Integrated injection molding machine injection intelligent manufacturing equipment for shoes

Abstract

The invention relates to the technical field of intelligent manufacturing equipment, and discloses integrated shoe injection molding machine injection intelligent manufacturing equipment which comprises a rack, a hopper assembly, a spiral conveying assembly, a fixed mold and movable mold assembly and an intelligent control part. The control part integrates a raw material state sensing module, a plasticizing quality evaluation module, an injection process modeling module, a deviation recognition module and a dynamic regulation module to construct a whole-process closed-loop control system; through integration of intelligent sensing, evaluation, identification and regulation, the size precision, surface smoothness and batch consistency of injection molding products for shoes can be remarkably improved, and waste products and rework caused by process fluctuation are reduced; and meanwhile, the autonomous optimization capability of the system reduces the debugging time and the dependence on skilled operators, so that the utilization rate of equipment is improved, and the energy consumption and the comprehensive production cost are reduced.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing equipment technology, specifically to an integrated intelligent manufacturing equipment for shoe injection molding machines. Background Technology

[0002] Injection molding is one of the key processes in footwear manufacturing. Its process stability and product quality are directly related to the dimensional accuracy, surface appearance, and production efficiency of the products. Traditional shoe injection molding machines typically consist of a frame, hopper, screw plasticizing system, mold, and electrical control unit. Their control methods mostly rely on preset process parameters and are adjusted in stages by human experience during production.

[0003] However, in actual production, the state of raw materials fluctuates from batch to batch, and the uniformity of the melt during plasticization is difficult to quantify and assess in real time. This leads to poor matching between the process parameters and the actual material state during the injection stage, easily causing problems such as incomplete filling, uneven shrinkage, and surface defects. In addition, existing control systems often lack the ability to integrate and analyze multi-source process data and provide closed-loop feedback, relying on operator experience for parameter correction. This results in a delayed response and limited adjustment accuracy, making it difficult to achieve continuous optimization and stable control of the production process.

[0004] In recent years, although some studies have attempted to introduce sensor monitoring or adopt rule-based control strategies in the injection molding process, the following limitations still exist:

[0005] Incomplete perception of raw material condition: Most systems only monitor conventional physical quantities such as temperature and pressure, lacking multi-dimensional real-time perception of raw material particle size, humidity, chemical composition and flow characteristics, making it difficult to comprehensively assess the impact of batch differences in raw materials on plasticizing quality.

[0006] The quality of plasticization lacks quantitative evaluation: In existing technologies, the quality of the plasticization process largely depends on human experience or post-process product inspection, lacking online and quantitative evaluation indicators for plasticization uniformity, which cannot provide timely and accurate basis for process adjustment.

[0007] Process adjustments rely on static models and human intervention: Traditional control methods are mostly based on static process models or fixed rules, which are difficult to adapt to changes in raw material conditions, environmental fluctuations and other factors; the adjustment process relies on human experience, has a slow response speed and is easily affected by subjective factors.

[0008] The system lacks full-process closed-loop intelligent collaboration: the control links of existing injection molding machines are relatively independent, and the data connection and intelligent closed-loop control of the entire process from raw material input to molding output have failed to be realized. As a result, process optimization is limited to local links, and the overall production stability and adaptability are insufficient.

[0009] Therefore, developing an integrated intelligent injection molding equipment that can achieve comprehensive perception of raw material status, online evaluation of plasticizing quality, intelligent identification of process deviations, and dynamic adjustment of key parameters based on real-time data is of great significance for improving the molding quality of footwear products, reducing scrap rates, and reducing reliance on skilled operators. It is also an urgent need for the current injection molding manufacturing industry to develop towards intelligence and precision. Summary of the Invention

[0010] To address the shortcomings of existing technologies, this invention provides an integrated intelligent manufacturing equipment for footwear injection molding machines. By integrating multi-dimensional raw material perception, quantitative evaluation of plasticization quality, and dynamic control based on reinforcement learning, a closed-loop intelligent control system is constructed for the entire process. This solves the problems of unstable product quality and high scrap rate caused by raw material fluctuations and reliance on manual experience in traditional footwear injection molding production.

[0011] To achieve the above objectives, the present invention provides the following technical solution: an integrated intelligent injection molding machine for footwear, comprising a frame; a hopper assembly installed on the top of the frame, which integrates a raw material crushing and pretreatment unit; a screw conveyor assembly fixedly installed in the middle of the frame, its inlet being sealed to the outlet of the hopper assembly via a flange; a fixed mold assembly and a moving mold assembly, respectively bolted to a fixed template and a movable template on the front side of the frame, forming a molding cavity; the outlet of the screw conveyor assembly being connected to the molding cavity via a hot runner system; and a control unit configured to perform intelligent closed-loop control of the entire process from raw material pretreatment and plasticizing to injection molding.

[0012] The control unit includes a raw material status sensing module, a plasticizing quality assessment module, an injection process modeling module, a deviation identification module, a dynamic control module, and a human-machine interaction terminal. The output of the raw material status sensing module is electrically connected to the input of the plasticizing quality assessment module, the output of the plasticizing quality assessment module is electrically connected to the input of the injection process modeling module, the output of the injection process modeling module is electrically connected to the input of the deviation identification module, the output of the deviation identification module is electrically connected to the input of the dynamic control module, and the output of the dynamic control module is electrically connected to the input of the human-machine interaction terminal.

[0013] Preferably, the raw material state sensing module includes a particle size image sensor, a humidity detection probe, an infrared spectrometer, and a vibration spectrum collector. The particle size image sensor is installed on the upper part of the inner wall of the hopper assembly, the humidity detection probe is embedded in the inner wall of the bottom discharge channel of the hopper assembly, the detection window of the infrared spectrometer faces the feeding area in the middle of the hopper assembly, and the vibration spectrum collector is fixed to the outer wall of the hopper assembly.

[0014] Preferably, the particle size image sensor acquires a projected image of the raw material particles every 0.5 seconds, and obtains the equivalent diameter distribution vector of the particles through an edge extraction algorithm; the humidity detection probe outputs a relative humidity value sequence at a frequency of 1Hz; and the infrared spectrometer operates at 1200-1800cm². -1 The wavenumber range is scanned to generate an absorption peak intensity vector; the vibration spectrum acquisition device has a sampling frequency of 5kHz, and outputs the main frequency energy ratio after fast Fourier transform.

[0015] Preferably, the plasticizing quality assessment module calculates the plasticizing uniformity index based on the multidimensional characteristics of the raw materials, and the calculation process includes:

[0016] Data alignment and time synchronization: The particle size distribution vector, humidity sequence, infrared absorption vector and vibration dominant frequency energy are aligned according to timestamps to form a four-dimensional synchronous observation matrix;

[0017] Feature normalization: Z-score standardization is applied to the data in each dimension.

[0018] Principal component dimensionality reduction: Perform principal component analysis on the standardized observation matrix, retain the first two principal components with a cumulative contribution rate ≥90%, and form two-dimensional plasticized input features;

[0019] Plasticization uniformity index calculation: Input the two-dimensional features into the pre-trained Gaussian process regression model and output the plasticization uniformity index U, U∈[0,1].

[0020] Preferably, the injection process modeling module is used to construct the theoretical response surface of the current batch injection process, and the calculation process includes:

[0021] Establish the process parameter vector P = (T1, T2, n, p) T Where T1 is the setpoint for the barrel front temperature, T2 is the setpoint for the nozzle temperature, n is the screw speed, and p is the back pressure value;

[0022] Based on the historical qualified product database, the radial basis function interpolation method is used to construct the mapping relationship F: P→U_theory;

[0023] By comparing the actual plasticizing uniformity index U with U_theory, the deviation vector ΔU = U-U_theory is defined.

[0024] If |ΔU|>0.05, the deviation identification module is triggered to enter the anomaly judgment process.

[0025] Preferably, the deviation identification module is used to quantify the degree of process deviation in the current injection cycle, and its execution logic includes:

[0026] A multi-source deviation feature set D = (ΔU, ΔT_melt, Δτ_fill, ΔV_shrink) is constructed, where ΔT_melt is the difference between the measured melt temperature and the set value, collected by a thermocouple installed at the outlet of the screw conveyor assembly; Δτ_fill is the difference between the actual mold filling time and the standard mold filling time, calculated by the displacement sensor of the moving mold assembly in combination with the mold opening and closing sequence; and ΔV_shrink is the deviation of the product volume shrinkage rate, estimated by the pressure-displacement combined sensor built into the fixed mold assembly.

[0027] Input D into a lightweight convolutional neural network, which contains two one-dimensional convolutional kernels with a kernel size of 3, 16 filters, and a fully connected layer, and outputs the process anomaly probability q∈[0,1].

[0028] When q > 0.7, it is determined to be a significant process deviation, and a set of control instructions is generated.

[0029] Preferably, the dynamic control module is used to generate and issue real-time adjustment instructions, and its control strategy includes:

[0030] Establish a control action space A = {ΔT1, ΔT2, Δn, Δp}, and limit the value range of each action variable to ±10% of the original set value;

[0031] The Q-learning algorithm is used to update the action value function Q(D,a) online, and the reward function r = -(|ΔU| + 0.3|ΔT_melt| + 0.2|Δτ_fill|).

[0032] Select the action 'a' that maximizes the Q value as the current control instruction;

[0033] The signal 'a' is decomposed into four independent control signals, which are sent to the barrel heating coil driver, the nozzle temperature control unit, the screw servo motor controller, and the back pressure proportional valve, respectively.

[0034] Preferably, the human-machine interface terminal is used to display the current plasticization uniformity index U, process abnormality probability q, control action a* and historical trend curve in real time, and supports operator manual intervention to control thresholds.

[0035] This invention provides an integrated intelligent injection molding machine for footwear. It offers the following advantages:

[0036] 1. This invention integrates five modules: raw material state sensing, plasticizing quality assessment, injection process modeling, deviation identification, and dynamic control, constructing a closed-loop control system for the entire process from raw material pretreatment to injection molding. The modules are interconnected via electrical signals, forming a complete data link of "sensing-assessment-modeling-identification-control," enabling real-time monitoring and autonomous adjustment of the injection process. This significantly reduces human intervention, ensures consistent production parameters for each batch, and thus improves the dimensional stability and surface quality of footwear products.

[0037] 2. This invention integrates a particle size image sensor, a humidity detection probe, an infrared spectrometer, and a vibration spectrum collector into the hopper assembly, enabling real-time acquisition of raw material state data from multiple dimensions, including physical morphology, moisture content, molecular structure, and flow characteristics. Through time alignment, normalization, and principal component analysis, key features are extracted and input into a Gaussian process regression model, outputting a plasticization uniformity index to achieve a quantitative evaluation of plasticization quality and provide a reliable basis for subsequent process adjustments.

[0038] 3. The injection process modeling module of the present invention constructs a mapping relationship between process parameters and theoretical plasticization uniformity based on historical qualified product data, and calculates the deviation between actual and theoretical values ​​in real time; the deviation identification module further integrates multi-source features such as melt temperature deviation, mold filling time deviation and volume shrinkage rate deviation, and outputs the process abnormality probability through a lightweight convolutional neural network. When q>0.7, it is judged as a significant process deviation, triggering the control mechanism to achieve early detection and rapid response to abnormal states.

[0039] 4. Through the integrated realization of the above-mentioned intelligent sensing, evaluation, identification and control, this invention can significantly improve the dimensional accuracy, surface smoothness and batch consistency of shoe injection molded products, and reduce scrap and rework caused by process fluctuations. At the same time, the system's autonomous optimization capability reduces debugging time and dependence on skilled operators, which is conducive to improving equipment utilization, reducing energy consumption and overall production costs. Attached Figure Description

[0040] Figure 1 This is a three-dimensional structural diagram of the present invention;

[0041] Figure 2 This is a rear-view stereoscopic structural diagram of the present invention;

[0042] Figure 3 This is a schematic diagram of the structure of the fixed mold and moving mold assembly of the present invention;

[0043] Figure 4 This is a structural block diagram of the control component of the present invention.

[0044] The components include: 1. Frame; 2. Hopper assembly; 3. Screw conveyor assembly; 4. Fixed mold assembly; 5. Moving mold assembly; and 6. Control components. Detailed Implementation

[0045] The technical solution of the present invention will now be clearly and completely described 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.

[0046] This invention provides an integrated intelligent manufacturing equipment for shoe injection molding machines, the overall structure of which is as follows: Figures 1-3 As shown, it mainly includes a frame 1, a hopper assembly 2, a screw conveyor assembly 3, a fixed mold assembly 4, a moving mold assembly 5, and a control component 6; each component is arranged in sequence according to the process flow in space, and achieves coordinated operation at the physical and signal levels through mechanical connections, sealed interfaces, and electrical circuits.

[0047] In this embodiment, it should be specifically explained that the frame 1 is the basic support structure of the entire equipment. The hopper assembly 2 is installed on the top of the frame 1 and integrates a raw material crushing and pretreatment unit inside. Its discharge port is connected to the feed port of the screw conveyor assembly 3 through a flange seal. The screw conveyor assembly 3 is fixedly installed in the middle of the frame 1. Its outer wall is provided with a partitioned heating ring. The discharge end is connected to the fixed mold assembly 4 through a hot runner system. The fixed mold assembly 4 and the moving mold assembly 5 are respectively fastened to the fixed template and the moving template on the front side of the frame 1 by bolts, which together form the forming mold cavity. The moving mold assembly 5 is driven by a drive mechanism to realize the mold opening and closing action.

[0048] Control component 6 is the core intelligent control unit of this invention. Its hardware mainly includes an industrial computer as the host computer, a programmable logic controller (PLC) as the slave computer, a high-speed multi-channel data acquisition card, and a human-machine interaction terminal. The software consists of a raw material status sensing module, a plasticizing quality assessment module, an injection process modeling module, a deviation identification module, and a dynamic control module. The data flow between the modules is shown in the attached figure. Figure 4 As shown.

[0049] This embodiment requires specific explanation regarding hardware configuration and communication:

[0050] The industrial computer carries the algorithm software of the aforementioned intelligent modules and is responsible for complex multi-source data fusion, model calculation and intelligent decision-making; it exchanges data with the PLC in real time and at high speed through industrial Ethernet such as EtherCAT or Profinet.

[0051] The PLC controller is responsible for receiving control instructions from the industrial computer and converting them into specific timing control signals to drive actuators such as heating coil drivers, screw servo motor controllers, and back pressure proportional valves; at the same time, the PLC collects the status feedback signals of each actuator in real time.

[0052] The data acquisition card is responsible for high-speed synchronous acquisition of analog and digital signals from various sensors in the hopper, screw, mold, etc. These sensors include image sensors, thermocouples, displacement sensors, etc.; the acquired signals are transmitted to an industrial computer for processing via PCIe or USB bus.

[0053] Human-machine interaction terminals are typically industrial touch screens that run monitoring software for parameter setting, status display, and manual intervention.

[0054] The control unit 6 is connected to the sensors and actuators in each component via shielded cables, forming a complete sensing and control network.

[0055] The raw material state sensing module is responsible for collecting multi-dimensional feature data of the initial raw materials, specifically:

[0056] Particle size image data: Projected images of raw material particles are captured at 0.5-second intervals using an image sensor installed on the upper inner wall of hopper assembly 2.

[0057] Humidity data: The moisture content on the surface of the raw material is collected at a frequency of 1Hz by a humidity detection probe embedded in the inner wall of the discharge channel at the bottom of the hopper assembly 2.

[0058] Molecular spectral data: The feeding area in the middle of the hopper was scanned using an infrared spectrometer to obtain data from 1200-1800 cm⁻¹. -1 Infrared absorption spectra within the wavenumber range generate molecular bond vibration characteristic vectors;

[0059] Vibration spectrum data: The vibration signal generated by the raw material falling and hitting the inner wall is recorded at a sampling frequency of 5kHz by a vibration spectrum acquisition device fixed to the outer wall of the hopper assembly 2, and the main frequency energy ratio is extracted by fast Fourier transform.

[0060] The above four types of timestamped data are synchronized by the raw material status sensing module to form a four-dimensional synchronous observation matrix;

[0061] Further explanation of the plasticizing quality assessment module: The plasticizing quality assessment module receives a four-dimensional synchronous observation matrix from the raw material state perception module. First, it performs Z-score standardization on the particle size distribution vector, humidity sequence, infrared absorption vector, and vibration dominant frequency energy to eliminate the dimensional differences of the data in each dimension. Then, it performs principal component analysis on the standardized matrix, retaining the first two principal components with a cumulative contribution rate of not less than 90%, forming a two-dimensional plasticizing input feature. This two-dimensional feature is input into a pre-trained Gaussian process regression model, and the model outputs the plasticizing uniformity index U of the current batch of raw materials, with the U value between 0 and 1.

[0062] In this embodiment, it should be specifically explained that the training dataset of the pre-trained Gaussian process regression model is obtained by using four-dimensional synchronous observation matrix data of a large number of different batches of raw materials collected under standard process parameters as input features, and using the "qualified" and "unqualified" labels (or quantified uniformity scores) determined by offline detection (including but not limited to melt flow rate test and product density measurement) after the end of the corresponding production cycle as output targets, and is trained through supervised learning.

[0063] Further explanation of the injection process modeling module: The injection process modeling module receives the plasticization uniformity index U and combines it with the currently set process parameter vector P = (T1, T2, n, p). T A comparative analysis is performed; where T1 is the barrel front-end temperature setpoint, T2 is the nozzle temperature setpoint, n is the screw speed, and p is the back pressure value. The injection process modeling module calls the mapping relationship between process parameters and plasticization uniformity index stored in the historical qualified product database, and constructs the theoretical response surface F: P→U_theory using the radial basis function interpolation method. Subsequently, the deviation ΔU between the actual plasticization uniformity index U and the theoretical value U_theory is calculated. If |ΔU| is greater than the preset threshold ε (ε=0.05), a trigger signal is sent to the deviation identification module.

[0064] In this embodiment, it should be specifically explained that the historical qualified product database is a complete set of data corresponding to multiple injection cycles in which the final products were judged to be qualified by quality inspection during equipment debugging and historical production stages. This includes the process parameter vector P of the cycle, the plasticization uniformity index U actually calculated, and the relevant sensor raw data. This database serves as the knowledge base for constructing theoretical response surfaces and optimization decisions, and can be continuously updated and expanded during system operation.

[0065] Further explanation of the deviation identification module: After receiving the trigger signal, the deviation identification module begins to collect a multi-source deviation feature set D = (ΔU, ΔT_melt, Δτ_fill, ΔV_shrink). ΔT_melt is obtained by real-time measurement of the melt temperature by a thermocouple installed at the outlet of the screw conveyor assembly 3 and compared with a set value. Δτ_fill is obtained by recording the time from the start of mold closing to complete closure via a displacement sensor on the moving mold assembly 5 and comparing it with the standard mold filling time. ΔV_shrink is obtained by collecting pressure changes and product shrinkage displacement within the mold cavity during the pressure holding stage using a pressure-displacement combined sensor built into the fixed mold assembly 4, and the volume shrinkage rate deviation is calculated by an algorithm. The above four-dimensional deviation features are input into a lightweight convolutional neural network, which contains two one-dimensional convolutional layers (with a kernel size of 3 and 16 filters) and a fully connected layer, ultimately outputting a process anomaly probability q. When q is greater than 0.7, a significant process deviation is determined in the current injection cycle, and a control command request is generated.

[0066] Further explanation of the dynamic control module: After receiving a control request, the dynamic control module initiates the Q-learning algorithm for online decision-making. The control action space A is defined as {ΔT1, ΔT2, Δn, Δp}, and the adjustment range of each action variable is limited to ±10% of the original set value; the reward function r is set as r = -(|ΔU| + 0.3|ΔT_melt| + 0.2|Δτ_fill|), which guides the algorithm to prioritize reducing the main deviation terms; the dynamic control module updates the action value function Q(D, a) based on the current deviation characteristics D and historical experience, and selects the action a* that maximizes the Q value as the optimal control strategy; subsequently, it decomposes it into four independent control signals: the ΔT1 signal is sent to the driver of the heating coil at the front end of the barrel to adjust the heating power; the ΔT2 signal is sent to the nozzle temperature control unit to adjust the nozzle temperature; the Δn signal is sent to the screw servo motor controller to change the screw speed; and the Δp signal is sent to the back pressure proportional valve to adjust the back pressure.

[0067] Further explanation of the human-machine interface terminal: The human-machine interface terminal displays the current plasticization uniformity index U, process abnormality probability q, selected control action a*, and historical trend curves of each parameter in real time; the operator can view the system operating status through the touch screen interface, and manually adjust the deviation identification threshold or intervene control strategy when necessary; the execution results of all control commands will be reflected in the next injection cycle, and the process parameters will be continuously optimized through closed-loop feedback.

[0068] In summary, the complete workflow of this embodiment is as follows:

[0069] In actual operation, the operator feeds the plastic raw material into hopper assembly 2, and the system operates according to the following closed-loop process:

[0070] S2 Raw Material Sensing and Assessment: The raw material state sensing module is activated, and it simultaneously collects particle size, humidity, spectrum, and vibration data and transmits them to the plasticizing quality assessment module. This module calculates and outputs the plasticizing uniformity index U.

[0071] S3 Modeling and Preliminary Deviation Judgment: The injection process modeling module receives U and compares it with the theoretical value U_theory corresponding to the current process parameter set P to calculate the deviation ΔU; if ΔU exceeds the threshold, deviation identification is triggered.

[0072] S4 Multi-Source Identification and Decision-Making: The deviation identification module collects real-time deviations such as melt temperature, filling time, and shrinkage rate, and calculates the anomaly probability q through a neural network; if q exceeds the threshold, dynamic adjustment is requested.

[0073] S5 Online Control and Execution: The dynamic control module uses reinforcement learning algorithms to calculate the optimal process parameter adjustment strategy a*, and sends the adjustment instructions to each actuator to optimize the next injection cycle;

[0074] S6 Human-Machine Interaction and Monitoring: The human-machine interaction terminal displays U, q, a* and historical curves of each parameter in real time. Operators can monitor the status and manually intervene when necessary. All control effects will be continuously optimized through closed-loop feedback in subsequent cycles.

[0075] Further explanation of this workflow: When starting up for the first time or changing the raw material grade, it first runs in "process parameter self-tuning mode": based on the process parameters of similar raw materials in the historical qualified product database as the initial setting, it executes several injection cycles. During this period, it mainly relies on the deviation identification module for monitoring, and the dynamic control module explores optimization with a small range, and quickly converges to a stable production state.

[0076] Furthermore, if the dynamic control module fails to reduce the process abnormality probability q below the threshold or detects hardware abnormalities such as sensor malfunctions when the control commands issued within N consecutive injection cycles (e.g., N=5), the system will automatically switch to "safety hold mode", lock the current process parameters and stop autonomous adjustment, and issue an audible and visual alarm through the human-machine interface terminal to prompt the operator to intervene and check.

[0077] It should be further noted that throughout the entire operation, each module strictly followed... Figure 4The data flow shown is executed sequentially, forming a closed-loop control link with strict time-series dependence. This invention integrates raw material state perception, plasticizing quality assessment, injection modeling, deviation identification, and dynamic control to construct a complete intelligent manufacturing system. The hardware components are clearly connected, the signal paths are clear, and the control logic is rigorous. It can achieve autonomous perception, intelligent judgment, and real-time adjustment of the entire shoe injection molding process without relying on human experience intervention, thus ensuring the stability of the production process and the consistency of the products.

[0078] All contents not described in detail in the specification are existing technologies known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited; conventional equipment can be used. Electrical control components not mentioned in this technical solution are existing technologies and are therefore not shown in the figures, nor will they be described here.

[0079] 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. An integrated intelligent manufacturing equipment for shoe injection molding machines, characterized in that, Includes a frame (1); a hopper assembly (2), installed on the top of the frame (1), which integrates a raw material crushing and pretreatment unit; a screw conveyor assembly (3), fixedly installed in the middle of the frame (1), whose inlet is connected to the outlet of the hopper assembly (2) through a flange seal; a fixed mold assembly (4) and a moving mold assembly (5), which are respectively fastened to the fixed template and the moving template on the front side of the frame (1) by bolts, together forming a molding cavity, and the outlet of the screw conveyor assembly (3) is connected to the molding cavity through a hot runner system; and a control component (6), which is configured to perform intelligent closed-loop control of the entire process from raw material pretreatment, plasticizing to injection molding. The control unit (6) includes a raw material state sensing module, a plasticizing quality assessment module, an injection process modeling module, a deviation identification module, a dynamic control module, and a human-machine interaction terminal. The output end of the raw material state sensing module is electrically connected to the input end of the plasticizing quality assessment module, the output end of the plasticizing quality assessment module is electrically connected to the input end of the injection process modeling module, the output end of the injection process modeling module is electrically connected to the input end of the deviation identification module, the output end of the deviation identification module is electrically connected to the input end of the dynamic control module, and the output end of the dynamic control module is electrically connected to the input end of the human-machine interaction terminal.

2. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The raw material state sensing module includes a particle size image sensor, a humidity detection probe, an infrared spectrometer, and a vibration spectrum collector. The particle size image sensor is installed on the upper part of the inner wall of the hopper assembly (2). The humidity detection probe is embedded in the inner wall of the bottom discharge channel of the hopper assembly (2). The detection window of the infrared spectrometer faces the feeding area in the middle of the hopper assembly (2). The vibration spectrum collector is fixed to the outer wall of the hopper assembly (2).

3. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 2, characterized in that, The particle size image sensor acquires a projected image of the raw material particles every 0.5 seconds, and obtains the equivalent diameter distribution vector of the particles through an edge extraction algorithm; the humidity detection probe outputs a relative humidity value sequence at a frequency of 1Hz; the infrared spectrometer operates at 1200-1800cm². -1 The wavenumber range is scanned to generate an absorption peak intensity vector; the vibration spectrum acquisition device has a sampling frequency of 5kHz, and outputs the main frequency energy ratio after fast Fourier transform.

4. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The plasticizing quality assessment module calculates the plasticizing uniformity index based on the multidimensional characteristics of raw materials. The calculation process includes: Data alignment and time synchronization: The particle size distribution vector, humidity sequence, infrared absorption vector and vibration dominant frequency energy are aligned according to timestamps to form a four-dimensional synchronous observation matrix; Feature normalization: Z-score standardization is applied to the data in each dimension. Principal component dimensionality reduction: Perform principal component analysis on the standardized observation matrix, retain the first two principal components with a cumulative contribution rate ≥90%, and form two-dimensional plasticized input features; Plasticization uniformity index calculation: Input the two-dimensional features into the pre-trained Gaussian process regression model and output the plasticization uniformity index U, U∈[0,1].

5. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The injection process modeling module is used to construct the theoretical response surface of the current batch injection process. The calculation process includes: Establish the process parameter vector P = (T1, T2, n, p) T Where T1 is the setpoint for the barrel front temperature, T2 is the setpoint for the nozzle temperature, n is the screw speed, and p is the back pressure value; Based on the historical qualified product database, the radial basis function interpolation method is used to construct the mapping relationship F: P→U_theory; By comparing the actual plasticizing uniformity index U with U_theory, the deviation vector ΔU = U-U_theory is defined. If |ΔU|>0.05, the deviation identification module is triggered to enter the anomaly judgment process.

6. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The deviation identification module is used to quantify the degree of process deviation in the current injection cycle, and its execution logic includes: Construct a multi-source deviation feature set D = (ΔU, ΔT_melt, Δτ_fill, ΔV_shrink), where ΔT_melt is the difference between the measured melt temperature and the set value, collected by a thermocouple installed at the outlet of the screw conveyor assembly (3); Δτ_fill is the difference between the actual filling time and the standard filling time, calculated by the displacement sensor of the moving mold assembly (5) in combination with the mold opening and closing sequence; ΔV_shrink is the product volume shrinkage rate deviation, estimated by the pressure-displacement combined sensor built into the fixed mold assembly (4); Input D into a lightweight convolutional neural network, which contains two one-dimensional convolutional kernels with a kernel size of 3, 16 filters, and a fully connected layer, and outputs the process anomaly probability q∈[0,1]. When q > 0.7, it is determined to be a significant process deviation, and a set of control instructions is generated.

7. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The dynamic control module is used to generate and issue real-time adjustment commands, and its control strategies include: Establish a control action space A = {ΔT1, ΔT2, Δn, Δp}, and limit the value range of each action variable to ±10% of the original set value; The Q-learning algorithm is used to update the action value function Q(D,a) online, and the reward function r = -(|ΔU| + 0.3|ΔT_melt| + 0.2|Δτ_fill|). Select the action 'a' that maximizes the Q value as the current control instruction; The signal 'a' is decomposed into four independent control signals, which are sent to the barrel heating coil driver, the nozzle temperature control unit, the screw servo motor controller, and the back pressure proportional valve, respectively.

8. The integrated intelligent manufacturing equipment for shoe injection molding machines according to claim 1, characterized in that, The human-machine interface terminal is used to display the current plasticization uniformity index U, process abnormality probability q, control action a*, and historical trend curve in real time, and supports operators to manually intervene in the control threshold.

Images