State determination device and state determination method

Abstract

The state determining device (1) has: a data obtaining section (100) that obtains predetermined physical quantity-related data as data representing a state related to an injection molding machine (4); a characteristic quantity calculating section (110) that calculates a characteristic quantity representing a characteristic of the state of the injection molding machine (4) from the physical quantity-related data; a statistical data calculating section (120) that calculates a statistical quantity as statistical data in accordance with a statistical condition that includes at least a statistical function for calculating a predetermined statistical quantity from a predetermined characteristic quantity; and a state determining section (140) that determines the state of the injection molding machine (4) from variations in a plurality of successive statistical data.

Description

TECHNICAL FIELD

[0001] The present application relates to a state determination device and a state determination method relating to an injection molding machine, and particularly relates to a state determination device and a state determination method that assist in determining the quality of a molded product molded by an injection molding machine. BACKGROUND

[0002] In production of molded products based on an injection molding machine, a determination condition relating to molding is set in advance, and the molded product after molding is determined for quality using the determination condition. For example, if the manufacturing lot of resin as the material of the molded product is switched, the plasticizing state of the resin in the injection cylinder varies, and as a result, sometimes a defect in the molded product occurs. In addition, due to wear of components such as a screw or depletion of lubricating grease to movable portions, sometimes a defect in the molded product occurs. Therefore, determination of the state of the injection molding machine that varies depending on the passage of time and environmental changes is performed based on changes in characteristic amounts such as injection time, peak pressure, metering time, metering position, and the like of the injection process in the molding cycle.

[0003] Compared with the characteristic amounts when the plasticizing state of the resin is optimal, even if some difference in the characteristic amounts occurs, as long as the difference is not significant, the molded product does not necessarily have an abnormality. Therefore, generally, an allowable range is set in the determination condition of the characteristic amounts. For example, in Patent Document 1, it is shown that quality determination is performed based on the maximum value and the minimum value of the measured data detected per molding cycle. In addition, in Patent Documents 2 to 4, it is shown that the characteristic amounts (e.g., actual values / operation data of injection time, peak pressure, metering position, and the like) are calculated from time series data, and based on the allowable range of the reference value, the deviation from the reference value, the average value, the standard deviation, and the like of the calculated characteristic amounts, it is determined whether it is normal (a good product) or abnormal (a defective product), and it is reported as an alarm (the possibility that a product has an abnormality).

[0004] PRIOR ART DOCUMENTS

[0005] PATENT DOCUMENTS

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 02-106315

[0007] Patent Document 2: Japanese Patent Application Laid-Open No. 06-231327

[0008] Patent Document 3: Japanese Patent Application Laid-Open No. 2002-079560

[0009] Patent Document 4: Japanese Patent Application Laid-Open No. 2003-039519 SUMMARY

[0010] PROBLEMS TO BE SOLVED BY THE INVENTION

[0011] The main causes of abnormalities (defects) of the molded product are various, and there are sudden main causes and medium / long-term main causes. As examples of the sudden main causes, there are damage of the sensor, mixing of foreign matter into the movable part, mixing of foreign matter into the production material, operation error of the operator, and the like. On the other hand, as examples of the medium / long-term main causes, there are wear, consumption, deterioration of the mechanism parts (wear of the screw, consumption of the belt, depletion of the grease of the movable part, annual deterioration of the electrical installation, wear of the metal mold, and the like), change of the production environment (deterioration of the production material (resin), change of humidity caused by the season, rainfall, and the like, change of the air temperature in the morning, noon, and evening, and the like), and the like. For example, the change of the air temperature in the morning, noon, and evening has an influence on the temperature control of the heated injection cylinder, and the plasticizing state of the resin in the injection cylinder fluctuates and sometimes causes defects of the molded product.

[0012] Thus, even if the conditions (program, injection speed, and the like) of the operating machine are the same, the characteristic amount calculated from the measured data fluctuates and a deviation occurs due to the influence of the environmental change such as the air temperature and the change over time. In the past, regarding the abnormalities related to the sudden / short-term main causes, the threshold value such as the upper limit value / lower limit value was set to the measured value obtained per molding cycle, or the characteristic amount / calculative amount calculated from the measured value, and the molding state was determined.

[0013] However, there has not been a sufficient response to determine the molding state which changes slowly over a long period of time, grasp the sign of the change of the state which changes a little at a time, and predict the change of the state in the future.

[0014] That is, it is desired to give a notification before the machine is damaged, to notify the state before defects of the molded product occur, and to improve the operation rate by preventive maintenance.

[0015] Means for solving the problem

[0016] The state determination device of the present application calculates the characteristic amount of the time series data (for example, pressure, current, speed, and the like) related to the molding operation of the injection molding machine per molding process (peak value and the like in the molding process), and calculates the statistical amount using the statistical function for the plurality of calculated characteristic amounts. Then, the molding state of the injection molding machine is determined from the fluctuation of the plurality of calculated statistical amounts.

[0017] Also, one embodiment of the present application is a state determination device that determines a state of an injection molding machine, the state determination device including: a data acquisition unit that acquires data on a predetermined physical quantity as data that indicates a state related to the injection molding machine; a feature quantity calculation unit that calculates a feature quantity that indicates a feature of the state of the injection molding machine, based on the data on the physical quantity; a feature quantity storage unit that stores the feature quantity; a statistical condition storage unit that stores a statistical condition that includes at least a statistical function for calculating a predetermined statistical quantity based on a predetermined feature quantity; a statistical data calculation unit that calculates a statistical quantity as statistical data, based on the feature quantity stored in the feature quantity storage unit, with reference to the statistical condition stored in the statistical condition storage unit; a statistical data storage unit that stores the statistical data; and a state determination unit that determines the state of the injection molding machine, based on a change in a plurality of pieces of statistical data that are continuous within the statistical data stored in the statistical data storage unit.

[0018] Another embodiment of the present application is a state determination method that determines a state of an injection molding machine, the state determination method including: a step of acquiring data on a predetermined physical quantity as data that indicates a state related to the injection molding machine; a step of calculating a feature quantity that indicates a feature of the state of the injection molding machine, based on the data on the physical quantity; a step of calculating a statistical quantity as statistical data, based on the calculated feature quantity, in accordance with a statistical condition that includes at least a statistical function for calculating a predetermined statistical quantity based on a predetermined feature quantity; and a step of determining the state of the injection molding machine, based on a change in a plurality of pieces of statistical data that are continuous within the calculated statistical data.

[0019] Effects of the Invention

[0020] According to one embodiment of the present application, determination of a molding state that changes slowly over a long period of time is possible, and a change in a future state can be predicted. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 FIG. 1 is a schematic hardware configuration diagram of a state determination device according to one embodiment of the present application.

[0022] Figure 2 FIG. 2 is a schematic configuration diagram of an injection molding machine.

[0023] Figure 3 FIG. 3 is a schematic functional block diagram of a state determination device according to a first embodiment of the present application.

[0024] Figure 4 FIG. 4 is a diagram that indicates an example of a molding cycle in which one molded product is manufactured.

[0025] Figure 5 FIG. 5 is a diagram that indicates an example of calculation of a feature quantity based on one time series data.

[0026] Figure 6 is a graph indicating an example of calculating a feature quantity from two or more time series data.

[0027] Figure 7 is a graph indicating an example of a statistical condition.

[0028] Figure 8A is a graph indicating a graph in which a feature quantity of each injection is plotted.

[0029] Figure 8B is a graph indicating a graph in which a statistical data calculated from a feature quantity is plotted.

[0030] Figure 9 is a graph indicating an example of statistical data stored in a statistical data storage section.

[0031] Figure 10 is a functional block diagram of a state determination section in a case where a state of an injection molding machine is determined by statistical analysis.

[0032] Figure 11 is a graph indicating an example of a determination condition.

[0033] Figure 12 is a functional block diagram of a state determination section in a case where a state of an injection molding machine is determined by machine learning.

[0034] Figure 13 is a graph indicating an example of a learning model.

[0035] Figure 14 is a graph indicating an example of an input screen of a statistical condition. DETAILED DESCRIPTION

[0036] Hereinafter, an embodiment of the present application will be described with reference to the accompanying drawings. Figure 1 Hereinafter, an embodiment of the present application will be described with reference to the accompanying drawings.

[0037] Figure 1 is a functional block diagram of a state determination section in a case where a state of an injection molding machine is determined by statistical analysis.

[0038] The state determination device 1 of this embodiment has a CPU 11, which is a processor that controls the state determination device 1 as a whole. The CPU 11 reads the system program stored in the ROM 12 via the bus 22 and controls the state determination device 1 as a whole according to the system program. Temporary calculation data, display data, and various data input from the outside are temporarily stored in the RAM 13.

[0039] The non-volatile memory 14 is composed of, for example, a memory backed up using a battery (not shown) or an SSD (SolidState Drive), and maintains its storage state even when the power supply to the state determination device 1 is disconnected. The non-volatile memory 14 stores data read from the external device 72 via interface 15, data input from the input device 71 via interface 18, and data obtained from the injection molding machine 4 via network 9. The stored data may include, for example, data related to physical quantities detected by various sensors 5 installed on the injection molding machine 4 controlled by the control device 3, such as motor current, voltage, torque, position, speed, acceleration, pressure inside the metal mold, temperature of the injection cylinder, resin flow rate, resin flow velocity, vibration of the drive unit, and sound. The data stored in the non-volatile memory 14 can also be expanded in RAM 13 during execution / use. Furthermore, various system programs, such as known parsing programs, are pre-written in ROM 12.

[0040] Interface 15 is used to connect the CPU 11 of the status determination device 1 to an external device 72, such as an external storage medium. It can read, for example, system programs, programs related to the operation of the injection molding machine 4, parameters, etc., from the external device 72. Furthermore, data created / edited on the status determination device 1 can be stored via the external device 72 in an external storage medium such as a CF card or USB memory (not shown).

[0041] Interface 20 is used to connect the CPU of the status determination device 1 to a wired or wireless network 9. Network 9 can communicate using technologies such as RS-485 serial communication, Ethernet communication, optical communication, wireless LAN, Wi-Fi, and Bluetooth. Network 9 connects to the control device 3 that controls the injection molding machine 4, the fog computer 6, the cloud server 7, etc., and exchanges data with the status determination device 1.

[0042] Data read into the memory, data obtained as a result of executing programs, etc., are output and displayed on the display device 70 via interface 17. In addition, the input device 71, which consists of a keyboard, pointer devices, etc., transmits instructions and data based on the operator's operation to the CPU 11 via interface 18.

[0043] Figure 2This is a schematic structural diagram of the injection molding machine 4. The injection molding machine 4 mainly consists of a mold clamping unit 401 and an injection unit 402. The mold clamping unit 401 has a movable pressure plate 416 and a fixed pressure plate 414. In addition, a movable side metal mold 412 is mounted on the movable pressure plate 416, and a fixed side metal mold 411 is mounted on the fixed pressure plate 414. On the other hand, the injection unit 402 consists of an injection cylinder 426, a hopper 436 for accumulating resin material supplied to the injection cylinder 426, and a nozzle 440 provided at the front end of the injection cylinder 426. In the molding cycle of manufacturing one molded article, in the mold clamping unit 401, the mold closing / molding action is performed by moving the movable pressure plate 416, and in the injection unit 402, the nozzle 440 is pressed against the fixed side metal mold 411 to inject resin into the metal mold. These actions are controlled by commands from the control device 3.

[0044] In addition, sensors 5 are installed in various parts of the injection molding machine 4 to detect physical quantities such as motor current, voltage, torque, position, speed, acceleration, internal pressure of the metal mold, temperature of the injection cylinder 426, resin flow rate, resin flow velocity, vibration of the drive unit, and sound, and transmit these data to the control device 3. In the control device 3, the detected physical quantities are stored in RAM, non-volatile memory, etc. (not shown), and sent to the status determination device 1 via network 9 as needed.

[0045] Figure 3 The functions of the state determination device 1 according to the first embodiment of the present invention are shown in a schematic block diagram. The functions of the state determination device 1 of this embodiment are described by... Figure 1 The state determination device 1 shown has a CPU 11 that executes a system program to control the actions of each part of the state determination device 1.

[0046] The state determination device 1 of this embodiment includes: a data acquisition unit 100, a feature quantity calculation unit 110, a statistical data calculation unit 120, and a state determination unit 140. Furthermore, in the RAM 13 or non-volatile memory 14 of the state determination device 1, the following are pre-prepared: an acquisition data storage unit 300 for storing data acquired by the data acquisition unit 100 from the control device 3, etc.; a feature quantity storage unit 310 for storing feature quantities calculated by the feature quantity calculation unit 110; a statistical condition storage unit 320 for pre-storing statistical conditions in the calculation of statistical data based on the statistical data calculated by the statistical data calculation unit 120; and a statistical data storage unit 330 for storing statistical data calculated by the statistical data calculation unit 120.

[0047] Data Acquisition Department 100 passed Figure 1The state determination device 1 shown has a CPU 11 that executes a system program read from ROM 12. This primarily involves CPU 11 performing arithmetic processing using RAM 13 and non-volatile memory 14, and input control processing based on interfaces 15, 18, or 20. The data acquisition unit 100 acquires data related to physical quantities detected by sensors 5 mounted on the injection molding machine 4, such as motor current, voltage, torque, position, speed, acceleration, pressure inside the metal mold, temperature of the injection cylinder 426, resin flow rate, resin flow velocity, vibration of the drive unit, and sound. The data acquired by the data acquisition unit 100 can be time-series data representing the value of the physical quantity for each predetermined cycle. When acquiring data related to physical quantities, the data acquisition unit 100 also acquires the production count (injection count) at the time the physical quantity was detected. This production count (injection count) can be the production count (injection count) since the last curing. The data acquisition unit 100 can also directly acquire data from the control device 3 that controls the injection molding machine 4 via network 9. The data acquisition unit 100 can also acquire and store data from external devices 72, fog computers 6, cloud servers 7, etc. The data acquisition unit 100 can also acquire data related to physical quantities for each process constituting one molding cycle of the injection molding machine 4. Figure 4 This is a diagram illustrating the molding cycle for manufacturing one molded product. Figure 4 In this process, the mold closing, mold opening, and ejection processes, which are part of the wireframe manufacturing process, are performed by the operation of the mold closing unit 401. Additionally, the injection, holding, metering, decompression, and cooling processes, which are part of the blank frame manufacturing process, are performed by the operation of the injection unit 402. The data acquisition unit 100 acquires physical quantity-related data in a manner that allows differentiation according to these processes. The physical quantity-related data acquired by the data acquisition unit 100 is stored in the data acquisition and storage unit 300 in association with the production count (injection count) based on the injection molding machine 4.

[0048] Feature Calculation Unit 110 via Figure 1 The state determination device 1 shown has a CPU 11 that executes a system program read from ROM 12. The CPU 11 primarily performs computational processing using RAM 13 and non-volatile memory 14. The feature quantity calculation unit 110 calculates feature quantities (injection time, peak pressure, peak pressure arrival position, metering pressure peak value, metering end position, mold closing time, mold opening time, etc.) of the physical quantities related to the state of the injection molding machine 4, according to the steps constituting the molding cycle of the injection molding machine 4, based on the data related to the physical quantities acquired by the data acquisition unit 100. The feature quantities calculated by the feature quantity calculation unit 110 represent the characteristics of the state of each step of the injection molding machine 4. Figure 5 It is a graph showing the pressure changes during the injection process.Figure 5 t1 represents the start time of the injection process, and t3 represents the end time of the injection process. The pressure begins to rise as resin in the injection cylinder is injected into the metal mold, and is then controlled by the control device 3 of the injection molding machine 4 to reach a predetermined target pressure P1. The predetermined target pressure P is manually set by the operator based on the operator's instructions, visually confirmed on the operation screen of the display device 70, and operated by the input device 71. Figure 5 As shown, the feature quantity calculation unit 110 calculates the peak value of the time series data representing the pressure obtained in the injection process and uses it as a feature quantity of the peak pressure in the injection process. Figure 6 This is a graph showing the changes in pressure and screw position during the injection process. For example... Figure 6 As shown, the feature quantity calculation unit 110 calculates the screw position at the peak pressure arrival time t2 in the injection process based on the peak pressure in the injection process, and uses this as the feature quantity of the peak pressure arrival position in the injection process. Thus, the feature quantity calculated by the feature quantity calculation unit 110 is sometimes calculated based on data related to predetermined physical quantities in a predetermined process, and sometimes based on data related to multiple physical quantities in a predetermined process. The feature quantity calculated by the feature quantity calculation unit 110 is stored in the feature quantity storage unit 310 in association with the production number (injection number) based on the injection molding machine 4.

[0049] Statistical Data Calculation Department 120 Figure 1 The state determination device 1 shown has a CPU 11 that executes a system program read from ROM 12. The CPU 11 primarily performs arithmetic operations using RAM 13 and non-volatile memory 14. The statistical data calculation unit 120 calculates statistical values, or statistical data, of the characteristic quantities representing the state of the injection molding machine 4 calculated by the characteristic quantity calculation unit 110. When calculating the statistical data, the statistical data calculation unit 120 refers to statistical conditions stored in the statistical condition storage unit 320.

[0050] The statistical conditions stored in the statistical conditions storage unit 320 determine the conditions for calculating statistics (e.g., mean, variance, etc.) based on characteristic quantities. Figure 7 These are examples of statistical conditions stored in the statistical condition storage unit 320. For example... Figure 7 As illustrated, a statistical condition is a condition that associates a characteristic quantity with a statistical function used to calculate a statistic based on that characteristic quantity. For example... Figure 7 As shown, statistical conditions can be defined according to the forming process of the forming cycle to which the characteristic quantity belongs. Additionally, as... Figure 7As shown, the statistical conditions can include the sample size of the characteristic quantities used in calculating the statistics. The statistical functions included in the statistical conditions can be, for example, weighted average, arithmetic mean, weighted harmonic mean, harmonic mean, trimmed mean, logarithmic mean, root mean square, minimum value, maximum value, median, weighted median, and most frequent value. Regarding these statistical functions, the injection molding machine 4 can be tested beforehand to analyze the correlation between the molding state of the molded product based on the injection molding machine 4 and the various statistics calculated based on the characteristic quantities. An appropriate function can be selected based on the analysis results. For example, if the maximum value of a predetermined characteristic quantity changes with the molding state of the molded product based on the injection molding machine 4, the maximum value can be selected as the statistical function for calculating the statistic of that characteristic quantity. Alternatively, if multiple characteristic quantities include deviation values ​​that deviate significantly from the average value of the characteristic quantities, the weighted median, most frequent value, etc., which are less affected by the deviation values, can be selected as the statistical function. Furthermore, if the value of a predetermined characteristic quantity deviates with the molding state of the molded product based on the injection molding machine 4, the standard deviation can be selected as the statistical function for calculating the statistic of that characteristic quantity. Furthermore, the statistical function used to handle deviations in the value of the characteristic quantity is not limited to the standard deviation; it can also be variance, mean deviation, coefficient of variation, etc. Thus, among the statistical conditions related to the predetermined characteristic quantity, it is preferable to select a statistical function useful for determining changes in the state of the injection molding machine 4.

[0051] Statistical conditions can also be such as Figure 14 As illustrated, the operator can manually set / update the input device 71 from the operation screen displayed on the display device 70. Figure 14 This display example shows a scenario where the operator selects a weighted average as the statistical function for calculating statistics based on injection time of the characteristic quantity, and a standard deviation as the statistical function for calculating statistics based on the peak pressure arrival position of the characteristic quantity. Furthermore, the sample size used in the statistical function's calculation of the statistics indicates that the injection time of the characteristic quantity is 30 injections and the peak pressure arrival position of the characteristic quantity is 10 injections. As a method for determining the sample size, when the value of the characteristic quantity, such as injection time or peak pressure arrival position in the injection process, changes with a small number of injections, a smaller value is selected as the sample size. Conversely, when the value of the characteristic quantity, such as mold opening time in the mold opening process, changes little over the molding cycle, or when the characteristic quantity, such as injection cylinder temperature, changes slowly over a large number of injections, a larger value, such as 90 injections, is selected as the sample size. In this way, the sample size can be appropriately selected based on how the characteristic quantity changes over the molding cycle (per injection).

[0052] The statistical data calculation unit 120 refers to the statistical conditions stored in the statistical conditions storage unit 320 and calculates statistical data as a statistical quantity of the characteristic quantity stored in the characteristic quantity storage unit 310 at a predetermined time. For example, the statistical data calculation unit 120 can calculate statistical data according to a predetermined molding cycle (every injection, every 10 injections, the number of samples set according to the statistical conditions, etc.). Figure 8A , Figure 8B Examples of statistics representing the location where peak pressure is reached. Figure 8A It plotted a graph showing the characteristic amounts of each injection. Figure 8B It is a chart that plots statistical data calculated based on characteristic quantities. For example... Figure 7 As illustrated, the statistical conditions (statistical condition number 3) for calculating the statistical data of the peak pressure arrival position determine the standard deviation as the statistical function and 10 injections as the sample size. At this time, the statistical data calculation unit 120 calculates the standard deviation of the characteristic quantity of the peak pressure arrival position calculated based on the injection for each of the 10 injections, and uses the result as the statistical data of the peak pressure arrival position. Furthermore, in the statistical conditions (statistical condition number 3), the injection process is determined as the molding process to which the characteristic quantity belongs. Therefore, the timing of the statistical data calculation unit 120 calculating the statistical data does not need to overlap with the injection process; that is, the statistical data can be calculated in processes after the injection process, such as the mold opening process, ejection process, etc. (Ref.) Figure 4 The statistical data calculation unit 120 stores the calculated statistical data in the statistical data storage unit 330. Furthermore, when determining the statistical function specified in the statistical conditions, the operator can visually confirm the result. Figure 8A The statistical function is selected appropriately based on the distribution of the characteristic quantities plotted in the graph.

[0053] Figure 9 This represents an example of statistical data stored in the statistical data storage unit 330. Figure 9 In this context, the counts from 1 to n correspond to the number of times the statistical data is calculated. That is, Figure 9 For example, after calculating and storing statistical data, n statistical data points are stored. Furthermore, the statistical data points are arranged in a manner that prioritizes the largest among the subsequently calculated statistical data points. Thus, it is preferable that the statistical data points calculated by the statistical data calculation unit 120 are stored in the statistical data storage unit 330 in a way that allows for control over their calculation order, i.e., obtaining the temporal order of the data related to the physical quantities that form the basis of the calculation. By storing the statistical data in a way that allows for control over the order of the statistical data, predetermined processing can be performed on multiple consecutive statistical data points.

[0054] Status determination unit 140 passes Figure 1The state determination device 1 shown has a CPU 11 that executes a system program read from ROM 12. The CPU 11 primarily performs computational processing using RAM 13 and non-volatile memory 14. The state determination unit 140 determines the state of the injection molding machine 4 based on changes in multiple consecutive statistical data stored in the statistical data storage unit 330. For example, the state determination unit 140 determines the state of the injection molding machine 4 based on changes in the most recent five statistical data points of injection time, metering time, mold closing time, and mold opening time. In another example, the state determination unit 140 determines the state of the injection molding machine 4 based on changes in the most recent five statistical data points of metering pressure peak, metering torque peak, and metering end position.

[0055] The status determination unit 140 can also determine changes by analyzing multiple consecutive statistical data stored in the statistical data storage unit 330. Figure 10 The state determination unit 140 performing statistical analysis is illustrated in a simplified block diagram. The state determination unit 140 performing statistical analysis includes a statistical analysis unit 141 and a determination condition storage unit 142.

[0056] The statistical analysis unit 141 performs statistical analysis on multiple consecutive statistical data based on the decision conditions stored in the decision condition storage unit 142. Figure 11 This represents an example of a decision condition stored in the decision condition storage unit 142. Decision conditions can be defined as a group of statistical data variation conditions and decision results when those conditions are met, based on the decision state. Figure 11 In the example, the determination condition for judging the "state of time related to the molding process" (determination condition number 1) defines a state of "molding time abnormality" if the condition that "the most recent five statistical data points for any one of injection time, metering time, mold closing time, and mold opening time continuously increase monotonically" are met. With this determination condition defined, the statistical analysis unit 141, each time new statistical data is calculated, obtains the five most recent statistical data points for injection time, metering time, mold closing time, and mold opening time, and determines whether the statistical data included in the obtained statistical data points is monotonically increasing. Furthermore, if any one of injection time, metering time, mold closing time, or mold opening time increases monotonically, the state determination unit 140 determines that the molding time is abnormal. Figure 11In another example, the determination condition for judging the "state of the metering process" (determination condition number 3) defines a state of "metering abnormality" if the sum of the five most recent statistical data points for any one of the metering pressure peak, metering torque peak, and metering end position increases by 10%. With this determination condition defined, the statistical analysis unit 141, each time new statistical data is calculated, obtains the five most recent statistical data points for each of the metering pressure peak, metering torque peak, and metering end position, and determines whether the sum of the increases among the statistical data points included in the obtained statistical data is 10% or more. Furthermore, if the sum of any one of the metering pressure peak, metering torque peak, and metering end position increases by 10% or more, the state determination unit 140 determines that there is an abnormality in the metering process.

[0057] The determination result of the status determination unit 140 can be displayed and output to the display device 70. Alternatively, the status determination unit 140 can also send the determination result to a host device such as the control device 3 of the injection molding machine 4, the fog computer 6, or the cloud server 7 via the network 9. Furthermore, if the status determination unit 140 determines an abnormality, it can stop the operation of the injection molding machine 4, reduce its speed, or limit the drive torque of the prime mover driving the drive unit of the injection molding machine 4. Therefore, the operation of the injection molding machine 4 can be stopped before molding defects increase, or a safe standby state can be established to prevent damage to the injection molding machine 4.

[0058] The state determination device 1 of this embodiment, having the above-described structure, can determine the molding state that changes slowly over a long period and can predict future changes in the state. For example, in the case of a sudden impact on sensor 5 or noise applied to the physical quantity detected by sensor 5, the feature quantity calculated by feature quantity calculation unit 110 may contain a deviation value. The statistical data calculated using statistical conditions for the feature quantity containing this deviation value is a value in which the influence of the deviation value of the feature quantity is reduced or the deviation value of the feature quantity is removed. Therefore, the slowly changing molding state can be determined with good accuracy. In addition, in the state determination device 1 of this embodiment, by judging the change state of statistical quantities obtained from multiple molding cycles, the progression of the molding state that changes little by little over time can be grasped, and the signs of an anomaly can be detected before it becomes an anomaly (alarm), and the signs of an anomaly can be notified to the operator. That is, notification can be made before the injection molding machine is damaged and before the molded product is defective, i.e., anomaly detection / prevention maintenance. The presence or absence of an anomaly can be detected before production is stopped due to an anomaly, thus improving operating rate, reducing costs, and improving work efficiency. For example, operators can detect abnormalities before the wear and tear on the screw or metal mold causes poor forming, and can prepare maintenance parts or perform maintenance operations such as replacing the corresponding parts with maintenance parts before the corresponding parts are damaged. Thus, the determination of the presence or absence of abnormalities is not based on the operator's experience and intuition, but on stable and reproducible determination based on numerical information.

[0059] As a variation of the state determination device 1 in this embodiment, the state determination unit 140 may also use machine learning technology to determine changes in a plurality of consecutive statistical data stored in the statistical data storage unit 330. Figure 12 The state determination unit 140 is illustrated in a simplified block diagram when a change determination is made based on an inference result using machine learning techniques. The state determination unit 140, which performs machine learning-based determinations, includes an inference unit 143 and a learning model storage unit 144.

[0060] The estimation unit 143 uses the learning model stored in the learning model storage unit 144 to make a state estimation based on multiple consecutive statistical data. Figure 13This illustrates an example of a learning model stored in the learning model storage unit 144. The learning model stored in the learning model storage unit 144 is a model learned using statistical data calculated from data obtained from both a normally functioning injection molding machine 4 and an injection molding machine 4 exhibiting abnormalities. The learning model can, for example, be a model learned through known supervised learning. In this case, known algorithms such as multilayer perceptrons, regressive coupled neural networks, and convolutional neural networks can be used as the machine learning algorithm. The definition of the label data and the threshold for state determination vary depending on the object of state determination and the type of machine learning algorithm; therefore, appropriate values ​​can be set by repeatedly performing trial actions in advance. For example, in... Figure 13 In the example, the learning model for "state estimation of molding process related time" (learning model number 1) is a learning model obtained by learning from teacher data, using the five most recent statistical data points (injection time, metering time, mold closing time, and mold opening time) obtained in advance from the injection molding machine 4 as input data and the ratio (0-100%) of the increase in molding process related time to normal values ​​as output data (label data). Each time new statistical data is calculated, the estimation unit 143 obtains five most recent consecutive statistical data points for injection time, metering time, mold closing time, and mold opening time, and inputs these statistical data points into the learning model to obtain its output (estimated value of anomaly). Furthermore, if the estimated anomaly is a threshold value of 10 or higher, the state determination unit 140 determines that there is an anomaly in the molding time. Figure 13 In another example, the learning model for "state estimation of the metering process" (learning model number 3) is a learning model obtained by learning from teacher data that uses "the most recent 10 statistical data points of metering pressure peak, metering torque peak, and the most recent 20 statistical data points of metering end position" obtained in advance from the injection molding machine 4 as input data and labels (0-100%) indicating the deviation of the weight of the molded product from the normal value as output data (label data). Each time new statistical data is calculated, the estimation unit 143 obtains the most recent 10 consecutive statistical data points of metering pressure peak, metering torque peak, and the most recent 20 consecutive statistical data points of metering end position, and inputs the obtained statistical data into the learning model to obtain its output (estimated value of anomaly). Furthermore, if the estimated anomaly is a threshold value of 30 or higher, the state determination unit 140 determines that there is an anomaly in the metering process. Thus, the learning model obtained by learning from a sequence of consecutive statistical data points as input data is a model that learns the correlation between the changes in multiple statistical data points and the state of the injection molding machine 4 (state of the molded product).

[0061] The learning model can be, for example, a model based on well-known unsupervised learning. In this case, well-known algorithms such as autoencoders and k-means can be used as the machine learning algorithm. Alternatively, the learning model can be, for example, a model based on well-known reinforcement learning. In this case, well-known algorithms such as Q-learning can be used as the machine learning algorithm.

[0062] The learning model can also be stored in a compressed state in the learning model storage unit 144 and decompressed for use during estimation processing. This allows for efficient use of the state determination device's memory and reduces the amount of memory required, thus offering cost reduction advantages. Alternatively, the learning model can be encrypted and stored in the learning model storage unit 144 and decrypted for use during estimation processing. This results in a state determination device 1 with strong security and information confidentiality.

[0063] Learning models can be created with different characteristics based on the type of learning data and the differences in learning algorithms. Different characteristics and differences such as computational load (computation time), the accuracy of the estimated value, and robustness (stability, robustness) to time series data can also be considered to prepare different learning models and use them appropriately. In this case, multiple different learning models are pre-made for the state to be determined. For example, if the computational load of state determination device 1 is high, a learning model with low computational load can be selected; or if the accuracy of the estimated value is required, a learning model with high estimation accuracy despite high computational load can be selected. The appropriate learning model can be used according to the situation.

[0064] Thus, the state determination device 1, which utilizes machine learning technology, is capable of determining a slowly changing, established state over a long period and predicting future state changes. Unlike statistically based analytical methods, machine learning pre-learns the correlation between statistical data and state changes as a learning model, thereby reducing the cost of pre-analyzing this relationship.

[0065] The present invention has been described above as an embodiment of the invention, but the present invention is not limited to the example of the above embodiment and can be implemented in various ways by applying appropriate modifications.

[0066] For example, when multiple injection molding machines 4 are interconnected via a network 9, data can be obtained from multiple injection molding machines and the status of each injection molding machine can be determined by a status determination device 1. Alternatively, the status determination device 1 can be configured on each control device of the multiple injection molding machines, and the status of each injection molding machine can be determined by the status determination device of each injection molding machine.

[0067] Symbol Explanation

[0068] 1. Status determination device;

[0069] 2. Machine learning device;

[0070] 3. Control device;

[0071] 4. Injection molding machine;

[0072] 5 sensors;

[0073] 6. Fog Computer;

[0074] 7. Cloud servers;

[0075] 9. Network;

[0076] 11 CPUs;

[0077] 12 ROM;

[0078] 13 RAM;

[0079] 14. Non-volatile memory;

[0080] Interfaces 15, 17, 18, and 20;

[0081] 22 bus;

[0082] 70 Display devices;

[0083] 71 Input device;

[0084] 72. External devices;

[0085] 100 Data Acquisition Department;

[0086] 110 Characteristic Quantity Calculation Unit;

[0087] 120 Statistical Data Calculation Department;

[0088] 140 Status Determination Unit;

[0089] 141 Statistical Analysis Department;

[0090] 142. Decision condition storage unit;

[0091] 143. Presumption section;

[0092] 144 Learning model storage unit;

[0093] 300 Acquire data storage department;

[0094] 310 Feature storage unit;

[0095] 320 Statistical Conditions Storage Department;

[0096] 330 Statistical Data Storage Department.

Claims

1. A state determination device for determining the state of an injection molding machine, characterized in that, The state determination device has: The data acquisition unit acquires data related to predetermined physical quantities as data representing the state of the injection molding machine. The feature quantity calculation unit calculates feature quantities representing the state of the injection molding machine based on data related to the physical quantities. A feature storage unit that stores the feature values; A statistical condition storage unit stores statistical conditions, which at least include statistical functions for calculating predetermined statistical quantities based on predetermined characteristic quantities; The statistical data calculation unit calculates statistical quantities as statistical data based on the feature quantities stored in the feature quantity storage unit and with reference to the statistical conditions stored in the statistical condition storage unit. The statistical data storage unit stores the statistical data; as well as The status determination unit determines the status of the injection molding machine based on the changes in multiple consecutive statistical data stored in the statistical data storage unit.

2. The state determination device according to claim 1, characterized in that, The state determination unit has: The determination condition storage unit stores determination conditions for determining the state of the injection molding machine; as well as The statistical analysis unit analyzes, in a statistical manner, whether multiple consecutive statistical data stored in the statistical data storage unit satisfy the judgment conditions stored in the judgment condition storage unit. The state determination unit determines the state of the injection molding machine based on the analysis results of the statistical analysis unit.

3. The state determination device according to claim 2, characterized in that, The determination criteria define any one of the following conditions related to a series of consecutive statistical data: the number of monotonically increasing data points, the number of monotonically decreasing data points, the rate of increase, and the rate of decrease.

4. The state determination device according to claim 1, characterized in that, The state determination unit has: The learning model storage unit stores a learning model that learns the correlation between multiple consecutive statistical data within the statistical data calculated by the statistical data calculation unit and the state of the injection molding machine when the statistical data was calculated. as well as The estimation unit estimates the state of the injection molding machine using the learning model based on multiple consecutive statistical data stored in the statistical data storage unit.

5. The state determination device according to claim 4, characterized in that, The learning model is a model that has been learned through at least one of the following learning methods: supervised learning, unsupervised learning, and reinforcement learning.

6. The state determination device according to claim 1, characterized in that, The statistical function is any one of variance, standard deviation, mean deviation, coefficient of variation, weighted average, weighted harmonic average, trimmed average, root mean square, minimum, maximum, most frequent value, and weighted median.

7. The state determination device according to claim 1, characterized in that, The result of the determination by the status determination unit is displayed and output to the display device.

8. The state determination device according to claim 1, characterized in that, If the state determination unit determines that the injection molding machine is in an abnormal state, it outputs at least one of the following signals: stopping the operation of the injection molding machine, slowing down the operation, or limiting the driving torque of the prime mover driving the injection molding machine.

9. The state determination device according to claim 1, characterized in that, The data acquisition unit is connected via a wired or wireless network and acquires data from multiple injection molding machines.

10. The state determination device according to claim 1, characterized in that, The status determination device is installed on a host device connected to the injection molding machine via a wired or wireless network.

11. A method for determining the state of an injection molding machine, characterized in that, The state determination method performs the following steps: The step of obtaining data related to predetermined physical quantities as data representing the state of the injection molding machine; The step of calculating the characteristic quantity representing the state of the injection molding machine based on the data related to the physical quantity; The step of calculating a statistic as statistical data based on the calculated characteristic quantity according to statistical conditions, wherein the statistical conditions include at least a statistical function for calculating a predetermined statistic based on a predetermined characteristic quantity; as well as The step of determining the state of the injection molding machine based on the changes in multiple consecutive statistical data within the calculated statistical data.

Images