Intelligent setting method for clamping force of injection molding machine

Abstract

The application relates to the field of data processing, in particular to an intelligent setting method of a mold clamping force of an injection molding machine, which comprises the following steps: collecting pressure and temperature data of each region of a mold cavity and preprocessing the data, constructing pressure sequences and temperature sequences of each region according to the sequence of collection time and forming feature sequences of the regions; dividing the feature sequences into a first time period and a second time period according to the working process of injection molding, calculating target similarity between any two regions according to the feature sequences of the same time period and feature weights of the same time period calculated; calculating a mold clamping force prediction value according to the target similarity, the pressure sequence of the second time period and the projected area of the mold; multiplying the predicted mold clamping force by a preset safety coefficient to obtain an actual mold clamping force before the next injection molding production, and completing intelligent setting based on the actual mold clamping force. Through the intelligent setting method, the mold clamping force can be more reasonable, so that product defects are reduced.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method for intelligently setting the clamping force of an injection molding machine. Background Technology

[0002] Clamping force refers to the force used by an injection molding machine to tightly lock the two halves of the mold during the injection process, preventing the mold from opening due to the high pressure generated by the molten plastic during injection. During injection molding, the molten plastic generates significant pressure after being injected into the mold cavity. If the clamping force is insufficient, the mold may open, leading to quality problems such as flash and dimensional instability in the product. Therefore, in order to ensure production safety and product quality during injection molding, intelligent setting of the clamping force is necessary.

[0003] Moldflow is a software used to simulate the plastic injection molding process. It can predict the flow of plastic melt in the mold, including pressure distribution. However, when predicting clamping force using Moldflow, existing technologies typically use the global average pressure from holding pressure to the completion of injection, ignoring the uneven pressure distribution in different areas of the mold and the complexity of the mold's stress structure. This leads to inaccurate predicted clamping force, which in turn causes deviations in the clamping force applied to the next injection cycle, affecting the stability of injection molding production and product quality. Summary of the Invention

[0004] To address the aforementioned technical problem of inaccurate mold clamping force prediction, the present invention provides the following technical solution.

[0005] A method for intelligently setting the clamping force of an injection molding machine includes:

[0006] The mold cavity is divided into multiple regions. Pressure and temperature data of each region at each time are collected and preprocessed. Based on a single region, a pressure sequence and a temperature sequence of that region are constructed according to the order of collection time. The pressure sequence and temperature sequence constitute the feature sequence of the region.

[0007] The feature sequence is divided into a first time period and a second time period according to the injection molding process. Based on the feature sequence of the same time period and the calculated feature weight of the same time period, the target similarity between any two regions is calculated. The target similarity is used to evaluate the similarity of feature distribution between two regions throughout the injection molding process.

[0008] The predicted clamping force is calculated based on the target similarity, the pressure sequence of the second time period, and the mold projection area.

[0009] Before the next injection molding production, the predicted clamping force is multiplied by the preset safety factor to obtain the actual clamping force, and the intelligent setting is completed based on the actual clamping force.

[0010] Preferably, the acquisition of the pressure and temperature data of each region at each time comprises:

[0011] According to the parting surface area of the mold, the mold is equally divided into regions, and a pressure sensor is arranged at the center of each divided region to collect pressure data in the mold cavity of the mold;

[0012] Synchronously, an infrared sensor is used to collect temperature data of the mold surface. For a region, at a certain time, the average value of the temperature values of all positions in the region is taken to obtain the temperature value of the region at that time.

[0013] Preferably, the first time period is from the start of injection filling to the start of pressure maintaining, and the second time period is from the end of pressure maintaining to the end of filling.

[0014] Preferably, the process of obtaining the feature weight comprises:

[0015] For a time period, the distinctness of the pressure sequence and the distinctness of the temperature sequence in the corresponding time period are respectively calculated based on the Manhattan distance between the feature sequences of all regions in the same time period;

[0016] According to the distinctness of the pressure sequence and the distinctness of the temperature sequence, the feature weight of the pressure sequence and the feature weight of the temperature sequence in the corresponding time period are obtained.

[0017] Preferably, the calculation of the target similarity between any two regions comprises:

[0018] For any two regions, the time period similarity of the two regions in the same time period is calculated based on the feature sequences and the weights of the feature sequences in the same time period;

[0019] Then, the time period similarity of the first time period and the similarity of the second time period are added and normalized to obtain the target similarity between any two regions.

[0020] Preferably, the Manhattan distance of the pressure sequence of the first time period between any two regions is calculated, and the length of the first time period is obtained; the Manhattan distances of the pressure sequence of the first time period between all any two regions are added and then divided by the length of the first time period to obtain the distinctness of the pressure sequence of the first time period.

[0021] According to the distinctness of the pressure sequence of the first time period, the distinctness of the temperature sequence of the first time period, the distinctness of the pressure sequence of the second time period, and the distinctness of the temperature sequence of the second time period can be calculated.

[0022] Preferably, the process of obtaining the time period similarity of the first time period comprises:

[0023] For the pressure sequence of the first period, the difference of the two regions on the feature, i.e. the absolute value difference, is calculated, the calculated difference is divided by the global maximum value of the feature, and the sum of all differences is multiplied by the weight of the feature to obtain the weighted difference sum of the pressure sequence;

[0024] The weighted difference sum of the temperature sequence of the first period is calculated according to the calculation method of the weighted difference sum of the pressure sequence, and the weighted difference sum of the pressure sequence and the weighted difference sum of the temperature sequence are added to obtain the period similarity of the two regions in the first period.

[0025] Preferably, the acquisition process of the mold clamping force prediction value comprises:

[0026] The attention weight of each region is calculated based on the target similarity ratio of each region to the region to which the most remote and the most adjacent sensor belongs;

[0027] The attention weight of each region, the average value of the pressure sequence of the second period, and the projected area of the region are multiplied, and then the sum of all regions is summed to obtain the mold clamping force prediction value.

[0028] The beneficial effects of the present application are:

[0029] The present application can more accurately reflect the actual state of the mold during the injection molding process by dividing the mold cavity into multiple regions, collecting pressure and temperature data of each region, and constructing feature sequences, avoiding errors caused by single data points or overall average values;

[0030] Secondly, the feature sequence is divided into a first period (filling to start of holding pressure) and a second period (holding pressure to end of filling), and feature weight and target similarity are calculated respectively, fully considering the dynamic changes of the injection molding process. The pressure and temperature distribution in the mold is different at different stages. This method can analyze according to the characteristics of each stage, so as to more accurately predict the mold clamping force;

[0031] In calculating the mold clamping force prediction value, target similarity, second period pressure sequence, and mold projected area are considered. Target similarity reflects the similarity of feature distribution between different regions, second period pressure sequence reflects the pressure state in the holding pressure stage, and mold projected area is directly related to the size of the mold clamping force. By combining these factors organically, the demand for mold clamping force can be more comprehensively evaluated, avoiding inaccurate setting due to neglecting some important factors, so as to ensure that the actual mold clamping force meets the production demand, ensuring the safety and stability of the injection molding process, improving product quality and production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1is a method flow chart of steps S1-step S4 in an intelligent setting method of mold clamping force of an injection molding machine. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application.

[0034] Reference Figure 1 An intelligent setting method of mold clamping force of an injection molding machine includes steps S1-step S4, and specifically as follows:

[0035] S1: divide the mold cavity into multiple regions, collect pressure and temperature data of each region at each time and pre-process, construct pressure sequence and temperature sequence of the region based on single region according to the order of collection time, and the pressure sequence and temperature sequence constitute the feature sequence of the region.

[0036] In one embodiment, first, based on the area of the mold parting surface, the mold cavity is divided into multiple regions with equal area, the parting surface is a plane in the mold for dividing the mold into two parts to take out the molded product, and the area of the parting surface is used as the basis for dividing the region, in order to make the divided regions evenly distributed in the mold cavity and representative.

[0037] Then, a pressure sensor is placed in the mold cavity corresponding to the center position of each region, and the pressure data of the region is collected in real time, and an infrared temperature sensor is used to collect the temperature data of the surface of the mold cavity. For a region, at a certain time, the average value of the temperature values of all positions in the region is taken to obtain the average temperature value of the region at that time. In this way, the temperature sequence of a region over time can be obtained.

[0038] Further, wavelet transform is used to denoise the collected pressure and temperature data.

[0039] Finally, the pressure sequence and temperature sequence of each region are obtained according to the collection time sequence, and the pressure sequence and temperature sequence of each region are combined together to constitute the feature sequence of the region.

[0040] It should be noted that the projection of the parting surface of the injection mold is usually irregularly shaped. In order to reasonably set the sensor on these irregular shapes, first, find the smallest rectangle containing the projection of the parting surface, and when the region center is not on the projection, select the position closest to the region center on the edge of the projection as the sensor position. In this way, it can be ensured that the sensor can effectively collect data, while avoiding inaccurate data collection due to improper sensor position.

[0041] S2: dividing the feature sequence into a first period and a second period according to the working process of injection molding, and calculating the target similarity of any two regions in the same period according to the feature sequence of the same period and the feature weight of the same period calculated; the target similarity is used to evaluate the feature distribution similarity of the two regions in the entire injection molding process.

[0042] The injection molding machine injects multiple stages, and the pressure distribution of different regions is different. Only analyzing the instantaneous pressure distribution in the mold after injection molding to identify the easy and difficult edge regions and predict the locking force has defects, and ignoring the influence of temperature in different stages will lead to incorrect region identification, for example, mistakenly considering the edge region with high pressure and high temperature in different stages as a difficult edge region as the center region of the injection point with high pressure, thereby causing the mold to have an edge phenomenon. Therefore, the pressure and temperature characteristics should be considered comprehensively, the target similarity is calculated to evaluate the feature distribution similarity of the two regions in the entire injection molding process, and then the possibility of region abnormality (such as edge phenomenon) is judged.

[0043] In one embodiment, the injection molding process is divided into two periods: the first period from filling to start of holding, and the second period from holding to end of filling, and then for the feature sequence of each region, the feature sequence of the first period and the second period can be obtained.

[0044] In the filling stage, the edge region may have high pressure due to the fast flow of plastic, but in the holding stage, the temperature of the edge region may be lower, causing the plastic to cool faster. The center region may have lower pressure in the filling stage, but in the holding stage, the temperature may be higher due to heat concentration. Therefore, by dividing the injection molding process into a first period and a second period, the pressure and temperature changes in different stages can be more comprehensively analyzed, and the feature distribution of each region can be more accurately identified and evaluated. This method not only considers the two-dimensional features of pressure and temperature, but also eliminates the influence of feature changes in different stages through period-by-period analysis, improving the identification accuracy of abnormal regions (such as edge phenomenon).

[0045] In order to measure the difference degree of the features of different regions in a certain period, in one embodiment, the Manhattan distance is used to calculate the discrimination degree of the pressure sequence and the discrimination degree of the temperature sequence in the same period, and the discrimination degree of the pressure sequence and the discrimination degree of the temperature sequence are used to more accurately identify and distinguish the features of different regions.

[0046] For example, taking the pressure sequence of the first period as an example, the discrimination degree of the pressure sequence satisfies the following relationship:

[0047]

[0048] In the formula, is the discrimination degree of the pressure sequence of the first period, represents the region and regions Manhattan distance of the first period pressure sequence, is the length of the first period (i.e., the number of time points), used to eliminate the influence of different period lengths on the Manhattan distance, is the total number of regions.

[0049] The distinctness of the first period temperature sequence is calculated in the same way as the distinctness of the first period pressure sequence, and the distinctness of the second period pressure sequence and the distinctness of the temperature sequence are calculated.

[0050] Further, the importance of features may change in different periods. For example, certain features may have strong explanatory power for the target in a certain period, but may become less important in another period. By comparing the distinctness of a feature with the distinctness of other features and calculating the weight, the relative importance of each feature in a certain period can be quantified. For example, in the first period, if the distinctness of the pressure sequence is high and the distinctness of the temperature sequence is low, the weight of the pressure sequence in that period will be greater, indicating that it contributes more to the target in that period.

[0051] For example, the ratio of the distinctness of the first period pressure sequence to the sum of the distinctness of the first period pressure sequence and the distinctness of the first period temperature sequence is taken as the feature weight of the first period pressure sequence. Similarly, the feature weight of the first period temperature sequence, and the feature weight of the second period pressure sequence and the feature weight of the temperature sequence are calculated.

[0052] Further, the period similarity between regions in a certain period is calculated, and the specific calculation process is as follows:

[0053] For the pressure sequence of the first period, the difference between the two regions on this feature, i.e., the absolute value difference, is calculated, and the calculated difference is divided by the global maximum value of the feature and summed for all differences, then multiplied by the weight of the feature to obtain the weighted difference sum of the pressure sequence. Similarly, the weighted difference sum of the first period temperature sequence is also calculated, and the weighted difference sum of the pressure sequence and the weighted difference sum of the temperature sequence are added to obtain the period similarity between the two regions in the first period. Similarly, the period similarity between the two regions in the second period can be obtained.

[0054] The above fusion of temperature and pressure dimensions can more comprehensively and accurately identify the differences between regions, improve the identification accuracy of abnormal regions, and avoid misjudgment or omission caused by single-dimensional analysis.

[0055] Furthermore, considering that the first time period typically focuses on local or short-term behavior—that is, during injection molding, the first time period may only consider the instantaneous pressure change or filling situation when the melt just enters the mold—the similarity of this time period can reflect whether the behavior of a region is consistent in the initial stage, but it may not capture subsequent long-term behavior. However, the second time period focuses more on overall or long-term behavior. The similarity of this time period can reflect whether the overall behavior of a region is consistent throughout the entire injection molding process, but it may not capture local differences in the initial stage. Additionally, during injection molding, the distance from the injection point to different areas of the mold varies, and the injection speed also affects the behavior of the region. For example, a region farther from the injection point may respond more slowly in the initial stage, but if only the similarity of the first time period is considered, it may be misjudged as an anomaly. By combining the similarity of the second time period, the influence of this single factor can be eliminated.

[0056] By calculating the similarity between two regions in the first and second time periods, and then further calculating the target similarity, the main purpose is to comprehensively consider information at different time scales, so as to more comprehensively and accurately reflect the behavioral characteristics and anomalies between regions.

[0057] By region and region For example, the target similarity between these two regions satisfies the following relationship:

[0058]

[0059] In the formula, Indicates the region and region Target similarity, For the region and region Temporal similarity in the first time period For the region and region The similarity of time periods in the second time period This indicates normalization processing.

[0060] Then, the target similarity between any two other regions is obtained according to the above calculation method.

[0061] If the target similarity between two regions is low (close to 0), it indicates that the behavior of the two regions differs greatly between the two time periods, which may indicate an anomaly. If the target similarity is high (close to 1), it indicates that the behavior of the two regions is relatively consistent between the two time periods, which is normal.

[0062] S3: The predicted clamping force is calculated based on the target similarity, the pressure sequence of the second time period, and the mold projection area.

[0063] Because the thickness varies in different parts of the injection mold (such as smooth areas, ribs, pins, etc.), uneven pressure distribution on the mold surface during injection molding is easily caused. Existing technology, Moldflow, usually takes the average value of the pressure field on the entire projected area, ignoring the complex structure inside the mold cavity. This results in local high pressure in non-injection points or edge areas being smoothed by other areas, leading to problems such as misjudgment of flash areas, insufficient clamping force to support local high pressure areas resulting in flash, or excessive clamping force leading to excessive energy consumption.

[0064] By calculating attention weights, regions exhibiting abnormal distribution characteristics can be identified. Because the melt flows outwards during injection molding, regions close to the injection point should have similar distribution characteristics. The greater the difference between a region and its neighbors, and the smaller the difference from more distant regions, the greater the likelihood of abnormal distribution characteristics in that region. Such regions should be given more attention to avoid issues like flash. Using attention weights and a second-time pressure sequence for clamping force prediction allows for a more accurate consideration of pressure distribution in different areas of the mold, thereby improving the accuracy of clamping force prediction, reducing flash occurrence, and lowering energy consumption.

[0065] Specifically, firstly, by calculating the Euclidean distance between each sensor and the injection point, the sensor closest to the injection point (nearest neighbor sensor) and the sensor farthest from the injection point (farthest remote sensor) are identified. When multiple sensors meet the criteria, further filtering is performed based on the distances between the sensors. It's worth noting that during the product design phase, engineers determine the optimal location for the plastic filling inlet based on the product's shape, size, and functional requirements; this location is the injection point.

[0066] Then, based on the region to which each sensor belongs, the target similarity of each region with the region to which its nearest and farthest sensor belongs is determined (calculated based on the target similarity in S2 above).

[0067] Next, based on the target similarity ratio between each region and the regions of its farthest and nearest sensors, the attention weight for each region is calculated. This satisfies the following relationship:

[0068]

[0069] In the formula, For the region Attention weights For the region Similarity to targets in the region where its most remote sensor is located. Indicates the region Similarity to the target in the region of its nearest sensor. This represents an exponential function with the natural constant e as its base.

[0070] When >1, it means the region is more similar to the feature distribution of the remote region, which means that this region has a higher probability of local anomalies and needs more attention. Since the melt flows from the injection point to the surrounding area during injection molding, the feature distribution of the regions close to the injection point should be more similar. If a region has a large difference from the adjacent regions and a small difference from the remote regions, it indicates that the region may have abnormal distribution characteristics and needs to avoid the flash. The exponential function amplifies the difference and enhances the attention of the abnormal region.

[0071] Further, using the attention weight of each region and the average value of the second period pressure sequence, combined with the mold projection area of each region, the mold clamping force prediction value is calculated. Then the mold clamping force prediction value satisfies the relationship as follows:

[0072]

[0073] In the formula, is the mold clamping force prediction value, is the attention weight of the region , is the average value of the second period pressure sequence of the region , is the mold projection area of the region , is the average value of the second period pressure sequence of the region , is the mold projection area of the region (along the projection area in the clamping direction, which determines the range of pressure, this parameter is obtained by existing technology such as CAD model or actual measurement), is the total number of regions.

[0074] Where, by introducing the attention mechanism, the importance of each region in the calculation of the mold clamping force can be more accurately reflected; since the injection molding process is a dynamic process, the pressure will change over time, by considering the average value of the pressure sequence, the pressure distribution in the actual injection molding process can be more comprehensively reflected, and secondly, the above second period usually corresponds to the middle stage of plastic filling the mold, at this time the plastic has partially filled the mold, and the pressure sequence may be more stable, more representative of the average pressure level in the entire injection molding process.

[0075] When calculating the mold clamping force, it is very important to consider the projection area, because the size of the mold clamping force is not only related to the pressure, but also related to the area of the pressure action. If the projection area of a region is larger, then even if the pressure is the same, the mold clamping force received by the region will be larger. Therefore, by multiplying the projection area of each region with the average pressure and attention weight of the region, the contribution of the region to the total mold clamping force can be more accurately calculated.

[0076] S4: Before the next injection production, the predicted clamping force is multiplied by the preset safety factor to obtain the actual clamping force, and intelligent setting is completed based on the actual clamping force.

[0077] In injection production, intelligent setting is achieved by predicting the clamping force and combining the safety factor, which can effectively balance the production efficiency and equipment safety.

[0078] Specifically, first, the safety factor is determined, by considering the fluctuations in material properties, mold wear, injection molding machine precision error and other factors in the production process, and the safety factor can be set according to experience and actual demand, for example, 1.2 or 1.3.

[0079] Then, the clamping force prediction value calculated in S3 above is multiplied by the safety factor to obtain an actual clamping force.

[0080] Finally, according to the calculated actual clamping force, the setting is made in the control system of the injection molding machine. For example, assuming that the clamping force prediction value is 500 and the safety factor is 1.2, the actual clamping force is 600, and then in the control interface of the injection molding machine, the clamping force is set to 600 tons, saved and trial produced.

[0081] It should be pointed out that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. An intelligent setting method of clamp force of an injection molding machine, characterized by, The method comprises the following steps: Divide the mold cavity into multiple regions, collect pressure and temperature data of each region at each time and preprocess, construct pressure sequence and temperature sequence of a single region according to the sequence of collection time, and the pressure sequence and temperature sequence constitute the feature sequence of the region; According to the working process of injection molding, the feature sequence is divided into a first period and a second period, for a period, the discrimination of the pressure sequence and the discrimination of the temperature sequence are calculated based on the Manhattan distance between the feature sequences of all regions in the same period, and the feature weight of the pressure sequence and the feature weight of the temperature sequence in the corresponding period are obtained according to the discrimination of the pressure sequence and the discrimination of the temperature sequence; For any two regions, the period similarity of the two regions in the same period is calculated based on the feature sequence and the weight of the feature sequence in the same period; Then, the target similarity between any two regions is obtained by adding and normalizing the period similarity of the first period and the similarity of the second period; The attention weight of each region is calculated based on the target similarity ratio of each region to the region to which the most remote and the nearest sensor belongs; The locking force prediction value is obtained by multiplying the average value of the pressure sequence of the second period, the projection area of the region and the attention weight of each region, and then summing all regions. Before the next injection production, the actual locking force is obtained by multiplying the predicted locking force by a preset safety factor, and the intelligent setting is completed based on the actual locking force.

2. The method of claim 1, wherein the method further comprises: The collection of pressure and temperature data of each region at each time comprises: The mold is equally divided according to the parting surface area, and a pressure sensor is arranged at the center of each divided region to collect pressure data in the mold cavity; Synchronously, an infrared sensor is used to collect temperature data on the surface of the mold, and for a region, the average value of the temperature values of all positions in the region at a certain time is obtained to obtain the temperature value of the region at that time.

3. The method of claim 2, wherein the method further comprises: The first period is from the filling of injection molding to the beginning of holding pressure, and the second period is from holding pressure to the end of filling.

4. The method of claim 3, wherein the method further comprises: The Manhattan distance of the pressure sequence of any two regions in the first period is calculated, and the length of the first period is obtained; the Manhattan distance of the pressure sequence of all arbitrary two regions in the first period is added and then divided by the length of the first period to obtain the discrimination of the pressure sequence in the first period. The discrimination of the temperature sequence in the first period can be calculated according to the discrimination of the pressure sequence in the first period, and the discrimination of the pressure sequence and the discrimination of the temperature sequence in the second period.

5. The method of claim 4, wherein the method further comprises: The process of obtaining the period similarity of the first period is as follows: For the pressure sequence of the first period, the difference between the two regions on the feature, i.e. the absolute value difference, is calculated, the calculated difference is divided by the global maximum value of the feature, and the sum of all differences is multiplied by the weight of the feature to obtain the weighted difference sum of the pressure sequence; The weighted difference sum of the temperature sequence in the first period is calculated according to the calculation method of the weighted difference sum of the pressure sequence, and the weighted difference sum of the pressure sequence and the weighted difference sum of the temperature sequence are added to obtain the period similarity of the two regions in the first period.

Images