Abnormality determination device, abnormality determination method, and abnormality determination program

Abstract

The abnormality determination device according to the present invention comprises: a measurement unit that measures a plurality of factors that affect the quality of a product produced in a production facility; a control unit; and a storage unit that stores: a causal relationship model in which a causal relationship between the factors is determined on the basis of feature quantities calculated from the factors when a normal product is produced in the production facility; a predictive formula using at least one of the feature quantities as an input and using other feature quantities different from the input as an output based on the causal relationship model; and a prediction range of the output. The control unit is configured to determine whether or not the feature amount that is the output, among the feature amounts calculated on the basis of the factors measured in the production process of the production facility, is within the prediction range.

Description

Technical Field

[0001] The present invention relates to an abnormality determination device, an abnormality determination method, and an abnormality determination program. Background Art

[0002] As a method of monitoring the state of equipment, Patent Document 1 discloses a system that makes use of a cause-effect relationship model obtained from manufacturing process data to facilitate verification based on domain knowledge.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2018-181158 Summary of the invention

[0006] Problems to be solved by the invention

[0007] However, the method is based on a certain degree of domain knowledge of phenomena that affect quality. Therefore, it is considered difficult for unskilled people or newly generated phenomena to use the method. The present invention is made to solve the above problems, and its purpose is to provide an abnormality determination device, an abnormality determination method, and an abnormality determination program, wherein even a skilled person can easily determine factors related to abnormalities that may occur in production equipment.

[0008] Technical means of solving problems

[0009] The abnormality determination device of the first aspect of the present invention includes: a measuring unit, which measures multiple factors that affect the quality of products produced in a production device; a control unit; and a storage unit, which stores: a causal relationship model, which determines the causal relationship between the factors based on feature quantities calculated based on the factors when normal products are produced in the production equipment; a prediction formula, which is based on the causal relationship model and takes at least one of the feature quantities as input and other feature quantities different from the input as output; and a prediction range of the output, the control unit being configured to determine whether the feature quantity that becomes the output among the feature quantities calculated based on the factors measured in the production process of the production equipment is within the prediction range.

[0010] The abnormality determination device according to a second aspect is the abnormality determination device according to the first aspect, wherein when the feature amount serving as the output is outside the prediction range, a factor related to the feature amount is notified.

[0011] The abnormality determination device according to a third aspect is the abnormality determination device according to the second aspect, and is configured to calculate a degree of deviation of the feature amount from the prediction range.

[0012] An abnormality determination device according to a fourth aspect is the abnormality determination device according to the third aspect, further comprising a display unit, wherein the control unit is configured to display the notified factor and the degree of deviation corresponding to the factor.

[0013] The abnormality determination device according to a fifth aspect is the abnormality determination device according to any one of the first to fourth aspects, wherein the control unit is configured to perform the determination using the measured feature amounts of the group of factors.

[0014] The abnormality determination device of the sixth aspect is the abnormality determination device according to any one of the first to fifth aspects, wherein the prediction range is specified based on the magnitude of deviation of multiple feature quantities derived from multiple normal data (for example, deviation of a prediction formula described later).

[0015] The abnormality determination device of the seventh aspect is the abnormality determination device according to any one of the first to fifth aspects, wherein other factors whose values ​​of the factors are outside the predicted range are output based on the factor data of the product determined to be abnormal, thereby inferring the cause of the abnormality.

[0016] The abnormality judgment method of the eighth aspect of the present invention includes the following steps: a step of determining the causal relationship between the factors and generating a causal relationship model based on the feature quantities calculated based on multiple factors that affect the quality of the products produced in the production equipment when the products are produced normally; a step of generating a prediction formula based on the causal relationship model, which takes at least one of the feature quantities as input and other feature quantities different from the input as output; a step of generating a prediction range of the output; a step of calculating the feature quantities based on the factors measured in the production process of the production equipment; and a step of judging whether the feature quantity that becomes the output among the calculated feature quantities is within the prediction range.

[0017] The abnormality judgment program of the ninth aspect of the present invention enables a computer to execute the following steps: a step of generating a causal relationship model by determining the causal relationship between the factors based on the feature quantities calculated based on multiple factors that affect the quality of the products produced in the production equipment when the products are produced normally; a step of generating a prediction formula based on the causal relationship model that takes at least one of the feature quantities as input and takes other feature quantities different from the input as output; a step of generating a prediction range of the output; a step of calculating the feature quantities based on the factors measured in the production process of the production equipment; and a step of judging whether the feature quantity that becomes the output among the calculated feature quantities is within the prediction range.

[0018] Effects of the Invention

[0019] According to the present invention, even a skilled person can easily identify factors related to abnormalities that may occur in production equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] [ Figure 1 ] is a block diagram schematically illustrating an example of a production system to which the present invention is applied.

[0021] [ Figure 2A ] is a schematic diagram of a joining mechanism included in a production equipment according to one embodiment of the present invention.

[0022] [ Figure 2B ] is a diagram showing the action of the joining mechanism.

[0023] [ Figure 3A ] is a diagram when an abnormality occurs in the joint mechanism.

[0024] [ Figure 3B ] is a diagram illustrating the operation of a joining mechanism in which an abnormality occurs.

[0025] [ Figure 4 ] is a block diagram showing the hardware structure of an abnormality determination device according to one embodiment of the present invention.

[0026] [ Figure 5 ] is a block diagram showing the functional structure of the abnormality determination device.

[0027] [ Figure 6 ] is an example of a causal model.

[0028] [ Figure 7 ] is a diagram showing an example of a prediction formula.

[0029] [ Figure 8 ] is a diagram showing an example of a prediction range.

[0030] [ Fig. 9] is a diagram for explaining the determination of abnormality.

[0031] [ Fig.10 ] is an example of the deviation displayed in the display device. DETAILED DESCRIPTION

[0032] Hereinafter, an embodiment in which the abnormality determination device of the present invention is applied to a production facility having a joining mechanism will be described with reference to the drawings. Figure 1 1 is a block diagram of a production system including an abnormality determination device and production equipment according to the present embodiment. In the present embodiment, the abnormality determination device 1 is configured to determine whether an abnormality has occurred in a joining mechanism included in the production equipment 3.

[0033] <1. Overview of the joining mechanism>

[0034] like Figure 2A As shown, the joining mechanism included in the production equipment 3 has a joining mechanism for joining a cylindrical part 1 arranged on the fixture surface with a cylindrical part 2. In the initial state, the part 1 is inserted into the through hole of the part 2. The part 1 is higher than the part 2 and protrudes upward from the through hole of the part 2. When from the state, as shown in FIG. Figure 2B As shown, when the pressing member is lowered by a predetermined distance by a servo motor or the like to press the upper surface of the component 1, the component 1 is plastically deformed. That is, the length of the component 1 in the vertical direction is shortened, and the length in the radial direction is increased. As a result, the outer peripheral surface of the component 1 that has increased in the radial direction is engaged with the inner peripheral surface of the through hole of the component 2 by friction.

[0035] When the engagement mechanism operates normally, as described above, part 1 engages with part 2, but for example Figure 3A If the fixture surfaces of parts 1 and 2 shown in the figure are tilted due to the mixing of foreign matter, Figure 3B As shown, when the pressing member presses the part 1, the pressing member may contact the part 2 before reaching the original descending distance, and the pressing may stop. As a result, an unfavorable situation (abnormality) occurs in which the part 1 cannot be joined to the part 2. In this embodiment, the determination of the occurrence of such abnormality is described.

[0036] In the joining mechanism of the present embodiment, as an example, the pressing force (N) of the pressing member, the descending speed (mm / s) of the pressing member, the height of the part 1 after joining, i.e., the part height (mm), the joining strength (N) between the part 1 and the part 2, and the difference (mm) in the height of the part 1 and the part 2 after joining, i.e., the part step (mm) are used as factors, and these are extracted from the joining mechanism as action state data. Among these, the pressing force, the part height, and the pressing force are predetermined and input into a computer such as a programmable logic controller (PLC) that controls the action of the joining mechanism. On the other hand, the descending time, the joining strength, and the part step are measured in real time or afterwards in the joining mechanism and sent to the abnormality judgment device 1 as action state data. Therefore, sensors for measuring various factors are provided in the joining mechanism. In addition, in the case of measuring various factors afterwards, sensors may not be provided in the joining mechanism.

[0037] <2. Abnormality determination device>

[0038] <2-1. Hardware structure>

[0039] Next, an example of the hardware configuration of the abnormality determination device 1 according to the present embodiment will be described. Figure 4 is a block diagram showing the hardware structure of the abnormality determination device. Figure 4 As shown, the abnormality determination device 1 is a computer to which a control unit 11, a storage unit 12, a communication interface 13, an external interface 14, an input device 15, and a driver 16 are electrically connected.

[0040] The control unit 11 includes a central processing unit (CPU), a random access memory (RAM), a read only memory (ROM), etc., and controls each structure according to information processing. The storage unit 12 is an auxiliary storage device such as a hard disk drive or a solid state drive, and stores a program 121 executed by the control unit 11, a causal relationship model 122, a prediction formula 123, a prediction range 124, prediction data 125, and action state data 126.

[0041] The program 121 is a program for calculating feature quantities from the operating state data extracted from the production equipment 3, and generating a causal relationship model between factors based on these feature quantities, or performing abnormality determination, etc.

[0042] The characteristic quantity that becomes the object of the causal relationship model 122 is based on the characteristic quantity of the factor that affects the quality of the product produced by the joint mechanism, and the causal relationship between the factors when the product is normally produced is constructed as a causal relationship model. In the abnormality determination device 1, the causal relationship model 122 is generated using the characteristic quantity calculated based on the action state data, but the causal relationship model generated in advance in the external device can also be stored.

[0043] The prediction formula 123 is used to calculate other feature quantities other than input from the feature quantities that become input among the obtained feature quantities, and is generated based on the causal relationship model 122. The prediction range 124 statistically specifies the range of the output obtained according to the prediction formula 123. The prediction data 125 is data when an abnormality occurs, and is data indicating the abnormal factor, feature quantity, and the degree of deviation of the feature quantity from the prediction range 124. In addition, as described above, the operation state data 126 is the operation state data sent from the production equipment 3.

[0044] The communication interface 13 is, for example, a wired local area network (LAN) module, a wireless LAN module, etc., and is an interface for wired communication or wireless communication. That is, the communication interface 13 is an example of a communication unit configured to communicate with other devices. The abnormality determination device 1 of this embodiment is connected to the production equipment 3 via the communication interface 13.

[0045] The external interface 14 is an interface for connecting to an external device, and is appropriately configured according to the external device to be connected. In this embodiment, the external interface 14 is connected to the display device 2. In addition, the display device 2 can use a known liquid crystal display, a touch screen display, etc.

[0046] The input device 15 is a device for performing input, such as a mouse or a keyboard.

[0047] The drive 16 is, for example, a CD drive or a DVD drive, and is a device for reading a program stored in the storage medium 17. The type of the drive 16 can be appropriately selected according to the type of the storage medium 17. In addition, at least a part of the various data 122 to 126 including the program 121 stored in the storage unit can also be stored in the storage medium 17.

[0048] The storage medium 17 is a medium that stores information such as programs through electrical, magnetic, optical, mechanical or chemical action in a manner that allows a computer or other device, machine or the like to read the recorded information such as programs. Figure 4In the embodiment, a disk-type storage medium such as a CD or a DVD is illustrated as an example of the storage medium 17. However, the type of the storage medium 17 is not limited to the disk type, and may be a type other than the disk type. As a storage medium other than the disk type, for example, a semiconductor memory such as a flash memory may be cited.

[0049] In addition, regarding the specific hardware structure of the abnormality determination device 1, the structure can be appropriately omitted, replaced, and added according to the implementation method. For example, the control unit 11 may also include multiple processors. The abnormality determination device 1 may also include multiple information processing devices. In addition, in the abnormality determination device 1, in addition to using an information processing device specially designed for the service provided, a general-purpose server device may also be used.

[0050] <2-2. Functional structure and determination of abnormality>

[0051] Next, the functional configuration (software configuration) of the abnormality determination device 1 will be described. Figure 5 This is an example of the functional structure of the abnormality determination device 1 of this embodiment. The control unit 11 of the abnormality determination device 1 expands the program 121 stored in the storage unit 12 to the RAM. Then, the control unit 11 uses the CPU to interpret and execute the program 121 expanded to the RAM, thereby controlling each structure. Figure 5 As shown, the abnormality determination device 1 of the present embodiment functions as a computer including a feature quantity acquisition unit 111 , a model construction unit 112 , an abnormality determination unit 113 , and a deviation calculation unit 114 .

[0052] <2-2-1. Acquisition of feature quantities>

[0053] The feature quantity acquisition unit 111 acquires the value of the feature quantity calculated based on the action state data 126 indicating the action state of the production equipment 3, respectively, when the joining mechanism joins the parts 1 and 2 normally, and when the two parts are not joined correctly. As the action state data 126 in normal state, data extracted at multiple time points are used, but as the action state data in abnormal state, data extracted at one or more time points where an abnormality is considered to have occurred are used. Then, the model construction unit 112 constructs a causal relationship model 122 indicating the causal relationship between factors based on a predetermined algorithm for deriving the correlation degree of each feature quantity when joining the parts 1 and 2 based on the acquired value of the feature quantity in normal state.

[0054] For example, the control unit 11 obtains the feature value as follows. First, the control unit 11 divides the collected action state data 126 according to each frame in order to define the processing range for calculating the feature value. For example, the control unit 11 may also divide the action state data 126 according to each frame of a certain time length. Among them, the joint mechanism is not necessarily limited to operating at a certain time interval. Therefore, if the action state data 126 is divided according to each frame of a certain time length, the action of the joint mechanism reflected in each frame may be misaligned.

[0055] Therefore, in this embodiment, the control unit 11 performs frame division on the action state data 126 according to each takt time. The takt time is the time taken to produce a specified number of products, that is, to join a specified number of parts 1 and parts 2. The takt time can be determined based on a signal for controlling a joining mechanism, such as a control signal for controlling the operation of each servo motor of the joining mechanism. In addition, the type of control signal is not particularly limited as long as it can be used to control the joining mechanism.

[0056] Next, the control unit 11 calculates the value of the feature quantity based on each frame of the action state data 126. For example, when the action state data 126 is quantitative data (fall time, bonding strength, part step difference) as described above, the control unit 11 can calculate the amplitude, maximum value, minimum value, average value, variance value, standard deviation, autocorrelation coefficient, maximum value of the power spectrum obtained by Fourier transformation, distortion degree, kurtosis, etc. within the frame as the feature quantity.

[0057] Furthermore, the feature quantity can be derived not only from a single action state data 126 but also from a plurality of action state data 126. For example, the control unit 11 can calculate the correlation coefficient, ratio, difference, synchronization misalignment, distance, etc. between the corresponding frames of the action state data 126 as the feature quantity.

[0058] In this way, the control unit 11 can normally obtain the values ​​of various feature quantities calculated from the operation state data 126. In addition, the feature quantities can also be calculated by further performing pre-processing such as normalization and outlier removal on the feature quantities obtained in the above manner.

[0059] <2-2-2. Construction of causal relationship model>

[0060] The model construction unit 112 processes each acquired feature quantity as a random variable, that is, sets each acquired feature quantity at each node, and can construct a causal relationship model in the following manner, for example.

[0061] (1) Construction of Graphical Model

[0062] (1-1) The hierarchical structure is specified based on the time series order in which each feature quantity is generated, etc.

[0063] (1-2) Based on the hierarchical structure and taking the partial correlation coefficient or the correlation error as a criterion, a directed graph is determined in which directed lines are used to connect the nodes.

[0064] In addition, there is no particular limitation on the method for selecting directed lines. There are methods that select based on a goodness of fit index (Goodness of Fit Index, GFI) or a standardized root mean square residual (SRMR), and there are also methods that select by focusing only on partial correlation coefficients, etc.

[0065] (2) Formulation based on structural equation modeling

[0066] (2-1) Based on the data, learn the various relationship equations of the causal structure obtained in (1-2). If it is linear, multiple regression can be used, and if it is nonlinear, support vector regression (SVR) can be used.

[0067] (2-2) In cases where this is not appropriate, you can also try to logarithmize the data or add interactions.

[0068] In addition, the method for constructing the causal relationship model is not limited to this. For example, the causal relationship between each factor can be derived by constructing a Bayesian network. When constructing a Bayesian network, a known method can be used. For example, when constructing a Bayesian network, a structure learning algorithm such as a greedy search algorithm, a stingy search algorithm, and an exhaustive search method can be used. In addition, the evaluation criteria of the constructed Bayesian network can use Akaike's. Information Criterion (AIC), C4.5, Cooper Herskovits Measure (CHM), Minimum Description Length (MDL), Maximum Likelihood (ML), etc. In addition, as a processing method for when missing values ​​are included in the learning data (action state data 126) used to construct the Bayesian network, a pairwise method, a list-wise method, etc. can be used.

[0069] For example, Figure 6The causal relationship model of the descent speed, component height, pressing force, descent time, bonding strength, and component step difference of the bonding mechanism of this embodiment is shown. According to the causal relationship model, it can be seen that the descent speed and component height have an impact on the descent time, and the descent time and pressing force have an impact on the bonding strength. That is, in the causal relationship model, the descent speed, component height, and pressing force become control factors (inputs), based on which, the descent time becomes an intermediate response (output), and the bonding strength and component step difference become the final response (output).

[0070] <2-2-3. Predictive construction>

[0071] Next, the model building unit 112 builds prediction formulas for the intermediate response and the final response based on the causal relationship model built in the above manner. Figure 3B The abnormality shown in FIG. 1 is eliminated by constructing a prediction formula for calculating the descent time based on the part height and the descent speed. Such a prediction formula can be constructed based on the obtained causal relationship model through multiple regression analysis, support vector machine (SVM), etc. In this embodiment, as a prediction formula, for example, the following formula (1) can be constructed.

[0072] Descent time = 0.52*part height-0.7*descent speed+0.03(1)

[0073] When the prediction formula is shown in the figure, it becomes Figure 7 In addition, it is also possible to construct prediction formulas for joint strength and part step difference. In other words, prediction formulas can be constructed for all responses from intermediate responses to final responses.

[0074] <2-2-4. Definition of forecast range>

[0075] Next, calculate the statistical distribution of the normal descent time. This becomes the prediction range. For example, the prediction interval can be specified by the distribution (reliable interval) that can be specified based on the obtained prediction formula, and the distribution that takes into account the deviation of the data (for example, 3σ). Here, as an example, the 95% prediction interval is used. For example, the predicted value of the descent time when the part height is 0.6mm and the descent speed is 0.03mm / s is 0.321s according to the prediction formula (1), so the 95% prediction interval is 0.163~0.479. Figure 8 is a diagram showing the prediction range. In addition, similarly to the above, the prediction range can be specified for all responses of the intermediate response and the final response (in this embodiment, the bonding strength and the part step difference). In this way, the prediction range can be specified by a set of deviations of the prediction formula and deviations of the data.

[0076] <2-2-5. Determination of abnormality>

[0077] Then, the abnormality determination unit 113 determines the abnormality based on the characteristic amount calculated from the action state data extracted when the joint mechanism is considered to have an abnormality. For example, when determining whether the descending time is abnormal, it is determined whether the descending speed corresponding to the obtained part height and the descending speed falls within the predicted range specified in the above manner. Fig. 9 As shown in the figure, when the part height is 0.6 mm and the descending speed is 0.03 mm / s, the measured descending time is 0.085 s. Since the descending time is outside the predicted range, it is determined that an abnormality has occurred. In addition, as described above, all responses of the intermediate response and the final response (in this embodiment, the bonding strength and the part step difference) are also determined to be abnormal. In this way, the determination of abnormality can be performed based on data at one point.

[0078] <2-2-6. Calculation of deviation>

[0079] Next, for each characteristic value, in order to determine the degree of abnormality, the deviation calculation unit 114 is used to calculate the deviation. For example, the deviation can be calculated by the ratio of the difference with the predicted value. That is, based on the numerical value, (predicted value (0.321) - measured value (0.085)) / predicted value (0.321) = -0.74 can be used as the deviation. In addition, for example, in the case where it is determined that the bonding strength is abnormal, the deviation is also calculated in the same way. The deviation calculated in this way is, for example, Fig.10 As shown, it can be displayed on the display device 2 in the form of a list.

[0080] As described above, an abnormality is determined. If an abnormality is confirmed, the cause of the abnormality is studied based on this. For example, when the falling time is shorter than the predicted value as described above, an abnormality such as that shown in FIG. 3 can be considered. In addition, even if the measured value is outside the predicted range, it is not immediately determined to be an abnormality, but the abnormality is determined based on the degree of deviation.

[0081] <3. Features>

[0082] (1) Since the cause-effect relationship model, prediction formula, and prediction range are constructed as described above, even an unskilled person can easily identify factors related to abnormalities that may occur in production equipment.

[0083] (2) According to the present embodiment, the abnormality is determined by constructing a prediction range, so that, for example, the abnormality can be determined without constructing a causal relationship model when the abnormality occurs. That is, as described above, the occurrence of the abnormality can be determined by extracting a feature quantity when the abnormality is considered to have occurred. Therefore, for example, it is not necessary to stop the production equipment 3 in order to obtain data for constructing a causal relationship model when the abnormality occurs. In particular, in the present embodiment, the abnormality can be determined using data (data at one point) extracted at a time point when the abnormality is considered to have occurred, so the cost or load required for the abnormality determination can be reduced.

[0084] (3) When an abnormality is considered to have occurred, since the deviation of each feature value is calculated, it is easy to determine the degree of abnormality. In addition, by displaying the deviation together with the intermediate response or the final response, the occurrence of an abnormality can be easily visually recognized.

[0085] <4. Modifications>

[0086] The above describes the embodiments of the present invention in detail, but the description so far is only an illustration of the present invention in all aspects. Of course, various improvements or modifications can be made without departing from the scope of the present invention. For example, the following changes can be made. In addition, the following structural elements identical to those of the embodiments are used with the same symbols, and the description of the same points as those of the embodiments is appropriately omitted. The following modifications can be appropriately combined.

[0087] <4-1>

[0088] In the above-described embodiment, the determination of an abnormality occurring in the process of joining parts 1 and 2 is taken as an example, but this is only an example and can of course be applied to other processes.

[0089] <4-2>

[0090] In the above embodiment, the quantitative data is used as the action state data and the feature quantity is calculated based on this, but the action state data 126 may be used as qualitative data such as the on / off state of the sensor. That is, the "on" time, "off" time, duty ratio, "on" times, "off" times, etc. in each frame may be calculated as the feature quantity.

[0091] <4-3>

[0092] The processes from the collection of the operation state data 126 to the calculation of the value of the feature amount can be performed by the production equipment 3 or various devices that control the production equipment 3 instead of by the abnormality determination device 1 .

[0093] <4-4>

[0094] In the above embodiment, when an abnormality is considered, the abnormality is determined based on a set (one point) of feature quantities, but the abnormality can be determined by calculating feature quantities based on the action state data 126 extracted at multiple time points and using these feature quantities in combination. For example, the average value of the feature quantities calculated based on multiple time points can be used.

[0095] <4-5>

[0096] In the above-described embodiment, the result of the determination is displayed on the display device 2, but the determination result may be notified by a method other than displaying on the display device 2. For example, the occurrence of an abnormality may be notified by notification based on sound or light, etc. In addition, the notification of the deviation is not essential, and only the factor of the abnormality may be notified, or only the determination may be performed and the result of the determination may be studied after production, etc.

[0097] Explanation of symbols

[0098] 1: Abnormality determination device

[0099] 11: Control Department

[0100] 12: Storage

[0101] 3: Production equipment

Claims

1. An abnormality determination device, comprising: A measuring unit that measures various factors that influence the quality of the products produced in the production facility; Control Department; as well as Storage Department, The storage unit stores: a causal relationship model that determines the causal relationship between the factors based on feature quantities calculated from the factors when normal products are produced in the production equipment; A prediction formula, based on the causal relationship model and taking at least one of the feature quantities as input and taking other feature quantities different from the input as output; as well as The predicted range of the output, The control unit is configured as follows: It is determined whether the feature quantity serving as the output among the feature quantities calculated based on the factors measured in the production process of the production equipment is within the prediction range.

2. The abnormality determination device according to claim 1, wherein the control unit is configured as follows: When the feature amount serving as the output is outside the prediction range, a factor related to the feature amount is notified. 3 . The abnormality determination device according to claim 2 , wherein the device is configured to calculate a degree of deviation of the feature amount from the prediction range.

4. The abnormality determination device according to claim 3, wherein the control unit is configured as follows: The notified factor and the degree of deviation corresponding to the factor are displayed.

5. The abnormality determination device according to claim 1, wherein the control unit is configured as follows: The determination is performed using the measured feature quantities of the group of factors. 6 . The abnormality determination device according to claim 1 , wherein the prediction range is specified based on a magnitude of deviation of a plurality of feature quantities derived from a plurality of normal data. 7 . The abnormality determination device according to claim 1 , wherein based on the factor of the product determined to be abnormal, other factors whose values ​​of the factor are outside the prediction range are output.

8. A method for determining an abnormality, comprising the following steps: A step of generating a causal relationship model by determining a causal relationship between a plurality of factors affecting the quality of products produced in a production facility based on the feature quantities calculated based on the feature quantities when the products are normally produced; A step of generating a prediction formula based on the causal relationship model, which takes at least one of the feature quantities as an input and takes other feature quantities different from the input as an output; the step of generating a predicted horizon for said output; A step of calculating the characteristic amount based on the factor measured in the production process of the production equipment; as well as A step of determining whether the feature quantity serving as the output among the calculated feature quantities is within the prediction range.

9. An abnormality determination program, Instruct the computer to perform the following steps: A step of generating a causal relationship model by determining a causal relationship between a plurality of factors affecting the quality of products produced in a production facility based on the feature quantities calculated based on the feature quantities when the products are normally produced; A step of generating a prediction formula based on the causal relationship model, which takes at least one of the feature quantities as an input and takes other feature quantities different from the input as an output; the step of generating a predicted horizon for said output; A step of calculating the characteristic amount based on the factor measured in the production process of the production equipment; as well as A step of determining whether the feature quantity serving as the output among the calculated feature quantities is within the prediction range.

Images