Method for extracting peaks through ai supervised learning
The AI supervised learning method addresses the challenge of detecting peak values across various devices by learning waveform relationships, enhancing predictive maintenance versatility and reducing downtime costs by automating peak detection and threshold analysis.
Patent Information
- Application Number
- PCT/KR2024/007433
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-06-02
- Filing Date
- 2024-05-31
- Publication Date
- 2024-12-05
AI Technical Summary
Existing predictive maintenance methods for devices require individually building programs to detect peak values from various types of waveform information, limiting their versatility and applicability to facilities with hundreds or thousands of devices performing diverse operations, leading to reduced effectiveness and increased downtime costs.
A peak extraction method through AI supervised learning that collects and analyzes waveform information from various devices, learning the relationship between peak values and waveform data, allowing for real-time peak detection and threshold analysis without the need for individual program development.
Enables accurate and efficient peak value detection and predictive maintenance across diverse devices and operations, enhancing maintenance versatility and reducing downtime-related losses by automating peak value acquisition and threshold analysis.
Smart Images

Figure KR2024007433_05122024_PF_FP_ABST
Abstract
Description
Peak extraction method using AI supervised learning
[0001] The present invention relates to a peak extraction method through AI supervised learning, and more particularly, to a peak extraction method through AI supervised learning, which collects waveform information in various forms from various types of devices and devices performing various operations, detects an accurate peak value from the collected waveform information, and then provides a large amount of the detected peak value and waveform information to AI so that the relationship between the waveform information and the peak value can be learned, and when real-time waveform information measured and collected in real time from the device is provided to AI, the peak value for the real-time waveform information can be easily obtained from the learned AI, thereby solving the practical problem of having to individually construct a program for detecting the peak value according to the form of the waveform information in order to obtain a peak value from waveform information collected in various forms in the past, thereby ensuring excellent usability for a predictive maintenance method of the device.
[0002] Generally, stable operation is very important for equipment (motors, pumps, conveyors, compressors, etc.) used for automation processes in facilities.
[0003] For example, in the facilities of a large-scale transport factory, hundreds or thousands of devices are installed and operate in conjunction with each other to continuously transport the materials to be transported. If any one of the devices among the many devices breaks down, a catastrophic situation can occur where the entire operation of the facility comes to a halt.
[0004] At this time, downtime due to equipment failure inevitably leads to enormous losses not only in equipment repair costs but also in operating costs and business effects wasted while the equipment is down.
[0005] According to recent data from the Ministry of Employment and Labor and the Korea Occupational Safety and Health Agency, the total number of casualties from industrial accidents annually is estimated at around 100,000, which, when converted to cost, results in an annual loss of 18 trillion won.
[0006] To avoid such unexpected downtime costs, the present applicant has developed and proposed a predictive maintenance method for many types of devices (drive units) that performs predictive maintenance of the devices based on waveform information measured and collected from the devices.
[0007] Among the predictive maintenance methods for the above devices, there is a technology that performs predictive maintenance of the device based on the peak value (the value in which the energy size is the maximum in the waveform information) extracted from the waveform information.
[0008] However, in order to accurately obtain peak values from waveform information collected in various forms from devices due to the various types of devices or the various operations performed by the devices, there was a practical difficulty in that a system (algorithm) capable of detecting peak values for each waveform type had to be individually built as a program.
[0009] Therefore, the predictive maintenance method of the device developed and proposed by the applicant was difficult to apply to facilities consisting of hundreds or thousands of devices of various types and performing various operations, and thus there was a problem that the versatility of the predictive maintenance method was somewhat reduced.
[0010] The present invention has been proposed to solve the above-mentioned problems, and its purpose is to collect waveform information in various forms from various types of devices and devices performing various operations, detect an accurate peak value from the collected waveform information, and then provide a large amount of the detected peak value and waveform information to AI so that the relationship between the waveform information and the peak value can be learned, and when real-time waveform information measured and collected in real time from the device is provided to AI, the peak value for the real-time waveform information can be easily obtained from the learned AI, thereby resolving the practical problem of having to individually build a program for detecting the peak value according to the form of the waveform information in order to obtain a peak value from waveform information collected in various forms in the past, thereby ensuring excellent usability for a predictive maintenance method of the device, thereby providing a peak extraction method through AI supervised learning.
[0011] In addition, the present invention provides a peak extraction method through AI supervised learning, which further learns the relationship between waveform information and the threshold value set for the waveform information and the time value detected from the waveform information based on the threshold value, and provides real-time waveform information and threshold value measured and collected in real time from the device to the AI side, and detects and provides the time value in which the energy value remains above the threshold value from the real-time waveform information based on the learned AI.
[0012] In order to achieve the above object, the present invention provides a peak extraction method through AI supervised learning, which comprises: a waveform collection step (S10) for collecting waveform information such as a graph showing information on changes in energy size over time measured in an operating state of a device; a detection step (S20) for detecting and collecting, as a peak value, a value where the energy size is formed to a maximum value in the waveform information collected in the waveform collection step (S10); a supervised learning step (S30) for providing the waveform information collected in the waveform collection step (S10) and the peak value detected from the waveform information in the detection step (S20) to the AI side to induce the AI side to learn the relationship between the waveform information and the peak value; and an acquisition step (S40) for collecting real-time waveform information showing information on changes in energy size over time measured in a device in a real-time operating state and providing the real-time waveform information to the AI side to obtain a peak value for the real-time waveform information from the AI side.
[0013] In addition, the energy measured and collected from the device is characterized by being used by selecting one of the following: current consumed in operating the device, vibration generated when operating the device, noise generated when operating the device, frequency of the power supply to the device, temperature, humidity, and pressure of the device when operating the device.
[0014] In addition, the detection step (S20) sets a threshold for the waveform information collected in the waveform collection step (S10), and measures and detects the time during which the energy value of the waveform information is maintained beyond the threshold.
[0015] The above-mentioned map learning step (S30) provides the AI side with the waveform information collected in the waveform collection step (S10) and the threshold value for the waveform information set in the detection step (S20) and the time value detected when the energy value of the waveform information is maintained beyond the threshold value, thereby inducing the AI side to learn the relationship between the waveform information, the threshold value set in the waveform information, and the time value detected from the waveform information.
[0016] The above acquisition step (S40) is characterized by collecting real-time waveform information representing information on changes in energy size over time measured in a real-time operating state, and providing a threshold value for the real-time waveform information along with the real-time waveform information to the AI side, thereby obtaining a time value for the real-time waveform information from the AI side.
[0017] As described above, according to the peak extraction method through AI supervised learning according to the present invention, waveform information in various forms is collected from various types of devices and devices performing various operations, and an accurate peak value is detected from the collected waveform information, and then the detected peak value and waveform information are provided in large quantities to the AI side so that the relationship between the waveform information and the peak value is learned, so that when real-time waveform information measured and collected in real time from the device is provided to the AI side, the peak value for the real-time waveform information is easily obtained from the learned AI, thereby solving the practical problem of having to individually build a program for detecting the peak value according to the form of the waveform information in order to obtain the peak value from the waveform information collected in various forms in the past, thereby ensuring excellent usability for the predictive maintenance method of the device.
[0018] In addition, the relationship between the waveform information and the threshold value set for the waveform information and the time value detected from the waveform information based on the threshold value is further learned through AI, so that when real-time waveform information and threshold value measured and collected in real time from the device are provided to the AI side, the time value at which the energy value exceeds the threshold value is detected and provided from the real-time waveform information and threshold value information learned from the AI.
[0019] Figure 1 is a block diagram of a peak extraction method using AI supervised learning according to an embodiment of the present invention.
[0020] Figures 2 to 8 are diagrams for explaining the peak extraction method through AI supervised learning shown in Figure 1.
[0021] A peak extraction method using AI supervised learning according to a preferred embodiment of the present invention is described in detail with reference to the attached drawings. Detailed descriptions of known functions and configurations that may unnecessarily obscure the gist of the present invention are omitted.
[0022] FIGS. 1 to 8 illustrate a peak extraction method through AI supervised learning according to an embodiment of the present invention. FIG. 1 is a block diagram of a peak extraction method through AI supervised learning according to an embodiment of the present invention, and FIGS. 2 to 8 are drawings for explaining the peak extraction method through AI supervised learning illustrated in FIG. 1, respectively.
[0023] As illustrated in Fig. 1, the peak extraction method (100) through AI supervised learning according to an embodiment of the present invention includes a waveform collection step (S10), a detection step (S20), a supervised learning step (S30), and an acquisition step (S40).
[0024] The above waveform collection step (S10) is a step for collecting waveform information such as a graph showing information on changes in energy size over time measured in the operating state of the device.
[0025] Here, in the waveform collection step (S10), the energy measured and collected from the device may be selected and used from among the current consumed in operating the device, vibration generated when operating the device, noise generated when operating the device, frequency of the power supply to the device, temperature, humidity, and pressure of the device when operating the device. In the peak extraction method (100) through AI supervised learning of the present invention, the current consumed in operating the device is selected and used as the energy measured from the device, but the selection is not limited to this current.
[0026] That is, when information on changes in the size (value) of current consumed by a device performing an operation in the waveform collection step (S10) is collected over time, a current (energy) waveform such as a graph is collected as waveform information. If waveform information is collected from various types of devices or from devices performing various operations, it is natural that various forms of waveform information can be collected from the devices.
[0027] For example, if the device is limited to a motor, and the motor performs various purposes and operations, for example, if the motor is used in a drilling device or an elevator, if information on changes in the amount of current consumed to drive the motor is collected over time, various types of waveform information, such as that shown in Fig. 2, can be collected.
[0028] The above detection step (S20) is a step of detecting and collecting the value where the energy size is formed as the maximum (peak) in the waveform information collected in the waveform collection step (S10) as a peak value.
[0029] That is, as illustrated in FIG. 3, in the detection step (S20), peak values for various types of waveform information collected in the waveform collection step (S10) are detected and collected, and at this time, the peak values for the collected waveform information are used as learning materials (data) for AI to learn in the supervised learning step (S30) described later. It is desirable to detect and collect peak values for a large amount of waveform information for efficient learning of the AI.
[0030] Here, a system (algorithm) for detecting the exact peak value of the waveform information collected in the waveform collection step (S10) is implemented as a program so that the detection step (S20) detects and collects the exact peak value from the waveform information.
[0031] Therefore, the learning objective of the AI that learns based on the peak value information of the waveform information detected and collected in the above detection step (S20) is accurately set.
[0032] If the peak value, which is the maximum current value, is not accurately detected and provided from the waveform information in the detection step (S20), it is difficult to perform the AI learning purpose of obtaining the peak value from the waveform information through the supervised learning step (S30) preferably, and this may lower the reliability of the peak value obtained from the real-time waveform information through the AI side learned in the acquisition step (S40) to be described later, so the detection step (S20) must detect and provide an accurate peak value from the waveform information.
[0033] In summary, the detection step (S20) must accurately detect the peak value for each waveform information collected in the waveform collection step (S10) and provide the peak value for the detected waveform information as learning material for AI learning.
[0034] The above map learning step (S30) is a step that provides the waveform information collected in the waveform collection step (S10) and the peak value detected from the waveform information in the detection step (S20) to the AI side, thereby inducing the AI side to learn the relationship between the waveform information and the peak value.
[0035] That is, as illustrated in FIG. 4, the AI clearly learns the relationship between waveform information and peak values through a large amount of learning data on various types of waveform information and peak values of the waveform information. At this time, the AI performs learning in a supervised learning manner through deep learning, but of course, it is not limited to performing learning in this manner.
[0036] Here, the AI can effectively identify and learn the relationship between waveform information and peak values as the provided learning materials (waveform information and peak values) are abundant. Therefore, it is desirable to collect, detect, and provide peak values together with a large amount of waveform information in the waveform collection step (S10) and the detection step (S20).
[0037] The above acquisition step (S40) is a step of collecting real-time waveform information representing information on changes in energy size over time measured from a device in a real-time operating state, and providing the real-time waveform information to the AI side to acquire a peak value for the real-time waveform information from the AI side.
[0038] That is, as illustrated in FIG. 5, the AI, which has learned the relationship between waveform information and peak values through the above-described waveform collection step (S10), detection step (S20), and supervised learning step (S30), automatically detects and provides the peak value for the real-time waveform information when real-time waveform information measured from a device in a real-time operating state is provided, thereby enabling the AI to very easily obtain the peak value for the real-time waveform information.
[0039] In summary, AI trained through the above process can very easily detect and provide peak values for each waveform information even when various types of waveform information are collected and provided from various types of devices and devices performing various operations.
[0040] For example, when real-time waveform information collected from the real-time operating status of a motor used in a drilling device or real-time waveform information collected from the real-time operating status of a motor used in an elevator is provided to the AI, the AI detects and provides peak values for each of the real-time waveform information of the motor, thereby enabling the peak values for various real-time waveform information to be obtained very easily.
[0041] Therefore, by training AI through the above process and enabling the trained AI to easily detect and acquire peak values from various types of waveform information, the practical difficulty of having to individually construct a system (algorithm) for detecting peak values according to the form of waveform information in order to acquire peak values from waveform information collected in various forms due to the type of device or various operations performed in the past can be resolved.
[0042] That is, the applicant has developed and proposed various types of predictive maintenance methods that measure and collect peak values from waveform information (current waveform) collected from devices and perform predictive maintenance of the devices based on the collected peak values. These predictive maintenance methods can be very effectively applied to facilities consisting of hundreds or thousands of devices through AI learned as described above.
[0043] 『"Precise predictive maintenance method for a driving part (Registration No. 10-2103143)", "Precise predictive maintenance method for a driving part (Registration No. 10-2103146)", "Precise predictive maintenance method for a driving part (Registration No. 10-2103147)", "Precise predictive maintenance method for a device through defining a constant speed for a peak (Registration No. 10-2477707)", "Precise predictive maintenance method for a device through defining a constant speed for a peak (Registration No. 10-2477708)"』
[0044] Meanwhile, the detection step (S20) sets a threshold value for the waveform information collected in the waveform collection step (S10), and measures and detects the time during which the energy value of the waveform information is maintained beyond the threshold value.
[0045] Here, the threshold value refers to a value that serves as a standard for detecting abnormal signs of the device in the predictive maintenance method of the device (driving unit) developed and proposed by the applicant. If the energy value (size) of the waveform information collected from the device exceeds the set threshold value and is maintained for a critical period of time, an abnormal sign of the device is suspected.
[0046] These thresholds may be set to various values depending on the type of device or the operation the device performs, and the waveform information collected in various forms.
[0047] That is, as illustrated in FIG. 6, in the detection step (S20), if the current value in the waveform information collected in the waveform collection step (S10) exceeds a set threshold value, the time for which the current value is maintained exceeding the threshold value is detected and collected, and the waveform information detected and collected in this manner and the threshold value and the time for which the current value of the waveform information is maintained exceeding the threshold value are used as learning materials (data) for AI to learn in the supervised learning step (S30). For efficient learning of the AI, it is preferable to detect and collect the threshold value for a large amount of waveform information and the time information for which the threshold value is maintained exceeding the threshold value.
[0048] Of course, the large amount of waveform information collected in the waveform collection step (S10) can be collected in various forms, such as waveform information exceeding a set threshold value and waveform information not exceeding the threshold value, and in the detection step (S20), if the current value of the waveform information exceeds the threshold value, the time value will be detected and collected, and if the current value of the waveform information does not exceed the threshold value, the time value will not be detected. Of course, such various pieces of information (learning materials) are provided to the AI side through the supervised learning step (S30) so that various cases can be learned.
[0049] Here, in the detection step (S20), if the current value of the waveform information does not exceed the set threshold value and thus the time value is not detected, the time value is provided as “0”, and when the waveform information and the threshold value are provided to the AI side learned in the acquisition step (S40), learning can be induced in a manner in which “0” is presented as the time value when the current value of the waveform information does not exceed the threshold value.
[0050] As described above, a threshold value for waveform information is set, and when the current value of the waveform information exceeds the set threshold value, a system (algorithm) for accurately detecting the time for which the exceeded current value is maintained is implemented as a program so that the detection step (S20) can accurately detect the time value for which the current value in the waveform information exceeds the threshold value.
[0051] Therefore, the learning objective of the AI is accurately set based on the time value information in which the current value of the waveform information is maintained beyond the threshold value set in the above detection step (S20).
[0052] If the time at which the current value of the waveform information is maintained beyond the threshold value set in the detection step (S20) is not accurately detected and provided, it is difficult to perform the AI learning purpose of accurately obtaining the time at which the current value is maintained beyond the threshold value set in the waveform information through the supervised learning step (S30) desirably, and this may lower the reliability of the time value obtained based on the real-time waveform information and the threshold value through the AI side learned in the acquisition step (S40), so the detection step (S20) must detect and provide an accurate time value at which the current value of the waveform information is maintained beyond the threshold value set.
[0053] In summary, the detection step (S20) must accurately detect the time during which the current value of each waveform information collected in the waveform collection step (S10) exceeds a set threshold value and is maintained, and provide the time value for the detected waveform information as learning material for AI learning.
[0054] Then, as illustrated in FIG. 7, the map learning step (S30) provides the AI side with the waveform information collected in the waveform collection step (S10) and the threshold value for the waveform information set in the detection step (S20) and the time value detected when the energy value is maintained exceeding the threshold value, thereby inducing the AI side to learn the relationship between the waveform information, the threshold value set in the waveform information, and the time value detected from the waveform information.
[0055] That is, the AI clearly learns the relationship between various types of waveform information, the threshold value set for the waveform information, and the time value detected from the waveform information, and at this time, the AI performs learning in a supervised learning manner through deep learning.
[0056] Here, the AI can effectively identify and learn the relationship between the waveform information, the threshold value set for the waveform information, and the time value detected from the waveform information as the learning materials (waveform information, threshold value, time value) provided are abundant. Therefore, in the waveform collection step (S10) and the detection step (S20), it is desirable to collect, detect, and provide the time value detected from the waveform information together with a large amount of waveform information and threshold value.
[0057] After that, the acquisition step (S40) collects real-time waveform information representing information on changes in energy size over time measured in a real-time operating state, and provides the AI side with the real-time waveform information and a threshold value for the real-time waveform information to obtain a time value for the real-time waveform information from the AI side.
[0058] That is, as shown in Fig. 8, the AI that has learned the relationship between the waveform information, the threshold value set for the waveform information, and the time value detected from the waveform information through the learning process as described above automatically detects and provides the time value at which the current value of the real-time waveform information exceeds the set threshold value when the real-time waveform information measured from the device in a real-time operating state and the threshold value for the waveform information are set and provided, thereby enabling the time value for the real-time waveform information to be obtained very easily through the AI.
[0059] Of course, if the current value of the real-time waveform information does not exceed the set threshold, the time value for the real-time waveform information will be acquired as “0” through AI.
[0060] In summary, the AI learned through the above process can very easily detect and provide a time value (the time for which the current value exceeds the threshold value) from each waveform information based on the relationship between the waveform information and the threshold value, even if waveform information of various forms is collected from various types of devices and devices performing various operations and the threshold value for the collected waveform information is set to various sizes (values).
[0061] For example, when real-time waveform information collected from the real-time operating status of a motor used in a drilling device and a threshold value set for the real-time waveform information, or real-time waveform information collected from the real-time operating status of a motor used in an elevator and a threshold value for the real-time waveform information, are provided to the AI, the AI detects and provides time values for each real-time waveform information based on the real-time waveform information of each motor and the set threshold value, thereby enabling the AI to very easily obtain various real-time waveform information and time values for threshold values.
[0062] Therefore, it is possible to resolve the practical difficulty of having to individually construct a system (algorithm) for detecting time values according to the form of the waveform information and the size (value) of the threshold value in order to obtain time values from waveform information based on waveform information collected in various forms due to the type of device or various operations performed and threshold values set in various ways according to the waveform information.
[0063] The peak extraction method (100) through AI supervised learning of the present invention, which is composed of the steps described above, collects waveform information in various forms from various types of devices and devices performing various operations, detects an accurate peak value from the collected waveform information, and then provides a large amount of the detected peak value and waveform information to the AI side to learn about the relationship between the waveform information and the peak value, so that when real-time waveform information measured and collected in real time from the device is provided to the AI side, the peak value for the real-time waveform information is easily obtained from the learned AI, thereby solving the practical problem of having to individually build a program for detecting the peak value according to the form of the waveform information in order to obtain a peak value from waveform information collected in various forms in the past, thereby ensuring excellent usability for the predictive maintenance method of the device.
[0064] In addition, the relationship between the waveform information and the threshold value set for the waveform information and the time value detected from the waveform information based on the threshold value is further learned through AI, so that when real-time waveform information and threshold value measured and collected in real time from the device are provided to the AI side, the time value at which the energy value exceeds the threshold value is detected and provided from the real-time waveform information and threshold value information learned from the AI.
[0065] While the present invention has been described with reference to the embodiments illustrated in the accompanying drawings, these are merely illustrative and are not limited to the above-described embodiments. Those skilled in the art will readily appreciate that various modifications and equivalent embodiments are possible. Furthermore, it should be understood that modifications are possible by those skilled in the art without departing from the spirit and scope of the present invention. Therefore, the scope of the claims in the present invention is not limited to the detailed description, but rather by the claims set forth below and their technical spirit.
[0066] The present invention can be used in the predictive maintenance industry of equipment.
Claims
1. A waveform collection step for collecting waveform information such as a graph showing information on changes in energy size over time measured in the operating state of the device; A detection step for detecting and collecting the value in which the energy size is formed as the maximum (peak) in the waveform information collected in the above waveform collection step as a peak value; A supervised learning step that provides the waveform information collected in the waveform collection step and the peak value detected from the waveform information in the detection step to the AI (Artificial Intelligence) side and induces the AI side to learn the relationship between the waveform information and the peak value; and A peak extraction method through AI supervised learning, characterized by comprising an acquisition step of collecting real-time waveform information representing information on changes in energy size over time measured from a device in a real-time operating state, and providing the real-time waveform information to the AI side to obtain a peak value for the real-time waveform information from the AI side.
2. In paragraph 1, The energy measured and collected from the device is: A peak extraction method through AI supervised learning characterized in that any one of the following is selected and used: current consumed in operating the device, vibration generated during device operation, noise generated during device operation, frequency of the device power supply, temperature, humidity, and pressure of the device during device operation.
3. In paragraph 1 or 2, The above detection step sets a threshold for the waveform information collected in the waveform collection step, and measures and detects the time during which the energy value of the waveform information is maintained beyond the threshold. The above-mentioned supervised learning step provides the AI side with the waveform information collected in the waveform collection step, the threshold value for the waveform information set in the detection step, and the time value detected when the energy value of the waveform information is maintained beyond the threshold value, thereby inducing the AI side to learn the relationship between the waveform information, the threshold value set for the waveform information, and the time value detected from the waveform information. The above acquisition step is a peak extraction method through AI supervised learning, characterized in that it collects real-time waveform information representing information on changes in energy size over time measured in a real-time operating state, and provides a threshold value for the real-time waveform information together with the real-time waveform information to the AI side, thereby obtaining a time value for the real-time waveform information from the AI side.
Citation Information
Patent Citations
System and Methods for Matching Cloth Seller via On-line
KR1020220053396A
Server, method and computer program for recognizing multiple objects included in image data
KR1020220138682A
Apparatus for cleaning sensor
KR1020240046969A
A method of detecting a text for detecting a scene text and an apparatus for performing the same
KR1020240113259A
Process abnormality detection device and method using analysis of control section temperature signal
WO2022030831A1
