CN117990966A - Generating system, computer-readable storage medium, and method for generating waveform evaluation model - Google Patents

Generating system, computer-readable storage medium, and method for generating waveform evaluation model Download PDF

Info

Publication number
CN117990966A
CN117990966A CN202311415222.5A CN202311415222A CN117990966A CN 117990966 A CN117990966 A CN 117990966A CN 202311415222 A CN202311415222 A CN 202311415222A CN 117990966 A CN117990966 A CN 117990966A
Authority
CN
China
Prior art keywords
waveform data
waveform
virtual
data
unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311415222.5A
Other languages
Chinese (zh)
Inventor
铃木良平
株丹亮
角谷拓也
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yaskawa Electric Corp
Original Assignee
Yaskawa Electric Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yaskawa Electric Corp filed Critical Yaskawa Electric Corp
Publication of CN117990966A publication Critical patent/CN117990966A/en
Pending legal-status Critical Current

Links

Abstract

The invention provides a generating system, comprising: a waveform data acquisition unit that acquires waveform data; an intention determination unit that determines an intention of a user; and a virtual waveform generation unit that generates virtual waveform data from the waveform data acquired by the waveform data acquisition unit by a method that reflects the intention of the user determined by the intention determination unit. The invention also provides a method for generating the waveform evaluation model executed by the computer, which comprises the following steps: a waveform data acquisition step of acquiring waveform data; an intention determining step of determining an intention of a user; a virtual waveform generation step of generating virtual waveform data from the waveform data acquired in the waveform data acquisition step by a method reflecting the intention of the user determined in the intention determination step; and a learning execution step of generating a waveform evaluation model that outputs an evaluation result of the input waveform data by executing machine learning using the virtual waveform data generated in the virtual waveform generation step.

Description

Generating system, computer-readable storage medium, and method for generating waveform evaluation model
Technical Field
The invention relates to a generating system, a computer-readable storage medium, and a method for generating a waveform evaluation model.
Background
Non-patent document 1 and non-patent document 2 describe a technique called data expansion for generating new data from existing data.
Prior art literature
Non-patent literature:
non-patent literature 1：Zhao,Zhengli,Zizhao,Zhang,Ting,Chen,Sameer,Singh,and Han,Zhang."Image Augmentations for GAN Training."(2020).
Non-patent literature 2：Brian Kenji Iwana,and Seiichi Uchida."An empirical survey of data augmentation for time series classification with neural networks".PLOS ONE16,no.7(2021):e 0254841.
Disclosure of Invention
According to one embodiment of the present invention, a generation system is provided. The generating system may include a waveform acquisition section that acquires waveform data. The generating system may include an intention determining portion that determines the intention of the user. The generating system may include a virtual waveform generating section that generates virtual waveform data from the waveform data acquired by the waveform data acquiring section by a method that reflects the intention of the user determined by the intention determining section.
The generating system may include a similarity display control section that performs control to display data to a user, the display data representing a similarity between the virtual waveform data generated from the waveform data by the virtual waveform generating section and the waveform data that is a generation source of the virtual waveform data. The virtual waveform generation unit may generate a plurality of the virtual waveform data from the waveform data by using at least any one of a plurality of algorithms, a plurality of parameter settings, and a random number, and the similarity display control unit may perform control to display the display data representing a similarity between the plurality of virtual waveform data generated from the waveform data by the virtual waveform generation unit and the waveform data as a generation source of the plurality of virtual waveform data to a user. The intention determining section may determine a plurality of combinations of algorithm and parameter settings in accordance with an instruction of the user, the virtual waveform generating section may generate, for each of the plurality of combinations determined by the intention determining section, a virtual waveform data set including the plurality of virtual waveform data from the waveform data using the algorithm and the parameter settings in the combination, and the similarity display control section may perform control to display the display data representing a similarity between each of the plurality of virtual waveform data sets generated by the virtual waveform generating section and the waveform data as a generation source to the user. The generating system may include a virtual waveform storing section that associates recipe information with each of the plurality of virtual waveform data sets generated by the virtual waveform generating section, the recipe information being capable of identifying a combination of the algorithm used in the generation of the virtual waveform data set and the parameter setting, and stores the plurality of virtual waveform data sets.
The arbitrary generation system may include a virtual waveform storage section that associates source waveform identification information capable of identifying the waveform data that is a generation source of the virtual waveform data set with each of the plurality of virtual waveform data sets generated by the virtual waveform generation section, and stores the plurality of virtual waveform data sets.
The arbitrary generation system may include: a range specification receiving unit configured to receive specification of a range of the similarity of the display data displayed by the similarity display control unit by the user; an in-range waveform selection unit configured to select a plurality of pieces of virtual waveform data corresponding to the range in which the similarity specified by the range specification reception unit is received; and a virtual waveform storage unit that stores a virtual waveform data set including the plurality of virtual waveform data selected by the in-range waveform selection unit.
In the above-described generation system, the similarity display control section may perform control to display, to the user, the display data representing a similarity between the virtual waveform data and the waveform data that is a generation source of the virtual waveform data, and other display data representing a similarity between the virtual waveform data and other waveform data different from the waveform data.
The arbitrary generation system may include: a preprocessing unit configured to perform at least one of a plurality of preprocessing on the waveform data; and a virtual waveform storage unit that associates preprocessing identification information capable of identifying the preprocessing with a virtual waveform data set including a plurality of pieces of virtual waveform data generated by the virtual waveform generation unit from the waveform data subjected to the preprocessing by the preprocessing unit, and stores the virtual waveform data set.
The arbitrary generation system may include: a preprocessing unit configured to perform at least one of a plurality of preprocessing on the waveform data; a learning execution unit configured to generate a waveform evaluation model that outputs an evaluation result of the input waveform data by executing machine learning using the virtual waveform data generated by the virtual waveform generation unit from the waveform data subjected to the preprocessing by the preprocessing unit; and a waveform evaluation model storage unit that associates preprocessing identification information capable of identifying the preprocessing with the waveform evaluation model generated by the learning execution unit, and stores the waveform evaluation model. The generation system may include: a model acquisition unit configured to acquire the waveform evaluation model associated with the preprocessing identification information stored in the waveform evaluation model storage unit; a waveform input unit for performing preprocessing indicated by the preprocessing identification information on input waveform data and inputting the preprocessing data into the waveform evaluation model; and an evaluation result output control unit that controls to output an evaluation result of the input waveform data output from the waveform evaluation model.
The arbitrary generation system may include: a virtual waveform storage unit that associates virtual identification information indicating that the virtual waveform data is data generated from the waveform data with the virtual waveform data generated by the virtual waveform generation unit, and stores the virtual waveform data; and a learning execution unit configured to generate a waveform evaluation model for outputting an evaluation result of the input waveform data by performing machine learning using the virtual waveform data generated by the virtual waveform generation unit, and perform evaluation of the waveform evaluation model without using the virtual waveform data.
According to an embodiment of the present invention, there is provided a program for causing a computer to function as the generation system. According to an embodiment of the present invention, there is provided a computer-readable storage medium storing the program.
According to an embodiment of the present invention, there is provided a waveform evaluation model generation method executed by a computer. The generating method may include a waveform data acquisition step of acquiring waveform data. The generating method may include an intention determining step of determining the intention of the user. The generating method may include a virtual waveform generating step of generating virtual waveform data from the waveform data acquired in the waveform data acquiring step using a method reflecting the user's intention determined in the intention determining step. The generating method may include a learning execution step of generating a waveform evaluation model that outputs an evaluation result of the input waveform data by executing machine learning using the virtual waveform data generated in the virtual waveform generation step.
In addition, the above summary does not list all necessary features of the present invention. In addition, sub-combinations of these feature sets can also form the inventive arrangements.
Drawings
Fig. 1 is an explanatory diagram for explaining a process in the system 10.
Fig. 2 shows an example of virtual abnormal waveform data.
Fig. 3 shows an example of a display related to the virtual waveform confirmation processing.
Fig. 4 shows an example of a display related to the virtual waveform confirmation processing.
Fig. 5 schematically shows an example of the functional structure of the system 10.
Fig. 6 schematically shows an example of the flow of the waveform evaluation model generation process of the system 10.
Fig. 7 schematically illustrates an example of the management data 190.
Fig. 8 is an explanatory diagram for explaining an expansion processing algorithm.
Fig. 9 is an explanatory diagram for explaining an expansion/contraction processing algorithm.
FIG. 10 is an explanatory diagram for describing a scale random deformation algorithm;
fig. 11 schematically shows an example of the display data 400 displayed by the similarity display control unit 170.
Fig. 12 schematically shows an example of a hardware configuration of a computer 1200 functioning as the system 10 or a part of the system 10.
Detailed Description
The present invention will be described below with reference to embodiments of the invention, but the following embodiments do not limit the invention according to the claims. The combination of the features described in the embodiments is not necessarily essential to the embodiments of the invention.
Fig. 1 is an explanatory diagram for explaining a process in the system 10. The system 10 may be implemented by a single device. The system 10 may also be implemented by a number of devices. System 10 may be an example of a generating system.
The system 10 generates a virtual waveform from an existing waveform. The system 10 has a function of generating a virtual abnormal waveform, for example, and thus enables AI learning even with a small amount of abnormal data. The system 10 has a virtual abnormal waveform generation function of generating and outputting a virtual abnormal waveform. The system 10 may have a discriminant learning function that learns the discriminant using virtual abnormal waveforms.
For example, abnormal data may not be sufficiently measured in a step before starting. In this case, even if an AI model is generated to automate the inspection process (automatically determine normal or abnormal), sufficient performance cannot be obtained in many cases. This is because it is easy to excessively learn less abnormal data, and sufficient determination accuracy cannot be obtained.
In contrast, the system 10 processes the abnormal waveform so as not to significantly impair the physical meaning, and generates a virtual abnormal waveform. The system 10 generates a large number of virtual abnormal waveforms subjected to various processing from a small number of abnormal waveforms.
For example, in a product manufacturing system, it is desired to check whether or not a manufactured product has a defect, but there are many cases in which the manufacturing process has just been started and the number of abnormal data is small. In order to accurately determine the normal and abnormal conditions, a certain amount of abnormal data is required, but if the required amount is continuously collected, 3 abnormal data among 1000 production steps having a process capability of 3σ (the possibility of deviation from the standard is about 0.3%), and 33000 production steps are required if 100 abnormal data are to be collected. Thus, the starting of the AI for inspection takes much time.
According to the system 10 of the present embodiment, a large amount of virtual abnormal data is generated at the stage of collecting a small amount of abnormal data, and thus an AI discriminator with sufficient accuracy can be generated, which can contribute to early start of the automatic inspection process.
The system 10 is not limited to abnormal waveforms, and may generate virtual normal waveforms from normal waveforms. The system 10 is not limited to the abnormal waveform and the normal waveform, and may generate a virtual waveform from an arbitrary waveform.
The system 10 uses, for example, waveform data of a fixed length periodically acquired by a motor or the like that performs repetitive operations as a processing target. The system 10 subjects, for example, data whose waveform has periodicity and whose characteristics appear after conversion to the frequency domain to processing.
The system 10 includes a waveform storage 102. The waveform storage unit 102 stores waveform data as a generation source of virtual waveform data. The waveform data may be time series data. For example, the waveform storage unit 102 stores waveform data generated by observing or measuring a target system.
The waveform storage unit 102 may store abnormal waveform data generated by observing or measuring the target system in a state where an abnormality occurs in the target system itself. The waveform storage unit 102 may store, for example, abnormal waveform data generated by observing or measuring the target system in a state where abnormality is intentionally generated in the target system. In the case where the target system is a system for manufacturing a product, the waveform storage unit 102 may store abnormal waveform data generated by observing or measuring the target system when an abnormality occurs in the product, and abnormal waveform data generated by observing or measuring the product.
The waveform storage 102 may store normal waveform data used for generating AI for discriminating between a normal waveform and an abnormal waveform. The waveform storage unit 102 may store normal waveform data generated by observing or measuring the target system in a state where the target system is operating normally. In the case where the target system is a system for manufacturing a product, the waveform storage unit 102 may store normal waveform data generated by observing or measuring the target system when the product is normal, and normal waveform data generated by observing or measuring the product.
The waveform storage unit 102 may store arbitrary waveform data regardless of normal waveform data and abnormal waveform data.
The system 10 may perform the dummy processing 104 on the waveform data stored in the waveform storage 102. The virtual augmentation process 104 may be performed by a virtual waveform generation section included in the system 10.
The virtual waveform generation unit virtually generates various virtual abnormal waveform data from the abnormal waveform data stored in the waveform storage unit 102, for example. The virtual waveform generation unit generates virtual abnormal waveform data by adding noise to the abnormal waveform data, expanding in the vertical axis direction of the graph, contracting in the vertical axis direction of the graph, expanding in the horizontal axis direction of the graph, contracting in the horizontal axis direction of the graph, and the like, for example. As a specific example, as shown in fig. 2, the virtual waveform generation section generates virtual abnormal waveform data 204 by expanding the abnormal waveform data 202 in the direction of the horizontal axis of the graph, or generates virtual abnormal waveform data 206 by expanding the abnormal waveform data 202 in the direction of the vertical axis of the graph.
As a specific example, the virtual waveform generation unit generates about 100 pieces of virtual abnormal waveform data from about 10 pieces of abnormal waveform data.
The virtual waveform generation unit may virtually generate various virtual normal waveform data from the normal waveform data in the same manner. The virtual waveform generation unit may similarly virtually generate various virtual waveform data from arbitrary waveform data.
The learning data storage 106 stores learning data for learning AI for discriminating between normal waveforms and abnormal waveforms. The learning data storage unit 106 may store the waveform data stored in the waveform storage unit 102 and the virtual waveform data generated by the virtual waveform generation unit as the learning data. For example, the learning data storage unit 106 stores the normal waveform data and a small amount of abnormal waveform data stored in the waveform storage unit 102, and the virtual abnormal waveform data generated from the small amount of abnormal waveform data by the virtual waveform generation unit.
The system 10 may include an input accepting portion that accepts user input 108 from a user of the system 10. In addition, the system 10 may include an algorithm storage 110, the algorithm storage 110 storing a plurality of algorithms for generating virtual waveform data from the original waveform data.
The virtual waveform generation unit may execute a selection process 112 of selecting one or more algorithms from among the algorithms stored in the algorithm storage unit 110, based on the user input 108 received by the input reception unit.
The virtual waveform generation section may execute parameter setting processing 114 of setting parameters for generating virtual waveform data. The virtual waveform generation unit may set a plurality of parameters such as a large parameter amplitude, a medium parameter amplitude, and a small parameter amplitude for one algorithm, for example. The virtual waveform generation unit may execute the parameter setting process 114 based on the user input 108 received by the input reception unit. The virtual waveform generation unit may automatically execute the parameter setting process 114. The virtual waveform generation unit executes the parameter setting process 114 based on, for example, a difference between the normal waveform data and the abnormal waveform data.
The virtual waveform generation unit applies the parameters set by the parameter setting process 114 to the algorithm selected by the selection process 112, and executes a generation process 116 of generating virtual waveform data. The virtual waveform generation unit sequentially applies the plurality of parameters set in the parameter setting process 114 to, for example, one algorithm selected in the selection process 112, and generates virtual waveform data. For example, the virtual waveform generation unit generates virtual waveform data by setting a parameter having a large amplitude, generates virtual waveform data by setting a parameter having a small amplitude, and generates virtual waveform data by setting a parameter having a large amplitude with respect to an algorithm for adding noise. The virtual waveform generation unit causes the data storage unit 118 to store the generated virtual waveform data.
As a specific example, the virtual waveform generation section generates virtual waveform data by adding a random numerical value based on a normal distribution with a mean value of 0 and a standard deviation of σ to the waveform data. In addition, as a specific example, the virtual waveform generation section generates virtual waveform data by accumulating random values based on a normal distribution having a mean value of 1 and a standard deviation of σ, on the waveform data. The virtual waveform generation unit can set the standard deviation σ by the parameter setting process 114. The virtual waveform generation unit sets one or a plurality of standard deviations σ in accordance with a user instruction, for example. The virtual waveform generation unit sets the standard deviation σ based on, for example, a difference between the normal waveform data and the abnormal waveform data when generating the virtual abnormal waveform data. For example, the larger the difference between the normal waveform data and the abnormal waveform data, the larger the standard deviation σ is made by the virtual waveform generating section.
The system 10 may use all of the virtual waveform data stored in the data storage unit 118 as learning data, or may use only a part of the virtual waveform data stored in the data storage unit 118 as learning data according to the intention of the user of the system 10.
The system 10 may include a virtual waveform confirmation processing section. The virtual waveform confirmation processing part executes the user intention reflecting processing 120 for the virtual waveform data stored in the data storage part 118.
The virtual waveform confirmation processing unit may execute the distribution display processing 122 for each combination of the algorithm and the parameter setting, and the distribution display processing 122 may display a distribution indicating the characteristics of the generated virtual waveform data to the user. For example, the virtual waveform confirmation processing unit displays the following histogram or graph to the user: the virtual waveform data visually showing each combination of the algorithm and the parameter setting is generated with a degree of deviation from the waveform data as the generation source of the virtual waveform data. As a specific example, the virtual waveform confirmation processing unit displays together for one algorithm: a histogram indicating a correlation between the plurality of virtual waveform data generated by setting the first parameter and the source waveform data, a histogram indicating a correlation between the plurality of virtual waveform data generated by setting the second parameter and the source waveform data, and a histogram indicating a correlation between the plurality of virtual waveform data generated by setting the third parameter and the source waveform data. The virtual waveform confirmation processing unit may similarly display histograms of a plurality of virtual waveform data generated by setting a plurality of parameters for other algorithms. By viewing the display, the user can study which of a plurality of combinations of algorithms and parameter settings is used.
For example, the virtual waveform confirmation processing unit displays the following histogram or graph to the user: deviations of the virtual abnormal waveform data from each combination of algorithm and parameter setting are visually displayed with respect to the abnormal waveform data as a generation source of the virtual abnormal waveform data. The virtual waveform confirmation processing unit may display the average waveform of the abnormal waveform data together with the average waveform of the virtual abnormal waveform data for each combination of the algorithm and the parameter setting. The virtual waveform confirmation processing unit may display a histogram or graph representing the statistical difference of the virtual abnormal waveform data of each combination of the abnormal waveform data and the algorithm and parameter setting in various viewpoints to the user.
For example, the virtual waveform confirmation processing unit may display the following histogram or graph to the user: deviations of the abnormal waveform data and the virtual abnormal waveform data for each combination of algorithm and parameter settings from the normal waveform data are visually displayed. The virtual waveform confirmation processing unit may display the average waveform of the normal waveform data, the average waveform of the abnormal waveform data, and the average waveform of the virtual abnormal waveform data for each combination of the algorithm and the parameter setting. The virtual waveform confirmation processing unit may display a histogram or graph representing the statistical difference of the virtual abnormal waveform data of each combination of the normal waveform data and the abnormal waveform data and the algorithm and parameter setting in various viewpoints to the user. As a specific example, the virtual waveform confirmation processing part may display the histogram 212 and the waveform 214 as illustrated in fig. 3 to the user.
The virtual waveform confirmation processing part may execute a selection process 124 of selecting at least a part from the virtual waveform data of which distribution is displayed by the distribution display process 122 according to an instruction of the user. The virtual waveform confirmation processing part may perform the individual display processing 126 of individually displaying the virtual waveform data selected by the selection processing 124. By viewing the display, the user can confirm the virtual waveform data individually, and can determine which material is used from among a plurality of combinations of algorithms and parameter settings. As a specific example, the virtual waveform confirmation processing part may provide the display 220 as shown in fig. 4 to the user.
The virtual waveform confirmation processing unit executes the range selection processing 128 in accordance with an instruction from the user who has reviewed the display of the distribution display processing 122 and the individual display processing 126, and the range selection processing 128 selects the use range in the distribution or the non-use range in the distribution. The virtual waveform confirmation processing unit reflects the data of the employed range as learning data with the selection result of the range selection processing 128 as the intention of the user. The virtual waveform confirmation processing unit may store the data of the usage range in the learning data storage unit 130.
The system 10 may include a learning execution portion 140, and the learning execution portion 140 generates the AI 142 that discriminates between the normal waveform and the abnormal waveform by machine learning. The learning execution unit 140 may generate the AI 142 by executing machine learning using the learning data stored in the learning data storage unit 106. The learning execution unit 140 may execute machine learning using the normal waveform data, the abnormal waveform data, and the virtual abnormal waveform data stored in the learning data storage unit 106. The learning execution unit 140 may change the number of virtual abnormal waveform data used for learning in response to an instruction from the user or the like. The learning execution unit 140 may execute machine learning in which the weight of the virtual abnormal waveform data is reduced with respect to the abnormal waveform data, in accordance with an instruction or the like of the user.
The learning execution unit 140 may generate the AI 142 by executing machine learning using the learning data stored in the learning data storage unit 130. The learning execution unit 140 may execute machine learning using the normal waveform data, the abnormal waveform data, and the virtual abnormal waveform data stored in the learning data storage unit 106. The learning execution unit 140 may change the number of virtual abnormal waveform data used for learning in response to an instruction from the user or the like. The learning execution unit 140 may execute machine learning with the weight of the virtual abnormal waveform data reduced on the abnormal waveform data in accordance with an instruction or the like of the user.
The system 10 may also include an algorithm storage 144 that stores a plurality of learning algorithms. The learning execution section 140 may learn the AI 142 using any one of a plurality of learning algorithms stored in the algorithm storage section 144. The learning execution unit 140 uses, for example, a learning algorithm corresponding to an algorithm for generating virtual waveform data among algorithms stored in the algorithm storage unit 110. The learning execution unit 140 uses, for example, a learning algorithm corresponding to parameter settings used for generating virtual waveform data. The learning execution unit 140 uses, for example, a learning algorithm corresponding to an algorithm for generating virtual waveform data and a parameter setting for generating virtual waveform data, among the algorithms stored in the algorithm storage unit 110. The learning execution unit 140 may use a learning algorithm selected according to an instruction from the user of the system 10.
The application 150 having the AI 142 is provided to, for example, a device that performs a discrimination process of discriminating a normal waveform and an abnormal waveform, and the device performs abnormality detection of the target system by executing the application 150. The application 150 having the AI 142 may be provided to any other device such as a device that performs failure prediction.
As described above, the system 10 may include a virtual waveform generation section that generates virtual waveform data from waveform data. The virtual waveform generation unit may generate virtual waveform data from the waveform data by a method that reflects the intention of the user. The virtual waveform generation unit may generate virtual waveform data from the waveform data using an algorithm corresponding to the user's intention among a plurality of algorithms. For example, the virtual waveform generation section generates virtual waveform data from waveform data using one or more algorithms selected by a user. The virtual waveform generation unit may generate virtual waveform data from the waveform data according to parameter settings according to the intention of the user. For example, the virtual waveform generation unit generates virtual waveform data from waveform data in accordance with one or more parameter settings selected by a user. The virtual waveform generation unit may generate virtual abnormal waveform data from the abnormal waveform data. The virtual waveform generation unit can process a small amount of abnormal waveform data so as not to significantly impair the physical meaning, and generate a large amount of abnormal waveform data.
As described above, the system 10 may include a virtual waveform confirmation processing section. The virtual waveform confirmation processing unit may display to the user: the distribution of the characteristics of the virtual waveform data generated by the virtual waveform generation unit according to each combination of the algorithm and the parameter setting is shown. For example, the virtual waveform confirmation processing unit displays the following histogram or graph to the user: the virtual waveform data of each combination of the visual display algorithm and the parameter setting is generated with a deviation of a degree from the waveform data which is the generation source of the virtual waveform data. The virtual waveform confirmation processing part may select virtual waveform data corresponding to a range selected by the user in the histogram displayed to the user as the data for learning. The virtual waveform confirmation processing unit may display virtual waveform data and source waveform data corresponding to a range selected by the user in a histogram displayed to the user so as to be able to be compared one by one. The virtual waveform confirmation processing unit may display the deviation of the virtual waveform data and the deviation of the source waveform data in parallel, the deviation corresponding to the range selected by the user.
As described above, the system 10 may include a learning execution section that performs machine learning using the virtual waveform data generated by the virtual waveform generation section. The learning execution unit may generate the AI that discriminates the normal waveform and the abnormal waveform by machine learning using the normal waveform data, the abnormal waveform data, and the virtual abnormal waveform data generated from the abnormal waveform data by the virtual waveform generation unit.
Fig. 5 schematically shows an example of the functional structure of the system 10. The system 10 includes a storage unit 11, a registration unit 12, a generation unit 13, a virtual waveform confirmation processing unit 14, a learning unit 15, and an evaluation execution unit 16. Furthermore, the system 10 need not necessarily include all of them.
The storage 11 may include a waveform storage 102, an algorithm storage 110, and a virtual waveform storage 168. The waveform storage section 102 stores waveform data. The waveform data may be time series data. The waveform data may be in any form. For example, the waveform data may be data represented by consecutive numerical values. The waveform storage 102 may store waveform data as a generation source of virtual waveform data. The waveform storage unit 102 may store evaluation waveform data for evaluating the waveform evaluation model.
The algorithm storage unit 110 stores an algorithm for generating virtual waveform data from waveform data. The algorithm storage section 110 may store a plurality of algorithms.
The virtual waveform storage 168 stores the virtual waveform data generated by the generation unit 13. The virtual waveform storage 168 may function as the data storage 118.
The registration section 12 performs various registrations. For example, the registration unit 12 registers waveform data. The registration unit 12 stores the registered waveform data in the waveform storage unit 102.
The registration unit 12 may register waveform data as a generation source of the virtual waveform data. The registration unit 12 registers, for example, so-called raw waveform data. The registration unit 12 registers waveform data generated by observing or measuring an arbitrary target system, for example.
The registration unit 12 may register abnormal waveform data generated by observing or measuring the target system in a state where an abnormality occurs in the target system itself. The registration unit 12 may register, for example, abnormal waveform data generated by observing or measuring the target system in a state where abnormality is intentionally generated in the target system. When the target system is a system for manufacturing a product, the registration unit 12 may register abnormal waveform data generated by observing or measuring the target system when an abnormality occurs in the product, or abnormal waveform data generated by observing or measuring the product.
The registration unit 12 may register normal waveform data generated by observing or measuring the target system in a state where the target system is operating normally. In the case where the target system is a system for manufacturing a product, the registration unit 12 may register normal waveform data generated by observing or measuring the target system when the product is normal, and normal waveform data generated by observing or measuring the product.
The generating unit 13 generates virtual waveform data from the waveform data stored in the waveform storage unit 102. The generating section 13 may generate one piece of virtual waveform data from one piece of waveform data. The generating section 13 may generate a plurality of virtual waveform data from one waveform data. The generating section 13 may generate a plurality of virtual waveform data from the plurality of waveform data. The generating section 13 may generate virtual waveform data, which is greater in number than the waveform data, from the plurality of waveform data. The generating section 13 may include a waveform acquiring section 160, an intention determining section 162, a preprocessing section 164, and a virtual waveform generating section 166.
The waveform acquisition unit 160 acquires waveform data as a generation source of virtual waveform data. The waveform acquisition unit 160 may acquire waveform data from the waveform storage unit 102.
The intention determining unit 162 determines the intention of the user. The user may be a person using the system 10. The intention determining portion 162 may determine the intention of the user to generate virtual waveform data from the waveform data.
The intention determining unit 162 determines an algorithm used when generating virtual waveform data from waveform data, for example, as the intention of the user. The intention determining unit 162 may determine one or more algorithms among the algorithms stored in the algorithm storage unit 110 according to an instruction of the user. The intention specifying unit 162 may specify a plurality of algorithms to be continuously applied to the waveform data and an order of application according to an instruction from the user.
The intention determining unit 162 determines, for example, a parameter setting used when generating virtual waveform data from waveform data as the intention of the user. The intention determining unit 162 may determine the parameter setting according to the instruction of the user. As a specific example, the intention determining portion 162 determines one or more parameter settings by accepting a user selection of a plurality of parameter settings. In addition, as a specific example, the intention determining section 162 determines one or more parameter settings by accepting user input of the parameter settings.
The intention determining unit 162 determines, for example, a combination of an algorithm used when generating virtual waveform data from waveform data and parameter setting as the intention of the user. The intention determining unit 162 may determine one or more combinations of the algorithm and the parameter setting according to the instruction of the user.
The intention specification unit 162 specifies preprocessing performed on waveform data as the intention of the user, for example, when generating virtual waveform data from the waveform data. The preprocessing may be a process of extracting features from waveform data. Examples of preprocessing include FFT (fast fourier transform), normalization, filtering, slicing, sampling frequency transformation, and dimensional transformation. The intention determination unit 162 may determine preprocessing to be continuously applied to the waveform data and the order of application according to the instruction of the user.
The intention determination unit 162 determines, as the intention of the user, a combination of at least any one of an algorithm and parameter settings used when generating virtual waveform data from waveform data, and preprocessing performed on the waveform data, for example.
The preprocessing unit 164 performs at least one of a plurality of preprocessing on the waveform data acquired by the waveform acquisition unit 160. The preprocessing section 164 may apply preprocessing determined by the intention determining section 162 to the waveform data acquired by the waveform acquiring section 160. When the intention determining unit 162 determines a plurality of preprocessing to be successively applied to the waveform data and an order to be applied, the preprocessing unit 164 may sequentially apply the plurality of preprocessing to the waveform data in this order.
The virtual waveform generation unit 166 generates virtual waveform data from the waveform data acquired by the waveform acquisition unit 160. For example, the virtual waveform generation unit 166 generates virtual abnormal waveform data from the abnormal waveform data. For example, the virtual waveform generation unit 166 generates virtual normal waveform data from normal waveform data.
The virtual waveform generation unit 166 may generate a plurality of virtual waveform data from one waveform data. The virtual waveform generation unit 166 may generate a plurality of waveform data from one waveform data by using at least any one of a plurality of algorithms, a plurality of parameter settings, and a random number. For example, the virtual waveform generation unit 166 generates a plurality of virtual waveform data by using a plurality of algorithms for one waveform data. For example, the virtual waveform generation unit 166 generates a plurality of virtual waveform data by using a plurality of parameter settings for one waveform data. For example, the virtual waveform generation unit 166 generates a plurality of virtual waveform data by using a random number for one waveform data. By using a plurality of algorithms, a plurality of parameter settings, random numbers, and the like, various kinds of variable virtual waveform data can be generated from one waveform data.
The virtual waveform generation section 166 may generate virtual waveform data from the waveform data by a method of reflecting the intention of the user determined by the intention determination section 162. The virtual waveform generation unit 166 generates virtual waveform data from the waveform data using, for example, one or more algorithms determined by the intention determination unit 162. When the intention determining unit 162 determines a plurality of algorithms to be continuously applied to the waveform data and the order of application, the virtual waveform generating unit 166 may generate virtual waveform data from the waveform data using the plurality of algorithms in this order. The virtual waveform generation unit 166 generates virtual waveform data from the waveform data, for example, using one or more parameter settings determined by the intention determination unit 162. The virtual waveform generation unit 166 generates virtual waveform data from the waveform data using the algorithm and the parameter setting, for example, for each of the combinations of the algorithm and the parameter setting determined by the intention determination unit 162. The virtual waveform generation unit 166 generates virtual waveform data, for example, from waveform data after the preprocessing unit 164 has performed one or more preprocessing steps determined by the intention determination unit 162. It is sometimes difficult to generate virtual waveform data suitable for learning from waveform data, but a variety of virtual waveform data can be generated by a method reflecting the intention of the user, whereby the possibility of generating virtual waveform data suitable for learning can be increased.
The virtual waveform generation unit 166 can process waveform data so as not to significantly impair physical meaning, and generate virtual waveform data. The virtual waveform generator 166 may generate virtual waveform data by adding noise to the waveform data, machining the entire or partial section of the waveform data, expanding the waveform data in the vertical axis direction of the graph, contracting the waveform data in the vertical axis direction of the graph, expanding the waveform data in the horizontal axis direction of the graph, contracting the waveform data in the horizontal axis direction of the graph, or the like.
The virtual waveform generation unit 166 generates virtual waveform data using, for example, an algorithm (sometimes referred to as a noise addition processing algorithm) that adds a random value to time-series waveform data, the random value being a random value based on a normal distribution having a mean value of 0 and a standard deviation of σ. When the noise addition processing algorithm is used, the virtual waveform generation unit 166 can generate a plurality of virtual waveform data from one waveform data by using a plurality of random values. When the noise addition processing algorithm is used, the virtual waveform generator 166 can generate a plurality of virtual waveform data from one waveform data by using a plurality of parameter settings such as a plurality of standard deviations σ.
The virtual waveform generation unit 166 generates virtual waveform data using, for example, an algorithm (sometimes referred to as a scale deformation algorithm) for accumulating random values based on a normal distribution having a mean value of 1 and a standard deviation of σ with respect to time-series waveform data. In the case of using the scale deformation algorithm, the virtual waveform generation section 166 can generate a plurality of virtual waveform data from one waveform data by using a plurality of random values. When the scale deformation algorithm is used, the virtual waveform generator 166 may generate a plurality of virtual waveform data from one waveform data by using a plurality of parameter settings such as a plurality of standard deviations σ.
The virtual waveform generation unit 166 determines, for example, a reduction_ratio, determines a start point within a range of "time-series data size× (1-reduction_ratio)" from 0 of time-series waveform data, cuts out a section from the start point to the end point with a point that has advanced by "time-series data size× (reduction_ratio)" from the determined start point, and generates virtual waveform data using an algorithm (sometimes referred to as an expansion processing algorithm) that enlarges the cut-out section so as to become the original size. When the dilation processing algorithm is used, the virtual waveform generator 166 may generate a plurality of pieces of virtual waveform data from one piece of waveform data by using a plurality of reduction ratios.
The virtual waveform generation unit 166 generates virtual waveform data using, for example, the following algorithm (sometimes referred to as an expansion/contraction processing algorithm): generating candidates of warp_ scales, randomly determining a value from among the candidates of warp_ scales, determining a window_ratio, calculating a "warp_size=window_ratio×time-series data size", randomly securing a region of warp_size in a range of 1 to "size in time-series direction-warp_size-1", determining a waveform preceding the region of warp_size "," a waveform following the region of warp_size ", and" combining again a waveform preceding the region of warp_size "," a waveform following the region of warp_size ", with respect to a region of" warp_size "to a size of" warp_size×warp_ scales ", and returning the waveform preceding the region of warp_size", the waveform of the region of warp_size after the enlargement or reduction, to the original size. In the case of using the expansion/contraction processing algorithm, the virtual waveform generation unit 166 may generate a plurality of virtual waveform data from one waveform data by using a plurality of parameter settings, for example, generate a plurality of candidates for warp_ scales, use a plurality of candidates for warp_ scales, or use a plurality of windows_ratio.
The virtual waveform generation unit 166 generates virtual waveform data using, for example, the following algorithm (sometimes referred to as a scale random deformation algorithm): the number of passing points is determined, a random value (y-axis information) of the same number as the determined number of passing points is determined based on a normal distribution having a mean value of 1 and a standard deviation of sigma, the passing points (x-axis information) are arranged at equal intervals by dividing the size of the time series direction by the number of passing points, a spline curve having the size of the time series direction of the waveform data as the x-axis is created using the y-axis information and the x-axis information, and the created spline curve and the waveform data are integrated. In the case of using the scale random deformation algorithm, the virtual waveform generation section 166 can generate a plurality of virtual waveform data from one waveform data by using a plurality of random values. In the case of using the scale random deformation algorithm, the virtual waveform generation unit 166 may generate a plurality of virtual waveform data from one waveform data by using a plurality of parameter settings such as the number of transit points of a plurality of patterns or using a plurality of standard deviations σ.
The virtual waveform generation section 166 may generate a virtual waveform data set including a plurality of virtual waveform data from one or more waveform data. The virtual waveform generation unit 166 may generate various virtual waveform data sets by using different algorithms, or using different parameter settings, or using different preprocessing.
For example, the virtual waveform generation unit 166 generates a virtual abnormal waveform data set including a plurality of virtual abnormal waveform data. The virtual waveform generation section 166 may generate a plurality of virtual abnormal waveform data sets. For example, the virtual waveform generation unit 166 generates a virtual normal waveform data set including a plurality of virtual normal waveform data.
The virtual waveform storage unit 168 stores the virtual waveform data generated by the virtual waveform generation unit 166. The virtual waveform storage 168 may store virtual identification information indicating data that is virtually generated in association with virtual waveform data. By associating the virtual identification information with the virtual waveform data, the virtual waveform data can be identified from the original waveform data.
The virtual waveform storage 168 may store virtual identification information in association with the virtual waveform data set. The virtual waveform storage 168 may store virtual identification information and abnormality identification information indicating that the virtual waveform is abnormal data in association with the virtual abnormal waveform data set. The virtual waveform storage 168 may store virtual identification information and normal identification information indicating that it is normal data in association with the virtual normal waveform data set.
The virtual waveform storage unit 168 may store a plurality of virtual waveform data sets generated by the virtual waveform generation unit 166 in association with an algorithm used when generating the virtual waveform data set, the data set, and recipe information (recipe information) capable of identifying preprocessing, respectively. For example, when the virtual waveform generation unit 166 generates a virtual waveform data set from waveform data using an algorithm and a parameter setting for each of a plurality of combinations of the algorithm and the parameter setting, the virtual waveform storage unit 168 associates each of the plurality of virtual waveform data sets with recipe information capable of identifying a combination of the algorithm and the parameter setting used for generating the virtual waveform data set, and stores the plurality of virtual waveform data sets. By referring to the recipe information, it is thereby possible to grasp under what conditions the virtual waveform data set is generated. In addition, the virtual waveform data set can be easily managed by the recipe information.
The virtual waveform storage 168 may store a plurality of virtual waveform data sets by associating source waveform identification information capable of identifying waveform data to be a generation source of the virtual waveform data set with the plurality of virtual waveform data sets generated by the virtual waveform generation unit 166. The virtual waveform storage 168 may store a plurality of virtual waveform data sets in association with source waveform identification information capable of identifying one waveform data when the virtual waveform generation unit 166 generates the plurality of virtual waveform data sets from the one waveform data. The virtual waveform storage 168 may store a plurality of virtual waveform data sets in association with source waveform identification information capable of identifying the plurality of waveform data sets when the virtual waveform generation unit 166 generates the plurality of virtual waveform data sets from the plurality of waveform data sets. In this way, when a certain virtual waveform data set is to be compared with the source waveform data, and when a certain virtual waveform data set is to be used for learning, the source waveform data can be easily specified when the source waveform data is to be used together.
In the case where the virtual waveform generation unit 166 generates a virtual waveform data set from the waveform data subjected to the preprocessing by the preprocessing unit 164, the virtual waveform storage unit 168 may store the virtual waveform data set in association with preprocessing identification information capable of identifying the preprocessing. Thus, it is possible to easily grasp, for a certain virtual waveform data set, what preprocessing is performed on the source waveform data.
The virtual waveform confirmation processing unit 14 performs processing for causing the user to confirm the virtual waveform data stored in the virtual waveform storage unit 168. The virtual waveform confirmation processing unit 14 may include a similarity display control unit 170, a range specification receiving unit 172, and an in-range waveform selecting unit 174.
The similarity display control section 170 performs control to display, to the user, display data indicating a similarity between the virtual waveform data generated from the waveform data by the virtual waveform generation section 166 and the waveform data that is the generation source of the virtual waveform data. The similarity display control unit 170 causes, for example, a display included in the system 10 to display data. The similarity display control unit 170 may transmit the display data to any device and cause the device to display the display data.
The similarity display control section 170 may perform control to display, to the user, display data representing the similarity between the plurality of virtual waveform data generated from the waveform data by the virtual waveform generation section 166 and the waveform data as the generation source. The similarity display control unit 170 may perform control to display, to the user, display data indicating how much the plurality of virtual waveform data are generated with respect to the waveform data that is the generation source of the plurality of virtual waveform data. The similarity display control unit 170 calculates, for example, the similarity between each of the plurality of pieces of virtual waveform data and the waveform data as the generation source, and displays display data including at least any one of a histogram and a graph indicating the number of pieces of virtual waveform data for each of the similarities. The similarity may be, for example, a value indicating that the closer to 1, the more similar to the source waveform data, and the closer to 0, the more apart from the source waveform data.
The similarity display control unit 170 may use, for example, a correlation coefficient between the waveform data and the virtual waveform data as a similarity between the waveform data and the virtual waveform data. The similarity display control unit 170 may calculate the similarity between the waveform data and the virtual waveform data by any method as long as the similarity between the waveform data and the virtual waveform data can be expressed.
In the case where a plurality of virtual waveform data are generated from one waveform data, the similarity display control section 170 may calculate the similarity of each of the plurality of virtual waveform data to the waveform data. In the case of generating a plurality of pieces of virtual waveform data from a plurality of pieces of waveform data, the similarity display control unit 170 may calculate, for each of the plurality of pieces of virtual waveform data, the similarity between the virtual waveform data and the plurality of pieces of waveform data, and average the calculated similarity as the similarity between the waveform data and the virtual waveform data.
The similarity display control section 170 may perform control of performing the following display to the user: display data representing a similarity between the virtual waveform data and waveform data that is a generation source of the virtual waveform data, and another display data representing a similarity between the virtual waveform data and another waveform data different from the waveform data that is the generation source of the virtual waveform data are displayed. The similarity display control unit 170 performs, for example, control for displaying the following to the user: display data representing the similarity of virtual abnormal waveform data generated from the abnormal waveform data and the abnormal waveform data, and display data representing the similarity of the virtual abnormal waveform data and the normal waveform data are displayed. In this way, in addition to the correlation between the generated virtual abnormal waveform data and the source abnormal waveform data, the user can be made aware of the correlation between the generated virtual abnormal waveform data and the normal waveform data. By grasping the correlation between the virtual abnormal waveform data and the normal waveform data, for example, it is possible to identify virtual abnormal waveform data having a low similarity to the normal waveform data and a high possibility of being effective for learning, among the virtual abnormal waveform data having the same similarity to the source abnormal waveform data.
The similarity display control section 170 may perform control to display, to the user, display data representing the similarity between each of the plurality of virtual waveform data sets generated by the virtual waveform generation section 166 and the waveform data as the generation source. The similarity display control unit 170 calculates, for each of a plurality of virtual waveform data sets, for example, the similarity between a plurality of virtual waveform data included in the virtual waveform data set and waveform data as a generation source, and displays display data including at least any one of a histogram and a graph showing the number of virtual waveform data of each similarity. Thus, for example, a histogram corresponding to each of the plurality of virtual waveform data sets can be presented to the user, and it is possible to investigate which virtual waveform data set is used for learning. Specifically, it is possible to select only a virtual waveform data set having a high similarity to the source waveform data among a plurality of virtual waveform data sets, or to try to include a virtual waveform data set having a low similarity to the source waveform data, or the like.
The range specification receiving unit 172 receives a user specification of a range of similarity of the display data displayed by the similarity display control unit 170. For example, the range specification receiving unit 172 receives specification of a range of the similarity with respect to the histogram or graph showing the number of virtual waveform data according to each similarity with the source waveform data, which is displayed by the similarity display control unit 170. The range specification accepting unit 172 accepts input of the user for the upper limit and the lower limit of the similarity, for example. The range specification receiving unit 172 receives, for example, an input of selecting a range of similarity for a histogram or a graph in display data by a user.
The in-range waveform selection unit 174 selects a plurality of pieces of virtual waveform data corresponding to the range in which the range designation accepting unit 172 accepts the designated similarity. For example, the in-range waveform selection unit 174 selects virtual waveform data corresponding to each similarity within the range in which the range designation reception unit 172 has received the designated similarity.
The virtual waveform storage 168 may store a virtual waveform data set including the plurality of virtual waveform data selected by the in-range waveform selection unit 174. The virtual waveform storage 168 may store a virtual waveform data set including the plurality of virtual waveform data selected by the in-range waveform selection unit 174 as a data set for learning. Thus, the user can easily investigate which range of virtual waveform data similar to the source waveform data is used, which range of virtual waveform data similar to the source waveform data is not used, and the like, among the plurality of virtual waveform data included in the one or more virtual waveform data sets.
The similarity display control unit 170 may perform control to display, to the user, display data capable of comparing the virtual waveform data set with waveform data as a generation source at a waveform level. For example, when a virtual waveform data set is generated from one waveform data, the similarity display control unit 170 displays display data in which one waveform data and a plurality of virtual waveform data included in the virtual waveform data set are arranged on a graph. For example, when a virtual waveform data set is generated from a plurality of waveform data sets, the similarity display control unit 170 displays display data in which a plurality of waveform data sets and a plurality of virtual waveform data included in the virtual waveform data set are arranged on a graph. For example, when generating a virtual waveform data set from a plurality of waveform data sets, the similarity display control unit 170 displays display data in which an average waveform of the plurality of waveform data sets and a plurality of virtual waveform data included in the virtual waveform data set are arranged on a graph.
The learning unit 15 performs machine learning using the virtual waveform data generated by the generating unit 13. The learning unit 15 may include a learning data storage unit 106, an algorithm storage unit 144, a learning execution unit 140, and a waveform evaluation model storage unit 146.
The learning data storage 106 stores learning data for learning. The learning data storage unit 106 may acquire the waveform data stored in the waveform storage unit 102 from the waveform storage unit 102 and store the waveform data. The learning data storage unit 106 may acquire the virtual waveform data stored in the virtual waveform storage unit 168 from the virtual waveform storage unit 168, and store the virtual waveform data.
The learning data storage unit 106 may acquire and store virtual waveform data selected as data for learning from among the plurality of virtual waveform data stored in the virtual waveform storage unit 168 from the virtual waveform storage unit 168. The virtual waveform storage unit 168 acquires, for example, from the virtual waveform storage unit 168, a virtual waveform data set selected as a data set for learning among the plurality of virtual waveform data sets stored in the virtual waveform storage unit 168, and stores the virtual waveform data set. The learning data storage unit 106 may acquire and store a virtual waveform data set stored by the virtual waveform storage unit 168 as a data set used for learning from the virtual waveform storage unit 168. The learning data storage unit 106 may acquire and store information such as recipe information, source waveform identification information, preprocessing identification information, virtual identification information, abnormality identification information, and normal identification information associated with the virtual waveform data set from the virtual waveform storage unit 168 together with the virtual waveform data set.
The algorithm storage section 144 stores a plurality of learning algorithms. Examples of the learning algorithm include NN (Neural Network ), VGG (Visual Geometry Group, visual geometry group), SVM (Support Vector Machine ), and the like, but any algorithm may be used.
The learning execution unit 140 generates a waveform evaluation model that outputs the evaluation result of the input waveform data by performing machine learning using the learning data stored in the learning data storage unit 106. The learning execution unit 140 may generate the waveform evaluation model using a learning algorithm specified by the user among the plurality of learning algorithms stored in the algorithm storage unit 144.
The learning execution unit 140 may evaluate the generated waveform evaluation model. The learning execution unit 140 performs evaluation of the waveform evaluation model using, for example, waveform data for evaluation registered in advance by the registration unit 12 and stored in the waveform storage unit 102. The learning execution unit 140 may perform evaluation of the waveform evaluation model using waveform data for evaluation selected by the user from among the plurality of waveform data for evaluation.
The learning execution unit 140 may evaluate the waveform evaluation model without using the virtual waveform data. For example, when virtual waveform data or a virtual waveform data set is selected as waveform data for evaluation, the learning execution unit 140 does not use the selected virtual waveform data for evaluation. For example, the learning execution unit 140 controls so that the user cannot select the virtual waveform data and the virtual waveform data set when selecting the waveform data for evaluation. Thus, the waveform data generated virtually can be not used for evaluation of the waveform evaluation model, and the possibility of deterioration in reliability of the evaluation of the waveform evaluation model can be reduced.
The waveform evaluation model storage unit 146 stores the waveform evaluation model generated by the learning execution unit 140. When the learning execution section 140 generates a waveform evaluation model by performing machine learning using the virtual waveform data associated with the preprocessing identification information, the waveform evaluation model storage section 146 may store the waveform evaluation model in association with the preprocessing identification information.
The evaluation execution unit 16 executes evaluation of waveform data using the waveform evaluation model generated by the learning unit 15. The evaluation execution unit 16 may include a model acquisition unit 180, an input waveform acquisition unit 182, a waveform input unit 184, and an evaluation result output control unit 186.
The model acquisition unit 180 acquires the waveform evaluation model stored in the waveform evaluation model storage unit 146. When the preprocessing identification information is associated with the waveform evaluation model, the model acquisition unit 180 acquires the waveform evaluation model and the preprocessing identification information.
The input waveform acquisition unit 182 acquires input waveform data to be evaluated. The input waveform acquisition section 182 may acquire input waveform data instructed to be input by the user.
The waveform input unit 184 inputs the input waveform data acquired by the input waveform acquisition unit 182 into the waveform evaluation model acquired by the model acquisition unit 180. When the model acquisition unit 180 acquires the waveform evaluation model and the preprocessing identification information, the waveform input unit 184 may apply preprocessing indicated by the preprocessing identification information to the input waveform data acquired by the input waveform acquisition unit 182 and input the result to the waveform evaluation model. This can improve the accuracy of evaluating the waveform, as compared with the case where preprocessing shown by the preprocessing identification information is not performed. In addition, when the evaluation execution unit 16 performs the evaluation, the load generated by the virtual waveform data generated by performing what preprocessing is performed on the waveform evaluation model acquired by the management model acquisition unit 180 can be eliminated.
The evaluation result output control section 186 controls to output the evaluation result of the input waveform data output from the waveform evaluation model. The evaluation result output control unit 186 causes, for example, a display of the system 10 to display the evaluation result of the input waveform data. The evaluation result output control unit 186 may transmit the evaluation result of the input waveform data to any device, and cause the device to display the evaluation result.
The system 10 may be constructed of one device. The system 10 may also be comprised of a plurality of devices. For example, the system 10 includes: a device including a storage unit 11, a registration unit 12, a generation unit 13, a virtual waveform confirmation processing unit 14, and a learning unit 15, and a device including an evaluation execution unit 16. In this case, the model acquisition unit 180 receives the waveform evaluation model stored in the waveform evaluation model storage unit 146 from the system 10. The apparatus including the evaluation execution unit 16 may be realized by a program that causes a conventional apparatus to function as the evaluation execution unit 16. By this program, the evaluation execution unit 16 is installed in a device that acquires the waveform evaluation model stored in the waveform evaluation model storage unit 146, and can realize waveform evaluation using the waveform evaluation model. When the preprocessing identification information is associated with the waveform evaluation model, the waveform input unit 184 automatically performs preprocessing indicated by the preprocessing identification information on the input waveform data acquired by the input waveform acquisition unit 182 and inputs the input waveform data to the waveform evaluation model, so that the user side of the evaluation execution unit 16 can save the time and effort for performing management to perform specific preprocessing when inputting the input waveform data to the waveform evaluation model, and can prevent deterioration in evaluation accuracy due to input of the input waveform data to the waveform evaluation model without performing specific preprocessing.
The system 10 may also include: a device including a storage unit 11, a registration unit 12, a generation unit 13, and a virtual waveform confirmation processing unit 14, a device including a learning unit 15, and a device including an evaluation execution unit 16. The system 10 may be implemented by other device configurations than these.
Fig. 6 schematically shows an example of a flow of the waveform evaluation model generation process of the system 10. Here, a process of generating a large number of virtual abnormal waveform data by using a small number of abnormal waveform data stored in the waveform storage unit 102 and generating a normal abnormal evaluation model for evaluating whether the input waveform data is a normal waveform or an abnormal waveform by using the virtual abnormal waveform data, the abnormal waveform data, and the normal waveform data will be described as an example.
In step 102 (step is sometimes omitted as S), the intention determining unit 162 determines the intention of the user. The intention determining unit 162 may determine a plurality of combinations of preprocessing, algorithm, and parameter setting according to the instruction of the user.
In S104, the preprocessing unit 164 and the virtual waveform generation unit 166 perform preprocessing on a small amount of abnormal waveform data for one of the combinations, and then generate a virtual abnormal waveform data set using an algorithm and parameter settings.
In S106, the virtual waveform storage 168 stores the virtual identification information, the source waveform identification information, the preprocessing identification information, and the recipe information in association with the virtual abnormal waveform data set generated in S104.
If the generation of the virtual abnormal waveform data set has not been completed for all of the combinations specified in S102 (S108: no), the flow returns to S104, and the preprocessing unit 164 and the virtual waveform generation unit 166 generate the virtual abnormal waveform data set for the next combination, and if it is determined that the virtual abnormal waveform data set has been completed (S108: yes), the flow proceeds to S110.
In S110, the similarity display control section 170 performs control to display, to the user, display data indicating the similarity between each of the plurality of virtual abnormal waveform data sets and a small amount of abnormal waveform data as a generation source. In S112, the range specification reception unit 172 and the in-range waveform selection unit 174 select virtual abnormal waveform data for learning. When the range specification receiving unit 172 and the in-range waveform selecting unit 174 receive the specification of the range of the similarity by the user, the plurality of pieces of virtual abnormal waveform data corresponding to the range in which the specified similarity is received are selected as the virtual abnormal waveform data for learning. When all the virtual abnormal waveform data sets are selected by the user, the range specification reception unit 172 and the in-range waveform selection unit 174 may select all the virtual abnormal waveform data included in all the virtual abnormal waveform data sets as the virtual abnormal waveform data for learning.
In S114, the learning execution unit 140 generates a normal abnormality evaluation model by executing machine learning using the virtual abnormal waveform data, the normal waveform data, and the abnormal waveform data selected by the range specification reception unit 172 and the in-range waveform selection unit 174 in S112. Further, after S102 to S112 are performed a plurality of times, the process may proceed to S114. In S116, the learning execution unit 140 evaluates the normal abnormality evaluation model generated in S114, using the waveform data for evaluation stored in the waveform storage unit 102.
When the registration instruction of the normal abnormality evaluation model of the user is received (yes in S118), the routine proceeds to S120, and when the registration instruction is not received (no in S118), the routine is terminated. In S120, the waveform evaluation model storage unit 146 stores the preprocessing identification information in association with the normal abnormality evaluation model generated in S114. Then, the process ends.
By the system 10 executing the processing shown in fig. 6, even if the number of abnormal waveform data is small, a plurality of pieces of virtual abnormal waveform data reflecting the user's intention can be generated, and the normal abnormal evaluation model can be generated using the virtual abnormal waveform data selected by the user while confirming the display data, and the generation of the normal abnormal evaluation model with high evaluation accuracy can be facilitated.
Fig. 7 schematically illustrates an example of management data 190 for managing virtual waveform data. The system 10 may generate a plurality of items and manage various data for each item. Thus, for example, when a plurality of types of data such as data for evaluating a normal abnormality of the ball screw and data for evaluating a normal abnormality of the motor are desired, management thereof can be easily performed.
For example, the registration unit 12 generates an item in accordance with an instruction from the user and registers waveform data. The registration unit 12 may register abnormal waveform data, normal waveform data, waveform data for evaluation, and the like. The waveform storage unit 102 stores waveform data in association with items. When the virtual waveform generation unit 166 generates a virtual waveform data set using waveform data corresponding to an item, the virtual waveform storage unit 168 stores the virtual waveform data set and associated information in association with the item. Examples of the associated information include the number of data included in the virtual waveform data set, source waveform identification information, preprocessing identification information, and recipe information.
Fig. 7 illustrates a state in which a data set of "abnormal data a of the ball screw" and a data set of "abnormal data B of the ball screw" are registered in an item of the name "virtual waveform of the ball screw". Based on the management data 190, the user can confirm that the number of data of the "abnormal data a of the ball screw" is 80, the source waveform is the original abnormal data a, no preprocessing is performed, the number of data of the "abnormal data B of the ball screw" is 80, the source waveform is the original abnormal data a, downsampling with the sampling rate of 6 is performed as preprocessing, the expansion and contraction processing algorithm with the warp_ scales of [0.5,2.0] and the window_ratio of 0.1 is used, and then what preprocessing, algorithm, and parameter setting are used to generate virtual waveform data as a study material can be performed.
Fig. 8 is an explanatory diagram for explaining an expansion processing algorithm. Taking the case where reduce_ratio=0.9 as an example, the virtual waveform data 308 is generated from the waveform data 300 will be described.
The virtual waveform generation unit 166 randomly determines a start point 303 in a range 302 of a data size 301× (1-reduction_ratio) of the time series of the waveform data 300 from the start point of the time series of the waveform data 300. The virtual waveform generation unit 166 sets a point, which is advanced from the start point 303 by the amount of the time-series data size 301×reduction_ratio, as the end point 304. The virtual waveform generation unit 166 cuts out a section 305 from the start point 303 to the end point 304. The virtual waveform generation unit 166 enlarges the cut section 305 to the original size as virtual waveform data 308.
Even if the reduce_ratio is fixed, by randomly determining the start point 303, it is possible to generate different virtual waveform data 308 from the same waveform data 300 each time. By altering the reduce_ratio, more varied virtual waveform data 308 can be generated.
Fig. 9 is an explanatory diagram for explaining an expansion/contraction processing algorithm. Here, description will be made taking, as an example, a case where warp_ scales = [0.5, 2], window_ratio=0.1, and virtual waveform data is generated from the waveform data 300.
The virtual waveform generation unit 166 randomly determines a value from the warp_ scales. The virtual waveform generation section 166 calculates the data size 301 of the warp_size=window_ratio×time series. The virtual waveform generation unit 166 randomly secures a region of the warp_size in a range 311 of 1 to "time-series data size 301-warp_size-1", and determines three regions, that is, a waveform 312 preceding the region of the warp_size, a region waveform 313 of the warp_size, and a waveform 314 following the region of the warp_size. The virtual waveform generation unit 166 generates virtual waveform data by enlarging or reducing the area of the waveform 313 to the size of "warp_size×warp_ scales" and combining the waveform 312, the enlarged or reduced waveform 313, and the waveform 314 again to return to the original size.
By randomly determining one value from warp_ scales, different virtual waveform data is generated each time even from the same waveform data 300. By presetting a plurality of candidates of warp_ scales, more varied virtual waveform data can be generated.
Fig. 10 is an explanatory diagram for describing a scale random deformation algorithm. Here, a case will be described in which the number of pass points is set to 5 and virtual waveform data is generated from waveform data 300.
The virtual waveform generation unit 166 determines a random value (y-axis information) equal to the number of transit points based on a normal distribution having a mean value of 1 and a standard deviation of σ, and equally divides the size of the time series direction by the number of transit points to arrange the transit points 321 (x-axis information). The virtual waveform generation unit 166 creates a spline curve 322 having the data size 301 of the time series of waveform data as the x-axis, using the y-axis information and the x-axis information. The virtual waveform generation unit 166 generates virtual waveform data by integrating the generated spline curve 322 and waveform data.
Even if the number of passing points is set to be fixed, by determining a random value as y-axis information, virtual waveform data is generated differently every time even from the same waveform data 300. By changing the number of transit points, more varied virtual waveform data can be generated.
Fig. 11 schematically shows an example of the display data 400 displayed by the similarity display control unit 170. In the display data 400, the horizontal axis represents similarity between waveform data as a generation source of virtual waveform data and virtual waveform data, and the vertical axis represents frequency number as the number of virtual waveform data.
In fig. 11, display data 400 including a graph 412 and a histogram 414 corresponding to a first virtual waveform dataset of three virtual waveform datasets, a graph 422 and a histogram 424 corresponding to a second virtual waveform dataset, and a graph 432 and a histogram 424 corresponding to a third virtual waveform dataset are illustrated.
The similarity display control unit 170 may switch the display of each of the graphs 412, 414, 422, 424, 432, and 434 on and off according to the instruction of the user. The similarity display control unit 170 may apply semitransparent displays to the graph 412, the histogram 414, the graph 422, the histogram 424, the graph 432, and the histogram 434, respectively.
The display data 400 can easily grasp how much each virtual waveform data set is similar to the source waveform data, how much deviation is generated, and the like.
The range specification receiving unit 172 may receive specification of a range of the similarity with respect to a virtual waveform data set selected by the user among the plurality of virtual waveform data sets. For example, when the user selects the graph 412 and the histogram 414, the range specification accepting section 172 accepts specification of the similarity range of the graph 412 and the histogram 414. The in-range waveform selection unit 174 may select virtual waveform data corresponding to a range in which the specified similarity is accepted, from among the graph 412 and the histogram 414. This enables the following operations: it is first determined which one of the plurality of virtual waveform data sets to be used, and then further selects which range of the virtual waveform data set to use.
The range specification receiving unit 172 may receive specification of a range of similarity to a plurality of virtual waveform data sets. For example, the range specification receiving unit 172 receives specification of the range of the similarity with respect to the graph 412, the histogram 414, the graph 422, the histogram 424, the graph 432, and the histogram 434. The in-range waveform selection unit 174 may select virtual waveform data corresponding to a range in which the specified similarity is accepted, from among the graphs 412, 414, 422, 424, 432, and 434.
In the above embodiment, the abnormal normal evaluation model is exemplified as an example of the waveform evaluation model, but is not limited thereto. The system 10 can generate various waveform evaluation models such as a failure time evaluation model of the failure time of the evaluation target system.
The system 10 is not limited to one-dimensional data such as waveform data, and may be configured to target two-dimensional data such as an image. For example, the system 10 may generate virtual two-dimensional data from two-dimensional data. The system 10 may apply the virtual waveform generation function described above to two-dimensional data on the premise that the horizontal axis of the two-dimensional data is time and the vertical axis is frequency. For example, the system 10 applies the above-described virtual waveform generation function to frequency components obtained by cutting out two-dimensional data for each column (in time). By performing the same processing for all the times and then combining, the virtual waveform generation function can be applied to the two-dimensional data.
Fig. 12 schematically shows an example of a hardware configuration of a computer 1200 functioning as the system 10 or a part of the system 10. The computer 1200 functioning as a part of the system 10 functions as a device including, for example, a storage unit 11, a registration unit 12, a generation unit 13, and a virtual waveform confirmation processing unit 14. The computer 1200 functioning as a part of the system 10 functions as a device including, for example, a storage unit 11, a registration unit 12, a generation unit 13, a virtual waveform confirmation processing unit 14, and a learning unit 15. The computer 1200 functioning as a part of the system 10 functions as a device including the learning unit 15, for example. The computer 1200 functioning as a part of the system 10 functions as a device including the evaluation execution unit 16, for example. The program installed on the computer 1200 can cause the computer 1200 to function as one or more "parts" of the apparatus according to the present embodiment, or can cause the computer 1200 to perform an operation associated with the apparatus according to the present embodiment or the one or more "parts", and/or can cause the computer 1200 to perform a process according to the present embodiment or a step of the process. Such programs may be executed by the CPU 1212 to cause the computer 1200 to perform certain operations associated with some or all of the blocks in the flowcharts and block diagrams described in this specification.
The computer 1200 according to the present embodiment includes a CPU 1212, a RAM 1214, and a graphics controller 1216, which are connected to each other through a main controller 1210. The computer 1200 further includes a communication interface 1222, a storage device 1224, an input/output unit such as a DVD drive and an IC card drive, which are connected to the host controller 1210 via the input/output controller 1220. The DVD drive may be a DVD-ROM drive, a DVD-RAM drive, or the like. Storage 1224 may be a hard disk drive, a solid state drive, or the like. The computer 1200 also includes a ROM 1230 and a conventional input-output unit such as a keyboard, which are connected to the input-output controller 1220 via an input-output chip 1240.
The CPU 1212 operates according to programs stored in the ROM 1230 and the RAM 1214, thereby controlling the respective units. The graphics controller 1216 acquires image data generated by the CPU 1212 into a frame buffer or the like provided in the RAM 1214 or itself, and causes the image data to be displayed on the display device 1218.
The communication interface 1222 communicates with other electronic devices via a network. The storage device 1224 stores programs and data used by the CPU 1212 in the computer 1200. The DVD drive reads a program or data from a DVD-ROM or the like and supplies it to the storage device 1224. The IC card driver reads and/or writes programs and data from and/or to the IC card.
The ROM 1230 stores a boot program or the like executed by the computer 1200 at the time of activation and/or a program depending on hardware of the computer 1200. The input/output chip 1240 may also connect various input/output units with the input/output controller 1220 via a USB port, a parallel port, a serial port, a keyboard port, a mouse port, etc.
The program is provided by a computer readable storage medium such as a DVD-ROM or an IC card. The program is read from a computer-readable storage medium, installed in a storage device 1224, RAM 1214, or ROM 1230, which are examples of computer-readable storage media, and executed by the CPU 1212. The information processing described in these programs is read by the computer 1200, and cooperation between the programs and the above-described various types of hardware resources is brought about. The apparatus or method may be configured to: the operation or processing of information is achieved through the use of the computer 1200.
For example, in the case of performing communication between the computer 1200 and an external device, the CPU 1212 may execute a communication program loaded into the RAM 1214 and command a communication process to the communication interface 1222 based on a process described in the communication program. The communication interface 1222 reads transmission data stored in a transmission buffer area provided in a recording medium such as a RAM 1214, a storage device 1224, a DVD-ROM, or an IC card, transmits the read transmission data to a network, or writes reception data received from the network to a reception buffer area provided on the recording medium, or the like, under the control of the CPU 1212.
In addition, the CPU 1212 may read all or a necessary portion of a file or a database stored in an external recording medium such as a storage device 1224, a DVD drive (DVD-ROM), an IC card, or the like into the RAM 1214, and perform various types of processing on data on the RAM 1214. Next, the CPU 1212 may write the processed data back to the external recording medium.
Various types of information such as various types of programs, data, tables, and databases may be stored in the recording medium, and may be subjected to information processing. The CPU 1212 can perform various types of processing including various types of operations described anywhere in the present disclosure specified by an instruction sequence of a program, information processing, condition judgment, conditional branching, unconditional branching, retrieval/replacement of information, and the like on data read from the RAM 1214, and write the result back to the RAM 1214. In addition, the CPU 1212 may retrieve information in files, databases, or the like within the recording medium. For example, in the case where a plurality of entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored in the recording medium, the CPU 1212 may retrieve an entry, from among the plurality of entries, for which the attribute value of the first attribute matches the specified condition, and read the attribute value of the second attribute stored in the entry, thereby acquiring the attribute value of the second attribute associated with the first attribute satisfying the predetermined condition.
The programs or software modules described above may be stored on the computer 1200 or in a computer readable storage medium near the computer 1200. In addition, a recording medium such as a hard disk or a RAM provided in a server system connected to a dedicated communication network or the internet can be used as a computer-readable storage medium, thereby providing the program to the computer 1200 via the network.
The blocks in the flowcharts and block diagrams in the present embodiment may represent steps of a process of performing the operation or "parts" of the apparatus having the function of performing the operation. The specific steps and "parts" may be implemented by dedicated circuitry, programmable circuitry provided in conjunction with computer-readable instructions stored on a computer-readable storage medium, and/or a processor provided in conjunction with computer-readable instructions stored on a computer-readable storage medium. The dedicated circuitry may comprise digital and/or analog hardware circuitry, as well as Integrated Circuits (ICs) and/or discrete circuits. The programmable circuitry may include reconfigurable hardware circuits such as Field Programmable Gate Arrays (FPGAs) and Programmable Logic Arrays (PLAs) including logical products, logical sums, exclusive or, nand, nor, and other logical operations, flip-flops, registers, and memory elements.
The computer-readable storage medium may comprise any tangible device capable of storing instructions for execution by a suitable device, with the result that the computer-readable storage medium having instructions stored therein comprises an article of manufacture including instructions which can be executed to produce an element for performing the operations specified in the flowchart or block diagram block or blocks. As examples of the computer readable storage medium, an electronic storage medium, a magnetic storage medium, an optical storage medium, an electromagnetic storage medium, a semiconductor storage medium, or the like may be included. As more specific examples of the computer-readable storage medium, a floppy disk (registered trademark), a magnetic disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an electrically erasable programmable read-only memory (EEPROM), a Static Random Access Memory (SRAM), a compact disc read-only memory (CD-ROM), a Digital Versatile Disk (DVD), a blu-ray (registered trademark) disk, a memory stick, an integrated circuit card, and the like may be included.
The computer readable instructions may include any of assembly instructions, instruction Set Architecture (ISA) instructions, machine dependent instructions, microcode, firmware instructions, state setting data, or source code or object code described in any combination of one or more programming languages including conventional procedural programming languages, such as Smalltalk (registered trademark), JAVA (registered trademark), c++, and the like, and "C" programming language or the same programming language.
The computer readable instructions are: the computer readable instructions for execution by a processor or programmable circuit of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such as the Local Area Network (LAN), the internet, or other Wide Area Network (WAN), may be provided to the processor or programmable circuit of the general purpose computer, special purpose computer, or other programmable data processing apparatus. Examples of a processor include a computer processor, a processing unit, a microprocessor, a digital signal processor, a controller, a microcontroller, and the like.
The present invention has been described above using the embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various alterations and modifications can be made to the above-described embodiments. It is apparent from the description of the claims that such modifications and improvements are also included in the technical scope of the present invention.
It should be noted that the order of execution of the respective processes such as the operations, the sequences, the steps, and the stages in the apparatus, the system, the program, and the method shown in the claims, the specification, and the drawings may be implemented in any order as long as the processes are not specifically indicated as "before", and the like, and the outputs of the processes before are not used in the processes after. The operation flows in the claims, specification, and drawings do not necessarily have to be performed in this order, even though "first", "next", etc. are described for convenience.
Reference numerals illustrate:
10 systems; 11 a storage unit; 12 registration part; 13 generation part; a 14 virtual waveform confirmation processing unit; 15 a learning unit; 16 an evaluation execution unit; 102 a waveform storage unit; 104, performing virtual augmentation treatment; 106 a learning data storage unit; 108 user input; 110 algorithm storage; 112 selecting a process; 114 parameter setting processing; 116 generating processing; 118 a data storage section; 120 user intent reflection processing; 122 a distributed display process; 124 selecting a process; 126 separate display processing; 128 range selection processing; 130 a learning data storage unit; 140 a learning execution unit; 142AI;144 algorithm storage; a 146 waveform evaluation model storage unit; 150 applications; a 160 waveform acquisition unit; 162 intention determining section; 164 a pretreatment section; 166 a virtual waveform generation unit; 168 virtual waveform storage; 170 a similarity display control section; 172 range designation receiving unit; 174 range waveform selecting section; 180 a model acquisition unit; 182 input waveform acquisition unit; a 184 waveform input unit; 186 an evaluation result output control unit; 190 managing the data; 202 abnormal waveform data; 204 virtual abnormal waveform data; 206 virtual abnormal waveform data; 212 histogram; a 214 waveform; 220; 300 waveform data; 301 data size; 302 range; 303 start point; 304 end point; interval 305; 308 virtual waveform data; 311 range; a 312 waveform; 313 waveforms; a 314 waveform; 321 via points; 322 spline curves; 400 display data; 412 graph; 414 histogram; 422 graph; 424 histogram; 432 graph; 434 histogram; 1200 computers; 1210 a master controller; 1212CPU;1214RAM;1216 a graphics controller; 1218 display device; 1220 input/output controller; 1222 a communication interface; 1224 storage means; 1230ROM;1240 input/output chips.

Claims (14)

1. A generation system, comprising:
A waveform acquisition unit that acquires waveform data;
An intention determination unit that determines an intention of a user; and
A virtual waveform generation unit that generates virtual waveform data from the waveform data acquired by the waveform acquisition unit by a method that reflects the intention of the user determined by the intention determination unit.
2. The generation system of claim 1, further comprising:
And a similarity display control unit that performs control to display data representing a similarity between the virtual waveform data generated from the waveform data by the virtual waveform generation unit and the waveform data that is a generation source of the virtual waveform data, to a user.
3. The generating system of claim 2, wherein,
The virtual waveform generation unit generates a plurality of virtual waveform data from the waveform data by using at least any one of a plurality of algorithms, a plurality of parameter settings, and a random number,
The similarity display control section performs control of displaying the display data representing a similarity between the plurality of virtual waveform data generated from the waveform data by the virtual waveform generating section and the waveform data as a generation source of the plurality of virtual waveform data to the user.
4. The generating system of claim 3, wherein,
The intention determining unit determines a plurality of combinations of algorithm and parameter settings according to the instruction of the user,
The virtual waveform generation unit generates a virtual waveform data set including the plurality of virtual waveform data from the waveform data using the algorithm and the parameter setting in each of the plurality of combinations determined by the intention determination unit,
The similarity display control section performs control of displaying, to the user, the display data representing a similarity between each of the plurality of virtual waveform data sets generated by the virtual waveform generation section and the waveform data as a generation source.
5. The generation system of claim 4, further comprising:
A virtual waveform storage unit that stores a plurality of virtual waveform data sets generated by the virtual waveform generation unit in association with each of the plurality of virtual waveform data sets, the recipe information being capable of identifying a combination of the algorithm used in the generation of the virtual waveform data sets and the parameter setting.
6. The generation system of claim 4, further comprising:
A virtual waveform storage unit that stores a plurality of virtual waveform data sets generated by the virtual waveform generation unit in association with each of the plurality of virtual waveform data sets, the source waveform identification information being capable of identifying the waveform data that is a generation source of the virtual waveform data sets.
7. The generating system of claim 3 or 4, further comprising:
a range specification receiving unit configured to receive specification of a range of the similarity of the display data displayed by the similarity display control unit by the user;
An in-range waveform selection unit configured to select a plurality of pieces of virtual waveform data corresponding to the range of the similarity that is accepted by the range designation accepting unit; and
And a virtual waveform storage unit configured to store a virtual waveform data set including the plurality of virtual waveform data selected by the in-range waveform selection unit.
8. The generating system as claimed in any one of claims 2 to 6, wherein,
The similarity display control section performs control of displaying, to the user, the display data representing a similarity between the virtual waveform data and the waveform data that is a generation source of the virtual waveform data and other display data representing a similarity between the virtual waveform data and other waveform data different from the waveform data.
9. The generation system of any of claims 1 to 4, further comprising:
A preprocessing unit configured to perform at least one of a plurality of preprocessing on the waveform data; and
And a virtual waveform storage unit configured to store a virtual waveform data set including a plurality of pieces of virtual waveform data generated by the virtual waveform generation unit from the waveform data subjected to the preprocessing by the preprocessing unit, in association with preprocessing identification information capable of identifying the preprocessing.
10. The generation system of any of claims 1 to 6, further comprising:
a preprocessing unit configured to perform at least one of a plurality of preprocessing on the waveform data;
A learning execution unit configured to generate a waveform evaluation model for outputting an evaluation result of the input waveform data generated by the virtual waveform generation unit from the waveform data subjected to the preprocessing by the preprocessing unit by executing machine learning using the virtual waveform data; and
And a waveform evaluation model storage unit that stores the waveform evaluation model in association with the waveform evaluation model generated by the learning execution unit, the preprocessing identification information being able to identify the preprocessing.
11. The generation system of claim 10, further comprising:
A model acquisition unit that acquires the waveform evaluation model associated with the preprocessing identification information stored in the waveform evaluation model storage unit;
A waveform input unit for performing preprocessing indicated by the preprocessing identification information on input waveform data and inputting the preprocessing data into the waveform evaluation model; and
And an evaluation result output control unit configured to control the output of the evaluation result of the input waveform data output from the waveform evaluation model.
12. The generation system of any of claims 1 to 6, further comprising:
A virtual waveform storage unit that stores virtual waveform data in association with the virtual waveform data generated by the virtual waveform generation unit, the virtual waveform data representing that the virtual waveform data is data generated from the waveform data; and
And a learning execution unit configured to generate a waveform evaluation model for outputting an evaluation result of the input waveform data by performing machine learning using the virtual waveform data generated by the virtual waveform generation unit, and perform evaluation of the waveform evaluation model without using the virtual waveform data.
13. A computer-readable storage medium storing a program for causing a computer to function as the generating system according to any one of claims 1 to 12.
14. A method for generating a waveform evaluation model includes:
A waveform data acquisition step of acquiring waveform data,
An intention determining step of determining an intention of a user;
A virtual waveform generation step of generating virtual waveform data from the waveform data acquired in the waveform data acquisition step by a method reflecting the intention of the user determined in the intention determination step; and
A learning execution step of generating a waveform evaluation model that outputs an evaluation result of the input waveform data by executing machine learning using the virtual waveform data, the waveform evaluation model being generated in the virtual waveform generation step.
CN202311415222.5A 2022-11-01 2023-10-30 Generating system, computer-readable storage medium, and method for generating waveform evaluation model Pending CN117990966A (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US202263421539P 2022-11-01 2022-11-01
US63/421,539 2022-11-01
JP2023-032308 2023-03-02

Publications (1)

Publication Number Publication Date
CN117990966A true CN117990966A (en) 2024-05-07

Family

ID=90899716

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311415222.5A Pending CN117990966A (en) 2022-11-01 2023-10-30 Generating system, computer-readable storage medium, and method for generating waveform evaluation model

Country Status (1)

Country Link
CN (1) CN117990966A (en)

Similar Documents

Publication Publication Date Title
JP6555061B2 (en) Clustering program, clustering method, and information processing apparatus
CN110352389B (en) Information processing apparatus and information processing method
US7657408B2 (en) Structural analysis apparatus, structural analysis method, and structural analysis program
CN105956628B (en) Data classification method and device for data classification
CN113092981B (en) Wafer data detection method and system, storage medium and test parameter adjustment method
JP2015184942A (en) Failure cause classification device
JP6737277B2 (en) Manufacturing process analysis device, manufacturing process analysis method, and manufacturing process analysis program
EP3862829A1 (en) State estimation device, system, and manufacturing method
KR102470763B1 (en) Data outlier detection apparatus and method
CN113298162A (en) Bridge health monitoring method and system based on K-means algorithm
US10877989B2 (en) Data conversion system and method of converting data
CN115033453A (en) Abnormality detection method, apparatus, device, storage medium, and program
CN115902579A (en) Method, apparatus, computer device and readable storage medium for chip classification
CN113723467A (en) Sample collection method, device and equipment for defect detection
JP7463055B2 (en) Abnormality diagnosis device, abnormality diagnosis method, abnormality diagnosis program, and recording medium
CN117990966A (en) Generating system, computer-readable storage medium, and method for generating waveform evaluation model
US20240144101A1 (en) Generation system, computer-readable storage medium, and method for generating waveform evaluation model
KR101696105B1 (en) Apparatus and Method for analyzing defect reason
JP2024066391A (en) GENERATION SYSTEM, PROGRAM, AND METHOD FOR GENERATION OF WAVEFORM EVALUATION MODEL
US10891332B2 (en) Instrumentation diagram data generation device, instrumentation diagram search system, and computer readable medium
JP7170937B2 (en) Data extraction device, data extraction method and data extraction program
CN111932142A (en) Method, device, equipment and storage medium for scheme grouping and data grouping
JP6529688B2 (en) Selection apparatus, selection method, and selection program
EP3712777A1 (en) Data classification device
US20180137270A1 (en) Method and apparatus for non-intrusive program tracing for embedded computing systems

Legal Events

Date Code Title Description
PB01 Publication