generated by the pressing actuator pressing the device into place, injects an AC excitation current containing the target frequency component into the voice coil of the speaker semi-finished product, synchronously acquires the response voltage signal at both ends of the voice coil, and performs frequency domain analysis to generate the original mechanical impedance sequence within the discrete calculation frame period. The original mechanical impedance sequence is smoothly reconstructed using a polynomial least squares algorithm, and a time-difference operation is performed based on the reconstructed impedance sequence to calculate and generate the transient mechanical impedance gradient vector and the mechanical impedance evolution acceleration vector. Monitor the zero-crossing attribute of the mechanical impedance evolution acceleration vector, and perform conditional logic judgment based on the number of consecutive zero-crossing frames to generate system state variables characterizing the solidification process. Use the system state variables as feedback parameters to reset the smooth window length parameter in the polynomial least squares algorithm. Based on the system state variables, the dynamic switching of the control sensitivity Jacobian matrix is performed in the preset multi-model state space control equations. The difference between the impedance gradient envelope reference and the transient mechanical impedance gradient vector is calculated, and algebraic solutions are performed in combination with the switched Jacobian matrix to output the control command adjustment vector. The control command adjustment vector is decomposed into a local compensation component and an upstream fluid compensation component. The pulse width modulation register or displacement timing register inside the programmable logic controller is overwritten based on the local compensation component, and the valve action time register inside the fluid dispensing device is overwritten based on the upstream fluid compensation component.
[0005] This invention calculates the equivalent mechanical stiffness using the electrical response parameters of the voice coil under forced vibration, determines the physical state of the material crossing the gel point by extracting the polarity reversal feature of the acceleration vector, and establishes a linkage mechanism between phase characteristics, filter window parameters, and control matrix weights to achieve adaptive matching of production control parameters to the real-time phase sensitivity of the adhesive material.
[0006] In a preferred embodiment of the present invention, the specific steps of performing frequency domain analysis to generate the original mechanical impedance sequence within the discrete calculation frame period are as follows: at the synchronization moment of the AC excitation current injection, the direct memory access channel is opened, and the response voltage signal and the actual current signal are continuously collected at a preset sampling rate. The continuously collected data is extracted and segmented and windowed according to the set discrete calculation frame period. For each sliced calculation frame, the fast Fourier transform algorithm is executed to extract the voltage frequency domain component and the current frequency domain component at the target frequency point. The complex impedance magnitude corresponding to the voltage frequency domain component and the current frequency domain component is calculated. The complex impedance magnitude corresponding to each calculation frame is combined into a one-dimensional data queue according to the time occurrence order to generate the original mechanical impedance sequence.
[0007] In a preferred embodiment of the present invention, the specific steps of applying the polynomial least squares algorithm to perform smooth reconstruction and difference operation on the original mechanical impedance sequence are as follows: a polynomial model is established based on the set of discrete data points within a dynamic smoothing window constructed by unidirectional backtracking from the current data frame to the historical data direction, and the least squares solution is solved to obtain the impedance characteristic value after polynomial smooth reconstruction; a first-order difference operation is performed on the continuous time dimension based on the reconstructed impedance characteristic value to extract the transient mechanical impedance gradient vector that maps the rate of change of material mechanical stiffness; a second-order difference operation is performed on the transient mechanical impedance gradient vector along the time axis to calculate and generate the mechanical impedance evolution acceleration vector.
[0008] In a preferred embodiment of the present invention, the specific steps of monitoring the zero-crossing attribute of the mechanical impedance evolution acceleration vector and performing conditional logic judgment are as follows: configuring a debouncing mechanism for the built-in accumulator; when the value of the mechanical impedance evolution acceleration vector in the current calculation frame is detected to be less than zero, the built-in accumulator increments; the value of the built-in accumulator is compared with a confirmation frame number threshold calibrated according to the mechanical response delay characteristics of the curing device; if the value of the built-in accumulator reaches the confirmation frame number threshold, it is determined that the adhesive material has crossed the gel point and a system state variable characterizing the solid phase-dominant stage is generated; if an acceleration value greater than or equal to zero occurs during the accumulation process, the built-in accumulator is immediately reset to zero and a system state variable characterizing the liquid phase-dominant stage is generated.
[0009] In a preferred embodiment of the present invention, the specific steps of resetting the smoothing window length parameter by using the system state variable as a feedback parameter are as follows: when the system state variable represents the liquid phase-dominant stage, the preset reference window length value in the polynomial least squares algorithm is maintained; when the system state variable flips and represents the solid phase-dominant stage, the smoothing window length parameter is reset in combination with the preset frequency doubling expansion coefficient to expand the sampling coverage range, adapt to the low-frequency impedance changes in the later stage of curing, and at the same time, a one-way hardware latching mechanism is triggered for the remaining processing cycle of the current product to prevent frequent state flips caused by subsequent local disturbances.
[0010] In a preferred embodiment of the present invention, the specific steps of dynamically switching the control sensitivity Jacobian matrix are as follows: based on the system state variables, the current material rheological phase state is identified; when in the liquid-phase dominant stage before crossing the gel point, a liquid-phase Jacobian matrix with high energy sensitivity weight is retrieved to match the control requirements of photothermal power on the initial reaction rate; when in the solid-phase dominant stage after crossing the gel point, the system is simultaneously switched to a solid-phase Jacobian matrix with high pressure holding sensitivity weight to adapt to the shrinkage stress control requirements of the polymer shaping stage.
[0011] In a preferred embodiment of the present invention, the specific steps of performing algebraic calculation to output the control command adjustment vector in combination with the switched Jacobian matrix are as follows: performing multidimensional incremental calculation by combining the control command reference state maintained by the previous calculation frame with the preset diagonal proportional gain matrix, and the switched Jacobian matrix with the impedance gradient error parameter; after obtaining the control command adjustment vector of the current calculation frame, introducing a preset maximum step size limiting function to clamp and truncate the command increment between adjacent calculation frames to avoid sudden changes in control command output caused by system state switching.
[0012] In a preferred embodiment of the present invention, the preparation step of overwriting the underlying register based on the local compensation component is as follows: the local compensation component is further decomposed into energy control primitives and voltage holding timing primitives; relying on a preset device memory mapping table that defines the linear proportional relationship between physical control increments and underlying register machine code, the above primitive values are converted into corresponding underlying hexadecimal operands; and the converted operands are packaged and encapsulated into bus read / write messages.
[0013] In a preferred embodiment of the present invention, the execution step of overwriting the underlying register is as follows: the bus read / write message is sent through the industrial real-time Ethernet bus to bypass the conventional ladder diagram scanning cycle of the programmable logic controller, directly address and overwrite the data bits of the pulse width modulation register to change the photothermal output duty cycle of the energy injection device, and directly address and overwrite the displacement timing register to change the pressure holding countdown parameters of the pressing actuator, thereby establishing a direct control path with lower response latency.
[0014] In a preferred embodiment of the present invention, the specific steps of overwriting the internal register of the fluid dispensing device based on the upstream fluid compensation component are as follows: configuring a sliding time window for the upstream fluid compensation component on a continuous production batch sequence, performing batch-dimensional integral operations to extract the cumulative integral value of fluid compensation, filtering out random errors in the discrete processing, comparing the absolute value of the cumulative integral value of fluid compensation with a baseline drift threshold set according to the dispensing quantity tolerance boundary, and when the baseline drift threshold is exceeded, converting the integral value into a valve action time adjustment word in combination with the feedforward proportional gain, overwriting the valve action time register inside the fluid dispensing device through a network link, and forcibly clearing the historical accumulated data in the sliding time window after the instruction is executed to reset the integrator and prevent the system from experiencing integral saturation.
[0015] A second aspect of the present invention provides a parameter-adaptive loudspeaker receiver production control system, comprising: The data acquisition module is used to receive hardware interrupt trigger signals and call external devices to inject AC excitation current into the voice coil of the speaker semi-finished product to obtain the response voltage signal, perform frequency domain analysis, and generate the original mechanical impedance sequence. The feature extraction module is used to perform smooth reconstruction and time-differential calculation on the original mechanical impedance sequence to generate transient mechanical impedance gradient vector and mechanical impedance evolution acceleration vector; The state determination module is used to monitor the zero-crossing attribute of the mechanical impedance evolution acceleration vector, generate system state variables, and feed them back to the feature extraction module to reset the smooth window length parameter. The equation solving module is used to dynamically switch the internally preset control sensitivity Jacobian matrix based on the system state variables and perform algebraic calculations in conjunction with the transient mechanical impedance gradient vector to output the control command adjustment vector; The register overwrite module is used to decompose the control command adjustment vector and overwrite the underlying control register and fluid equipment register respectively. This control system combines feature data extraction and multi-model control mechanisms to achieve online determination of the solidification phase of the fluid and closed-loop adjustment of process parameters.
[0016] In a preferred embodiment of the present invention, the control system includes a hardware structure for mounting various modules. The hardware structure includes an industrial control host computer with an internal peripheral communication bus and an external industrial real-time Ethernet bus; a programmable logic controller (PLC) connected to the industrial control host computer via the industrial real-time Ethernet bus and equipped with an electrical control terminal; an energy injection device connected to the electrical control terminal of the PLC and a pressing actuator; a fluid dispensing device connected to the industrial control host computer or the PLC via a network link; a test probe fixture connected to the mechanical displacement end of the pressing actuator; and a signal generator and a data acquisition card electrically connected to the detection pins of the test probe fixture.
[0017] In a preferred embodiment of the present invention, after the data acquisition module receives a hardware interrupt trigger pulse sent directly by the programmable logic controller across the main scan cycle, it immediately wakes up the suspended direct memory access acquisition thread and drives the underlying transmission channel to transport dual-channel sequence data to ensure the alignment of the start time of the acquisition cycle.
[0018] In a preferred embodiment of the present invention, when the feature extraction module performs smooth reconstruction using the polynomial least squares algorithm, its internally configured sliding window strictly follows the principle of unidirectional truncation along the historical time sequence direction from the current data acquisition frame, preventing the solution process from calling non-occurring sequence data that does not conform to causal relationships, so as to maintain the timing stability of the closed-loop control system.
[0019] In a preferred embodiment of the present invention, the state determination module maintains the monotonicity of the state determination result by configuring a one-way hardware latch mechanism associated with the current physical process cycle. When the internal accumulation logic confirms that the system state variable representing the curing process has completed the critical transition from liquid phase characteristics to solid phase characteristics, the value of the state variable is forcibly locked until the current speaker single-piece processing operation ends.
[0020] In a preferred embodiment of the present invention, the equation solving module is configured with a maximum step size limiting function component. Within the calculation frame in which the system state variables drive the internal control sensitivity Jacobian matrix to complete the model switching, the amplitude of the control command adjustment vector increment generated by the solution is truncated to maintain the motion stability of the external physical actuator.
[0021] In a preferred embodiment of the present invention, the register overwrite module uses a device memory mapping table divided into system memory regions to directly convert physical control increments into machine instructions for the underlying hardware addressing architecture, thereby reducing communication protocol conversion time.
[0022] In a preferred embodiment of the present invention, the register overwrite module includes an integrator unit associated with consecutive production batches. After the monitoring module confirms that the cumulative integral value of fluid compensation exceeds the baseline drift threshold and completes the remote register parameter modification action, it synchronously issues a clear instruction to its own memory management area to overwrite the stored historical compensation accumulation data.
[0023] This invention provides a method and system for controlling the production of loudspeaker receivers based on parameter adaptation. It has the following beneficial effects: 1. This invention obtains a mechanical impedance sequence by injecting an AC excitation current into the voice coil, and extracts the mechanical impedance evolution acceleration vector through differential operation. The zero-crossing property of this vector is used to determine the solid-liquid phase critical point of the adhesive material. Combined with the system state variables generated by the determination, the smoothing window length parameter is adaptively reset. Without adding additional sensing hardware, online identification of material rheological phase changes is achieved. Furthermore, the filtering scale is dynamically adjusted according to the data fluctuation characteristics of different curing stages, which improves the accuracy of curing state determination.
[0024] 2. Based on the system state variables generated by monitoring, this invention dynamically switches the control sensitivity Jacobian matrix in the preset multi-model control equations, and combines the transient mechanical impedance gradient vector to solve the control command. This mechanism enables the system to focus on energy input adjustment in the liquid phase and pressure holding parameter adjustment in the solid phase, so that the control variable output of the physical device matches the actual process requirements of the adhesive material in different crosslinking stages, reduces the internal stress shift caused by curing shrinkage, and improves the acoustic performance consistency of the finished micro loudspeaker.
[0025] 3. This invention decomposes and processes the control command adjustment vector separately. By directly overwriting the local compensation component into the underlying control register, it skips the periodic scanning process of the conventional controller ladder diagram and shortens the response delay of the underlying physical action. At the same time, it uses batch-dimensional sliding time window integration to overwrite the valve action time register of the dispensing equipment for the upstream fluid compensation component, which filters out random interference in the single-piece processing process, corrects the feed baseline drift of the upstream fluid equipment, and prevents the continuous accumulation of production errors. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the data acquisition and frequency domain analysis process of the present invention; Figure 4 This is a timing diagram illustrating the rheological phase state determination and parameter reconstruction according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the dynamic solution logic for the state-space control equations of the present invention; Figure 6 This is a schematic diagram of the cross-level register full-link overwrite logic of the present invention; Figure 7 This is a schematic diagram comparing the consistency distribution of loudspeaker resonant frequencies between the experimental group and the control group of this invention; Figure 8 This is a schematic diagram of the closed-loop dynamic overwrite trajectory of the valve action time in the fluid dispensing device of the present invention.
[0027] Among them, 100 is the data acquisition module; 200 is the feature extraction module; 300 is the state determination module; 400 is the equation solving module; and 500 is the register overwrite module. Detailed Implementation
[0028] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0029] Please see the appendix Figure 1 , Figure 1 This is a schematic diagram of the system architecture of a speaker receiver production control system based on parameter adaptation according to an embodiment of the present invention. The present invention provides a speaker receiver production control system based on parameter adaptation, the system including: a data acquisition module 100, a feature extraction module 200, a state determination module 300, an equation solving module 400, a register overwrite module 500, and a hardware structure for mounting each module.
[0030] The hardware structure includes: an industrial control host computer, a programmable logic controller, an energy injection device, a pressing actuator, a fluid dispensing device, a test probe fixture, a signal generator, and a data acquisition card.
[0031] The industrial control host computer is equipped with an internal peripheral communication bus and an external industrial real-time Ethernet bus.
[0032] The programmable logic controller is connected to the industrial control host computer via an industrial real-time Ethernet bus and is equipped with an electrical control terminal that connects the energy injection device and the pressing actuator.
[0033] The fluid dispensing equipment is connected to an industrial control host computer or programmable logic controller via a network link.
[0034] The test probe fixture is connected to the mechanical displacement end of the pressing actuator, and its probe pin is electrically connected to the data input end of the data acquisition card and the signal output end of the signal generator.
[0035] The data acquisition module 100 runs in the industrial control host computer and is used to receive the hardware interrupt trigger signal forwarded by the programmable logic controller, call the signal generator to inject AC excitation current containing the target frequency component into the voice coil of the speaker semi-finished product through the test probe fixture, and open the direct memory access channel of the data acquisition card to obtain the response voltage signal, thereby generating the original mechanical impedance sequence.
[0036] The feature extraction module 200 relies on the processor of the industrial control host computer to perform smooth reconstruction and differential operation on the original mechanical impedance sequence, and calculate and generate the transient mechanical impedance gradient vector and the mechanical impedance evolution acceleration vector.
[0037] The state determination module 300 runs in the industrial control host computer and is used to monitor the zero-crossing property of the mechanical impedance evolution acceleration vector, generate system state variables and write them into the state storage area of the industrial control host computer, and feed these variables back to the feature extraction module 200 to adjust the smoothing window parameters.
[0038] The equation solving module 400 runs in the industrial control host computer. It is used to retrieve the corresponding control sensitivity Jacobian matrix from the internally preset state space equation model based on the system state variables, and perform the solution in combination with the transient mechanical impedance gradient vector to generate the control command adjustment vector.
[0039] The register overwrite module 500 is associated with the communication port of the industrial control host computer. It is used to decompose the control command adjustment vector, overwrite the pulse width modulation register or displacement timing register inside the programmable logic controller through the industrial real-time Ethernet bus to change the photothermal output and pressure holding time, and overwrite the valve action time register inside the fluid dispensing equipment through the network link.
[0040] See attached document Figure 2 , Figure 2This is a schematic flowchart of a speaker receiver production control method based on parameter adaptation according to an embodiment of the present invention. The present invention provides a speaker receiver production control method based on parameter adaptation, comprising the following steps: S1 receives the hardware interrupt trigger signal generated by the pressing actuator pressing down to the position, injects AC excitation current containing the target frequency component into the voice coil of the speaker semi-finished product, synchronously acquires the response voltage signal at both ends of the voice coil, and performs frequency domain analysis to generate the original mechanical impedance sequence within the discrete calculation frame period; S2, apply the polynomial least squares algorithm to smoothly reconstruct the original mechanical impedance sequence, perform time dimension difference operation based on the reconstructed impedance sequence, and calculate and generate the transient mechanical impedance gradient vector and the mechanical impedance evolution acceleration vector. S3 monitors the numerical zero-crossing attribute of the mechanical impedance evolution acceleration vector, performs conditional logic judgment based on the number of consecutive zero-crossing frames, generates system state variables characterizing the solidification process, and uses the system state variables as feedback parameters to reset the smooth window length parameter in the polynomial least squares algorithm. S4, based on system state variables, performs dynamic switching of the control sensitivity Jacobian matrix in the preset multi-model state space control equations, calculates the difference between the impedance gradient envelope reference and the transient mechanical impedance gradient vector, and performs algebraic solution in combination with the switched matrix, outputting the control command adjustment vector. S5 decomposes the control command adjustment vector into a local compensation component and an upstream fluid compensation component. Based on the local compensation component, it overwrites the pulse width modulation register or displacement timing register inside the programmable logic controller, and based on the upstream fluid compensation component, it overwrites the valve action time register inside the fluid dispensing device.
[0041] Please see the appendix Figure 3 , Figure 3 This is a schematic diagram of the data acquisition and frequency domain analysis process according to an embodiment of the present invention. The data acquisition module 100 monitors the level status on the peripheral communication bus of the industrial control host computer. When the pressing actuator drives the test probe fixture to press down and contact the voice coil pad of the speaker semi-finished product and reach the preset physical displacement pole, the electrical control terminal of the programmable logic controller captures the mechanical positioning action and converts it into a level transition pulse. The tolerance compression margin corresponding to the physical displacement pole is in the range of 0.1 mm to 0.5 mm. The pulse is sent directly to the industrial control host computer as a hardware interrupt trigger signal across the main scan cycle. The data acquisition module 100 receives the hardware interrupt trigger signal and wakes up the direct memory access acquisition thread that is in a suspended state.
[0042] As a preferred method, the data acquisition module 100 sends waveform control words to the signal generator via the peripheral communication bus, controlling the signal generator to inject sweep frequency excitation current into the voice coil of the speaker semi-finished product through the probe pin of the test probe fixture. As for the specific circuit structure and impedance matching method of the signal generator to generate a specific frequency band AC constant current signal, those skilled in the art can refer to the electronic manual for conventional configuration based on the nominal DC resistance of the speaker voice coil. The process of generating AC signals is a well-known technology in the field and will not be described in detail here.
[0043] In this embodiment, the speaker voice coil generates microscopic forced vibration under the action of AC excitation current. Since the voice coil is in close contact with the curing adhesive, the physical phase change of the adhesive during the rheological process will directly cause the change of the equivalent stiffness and damping of the voice coil mechanical system, which will then be deterministically reflected in the response voltage fluctuation at the electrical terminal. At the synchronous moment of the excitation current injection, the data acquisition module 100 calls the data acquisition card to open the direct memory access channel, continuously acquires the response voltage signal at both ends of the voice coil and the actual current signal flowing through the voice coil at a preset sampling rate, and transfers the dual-channel sequence data to the pre-allocated ring data buffer in the memory of the industrial control host computer. The actual current signal flowing through the voice coil obtains the voltage drop data through the high-precision non-inductive shunt resistor connected in series in the test probe fixture detection return loop and extracts it according to Ohm's law.
[0044] The specific value of the aforementioned preset sampling rate is determined based on the Nyquist sampling theorem and is usually set to 3-5 times the highest frequency component of the target to balance anti-aliasing effect and memory overhead. The data acquisition module 100 extracts continuously acquired data from the circular data buffer and performs frame division and windowing truncation processing according to the discrete calculation frame period. The discrete calculation frame period is defined as the time span for processing a single discrete data, and its value is usually set between 10-50 milliseconds. The parameter setting of this duration must accurately cover the lowest frequency period of the AC excitation signal and be consistent with the basic scheduling rhythm of the industrial control host computer.
[0045] For each sliced calculation frame, the data acquisition module 100 performs a fast Fourier transform algorithm to extract the voltage and current frequency domain components at the target frequency point and calculates the corresponding complex impedance magnitude. This calculation process satisfies the following logical expression: ; In the formula, For the first The specific values of the original mechanical impedance sequence generated for each discrete computation frame; The calculation frame index number is the discrete-time dimension. Mathematical operation symbols for solving the modulus of complex variables; For the first The complex voltage frequency domain component at the target frequency extracted by transformation in each calculation frame; The target frequency parameter is obtained through preliminary testing and calibration based on the resonant frequency of the speaker's voice coil structure. For the first The complex current frequency domain components at the target frequency in each calculation frame.
[0046] Based on the above complex domain analysis process, the data acquisition module 100 combines the impedance magnitude values calculated by each discrete calculation frame into a one-dimensional data queue according to the time occurrence order, thereby generating a complete original mechanical impedance sequence containing features that change with time, and continuously writes the original mechanical impedance sequence into a specific shared memory region. The feature extraction module 200 directly reads the updated original mechanical impedance sequence by monitoring the memory pointer address changes of the shared memory region.
[0047] The feature extraction module 200 reads the original mechanical impedance sequence output by the data acquisition module 100 from the shared memory area. High-frequency electromagnetic interference in the industrial field and mechanical forced vibration caused by the operation of the pressing actuator will mix unrelated physical high-frequency noise into the original impedance data. Directly performing differentiation operations on the original data will cause the noise to be amplified exponentially at the numerical level. The feature extraction module 200 applies a polynomial least squares algorithm to perform sliding window smooth reconstruction of the original mechanical impedance sequence. As a preferred method, this algorithm specifically uses a Savitzky-Golay smoothing filter to realize the local polynomial fitting operation of the time series.
[0048] The feature extraction module 200 establishes a polynomial model based on the set of data points within the sliding window and solves for the least squares solution. Its reconstruction calculation satisfies the following mathematical expression: ; In the formula, For the th after polynomial smooth reconstruction The impedance characteristic value of the frame; The normalization factor constant corresponding to the polynomial smoothing filtering algorithm is determined by the total number of data points in the sliding window and is preferably limited to an odd number ≥3. This is a local summation index variable used to locate relative positions within the sliding window; For the first The half-window length variable of the dynamic smoothing window corresponding to the frame; This is the initial lower limit value for the local summation of this dynamic smoothing window; For local summation index in The polynomial fitting filter coefficient values corresponding to the position; Offset to the current computed frame index To meet the causality constraints of the system's real-time high-frequency closed-loop control, the original mechanical impedance data points at each location are constructed by unidirectionally backtracking from the current data frame to the historical data during actual calculation, in order to avoid calling future data that has not yet occurred.
[0049] The polymer cross-linking reaction triggered by glue curing causes a continuous change in the system's equivalent mechanical stiffness. The feature extraction module 200 performs a first-order difference operation in the continuous-time dimension based on the reconstructed impedance sequence to extract the transient mechanical impedance gradient vector. This operation process satisfies the following logical expression: ; In the formula, For the first The specific value of the transient mechanical impedance gradient vector corresponding to the frame; The calculation frame index number is the discrete-time dimension. For the current number The smoothed reconstruction impedance value corresponding to each discrete computation frame; The smooth reconstructed impedance value generated and cached for the previous discrete computation frame; A fixed time interval parameter for a pre-defined discrete calculation frame period.
[0050] This gradient vector maps the dynamic rate of change of the speaker voice coil's mechanical stiffness over time at the glue interface. During the algorithm startup phase, when the index... When the value is equal to 1, due to the lack of preceding data frames, the system calls the three frames of smoothed background impedance data continuously collected in the mechanical silent state before the excitation current is injected, performs the averaging operation, and assigns it to the initial state for boundary condition overflow prevention.
[0051] After acquiring the transient mechanical impedance gradient vector, the feature extraction module 200 performs a second-order difference extraction operation along the time axis to calculate and generate a mechanical impedance evolution acceleration vector that maps the phase evolution characteristics of the crosslinking reaction. This calculation logic conforms to the following relationship: ; In the formula, For the first Numerical results of the mechanical impedance evolution acceleration vector corresponding to each discrete computation frame; The calculation frame index number is the discrete-time dimension. For the current number The transient mechanical impedance gradient value corresponding to each discrete calculation frame; The transient mechanical impedance gradient value generated and cached for the previous discrete computation frame; The fixed time interval parameter is the discrete calculation frame period set above.
[0052] The acceleration vector is used to extract the microscopic physical features of the inflection point of the crosslinking reaction rate through second-order differentiation. The feature extraction module 200 synchronously writes the generated transient mechanical impedance gradient vector and mechanical impedance evolution acceleration vector into a specific state storage area of the industrial control host computer for the state determination module 300 to read and call.
[0053] Please see the appendix Figure 4 , Figure 4 This is a timing diagram of rheological phase determination and parameter reconstruction according to an embodiment of the present invention. The state determination module 300 obtains the mechanical impedance evolution acceleration vector calculated and output by the feature extraction module 200 from the state storage area. During the curing process, the polymer adhesive undergoes a phase transition from a viscous fluid to an elastic solid. At the critical point of this physical transition, i.e. the gel point defined by rheology, the cross-linking reaction rate inside the material reaches a global extreme value and is manifested as the peak value of the mechanical equivalent stiffness growth rate. When mapped to the electrical detection end, it is shown that the transient mechanical impedance gradient vector reaches a maximum value and its second derivative, i.e. the mechanical impedance evolution acceleration vector, undergoes a deterministic reversal of positive and negative polarity. As a preferred approach, the state determination module 300 establishes a state mapping mechanism between the underlying physical phase and the digital control domain based on this positive and negative polarity reversal feature.
[0054] In industrial curing environments, mechanical vibrations and thermal disturbances can cause localized data jumps. A polarity reversal in a single data frame can easily lead to out-of-bounds misjudgments in the state determination algorithm. The state determination module 300, equipped with a built-in accumulator, employs a built-in anti-jitter mechanism to execute conditional logic determination. When the value of the mechanical impedance evolution acceleration vector is detected to be less than zero, the accumulator increments in the current calculation frame and compares it with a preset confirmation frame threshold. This confirmation frame threshold is calibrated to three to eight consecutive frames based on the mechanical response delay characteristics of the curing equipment. When the accumulator value reaches this threshold, the state determination module 300 confirms that the gel point crossing event has actually occurred and generates a system state variable. The specific determination logic for this system state variable satisfies the following conditional expression: ; In the formula, For the first The system state variables output by frame calculation; This is a system state constant characterizing the liquid-phase-dominated stage before the glue reaches its gel point; The system state constant is used to characterize the state of the glue after it has crossed the gel point and entered the solid-phase-dominated stage. This is the relative offset index variable used in the anti-shake mechanism to sum up back to historical frames; The threshold number of confirmation frames is determined based on the mechanical response delay characteristics of the curing equipment; The upper bound value for summation of historical frames; This is a unit step function whose function value is one when the independent variable is greater than or equal to zero, and zero otherwise. Backtracking forward to offset the current computation frame The mechanical impedance evolution acceleration vector at each location.
[0055] In this embodiment, if an acceleration value greater than or equal to zero occurs during the accumulation process, the accumulator is immediately reset to zero. At this time, the system state variable equal to zero indicates that the glue is in the liquid-phase dominant stage before reaching the gel point. The system state variable equal to one indicates that the glue has crossed the gel point and entered the solid-phase dominant stage. This state variable is triggered by a one-way hardware latching mechanism throughout the entire remaining processing cycle of the current speaker semi-finished product to shield the state bounce oscillation caused by subsequent local physical disturbances. The state determination module 300 synchronously writes the generated system state variable into the state storage area of the industrial control host computer. At the same time, the state determination module 300 sends the updated system state variable as a feedback parameter to the feature extraction. Module 200 triggers the underlying closed-loop parameter reset logic. The feature extraction module 200 dynamically switches the half-window length parameter in the polynomial least squares algorithm based on the received system state variable value. When the system state variable is equal to zero, the feature extraction module 200 maintains the preset reference window length value. This reference window length value is usually set to include an interval of 5 to 11 discrete data points. When the system state variable flips to 1, the feature extraction module 200 resets the reference window length in combination with the frequency doubling coefficient. The value range of the frequency doubling coefficient is set to 3.0 to 5.0. Through this closed-loop parameter reset logic, the smoothing filter algorithm parameters are adaptively tracked to the underlying rheological phase state.
[0056] Please see the appendix Figure 5 , Figure 5 This is a schematic diagram of the dynamic solution logic of the state-space control equation according to an embodiment of the present invention. The equation solution module 400 synchronously reads the system state variables generated by the state determination module 300 and the transient mechanical impedance gradient vector output by the feature extraction module 200 from the memory of the industrial control host computer. The module has a pre-stored impedance gradient envelope reference sequence, which is constructed by offline scanning and extracting the impedance gradient mean curve of a large number of physical test qualified samples in the early stage. The equation solution module 400 performs a subtraction operation between the transient mechanical impedance gradient vector of the current calculation frame and the impedance gradient envelope reference of the corresponding time node to extract the impedance gradient error parameter characterizing the deviation of the curing process. The corresponding time node is offset and aligned with the system clock stamp of the hardware interrupt trigger pulse of the pressing actuator captured by the data acquisition module as the absolute time synchronization origin. The error parameter serves as the basic input data of the multi-model state-space control equation to drive the system to generate a control command adjustment vector containing local energy compensation components, pressure holding timing compensation components, and upstream fluid compensation components.
[0057] As a preferred approach, the multi-model state-space control equations embed multiple sets of control sensitivity Jacobian matrices that strictly correspond to the rheological phase state of the adhesive. The elements within these matrices characterize the local partial derivative relationship between the system's physical control input and the impedance gradient output, and the specific values are obtained through a step control perturbation test applied in the early stage. In the liquid-phase dominant stage before crossing the gel point, the photothermal power output by the energy injection device exhibits high control sensitivity to the crosslinking reaction rate, while the influence of mechanical pressure is relatively weak. After crossing the gel point and entering the solid-phase dominant stage, the crosslinking network of the adhesive molecular chains solidifies, leading to a decrease in photothermal sensitivity, while the mechanical holding time plays a dominant role in the final curing shrinkage rate. The equation solving module 400 performs dynamic hard switching of the Jacobian matrix based on the received system state variables. When the system state variables are equal to zero, the system retrieves the liquid-phase Jacobian matrix configured with high energy sensitivity weights. When the system state variables are equal to one, the system synchronously switches to the solid-phase Jacobian matrix configured with high holding pressure sensitivity weights.
[0058] In this embodiment, the equation solving module 400 combines the switched control sensitivity Jacobian matrix and impedance gradient error parameter to perform algebraic calculation and output control command adjustment vector. The incremental feedback calculation process of this multidimensional parameter satisfies the following state-space control equation: ; In the formula, For the first The frame-generated control command adjustment vector includes multi-dimensional compensation components; The calculation frame index number is the discrete-time dimension. The control command reference state vector maintained in the previous calculation frame; The diagonal proportional gain matrix is used to maintain the closed-loop stability of the system; The control sensitivity Jacobian matrix is dynamically selected based on the system state variables; For the current number The system state variables output by the frame determination; The Moore-Penrose generalized inverse matrix operation symbol in matrix operations; For the first The impedance gradient error parameter, which characterizes the deviation of the solidification state, is extracted by frame calculation.
[0059] After obtaining the control command adjustment vector of the current calculation frame, the equation solving module 400 introduces a preset maximum step size limiting function to clamp and truncate the command increment of adjacent frames. Then, the equation solving module 400 passes the control command adjustment vector output by the solution to the next process step for physical mapping execution of the underlying control domain.
[0060] Please see the appendix Figure 6 , Figure 6This is a schematic diagram of the cross-level register full-link overwrite logic according to an embodiment of the present invention. The register overwrite module 500 and the analytical equation solving module 400 calculate the output control instruction adjustment vector and perform dimensional decomposition operations to extract the local compensation component and the upstream fluid compensation component. The local compensation component is further divided by the system into energy control primitives and pressure holding timing primitives. The register overwrite module 500 converts the above primitive values into low-level hexadecimal operands for the programmable logic controller based on a preset device memory mapping table. The device memory mapping table internally defines the linear proportional relationship between the physical control increment and the low-level register binary machine code. The register overwrite module 500 packages and encapsulates the converted operands into standard industrial bus read / write messages and directly addresses the corresponding memory block of the programmable logic controller through the industrial real-time Ethernet bus.
[0061] As a preferred approach, the system directly intervenes in the photothermal output duty cycle control of the energy injection device by rewriting the data bits of the pulse width modulation register and simultaneously overwrites the displacement timing register to change the pressure holding countdown parameters of the pressing actuator. In conventional industrial control architectures, upper-level instructions usually need to wait for the programmable logic controller to complete a full ladder diagram scan cycle before they can take effect and be executed. In this embodiment, the communication interaction process bypasses this conventional scan cycle and directly addresses and operates the underlying hardware registers. Based on this, the system can forcibly take over the physical execution trajectory of the current speaker semi-finished product curing station within a microsecond response time. This direct overwriting mechanism of the underlying registers reduces the handshake delay of industrial communication and establishes a single-machine high-frequency closed-loop control link based on real-time status feedback.
[0062] In this embodiment, the system extends the control domain from a single-machine closed loop to a feedforward correction process at the production line level. The register overwrite module 500 configures a sliding time window on the continuous production batch sequence for the aforementioned decomposed upstream fluid compensation component and performs batch-dimensional integral decision calculation. This discrete integral calculation process satisfies the following state evolution expression: ; In the formula, As of the end of the The cumulative integral value of fluid compensation calculated for each production batch; This serves as the global index number for product batches during continuous production. Use the index variable for batch summation within the sliding time window; The batch length constant for the pre-configured sliding time window, with a value ranging from 20 to 50 product batches; The lower limit value for the initial summation of batch traversal within the sliding time window; For the first The numerical value of the single fluid compensation component extracted for each production batch.
[0063] This calculation process filters out high-frequency random disturbances introduced by a single production run and extracts the low-frequency baseline drift trend that objectively exists in the upstream material distribution system.
[0064] The register overwrite module 500 compares the absolute value of the fluid compensation cumulative integral value with a preset baseline drift threshold. This baseline drift threshold is usually set according to the factory-calibrated dispensing volume tolerance boundary of the fluid dispensing equipment, and its specific value generally corresponds to 5%-15% of the standard single dispensing physical volume. When the absolute value of the cumulative integral value exceeds the baseline drift threshold, the register overwrite module 500, in conjunction with the feedforward proportional gain, converts the integral value into a small valve action time adjustment word. The specific value range of the feedforward proportional gain is calibrated to be between 0.01 and 0.1. The system sends the adjustment word to the fluid dispensing equipment in the upstream process through the transmission control protocol network link and directly overwrites its internal valve action time register. After the overwrite instruction is executed, the register overwrite module 500 forcibly clears the historical accumulated data in the sliding time window, resetting the integrator to zero. This zeroing mechanism cuts off the continuous integration process from a mathematical logic perspective and avoids the system falling into the control dead zone of integral saturation or overcompensation.
[0065] Please see the appendix Figure 7 and attached Figure 8 .
[0066] The present invention further provides a specific industrial application embodiment, as well as corresponding experimental verification and effect comparison.
[0067] I. Application Examples: This embodiment is applied to the bonding and curing production line of voice coil and diaphragm of miniature loudspeakers. Ultraviolet light and heat dual-curing adhesive is selected as the material medium. The basic scheduling cycle parameter of the industrial control host computer is set to 20 milliseconds, and the batch length constant of the sliding time window configured in the register overwrite module 500 is set to 30 product batches.
[0068] When the pressing actuator drives the test probe fixture to press down on the voice coil pad of the speaker semi-finished product and the tolerance compression margin corresponding to the physical displacement pole reaches 0.2 mm, the electrical control terminal of the programmable logic controller generates a hardware interrupt trigger signal. The data acquisition module 100 receives the hardware interrupt trigger signal and wakes up the direct memory access acquisition thread that is in a suspended state. The data acquisition module 100 injects a target frequency AC excitation current of 1 kHz into the voice coil and extracts the response voltage signal and the actual current signal within a discrete calculation frame period of 20 milliseconds to generate the original mechanical impedance sequence. The feature extraction module 200 uses a sliding window containing 5 data points to smoothly reconstruct the original mechanical impedance sequence and continuously performs a second differential operation along the time dimension to generate a mechanical impedance evolution acceleration vector.
[0069] The state determination module 300 monitors the specific value of the mechanical impedance evolution acceleration vector. When it detects that the mechanical impedance evolution acceleration vector values corresponding to four consecutive calculation frames are all less than zero, the state determination module 300 resets the corresponding system state variable parameter from 0 to 1. The feature extraction module 200 receives the system state variable value and expands and resets the half-window length parameter inside the polynomial least squares algorithm from 5 discrete data points to 15 discrete data points.
[0070] When the system state variable is equal to 0, the equation solving module 400 retrieves the liquid phase Jacobian matrix to calculate the control command adjustment vector. The register overwrite module 500 overwrites the pulse width modulation register according to the local compensation component in the control command adjustment vector and increases the photothermal output power parameter of the energy injection device by 5%. When the system state variable is equal to 1, the equation solving module 400 introduces the maximum step size limiting function and retrieves the solid phase Jacobian matrix to calculate the corresponding control command adjustment vector. The register overwrite module 500 overwrites the displacement timing register according to the local compensation component and extends the holding time parameter of the pressing actuator by 0.2 seconds. After completing 30 batches of continuous production operations, the register overwrite module 500 performs integral decision calculation on the upstream fluid compensation component and generates the valve action time adjustment word. The register overwrite module 500 overwrites the valve action time register inside the fluid dispensing equipment through the network link and reduces the valve opening time value of a single dispensing action by 3 milliseconds.
[0071] II. Experimental Verification and Effect Comparison: A total of 20,000 speaker semi-finished products were produced. Of these, 10,000 were produced using the traditional fixed-parameter open-loop control method (control group), and the other 10,000 were produced using the parameter adaptive control system of this invention (experimental group).
[0072] The verification analysis indicators selected are the standard deviation of resonant frequency consistency, the defect rate of total harmonic distortion exceeding the standard, the average production cycle time per piece, and the size parameter of the thickness range of the cured adhesive line.
[0073] The experimental results are shown in the table below.
[0074]
[0075] From the table above, we can see that: The standard deviation of resonant frequency consistency of the experimental group test samples decreased from 18.5 Hz to 4.2 Hz compared to the control group test samples. The total harmonic distortion exceeding the standard defect rate of the experimental group test samples decreased from 3.45% to 0.62% compared to the control group test samples. The average single-piece production cycle time of the experimental group test samples decreased from 4.8 seconds to 3.5 seconds compared to the control group test samples.
[0076] The thickness range of the cured adhesive line in the experimental group decreased from 0.15 mm to 0.04 mm compared to the control group.
[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for controlling the production of loudspeaker receivers based on parameter adaptation, characterized in that, Includes the following steps: An AC excitation current is injected into the voice coil of the speaker semi-finished product and a response voltage signal is obtained to generate the original mechanical impedance sequence; The original mechanical impedance sequence is smoothly reconstructed and a time-dimension quadratic difference operation is performed to generate a transient mechanical impedance gradient vector and a mechanical impedance evolution acceleration vector. Monitor the zero-crossing property of the mechanical impedance evolution acceleration vector, generate system state variables, and reset the smoothing window length parameter; Based on the system state variable switching control sensitivity Jacobian matrix, and combined with the transient mechanical impedance gradient vector, the output control command adjustment vector is calculated. The control command adjustment vector is decomposed into a local compensation component and an upstream fluid compensation component. The underlying control register is overwritten based on the local compensation component, and the valve action time register of the fluid dispensing device is overwritten based on the upstream fluid compensation component. The steps based on the system state variable switching control sensitivity Jacobian matrix are as follows: When the system state variables are in the liquid-dominant stage, the liquid phase Jacobian matrix is retrieved, and the energy sensitivity weight of the liquid phase Jacobian matrix is higher than the pressure holding sensitivity weight. When the system state variables are in the solid-phase dominant stage, switch to the solid-phase Jacobian matrix, where the pressure holding sensitivity weight is higher than the energy sensitivity weight. The step of calculating the output control command adjustment vector by combining the transient mechanical impedance gradient vector is as follows: Multidimensional incremental solution is performed by combining the preset diagonal proportional gain matrix, the switched Jacobian matrix, and the error parameters of the impedance gradient envelope reference and the transient mechanical impedance gradient vector. After obtaining the control command adjustment vector of the current calculation frame, a preset maximum step size limiting function is introduced to truncate the magnitude of the command increment between adjacent calculation frames.
2. The speaker receiver production control method based on parameter adaptation according to claim 1, characterized in that, The step of injecting AC excitation current into the voice coil of the speaker semi-finished product and obtaining the response voltage signal to generate the original mechanical impedance sequence is as follows: The response voltage signal and the actual current signal are continuously acquired at a preset sampling rate, and frame truncation is performed according to the set discrete calculation frame period. The Fast Fourier Transform algorithm is executed on the sliced computation frame to extract the voltage frequency domain component and the current frequency domain component at the target frequency point; The complex impedance magnitudes corresponding to the voltage frequency domain component and the current frequency domain component are calculated, and the complex impedance magnitudes within consecutive calculation frames are combined in chronological order to generate the original mechanical impedance sequence.
3. The speaker receiver production control method based on parameter adaptation according to claim 1, characterized in that, The steps for smoothly reconstructing the original mechanical impedance sequence and performing a time-dimension quadratic difference operation are as follows: A smooth window is constructed by tracing back from the current data frame to the historical data direction. A polynomial model is established for the discrete data points within the window and the least squares solution is obtained to obtain the impedance characteristic value after polynomial smooth reconstruction. Based on the reconstructed impedance eigenvalues, perform a first-order difference operation in the continuous-time dimension to extract the transient mechanical impedance gradient vector; A second difference operation is performed on the transient mechanical impedance gradient vector along the time axis to calculate and generate the mechanical impedance evolution acceleration vector.
4. The speaker receiver production control method based on parameter adaptation according to claim 1, characterized in that, The steps for monitoring the zero-crossing attribute of the mechanical impedance evolution acceleration vector and generating system state variables are as follows: When the value of the mechanical impedance evolution acceleration vector in the current calculation frame is detected to be less than zero, the built-in accumulator increments. The value of the built-in accumulator is compared with a preset acknowledgment frame count threshold; If the value of the built-in accumulator reaches the threshold of the number of confirmed frames, a system state variable representing the solid-phase dominant stage is generated. If an acceleration value greater than or equal to zero occurs during the accumulation process, the built-in accumulator is reset to zero, and a system state variable characterizing the liquid-dominated stage is generated.
5. The speaker receiver production control method based on parameter adaptation according to claim 4, characterized in that, The step of resetting the smooth window length parameter is as follows: When the system state variables represent the liquid-dominant stage, the preset reference window length value in the polynomial least squares algorithm is maintained. When the system state variable represents the solid-phase dominant stage, the smooth window length parameter is reset in combination with the preset frequency doubling expansion coefficient to expand the sampling coverage range, and at the same time, a one-way hardware latching mechanism is triggered to lock the system state variable until the end of the current product processing cycle.
6. The speaker receiver production control method based on parameter adaptation according to claim 1, characterized in that, The step of overwriting the underlying control register based on the local compensation component is as follows: The local compensation component is decomposed into energy control primitives and voltage holding timing primitives. The primitives are converted into hexadecimal operands and encapsulated into bus read / write messages through a preset device memory mapping table. The bus read / write message is sent through the communication bus to directly address and overwrite the pulse width modulation register to change the photothermal output duty cycle, and to directly address and overwrite the displacement timing register to change the voltage holding countdown parameters.
7. The speaker receiver production control method based on parameter adaptation according to claim 1, characterized in that, The step of overwriting the valve action time register of the fluid dispensing device based on the upstream fluid compensation component is as follows: A sliding time window is configured on the continuous production batch sequence, and batch-dimensional integration is performed on the upstream fluid compensation component to extract the cumulative integral value of fluid compensation. When the absolute value of the fluid compensation cumulative integral value is greater than the preset baseline drift threshold, the integral value is converted into a valve action time adjustment word in combination with the feedforward proportional gain. The valve action time register is overwritten through the network link, and the historical accumulated data in the sliding time window is cleared to reset the integrator.
8. A speaker receiver production control system based on parameter adaptation, characterized in that, The speaker receiver production control method based on parameter adaptation according to any one of claims 1-7 includes: The data acquisition module is used to inject AC excitation current into the voice coil of the speaker semi-finished product and acquire the response voltage signal to generate the original mechanical impedance sequence; The feature extraction module is used to smoothly reconstruct the original mechanical impedance sequence and perform a time-dimension quadratic difference operation to generate a transient mechanical impedance gradient vector and a mechanical impedance evolution acceleration vector. The state determination module is used to monitor the zero-crossing attribute of the mechanical impedance evolution acceleration vector, generate system state variables, and reset the smoothing window length parameter. The equation solving module is used to solve and output the control command adjustment vector based on the control sensitivity Jacobian matrix of the system state variable switching and the transient mechanical impedance gradient vector. The register overwrite module is used to decompose the control command adjustment vector into a local compensation component and an upstream fluid compensation component, overwrite the underlying control register based on the local compensation component, and overwrite the valve action time register of the fluid dispensing device based on the upstream fluid compensation component.
Citation Information
Patent Citations
Method and arrangement for controlling an electro-acoustical transducer
CN104756519A
Loudspeaker diaphragm displacement control method and control system
CN116634330A
