KR102720965B1 - System and method for degradation detection of polymer component based on deep learning - Google Patents

Abstract

A system and method for detecting degradation characteristics of polymer components based on deep learning are disclosed. The present invention relates to a system for detecting degradation characteristics of polymer components, comprising: a spectrometer which generates spectrum data for a sample extracted from the polymer component; and a user terminal which receives the spectrum data from the spectrometer, detects a degradation state of the sample based on the spectrum data, and guides a user about the degradation state and a countermeasure according to the degradation state, wherein the user terminal learns normal state data and abnormal state data input in advance related to the sample, and can detect the degradation state based on the learned result value and the measurement data.

Description

{SYSTEM AND METHOD FOR DEGRADATION DETECTION OF POLYMER COMPONENT BASED ON DEEP LEARNING}

The present invention relates to a system and method for detecting degradation characteristics of polymer components based on deep learning. More specifically, the present invention relates to a system and method for learning a spectrum of a sample of a polymer component and detecting degradation characteristics of the polymer component based on the spectrum.

Deterioration means that the chemical structure of a material is changed harmfully by heat or light, and especially the physical properties are permanently changed, resulting in a decrease in properties. The deterioration factors of polymer components include temperature, voltage, current, pressure, vibration, moisture, chemical action, and radiation, and in particular, thermal deterioration and radiation deterioration have the greatest impact.

Thermal degradation refers to a phenomenon that occurs when the rate of oxidation and decomposition reactions increases as temperature increases. Radiation degradation refers to a phenomenon in which organic polymer materials deteriorate by causing cross-linking or decomposition between molecules when exposed to radiation. It is known that this is caused by diffusion of internal oxygen atoms.

Stressors that promote deterioration include:

First, there is deterioration due to thermal stress. Temperature rise that promotes chemical reaction increases the rate of deterioration and is the most common deterioration factor that shortens the life of the material. The temperature rise limit of the device is also often determined from the perspective of thermal deterioration of the material.

Second, there is deterioration due to electric stress. This is caused by the electric field applied to the material of the device, and is caused by various causes such as conduction current, dielectric loss, electromagnetic force, electrostatic force, or partial discharge. The conduction current, in addition to exhibiting a thermal effect as Joule heat, causes an electrochemical effect in ion conduction, thereby deteriorating the material, and the dielectric loss occurs under an alternating electric field and causes thermal deterioration. In addition, electromagnetic force and electrostatic force are forces generated by high voltages such as short-circuit currents, and exhibit a mechanical effect, and when a partial discharge of gas or liquid occurs in a high field, thermal action, particle impact action, and chemical action by excited molecules or ions occur, causing deterioration.

Third, there is deterioration due to mechanical stress. This is mainly caused by mechanical stress, vibration, etc., and in addition to these external mechanical forces, it is caused by thermal distortion due to differences in thermal expansion coefficients, and electromagnetic stress due to short-circuit current.

Fourth, there is deterioration due to environmental stress. Physical and chemical deterioration is accelerated in high-energy radiation environments such as neutron rays, gamma rays, x-rays, electron rays, etc., from nuclear reactors, radioactive elements, or particle accelerators. In addition, hydrolysis due to reactive substances, moisture absorption, and erosion by microorganisms are also noted. Furthermore, deterioration is accelerated by strong ultraviolet irradiation in the natural environment.

However, in general, the above-mentioned factors often act in combination, compared to the deterioration in which each factor acts alone. In this case, the deterioration is approximately caused by overlapping deterioration when alone, and in cases where a large deterioration is promoted by simple overlapping when alone. In particular, in cases where a large deterioration is promoted, it is treated as a complex deterioration.

In addition, the artificial intelligence (AI) system is a computer system that implements human-level intelligence, and unlike existing rule-based smart systems, it is a system in which machines learn, judge, and become smarter on their own. The more an AI system is used, the more its recognition rate improves and it can understand user preferences more accurately, so existing rule-based smart systems are gradually being replaced by deep learning-based AI systems.

Artificial intelligence technology consists of machine learning (deep learning) and element technologies utilizing machine learning. Machine learning is an algorithm technology that classifies/learns the characteristics of input data on its own, and element technologies are technologies that utilize machine learning algorithms such as deep learning, and consist of technology fields such as linguistic understanding, visual understanding, inference/prediction, knowledge expression, and motion control.

In this way, with the development of artificial intelligence technology, the demand for detecting the deterioration status of polymer components through artificial intelligence technology is increasing.

Republic of Korea Patent Publication No. 10-2014-0085425 (Title of invention: Deterioration analysis method)

The purpose of the present invention is to detect pressure applied to the sole of a shoe, learn the pressure, and automatically detect the wear condition of the sole of the shoe.

In addition, an object of the present invention is to automatically classify measurement data when a shoe sole is detected to be in a worn state and to provide the user with a corresponding countermeasure.

In order to solve the above-described problem, the present invention provides a system for detecting a degradation characteristic of a polymer component, comprising: a spectrometer for generating a first spectrum image for a sample extracted from the polymer component; and a user terminal for learning the first spectrum image through a deep neural network, extracting feature points for the first spectrum image, and detecting a degradation state of the sample based on the feature points, wherein the user terminal learns a pre-labeled second spectrum image, and detects the degradation state based on the learned result value and the first spectrum image, wherein the second spectrum image may be generated based on a functional group known in advance to be included in the polymer component.

In addition, the user terminal may obtain a plurality of coordinate information for a plurality of peaks included in the first spectrum image and the second spectrum image through the deep neural network, perform a K-center clustering process on the plurality of coordinate information, ultimately generate two clusters based on the K-center clustering process, and derive a similarity based on the Euclidean distance between the two clusters.

Additionally, the user terminal may extract an average value of each of the two clusters, extract a final distance value between each average value, and detect the deterioration state based on the final distance value.

Additionally, the deep neural network may include an image filter using a CNN algorithm.

Additionally, the user terminal may include a camera module for capturing the first spectrum image, and the camera module may include a housing including a through hole in a side wall and a lens installed in the through hole.

In addition, in order to solve the above-described problem, the present invention may include a step of learning a second spectrum image that is pre-labeled for a functional group included in a polymer component in a normal state, a step of learning a first spectrum image for a target sample, a step of detecting a deterioration state of the target sample based on the learned result value and the first spectrum image, and a step of displaying the deterioration state.

In addition, in order to solve the above-described problem, the present invention may be a non-transitory computer-readable medium storing instructions, which, when executed by a processor, cause the processor to perform the above-described method.

The present invention has the effect of enabling quantitative analysis of deterioration of polymer components by detecting the deterioration status of polymer components through artificial intelligence.

In addition, the present invention has the effect of providing a basis for predicting failure or repair cycle according to the deterioration state of heterogeneous joint parts in which metal and plastic are combined, such as tires and automobile parts containing plastic parts.

In addition, the present invention has the effect of generating an accurate spectrum image by protecting a lens from external contaminants by using a user terminal that captures a spectrum image through a camera including a unique coating layer.

In addition, the present invention has the effect of enabling more accurate detection of the deterioration state by proposing a deterioration state detection algorithm utilizing a unique correction constant.

The effects obtainable according to the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by a person having ordinary skill in the art to which this specification pertains from the outer surface of the protective member below.

FIG. 1 is a diagram illustrating a system for detecting degradation characteristics of a polymer component based on deep learning according to the present invention.
Figure 2 schematically illustrates a user terminal according to the present invention.
Figure 3 illustrates the functional configuration of a memory according to the present invention.
Figure 4 illustrates a graph that applies a K-centered clustering process execution module according to the present invention.
Figure 5 is an example comparing Euclidean distance values for each mode according to the present invention.
Figure 6 is a diagram illustrating a diffraction grating spectrometer.
Figure 7 is a diagram illustrating a filter type spectrometer.
Figure 8 shows an example of a spectral spectrum according to the present invention.
Figure 9 illustrates a camera module according to the present invention.
FIG. 10 is a drawing showing an example of an image correction filter according to the present invention.
Figure 11 illustrates a method for detecting degradation characteristics of a polymer component based on deep learning according to the present invention.
The accompanying drawings, which are incorporated in and are intended to aid in the understanding of the present specification and are a part of the detailed description, illustrate embodiments of the present specification and, together with the detailed description, serve to explain the technical features of the present specification.

Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Regardless of the drawing symbols, identical or similar components will be given the same reference numerals and redundant descriptions thereof will be omitted.

In addition, when describing the embodiments disclosed in this specification, if it is determined that a detailed description of a related known technology may obscure the gist of the embodiments disclosed in this specification, the detailed description is omitted.

In addition, the attached drawings are only intended to facilitate easy understanding of the embodiments disclosed in this specification, and the technical ideas disclosed in this specification are not limited by the attached drawings, and should be understood to include all modifications, equivalents, or substitutes included in the spirit and technical scope of the present invention.

Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The terms are used only to distinguish one component from another.

Singular expressions include plural expressions unless the context clearly indicates otherwise.

In this application, it should be understood that the terms “comprises” or “has” and the like are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof that is external to the protection portion in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

Hereinafter, based on the above-described contents, a deep learning-based polymer component degradation characteristic detection system according to a preferred embodiment of the present specification will be described in detail as follows.

FIG. 1 is a diagram illustrating a system for detecting degradation characteristics of a polymer component based on deep learning according to the present invention.

According to FIG. 1, the system according to the present invention may include a user terminal (100) and a spectrometer (200). The user terminal (100) according to the present invention may refer to any computing device that a user may carry or use, such as a smart phone, a mobile phone, a PDA, a laptop, a computer, etc. In addition, the spectrometer (200) according to the present invention may be an infrared (IR) spectrometer or a Fourier transform spectrometer. In addition, the spectrometer (200) according to the present invention may refer to any device that can identify the components of a polymer through a spectrum of light.

The spectrometer (200) according to the present invention can generate a first spectrum image for a sample extracted from a polymer component. The first spectrum image according to the present invention can mean data or an image of a spectrum generated by emitting light (such as a laser) of a specific component toward a target sample.

The user terminal (100) according to the present invention can learn a first spectrum image through a deep neural network, extract feature points for the first spectrum image, and detect the deterioration state of a sample based on the feature points.

The deep neural network of the present invention refers to a system or network that constructs one or more layers in one or more computers to perform judgment based on multiple data. For example, the deep neural network can be implemented as a set of layers including a convolutional pooling layer, a locally connected layer, and a fully-connected layer. For example, the overall structure of the deep neural network can be formed as a convolutional neural network (i.e., CNN) structure in which a convolutional pooling layer is followed by a locally connected layer, and a fully-connected layer is followed by a locally connected layer. In addition, the deep neural network can be formed as a recurrent neural network (RNN) structure that is recursively connected as edges pointing to themselves are included in nodes of each layer, for example. The deep neural network can include various judgment criteria and can add new judgment criteria through analysis of input images. However, the structure of the deep neural network according to the embodiment of the present invention is not limited thereto, and may be formed as a neural network of various structures.

The deterioration state according to the present invention refers to an abnormal state of a polymer component, and specifically, may refer to a state in which performance is reduced due to heat generation, chemical reaction, etc. For example, when excessive friction occurs on a polymer component, or a temperature exceeding a threshold occurs, the polymer component may undergo a chemical reaction with other components in the air. In this way, a chemical reaction on the polymer component may cause a change in the components of the polymer component, and a state in which performance is reduced due to such a change in the components may be referred to as a deterioration state.

In addition, the user terminal (100) according to the present invention can learn a pre-labeled second spectrum image and detect the deterioration state based on the learned result value and the first spectrum image.

The pre-labeled second spectrum image according to the present invention may mean the following two spectrum images.

The second spectrum image may be big data collected in advance for the same type of polymer component as the polymer component according to the present invention. The pre-labeled second spectrum image may include an image including a normal state spectrum for a sample of the polymer component labeled as not deteriorated. In addition, the pre-labeled second spectrum image may include an image including an abnormal state spectrum for a sample of the polymer component labeled as deteriorated. In this case, the labeling may mean that it has been verified by an expert for the corresponding polymer component. That is, the user terminal (100) according to the present invention may include a deep neural network that has learned the pre-labeled big data, that is, the second spectrum image.

The first and second spectrum images according to the present invention are images including a graph, in which the unit of the x-axis may be wave numbers and the unit of the y-axis may be transmission of wavelengths, and details are described later in FIG. 8.

Figure 2 schematically illustrates a user terminal according to the present invention.

According to FIG. 2, the user terminal (100) according to the present invention may include a processor, a memory, a communication module, a display unit, and a camera module. The camera module according to the present invention is described below in FIG. 9.

The processor (110) can control other components by executing instructions stored in the memory (120). The processor (110) can perform instructions stored in the memory (120).

The processor (110) is a configuration that can perform calculations and control other devices. Mainly, it can mean a central processing unit (CPU), an application processor (AP), a graphics processing unit (GPU), etc. In addition, the CPU, AP, or GPU can include one or more cores therein, and the CPU, AP, or GPU can operate using an operating voltage and a clock signal. However, while the CPU or AP is composed of several cores optimized for serial processing, the GPU can be composed of thousands of smaller and more efficient cores designed for parallel processing.

The processor (110) can process signals, data, information, etc. input or output through the components discussed above, or can operate an application program stored in the memory (120) to provide or process appropriate information or functions to the user.

The memory (120) stores data that supports various functions of the user terminal (100). The memory (120) can store a plurality of application programs (or applications) that are run on the user terminal (100), data for the operation of the user terminal (100), and commands. At least some of these application programs can be downloaded from an external user terminal (100) via wireless communication. In addition, the application programs can be stored in the memory (120), installed on the user terminal (100), and driven by the processor (110) to perform the operation (or function) of the user terminal (100).

The memory (120) may include at least one type of storage medium among a flash memory type, a hard disk type, an SSD (Solid State Disk type), an SDD (Silicon Disk Drive type), a multimedia card micro type, a card type memory (for example, an SD or XD memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory (120) may also include a web storage that performs a storage function on the Internet.

The communication module (130) transmits and receives information to and from a base station or a camera including a communication function via an antenna. The communication module (130) may include a modulation unit, a demodulation unit, a signal processing unit, etc.

Wireless communication can refer to communication using communication facilities that communication companies have previously installed and a wireless communication network that uses the frequencies of the communication facilities. At this time, the communication module (130) can be used in various wireless communication systems such as CDMA (code division multiple access), FDMA (frequency division multiple access), TDMA (time division multiple access), OFDMA (orthogonal frequency division multiple access), SC-FDMA (single carrier frequency division multiple access), and moreover, the communication module (130) can also be used in 3GPP (3rd generation partnership project) LTE (long term evolution), etc. In addition, not only 5G communication that is currently being commercialized but also 6G that is scheduled to be commercialized in the future can be used. However, this specification can utilize an existing communication network without being restricted by such wireless communication methods.

Additionally, short range communication technologies such as Bluetooth, BLE (Bluetooth Low Energy), Beacon, RFID (Radio Frequency Identification), NFC (Near Field Communication), Infrared Data Association (IrDA), UWB (Ultra Wideband), and ZigBee can be used.

The display unit (140) according to the present invention may refer to a device that receives screen data from the processor (110) and displays it so that the user can confirm it through the senses. The display unit (140) may include a self-luminous display panel or a non-luminous display panel. As a self-luminous display panel, an OLED panel that does not require a backlight may be exemplified, and as a non-luminous display panel, an LCD panel that requires a backlight may be exemplified, but is not limited thereto.

The camera module (150) according to the present invention may be a configuration included in a user terminal (100) that can generate image data through an optical module. Specific details thereof will be described later.

FIG. 3 illustrates the functional configuration of a memory according to the present invention, and FIG. 4 illustrates a graph to which a K-centered clustering process execution module according to the present invention is applied.

According to FIG. 3, the functional configuration included in the memory (120) according to the present invention may include a CNN learning module (121), a K-center clustering process execution module (122), and a data storage module (123).

The data storage module (123) according to the present invention may be a module that stores all big data that is a learning target according to the present invention. Other functional components included in the memory (120) according to the present invention may receive big data from the data storage module (123). The data storage module (123) according to the present invention may be a module that stores big data labeled by a user.

The CNN learning module (121) according to the present invention may be a module that learns big data learned in a data storage module. Specifically, the CNN learning module (121) according to the present invention may learn a pre-labeled second spectrum image. At this time, the CNN learning module (121) may also learn the first spectrum image, and like the second spectrum image, feature points for the first spectrum image may also be extracted. The CNN learning module (121) according to the present invention may include a CNN algorithm utilizing an image filter, which will be described later in FIG. 9.

The K-centered clustering process execution module (122) according to the present invention may be a module that measures and classifies the similarity between the first spectrum image described above and the second spectrum image described above. Specifically, the K-centered clustering process execution module (122) may generate K clusters and classify data by judging the similarity based on the distance values of the generated clusters.

The K-center clustering process refers to clustering given data into K clusters. This K-center clustering process randomly selects a sample value from a given data sample and measures the distance from that sample value to other sample data. At this time, the closest center is selected from each data sample, and the distance from that sample to other data sample values is calculated again. This process is repeated to achieve clustering. In other words, multiple clusters are created. This clustering technique can be mainly implemented in Python. Clustering is used as one of the algorithms that can be used in signal analysis.

According to Fig. 4, as a result of performing the K-center clustering process, two clusters are ultimately created. As a result of performing the K-center clustering process according to the present invention, the measurement values can be classified into clustering 1 and clustering 2.

At this time, the data to be clustered may be multiple coordinate information for multiple peaks included in the spectrum image. That is, the user terminal (100) according to the present invention may perform a K-centered clustering process for multiple coordinate information for multiple peaks.

The user terminal (100) according to the present invention ultimately creates two clusters based on the K-centered clustering process, and can derive the similarity based on the Euclidean distance between the two clusters.

In addition, the user terminal (100) according to the present invention can extract the average value of each of the two clusters, extract the final distance value between each average value, and detect the deterioration state based on the final distance value. A specific method of detecting the deterioration state will be described later.

The K-center clustering process execution module (122) according to the present invention can calculate the first center value of clustering 1 (clustering for the first difference value) through the average calculation method, and can calculate the second center value of clustering 2 (clustering for the second difference value) through the average calculation method. The K-center clustering process execution module according to the present invention can calculate the final distance value between the first center value and the second center value, and if the size of the final distance value is greater than a predetermined value, it can be determined that the similarity of each cluster is lower than the predetermined value.

That is, the K-center clustering process execution module (122) according to the present invention calculates the final distance value between the first center value and the second center value, and if the size of the final distance value is greater than a predetermined value, it can be determined that a deterioration state has occurred in the polymer component that is the source of the corresponding sample. In addition, depending on the type of the final distance value that serves as the criterion for the occurrence of the deterioration state, the polymer component according to the present invention can be classified by mode according to the deterioration state.

The K-center clustering process execution module (122) according to the present invention can perform the K-center clustering process using feature 1 as the x-axis coordinate and feature 2 as the y-axis coordinate. The K-center clustering process execution module (122) according to the present invention can calculate the first center value of cluster 1 and the second center value of cluster 2, and calculate the Euclidean distance between the first center value and the second center value. The Euclidean distance can be defined by the following mathematical expression 1.

[Mathematical Formula 1]

(where d is a distance value, (x, y) is a coordinate value, x corresponds to the wave number, and y corresponds to the measured transmission, and the transmission is a value from 0 to 1).

The system according to the present invention can extract the final distance values of cluster 1 and cluster 2, and detect the deterioration state of the corresponding polymer component based on the final distance values.

At this time, the final distance value according to the present invention can be extracted in the following order.

(1) Extract the first mean (x1, y1) of cluster 1 and the second mean (x2, y2) of cluster 2.

(2) The first average value and the second average value are each coordinate values, and the final distance value (L) between the first average value and the second average value is extracted based on the following mathematical formula 2.

[Mathematical formula 2]

(However, (x1, y1) is a coordinate value generated based on the average value of cluster 1 (= first center value), and (x2, y2) is a coordinate value generated based on the average value of cluster 2 (= second center value))

The present invention can define that the deterioration state of a polymer component is detected when the extracted final distance value (L) is greater than a preset distance value.

In general, when a deterioration state of a polymer component occurs, it is more efficient to multiply the final distance value (L) by a correction factor and compare it with a preset distance value based on this. This is called a corrected final distance value (L') and can be extracted based on the following mathematical expressions 3 and 4.

[Mathematical Formula 3]

(However, (x1, y1) is a coordinate value generated based on the average value of cluster 1, and (x2, y2) is a coordinate value generated based on the average value of cluster 2)

[Mathematical Formula 4]

At this time, α is a correction constant for more accurate determination of the ideal state, and the unit of α can be ignored.

In this way, when the deterioration state is judged based on the corrected final distance value (L') using mathematical expressions 3 and 4, there is an effect in which the deterioration state of the polymer component can be derived more accurately.

[Experimental example]

The results of comparing the accuracy of a system for detecting the deterioration state of a polymer component based on a Euclidean distance generated using a correction constant according to the present invention and a system for detecting the deterioration state of a polymer component based on a Euclidean distance without using a correction constant are as shown in Table 1 below.

At this time, the accuracy is calculated based on a survey of 10 experts in the relevant field, and may be a value calculated by surveying and experimenting 10 experts in a total of 100 cases in the same environment. The accuracy may be a comparison of the deterioration state detected by the system according to the present invention and the actual deterioration state.

Correction factor α not used Using correction factor α Accuracy 83 98

(unit: %)

Table 1 shows the accuracy of cases according to the present invention. As a result of the experiment, the case where the correction factor α is used has significantly higher accuracy than the case where the correction factor α is not used.

Figure 5 is an example comparing Euclidean distance values for each mode according to the present invention.

According to Fig. 5, the polymer component according to the present invention can be classified by mode according to the degree of deterioration. The mode according to the present invention can be classified based on the final distance value described above regarding the degree of deterioration.

Specifically, the first mode may mean a stage where a slight deterioration has occurred, the second mode may mean a stage where the deterioration has progressed and replacement is required within a certain period of time, and the third mode may mean a stage where operation using the polymer component has been stopped and replacement is required immediately because the deterioration has progressed significantly.

That is, the first to third modes according to the present invention are classified according to the coping method according to the deterioration state, and the coping method according to each mode is different.

According to FIG. 5, the deterioration state according to the present invention can be divided into the first mode to the third mode. The final distance value in the first mode has a Euclidean distance value of 12 or more and less than 25, the final distance value in the second mode has a Euclidean distance value of 25 or more and less than 39, and the final distance value in the third mode can have a Euclidean distance value of 39 or more. In this case, the above-described preset Euclidean distance value, which is a criterion for detecting that the deterioration state has occurred, can be 12.

That is, the first to third modes can be characterized as being divided based on the premise that the polymer component is detected as being in a deteriorated state and classified according to the coping method. The distance value according to Fig. 5 can be understood as an example.

Figure 6 is a drawing illustrating a diffraction grating type spectrometer, and Figure 7 is a drawing illustrating a filter type spectrometer.

In a narrow sense, a spectrometer is a device that measures the intensity distribution of electromagnetic waves by breaking them down into differences in wavelength, but in a broad sense, it is a term that includes energy analysis devices for particle beams such as electron beams as well as electromagnetic waves.

According to Fig. 6, a spectrometer generally refers to a device that measures the intensity of light emitted by an analysis target by wavelength, or light reflected or transmitted by an external light source. As shown in Fig. 6, a spectrometer using a diffraction grating is mainly used, and in this case, the resolution is very excellent, down to 0.1 nm or less, but it is bulky and expensive. The degree to which a spectrometer accurately and precisely displays wavelength information is called resolution, and this resolution is evaluated as an important factor in evaluating the performance of a spectrometer.

According to Fig. 7, in contrast, a filter-type spectrometer using a transmission filter array can be miniaturized, but has a resolution of 3 to 10 nm, which limits its practical application, and requires the use of a two-dimensional array photodetector, making it expensive, especially in the infrared region. A filter-type spectrometer (200) is composed of an optical filter array (210) having predetermined different transmission characteristics to transmit light emitted, reflected, or transmitted from an analysis target, a photodetector array (220) that detects signals passing through a plurality of filters, and a control unit (230) that analyzes and processes the detected signals.

According to Fig. 7, the resolution of such a filter-type spectrometer improves as the optical correlation between filters is small and the number of filters increases, so it is necessary to have a plurality of optical filters with different characteristics in order to improve the resolution. In addition, in order to detect a signal passing through the filter array (210) constituting the spectrometer, a photodetector array (220) corresponding to the filter array (210) is required. For example, as shown in Fig. 2, a filter-type spectrometer (200) equipped with a two-dimensional optical filter array (210) must be equipped with a two-dimensional array-type photodetector (220), which accounts for the largest proportion of the cost of the filter-type spectrometer.

Figure 8 shows an example of a spectral spectrum according to the present invention.

According to Fig. 8, the spectral spectrum can display the characteristics of the component indicated by the corresponding peak according to the form of the spectrum. This is known in the prior art, and the second spectrum image according to the present invention can mean an image in which each peak is labeled with which component it means, as in Fig. 8. At this time, the spectrum image of Fig. 8 is an example showing the transmittance by functional group in an IR spectrometer.

Conversely, the present invention may use a spectrum image based on absorbance rather than transmittance, and since transmittance and absorbance are distinct from each other, only a spectrum image for one should be used. In addition, although there are various types of spectrometers for generating a spectrum image according to the present invention, such as an infrared spectrometer, one type should be used.

Figure 9 illustrates a camera module according to the present invention.

According to FIG. 9, the camera module (150) can generate image data from an optical image. The camera module (150) can include a housing (1324) including a through hole in a side wall, a lens (1321) installed in the through hole, and a driving unit (1323) that drives the lens (1321). The through hole can be formed to a size corresponding to the diameter of the lens (1321). The lens (1321) can be inserted into the through hole.

The lens driving unit (1323) may be configured to control the lens (1321) to move forward or backward. The lens (1321) and the lens driving unit (1323) may be connected in a conventionally known manner, and the lens (1321) may be controlled by the lens driving unit (1323) in a conventionally known manner.

To obtain various images, the lens (1321) needs to be exposed to the outside of the camera module (150) or housing (1324).

In particular, the lens (1321) according to the present invention should be positioned so as to be exposed to the outside of the housing (1324). Therefore, a lens with strong contamination resistance is required. The present invention proposes a coating layer for coating the lens to solve this problem.

Preferably, the lens (1321) may be coated on its surface with a coating composition including an acrylic compound represented by the following chemical formula 1; an organic solvent, inorganic particles, and a dispersant.

[Chemical Formula 1]

Here,

n and m are equal to or different from each other and are each independently an integer from 1 to 100,

L ₁ is a biphenylene group.

Since the lens (1321) can exhibit excellent water-repellent and contamination-resistant properties when coated with the above coating composition, even if the lens (1321) installed on the exterior of the vehicle is exposed to a polluted environment for a long period of time, images or videos that can be used as road information can be collected.

The above inorganic particles may be selected from the group consisting of silica, alumina, and mixtures thereof. The average diameter of the above inorganic particles is 70 to 100 μm, but is not limited to the above example. The above inorganic particles may be formed as a coating layer (1322) on the surface of the lens (1321), and may improve physical strength and maintain viscosity within a certain range to improve formability.

The above organic solvent is selected from the group consisting of methyl ethyl ketone (MEK), toluene, and mixtures thereof, and preferably methyl ethyl ketone can be used, but is not limited to the above examples.

As the above dispersant, a polyester series dispersant can be used, and specifically, TEGO-Disperse 670 (manufacturer: EVONIK) can be used as a polyester series dispersion stabilizer composed of a copolymer of 2-methoxypropyl acetate and 1-methoxy-2-propyl acetate, but is not limited to the above examples, and any dispersant obvious to those skilled in the art can be used without limitation.

The above coating composition may additionally contain a stabilizer as other additives, and the stabilizer may contain an ultraviolet absorber, an antioxidant, etc., but is not limited to the above examples and may be used without limitation.

The coating composition for forming the above coating layer (1322) may more specifically include an acrylic compound represented by the above chemical formula 1; an organic solvent, inorganic particles, and a dispersant.

The above coating composition may contain 40 to 60 parts by weight of an acrylic compound represented by the above chemical formula 1, 20 to 40 parts by weight of inorganic particles, and 5 to 15 parts by weight of a dispersant, per 100 parts by weight of an organic solvent. Within the above range, a synergistic effect is expressed to a critical degree due to the interaction of each component in terms of a water-repellent effect, and outside the above range, the synergistic effect is drastically reduced or becomes almost non-existent.

More preferably, the viscosity of the coating composition is 1500 to 1800 cP. If the viscosity is less than 1500 cP, there is a problem that when applied to the surface of a lens (1321), it flows down and it is not easy to form a coating layer (1322). If the viscosity exceeds 1800 cP, there is a problem that it is not easy to form a uniform coating layer (1322).

[Manufacturing Example 1: Manufacturing of coating layer]

1. Preparation of coating composition

A coating composition was prepared by mixing an acrylic compound represented by the following chemical formula 1, inorganic particles, and a dispersant into methyl ethyl ketone:

[Chemical Formula 1]

Here,

n and m are equal to or different from each other and are each independently an integer from 1 to 100,

L ₁ is a biphenylene group.

A more specific composition of the above anti-static composition is as shown in Table 2 below.

TX1 TX2 TX3 TX4 TX5 Organic solvent 100 100 100 100 100 Acrylic compounds 30 40 50 60 70 Weapon particles 10 20 30 40 50 Dispersant 1 5 10 15 20

(unit weight)

2. Manufacturing of coating layer

After applying the coating composition of DX1 to DX5 to one surface of the lens (1321), it was cured to form a coating layer (1322).

[Experimental example]

1. Evaluation of surface appearance

Due to the difference in viscosity of the coating composition, a sensory evaluation was conducted to determine whether a uniform surface was formed after the coating layer (1322) was manufactured. An evaluation was conducted to determine whether a uniform coating layer (1322) was formed, and the evaluation was conducted based on the following criteria.

○: Formation of a uniform coating layer

×: Formation of an uneven coating layer

TX1 TX2 TX3 TX4 TX5 Sensory evaluation X ○ ○ ○ X

When forming a coating layer (1322), if the viscosity is below a certain level, flow occurs on the surface of the lens (1321), and in many cases, it is difficult to form a uniform coating layer (1322) after the curing process. Accordingly, a problem of low production yield may occur. In addition, if the viscosity is too high, it is difficult to uniformly apply the composition, and thus it is impossible to form a uniform coating layer (1322).

2. Measurement of the water repellency angle

After forming a coating layer (1322) on the surface of the lens (1321), the results of measuring the water repellency angle are as shown in Table 4 below.

Advancing contact angle (") Stationary contact angle (") Backward contact angle (") TX1 117.1±2.9 112.1±4.1 < 10 TX2 132.4±1.5 131.5±2.7 141.7±3.4 TX3 138.9±3.0 138.9±2.7 139.8±3.7 TX4 136.9±2.0 135.6±2.6 140.4±3.4 TX5 116.9±0.7 115.4±3.0 < 10

As shown in Table 4 above, after forming a coating layer (1322) using the coating compositions of TX1 to TX5, the results of measuring the contact angle were confirmed. The receding contact angles of TX1 and TX5 were measured to be less than 10 degrees. That is, it was confirmed that when the optimal range for producing the coating composition was exceeded, a pinning phenomenon of water droplets occurred. On the other hand, it was confirmed that the pinning phenomenon did not occur in TX2 to 4, confirming that excellent waterproofing effects could be exhibited.

3. Evaluation of contamination resistance

A lens (1321) having a coating layer (1322) formed according to the above embodiment in an external environment was attached to a model camera and exposed to an external environment for 40 days. As a comparative example (Con), an identical lens (1321) without a coating layer (1322) formed was used, and the model camera for each example was installed in the same external environment.

Thereafter, the degree of contamination of the lens (1321) before and after the experiment was evaluated objectively, and the results were evaluated on an index of 1 to 10 compared to a comparative example in which a coating layer (1322) was not formed, and are shown in Table 5 below. The lower the number of the index below, the better the contamination resistance.

Con TX1 TX2 TX3 TX4 TX5 My contamination resistance 10 7 3 3 3 8

(Unit: Index)

Referring to Table 5 above, it can be seen that when a coating layer (1322) is formed on a lens (1321), image data can be collected in a form that is easy to analyze for a long period of time based on high contamination resistance even when the lens (1321) is exposed to the outside while installing the camera module in an external environment. In particular, it can be confirmed that the contamination resistance by the coating layer (1322) is very excellent in the case of TX2 to TX4.

FIG. 10 is a drawing showing an example of an image correction filter according to the present invention.

According to Fig. 10, the types and functions of filters are shown. That is, the CNN algorithm can be a learning algorithm that uses multiple layers. In addition, the CNN algorithm can automatically learn a filter that maximizes image classification accuracy, and can perform a classification operation on the filtered image after applying a filtering technique to the original image by adding new layers called convolutional layers and pooling layers before the fully-connected layer. The CNN algorithm can be configured to perform a classification operation on the filtered image after applying a filtering technique to the original image by adding new layers called convolutional layers and pooling layers before the fully-connected layer.

At this time, the calculation formula for the image filter using the CNN algorithm is as shown in mathematical formula 5 below.

[Mathematical Formula 5]

(step,

: pixel of the i-th row and j-th column of the filtered image expressed as a matrix,

: filter,

: image,

: Height of the filter (number of rows),

: is the width of the filter (number of columns).

Figure 11 illustrates a method for detecting degradation characteristics of a polymer component based on deep learning according to the present invention.

In addition, the subject performing the method for detecting degradation characteristics of a polymer component based on deep learning according to the present invention may be the user terminal (100 of FIG. 1) described above or a processor included in the user terminal (100 of FIG. 1).

According to FIG. 11, the method for detecting degradation characteristics of a polymer component based on deep learning according to the present invention may include a step (S1100) of learning a second spectrum image that is pre-labeled for a functional group included in a polymer component in a normal state, a step (S1200) of learning a first spectrum image for a target sample, a step (S1300) of detecting a degradation state of the target sample based on the learned result value and the first spectrum image, and a step (S1400) of displaying the degradation state to a user.

In the description of the method for detecting degradation characteristics of a polymer component based on deep learning according to the present invention, any content that is identical or overlapping with the description of the system for detecting degradation characteristics of a polymer component based on deep learning described above may be omitted.

The above-described present invention can be modeled as a computer-readable code on a medium in which a program is recorded. The computer-readable medium includes all kinds of recording devices that store data that can be read by a computer system. Examples of the computer-readable medium include a hard disk drive (HDD), a solid state disk (SSD), a silicon disk drive (SDD), a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like, and also includes those modeled in the form of a carrier wave (e.g., transmission via the Internet). Accordingly, the above detailed description should not be construed as limiting in all respects, but should be considered as illustrative. The scope of the present disclosure should be determined by a reasonable interpretation of the appended claims, and all changes within the equivalency range of the present disclosure are intended to be included in the scope of the present disclosure.

Any of the embodiments or other embodiments of the present invention described above are not mutually exclusive or distinct. Any of the embodiments or other embodiments of the present invention described above may have their respective configurations or functions combined or used together.

The above detailed description should not be construed as restrictive in all respects but should be considered as illustrative. The scope of the invention should be determined by a reasonable interpretation of the appended claims, and all changes coming within the equivalent scope of the invention are intended to be embraced within the scope of the invention.

100: User terminal
200: Spectroscope

Claims (7)

In a system for detecting the deterioration characteristics of polymer components,
A spectrometer for generating a first spectrum image of a sample extracted from said polymer component; and
A user terminal that learns the first spectrum image through a deep neural network, extracts feature points for the first spectrum image, and detects the deterioration state of the sample based on the feature points; including:
The above user terminal,
Learn the pre-labeled second spectrum image, and detect the deterioration state based on the learned result value and the first spectrum image.
The above second spectrum image is generated based on functional groups known in advance to be included in the polymer component,
The above user terminal,
Obtaining multiple coordinate information for multiple peaks included in the first spectrum image and the second spectrum image through the deep neural network, performing a K-centered clustering process on the multiple coordinate information, finally generating two clusters based on the K-centered clustering process, deriving a similarity based on the Euclidean distance between the two clusters, extracting an average value of each of the two clusters, extracting an average value of each of the two clusters, extracting a distance value between each average value, and detecting the deterioration state based on the distance value,
The distance value between (i) the wavenumber of the spectrum included in the first spectrum image and the second spectrum image is calculated by performing a clustering process using (i) the wavenumber of the spectrum as the x-coordinate and (ii) the transmittance of the wavelength as the y-coordinate, and (ii) the transmittance of the wavelength as the y-coordinate.
Multiplying the above distance value by a correction coefficient and detecting the deterioration state according to the corrected distance value.
A deep learning-based degradation characteristic detection system for polymer components.
delete delete In the first paragraph,
The above deep neural network,
Contains an image filter using a CNN algorithm,
A deep learning-based degradation characteristic detection system for polymer components.
In the first paragraph,
The above user terminal,
comprising a camera module for capturing the first spectrum image;
The above camera module,
a housing including a through hole in the side wall; and
A lens installed in the above through hole; including;
A deep learning-based degradation characteristic detection system for polymer components.
A step of learning a pre-labeled second spectrum image for functional groups included in a polymer component in a normal state;
A step of learning a first spectrum image for a target sample;
A step of detecting the deterioration state of the target sample based on the learned result value and the first spectrum image; and
A step of displaying the above deterioration state; comprising:
The step of detecting the above deterioration condition is:
A method of obtaining multiple coordinate information for multiple peaks included in the first spectrum image and the second spectrum image through a deep neural network, performing a K-centered clustering process on the multiple coordinate information, finally generating two clusters based on the K-centered clustering process, deriving a similarity based on the Euclidean distance between the two clusters, extracting an average value of each of the two clusters, extracting an average value of each of the two clusters, extracting a distance value between each average value, and detecting the deterioration state based on the distance value,
The distance value between (i) the wavenumber of the spectrum included in the first spectrum image and the second spectrum image is calculated by performing a clustering process using (i) the wavenumber of the spectrum as the x-coordinate and (ii) the transmittance of the wavelength as the y-coordinate, and (ii) the transmittance of the wavelength as the y-coordinate.
Multiplying the above distance value by a correction coefficient and detecting the deterioration state according to the corrected distance value.
A method for detecting degradation characteristics of polymer components based on deep learning.
As a non-transitory computer-readable medium storing instructions,
A non-transitory computer-readable medium having instructions that, when executed by a processor, cause the processor to perform the method of claim 6.