Injection molding process parameter self-adaptive compensation system

Abstract

The invention relates to the technical field of precise injection molding control, and discloses an injection molding process parameter self-adaptive compensation system which comprises a multi-source sensing module used for collecting mold cavity pressure, melt temperature and screw displacement information in the injection molding process in real time; the data fusion module is used for performing time sequence synchronization, space registration and noise suppression on the multi-source sensing data; the dynamic modeling module is used for constructing a dynamic model of an injection molding process state based on the fused process data and outputting a state estimation value; and the stability analysis module is used for carrying out system stability analysis based on the dynamic model and generating a control compensation strategy. Cooperative control of precise positioning and high-pressure response is realized through a composite driving framework and a dynamic disturbance compensation technology; an incremental learning algorithm and a multi-dimensional quality evaluation system are combined, a process parameter self-evolution mechanism is constructed, material characteristic drift and complex working condition disturbance are effectively dealt with, and closed-loop optimization from process control to quality feedback is formed.

Description

Technical Field

[0001] The present invention relates to the technical field of precision injection molding control, and particularly to an adaptive compensation system for injection molding process parameters. Background Art

[0002] Precision injection molding technology is a key process for mass manufacturing of complex plastic components and is widely used in fields such as electronic components and medical devices. With the increasing complexity of product structures, the manufacturing process needs to simultaneously meet the requirements of micron-level dimensional accuracy and high repeatability, which poses higher challenges to the dynamic response ability of the process control system.

[0003] In the current technical system, the modeling method based on fixed parameters is difficult to adapt to the fluctuations of material properties, resulting in a steady-state error in pressure control during the injection stage; traditional hydraulic or pure electric actuators have a performance compromise between high-speed high-pressure and precision positioning, restricting the expansion of the process window; quality inspection mostly relies on single-point sensor data, and there are blind spots in the assessment of internal defects and three-dimensional topography; the adjustment of process parameters depends on manual experience, and repeated trial and error is required during product changeover. These technical shortcomings form a mutually restrictive relationship - model errors reduce control accuracy, the limitations of actuators exacerbate quality fluctuations, and the lack of detection means further amplifies process uncertainty.

[0004] The core contradiction faced by the industry lies in the mismatch between the static characteristics of the existing technical framework and the dynamic requirements of the manufacturing process. Especially in the scenario of multi-variety and small-batch production, it is difficult for the overall system efficiency to break through the bottleneck. Summary of the Invention

[0005] In view of the deficiencies of the existing technology, the present invention provides an adaptive compensation system for injection molding process parameters, which solves the problem that it is difficult to achieve both process stability and efficiency due to the mismatch of dynamic modeling in injection molding equipment under complex working conditions, the difficulty in coordinating multi-objective control, and the lack of quality closed-loop feedback.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: An adaptive compensation system for injection molding process parameters, comprising: A multi-source sensing module for real-time collecting information on cavity pressure, melt temperature, and screw displacement during the injection molding process; A data fusion module for performing time series synchronization, spatial registration, and noise suppression on multi-source sensing data; A dynamic modeling module for constructing a dynamic model of the injection molding process state based on the fused process data and outputting a state estimate value; A stability analysis module for performing system stability analysis based on the dynamic model and generating a control compensation strategy; A compensation decision module for comprehensively considering the control compensation strategy and disturbance estimation information and outputting an injection molding machine control instruction; An execution drive module, configured to drive an injection molding device to perform corresponding actions according to control instructions; A quality assessment module, configured to evaluate the molding quality based on feedback signals during the molding process and output an assessment result, where the assessment result is used to determine whether the process meets the set quality requirements; A parameter optimization module, configured to update model parameters based on the quality assessment result to achieve adaptive adjustment of the system.

[0007] Preferably, the multi-source sensing module includes: A plurality of piezoelectric pressure sensors, which are arranged in the mold cavity and used to obtain multi-point pressure change data in the mold cavity; A temperature sensing unit, which adopts the infrared temperature measurement principle and is used for non-contact measurement of the melt temperature; A high-resolution displacement sensor, which is used to measure the position and speed changes of the screw during the injection process to comprehensively sense the dynamic parameters of the injection molding process.

[0008] Preferably, the data fusion module includes: A high-precision clock synchronization unit, which is used to achieve time alignment of the signals collected by different sensors; A spatial registration unit, which is used to uniformly map the collected multi-point data to a unified reference coordinate system; A multi-stage signal noise reduction processing unit, which is used to remove random interference in the collected signals and retain stable and effective dynamic features.

[0009] Preferably, the dynamic modeling module includes: A differential-algebraic equation modeling unit, which is used to construct a process model covering the coupling relationship of pressure, temperature, and speed variables; A structure analysis unit, which is used to process variable constraints and high-order differential relationships existing in the modeling process; A real-time calculation engine, which is used to quickly solve the modeling equations within the injection molding control cycle and output state estimation values.

[0010] Preferably, the stability analysis module includes: A time-delay modeling unit, which is used to identify the influence of non-ideal factors on the system dynamics, where the non-ideal factors include execution delay and transmission lag; A stability judgment unit, which is used to perform stability analysis based on the system model and design a control law that meets the convergence conditions; A control signal generation unit, which is used to convert the output of the control law into control instructions that can be executed by the injection molding device.

[0011] Preferably, the compensation decision module includes: A disturbance observation unit, which is used to identify external disturbances and modeling deviations during the injection molding process; A control instruction integration unit for fusing disturbance observation information with the control strategy output by stability analysis; An execution constraint management unit for performing amplitude limitation and rate limitation on control signals.

[0012] Preferably, the execution drive module includes: A servo driver for precisely completing small-displacement control operations; A hydraulic auxiliary control device for providing a strong driving response during the high-pressure injection stage; A feedback mechanism unit for real-time collecting the results of execution actions and feeding them back to the compensation decision module.

[0013] Preferably, the quality assessment module includes: A volume ratio measurement unit for evaluating whether the filling of the molded part is complete; A geometric error analysis unit for detecting precision deviations in the dimensions of the molded part; A non-destructive imaging detection unit for identifying surface flaws and internal defects.

[0014] Preferably, the parameter optimization module includes: An incremental learning unit for dynamically adjusting modeling parameters based on real-time quality assessment results; A feasible region update unit for expanding or contracting the parameter space according to control feedback; A knowledge base module for recording and maintaining the mapping relationship between process parameters and quality output to support subsequent control optimization.

[0015] Preferably, the parameter optimization module further includes: A monitoring mechanism unit for periodically evaluating the system control performance and the trend of molding quality; A parameter calibration unit for triggering the recalibration of key modeling parameters when a working condition deviation is detected; A parameter update and feedback mechanism for automatically transmitting the calibrated parameters to the dynamic modeling module.

[0016] The present invention provides an injection molding process parameter adaptive compensation system, which has the following beneficial effects: 1. Through the dual-drive mechanism based on incremental learning and feasible region update, the present invention establishes a system for autonomous accumulation of process knowledge. Compared with the fixed-parameter control scheme, it can effectively cope with time-varying working conditions such as material property drift and reduce the frequency of manual intervention.

[0017] 2. Through the extended state observer and instruction fusion technology, the present invention realizes real-time dynamic compensation for modeling deviation and external disturbance. Compared with the traditional feed-forward compensation scheme, it can effectively solve the problem of tracking lag of time-varying disturbance and significantly suppress the overshoot phenomenon.

[0018] 3. The present invention adopts a servo-hydraulic composite drive architecture and combines rate constraint management technology to break through the contradiction between precise positioning and strong driving in the existing single actuator solution, and takes into account micron-level positioning accuracy and megapascal-level high-pressure response.

[0019] 4. The present invention constructs a quality evaluation system that combines flow integration, optical scanning, and non-destructive imaging. Compared with single detection means, it synchronously solves the problem of collaborative recognition of surface defects, internal flaws, and dimensional deviations, and forms a closed-loop monitoring of the manufacturing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the process framework diagram of the system of the present invention; Figure 2 is the framework diagram of the multi-source sensing module of the present invention; Figure 3 is the framework diagram of the data fusion module of the present invention; Figure 4 is the framework diagram of the dynamic modeling module of the present invention; Figure 5 is the framework diagram of the stability analysis module of the present invention; Figure 6 is the framework diagram of the compensation decision module of the present invention; Figure 7 is the framework diagram of the execution drive module of the present invention; Figure 8 is the framework diagram of the quality evaluation module of the present invention; Figure 9 is the framework diagram of the parameter optimization module of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Please refer to the attached Figure 1 , the embodiment of the present invention provides an injection molding process parameter adaptive compensation system, including: Please refer to the attached Figure 2 , a multi-source sensing module for real-time collecting cavity pressure, melt temperature, and screw displacement information during the injection molding process; The multi-source sensing module includes a cavity pressure sensing part, a melt temperature measuring part, and a screw displacement monitoring part. The cavity pressure sensing part consists of 8 piezoelectric pressure sensors arranged in a circular array, which are exemplarily equidistantly arranged circumferentially at the end of the cavity flow channel, and the sensor spacing is 1 / 8 of the cavity diameter. The output signals of each sensor are transmitted to the data fusion module through shielded cables, and its spatial layout satisfies the following relationship: ; where, represents the polar angle coordinate of the th sensor in the cross-section of the cavity. This layout method can realize 360° coverage monitoring of the cavity pressure field.

[0023] The melt temperature measuring part uses a dual-band infrared temperature measuring device, preferably including two detection channels of 1.55μm and 1.65μm. The optical probe is installed at a 45° angle with the melt flow channel through a quartz observation window, and its radiation received energy satisfies: ; where, is the detection energy, is the emissivity, is the Planck constant, is the speed of light, is the Boltzmann constant, , are the radiation energies detected at wavelengths of and respectively, is the emissivity of the melt at wavelength and temperature . By calculating the dual-band energy ratio, the influence of emissivity fluctuation can be eliminated, and the temperature measurement accuracy of ±0.5°C can be achieved.

[0024] The screw displacement monitoring part uses a magnetostrictive displacement sensor, whose waveguide rod is installed parallel to the piston rod of the injection cylinder, and the distance between the measuring head and the magnetic ring is kept at 3 - 5mm. The displacement L and the detection pulse time delay satisfy a linear relationship: ; where, is the propagation speed of the magnetostrictive wave in the waveguide rod (typical value 2800m / s), is the screw displacement. This sensor can capture the micron-level displacement change of the screw at a sampling frequency of 10kHz.

[0025] Each sensing unit achieves time synchronization through the PTP protocol. Exemplarily, the IEEE 1588-2008 standard is adopted. The master clock periodically sends synchronization messages, and the slave clock achieves a synchronization accuracy at the microsecond level through offset compensation calculation. Spatial registration is completed by establishing a coordinate transformation matrix from the positions of each pressure sensor to the center of the mold cavity. Further, the Gaussian weighting algorithm is used to calculate the equivalent center pressure value: ; where, is the position vector of the th sensor, takes 1 / 3 of the mold cavity radius, is the equivalent mold cavity center pressure, is the weight coefficient of the th sensor, is the th sensor, and

[0026] is the pressure value measured by the

[0027] th sensor. This processing can effectively suppress the interference of local pressure fluctuations on the overall state judgment.

[0028] The signal transmission adopts a dual-redundant CAN bus architecture, and each sensor node is equipped with an independent power isolation module. Exemplarily, the pressure signal sampling rate is 20 kHz, the temperature signal is 1 kHz, and the displacement signal is 10 kHz. The time alignment of multi-rate data is achieved through the buffer queue of the data fusion module.

[0029] Please refer to the attached Figure 3 , the data fusion module is used to perform time sequence synchronization, spatial registration, and noise suppression on multi-source sensing data; The data fusion module includes a clock synchronization unit, a spatial registration unit, and a signal noise reduction unit. The clock synchronization unit constructs a master-slave network architecture using the IEEE 1588-2008 Precision Time Protocol. The master clock node periodically sends Sync messages and Follow_Up messages, and the slave node calculates the clock offset through the following formula: ; where, is the host sending time, is the slave receiving time, is the slave response time, is the master reception time, is the slave local time, is the actual transmission time of the Sync message recorded by the master clock in the Follow_Up message. The offset compensation is implemented by a phase-locked loop algorithm. Exemplarily, a second-order loop filter is adopted, and its transfer function is: ; In the formula, is the damping ratio, taking 0.707, is the natural frequency, taking 1.5 rad / s, which can achieve a synchronization accuracy at the microsecond level.

[0030] The spatial registration unit establishes a three-dimensional coordinate system with the center of the mold cavity as the origin, and the position coordinates of each pressure sensor are obtained through conversion of pre-calibrated parameters. For the measurement value of the th sensor, its spatial weight calculation follows the Gaussian distribution formula: ; In the formula, is the sensor position vector, takes the same value as the pressure sensor layout parameters. The calculated equivalent pressure value after registration is: ; In the formula, is the equivalent mold cavity pressure after spatial registration.

[0031] The signal noise reduction unit implements a three-level processing process: First, wavelet threshold denoising is used to eliminate high-frequency noise. The sym8 wavelet basis is selected for 5-layer decomposition, and the threshold setting adopts: ; In the formula, is the threshold of the th layer, is the standard deviation of the noise estimation of this layer, is the signal length. Subsequently, moving average filtering is used to suppress periodic interference, and the window width is 10 sampling points. Finally, Kalman filtering is used for dynamic noise suppression, and the state equation is defined as: ; ; In the formula, is the state transition matrix, is the control matrix, is the observation matrix, and are the process noise and observation noise, is The state vector at a moment, is the control input vector at a moment, is the process noise, and the observation vector at a moment, is the observation noise. Exemplarily, set Q = 0.01I and R = 0.1I for the pressure signal to achieve a signal-to-noise ratio improvement of ≥15 dB.

[0032] Each processing unit realizes data interaction through a dual-port RAM. The clock synchronization unit outputs data packets with timestamps and stores them in a circular buffer. The spatial registration unit extracts data from the buffer for coordinate transformation, and the noise reduction unit calls different processing algorithms according to the signal type. The data flow control adopts a priority scheduling mechanism to ensure that the processing delay of the pressure signal is ≤50 μs and that of the temperature signal is ≤1 ms.

[0033] The calibration process includes spatial coordinate calibration: measuring the actual positions of each sensor with a laser tracker, and correcting the deviation from the theoretical coordinates through a least squares fitting correction matrix. The time synchronization accuracy verification uses an oscilloscope to compare the trigger signal with the timestamp of the synchronized data, and the error is controlled within ±0.5 μs.

[0034] This embodiment can achieve the spatio-temporal consistency alignment of multi-source data through a hierarchical processing architecture; the wavelet-Kalman composite noise reduction algorithm can effectively separate noise from the effective signal; the pre-calibration mechanism ensures the accuracy of spatial registration.

[0035] Please refer to Appendix Figure 4 , a dynamic modeling module, which is used to construct a dynamic model of the injection molding process state based on the fused process data and output a state estimate value; The dynamic modeling module includes a differential-algebraic modeling unit, a structural analysis unit, and a real-time solution unit. The differential-algebraic modeling unit constructs a coupled equation including pressure , temperature , flow rate . Exemplarily, the modified Hele-Shaw model is used to describe the melt flow: ; In the formula, is the flow conductivity tensor, is the thermal expansion coefficient, represents the viscosity change term related to the shear rate, represents the partial derivative operator with respect to time , is the partial differential symbol. The structural analysis unit processes the constraint relationship between velocity and pressure in the equation, and introduces the Lagrange multiplier to construct an augmented equation: ; In the formula, is the melt density, is the viscous stress tensor, is the filling front constraint function, is the material derivative, is the Lagrange multiplier. The real-time solution part adopts the variable step-size Rosenbrock algorithm to convert the differential-algebraic equations into a stiff ODE system: ; In the formula, is the mass matrix, is the state vector, is the non-linear function vector, which is obtained by discretizing the differential-algebraic equations. The Jacobian matrix is generated by automatic differentiation technology and is implemented by using the ADOL-C library exemplarily.

[0036] The spatial discretization uses an unstructured finite volume grid, and the time step is dynamically adjusted according to the Courant number: ; In the formula, is the characteristic length of the th grid, is the flow velocity in the th grid cell, is the total number of grid cells. The solver realizes MPI parallel computing through the PETSc library, and completes 20 million degrees of freedom calculation within each control cycle, with a delay ≤ 2 ms.

[0037] Online update mechanism of model parameters: When the parameter optimization module issues new parameters, update , and other coefficients through the shared memory interface. Exemplarily, the viscosity model adopts the Cross-WLF formula: , where is calculated by the Arrhenius equation from the temperature , is the shear rate, and are the optimized parameters.

[0038] The verification process includes a static pressure test: Inject a standard fluid with a known viscosity into the mold cavity, compare the simulated pressure distribution with the measured data of the sensor, and control the error within ±3%. The dynamic response test verifies that the tracking delay of the state estimation value ≤ 5 ms by step-changing the injection speed.

[0039] In this embodiment, a multi-physics field coupling model is constructed through differential algebraic equations, which can accurately describe the dynamic characteristics of the injection molding process; the rigid ODE solving algorithm ensures the real-time requirement; and the parameter online update mechanism realizes the adaptive adjustment of the model.

[0040] Please refer to the appendix Figure 5 , a stability analysis module, which is used to perform system stability analysis based on the dynamic model and generate a control compensation strategy; The stability analysis module includes a time-delay compensation part, a stability discrimination part and a control generation part. The time-delay compensation part constructs an augmented state model including execution delay and sensing delay . Exemplarily, a third-order Pade approximation is used to express the time-delay link: ; where is the total delay time, is the Laplace variable. The augmented system state equation is extended to: ; ; where includes the original state and the time-delay state . The matrix dimension is determined according to the augmentation rule, is the augmented state vector, is the augmented system matrix, is the augmented input matrix, is the augmented output matrix.

[0041] The stability discrimination part adopts the Lyapunov-Krasovskii functional method to construct a functional in the following form: ; where , are positive definite matrices, is the value of the augmented state at time . By solving the linear matrix inequality: ; the maximum allowable time delay that makes the system stable is obtained, where is the time-delay correlation matrix. Exemplarily, the YALMIP toolbox is used for solving, and the convergence tolerance is set to 1e-6.

[0042] The control generation part designs a predictive compensation controller, and its control law is: ; where, is the feedback gain matrix, is the state prediction value, is the state prediction value at the future time instant. The predictor adopts the Adams - Bashforth two - step method: ; wherein, is the solution step size of the dynamic modeling module, is the system dynamic equation, coming from the of the dynamic modeling module. The gain matrix K is calculated by the pole placement method. Exemplarily, the closed - loop poles are placed in the region.

[0043] The signal conversion unit maps the control quantity to the injection pressure set value and the screw speed , and the conversion relationship satisfies: ; In the formula, , are calibration coefficients, sat is the limiting function, , are the components of the control quantity , corresponding to the pressure and speed commands respectively, is the speed control proportional coefficient. The execution constraint management sets , according to the physical limits of the equipment.

[0044] The online verification process includes a step - response test: when the set value suddenly changes by 20%, the overshoot of the actual pressure and speed is monitored to be ≤5%, and the adjustment time is ≤0.5 s. The stability boundary verification is carried out by gradually increasing the time delay until the system diverges, and the deviation between the measured and the theoretical value is ≤8%.

[0045] This embodiment can effectively overcome the influence of execution delay through the time - delay compensation model; the Lyapunov - Krasovskii method ensures the strictness of the stability criterion; the predictive control algorithm realizes the precise compensation of the time - delay system.

[0046] Please refer to Appendix Figure 6 , the compensation decision module, which is used to integrate the control compensation strategy and the disturbance estimation information and output the injection molding machine control instruction; The compensation decision module includes a disturbance observation part, an instruction fusion part and a constraint management part. The disturbance observation part uses an extended state observer (ESO) to estimate the total disturbance , and its dynamic equation is constructed as: ; wherein is the state estimate, is the disturbance estimate , is the observer gain coefficient, is the observation bandwidth, is the control gain, associated with the matrix of the stability analysis module.

[0047] The instruction fusion unit superimposes the disturbance compensation amount on the control instruction to generate a composite control amount: ; wherein, is the control law output from the stability analysis module. Exemplarily, a feedforward-feedback composite architecture is adopted, and the feedforward term is calculated based on the inverse model of the dynamic modeling module: ; wherein, is the pseudo-inverse of the input matrix, is the reference trajectory, is the desired acceleration.

[0048] The constraint management unit implements a dual limit strategy. The amplitude limit adopts a piecewise saturation function: ; wherein, is the lower limit of the control pressure, is the upper limit of the control pressure.

[0049] The rate limit is achieved through differential algebraic constraints: ; wherein, is the difference in the change of the control quantity, is the maximum allowable change rate, is the control period.

[0050] Exemplarily, a rate-limited integrator is adopted, and its discrete form is: ; wherein is the control period, which is consistent with the step size of the real-time calculation engine.

[0051] The data interface connects each module through a DMA channel. The disturbance observation unit obtains from the data fusion module, the instruction fusion unit receives the of the stability analysis module, and the constraint management unit outputs the final to the injection molding machine PLC. The communication protocol adopts EtherCAT, and the cycle jitter ≤ 10 μs.

[0052] The calibration process includes a disturbance rejection test: applying a 20% step pressure disturbance, with the observer response time ≤ 0.1 s and the steady-state compensation error ≤ 2%. The constraint effectiveness verification is carried out through over-limit command injection, and the measured output is strictly limited within to the range.

[0053] In this embodiment, the extended state observer can estimate and compensate the unmodeled disturbance in real time; the feedforward-feedback composite architecture improves the tracking accuracy; and the dual constraint mechanism ensures the safety of the control command.

[0054] Please refer to the appendix Figure 7 , an execution drive module, configured to drive the injection molding equipment to perform corresponding actions according to the control command; The execution drive module includes a servo drive unit, a hydraulic enhancement unit, and a feedback acquisition unit. The servo drive unit adopts a three-loop control architecture, and the position loop control law is: ; where is the position error, , , are the PID parameters, which are compatible with the rate limit parameters of the compensation decision module, is the control voltage of the servo motor. The speed loop bandwidth is set to 200 Hz, and the current loop response time ≤ 50 μs.

[0055] The hydraulic enhancement unit is activated when the injection pressure > 100 MPa, and the pressure-flow equation is constructed as: ; where is the effective area of the piston, is the screw speed, is the volume of the oil chamber, is the elastic modulus of the oil, is the hydraulic oil flow rate, is the pressure change rate. The proportional valve control algorithm adopts: ; where , , the output current is limited within the range of 4 - 20 mA, is the proportional valve control current, is the pressure set value, is the actual pressure.

[0056] The feedback acquisition unit reads the position data of the grating scale through the SSI interface, with a resolution of 0.1μm and a sampling rate of 1kHz. The pressure signal is converted by a 4-20mA transmitter, and the AD sampling accuracy is 16bit. The data timestamp is aligned with the clock synchronization unit of the data fusion module, and the deviation ≤ 1μs.

[0057] The electromechanical interface adopts a modular design: the servo motor is directly connected to the ball screw, with a transmission ratio of 5:1 and a screw stroke of 250mm. The hydraulic system includes an accumulator and a pressure relay. When it exceeds 150MPa, the emergency unloading valve is triggered, and the response time ≤ 2ms.

[0058] The verification test includes a positioning accuracy test: for a 100mm stroke command, the measured repeat positioning accuracy is ±0.005mm. The pressure tracking test: at a set value of 120MPa, the steady-state fluctuation ≤ ±0.5MPa. The step response test shows that the acceleration time for 90% of the stroke ≤ 80ms, and the overshoot ≤ 1%.

[0059] This embodiment can achieve precise position control through a three-loop control architecture; the hydraulic pressure equation accurately describes the dynamic characteristics of the system; the high-precision feedback mechanism ensures the controllability of the execution results.

[0060] Please refer to the attached Figure 8 , a quality evaluation module, which is used to evaluate the molding quality based on the feedback signal during the molding process and output an evaluation result, and the evaluation result is used to determine whether the process meets the set quality requirements; The quality evaluation module includes a volume ratio calculation unit, a morphology analysis unit, and a defect detection unit. The volume ratio calculation unit calculates the actual filling volume through the integral flow rate during the injection stage: ; where is the flowmeter data from the execution drive module, is the filling time, is synchronized with the control period, is the actual injection volume. The theoretical volume of the mold cavity , where is the cross-sectional area of the mold cavity, is the runner length.

[0061] The morphology analysis unit adopts a laser scanning point cloud registration algorithm and defines a geometric error index: ; where is the measured point cloud, is the CAD model point cloud, is the rigid body transformation matrix, is the average shape error. The optimal transformation is solved by the ICP algorithm, and the convergence threshold is set to 0.01 mm.

[0062] The defect detection unit adopts the X-ray attenuation model: ; In the formula, is the linear attenuation coefficient of the material, is the product thickness, is the ray intensity. The defect determination condition is: ; Among them, is the system noise level, and the reference image comes from the standard qualified part database, is the measured gray value.

[0063] The data fusion interface receives from: 1) at the injection end point, 2) the point cloud data of the optical measuring machine, and 3) the gray image of the X-ray machine. The evaluation result is output as a three-dimensional quality matrix: ; In the formula, , is the maximum shape error, is the number of defect pixels.

[0064] The on-line test process includes a short-shot experiment: when is satisfied, an underfill alarm is triggered. The verification of the standard part shows that the geometric error detection accuracy reaches ±0.02 mm, and the porosity defect identification sensitivity is Φ0.3 mm.

[0065] This embodiment realizes the quantitative evaluation of filling integrity through flow integration; the point cloud registration algorithm can accurately detect dimensional deviations; the X-ray attenuation model can effectively identify internal defects.

[0066] Please refer to Appendix Figure 9 , a parameter optimization module, which is used to update the model parameters based on the quality evaluation result to achieve the adaptive adjustment of the system.

[0067] The parameter optimization module includes an incremental learning unit, a feasible region management unit, and a knowledge base unit. The incremental learning unit updates the melt viscosity parameter of the dynamic modeling module by using the recursive least squares method: ; ; ; In the formula, is the viscosity model parameter, is the shear rate and temperature, is the forgetting factor, is the covariance matrix, is the actual viscosity measurement value (unit: Pa·s), from the on-line rheometer.

[0068] The feasible region management department updates the parameter constraints through the ellipsoid algorithm: ; where, is the parameter covariance matrix, corresponds to 95% confidence level, is the current parameter estimate value. When the parameter touches the boundary, the expansion coefficient amplifies the ellipsoid: ; The knowledge base department constructs the process parameter-quality mapping relationship: ; In the formula, is the historical process parameter vector, is the corresponding quality evaluation result.

[0069] Adopt index to implement the nearest neighbor query, and the similarity threshold is set to .

[0070] The monitoring mechanism department calculates the sliding variance of the quality index: ; where, is the mean value of the quality index within the window, benchmark variance; When , calibration is triggered, and the window width is cycles.

[0071] The parameter calibration department uses the trust region algorithm to solve: s.t. ; where, is the output prediction of the dynamic modeling module, is the current feasible region constraint; The Jacobian matrix is calculated by the finite difference method, and the step size is . The feedback mechanism writes the updated θ into the dynamic modeling module through the OPCUA protocol, and stores it with the version number incremented.

[0072] The verification case shows that when the viscosity deviation is 15% due to the change of the material batch, the system completes the parameter correction within 10 cycles, and the fluctuation of the holding pressure decreases from ±8% to ±2%. After the knowledge base accumulates 1000 groups of data, the parameter initialization time is shortened by 60%.

[0073] In this embodiment, the recursive least squares method can be used to achieve online update of model parameters; the ellipsoid constraint ensures the safety of parameter adjustment; and the process knowledge base improves the optimization efficiency.

[0074] Working principle: During the injection molding process, the system first relies on the multi-source sensing module to collect data. This module consists of multiple sensing units, including piezoelectric pressure sensors arranged in the mold cavity, infrared temperature sensing units for non-contact measurement of the melt temperature, and high-resolution displacement sensors for measuring the position and speed changes of the screw. Through these sensors, the system can monitor the multi-point pressure changes in the mold cavity, the temperature distribution of the melt, and the dynamic behavior of the screw in real time.

[0075] These collected raw signals will be input into the data fusion module. In this module, through the high-precision clock synchronization unit, the system accurately aligns the signals from different sensors along the time axis to ensure the temporal consistency of each data item. At the same time, the spatial registration unit performs unified coordinate transformation on the measurement data from different positions to ensure that the data can be processed under the same reference framework. To improve the accuracy and usability of the data, the system also removes sensor noise and external interference through the multi-stage signal denoising processing unit, retaining the effective dynamic feature data.

[0076] Next, the fused data is transmitted to the dynamic modeling module. The differential algebraic equation modeling unit in it uses a mathematical model to model the coupling relationship between variables such as mold cavity pressure, melt temperature, and screw displacement, constructing a multi-dimensional and dynamic injection molding process model. This model can not only reflect the current process state of the system but also predict future dynamic behaviors. Since the variables in the injection molding process are often affected by physical constraints and higher-order differential relationships, the structural analysis unit will consider these factors during modeling to ensure the accuracy and rationality of the model. Finally, the system quickly solves these equations through a real-time calculation engine to obtain the state estimation values of the current injection molding process, such as the mold cavity filling progress and the melt flow rate.

[0077] Once an accurate state estimation is obtained, the stability analysis module begins to play its role. This module first evaluates the impact of non-ideal factors such as time-delay effects and controller response delays in the injection molding process on the system. Through the analysis of the system stability boundary, the stability judgment unit can determine whether the current control strategy can still ensure the stable operation of the system and design a set of adaptive compensation control strategies to cope with the disturbances that the system may encounter. The generated control signals will pass through the control signal generation unit and be converted into specific injection molding equipment execution instructions.

[0078] These control instructions are then transmitted to the compensation decision-making module, where the system further processes them based on real-time process status and disturbance estimation information. The disturbance observation unit monitors external disturbances (such as material fluctuations, environmental changes) and any model deviations that may affect the process quality, thereby identifying and quantifying the disturbance factors. The control instruction integration unit fuses this disturbance information with the previously generated compensation control strategy to form the final compensation decision. To ensure the safety and reliability of the control instructions in practical applications, the execution constraint management unit appropriately limits the amplitude and rate of the control signal to avoid over-control or the instruction exceeding the physical execution range of the device.

[0079] The control instructions output by the compensation decision-making module are finally received and executed by the execution drive module. This module includes a servo drive and a hydraulic auxiliary control device. The servo drive is used to precisely control small displacement adjustments, while the hydraulic device provides a strong drive response during the high-pressure injection stage to ensure that the power requirements during the injection molding process are met. At the same time, the feedback mechanism unit is responsible for collecting the execution results in real time and feeding this data back to the compensation decision-making module, thereby achieving closed-loop control. In this way, the system can continuously adjust the compensation strategy according to the actual execution effect to ensure precise control of the injection molding process.

[0080] After each round of injection molding, the system evaluates the quality of the molded part. The quality evaluation module determines the filling integrity of the molded part through the volume ratio measurement unit, checks the dimensional accuracy of the molded part using the geometric error analysis unit, and identifies surface defects and internal defects through the non-destructive imaging detection unit. The quality evaluation results are not only used to determine whether the finished products of this batch meet the quality standards but also transmitted as feedback information to the parameter optimization module. The parameter optimization module dynamically adjusts the modeling parameters of the system based on this feedback information. The incremental learning unit gradually optimizes the modeling process according to the quality evaluation results, while the feasible region update unit expands or contracts the parameter space through control feedback to improve the response ability of the control system. The knowledge base module records and maintains the relationship between historical parameters and quality outputs, continuously accumulating knowledge of process control to support subsequent optimization and adjustment.

[0081] When the system detects a deviation in the working conditions through the monitoring mechanism unit, the parameter optimization module will be automatically activated, triggering the parameter calibration unit to recalibrate the key parameters. The calibrated parameters will be transmitted to the dynamic modeling module through the parameter update feedback mechanism to ensure the system's adaptive adjustment ability and continuous optimization.

[0082] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An injection molding process parameter adaptive compensation system, characterized in that, It includes: A multi-source sensing module for real-time acquisition of cavity pressure, melt temperature, and screw displacement information during the injection molding process; A data fusion module for time-series synchronization, spatial registration, and noise suppression of multi-source sensing data; A dynamic modeling module for constructing a dynamic model of the injection molding process state based on the fused process data and outputting state estimation values; A stability analysis module for performing system stability analysis based on the dynamic model and generating a control compensation strategy; A compensation decision module for comprehensively integrating the control compensation strategy and disturbance estimation information and outputting an injection molding machine control instruction; An execution drive module for driving the injection molding equipment to perform corresponding actions according to the control instruction; A quality assessment module for evaluating the molding quality based on the feedback signal of the molding process and outputting an assessment result, which is used to determine whether the process meets the set quality requirements; A parameter optimization module for updating the model parameters based on the quality assessment result to achieve system adaptive adjustment.

2. The adaptive compensation system for injection molding process parameters according to claim 1, wherein The multi-source sensing module includes: Multiple piezoelectric pressure sensors disposed in the cavity for obtaining multi-point pressure change data in the cavity; A temperature sensing unit that uses the infrared temperature measurement principle for non-contact measurement of the melt temperature; A high-resolution displacement sensor for measuring the position and speed changes of the screw during the injection process to comprehensively sense the dynamic parameters of the injection molding process.

3. An injection molding process parameter adaptive compensation system according to claim 1, characterized in that The data fusion module includes: A high-precision clock synchronization unit for achieving time alignment of the signals collected by different sensors; A spatial registration unit for uniformly mapping the collected multi-point data to a unified reference coordinate system; A multi-stage signal noise reduction processing unit for removing random interference in the collected signals and retaining stable and effective dynamic features.

4. An injection molding process parameter adaptive compensation system according to claim 1, characterized in that The dynamic modeling module includes: A differential-algebraic equation modeling unit for constructing a process model covering the coupling relationship of pressure, temperature, and speed variables; A structure analysis unit for dealing with variable constraints and high-order differential relationships existing in the modeling process; A real-time calculation engine for quickly solving the modeling equations within the injection molding control cycle and outputting state estimation values.

5. An adaptive compensation system for injection molding process parameters according to claim 1, characterized in that The stability analysis module includes: A time-delay modeling unit for identifying the impact of non-ideal factors on the system dynamics, where the non-ideal factors include execution delay and transmission lag; A stability judgment unit for performing stability analysis based on the system model and designing a control law that meets the convergence conditions; A control signal generation unit for converting the output of the control law into a control instruction executable by the injection molding equipment.

6. The adaptive compensation system for injection molding process parameters according to claim 1, wherein The compensation decision module includes: A disturbance observation unit for identifying external disturbances and modeling deviations during the injection molding process; A control instruction integration unit for fusing the disturbance observation information and the control strategy output by the stability analysis; An execution constraint management unit for performing amplitude limitation and rate limitation on the control signal.

7. An injection molding process parameter adaptive compensation system according to claim 1, characterized in that The execution drive module includes: A servo driver for precisely completing small-displacement control operations; A hydraulic auxiliary control device for providing a strong drive response during the high-pressure injection stage; A feedback mechanism unit for real-time acquisition of the execution action result and feedback to the compensation decision module.

8. An injection molding process parameter adaptive compensation system according to claim 1, characterized in that, The quality assessment module includes: Volume ratio measurement unit for evaluating whether the filling of the formed part is complete; Geometric error analysis unit for detecting the precision deviation of the formed part in terms of dimensions; Non-destructive imaging detection unit for identifying surface defects and internal flaws.

9. The adaptive compensation system for injection molding process parameters according to claim 1, characterized in that, The parameter optimization module includes: Incremental learning unit for dynamically adjusting the modeling parameters based on the real-time quality assessment results; Feasible region update unit for expanding or contracting the parameter space according to the control feedback; Knowledge base module for recording and maintaining the mapping relationship between process parameters and quality output to support subsequent control optimization.

10. The adaptive compensation system for injection molding process parameters according to claim 1, wherein The parameter optimization module further includes: Monitoring mechanism unit for periodically evaluating the system control performance and the trend of the formed part quality; Parameter calibration unit for triggering the recalibration of key modeling parameters when the working condition deviation is detected; Parameter update and feedback mechanism for automatically transmitting the calibrated parameters to the dynamic modeling module.

Images