Method and apparatus for monitoring a periodic manufacturing process

Abstract

The invention relates to a method for monitoring a periodic manufacturing process having a plurality of cycles, at least one single product being manufactured in each cycle; wherein, in a first method step, at least one time sequence of sensor data of the manufacturing process is automatically provided for at least one cycle, the sensor data having an effect relationship with the stability and anomalies of the manufacturing process; in a second method step, the stability of the manufacturing process is determined for the period, and the sensor data is marked as stable sensor data when the manufacturing process is stable; in the third method step, the stable sensor data is subjected to automatic dimensionality reduction to obtain point data; in a fourth method step, a density distribution of the point data is automatically formed, which density distribution has at least one stable region of the point data and at least one abnormal region of the point data, and the point data in the abnormal region corresponds to an abnormality of the manufacturing process.

Description

Technical Field

[0001] The present invention relates to a method and a device for monitoring a cyclic manufacturing process. Background Art

[0002] A cyclic manufacturing process is characterized by repetitive actions during the manufacturing process of a single product. These actions are fully executed and repeated in cycles. By continuously repeating these actions, a large number of identical single products can be produced, thereby reducing the manufacturing cost. Examples of cyclic manufacturing processes include forming processes such as casting, injection molding, sintering, etc. In the following, the present invention will be described by way of an injection molding process, but this does not limit the application of the present invention in other forming processes.

[0003] In an injection molding process, an injection molding machine is used to produce single products. For this purpose, each injection molding machine has an injection molding die with at least one cavity. The material is injected into the cavity of the injection molding die under pressure. The injected material is plasticized and forms the shape of the cavity. The injected material cools and forms a single product. This injection molding process needs to be monitored. Here, the internal pressure in the cavity of the injection molding die is an important parameter affecting the quality of the single product. If the single products meet at least one quality characteristic in a quality inspection, they are good products. Otherwise, the single products are defective products. Abnormal situations in the injection molding process, such as too high or too low internal pressure in the die, will result in defective products, thereby increasing the manufacturing cost and should be avoided.

[0004] For this purpose, patent document WO2022 / 258239A1 discloses a method for determining an abnormality in a cyclic manufacturing process (such as an injection molding process). For each cycle, the time series of the pressure values of the internal pressure in the die is reduced to at least one characteristic value, which is automatically segmented in a decision tree for multiple cycles. Assuming that good products are very similar in nature, while defective products are more different from each other, good products cannot be segmented as quickly as defective products in the decision tree. Therefore, the decision tree for good products has a relatively large depth. Then, in order to determine an abnormality in the injection molding process, for the current cycle, the depth of the decision tree for the current characteristic value is compared with the average depth of the decision tree for the same characteristic value in the previous cycle. When the depth is small, the current cycle has an abnormality in the injection molding process. Summary of the Invention

[0005] The object of the present invention is to simplify and improve the method for monitoring a cyclic manufacturing process known from document WO2022 / 258239A1. The present invention also provides a device using this simplified and improved method.

[0006] The object of the present invention is achieved by a method and a device for monitoring a cyclic manufacturing process according to the present invention.

[0007] The present invention relates to a method for monitoring a periodic manufacturing process having a plurality of cycles, wherein at least one single-piece product is produced in each cycle; wherein, in a first method step, for at least one cycle, at least one time series of sensor data of the manufacturing process is automatically provided, and the sensor data has an effective relationship with the stability and anomalies of the manufacturing process; in a second method step, it is determined that the manufacturing process is stable for the cycle, and when the manufacturing process is stable, the sensor data is marked as stable sensor data; in a third method step, the stable sensor data is automatically reduced to point data; and in a fourth method step, a density distribution of the point data is automatically formed, the density distribution includes at least one stable region of the point data and at least one anomaly region of the point data, and the point data in the anomaly region corresponds to an anomaly of the manufacturing process.

[0008] The present invention also relates to a device for performing a method for monitoring a periodic manufacturing process having a plurality of cycles, the device manufacturing at least one single-piece product in each cycle. The device includes: at least one sensor unit that generates and automatically provides at least one time series of sensor data for at least one cycle in order to perform the first method step, and the sensor data has an effective relationship with the stability and anomalies of the manufacturing process; at least one evaluation unit in which a computer program is loaded, and the loaded computer program causes the evaluation unit to automatically load the sensor data, wherein the loaded computer program drives the evaluation unit to mark the sensor data as stable sensor data when it is determined that the manufacturing process is stable in order to perform the second method step; wherein the loaded computer program drives the evaluation unit to automatically reduce the stable sensor data to point data in order to perform the third method step; and wherein the loaded computer program drives the evaluation unit to automatically form a density distribution using the point data in order to perform the fourth method step, the density distribution has at least one stable region of the point data and at least one anomaly region of the point data, and the point data in the anomaly region corresponds to an anomaly of the manufacturing process.

[0009] Contrary to the teachings of document WO2022 / 258239A1, instead of inferring anomalies in the manufacturing process based on the depth of the decision tree of eigenvalue, the present invention has chosen another method, in which, when it is determined that the manufacturing process is stable, the sensor data of the manufacturing process is labeled as stable sensor data and reduced to point data, and this point data is then divided into a stable region and an abnormal region according to the density distribution. The density distribution is a frequency distribution known in mathematical statistics. The density distribution describes the degree of density of the point data distributed around the average value of the point data. The point data close to the average value is located in the stable region. On the contrary, the point data far from the average value is located in the abnormal region. The advantage of the present invention is that once the stable region and the abnormal region of the point data are formed, in subsequent cycles, without determining the stability of the manufacturing process and without labeling the stable sensor data, the next sensor data can be reduced to point data, and it can be determined whether an anomaly occurs when manufacturing subsequent individual products only based on their positions in the stable region or the abnormal region. The dimensionality reduction of the sensor data can be simply, quickly, and automatically performed without much computational effort. Forming the density distribution also does not require much computational effort and can also be simply, quickly, and automatically performed. Therefore, this method can be economically and efficiently integrated into the existing periodic manufacturing process and provide real-time quality prediction for the manufactured individual products, without having to check whether the individual products meet the quality characteristics through separate method steps, thus saving workload and cost.

[0010] A further development of the subject matter of the present invention is presented below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] An exemplary description of the present invention will be made below with reference to the drawings. Among them

[0012] Figure 1 A part of an injection molding machine 1 for monitoring a periodic manufacturing process is schematically shown;

[0013] Figure 2 Shows the use according to Figure 1 Three in-mold pressure curves Y1, Y2, Y3 of a periodic manufacturing process performed by the injection molding machine 1;

[0014] Figure 3 Shows the use Figure 1 A flowchart of a plurality of method steps M1 to M7 of a method M for monitoring a periodic manufacturing process in the case of the injection molding machine 1 shown;

[0015] Figure 4 Shows the illustration of dimensionality reduction and determination of a stable region SA and an abnormal region AA in method steps M3 and M4 of the method M according to Figure 3 ;

[0016] Figure 5shows a flowchart of a plurality of method steps M8 to M13 of a method M for monitoring a periodic manufacturing process in the case of using Figure 1 the injection molding machine 1 shown;

[0017] Figure 6 shows a diagram for determining an effect relationship R between machine setting data MPD and process parameter data PPD in method step M9 of method M according to Figure 5 ;

[0018] Figure 7 shows a diagram for determining stable machine setting data SMPD and abnormal machine setting data AMPD in method step M11 of method M according to Figure 5 ;

[0019] Figure 8 shows a diagram for sorting residuals RE according to dimension SZ and symbol SI in method step M13 of method M according to Figure 5 ; and

[0020] Figure 9 shows a flowchart of a plurality of method steps M14 to M19 of a method M for monitoring a periodic manufacturing process in the case of using Figure 1 the injection molding machine 1 shown.

[0021] Like reference numerals denote like objects in the drawings.

[0022] The list of reference numerals is as follows:

[0023] 1: Injection molding machine

[0024] 10: Injection device

[0025] 11: Injection molding die

[0026] 12: Control unit

[0027] 13: Sensor unit

[0028] 14: Evaluation unit

[0029] 100: Screw

[0030] 101: Nozzle

[0031] 110: Cavity

[0032] 140: Data processor

[0033] 141: Data memory

[0034] 142: Output unit

[0035] 143: Input unit

[0036] AA: Abnormal area

[0037] AN: Abnormal

[0038] ANI: Abnormal information

[0039] AMPD: Abnormal machine setting data

[0040] APPD: Abnormal process parameter data

[0041] CP: Computer program

[0042] D: Device

[0043] DD: Density distribution

[0044] E: Error

[0045] EC: Error category

[0046] ECI: Error category information ED: Error data

[0047] EM: Error model

[0048] EXD: Error sensor data g: Error index

[0049] GP: Good product

[0050] h: Process parameter index

[0051] i: Machine setting parameter index I: Injection stage

[0052] Ⅱ: Holding pressure stage

[0053] Ⅲ: Cooling stage

[0054] INT: Integration

[0055] INF: Process information

[0056] j: Sensor data index k: Cycle index

[0057] l: Cycle - number

[0058] m: Sensor data - number M: Method M1 to M19: Method steps MLM: Machine learning model MP: Machine setting parameter MPD: Machine setting data MT: Melt

[0059] MV: Mean value

[0060] n: Machine setting parameter - number o: Process parameter - number P: Mold internal pressure

[0061] PI: Initial mold internal pressure PⅡ: Filling pressure PⅢ: Closing point pressure

[0062] PD, PD’: Point data

[0063] Pmax: Maximum in-mold pressure

[0064] PP: Process parameter

[0065] PPD: Process parameter data

[0066] q: Single-piece product error - number

[0067] R: Effect relationship

[0068] RE: Residual

[0069] RM: Regression model

[0070] SA: Stable region

[0071] SI: Symbol

[0072] SMPD: Stable machine setting data

[0073] SPPD: Stable process parameter data

[0074] ST: Stable

[0075] STD: Stable confirmation data

[0076] SXD: Stable sensor data

[0077] SZ: Size

[0078] t: Time

[0079] tI: Start of injection phase

[0080] tⅡ: Switching time point

[0081] tⅢ: Closing point

[0082] tIV: End of post-cooling phase

[0083] TD: Boundary distance

[0084] W, W’: Single-piece product

[0085] XD, XD’: Sensor data

[0086] Y: In-mold pressure curve

[0087] Z, Z’: Cycle Specific implementation method

[0088] Figure 1Schematically shown is a part of an injection molding machine 1 that is commercially available and known to those skilled in the art for monitoring the injection molding process. The injection molding process is an example of a cyclic manufacturing process. The characteristic of such a manufacturing process is that actions are repeated in a plurality of cycles Z. A single-piece product W is manufactured in each cycle Z. The cycle Z and the single-piece product W are also represented by the cycle index k as Zk, k = 1... l, and Wk, k = 1... l. The cycle index k represents a single cycle Z, and the number of cycles l represents the number of cycles Z. For an injection molding process with a typical cycle duration of 10 seconds, continuous operation of the injection molding machine 1 for 24 hours will produce 6240 single-piece products W.

[0089] Each cycle Zk, k = 1... l includes phases I to III. The first phase I is also called the injection phase I. The second phase II is also called the holding pressure phase II. The third phase III is also called the post-cooling phase III.

[0090] The injection molding machine 1 has at least one injection device 10 as a component, and the injection device has a screw 100 and a nozzle 101. The screw 100 is used to liquefy the material into a melt MT and move it towards the nozzle 101. The melt MT can be made of plastic, metal, ceramic, etc.

[0091] The injection molding machine 1 also has at least one injection molding die 11 as a component, and the injection molding die has at least one cavity 110. In the injection phase I, the melt MT is injected into the cavity 110 through the nozzle 101 under pressure. In the holding pressure phase II, the melt MT injected into the injection molding die 11 solidifies in the cavity 110. In the post-cooling phase III, the substantially solidified melt MT cools in the cavity 110. At the end of the manufacturing process, the manufactured single-piece product Wk, k = 1... l is ejected from the cavity 110.

[0092] The injection molding machine 1 has at least one control unit 12 as a component. The control unit 12 controls the injection molding process through at least one machine setting parameter MP. The machine setting parameter MP is also represented by the cycle index k and the machine setting parameter index i as Mpki, k = 1... l, i = 1... n. The machine setting parameter index i represents a single machine setting parameter Mpki, k = 1... l, i = 1... n, and the number of machine setting parameters n represents the number of machine setting parameters Mpki, k = 1... l, i = 1... n. Specifically, the machine setting parameter Mpki, k = 1... l, i = 1... n includes:

[0093] - The first machine setting parameter Mpki, k = 1... l, i = 1 is the metering speed of the screw 100.

[0094] - The second machine setting parameter Mpki, k = 1... l, i = 2 is the injection speed of the melt MT into the cavity 110.

[0095] - The third machine setting parameter Mpki, where k = 1...l and i = 3, is the switching time point tⅡ from the injection phase Ⅰ to the holding pressure phase Ⅱ.

[0096] - The fourth machine setting parameter Mpki, where k = 1...l and i = 4, is the unloading movement of the screw 100.

[0097] - The fifth machine setting parameter Mpki, where k = 1...l and i = 5, is the holding pressure target value in the holding pressure phase Ⅱ.

[0098] - The sixth machine setting parameter Mpki, where k = 1...l and i = 6, is the temperature of the melt MT.

[0099] - The seventh machine setting parameter Mpki, where k = 1...l and i = 7, is the temperature of the injection molding die 11.

[0100] The control unit 12 generates machine setting data MPD for the machine setting parameters Mpki, where k = 1...l and i = 1...n. The machine setting data MPD is also represented as MPDki, where k = 1...l and i = 1...n, using the cycle index k and the machine setting parameter index i. The machine setting data MPDki, where k = 1...l and i = 1...n, is digital data.

[0101] The injection molding machine 1 has at least one sensor unit 13 as a component. The sensor unit 13 is arranged on the cavity 110. The sensor unit 13 can be a pressure sensor, a temperature sensor, etc. Preferably, the sensor unit 13 is a pressure sensor that measures the time course of the internal mold pressure P in the cavity 110. The pressure sensor can be a piezoelectric pressure sensor, a piezoresistive pressure sensor, etc. Preferably, the pressure sensor is a piezoelectric pressure sensor that is electrically connected to an amplifier unit. The piezoelectric pressure sensor and the amplifier unit generate a time series of sensor data XD for the measured time course of the internal mold pressure P. The sensor data XD is digital data. The sensor data XD is also represented as XDkj with a cycle index k and a sensor data index j, where k = 1...l and j = 1...m. The sensor data index j represents an individual sensor data XDkj, k = 1...l, j = 1...m, and the sensor data - digit m represents the number of the sensor data XDkj, k = 1...l, j = 1...m. The sensor data XDkj, k = 1...l, j = 1...m follow each other successively over time tj, j = 1...m and preferably have a constant time interval between each other. The piezoelectric pressure sensor measures the internal mold pressure P as the polarization charge. An individual sensor data element XDkj, k = 1...l, j = 1...m represents the amount of polarization charge at time tj, j = 1...m. The amount of polarization charge is proportional to the magnitude of the internal mold pressure P. The piezoelectric pressure sensor typically measures the internal mold pressure P with a measurement accuracy of 1%. The piezoelectric pressure sensor measures the internal mold pressure P with a time resolution of less than / equal to 0.01 Hz. For an injection molding process with a typical cycle time of 10 seconds, the piezoelectric pressure sensor measures the internal mold pressure P at least 1000 times and generates at least 1000 time series of sensor data XDkj, k = 1...l, j = 1...m.

[0102] The injection molding machine 1 has at least one evaluation unit 14 as a component. The evaluation unit 14 has at least one data processor 140, at least one data memory 141, at least one output unit 142, and at least one input unit 143. At least one computer program CP is stored in the data memory 141 and can be loaded into the data processor 140. The evaluation unit 14 is connected to the control unit 12 and the sensor unit 13 via signal lines. The evaluation unit 14 receives machine setting data MPDki, k = 1...l, i = 1...n from the control unit 12 via a signal line and receives a time series of sensor data XDkj, k = 1...l, j = 1...m from the sensor unit 13.

[0103] The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to load the time series of sensor data XDkj for cycles Zk, k = 1... l, j = 1... m into the data processor 140 and evaluate the loaded sensor data XDkj, k = 1... l, j = 1... m. The first evaluation result of the sensor data XDkj, k = 1... l, j = 1... m is to determine at least one in-mold pressure curve Yk, k = 1... l, as Figure 2 shown. This in-mold pressure curve Yk, k = 1... l can be displayed on the output unit 142. Preferably, the output unit 142 is a screen so that the operator of the injection molding machine 1 can learn about the in-mold pressure curve Yk, k = 1... l displayed on the screen. The display of the in-mold pressure curve Yk, k = 1... l has a vertical axis and a horizontal axis. The vertical axis represents the in-mold pressure P, and the horizontal axis represents the time t. The area under this in-mold pressure curve Yk, k = 1... l is called the integral INT.

[0104] The injection molding process will be described below based on the in-mold pressure curve Yk, k = 1... l.

[0105] The injection phase I starts at the time point tI with an initial in-mold pressure PI and ends at the time point tII with a filling pressure PII. In the injection phase I, the injection device 10 liquefies the material into a melt MT and presses the melt MT towards the nozzle 101 using the screw 100. The melt MT is injected into the cavity 110 of the injection molding die 11 through the nozzle 101. The higher the injection speed, the faster the melt MT fills the cavity 110. When filling the cavity 110 with the melt MT, the in-mold pressure P rises to the maximum in-mold pressure Pmax in a short time. Shortly before detecting the maximum in-mold pressure Pmax, the cavity 110 is completely filled with the melt MT, and the filling pressure PII is measured. The time point tII when the cavity is completely filled with the melt is the switching time point tII. The injection phase I ends.

[0106] The holding pressure phase II starts at the switching time point tII. In the holding pressure phase II, the injection device 10 applies a holding pressure (Nachdruck) to the melt MT in the cavity 110 at the nozzle 101. At the same time, more melt MT is made to flow into the cavity 110 to compensate for the shrinkage of the cooled melt MT. Here, the screw 100 performs an unloading movement. The holding pressure affects the magnitude of the in-mold pressure P, especially in the range of the maximum in-mold pressure Pmax, within which the melt MT is compressed for a short time and the in-mold pressure P is greater than the filling pressure PII. Here, the control unit 12 reduces the in-mold pressure P to the seal point pressure PIII. The melt MT solidifies and cools in the cavity 110. The holding pressure phase II ends at the time point tIII with the seal point pressure PIII. The holding time affects the rate at which the in-mold pressure P drops.

[0107] The after-cooling phase III starts at time point tIII, and the solidified melt MT is further cooled. The time point tIII is also referred to as the closing point tIII, at which the melt MT solidifies in the region of the nozzle 101 of the injection device 10, such that the melt MT can no longer flow into the cavity 110, and the cavity 110 is sealed. The injection molding die 11 can be cooled by a coolant. The injection molding machine 1 can more or less strongly cool the die temperature of the cavity 110 in the holding pressure phase II and the after-cooling phase III. The after-cooling phase III ends at time point tIV, at which the manufactured single-piece product W is ejected from the cavity 110.

[0108] Another evaluation result of the sensor data XDkj, k = 1...l, j = 1...m is to determine at least one process parameter PP, which can be obtained from the in-die pressure curve Yk, k = 1...l. The process parameter PP is also expressed as PPkh, k = 1...l, h = 1...o by using the cycle index k and the process parameter index m. The process parameter index h represents the individual process parameter PPkh, k = 1...l, h = 1...o, and the process parameter number o represents the number of the process parameters PPkh, k = 1...l, h = 1...o. Specifically, the process parameters PPkh, k = 1...l, h = 1...o include:

[0109] - The first process parameter PPkh, k = 1...l, h = 1 is the maximum in-die pressure Pmax.

[0110] - The second process parameter PPkh, k = 1...l, h = 2 is the integral INT of the in-die pressure curve Yk, k = 1...l.

[0111] - The third process parameter PPkh, k = 1...l, h = 3 is the point data PDk, k = 1...l described below.

[0112] Figure 2 The three in-die pressure curves Yk, k = 1...3 of three cycles Zk, k = 1...3 are shown when manufacturing three single-piece products Wk, k = 1...3. The first in-die pressure curve Y1 is shown as a dotted line. The second in-die pressure curve Y2 is shown as a solid line. The third in-die pressure curve Y2 is shown as a dashed line. The three in-die pressure curves Yk, k = 1...3 have different shapes from each other. Therefore, the first in-die pressure curve Y1 has a slower pressure rise when filling the cavity 110 than the second and third in-die pressure curves Y2, Y3. The third in-die pressure curve Y3 also has a higher maximum in-die pressure Pmax than the first two in-die pressure curves Y1, Y2. And the third in-die pressure curve Y3 has a larger integral INT than the first two in-die pressure curves Y1, Y2.

[0113] The shape of the first die internal pressure curve Y1, where the pressure rises relatively slowly when filling the cavity 110, indicates an anomaly AN in the manufacturing process. Also, the shape of the third die internal pressure curve Y3, which has a relatively high maximum die internal pressure Pmax, indicates an anomaly AN in the manufacturing process. Additionally, the shape of the third die internal pressure curve Y3, which has a relatively large integral INT, indicates an anomaly AN in the manufacturing process. Only the shape of the second die internal pressure curve Y2, where the pressure rise is not relatively slow, the maximum die internal pressure Pmax is not relatively high, and the integral INT is not relatively large, indicates that the manufacturing process exhibits stability ST.

[0114] Figure 3 A flowchart showing multiple method steps M1 to M7 of a method M for monitoring a periodic manufacturing process having multiple cycles Zk, k = 1... l, where at least one single-piece product WZk, k = 1... l is manufactured in each cycle Zk, k = 1... l.

[0115] In a first method step M1, at least one time series of sensor data XDkj, k = 1... l, j = 1... m of the manufacturing process is automatically provided for each cycle Zk, k = 1... l. These sensor data XDkj, k = 1... l, j = 1... m have an effect relationship R with the stability ST and anomaly AN of the manufacturing process. In the case where the manufacturing process exhibits stability ST, the single-piece products WZk, k = 1... l are error-free. In the case where the manufacturing process exhibits an anomaly AN, the single-piece products WZk, k = 1... l are not error-free. To perform the first method step M1, the sensor unit 13 generates the sensor data XDkj, k = 1... l, j = 1... m. The provision of the sensor data XDkj, k = 1... l, j = 1... m is carried out automatically in the evaluation unit 14. The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to automatically load the sensor data XDkj, k = 1... l, j = 1... m into the data processor 140. "Automatically" in the sense of the present invention means that the components of the injection molding machine 1 operate automatically without the intervention of the operator of the injection molding machine 1.

[0116] In a second method step M2, for each cycle Zk, k = 1... l, the stability ST of the manufacturing process is determined. Preferably, the stability ST of the manufacturing process is determined by an operator of the injection molding machine 1. For this purpose, the operator has various options. The operator can visually inspect the individual products WZk, k = 1... l manufactured in a cycle Zk, k = 1... l and determine whether the individual products WZk, k = 1... l are error-free. The operator can also visually inspect the in-mold pressure curves Yk, k = 1... l displayed on the output unit 142 and determine an indication of the stability ST of the manufacturing process based on the shape of the in-mold pressure curves Yk, k = 1... l.

[0117] In the case where the manufacturing process is confirmed to be stable ST, the operator of the injection molding machine 1 generates stability confirmation data STDk, k = 1... l. The stability confirmation data STDk, k = 1... l is digital data. The stability confirmation data STDk, k = 1... l can be generated in various ways. Thus, the operator can input the stability confirmation data STDk, k = 1... l into the evaluation unit 14 via the input unit 143. The input unit 143 is a keyboard, a touch screen, etc. The operator can input the stability confirmation data STDk, k = 1... l by operating a combination of keys on the keyboard or can touch a specific area of the touch screen.

[0118] The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to automatically load the stability confirmation data STDk, k = 1... l into the data processor 140 and, in the case where the manufacturing process is confirmed to be stable ST, mark the sensor data XDkj, k = 1... l, j = 1... m as stable sensor data SXDkj, k = 1... l, j = 1... m.

[0119] In a third method step M3, the stable sensor data SXDkj, k = 1... l, j = 1... m is automatically reduced in dimension to point data PDk, k = 1... l. The dimensionality reduction is performed automatically in the evaluation unit 14. The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to project the stable sensor data SXDkj, k = 1... l, j = 1... m into a space having at least two principal component axes (Hauptkomponentenachsen). These two principal component axes are the result of neural network calculations or the directions of maximum dispersion of the stable sensor data SXDkj, k = 1... l, j = 1... m. The result of this projection is the point data PDk, k = 1... l. The dimensionality reduction can be presented to the operator of the injection molding machine 1 on the output unit 142, as Figure 4As shown. The two principal component axes are plotted as the ordinate and abscissa. The ordinate represents the first process parameter PPkh, k = 1... l, h = 1, which is the maximum in-mold pressure Pmax, as the first principal component. The abscissa represents the second process parameter PPkh, k = 1... l, h = 2, which is the integral INT of the in-mold pressure curve Yk, k = 1... l, as the second principal component. The point data PDk, k = 1... l are plotted as triangles.

[0120] In the fourth method step M4, a density distribution DD of the point data PDk, k = 1... l is automatically formed. The formation of the density distribution DD is carried out automatically in the evaluation unit 14. The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to automatically load the point data PDk, k = 1... l into the data processor 140 and form the density distribution DD using the point data PDk, k = 1... l. The density distribution DD has at least one stable region SA of the point data PDk, k = 1... l and at least one outlier region AA of the point data PDk, k = 1... l. The density distribution is a frequency distribution known in mathematical statistics. The density distribution indicates the degree of concentration of the point data PDk, k = 1... l around the mean value MV of the point data. The point data PDk, k = 1... l close to the mean value MV are located in the stable region. In contrast, the point data PDk, k = 1... l far from the mean value MV are located in the outlier region. The mean value MV can be determined in various ways. Thus, the mean value MV can be the arithmetic mean, geometric mean, etc. However, the mean value MV can also be the value in the point data PDk, k = 1... l that most probably corresponds to the manufacturing process showing stability ST. The point data PDk, k = 1... l in the outlier region AA correspond to the anomalies AN of the manufacturing process.

[0121] The density distribution DD can be presented to the operator of the injection molding machine 1 on the output unit 142, as Figure 4 shown. The density distribution DD is marked as a dashed ellipse. The density distribution DD has the mean value MV of the point data PDk, k = 1... l. The mean value MV is marked as a black circle. In the region near the mean value MV, the spatial distribution of the point data PDk, k = 1... l is the densest. The density distribution DD has a boundary distance TD to the mean value MV. The boundary distance TD is marked as a dotted ellipse. The point data PDk, k = 1... l located outside the boundary distance TD of the mean value MV correspond to the anomalies AN of the manufacturing process. The boundary distance TD can be determined in various ways. Thus, the boundary distance TD can be the Euclidean distance, Mahalanobis distance, etc.

[0122] In a fifth method step M5, at least one time series of further sensor data XD'j, j = 1...m of a manufacturing process of a further single-piece product W' is automatically provided for a further cycle Z'. This further sensor data XD'j, j = 1...m is generated by the sensor unit 13 and automatically provided to the evaluation unit 14. A computer program CP loaded into the data processor 140 drives the evaluation unit 14 to load this further sensor data XD'j, j = 1...m into the data processor 140.

[0123] In a sixth method step M6, the time series of the further sensor data XD'j, j = 1...m of this further single-piece product W' is automatically reduced in dimension to a further point data PD'. A computer program CP loaded into the data processor 140 drives the evaluation unit 14 to project this further sensor data XD'j, j = 1...m into a space having at least two principal component axes. This has been described in a third method step M3 and is shown in Figure 4 This further point data PD' is depicted as a square.

[0124] In a seventh method step M7, it is automatically determined whether this further point data PD' lies in a stable region SA or whether this further point data PD' lies in an anomaly region AA. The determination of the position of this further point data PD' in the stable region SA or in the anomaly region AA is carried out by the evaluation unit 14 and is driven by a computer program CP loaded into the data processor 140. In Figure 4 this case, this further point data PD' lies in the stable region SA. The evaluation unit 14 generates process information INF which states whether there is stability ST or an anomaly AN in the manufacturing process of the further single-piece product W'.

[0125] Figure 5 The flowcharts of a plurality of method steps M8 to M13 of a method M for monitoring a periodic manufacturing process having a plurality of cycles Zk, k = 1...l are shown, where at least one single-piece product WZk, k = 1...l is manufactured in each cycle Zk, k = 1...l.

[0126] In the eighth method step M8, in a plurality of cycles Zk, k = 1... l, at least one machine setting parameter Mpki, k = 1... l, i = 1... n is changed in each cycle Zk, k = 1... l, and for each cycle Zk, k = 1... l of the change in the machine setting parameter, machine setting data MPDki, k = 1... l, i = 1... n of the change in the machine setting parameter is provided. The execution of the eighth method step M8 is carried out by the control unit 12. The control unit provides the machine setting data MPDki, k = 1... l, i = 1... n of the change in the machine setting parameter to the evaluation unit 14 for each cycle Zk, k = 1... l of the change in the machine setting parameter.

[0127] Furthermore, for each cycle Zk, k = 1... l of the change in the machine setting parameter, process parameters PPkh, k = 1... l, h = 1... o of the change in the machine setting parameter are determined, and process parameter data PPDkh, k = 1... l, h = 1... o is provided for the determined process parameters PPkh, k = 1... l, h = 1... o. The determination of the process parameters PPkh, k = 1... l, h = 1... o and the provision of the process parameter data PPDkh, k = 1... l, h = 1... o (which describes the additional principal component directions of the space described in the sixth method step M6) are carried out automatically by the evaluation unit 14 and are driven by a computer program CP loaded into the data processor 140.

[0128] In the ninth method step M9, a regression model RM is provided, and the machine setting data MPDki, k = 1... l, i = 1... n and the process parameter data PPDkh, k = 1... l, h = 1... o are automatically input into the regression model RM. Here, the regression model RM determines the effect relationship Rkih, k = 1... l, i = 1... n, h = 1... o between the machine setting data MPDki, k = 1... l, i = 1... n and the process parameter data PPDkh, k = 1... l, h = 1... o. The execution of the ninth method step M9 by the evaluation unit 14 is driven by a computer program CP loaded into the data processor 140. If only a single combination of the machine setting parameters Mpki, k = 1... l, i = 1... n (parameters unchanged) is provided in a plurality of cycles Zk, k = 1... l, i = 1... n, then point calibration is carried out, in which only the average value and the standard deviation of the process parameter data PPDkh, k = 1... l, h = 1... o are determined.

[0129] Figure 6A graphical illustration for determining the effect relationship Rkih, where k = 1...l, i = 1...n, h = 1...o is shown. The machine setting data MPDki, where k = 1...l, i = 1...n is plotted on the vertical axis, and the process parameter data PPDkh, where k = 1...l, h = 1...o is plotted on the horizontal axis. According to the least squares method, the effect relationship Rkih, where k = 1...l, i = 1...n, h = 1...o is determined between the machine setting data MPDki, where k = 1...l, i = 1...n and the process parameter data PPDkh, where k = 1...l, h = 1...o. Preferably, the regression model RM is a multivariate regression model with multiple independent variables and multiple dependent variables, and there is a substantially linear relationship between these independent variables and dependent variables. In the case of point calibration, the average value PPDkhM of the calibration and its standard deviation are simply determined as the effect relationship.

[0130] In the tenth method step M10, the process parameters PPkh, where k = 1...l, h = 1...o are determined for a plurality of additional cycles Z’k, where k = 1...l, and process parameter data PPDkh, where k = 1...l, h = 1...o is provided for the determined process parameters PPkh, where k = 1...l, h = 1...o. The determination of the process parameters PPkh, where k = 1...l, h = 1...o and the provision of the process parameter data PPDkh, where k = 1...l, h = 1...o are automatically performed by the evaluation unit 14 and driven by a computer program CP loaded into the data processor 140.

[0131] In the tenth method step M10, the stability ST of the manufacturing process is also determined for each additional cycle Z’k, where k = 1...l. Preferably, the determination of the stability ST of the manufacturing process is performed by the operator of the injection molding machine 1, which has been described in the second method step M2 and is hereby referred to. In the case where the manufacturing process exhibits stability ST, the operator of the injection molding machine 1 generates stability confirmation data SDk, where k = 1...l. This has also been described in the second method step M2 and is hereby referred to.

[0132] In the case where the manufacturing process exhibits stable ST, the process parameter data PPDkh, k = 1... l, h = 1... o are automatically labeled as stable process parameter data SPPDkh, k = 1... l, h = 1... o. In the case where the manufacturing process exhibits anomalies AN, the process parameter data PPDkh, k = 1... l, h = 1... o are automatically labeled as anomalous process parameter data APPDkh, k = 1... l, h = 1... o. Labeling the sensor data XDkj, k = 1... l, j = 1... m as stable process parameter data SPPDkh, k = 1... l, h = 1... o or anomalous process parameter data APPDkh, k = 1... l, h = 1... o is performed in the evaluation unit 14 and is driven by a computer program CP loaded into the data processor 140.

[0133] In the eleventh method step M11, the stable process parameter data SPPDkh, k = 1... l, h = 1... o and the anomalous process parameter data APPDkh, k = 1... l, h = 1... o are input into the regression model RM, and through the effect relationship Rkih, k = 1... l, i = 1... n, h = 1... o, stable machine setting data SMPDki, k = 1... l, i = 1... n are determined for the stable process parameter data SPPDkh, k = 1... l, h = 1... o, and anomalous machine setting data AMPDki, k = 1... l, i = 1... n are determined for the anomalous process parameter data APPDkh, k = 1... l, h = 1... o. The stable machine setting data SMPDki, k = 1... l, i = 1... n and the anomalous machine setting data AMPDki, k = 1... l, i = 1... n are also simply referred to as stable machine setting data SMPD and anomalous machine data AMPD in the absence of the cycle index k and the machine setting parameter index i. The evaluation unit 14 executes the eleventh method step M11 driven by a computer program CP loaded into the data processor 140.

[0134] Figure 7 A graphical illustration showing the determination of the stable machine setting data SMPDki, k = 1... l, i = 1... n and the anomalous machine setting data AMPDki, k = 1... l, i = 1... n. From according to Figure 6Starting from the effect relationship Rkih, where k = 1...l, i = 1...n, h = 1...o, the stable process parameter data SPPDkh, k = 1...l, h = 1...o and the abnormal process parameter data APPDkh, k = 1...l, h = 1...o are plotted on the abscissa. According to the effect relationship Rkih, k = 1...l, i = 1...n, h = 1...o, the corresponding stable machine setting data SMPDki, k = 1...l, i = 1...n and the corresponding abnormal machine setting data AMPDki, k = 1...l, i = 1...n are determined on the ordinate. The corresponding relationship is represented by an arrow.

[0135] In the twelfth method step M12, the residual Rekih, k = 1...l, i = 1...n, h = 1...o of the abnormal machine setting data AMPDki, k = 1...l, i = 1...n with respect to the effect relationship Rkih, k = 1...l, i = 1...n, h = 1...o is formed. The residual Rekih, k = 1...l, i = 1...n, h = 1...o is the difference between the abnormal machine setting data AMPDki, k = 1...l, i = 1...n and the effect relationship Rkih, k = 1...l, i = 1...n, h = 1...o. The residual Rekih, k = 1...l, i = 1...n, h = 1...o is also only called the residual RE in the case of no cycle index k, no machine setting parameter index i, and no process parameter index h. In the case of point calibration, the deviation from the calibration center point is provided as a multiple of the determined standard deviation as the residual. This provides the operator of the injection molding machine 1 with information about the direction and magnitude of the parameter deviation in the case of the abnormal AN. The evaluation unit 14 executes the twelfth method step M12 and is driven by the computer program CP loaded into the data processor 140.

[0136] In the thirteenth method step M13, the residuals Rekih, k = 1...l, i = 1...n, h = 1...o are sorted according to the size SZ and the sign SI. The evaluation unit 14 executes the thirteenth method step M13 and is driven by the computer program CP loaded into the data processor 140. The evaluation unit 14 generates the abnormal information ANIk, k = 1...l, which explains which residual Rekih, k = 1...l, i = 1...n, h = 1...o of the machine setting parameter Mpki, k = 1...l, i = 1...n is the highest according to the sorting of the size SZ and the sign SI, and this forms the cause of the abnormal AN of the manufacturing process. In the case of no cycle index k, the abnormal information ANIk, k = 1...l is also called ANI.

[0137] Figure 8Shows an illustration of sorting the residuals Rekih, k = 1... l, i = 1... n, h = 1... o according to the size SZ and the sign SI. In Figure 8 it, the size SZ is plotted on the abscissa and the sign SI is plotted on the ordinate. Four residuals Rekih, k = 1... l, i = 1... n, h = 1... o are shown:

[0138] - The residual Rekih, k = 1... l, i = 5, h = 1... o, which corresponds to the fifth machine setting parameter Mpki, k = 1... l, i = 5, i.e., the temperature of the melt MT, by the machine setting parameter index i = 5, has a negative sign SI and a relatively minimum size SZ.

[0139] - The residual Rekih, k = 1... l, i = 3, h = 1... o, which corresponds to the third machine setting parameter Mpki, k = 1... l, i = 3, i.e., the switching time point tⅡ, by the machine setting parameter index i = 3, has a positive sign SI and a relatively second smallest size SZ.

[0140] - The residual Rekih, k = 1... l, i = 2, h = 1... o, which corresponds to the second machine setting parameter Mpki, k = 1... l, i = 2, i.e., the injection speed of the melt MT into the cavity 110, by the machine setting parameter index i = 2, has a positive sign SI and a relatively second largest size SZ.

[0141] - The residual Rekih, k = 1... l, i = 1, h = 1... o, which corresponds to the first machine setting parameter Mpki, k = 1... l, i = 1, i.e., the metering speed of the screw 100, by the machine setting parameter index i = 1, has a positive sign SI and a relatively largest size SZ.

[0142] Figure 9 Shows a flowchart of a plurality of method steps M14 to M19 of the method M for monitoring a periodic manufacturing process.

[0143] In the fourteenth method step M14, at least one single-piece product WZk, k = 1... l is manufactured in each of the plurality of cycles Zk, k = 1... l, and at least one time series of the sensor data XDkj, k = 1... l, j = 1... m of the manufacturing process is automatically provided for each cycle Zk, k = 1... l. The provision of the sensor data XDkj, k = 1... l, j = 1... m is carried out in the evaluation unit 14. The computer program CP loaded into the data processor 140 drives the evaluation unit 14 to load the sensor data XDkj, k = 1... l, j = 1... m into the data processor 140.

[0144] In the fifteenth method step M15, for a plurality of cycles Zk, k = 1... l, the stability ST or abnormality AN of the manufacturing process is determined. In the case where the manufacturing process exhibits an abnormality AN, at least one error E can be determined. Preferably, a plurality of errors E are predefined. In the case with cycle index k and error index g, the error E is also denoted as Ekg, k = 1... l, g = 1... q. The error index g represents an individual error Ekg, k = 1... l, g = 1... q, and the number of errors q represents the quantity of the errors Ekg, k = 1... l, g = 1... q. Specifically, the errors Ekg, k = 1... l, g = 1... q include:

[0145] - The first error Ekg, k = 1... l, g = 1 is the deviation from the predefined weight of a single-piece product WZk, k = 1... l.

[0146] - The second error Ekg, k = 1... l, g = 2 is the deviation from the predefined dimensional stability (Masshaltigkeit) of a single-piece product WZk, k = 1... l.

[0147] - The third error Ekg, k = 1... l, g = 3 is the deviation from the predefined dimensions of a single-piece product WZk, k = 1... l.

[0148] - The fourth error Ekg, k = 1... l, g = 4 is the burr formation on a single-piece product WZk, k = 1... l.

[0149] - The fifth error Ekg, k = 1... l, g = 5 is the deviation from the predefined filling of the cavity 110 during the injection molding process.

[0150] - The sixth error Ekg, k = 1... l, g = 6 is the burn mark (Brandstelle) on a single-piece product WZk, k = 1... l.

[0151] - The seventh error Ekg, k = 1... l, g = 7 is the blocked cooling channel of the injection molding machine 1.

[0152] - The eighth error Ekg, k = 1... l, g = 8 is the damaged heating belt of the injection molding machine 1.

[0153] - The ninth error Ekg, k = 1... l, g = 9 is the damaged check valve of the screw 100 of the injection molding machine 1.

[0154] - The tenth error Ekg, k = 1... l, g = 10 is the blockage of the nozzle 101 of the injection molding machine 1.

[0155] - The eleventh error Ekg, k = 1...l, g = 11 is the viscosity fluctuation of the melt MT of the injection molding machine 1.

[0156] Preferably, the fifteenth method step M15 is performed by the operator of the injection molding machine 1. Thus, the operator can visually inspect the individual products WZk, k = 1...l manufactured in a cycle Zk, k = 1...l and determine whether the individual products WZk, k = 1...l are error-free. The operator can also inspect the injection molding machine 1 and determine whether the injection molding machine 1 has no errors in the cycle Zk, k = 1…l. For this purpose, the injection molding machine 1 also has additional sensor units and output units (not shown in the figure) for determining blockages in the cooling channels, damage to the heating bands, damage to the check valve of the screw 100, blockages in the nozzle 101, or viscosity fluctuations of the melt MT. In the case of an anomaly AN in the manufacturing process, the operator determines the errors Ekg, k = 1...l, g = 1...q. For the determined errors Ekg, k = 1...l, g = 1...q, the operator generates error data Edkg, k = 1...l, g = 1...q. The operator can input the error data Edkg, k = 1...l, g = 1...q into the evaluation unit 14 via the input unit 143. The error data Edkg, k = 1...l, g = 1...q are digital data.

[0157] In the sixteenth method step M16, an error model EM is provided, which is trained using the error data Edkg, k = 1...l, g = 1...q. Here, the error model EM determines an error class Ecg, g = 1...q for each error index g of the error data Edkg, k = 1...l, g = 1...q. Providing the error model EM is performed in the evaluation unit 14. To determine the error class Ecg, g = 1...q, a computer program CP loaded into the data processor 140 drives the evaluation unit 14 to load the error model EM and the error data Edkg, k = 1...l, g = 1...q into the data processor 140 and train the error model EM using the error data Edkg, k = 1...l, g = 1...q.

[0158] In the seventeenth method step M17, the sensor data XDkj, k = 1... l, j = 1... m of the manufacturing process of the individual products WZk, k = 1... l with the errors Ekg, k = 1... l, g = 1... q or the sensor data XDkj, k = 1... l, j = 1... m of the cycles Zk, k = 1... l of the injection molding machine 1 are classified as error sensor data EXDkjg, k = 1... l, j = 1... m, g = 1... q for the error classes Ecg, g = 1... q determined for the errors Ekg, k = 1... l, g = 1... q, where the injection molding machine 1 has the errors Ekg, k = 1... l, g = 1... q in the cycles Zk, k = 1... l. The seventeenth method step M17 is carried out by the evaluation unit 14 and is driven by a computer program CP loaded into the data processor 140. In the case where there is no cycle index k, no sensor data index j, and no error index g, the error sensor data EXDkjg, k = 1... l, j = 1... m, g = 1... q is also only referred to as error sensor data EXD.

[0159] In the eighteenth method step M18, for another cycle Z', at least one time series of further sensor data XD'j, j = 1... m of the manufacturing process of another individual product W' is provided on the injection molding machine 1. The eighteenth method step M18 is carried out by the evaluation unit 14 and is driven by a computer program CP loaded into the data processor 140.

[0160] In the nineteenth method step M19, the further sensor data XD'j, j = 1... m is input into the error model EM. The error model EM determines whether the further sensor data XD'j, j = 1... m can be classified into the error classes Ecg, g = 1... q. And if the further sensor data XD'j, j = 1... m can be classified into the error classes Ecg, g = 1... q, error class information ECI' is generated, which indicates the error classes Ecg, g = 1... q for the other individual product W' or for the injection molding machine 1. The nineteenth method step M19 is carried out by the evaluation unit 14 and is driven by a computer program CP loaded into the data processor 140.

[0161] Preferably, the process information INF, the exception information ANI, and the error class information ECI' are output on the output unit 142 of the operator of the injection molding machine 1.

Claims

1. A method (M) for monitoring a cyclical manufacturing process having a plurality of cycles (Zk, k=1 ... l), wherein at least one single product (WZk, k=1 ... l) is manufactured in each cycle (Zk, k=1 ... l); in, In a first method step (M1), at least one time series of sensor data (XDkj, k=1 ... l, j=1 ... m) of the manufacturing process is automatically provided for at least one period (Zk, k=1 ... l), the sensor data (XDkj, k=1 ... l, j=1 ... m) having an effect relationship (R) with stability (ST) and anomalies (AN) of the manufacturing process; It is characterized in that In a second method step (M2), a stability (ST) of the manufacturing process is determined for the period (Zk (k=1 ... l)), and if the manufacturing process is stable (ST), the sensor data (XDkj, k=1 ... l, j=1 ... m) are marked as stable sensor data (SXDkj, k=1 ... l, j=1 ... m); In a third method step (M3), the stable sensor data (SXDkj, k=1 ... l, j=1 ... m) are automatically reduced to point data (PDk, k=1 ... l); and In the fourth method step (M4), a density distribution (DD) of the point data (PDk, k=1...l) is automatically formed, the density distribution (DD) includes at least one stable area (SA) of the point data (PDk, k=1...l) and at least one abnormal area (AA) of the point data (PDk, k=1...l), and the point data (PDk, k=1...l) in the abnormal area (AA) corresponds to an anomaly (AN) of the manufacturing process.

2. The method (M) according to claim 1, characterized in that In the second method step (M2), the stability (ST) of the manufacturing process is determined by checking the error-freeness of the individual products (WZk, k=1 . . . l) manufactured in the cycle (Zk, k=1 . . . l); Alternatively, the periodic manufacturing process is an injection molding process, and the time series of sensor data (XDkj, k=1...l, j=1...m) is evaluated to obtain a mold internal pressure curve (Yk, k=1...l), and in the second method step (M2), the stability (ST) of the manufacturing process is determined by examining the shape of the mold internal pressure curve (Yk, k=1...l) of a period (Zk, k=1...l).

3. The method (M) according to claim 1 or 2, characterized in that The stable area (SA) determined in the fourth method step (M4) includes the average value (MV) of the point data (PDk, k=1...l) and the boundary distance (TD) to the average value (MV), and the point data (PDk, k=1...l) of the abnormal area (AA) is located outside the boundary distance (TD) to the average value (MV).

4. The method (M) according to any one of claims 1 to 3, characterized in that In a fifth method step (M5), at least one time series of further sensor data (XD'j, j=1 . . . m) of a production process of a further individual product (W') is automatically provided for a further cycle (Z'); In a sixth method step (M6), the time series of further sensor data (XD'j, j=1...m) of the further individual product (W') is automatically reduced in dimension to further point data (PD'); and In the seventh method step (M7), it is automatically determined whether the other point data (PD') is located in the stable area (SA) or whether the other point data (PD') is located in the abnormal area (AA), and process information (INF) is generated, wherein the process information (INF) indicates whether there is stability (ST) or abnormality (AN) in the manufacturing process of the other single product (W').

5. The method (M) according to claim 4, characterized in that The cyclical manufacturing process is an injection molding process, in which a single product (WZk, k=1...l) is manufactured according to a cycle (Zk, k=1...l) by means of an injection molding machine (1); The injection molding machine (1) comprises at least one injection device (10) and at least one injection molding mold (11), wherein the injection device (10) comprises a screw (100), and the injection molding mold (11) comprises at least one cavity (110); Each cycle (Zk, k=1...l) of the injection molding process comprises an injection phase (I), a holding phase (II) and a residual cooling phase (III), wherein the injection phase (I), the holding phase (II) and the residual cooling phase (III) are controlled by the injection molding machine (1) via at least one of the following machine setting parameters (MPki, k=1...l, i=1...n): The first machine setting parameter (MPki, k=1...l, i=1) is the metering speed of the screw (100) of the injection device (10); The second machine setting parameter (MPki, k=1...l, i=2) is the injection speed of the melt (MT) into the cavity (110) of the injection molding mold (11); The third machine setting parameter (MPki, k=1 ... l, i=3) is the switching time point (tII) from the injection phase (I) to the pressure holding phase (II); A fourth machine setting variable (MPki, k=1 ... l, i=4) is the unloading movement of the screw (100) of the injection device (10); The fifth machine setting parameter (MPki, k=1...l, i=5) is the target value of the holding pressure in the holding pressure stage (II); a sixth machine setting variable (MPki, k=1 ... l, i=6) is the temperature of the melt (MT); and The seventh machine setting parameter (MPki, k=1...l, i=7) is the temperature of the injection molding mold (11).

6. The method (M) according to claim 5, characterized in that The injection molding machine (1) has at least one evaluation unit (14), the evaluation unit (14) has at least one data processor (140) and at least one data memory (141), and at least one computer program (CP) is stored in the data memory (141) and can be loaded into the data processor (140), the computer program (CP) loaded into the data processor (140) drives the evaluation unit (14) to load a time series of sensor data (XDkj, k=1...l, j=1...m) of a period (Zk, k=1...l) into the data processor (140) and evaluate the loaded sensor data (XDkj, k=1...l, j=1...m); wherein, A first evaluation result is the determination of at least one mold internal pressure curve (Yk, k=1 ... l), and a further evaluation result is the determination of at least one of the following process parameters (PPkh, k=1 ... l, h=1 ... o): The first process parameter (PPkh, k=1...l, h=1) is the maximum mold internal pressure (Pmax); The second process parameter (PPkh, k=1...1, h=2) is the integral of the mold internal pressure curve (Yk, k=1...1); and The third process parameter (PPkh, k=1...l, h=3) is the point data (PDk, k=1...l).

7. The method (M) according to claim 6, characterized in that In an eighth method step (M8), at least one machine setting parameter (MPki, k=1...l, i=1...n) is changed in a plurality of cycles (Zk, k=1...l), and for each cycle (Zk, k=1...l) of the machine setting parameter change, machine setting data (MPDki, k=1...l, i=1...n) of the machine setting parameter change are provided, and for each cycle (Zk, k=1...l) of the machine setting parameter change, process parameters (PPkh, k=1...l, h=1...o) of the machine setting parameter change are determined, and process parameter data (PPDkh, k=1...l, h=1...o) are automatically provided for the determined process parameters (PPkh, k=1...l, h=1...o); and In the ninth method step (M9), a regression model (RM) is provided, and the machine setting data (MPDki, k=1...l, i=1...n) and the process parameter data (PPDkh, k=1...l, h=1...o) are automatically input into the regression model (RM), and the regression model (RM) thereby determines the effect relationship (Rkih, k=1...l, i=1...n, h=1...o) between the machine setting data (MPDki, k=1...l, i=1...n) and the process parameter data (PPDkh, k=1...l, h=1...o).

8. The method (M) according to claim 7, characterized in that In a tenth method step (M10), process parameters (PPkh, k=1 ... l, h=1 ... o) are determined for a plurality of further cycles (Z'k, k=1 ... l), and process parameter data (PPDkh, k=1 ... l, h=1 ... o) are automatically provided for the determined process parameters (PPkh, k=1 ... l, h=1 ... o), and for each further cycle (Z'k, k=1 ... l) a manufacturing process stabilization (ST) is determined, in the case of a stable manufacturing process (ST), the process parameter data (PPDkh, k=1 ... l, h=1 ... o) are automatically marked as stable process parameter data (SPPDkh, k=1 ... l, h=1 ... o), in the case of an abnormal manufacturing process (AN), the process parameter data (PPDkh, k=1 ... l, h=1 ... o) are automatically marked as abnormal process parameter data (APPDkh, k=1 ... l, h=1 ... o); In an eleventh method step (M11), the stable process parameter data (SPPDkh, k=1 ... l, h=1 ... o) and the abnormal process parameter data (APPDkh, k=1 ... l, h=1 ... o) are input into the regression model (RM), and corresponding stable machine setting data (SMPDki, k=1 ... l, i=1 ... n) are determined for the stable process parameter data (SPPDkh, k=1 ... l, h=1 ... o) by means of the effect relationship (Rkih, k=1 ... l, i=1 ... n, h=1 ... o), and corresponding abnormal machine setting data (AMPDki, k=1 ... l, i=1 ... n) are determined for the abnormal process parameter data (APPDkh, k=1 ... l, h=1 ... o); In a twelfth method step (M12), a residual (REkih, k=1 ... l, i=1 ... n, h=1 ... o) of the abnormal machine setting data (AMPDki, k=1 ... l, i=1 ... n) with respect to the effect relationship (Rkih, k=1 ... l, i=1 ... n, h=1 ... o) is formed; and In the thirteenth method step (M13), the residuals (REkih, k=1...l, i=1...n, h=1...o) are sorted according to size (SZ) and sign (SI), and exception information (ANIk, k=1...l) is generated, which indicates which machine setting parameter (MPki, k=1...l, i=1...n) has the highest residual (REkih, k=1...l, i=1...n, h=1...o) sorted according to size (SZ) and sign (SI), which forms the cause of the manufacturing process abnormality (AN).

9. The method (M) according to claim 6, characterized in that In a fourteenth method step (M14), at least one individual product (WZk, k=1 ... l) is manufactured in each cycle (Zk, k=1 ... l) of a plurality of cycles (Zk, k=1 ... l), and at least one time series of sensor data (XDkj, k=1 ... l, j=1 ... m) of the manufacturing process is automatically provided for each cycle (Zk, k=1 ... l), In a fifteenth method step (M15), stability (ST) or anomaly (AN) of the manufacturing process is determined, and in the case of anomaly (AN) of the manufacturing process, at least one of the following errors (Ekg, k=1 ... l, g=1 ... q) is determined: The first error (Ekg, k=1 ... l, g=1) is the deviation from a predefined weight of the single product (WZk, k=1 ... l); The second error (Ekg, k=1...l, g=2) is the deviation from the predefined dimensional stability of the single product (WZk, k=1...l); The third error (Ekg, k=1...l, g=3) is the deviation from the predefined dimension of the single product (WZk, k=1...l); The fourth error (Ekg, k=1...l, g=4) is the burr formation on the single product (WZk, k=1...l); A fifth error (Ekg, k=1 ... l, g=5) is a deviation from a predefined filling of the cavity (110) during the injection molding process; The sixth error (Ekg, k=1...l, g=6) is the burn position on the single product (WZk, k=1...l); The seventh error (Ekg, k = 1 ... l, g = 7) is a blocked cooling channel of the injection molding machine (1); The eighth error (Ekg, k = 1 ... l, g = 8) is a damaged heating belt of the injection molding machine (1); The ninth error (Ekg, k=1...l, g=9) is a damaged check valve of the screw (100) of the injection molding machine (1); The tenth error (Ekg, k=1...l, g=10) is the blockage of the nozzle (101) of the injection molding machine (1); and The eleventh error (Ekg, k = 1 ... l, g = 11) is the viscosity fluctuation of the melt (MT) of the injection molding machine (1); Providing error data (EDkg,k=1 ... l, g=1 ... q) for the determined error (Ekg,k=1 ... l, g=1 ... q); In a sixteenth method step (M16), an error model (EM) is provided, the error model (EM) being trained with the error data (EDkg, k=1 ... l, g=1 ... q), wherein the error model (EM) determines an error class (ECg, g=1 ... q) for each error index (g) of the error data (EDkg, k=1 ... l, g=1 ... q); and In the seventeenth method step (M17), the sensor data (XDkj, k=1...l, j=1...m) of the manufacturing process of a single product (WZk, k=1...l) having an error (Ekg, k=1...l, g=1...q) are classified as error sensor data (EXDkjg, k=1...l, j=1...m, g=1...q) into an error category (ECg, g=1...q) determined for the error (Ekg, k=1...l, g=1...q).

10. The method (M) according to claim 9, characterized in that In an eighteenth method step (M18), at least one time series of further sensor data (XD'j, j=1 . . . m) of a manufacturing process of a further individual product (W') is provided for a further period (Z'); and In the nineteenth method step (M19), the other sensor data (XD'j, j=1...m) is input into the error model (EM), and the error model (EM) thereby determines whether the other sensor data (XD'j, j=1...m) can be classified into an error category (ECg, g=1...q), and if the other sensor data (XD'j, j=1...m) can be classified into an error category (ECg, g=1...q), error category information (ECI') is generated, which describes the error category (ECg, g=1...q) for the other single product (W').

11. A device (D) for carrying out a method (M) for monitoring a periodic manufacturing process having a plurality of cycles (Zk, k=1 ... l), said device (D) manufacturing at least one single product (WZk, k=1 ... l) in each cycle (Zk, k=1 ... l); The device (D) has at least one sensor unit (13), which generates and automatically provides at least one time series of sensor data (XDkj, k=1...l, j=1...m) for at least one cycle (Zk, k=1...l) for performing the first method step (M1), wherein the sensor data (XDkj, k=1...l, j=1...m) have an effect relationship (R) with stability (ST) and abnormality (ST) of the manufacturing process; The device (D) has at least one evaluation unit (14), in which a computer program (CP) is loaded, and the loaded computer program (CP) drives the evaluation unit (14) to automatically load the sensor data (XDkj, k=1...l, j=1...m); It is characterized in that The loaded computer program (CP) drives the evaluation unit (14) to mark the sensor data (XDkj, k=1 ... l, j=1 ... m) as stable sensor data (SXDkj, k=1 ... l, j=1 ... m) when determining that the production process is stable (ST) in order to perform the second method step (M2); The loaded computer program (CP) drives the evaluation unit (14) to automatically reduce the dimension of the stable sensor data (SXDkj, k=1...l, j=1...m) to point data (PDk, k=1...l) in order to perform the third method step (M3); and The loaded computer program (CP) drives the evaluation unit (14) to automatically form a density distribution (DD) using the point data (PDk, k=1...l) in order to execute the fourth method step (M4), wherein the density distribution (DD) has at least one stable area (SA) of the point data (PDk, k=1...l) and at least one abnormal area (AA) of the point data (PDk, k=1...l), and the point data (PDk, k=1...l) in the abnormal area (AA) corresponds to anomalies (AN) in the manufacturing process.

12. The device (D) according to claim 11, characterized in that The sensor unit (13) generates and provides at least one time series of further sensor data (XD'j, j=1...m) of a manufacturing process of a further individual product (W') for a further cycle (Z') in order to perform a fifth method step (M5); The loaded computer program (CP) drives the evaluation unit (14) to reduce the time series of the further sensor data (XD'j, j=1...m) of the further individual product (W') into further point data (PD') in order to perform a sixth method step (M6); and The loaded computer program (CP) drives the evaluation unit (14) to determine whether the other point data (PD') is located in the stable area (SA) or whether the other point data (PD') is located in the abnormal area (AA) in order to execute the seventh method step (M7), and the evaluation unit (14) generates process information (INF), which indicates whether there is an anomaly (AN) in the manufacturing process of the other single product (W').

13. The device (D) according to claim 11, characterized in that The device (D) is an injection molding machine (1), and the periodic manufacturing process is an injection molding process, and the injection molding machine (1) manufactures individual products (WZk, k=1...l) according to a period (Zk, k=1...l); The injection molding machine (1) comprises at least one injection device (10) and at least one injection molding mold (11), wherein the injection device (10) comprises a screw (100), and the injection molding mold (11) comprises at least one cavity (110); Each cycle (Zk, k=1...l) of the injection molding process comprises an injection phase (I), a pressure holding phase (II) and a residual cooling phase (III), and the injection molding machine (1) controls the injection phase (I), the pressure holding phase (II) and the residual cooling phase (III) by at least one of the following machine setting parameters (MPki, k=1...l, i=1...n): The first machine setting parameter (MPki, k=1...l, i=1) is the metering speed of the screw (100) of the injection device (10); The second machine setting parameter (MPki, k=1...l, i=2) is the injection speed of the melt (MT) into the cavity (110) of the injection molding mold (11); The third machine setting parameter (MPki, k=1 ... l, i=3) is the switching time point (tII) from the injection phase (I) to the pressure holding phase (II); A fourth machine setting variable (MPki, k=1 ... l, i=4) is the unloading movement of the screw (100) of the injection device (10); The fifth machine setting parameter (MPki, k=1...l, i=5) is a target value of the holding pressure in the holding pressure stage (II); and The sixth machine setting parameter (MPki, k=1 ... l, i=6) is the temperature of the melt (MT); The injection molding machine (1) changes at least one machine setting variable (MPki, k=1...l, i=1...n) in a plurality of cycles (Zk, k=1...l) for executing the eighth method step (M8), and the evaluation unit (14) provides machine setting data (MPDki, k=1...l, i=1...n) of the machine setting variable change for each cycle (Zk, k=1...l) of the machine setting variable change; The loaded computer program (CP) drives the evaluation unit (14) to load the machine setting data (MPDki, k=1...l, i=1...n) of the machine setting variable change in order to execute the eighth method step (M8); The loaded computer program (CP) drives the evaluation unit (14) to evaluate the loaded sensor data (XDkj, k=1 ... l, j=1 ... m) in order to execute the eighth method step (M8), and a first evaluation result is to determine at least one mold internal pressure curve (Yk, k=1 ... l), and another evaluation result is to determine at least one of the following process parameters (PPkh, k=1 ... l, h=1 ... o): The first process parameter (PPkh, k = 1 ... l, h = 1) is the maximum mold internal pressure (Pmax); The second process parameter (PPkh, k=1...l, h=2) is the integral of the mold internal pressure curve (Yk, k=1...l); and The third process parameter (PPkh, k=1...l, h=3) is the point data (PDk, k=1...l); The loaded computer program (CP) drives the evaluation unit (14) to determine the process parameters (PPkh, k=1...l, h=1...o) of the machine setting variable change for each cycle (Zk, k=1...l) of the machine setting variable change in order to perform the eighth method step (M8), and to provide process parameter data (PPDkh, k=1...l, h=1...o) for the determined process parameters (PPkh, k=1...l, h=1...o); and The evaluation unit (14) has a regression model (RM), and the loaded computer program (CP) drives the evaluation unit (14) to input the machine setting data (MPDki, k=1...l, i=1...n) and the process parameter data (PPDkh, k=1...l, h=1...o) into the regression model (RM) in order to execute the ninth method step (M9), and use the regression model (RM) to determine the effect relationship (Rkih, k=1...l, i=1...n, h=1...o) between the machine setting data (MPDki, k=1...l, i=1...n) and the process parameter data (PPDkh, k=1...l, h=1...o).

14. The device (D) according to claim 13, characterized in that The loaded computer program (CP) drives the evaluation unit (14) to automatically determine process parameters (PPkh, k=1...l, h=1...o) for a plurality of additional cycles (Z'k, k=1...l) in order to perform the tenth method step (M10), and automatically provides process parameter data (PPDkh, k=1...l, h=1...o) for the determined process parameters (PPkh, k=1...l, h=1...o), and drives the evaluation unit (14) to automatically mark the process parameter data (PPDkh, k=1...l, h=1...o) as stable process parameter data (SPPDkh, k=1...l, h=1...o) when it is determined that the manufacturing process is stable (ST), and to automatically mark the process parameter data (PPDkh, k=1...l, h=1...o) as abnormal process parameter data (APPDkh, k=1...l, h=1...o) when the manufacturing process is abnormal (AN); The loaded computer program (CP) drives the evaluation unit (14) to input the stable process parameter data (SPPDkh, k=1...l, h=1...o) and the abnormal process parameter data (APPDkh, k=1...l, h=1...o) into the regression model (RM) in order to execute the eleventh method step (M11), and through the effect relationship (Rkih, k=1...l, i=1...n, h=1...o), determine the corresponding stable machine setting data (SMPDki, k=1...l, i=1...n) with respect to the stable process parameter data (SPPDkh, k=1...l, h=1...o), and determine the corresponding abnormal machine setting data (AMPDki, k=1...l, i=1...n) with respect to the abnormal process parameter data (APPDkh, k=1...l, h=1...o). The loaded computer program (CP) drives the evaluation unit (14) to form a residual (REkih, k=1...l, i=1...n, h=1...o) of the abnormal machine setting data (AMPDki, k=1...l, i=1...n) relative to the effect relationship (Rkih, k=1...l, i=1...n, h=1...o) in order to perform a twelfth method step (M12); and The loaded computer program (CP) drives the evaluation unit (14) to sort the residuals (Rekih, k=1...l, i=1...n, h=1...o) according to size (SZ) and sign (SI) in order to execute the thirteenth method step (M13), and generates exception information (ANIk, k=1...l), which indicates which machine setting parameter (Mpki, k=1...l, i=1...n) has the highest residual (REkih, k=1...l, i=1...n, h=1...o) in terms of size (SZ) and sign (SI), which is the cause of the manufacturing process abnormality (AN).

15. The device (D) according to claim 11, characterized in that The device (D) is an injection molding machine (1), and the periodic manufacturing process is an injection molding process, and the injection molding machine (1) manufactures individual products (WZk, k=1...l) according to a period (Zk, k=1...l); The injection molding machine (1) manufactures at least one single product (WZk, k=1...l) in each cycle (Zk, k=1...l) of a plurality of cycles (Zk, k=1...l) for the purpose of performing the fourteenth method step (M14), and generates and provides at least one time series of sensor data (XDkj, k=1...l, j=1...m) of the manufacturing process for each cycle (Zk, k=1...l), In the case where it is determined that there is an anomaly (AN) in the manufacturing process, at least one of the following errors (Ekg, k = 1 ... l, g = 1 ... q) is determined: The first error (Ekg, k=1 ... l, g=1) is the deviation from a predefined weight of the single product (WZk, k=1 ... l); The second error (Ekg, k=1...l, g=2) is the deviation from the predefined dimensional stability of the single product (WZk, k=1...l); The third error (Ekg, k=1...l, g=3) is the deviation from a predefined dimension of the single product (WZk, k=1...l); The fourth error (Ekg, k=1...l, g=4) is the burr formation on the single product (WZk, k=1...l); A fifth error (Ekg, k=1 ... l, g=5) is a deviation from a predefined filling of the cavity (110) during the injection molding process; The sixth error (Ekg, k=1...l, g=6) is the burn position on the single product (WZk, k=1...l); The seventh error (Ekg, k = 1 ... l, g = 7) is a blocked cooling channel of the injection molding machine (1); The eighth error (Ekg, k = 1 ... l, g = 8) is a damaged heating belt of the injection molding machine (1); The ninth error (Ekg, k=1...l, g=9) is a damaged check valve of the screw (100) of the injection molding machine (1); The tenth error (Ekg, k=1...l, g=10) is the blockage of the nozzle (101) of the injection molding machine (1); and An eleventh error (Ekg, k=1 ... l, g=11) is a viscosity fluctuation of the melt (MT) of the injection molding machine (1); Providing error data (Edkg,k=1 ... l, g=1 ... q) for the determined error (Ekg,k=1 ... l, g=1 ... q); The loaded computer program (CP) drives the evaluation unit (14) to load the sensor data (XDkj, k=1 ... l, j=1 ... m) and the error data (EDkg, k=1 ... l, g=1 ... q) in order to execute the fourteenth method step (M14) and in order to execute the fifteenth method step (M15); The evaluation unit (14) has an error model (EM), and the loaded computer program (CP) drives the evaluation unit (14) to train the error model (EM) using the error data (EDkg, k=1 ... l, g=1 ... q) in order to perform a sixteenth method step (M16), and to determine an error class (ECg, g=1 ... q) for each error index (g) of the error data (EDkg, k=1 ... l, g=1 ... q) using the error model (EM); and The loaded computer program (CP) drives the evaluation unit (14) to classify the sensor data (XDkj, k=1...l, j=1...m) of the manufacturing process of a single product (WZk, k=1...l) with an error (Ekg, k=1...l, g=1...q) or the sensor data (XDkj, k=1...l, j=1...m) of a cycle (Zk, k=1...l) of the injection molding machine (1) with an error (Ekg, k=1...l, g=1...q) as error sensor data (EXDkjg, k=1...l, j=1...m, g=1...q) into an error category (ECg, g=1...q) determined for the error (Ekg, k=1...l, g=1...q).

16. The device (D) according to claim 15, characterized in that The sensor unit (13) provides at least one time series of further sensor data (XD'j, j=1...m) of a manufacturing process of a further individual product (W') for a further cycle (Z') in order to perform an eighteenth method step (M18); and The loaded computer program (CP) drives the evaluation unit (14) to input the other sensor data (XD'j, j=1...m) into the error model (EM) in order to execute the nineteenth method step (M19), and uses the error model (EM) to determine whether the other sensor data (XD'j, j=1...m) can be classified into an error category (ECg, g=1...q), and if the other sensor data (XD'j, j=1...m) can be classified into the error category (ECg, g=1...q), the evaluation unit (14) is driven to generate error category information (ECI'), which describes the error category (ECg, g=1...q) for the other single product (W').

Images