Failure prediction model generation method, monitoring method, and battery manufacturing method

Abstract

The technical idea of the present invention provides a monitoring method comprising the steps of: generating a failure prediction model for production equipment; and monitoring the production equipment on the basis of the failure prediction model. The step for generating the failure prediction model comprises the steps of: generating frequency domain data by processing, through a frequency analysis, time domain vibration data transmitted from a vibration sensor mounted in the production equipment; generating frequency domain of interest data by extracting, from the frequency domain data, data in a frequency band of interest for each component; generating production process-related data by determining a filtering period satisfying a process condition of the production equipment, and extracting data in the filtering period from the frequency domain of interest data; and generating index data by standardizing the production process-related data.

Description

Method for creating a failure prediction model, monitoring method, and battery manufacturing method

[0001] The present invention relates to a method for generating a failure prediction model, a monitoring method, and a battery manufacturing method, and more particularly, to a method for generating a failure prediction model for predicting equipment failure based on vibration data, a monitoring method including the failure prediction model generation method, and a battery manufacturing method including the failure prediction model generation method.

[0002] This application claims the benefit of priority to Republic of Korea Patent Application No. 10-2023-0167381, filed on November 28, 2023, and Republic of Korea Patent Application No. 10-2024-0105841, filed on August 8, 2024, all of which are incorporated herein by reference.

[0003] One method for monitoring equipment failures involves setting a control line (or threshold) based on data acquired at a low sampling rate, and generating a failure alert when the detected value exceeds the control line. However, because equipment contains a variety of interconnected components and the factors that indicate equipment condition vary widely, detecting failures individually can be challenging.

[0004] The technical problem to be achieved by the present invention is to provide a method for generating a failure prediction model.

[0005] The technical problem to be achieved by the present invention is to provide a monitoring method including a method for generating a failure prediction model.

[0006] The technical problem to be solved by the present invention is to provide a battery manufacturing method including a method for generating a failure prediction model.

[0007] In order to solve the above-described problem, the technical idea of ​​the present invention includes a step of creating a failure prediction model for a production facility; and a step of monitoring the production facility based on the failure prediction model; wherein the step of creating the failure prediction model includes a step of processing time-domain vibration data transmitted from a vibration sensor mounted on the production facility through frequency analysis to create frequency-domain data; a step of extracting data in a frequency band of interest for each component from the frequency-domain data to create frequency-domain data of interest; a step of determining a filtering section that satisfies a process condition of the production facility, and extracting data in the filtering section from the frequency-domain data of interest to create production process-related data; and a step of standardizing the production process-related data to create index data.

[0008] In exemplary embodiments, the frequency domain data is characterized in that it is generated by processing time domain acceleration data or time domain velocity data with a fast Fourier transform.

[0009] In exemplary embodiments, the interest frequency domain data is characterized in that it is generated by extracting data in the interest frequency band for each component from the data in the frequency domain, and calculating statistics using the root mean square for the extracted data.

[0010] In exemplary embodiments, the filtering section is characterized in that the rotational speed of the parts equipped in the production facility is determined as a section within a target range.

[0011] In exemplary embodiments, the production process related data is characterized in that it is generated by extracting data in the filtering section from the frequency domain data of interest and calculating an average value for each window determined based on other process conditions of the production facility for the extracted data.

[0012] In exemplary embodiments, the window is characterized in that it is determined as a time interval during which a set of items is processed in the production facility.

[0013] In exemplary embodiments, the step of generating the index data is characterized by including the step of calculating a Z-score for the production process associated data.

[0014] In exemplary embodiments, the method further comprises a step of determining a management line that serves as a criterion for determining a failure of the production facility from the index data.

[0015] In exemplary embodiments, the method further comprises, after the step of generating the failure prediction model, a step of evaluating the failure prediction model.

[0016] In exemplary embodiments, the vibration sensor is characterized by including an acceleration sensor.

[0017] In exemplary embodiments, the production facility is characterized by including a mixer for producing electrode slurry for secondary batteries.

[0018] In exemplary embodiments, the mixer is configured to agitate a material for electrode slurry, and the mixer is configured to perform a first agitating process of stirring the material by rotating the stirring blade at a first agitating speed and a second agitating process of stirring the material by rotating the stirring blade at a second agitating speed greater than the first agitating speed, and the filtering section is characterized in that the mixer is determined to be a section in which the second agitating process is performed.

[0019] In order to solve the above-described problem, the technical idea of ​​the present invention provides a method for generating a failure prediction model for a production facility, comprising: a step of processing time-domain vibration data transmitted from a vibration sensor mounted on the production facility through frequency analysis to generate frequency-domain data; a step of extracting data in a frequency band of interest for each component from the frequency-domain data, and calculating statistics using a root mean square for the extracted data to generate frequency-domain data of interest; a step of extracting data in a filtering section determined based on a process condition of the production facility from the data in the frequency-domain of interest, and calculating an average value of the extracted data for each window determined based on another process condition of the production facility to generate production process-related data; a step of calculating a standard score for the production process-related data to generate index data; and a step of determining a management line that serves as a criterion for determining a failure of the production facility from the index data.

[0020] In exemplary embodiments, the filtering section is characterized in that the rotational speed of the parts equipped in the production facility is determined as a section within a target range.

[0021] In exemplary embodiments, the window is characterized in that it is determined as a time interval during which a set of items is processed in the production facility.

[0022] In exemplary embodiments, the production facility is characterized by including a mixer for producing electrode slurry for secondary batteries.

[0023] In exemplary embodiments, the mixer is configured to agitate a material for electrode slurry, and the mixer is configured to perform a first agitating process of stirring the material by rotating the stirring blade at a first agitating speed and a second agitating process of stirring the material by rotating the stirring blade at a second agitating speed greater than the first agitating speed, and the filtering section is characterized in that the mixer is determined to be a section in which the second agitating process is performed.

[0024] In order to solve the above-described problem, the technical idea of ​​the present invention provides a battery manufacturing method, including the steps of: generating a failure prediction model for a production facility; performing a battery manufacturing process with the production facility; and monitoring the production facility based on the failure prediction model, wherein the production facility is a mixer configured to rotate a stirring blade to stir a material contained in a stirring vessel, and the step of generating the failure prediction model includes: processing time-domain vibration data transmitted from a vibration sensor mounted on the production facility through frequency analysis to generate frequency-domain data; extracting data in a frequency band of interest for each component from the frequency-domain data to generate frequency-domain data of interest; determining a filtering section that satisfies a process condition of the production facility, and extracting data in the filtering section from the frequency-domain data of interest to generate production process-related data; and standardizing the production process-related data to generate index data.

[0025] According to exemplary embodiments of the present invention, a failure prediction model can be established to predict failures in production equipment based on vibration data acquired from a vibration sensor. By monitoring production equipment based on the failure prediction model, failures in production equipment can be predicted in advance, gradual deterioration of production equipment can be diagnosed, and defective components in production equipment can be identified.

[0026] The effects that can be obtained from the exemplary embodiments of the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly derived and understood by those skilled in the art to which the exemplary embodiments of the present disclosure pertain from the following description. In other words, unintended effects resulting from practicing the exemplary embodiments of the present disclosure can also be derived by those skilled in the art from the exemplary embodiments of the present disclosure.

[0027] FIG. 1 is a schematic diagram showing a manufacturing device according to exemplary embodiments of the present invention.

[0028] Figure 2 is a flowchart illustrating a monitoring method according to exemplary embodiments of the present invention.

[0029] FIG. 3 is a diagram illustrating a failure prediction model generation unit according to exemplary embodiments of the present invention.

[0030] FIG. 4 is a flowchart illustrating a method for generating a failure prediction model according to exemplary embodiments of the present invention.

[0031] FIG. 5A is a diagram showing a first graph representing first interest frequency domain data related to a first interest component, which is one of a plurality of components, and a second graph representing a section of the first graph indicated by “AA” in an enlarged manner.

[0032] FIG. 5b is a diagram showing a third graph representing first production process related data related to a first part of interest and a fourth graph representing a section indicated by “AA” in the third graph in an enlarged form.

[0033] FIG. 5c is a diagram showing a fifth graph representing first index data related to a first component of interest and a sixth graph representing a section indicated by “AA” in the fifth graph in an enlarged manner.

[0034] Figure 6 is a flowchart illustrating a battery manufacturing method according to exemplary embodiments of the present invention.

[0035] Figure 7 is a schematic diagram showing a battery manufacturing facility according to exemplary embodiments of the present invention.

[0036] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. Prior to this, it should be noted that the terms and words used in this specification and claims should not be construed as limited to their conventional or dictionary meanings. Based on the principle that the inventor can appropriately define the concepts of terms to best explain his or her invention, they should be interpreted in a way that aligns with the technical spirit of the present invention.

[0037] Accordingly, the embodiments described in this specification and the configurations illustrated in the drawings are merely the most preferred embodiments of the present invention and do not represent all of the technical ideas of the present invention. Therefore, it should be understood that there may be various equivalents and modified examples that can replace them at the time of filing this application.

[0038] In addition, when describing the present invention, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present invention, the detailed description is omitted.

[0039] Since the embodiments of the present invention are provided to more fully explain the present invention to those skilled in the art, the shapes and sizes of components in the drawings may be exaggerated, omitted, or schematically illustrated for clearer explanation. Accordingly, the sizes and proportions of each component do not fully reflect the actual sizes or proportions.

[0040]

[0041] (Example 1)

[0042] FIG. 1 is a schematic diagram showing a manufacturing device (10) according to exemplary embodiments of the present invention.

[0043] Referring to FIG. 1, the manufacturing device (10) may include a production facility (100), a failure prediction model generation unit (200), and a monitoring unit (300).

[0044] A production facility (100) can manufacture a product by performing a production process on the product. The production facility (100) can include a plurality of parts (110) and a vibration sensor (130).

[0045] In exemplary embodiments, the production facility (100) is a facility for manufacturing a secondary battery, and the production facility (100) may be configured to manufacture a secondary battery or a component constituting a secondary battery. In exemplary embodiments, the production facility (100) may include a facility for manufacturing an electrode slurry for a secondary battery. In exemplary embodiments, the production facility (100) may include a mixer configured to mix materials constituting the electrode slurry for a secondary battery (e.g., a conductive material, a binder, an active material, etc.).

[0046] The vibration sensor (130) is mounted on the production facility (100), detects vibration of the production facility (100), and can generate time domain vibration data (VD1, VD2) acquired at a high sampling rate. The vibration sensor (130) may include an acceleration sensor, a velocity sensor, and / or a displacement sensor. The time domain vibration data (VD1, VD2) generated by the vibration sensor (130) may include acceleration data, velocity data, and / or displacement data. The vibration sensor (130) may include a non-contact sensor and / or a contact sensor.

[0047] The failure prediction model generation unit (200) can receive and store vibration data (VD1) transmitted from the vibration sensor (130), and can generate a failure prediction model (FM) for predicting failure of the production facility (100) based on the vibration data (VD1) and monitoring the status of the production facility (100).

[0048] The monitoring unit (300) can receive and store the failure prediction model (FM) transmitted from the failure prediction model generation unit (200), and can receive and store vibration data (VD2) from the production facility (100). While the production process for an item is in progress in the production facility (100), the monitoring unit (300) monitors the production facility (100) based on the failure prediction model (FM) and vibration data (VD2), thereby being able to predict failure of the production facility (100) in advance, and identify a component (110) that causes failure of the production facility (100).

[0049] The monitoring unit (300) may include at least one memory device configured to store data, at least one processor configured to process data, and a control panel. For example, the memory device may include a random access memory (RAM) and / or a read only memory (ROM). The processor may include a central processing unit (CPU), a microprocessor unit (MPU), and / or a graphic processing unit (GPU). The monitoring unit (300) may include a computer and / or a server.

[0050] Figure 2 is a flowchart illustrating a monitoring method according to exemplary embodiments of the present invention.

[0051] Referring to FIGS. 1 and 2, a monitoring method according to exemplary embodiments may include a step (S100) of generating a failure prediction model (FM) and a step (S200) of performing monitoring on a production facility (100) based on the failure prediction model (FM).

[0052] In the step (S100) of generating a failure prediction model (FM), while operating the production facility (100) to perform a production process for an item, vibration of the production facility (100) is detected for a predetermined period of time using a vibration sensor (130), and a failure prediction model (FM) can be generated based on vibration data (VD1) acquired from the vibration sensor (130) for a predetermined period of time. The step S100 will be described in detail later.

[0053] In the step (S200) of monitoring a production facility (100) based on a failure prediction model (FM), while operating the production facility (100) to perform a production process for an item, vibration of the production facility (100) is detected by a vibration sensor (130), and the production facility (100) can be monitored based on the failure prediction model (FM) and vibration data (VD2). The monitoring unit (300) can predict a failure of the production facility (100) in advance based on the failure prediction model (FM) and vibration data (VD2) and identify a component (110) that causes a failure of the production facility (100).

[0054] FIG. 3 is a configuration diagram showing a failure prediction model generation unit (200) according to exemplary embodiments of the present invention.

[0055] Referring to FIGS. 1 and 3, the failure prediction model generation unit (200) may include a data processing unit (210) including first to fifth data processing units (211, 212, 213, 214, 215), an interest frequency band input unit (231), a process condition input unit (233), a memory device (250), and a model evaluation unit (260). The first to fifth data processing units (211, 212, 213, 214, 215) and the model evaluation unit (260) may include a processor for processing data, for example, a processor such as a CPU, an MPU, or a GPU. The memory device (250) may include a RAM and / or a ROM.

[0056] The first data processing unit (211) may receive time domain vibration data (VD1) from a vibration sensor (130) mounted on the production facility (100) and generate frequency domain data (D1) from the time domain vibration data (VD1) through frequency analysis. In exemplary embodiments, the first data processing unit (211) may convert the time domain vibration data (VD1) into frequency domain data (D1) based on a fast Fourier transform (FFT). The frequency domain data (D1) may be frequency domain acceleration data or frequency domain velocity data.

[0057] In exemplary embodiments, the first data processing unit (211) may perform preprocessing on the time-domain vibration data (VD1) prior to processing the data based on frequency analysis. For example, if the time-domain vibration data (VD1) is acceleration data, the first data processing unit (211) may integrate the acceleration data to generate velocity data.

[0058] The second data processing unit (212) can receive data (FD) on the frequency band of interest for each component from the frequency band of interest input unit (231). The frequency band of interest for each component can include a plurality of different frequency bands of interest, and each of the plurality of frequency bands of interest can be related to one component among the plurality of components (110) provided in the production facility (100). Each frequency band of interest can correspond to a frequency band suitable for identifying a failure of a component of interest selected from among the components (110) of the production facility (100). In exemplary embodiments, the component of interest of the production facility (100) can be a rotating component configured to rotate about a rotation axis, such as a bearing, and the frequency band of interest can include a defect frequency of the rotating component.

[0059] The second data processing unit (212) can extract data within the frequency band of interest for each component from the frequency domain data (D1) transmitted from the first data processing unit (211) and generate frequency domain data of interest (D2). For example, when the production facility (100) includes a first component of interest and a second component of interest, the first frequency band of interest related to the first component of interest and the second frequency band of interest related to the second component of interest may be different from each other. In this case, the second data processing unit (212) can generate first frequency domain data of interest (D2a) within the first frequency band of interest and second frequency domain data of interest within the second frequency band of interest from the frequency domain data (D1).

[0060] The second data processing unit (212) can extract data in the frequency region of interest for each component from the frequency domain data (D1) and calculate statistics using the root mean square (RMS) for the extracted data to generate the frequency region of interest data (D2). In this case, the value of each data point of the frequency region of interest data (D2) may correspond to the RMS value. For example, the second data processing unit (212) can extract data in the frequency region of interest for each component from the frequency domain data (D1) and calculate the RMS for the extracted data for each predetermined window. For example, the window may be several to several thousand ms, but is not limited thereto.

[0061] The third data processing unit (213) can receive production process data (PD) regarding process conditions of the production facility (100) from the process condition input unit (233). The production process data (PD) may include data regarding a production process recipe of the production facility (100) and / or process data related to a plurality of components (110) provided in the production facility (100). For example, the production process data (PD) may include data regarding current, voltage, pressure, temperature, speed, revolutions per minute (RPM), etc. The production process data (PD) may be data acquired at a low sampling rate. For example, the process condition input unit (233) may receive and store production process data (PD) related to components of interest of the production facility (100) from the production facility (100), and transmit the production process data (PD) to the third data processing unit (213).

[0062] The third data processing unit (213) can determine a filtering section based on the production process data (PD) transmitted from the process condition input unit (233), and extract data within the filtering section from the interest frequency domain data (D2) to generate production process-related data (D3). The filtering section can be determined based on the process recipe of the production facility (100). Production conditions more suitable for detecting defects and / or deterioration of the part of interest can be determined, and a time section during which the production facility (100) operates under the determined production conditions can be the filtering section. In exemplary embodiments, when the part of interest is a rotating part configured to rotate about a rotational axis, such as a bearing, the filtering section can be determined as a time section during which the rpm of the rotating part is within a target rpm range. For example, when the rotating part is configured to operate in a low-rpm mode, a medium-rpm mode, and a high-rpm mode, the filtering section can be determined as a time section during which the rotating part is operated in the high-rpm mode. In exemplary embodiments, the component of interest may be a rotating component, such as a bearing, provided in a mixer for manufacturing electrode slurry for secondary batteries.

[0063] In exemplary embodiments, the production facility (100) may be a mixer for manufacturing electrode slurry for secondary batteries. The mixer may include a mixing vessel (510 in FIG. 7) for containing materials for manufacturing electrode slurry (e.g., solvent, active material, conductive material, and binder), a stirring blade (520 in FIG. 7) for stirring the materials contained in the mixing vessel (510), a rotary actuator (530 in FIG. 7) for rotating the stirring blade (520), and a bearing (540 in FIG. 7) mounted on a rotational axis of the stirring blade (520) for rotating in conjunction with the stirring blade (520). In exemplary embodiments, the component of interest may be the bearing (540). The mixer may be configured to perform a feeding process of feeding materials for manufacturing an electrode slurry (e.g., a solvent, an active material, a conductive material, and a binder) into a mixing vessel (510), a first stirring process of stirring the materials contained in the mixing vessel (510) while rotating a stirring blade (520) at a first stirring speed, and a second stirring process of stirring the materials contained in the mixing vessel (510) at a second stirring speed greater than the first stirring speed. The first stirring process may be a low-speed stirring process of stirring the materials at a relatively low stirring speed (e.g., 100 rpm to 500 rpm). The second stirring process may be a high-speed stirring process that is performed subsequent to the first stirring process and stirs the materials at a relatively high stirring speed (e.g., 1000 rpm to 2000 rpm). In exemplary embodiments, when extracting data within the filtering section from the interest frequency domain data (D2) to generate production process associated data (D3), the filtering section may be determined as a time section during which the second stirring process is performed in the mixer, and data acquired during the time section during which the second stirring process is performed in the mixer among the interest frequency domain data (D2) may become the production process associated data (D3).The third data processing unit (213) can extract data within the filtering section from the interest frequency domain data (D2), and perform an average processing on the extracted data for each predetermined window to generate production process-related data (D3). The window can be determined based on the process conditions of the production facility (100). The process conditions of the production facility (100) for determining the filtering section and the process conditions of the production facility (100) for determining the window can be different from each other. For example, when the production facility (100) is configured to process a set of products through a single production process, the window can be determined as a time period during which the set of products is processed. For example, in order to generate the production process-related data (D3), the window can be determined based on the process conditions of the production facility (100), data within the filtering section from the interest frequency domain data (D2), and an operation can be performed to obtain an average value for each window on the extracted data. For example, when the production facility (100) is a mixer configured to perform a mixing process for manufacturing an electrode slurry, the window can be determined as a time interval during which a mixing process is performed for a predetermined amount of material for the electrode slurry.

[0064] The fourth data processing unit (214) can standardize the production process-related data (D3) transmitted from the third data processing unit (213) to generate index data (D4). The index data (D4) can be obtained by standardizing the production process-related data (D3) based on the normal operating state of the production facility (100). Since the index data (D4) is obtained through standardization, it becomes an indicator that can be commonly applied even when there is a sensing deviation by sensor or a deviation by facility. The standardization may include obtaining a standard score for the production process-related data (D3). In exemplary embodiments, the standardization may include obtaining a Z-score for the production process-related data (D3). The Z-score can be obtained through the following equation (1).

[0065] ... Equation (1)

[0066] In the above equation (1), x(i) represents a value at each data point of the production process associated data (D3), μ represents an average value of the steady-state production process associated data generated based on vibration data acquired during normal operation of the production facility (100), σ represents a standard deviation value of the steady-state production process associated data generated based on vibration data acquired during normal operation of the production facility (100), and Z(i) represents a Z-score corresponding to each data point of the production process associated data (D3).

[0067] The fifth data processing unit (215) can determine a management line (or threshold) that serves as a criterion for determining a failure of the production facility (100) for the index data (D4) transmitted from the fourth data processing unit (214). When the monitoring unit (300) monitors the production facility (100), if the value of the index data generated based on the vibration data (VD2) acquired from the vibration sensor (130) exceeds the management line, the monitoring unit (300) can determine that a defect has occurred in a specific component of the production facility (100). For example, determining the management line may include determining a management line for a warning notification and a management line that serves as a criterion for determining a failure.

[0068] When the management line for the index data (D4) is determined by the fifth data processing unit (215), the data processing unit (210) can output a failure prediction model (FM). The data processing unit (210) transmits the failure prediction model (FM) to a memory device (250), and the memory device (250) can store the failure prediction model (FM). The failure prediction model (FM) stored in the memory device (250) can be provided to the monitoring unit (300) through the output unit of the failure prediction model (FM).

[0069] The model evaluation unit (260) can perform an evaluation task on the failure prediction model (FM) completed by the data processing unit (210). The model evaluation unit (260) can perform an evaluation on the failure prediction model (FM) based on past data acquired from the vibration sensor (130) of the production facility (100). For example, after loading past vibration data acquired during a period in which a failure occurred in the production facility (100), the past vibration data can be applied to the failure prediction model (FM) to determine whether a failure of the production facility (100) is detected in advance. In addition, after loading past vibration data acquired during normal operation of the production facility (100), the past vibration data can be applied to the failure prediction model (FM) to determine whether it is determined to be in a normal state. If the failure prediction model (FM) is determined to be inappropriate by the model evaluation unit (260), the data processing unit (210) can discard the existing failure prediction model (FM) and generate a new failure prediction model. If the failure prediction model (FM) is determined to be suitable through the model evaluation unit (260), the failure prediction model (FM) can be provided to the monitoring unit (300).

[0070]

[0071] (Example 2)

[0072] FIG. 4 is a flowchart illustrating a method for generating a failure prediction model according to exemplary embodiments of the present invention.

[0073] FIG. 5a is a diagram showing a first graph (410) representing first interest frequency domain data (D2a) related to a first interest component, which is one of a plurality of components (110), and a second graph (411) representing a section indicated by “AA” of the first graph (410) in an enlarged form.

[0074] FIG. 5b is a diagram showing a third graph (420) representing first production process related data (D3a) related to a first part of interest, and a fourth graph (421) showing a section of the third graph (420) indicated by “AA” in an enlarged form.

[0075] FIG. 5c is a drawing showing a fifth graph (430) representing first index data (D4a) related to a first part of interest and a sixth graph (431) showing a section indicated by “AA” of the fifth graph (430) in an enlarged form.

[0076] Hereinafter, a method for generating a failure prediction model according to exemplary embodiments of the present invention will be described with reference to FIGS. 4, 5a to 5c and FIGS. 1 to 3.

[0077] First, the production facility (100) is operated, and time-domain vibration data (VD1) is acquired for a predetermined period of time using a vibration sensor (130) mounted on the production facility (100) (S110). The vibration sensor (130) can transmit the time-domain vibration data (VD1) to a failure prediction model generation unit (200).

[0078] Next, the first data processing unit (211) of the failure prediction model generation unit (200) can process time domain vibration data (VD1) based on FFT to generate frequency domain data (D1) (S120).

[0079] Next, the second data processing unit (212) extracts data in the frequency band of interest for each component from the frequency domain data (D1) and processes the extracted data through RMS analysis to generate frequency domain data of interest (D2) (S130).

[0080] In Fig. 5a, a first graph (410) and a second graph (411) for first interest frequency domain data (D2a) related to a first component of interest are illustrated. Specifically, data in a first interest frequency band related to the first component of interest is extracted from the frequency domain data (D1), and the extracted data is processed through RMS analysis, thereby generating first interest frequency domain data (D2a) related to the first component of interest. The first component of interest may correspond to a rotating component having a rotational axis (e.g., a bearing).

[0081] Next, the third data processing unit (213) determines a filtering section and window that satisfy the process conditions of the production facility (100), extracts data in the filtering section from the interest frequency region data (D2), and calculates an average value for each window for the extracted data to generate production process-related data (D3) (S140).

[0082] In Fig. 5b, a third graph (420) and a fourth graph (421) for first production process related data (D3a) generated from first interest frequency domain data (D2a) related to a first part of interest are illustrated. A time period during which the rpm of the first part of interest operates within a predetermined target range may be determined as a first filtering period (415 in Fig. 5a). In addition, when the production facility (100) is configured to process a set of products through a single production process, the window (425) may be determined as a time period during which a set of products is processed in the production facility (100). Once the first filtering section (415) and window (425) are determined, data within the first filtering section (415) is extracted from the first interest frequency domain data (D2a), and an average value is calculated for each predetermined window (425) from the extracted data, thereby generating first production process-related data (D3a) related to the first part of interest.

[0083] Next, the fourth data processing unit (214) standardizes the production process-related data (D3) based on the normal operating state of the production facility (100) to generate index data (D4) (S150). The index data (D4) can be generated by calculating a Z-score for the production process-related data (D3). The steady-state production process-related data can be calculated based on vibration data acquired in the normal operating state of the production facility (100), the average value and standard deviation value of the steady-state production process-related data can be calculated, and the Z-score for the production process-related data (D3) can be calculated using the average value and standard deviation value of the steady-state production process-related data. The steady-state production process-related data can be obtained through steps substantially the same as steps S110 to S140 described above.

[0084] In Fig. 5c, a fifth graph (430) and a sixth graph (431) are illustrated for first index data (D4a) generated from first interest frequency domain data (D2a) related to a first part of interest. By calculating a Z-score for first production process-related data (D3a) related to the first part of interest, the first index data (D4a) related to the first part of interest can be generated.

[0085] Next, a management line is determined from the index data (D4) (S160). Determining the management line may include determining a management line for warning notification and a management line that serves as a criterion for fault determination.

[0086] As illustrated in Fig. 5c, a first management line (ML1) and a second management line (ML2) can be determined for the first index data (D4a) related to the first component of interest. The first management line (ML1) is when the value of the first index data (D4a) is the first value (V1), and the second management line (ML2) is when the value of the first index data (D4a) is the second value (V2) greater than the first value (V1). When monitoring the production facility (100), index data for monitoring is generated based on vibration data (VD2), and when the value of the index data for monitoring is between the first value (V1) of the first management line (ML1) and the second value (V2) of the second management line (ML2), a warning notification can be generated, and when the value of the index data for monitoring exceeds the second value (V2) of the second management line (ML2), a failure notification can be generated. The above monitoring index data can be obtained through steps substantially identical to steps S110 to S150 described above.

[0087] Once the step of determining a management line for the index data (D4) is completed, the failure prediction model generation unit (200) can generate and store a failure prediction model (FM). The model evaluation unit (260) can evaluate the suitability of the failure prediction model (FM) using past data, and the failure prediction model (FM) that has passed the suitability evaluation of the model evaluation unit (260) can be provided to the monitoring unit (300). The monitoring unit (300) can perform monitoring of the production facility (100) based on the failure prediction model (FM).

[0088] According to exemplary embodiments of the present invention, a failure prediction model (FM) capable of predicting a failure of a production facility (100) can be established based on vibration data (VD1) acquired from a vibration sensor (130). By monitoring the production facility (100) based on the failure prediction model (FM), a failure of the production facility (100) can be predicted in advance, a gradual deterioration phenomenon of the production facility (100) can be diagnosed, and a failed component of the production facility (100) can be identified.

[0089]

[0090] (Example 3)

[0091] Figure 6 is a flowchart illustrating a battery manufacturing method according to exemplary embodiments of the present invention. Figure 7 is a schematic diagram illustrating a battery manufacturing facility (500) according to exemplary embodiments of the present invention. Hereinafter, a battery manufacturing method according to exemplary embodiments will be described with reference to Figures 1 to 7.

[0092] A battery manufacturing method according to exemplary embodiments may include a step (S310) of generating a failure prediction model (FM) for a battery manufacturing facility (500), a step (S320) of performing a battery manufacturing process, and a step (S330) of monitoring the battery manufacturing facility (500) based on the failure prediction model (FM). The battery manufacturing facility (500) may correspond to the production facility (100) of FIG. 1.

[0093] In step S310, while operating the battery manufacturing facility (500) to proceed with the battery manufacturing process, vibration of the battery manufacturing facility (500) is detected for a predetermined period of time using a vibration sensor (130), and a failure prediction model (FM) can be generated based on vibration data (VD1) acquired from the vibration sensor (130) for the predetermined period of time. Step S310 may include steps S110 to S160 described with reference to FIG. 4.

[0094] In exemplary embodiments, the battery manufacturing facility (500) may include a mixer configured to perform a mixing process for manufacturing an electrode slurry for a secondary battery. The mixer may include a mixing vessel (510) that accommodates materials (590) for manufacturing an electrode slurry (e.g., a solvent, an active material, a conductive material, and a binder), a stirring blade (520) that agitates the materials (590) accommodated in the mixing vessel (510), a rotary actuator (530) that rotates the stirring blade (520), and a bearing (540) that is mounted on a rotational axis of the stirring blade (520) and rotates in conjunction with the stirring blade (520).

[0095] The mixer may be configured to perform a feeding process of feeding a material (590) for producing an electrode slurry into a mixing vessel (510), a first stirring process of stirring the material (590) contained in the mixing vessel (510) while rotating a stirring blade (520) at a first stirring speed, and a second stirring process of stirring the material (590) contained in the mixing vessel (510) at a second stirring speed greater than the first stirring speed. The first stirring process may be a low-speed stirring process of stirring the material (590) at a relatively low stirring speed (e.g., 100 rpm to 500 rpm). The second stirring process may be a high-speed stirring process that is performed subsequent to the first stirring process and stirs the material (590) at a relatively high stirring speed (e.g., 1000 rpm to 2000 rpm). In exemplary embodiments, in step S140, when extracting data within the filtering section from the interest frequency domain data (D2) to generate production process-related data (D3), the filtering section may be determined as a section in which the second stirring process is performed in the mixer.

[0096] In step S320, the battery manufacturing equipment (500) is operated to perform the battery manufacturing process. The battery manufacturing process may include a mixing process using a mixer.

[0097] At step S330, while the mixer is performing a mixing process, the vibration of the mixer is detected by a vibration sensor (130), and the battery manufacturing equipment (500) can be monitored based on a failure prediction model (FM) and vibration data (VD2). The monitoring unit (300) can predict a failure of the mixer in advance based on the failure prediction model (FM) and vibration data (VD2) and identify a component (110) that causes a failure of the battery manufacturing equipment (500).

[0098] According to exemplary embodiments of the present invention, a failure prediction model (FM) capable of predicting a failure of a battery manufacturing facility (500) configured to perform a battery manufacturing process can be established based on vibration data (VD1) acquired from a vibration sensor (130). By monitoring the battery manufacturing facility (500) based on the failure prediction model (FM), a failure of the battery manufacturing facility (500) can be predicted in advance, a gradual deterioration phenomenon of the battery manufacturing facility (500) can be diagnosed, and a failed component of the battery manufacturing facility (500) can be identified.

[0099]

[0100] The present invention has been described in more detail through drawings and examples. However, the configurations described in the drawings or examples described in this specification are merely embodiments of the present invention and do not represent all of the technical ideas of the present invention. Therefore, it should be understood that various equivalents and modified examples may exist as of the time of this application.

Claims

1. A step for creating a failure prediction model for production equipment; and A step of monitoring the production facility based on the above failure prediction model; Including, The steps for generating the above failure prediction model are: A step of processing time domain vibration data transmitted from a vibration sensor mounted on the above production facility through frequency analysis to generate frequency domain data; A step of extracting data in a frequency band of interest for each component from the data in the frequency domain above to generate data in a frequency domain of interest; A step of determining a filtering section that satisfies the process conditions of the above production facility, extracting data in the filtering section from the interest frequency domain data, and generating production process-related data; and A step of generating index data by standardizing the above production process related data; A monitoring method comprising:

2. In paragraph 1, A monitoring method characterized in that the above frequency domain data is generated by processing time domain acceleration data or time domain velocity data using a fast Fourier transform.

3. In paragraph 1, A monitoring method characterized in that the above interest frequency domain data is generated by extracting data in the interest frequency band for each component from the data in the frequency domain and calculating statistics using the root mean square for the extracted data.

4. In paragraph 1, A monitoring method characterized in that the above filtering section is determined as a section in which the rotational speed per minute of the parts equipped in the production facility is within the target range.

5. In paragraph 1, A monitoring method characterized in that the above production process related data is generated by extracting data in the filtering section from the above interest frequency domain data and calculating an average value for each window determined based on other process conditions of the production facility for the extracted data.

6. In paragraph 5, A monitoring method characterized in that the above window is determined as a time interval during which a set of goods is processed in the above production facility.

7. In paragraph 1, A monitoring method, characterized in that the step of generating the above index data includes the step of calculating a Z-score for the above production process associated data.

8. In paragraph 1, A monitoring method characterized by further comprising a step of determining a management line that serves as a criterion for determining a failure of the production facility from the above index data.

9. In paragraph 1, A monitoring method characterized in that, after the step of generating the above-mentioned failure prediction model, it further includes a step of evaluating the above-mentioned failure prediction model.

10. In paragraph 1, A monitoring method, characterized in that the vibration sensor includes an acceleration sensor.

11. In paragraph 1, A monitoring method, characterized in that the above production facility includes a mixer for manufacturing an electrode slurry for a secondary battery.

12. In paragraph 11, The above mixer is configured to stir the material for electrode slurry, The above mixer is configured to perform a first stirring process of stirring the material by rotating the stirring blade at a first stirring speed and a second stirring process of stirring the material by rotating the stirring blade at a second stirring speed greater than the first stirring speed. A monitoring method, characterized in that the above filtering section is determined as a section in which the mixer performs the second stirring process.

13. A method for creating a failure prediction model for production equipment, A step of processing time domain vibration data transmitted from a vibration sensor mounted on the above production facility through frequency analysis to generate frequency domain data; A step of extracting data in a frequency band of interest for each component from the data in the frequency domain above, and generating data in a frequency domain of interest by calculating statistics using the root mean square for the extracted data; A step of extracting data in a filtering section determined based on process conditions of the production facility from the above-mentioned interest frequency domain data, and calculating an average value of the extracted data for each window determined based on other process conditions of the production facility to generate production process-related data; A step of generating index data by calculating a standard score for the above production process related data; and A step of determining a management line that serves as a criterion for determining a failure of the production facility from the above index data; A method for generating a failure prediction model including:

14. In paragraph 13, A method for generating a failure prediction model, characterized in that the above filtering section is determined as a section in which the rotational speed per minute of the parts equipped in the production facility is within the target range.

15. In paragraph 13, A method for generating a failure prediction model, wherein the above window is determined as a time interval during which a set of products is processed in the above production facility.

16. In paragraph 13, A method for generating a failure prediction model, characterized in that the above production facility includes a mixer for manufacturing electrode slurry for secondary batteries.

17. In paragraph 16, The above mixer is configured to stir the material for electrode slurry, The above mixer is configured to perform a first stirring process of stirring the material by rotating the stirring blade at a first stirring speed and a second stirring process of stirring the material by rotating the stirring blade at a second stirring speed greater than the first stirring speed. A method for generating a failure prediction model, characterized in that the above filtering section is determined as a section in which the mixer performs the second stirring process.

18. Step of creating a failure prediction model for production equipment; A step of performing a battery manufacturing process using the above production facility; and A step of monitoring the production facility based on the above failure prediction model; Including, The above production facility is a mixer configured to rotate a stirring blade to stir materials contained in a stirring container, The steps for generating the above failure prediction model are: A step of processing time domain vibration data transmitted from a vibration sensor mounted on the above production facility through frequency analysis to generate frequency domain data; A step of extracting data in a frequency band of interest for each component from the data in the frequency domain above to generate data in a frequency domain of interest; A step of determining a filtering section that satisfies the process conditions of the above production facility, extracting data in the filtering section from the interest frequency domain data, and generating production process-related data; and A step of generating index data by standardizing the above production process related data; A method for manufacturing a battery comprising:

Images