A multi-mode collaborative intelligent forming control system

Abstract

The present invention discloses a multi-mode collaborative intelligent molding control system, which specifically relates to the field of production strategy control and is used to solve the problems in the existing mold molding process that it is difficult to dynamically adjust the control strategy according to different mold characteristics, and the molding cycle segment division cannot be intelligently matched. The preloading module collects the mold pressing displacement parameters and thermal response curves during the molding cycle, the response hysteresis analysis module analyzes the force response and structural deformation absorption capacity of the mold, the thermal inertia analysis module identifies the mold thermal inertia window, and jointly constructs the segment division strategy space of the current molding cycle. The control module sets the timing boundaries of the pressure rise section, the pressure stabilization section and the cooling section to achieve precise control of the molding process. The system further constructs a dual-axis parameter space based on the hysteresis attribute and the thermal inertia window through the strategy space update module, executes strategy updates, and forms a strategy adjustment network that can adapt to the multi-mode collaborative processing process.

Description

Technical Field

[0001] The present invention relates to the technical field of production strategy control, and more specifically, to a multi-mode collaborative intelligent molding control system. Background Art

[0002] In the multi-mold collaborative molding production system, in response to the application requirements of mold combinations with significant differences in structural characteristics, such as small embryo molds, steam molding molds, and secondary molding molds, existing production lines are gradually tending to support mixed loading and rapid switching operations of multiple types of molds to improve equipment utilization and flexible manufacturing capabilities. In actual production, there are often multiple molds that are put into use in turn on the same molding platform, with a high switching frequency, dynamic changes in task sequence, and continuous changes in molding targets and structural types within a short period of time, forming a complex control scenario for multi-mold mixed cutting. In this scenario, there are natural differences in the molding path structure, boundary reaction characteristics, thermal inertia, and force feedback of various molds. The traditional method of relying on mold numbers to load fixed process parameter templates is difficult to adapt to the real-time changing structural response state, and cannot handle the control response offset caused by mold aging, deformation, or repair. This leads to a decrease in molding stability and extended parameter debugging time, which seriously restricts the production line beat and product consistency control capabilities. In addition, existing control systems generally lack the ability to identify mold control behavior online, and are unable to perceive the dynamic deviation between mold behavior and process control objectives during the actual response process. It is difficult to support the intelligent ability to instantly derive control strategies based on the current true state of the mold, thus forming a key control bottleneck of "inaccurate identification, inappropriate parameters, and unchanged strategies" in the multi-mold mixed cutting process. It is urgent to build a new intelligent control planning method with perception capabilities, self-generated strategy capabilities and process correction capabilities. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a multi-mode collaborative intelligent molding control system to solve the problems raised in the above-mentioned background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A multi-mode collaborative intelligent forming control system includes a preloading module, a response hysteresis analysis module, a thermal inertia analysis module, a strategy space construction module, a control module, and a strategy space update module, wherein:

[0006] The preloading module performs preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and collects the mold press displacement parameters and thermal response curve;

[0007] The response hysteresis analysis module combines press-fit displacement parameters to analyze the mold's force response speed and structural deformation absorption capacity, and outputs the mold's response hysteresis properties.

[0008] The thermal inertia analysis module analyzes the slope change of the thermal response curve and identifies the thermal inertia window of the closed mold;

[0009] The strategy space construction module constructs the segment division strategy space of the current molding cycle of the mold based on the hysteresis attribute and thermal inertia window, and sets the time ratio of the pressure rise section, pressure stabilization section and cooling section;

[0010] The control module converts the time ratio of the pressure rise section, the pressure stabilization section and the cooling section into the section boundary timing sequence, and adjusts the control trigger point of the section within the molding cycle;

[0011] The strategy space update module establishes a strategy update model, updates the strategy space of the dual-axis parameter based on the hysteresis property and the thermal inertia window, and generates a strategy adjustment network for the steady-state control of the multi-die forming process.

[0012] In a preferred embodiment, the preloading module performs a preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and the collection of mold press displacement parameters and thermal response curves specifically includes:

[0013] Initialize the press-fit drive, apply a preset standard low-pressure closing command to the clamped mold, and record the press-fit displacement at the same time;

[0014] Calculate the change rate of the press-fit displacement over time. When the change rate is continuously converged, it is marked that the mold has reached the closed state. The press-fit displacement change rate sequence and the closing completion time are used as press-fit displacement parameters.

[0015] When the mold is closed, the heat source is started and the rise of the mold surface temperature over time is monitored to generate a thermal response curve.

[0016] In a preferred embodiment, the response hysteresis analysis module analyzes the force response speed and structural deformation absorption capacity of the mold in combination with the press-fitting displacement parameters, and outputs the response hysteresis properties of the mold, specifically including:

[0017] Among the press-fitting displacement parameters, the average displacement change rate in the response time interval from preloading start to closing completion is extracted, and the average displacement change rate is defined as the force response speed index of the mold;

[0018] The average change rate fluctuation amplitude before and after the maximum displacement change rate is extracted, and the structural deformation absorption capacity index of the mold is defined according to the normalized ratio between the force response speed index and the average change rate fluctuation amplitude.

[0019] The force response speed index and the structural deformation absorption capacity index are used as dual input parameters to perform weighted comprehensive calculations and output the numerical expression of the mold hysteresis property.

[0020] In a preferred embodiment, the thermal inertia analysis module performs slope change analysis on the thermal response curve to identify the thermal inertia window of the closed mold, specifically including:

[0021] Obtain a thermal response curve, and smooth the thermal response curve based on a sliding window average method;

[0022] Preset the slope stable interval, and mark the first time point when the slope of the thermal response curve enters the stable interval as the starting point of the thermal stable section;

[0023] The time span between the starting point of the thermal response curve and the starting point of the thermal stability section is calculated, and this time span is defined as the thermal inertia window of the closed mold.

[0024] In a preferred embodiment, the strategy space construction module constructs a segment division strategy space of the current molding cycle of the mold based on the hysteresis attribute and the thermal inertia window, and sets the time ratio of the pressure rise segment, the pressure stabilization segment, and the cooling segment, specifically including:

[0025] Establish a two-dimensional coordinate system in the strategy space with the hysteresis attribute as the horizontal axis and the thermal inertia window as the vertical axis;

[0026] The parameter space corresponding to the coordinate system is divided into rectangular strategic areas, and each strategic area is preset with a set of segment division strategies for the molding cycle;

[0027] Set the boundary scale of the rectangular strategy area and use the interval overlapping construction method to remove the boundary fuzzy area so that the rectangular strategy area completely covers the parameter space;

[0028] The hysteresis properties and thermal inertia window of the current mold cycle are obtained and mapped to the corresponding strategy area in the two-dimensional coordinate system. The time ratio of the pressure rise section, pressure stabilization section and cooling section of the corresponding strategy is extracted through the coordinate hit results.

[0029] In a preferred embodiment, the vertical axis of the two-dimensional coordinate system of the strategy space takes the thermal inertia window duration at the first hot start as the maximum time constraint.

[0030] In a preferred embodiment, the control module converts the time ratio of the pressure rise section, the pressure stabilization section, and the cooling section into a segment boundary timing sequence, and adjusts the control trigger point of the segment within the molding cycle specifically including:

[0031] The time ratio of the pressure rise section, the pressure stabilization section and the cooling section is mapped to the total time axis of the molding cycle to form two timing boundaries: the end of the pressure rise section and the end of the pressure stabilization section.

[0032] The two timing boundaries are injected into the thermal control and pressure control links to reconstruct the control trigger points of the section adjustment channels within the molding cycle.

[0033] In a preferred embodiment, the strategy space update module establishes a strategy update model, performs strategy update on the dual-axis parameter strategy space based on the hysteresis attribute and the thermal inertia window, and generates a strategy adjustment network for steady-state control of the multi-mold molding process, specifically including:

[0034] A reinforcement learning-based strategy update model was established to obtain molding cycle segment division strategies for different types of molds, and the time ratio structure of each strategy group was used as the state input for model recognition.

[0035] The single time adjustment operation of the pressure rise section, pressure stabilization section and cooling section is set as a discrete action set, including three operation forms: increase, decrease or keep the section time unchanged;

[0036] Through the mold processing forming rate feedback strategy update model, the strategy update operation is performed in the two-dimensional strategy coordinate space. The segment time ratio parameters of the corresponding coordinate points are corrected and updated through the strategy gradient method. The updated strategy area is generated and injected into the next cycle for use.

[0037] All strategy update results are recorded as a dynamic strategy map in the form of coordinate points, and a real-time control strategy adjustment network is constructed for the hysteresis properties and thermal inertia windows of different mold types.

[0038] In a preferred embodiment, during the execution of the strategy update process in the two-dimensional strategy coordinate space, a maximum cooling section length compression threshold is preset. When the compression degree exceeds the threshold, the molding cycle length is increased by an equal amount to cover the excess portion.

[0039] The technical effects and advantages of the multi-mode collaborative intelligent forming control system of the present invention are as follows:

[0040] By incorporating a dual analysis mechanism based on mold response hysteresis and a closed thermal inertia window, this system achieves precise quantification and identification of mold dynamic characteristics during multi-mold molding. This overcomes the limitations of traditional processes, where mold segmentation relies on empirical assumptions and lacks feedback adjustment capabilities. The strategy space construction module combines the hysteresis property and thermal inertia window to construct a dual-axis parameter control strategy space. This allows for adaptive adjustment of the time ratios for the pressure rise, pressure stabilization, and cooling stages to accommodate diverse mold response characteristics. The control module converts these time ratios into boundary control sequences within the molding cycle, ensuring that each stage executes under optimal thermal and structural response conditions. Furthermore, the strategy space update module incorporates a strategy adjustment network, enabling dynamic optimization and updating of the strategy space during the molding process. This enhances the system's robustness and stability to multi-mold fluctuations, mold variability, and operating condition disturbances, significantly improving the yield rate and control accuracy of the molding process. The system possesses strong engineering applicability and industrial application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1This is a structural schematic diagram of a multi-mode collaborative intelligent forming control system of the present invention. DETAILED DESCRIPTION

[0042] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0043] Example 1

[0044] Figure 1 The present invention provides a multi-mode collaborative intelligent forming control system, which includes a preloading module, a response hysteresis analysis module, a thermal inertia analysis module, a strategy space construction module, a control module, and a strategy space update module, wherein:

[0045] The preloading module performs preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and collects the mold press displacement parameters and thermal response curve;

[0046] The response hysteresis analysis module combines press-fit displacement parameters to analyze the mold's force response speed and structural deformation absorption capacity, and outputs the mold's response hysteresis properties.

[0047] The thermal inertia analysis module analyzes the slope change of the thermal response curve and identifies the thermal inertia window of the closed mold;

[0048] The strategy space construction module constructs the segment division strategy space of the current molding cycle of the mold based on the hysteresis attribute and thermal inertia window, and sets the time ratio of the pressure rise section, pressure stabilization section and cooling section;

[0049] The control module converts the time ratio of the pressure rise section, the pressure stabilization section and the cooling section into the section boundary timing sequence, and adjusts the control trigger point of the section within the molding cycle;

[0050] The strategy space update module establishes a strategy update model, updates the strategy space of the dual-axis parameter based on the hysteresis property and the thermal inertia window, and generates a strategy adjustment network for the steady-state control of the multi-die forming process.

[0051] The preloading module performs preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and collects the mold press displacement parameters and thermal response curve.

[0052] During the start-up phase of the molding cycle, mold press-fitting initialization is performed. Specifically, the target mold is clamped between the press-fitting mechanism's fixed bracket and the drive arm, ensuring that the mold's spatial positioning accuracy meets closing requirements. The press-fitting drive is controlled by a hydraulic system. During this initialization phase, a low-pressure closing command is issued to control the hydraulic cylinder to slowly advance the drive arm, achieving a gradual press-fit closing of the mold. The low-pressure setting should ensure that there is no risk of plastic deformation of the mold. It is typically set within 10% to 20% of the mold's maximum load-bearing capacity, and a stable loading process is maintained for at least 5 seconds to fully expose the structural response characteristics of the mold components under initial loading. During the press-fitting process, real-time displacement data of the drive arm is collected, recording the change in distance between the front end of the drive arm and a fixed reference point. This displacement data is collected by a linear displacement sensor mounted on the drive mechanism. The sampling frequency is recommended to be at least 52Hz to ensure continuity during the press-fitting dynamic process. The collected press-fitting displacement sequence is recorded as a reflection of the mold's force response during the press-fitting actuation process.

[0053] After completing the initial press-fit displacement acquisition, a differential operation is performed on the acquired displacement time series to calculate the displacement changes between adjacent time points, thereby generating a rate of change curve for the press-fit displacement. This rate of change curve reflects the changing characteristics of the mold's response speed during the press-fit process. Typically, during the initial press-fit process, the mold structure exhibits significant elastic absorption and clearance adjustment, manifested by a large rate of change. As the press-fit process approaches the closing stage, the mold structure stabilizes, and the displacement changes generated by the press-fit drive decrease significantly, manifested by a continuously decreasing rate of change that gradually converges to a low value. The standard for determining mold closing status is set as follows: when the fluctuation amplitude of the press-fit displacement rate of change for at least 10 consecutive sampling time points is below a preset convergence threshold (e.g., an absolute value of the rate of change of less than 1% mm / s), the mold is marked as closed. This judgment process avoids false positives caused by minor vibrations or mechanical rebound, ensuring engineering reliability of the closing determination. The system clock time when closing is completed is recorded as the closing completion time point of the press-fitting process. The sequence of the press-fitting displacement change rate from the start of loading to the completion of closing is retained and used as the press-fitting displacement parameter to characterize the absorption process and response rate characteristics of the mold structure.

[0054] Under the condition of the mold being closed, the heat source is started based on the confirmation of mold closure to perform standardized heating treatment on the mold. The heat source can be in the form of a resistance heating element, an infrared radiation source or a hot oil circulation pipeline. The specific selection depends on the molding process requirements and the mold structure layout. The constant heating power mode is adopted when the heat source is started. The temperature acquisition system is started synchronously, and thermocouples or infrared temperature sensors are used to monitor the temperature change data of multiple typical points on the outer surface of the mold in real time. The sensor points should cover the center and edge areas of the mold, and the number of points is usually 4 to 6 to ensure that the temperature response is representative. The sampling period is recommended to be set to 1 second, and data recording is continued until the mold temperature reaches the steady-state temperature range specified by the process. At the same time, a temperature rise curve is formed according to the temperature change process over time.

[0055] The response hysteresis analysis module analyzes the force response speed and structural deformation absorption capacity of the mold in combination with the press-fitting displacement parameters, and outputs the response hysteresis properties of the mold.

[0056] Obtain a complete time series of the press-fit displacement rate change from the start of preload press-fit to the completion of mold closure. This response interval is defined as the time range starting from the initial low-pressure closing command and ending at the time the press-fit displacement rate converges and mold closure is completed. Within this time range, the displacement rate change values ​​of all sampling points are averaged to obtain the average displacement rate change of the mold during the entire press-fit response period. This average value physically reflects the overall force response speed of the mold structure during the low-pressure loading phase and has clear engineering implications. A high average rate value indicates a rapid response to external forces, with clear displacement feedback, suggesting relatively low internal structural rigidity or the presence of large gap absorption. Conversely, a low average rate value indicates a sluggish overall structural response, possibly characterized by high overall stiffness or damping characteristics. Therefore, this average value is defined as the mold's force response speed indicator, expressed in a unified unit (e.g., millimeters per second).

[0057] After obtaining the press-fit displacement rate sequence and extracting the force response speed index, the phase of the displacement rate sequence with the most dramatic dynamic changes is further analyzed. This means identifying the main response moment corresponding to the maximum displacement rate value. This moment typically occurs between the initial and mid-stages of press-fit loading, reflecting the stage when the mold structure absorbs the most external forces. A symmetrical time window is set around this main response moment, for example, with five sampling points before and after it, for a total of 11 points of displacement data. The average fluctuation amplitude of the displacement values ​​within this time window is calculated, specifically the degree of dispersion between these sampling points and the window average. This average fluctuation amplitude reflects the mold's response stability or volatility within the main response interval. Large fluctuations typically indicate significant nonlinear absorption behavior in the structure, such as gap buffering, local plastic yielding, or relative displacement of multiple components. This fluctuation amplitude is then normalized with the aforementioned force response speed index and correlated using a ratio to define the mold's structural deformation absorption capacity index. This index quantitatively indicates whether the mold has strong internal energy absorption during press-fit loading. The smaller the ratio, the more unstable the mold structure's response to external forces; the larger the ratio, the more stable and rigid the mold structure.

[0058] After extracting the force response speed index and the structural deformation absorption degree index, it is necessary to clarify the physical and logical relationship between these two indicators and the mold hysteresis property. Specifically, the force response speed index reflects the mold's responsiveness during press loading. A higher value indicates a faster displacement response to external forces, a more compact mold rigidity, and no significant delayed response. The structural deformation absorption degree index measures the mold's internal structure's ability to directly absorb energy during the main response phase. A higher value indicates a lack of cushioning or nonlinear hysteresis. Therefore, both indicators are negatively correlated with hysteresis: higher values ​​indicate lower mold hysteresis values. Based on this physical relationship, an inverse reduction function must be constructed when calculating mold hysteresis properties, avoiding the direct use of a linear weighted model. In this implementation, the two indicators are processed using a normalized inverse function. This involves normalizing the force response speed index and the structural deformation absorption degree index separately, taking their inverses, and then weighting them together according to a weighting coefficient. For example, if the normalized force response speed index is weighted 50% and the structural deformation absorption index is weighted 50%, the mold hysteresis property value can be composed of the weighted sum of the two normalized reciprocals. This method effectively avoids the additive model results with physical errors and ensures that the hysteresis value forms a consistent negative correlation with the response speed and absorption capacity.

[0059] The resulting hysteresis attribute value is a dimensionless parameter. A larger value indicates a more pronounced mold response hysteresis during press loading, indicating more structural buffering or deformation energy absorption, reflecting a degree of dynamic instability in the mold's response. Conversely, when the hysteresis attribute value approaches zero, it indicates a fast and stable mold response, with no significant delay or absorption effects in the internal structure, making it suitable for high-precision molding control scenarios.

[0060] The thermal inertia analysis module performs slope change analysis on the thermal response curve to identify the thermal inertia window of the closed mold.

[0061] A sliding window averaging operation is performed on the thermal response curve to eliminate possible measurement noise and short-term fluctuations. The width of the sliding window is set to 5 to 10 consecutive time points, and the specific value is set based on the sampling frequency of the temperature sensor and the stability of the temperature change rate. This ensures that smoothing is achieved while maintaining the overall shape of the curve. The sliding window processing method is a fixed-length window recursive processing. That is, starting from the starting point of the curve, the data points in each continuous window are averaged to generate the smoothed value at the corresponding time point, and then a new smoothed thermal response curve is constructed. The curve visually presents a continuous and smooth temperature increase trend, effectively removing nonlinear disturbance factors such as jumps and oscillations that may exist in the original curve.

[0062] To identify the point at which the mold transitions from a rapid temperature rise to a stable state, the first-order derivative of the smoothed thermal response curve is extracted to generate a time-varying temperature slope sequence. A stable interval for the thermal response slope is set to control the consistency of slope identification and the engineering feasibility of the judgment threshold. This interval is set to maintain a stable temperature slope between 1% and 3% degrees Celsius per second for at least 30 seconds. The upper and lower thresholds of this interval are adjusted based on the thermal conductivity, structural mass, and heat capacity of the mold material, and are validated using historical data and empirical statistics during the setting process. The slope sequence is scanned point by point from the starting point. When the slope value first enters the preset stable interval and remains stable for at least 10 consecutive time points, the current time point is determined to be the starting point of the thermal response entering the stable section. This time point has a clear physical meaning: the mold's overall structure has completed its initial thermal response, its heat conduction process has reached equilibrium, and the subsequent temperature rise rate no longer increases significantly, thus entering a stable state.

[0063] After identifying the starting time of the thermal stability section, the time span between the two is calculated, combined with the starting time of the thermal response when the mold is closed. This time span reflects the complete response process of the mold from the initial temperature to the thermally stable state in the closed state, reflecting the response inertia of the mold's structure and material system to thermal energy in the absence of external disturbances. The size of the thermal inertia window directly reflects the response time required for the mold to reach thermal stability after closure. A larger window value indicates greater hysteresis in the mold's thermal response process after closure, which may be related to factors such as thicker mold walls, complex heat conduction paths, or high heat capacity. Conversely, a smaller window value indicates high heat conduction efficiency and a compact structure, making the mold suitable for rapid thermal control strategy intervention. In the actual molding control process, the thermal inertia window will serve as one of the key input parameters for the thermal control strategy to enter the judgment stage, guiding the time synchronization and switching node setting of the subsequent segment control logic.

[0064] The strategy space construction module constructs a segment division strategy space of the current molding cycle of the mold according to the hysteresis attribute and the thermal inertia window, and sets the time ratio of the pressure rise section, the pressure stabilization section and the cooling section.

[0065] Define engineering-quantifiable ranges for two key control parameters: mold hysteresis and thermal inertia. The mold hysteresis range is determined by combining a score of force response speed and structural deformation absorption, typically normalized to a dimensionless interval between 0 and 1. The thermal inertia window, measured in seconds, is defined as the time it takes for the mold to stabilize from closing to reaching temperature stability. Its range is capped based on differences in mold material and structure. The maximum thermal inertia window duration measured during the first hot start is used as the vertical axis limit to avoid spatial delineation exceeding the mold's achievable response range. A two-dimensional coordinate space is constructed, with the hysteresis property as the horizontal axis variable and the thermal inertia window as the vertical axis variable, encompassing typical mold response characteristics. This coordinate system has fixed boundaries, with the horizontal dimension typically ranging from 0 to 1 and the vertical dimension based on the maximum value of the thermal inertia window at initial startup. For example, if the maximum value is 240 seconds, the vertical axis is divided from 0 to 240 seconds. The horizontal axis represents the mechanical hysteresis response capability of the structure, while the vertical axis represents the inertia strength of the thermal process. In molding cycle control, both dimensions jointly influence the selection of segment control strategies.

[0066] Based on the constructed two-dimensional coordinate space, to achieve structured and rapid indexing of strategy management, the entire parameter plane is discretized and divided into several strategic control segments using rectangular strategy regions. Along the horizontal axis (hysteresis attribute), these segments are typically divided into 5 to 10 equally spaced intervals, while along the vertical axis (thermal inertia window), they are divided into equal-span segments based on the upper limit of the thermal inertia window. The resulting rectangular regions are unique mapping blocks for each hysteresis-inertia parameter combination. The total number of strategy regions is the product of the number of horizontal and vertical partitions. Each strategy region is pre-configured with a set of molding cycle segmentation strategies, including a three-stage time allocation scheme for the pressure rise stage, pressure stabilization stage, and cooling stage. For example, the time allocations for all strategies are set to 100%, which is used to construct a standardized intra-cycle segment control time distribution. Once this strategy library is established, parameter coordinates can be input to perform a hit match in the strategy space and quickly extract a control strategy.

[0067] When partitioning the parameter space, it is necessary to address the issue of ambiguous policy assignments caused by parameter values ​​falling within critical points at the partition boundaries. To ensure that any parameter combination falls within a defined policy region and to prevent critical values ​​from falling into blank areas shared by multiple regions or belonging to no region, an overlapping interval construction method is employed. This method creates a small overlap between adjacent policy regions. The overlap width can be set based on the actual parameter distribution, typically 5% to 10% of the original interval span. After this overlapping construction, parameter values ​​within the overlapping intervals are prioritized to the lower bound (leftward on the horizontal axis). Specifically, when a parameter value is at the midpoint of the overlapping region or at the lower bound, it is assigned to a policy region within the lower interval, thus ensuring continuity and stability in policy assignments. This construction method effectively avoids ambiguous region assignments and improves the robustness of policy matching at parameter critical values, resulting in smoother control switching characteristics in the policy space.

[0068] Through response hysteresis analysis and thermal inertia analysis, the hysteresis attribute value and thermal inertia window duration of the mold's current cycle are obtained, respectively. This parameter pair is used as coordinate input and directly mapped into the pre-built strategy space coordinate system, locating the corresponding rectangular strategy area. Upon successful hit, the three-segment time ratio scheme corresponding to the strategy area is automatically extracted from the strategy.

[0069] The control module converts the time ratio of the pressure rise section, the pressure stabilization section and the cooling section into a section boundary timing sequence, and adjusts the control trigger point of the section within the molding cycle.

[0070] Obtain the total molding duration set for this cycle. The total molding cycle duration is set during the control task configuration phase and is based on mold materials, product specifications, and processing requirements. For example, it can be 180 seconds, 240 seconds, or other standard cycle values. In this embodiment, the current cycle duration is set to 240 seconds. Based on the time ratio results previously extracted from the strategy library, such as 25% for the pressure rise segment, 55% for the pressure stabilization segment, and 20% for the cooling segment, they are proportionally mapped to a 240-second timeline. The specific calculation process is as follows: the pressure rise segment duration is 240 seconds × 25% = 60 seconds, the pressure stabilization segment is 240 seconds × 55% = 132 seconds, and the cooling segment is the remaining 48 seconds. This provides a clear duration for the three control segments. Based on the three-segment duration, the two timing boundaries of the pressure rise segment end time and the pressure stabilization segment end time are calculated to be the 60th second and the 192th second, respectively. These two boundaries serve as key timing points in the molding cycle and determine the segmented adjustment nodes for molding state control. Through control logic, the 60th second is set as the trigger point for the voltage-boosting phase end signal, and the 192nd second is set as the trigger point for the voltage-stabilizing phase end signal. These two timing nodes are loaded into the control schedule at the beginning of the cycle execution, and are precisely matched based on the system time.

[0071] The two timing boundaries are injected into the thermal control link and the pressure control link respectively, serving as the starting basis for control and regulation of each channel. For the pressure control channel, the system performs a constant pressurization operation from 0 to 60 seconds. After 60 seconds, the pressure is maintained at the set steady state until 192 seconds. At 192 seconds, the system executes the pressure relief instruction, starting the residual pressure release process in the cooling phase. In the thermal control link, the system implements the temperature maintenance strategy according to this timing, starting with the heat preservation maintenance mechanism at the starting point of the pressure stabilization phase, and switching to the rapid cooling control strategy at 192 seconds. In conjunction with the cooling preparation before the mold is opened, this realizes the staged progressive adjustment of the thermal control process.

[0072] The strategy space updating module establishes a strategy updating model, performs strategy updating on the dual-axis parameter strategy space based on the hysteresis attribute and the thermal inertia window, and generates a strategy regulating network for steady-state control of the multi-die forming process.

[0073] A policy update model tailored to the characteristics of multiple mold types was established, employing reinforcement learning to achieve adaptive optimization of the policy space. By collecting and organizing molding process data from different mold types, the timing structures of the pressure-increasing, pressure-stabilizing, and cooling-down phases used in each were identified and uniformly converted into a time-ratio representation based on the total molding cycle duration. Specifically, for any mold type, a typical molding cycle partition was obtained, and the duration of each phase was normalized to a percentage of the total cycle time. For example, the pressure-increasing phase was assigned a value of 25%, the pressure-stabilizing phase was assigned a value of 50%, and the cooling-down phase was assigned a value of 25%, thereby forming a standardized state vector input. This representation eliminates differences in the absolute time dimension between different molds, making the state space representation more uniform and comparable, and facilitating training convergence and improving state matching efficiency of the reinforcement learning model. This state input vector serves as the current state representation of the policy update model. A reinforcement learning algorithm is used to construct a state-action-reward mapping, and the model policy is continuously updated iteratively. At the same time, to ensure that the state input can accurately reflect the physical characteristics and process history of the mold, the mold's hysteresis properties and thermal inertia window indicators need to be bound to the state vector as the key basis for identifying the differences in thermal behaviors of different molds.

[0074] Define the action set for the reinforcement learning model. The action set is a set of specific policy adjustment behaviors that the model can take after identifying the current state. In the specific technical solution, the action set is limited to discrete fine-tuning operations on the time ratio structure of the pressure rise, pressure stabilization, and cooling stages. Specifically, for each control segment within the molding cycle, three basic action types are defined: increasing the time ratio for that segment; decreasing the time ratio for that segment; and maintaining the same time ratio without adjustment. These three actions constitute the complete operation space for that segment. By setting these action combinations for each of the three segments, a discrete action set with 27 possible (3×3×3) action arrangements is formed. To avoid excessive perturbations to the original policy structure during the policy update process, the ratio change for each fine-tuning operation is set to a fixed percentage, with a default setting of 2% to 5% of the original parameters. This design ensures the gradual nature of the policy update, reduces the risks associated with extreme strategies, and ensures stability during model training and controllability of the control curve. In addition, when increasing the allocation of a certain segment, the allocation time must be deducted from other segments by an equal amount at the same time, so as to maintain the constraint that the total allocation is 100% and avoid the situation where the overall strategy is unbalanced due to mutual squeezing between segments.

[0075] A forming rate feedback mechanism is constructed to drive the iterative optimization process of the policy update model. After each mold processing cycle, forming rate feedback data generated by the current policy structure is obtained in real time. The forming rate is the ratio of the number of qualified parts to the total number of parts produced per cycle. This metric directly reflects the impact of the policy structure on the efficiency of mold thermal control and pressure regulation and is a key criterion for evaluating policy performance. This forming rate serves as an immediate reward signal in the reinforcement learning model, driving policy gradient algorithm updates. Specifically, when the policy matching structure corresponding to the current action sequence yields a higher forming rate reward, the value function of this state-action pair is adjusted upward, giving it priority in subsequent policy selection. During the update process, the policy model performs policy mapping based on the current mold's hysteresis properties and the two-dimensional coordinate space of the thermal inertia window. It locks the current state point and performs policy gradient propagation within its neighborhood. The policy matching parameters are modified using the policy gradient method. The direction and magnitude of the adjustment are derived from the gradient estimate of the historical reward curve, ensuring that each update converges toward the desired performance improvement. The revised strategy ratio structure will be used as the execution strategy for the new cycle and injected into the molding cycle control process to form a closed-loop feedback optimization mechanism.

[0076] Based on the coordinate point representation of the reinforcement learning model during the policy update process, after each policy update, the hysteresis property and thermal inertia window indicator corresponding to the current mold type are used as coordinate axes to construct dimensions. The updated strategy time ratio structure is plotted in this two-dimensional space, forming a structured coordinate point record. Each coordinate point represents a set of molding control strategies suitable for a mold with specific thermal response characteristics, including the ratio parameters for the pressure rise, pressure stabilization, and cooling stages, the update gradient direction, and the corresponding molding rate feedback value. Ultimately, a dynamic strategy map based on the strategy evolution trajectory is constructed. To ensure the map's continuous evolution capability, new coordinate points are dynamically added to the map after each cycle update, forming a traceable trajectory sequence of historical strategies. If a mold type undergoes multiple strategy evolution stages over different cycles, its coordinate trajectory will appear as a continuous line segment. Based on this trajectory, its strategy convergence trend and optimal strategy convergence point are analyzed, providing a reference path for subsequent similar mold types. In addition, to enhance the engineering applicability of the atlas, the atlas has real-time query and local strategy migration capabilities. That is, when a new mold thermal characteristic indicator is first identified, the initial strategy configuration is called by calling the nearest strategy point through a similarity matching mechanism with existing coordinate points in the atlas. The following is a specific process example:

[0077] A composite metal preform mold with hysteresis properties was selected for molding using a set of supercritical foamed polypropylene beads with high density control requirements as raw material. Before processing began, the mold's hysteresis properties were measured in the strategic coordinate space based on its historical press-fit response. To prevent uneven mold density caused by cavity pressure deviation during the foaming phase, the pressure ramp-up phase was set to 35% of the total cycle, the pressure stabilization phase to 45%, and the cooling phase to 20% during the strategic space construction process. During the actual molding process, the system first preheated the mold cavity to 145°C to stimulate gas release activity in the supercritical bead core. Pressure ramp control was then initiated at an initial injection pressure of 2.5 bar, causing the mold to rapidly increase in thermal deformation. After approximately 5.5 seconds of pressure ramping, the bead shell structure reached critical expansion. The system then adjusted the pressure stabilization phase control parameters based on the real-time thermal inertia response, maintaining the mold cavity at a constant temperature of 145°C and a stable pressure of 4.2 bar for 9 seconds, allowing the beads to fully expand and form within the cavity. The thermal inertia window during this period is 12 seconds, and this parameter is also used as the vertical axis positioning point of the strategic space coordinate. After the stabilization phase, the system enters the cooling phase control, drops the mold wall temperature to 95°C and gradually reduces the pressure to 1.2 bar, and maintains this state for 5 seconds for shape setting and strength consolidation. Because the initial pressure rise phase was slightly longer, the expansion rate of individual beads exceeded the mold strain absorption range. The system marked the boundary position of the batch of beads in the current cycle as having a slight density deviation, and the molding rate was statistically 96.2%. The control system adjusted the strategic coordinate area based on the feedback, shortening the pressure rise phase by 0.8 seconds and extending the stabilization phase by 0.6 seconds in the next cycle to optimize the subsequent foaming uniformity.

[0078] The essential goal of constructing this strategy graph is to achieve structured management of control strategies and a strategy inheritance mechanism for mold categories. By recording each step of the strategy update process as a path node in the graph, the evolution trend and control-sensitive areas of a certain type of mold strategy during actual processing can be quickly identified. Especially when the hysteresis attribute is close to the boundary interval or the thermal inertia span is large, the graph is used to extract stable areas in the historical strategy to avoid the problem of unstable molding performance caused by frequent fluctuations in strategy adjustment. At the same time, the graph also provides basic support for the subsequent construction of the strategy adjustment network. The network uses the strategy graph as the main structure and the mold thermal feature classification as the guiding dimension to establish a set of strategy adjustment paths with strategy nodes as the weight distribution. Each strategy adjustment path corresponds to a strategy adjustment curve adapted for a type of mold. During operation, the network can call the adapted strategy path in the graph for dynamic adjustment based on the real-time identification of the mold thermal response state.

[0079] To prevent excessive compression of the cooling-down time during strategy optimization, which could lead to insufficient heat release and mold cooling during the molding process, a clear limit on the degree of compression during the cooling-down time is required to ensure that the optimized time still meets thermal stability requirements. To this end, the system pre-sets a maximum allowable threshold for cooling-down time compression. This threshold is based on historical empirical data, experimental verification results, and the thermal inertia response characteristics of typical molds. The threshold setting principle is: the cooling-down time should not be less than 20% of the initial setting. For example, if the original cooling-down time is 20 seconds, the compression threshold should be no less than 4 seconds. Once the threshold is set, the compression level is determined during each round of time allocation structure adjustment during the strategy update process. During execution, if the cooling-down time allocation in a given strategy update results in a round of execution falls below the minimum duration limit corresponding to the threshold, an over-limit handling mechanism is automatically triggered. This mechanism does not adjust the allocation structure to adapt to the original molding cycle. Instead, it increases the total molding cycle time, replenishing the compressed portion in isochronous increments. For example, if the cooling section should last for 10 seconds but the actual ratio is only 8 seconds, the system will increase the total cycle time by 2 seconds and keep the relative ratios in each section unchanged, only to ensure the integrity of the minimum cooling time.

[0080] This compression threshold processing not only serves as a protective boundary during strategy updates but also acts as a risk buffer within the dynamic control network. When multiple rounds of strategy updates continuously converge toward compression during the transition to the cooling stage, the system will trigger control intervention in advance based on the compression threshold to suppress process instability that may be caused by excessive compression. Therefore, the compression threshold mechanism and the strategy map together form a safety boundary layer within the strategy adjustment network, ensuring that all parameters remain within a controllable and executable range during the strategy evolution process, effectively improving the process stability and product consistency of the entire molding control solution.

[0081] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0082] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0083] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0084] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0085] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0086] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0087] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0088] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0089] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0090] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-mode collaborative intelligent forming control system, characterized in that: It includes a preloading module, a response hysteresis analysis module, a thermal inertia analysis module, a strategy space construction module, a control module, and a strategy space update module, among which: The preloading module performs preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and collects the mold press displacement parameters and thermal response curve; The response hysteresis analysis module combines press-fit displacement parameters to analyze the mold's force response speed and structural deformation absorption capacity, and outputs the mold's response hysteresis properties. The thermal inertia analysis module analyzes the slope change of the thermal response curve and identifies the thermal inertia window of the closed mold; The strategy space construction module constructs the segment division strategy space of the current molding cycle of the mold based on the hysteresis attribute and thermal inertia window, and sets the time ratio of the pressure rise section, pressure stabilization section and cooling section; The control module converts the time ratio of the pressure rise section, the pressure stabilization section and the cooling section into the section boundary timing sequence, and adjusts the control trigger point of the section within the molding cycle; The strategy space update module establishes a strategy update model, updates the strategy space of the dual-axis parameter based on the hysteresis attribute and the thermal inertia window, and generates a strategy adjustment network for the steady-state control of the multi-die forming process. The strategy space construction module constructs the segment division strategy space of the current molding cycle of the mold based on the hysteresis attribute and the thermal inertia window, and sets the time ratio of the pressure rise section, the pressure stabilization section, and the cooling section, specifically including: Establish a two-dimensional coordinate system in the strategy space with the hysteresis attribute as the horizontal axis and the thermal inertia window as the vertical axis; The parameter space corresponding to the coordinate system is divided into rectangular strategic areas, and each strategic area is preset with a set of segment division strategies for the molding cycle; Set the boundary scale of the rectangular strategy area and use the interval overlapping construction method to remove the boundary fuzzy area so that the rectangular strategy area completely covers the parameter space; Obtain the hysteresis properties and thermal inertia window of the current mold cycle and map them to the corresponding strategy area in the two-dimensional coordinate system. Extract the time ratio of the pressure rise section, pressure stabilization section, and cooling section of the corresponding strategy based on the coordinate hit results. The strategy space update module establishes a strategy update model, performs strategy update on the dual-axis parameter strategy space based on the hysteresis attribute and the thermal inertia window, and generates a strategy adjustment network for the steady-state control of the multi-mold forming process. Specifically, the strategy update module includes: A reinforcement learning-based strategy update model was established to obtain molding cycle segment division strategies for different types of molds, and the time ratio structure of each strategy group was used as the state input for model recognition. The single time adjustment operation of the pressure rise section, pressure stabilization section and cooling section is set as a discrete action set, including three operation forms: increase, decrease or keep the section time unchanged; Through the mold processing forming rate feedback strategy update model, the strategy update operation is performed in the two-dimensional strategy coordinate space. The segment time ratio parameters of the corresponding coordinate points are corrected and updated through the strategy gradient method. The updated strategy area is generated and injected into the next cycle for use. All strategy update results are recorded as a dynamic strategy map in the form of coordinate points, and a real-time control strategy adjustment network is constructed for the hysteresis properties and thermal inertia windows of different mold types.

2. A multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: The preloading module performs a preset standard low-pressure preloading and hot start on the clamping mold during the molding cycle, and collects the mold press displacement parameters and thermal response curve, specifically including: Initialize the press-fit drive, apply a preset standard low-pressure closing command to the clamped mold, and record the press-fit displacement at the same time; Calculate the change rate of the press-fit displacement over time. When the change rate is continuously converged, it is marked that the mold has reached the closed state. The press-fit displacement change rate sequence and the closing completion time are used as press-fit displacement parameters. When the mold is closed, the heat source is started and the rise of the mold surface temperature over time is monitored to generate a thermal response curve.

3. The multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: The response hysteresis analysis module combines the press-fit displacement parameters to analyze the mold force response speed and structural deformation absorption capacity, and outputs the mold response hysteresis properties, including: Among the press-fitting displacement parameters, the average displacement change rate in the response time interval from preloading start to closing completion is extracted, and the average displacement change rate is defined as the force response speed index of the mold; The average change rate fluctuation amplitude before and after the maximum displacement change rate is extracted, and the structural deformation absorption capacity index of the mold is defined according to the normalized ratio between the force response speed index and the average change rate fluctuation amplitude. The force response speed index and the structural deformation absorption capacity index are used as dual input parameters to perform weighted comprehensive calculations and output the numerical expression of the mold hysteresis property.

4. The multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: The thermal inertia analysis module performs slope change analysis on the thermal response curve to identify the thermal inertia window of the closed mold, specifically including: Obtain a thermal response curve, and smooth the thermal response curve based on a sliding window average method; Preset the slope stable interval, and mark the first time point when the slope of the thermal response curve enters the stable interval as the starting point of the thermal stable section; The time span between the starting point of the thermal response curve and the starting point of the thermal stability section is calculated, and this time span is defined as the thermal inertia window of the closed mold.

5. The multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: The vertical axis of the two-dimensional coordinate system of the strategy space takes the thermal inertia window duration at the first hot start as the maximum time constraint.

6. The multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: The control module converts the time ratio of the pressure rise section, the pressure stabilization section and the cooling section into a section boundary timing sequence, and adjusts the control trigger point of the section within the molding cycle, specifically including: The time ratio of the pressure rise section, the pressure stabilization section and the cooling section is mapped to the total time axis of the molding cycle to form two timing boundaries: the end of the pressure rise section and the end of the pressure stabilization section. The two timing boundaries are injected into the thermal control and pressure control links to reconstruct the control trigger points of the section adjustment channels within the molding cycle.

7. The multi-mode collaborative intelligent forming control system according to claim 1, characterized in that: In the process of executing the strategy update in the two-dimensional strategy coordinate space, a maximum cooling section length compression threshold is preset. When the compression degree exceeds the threshold, the molding cycle length is increased by an equal amount to cover the excess portion.

Images