A system and method for detecting and analyzing the implanting of special-shaped bristles on a toothbrush
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGSU AIKESI AUTOMATION TECH CO LTD
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-24
AI Technical Summary
Existing methods for inspecting the bristle implantation process of irregularly shaped toothbrushes are post-process inspections, which make it difficult to achieve real-time feedback and adaptive adjustments, resulting in a high scrap rate.
A six-dimensional force sensor is used to collect real-time six-dimensional force time-series waveform data for hair implantation, train a machine learning model to predict hair implantation indicators, and optimize the parameters of the hair implantation equipment through backpropagation to achieve adaptive adjustment.
It enables real-time monitoring of the tufting process, predicts whether the tufting indicators are up to standard, and improves tufting quality and production efficiency by adaptively adjusting equipment parameters.
Smart Images
Figure CN122439993A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of irregular bristle processing and inspection technology, specifically to a system and method for detecting and analyzing irregular bristle toothbrush bristle processing. Background Technology
[0002] The manufacturing of irregularly shaped toothbrushes is a sophisticated automated process, mainly consisting of three stages: brush head preparation, bristle implantation, and subsequent finishing. In the bristle implantation stage, advanced automated bristle implantation equipment integrates a six-dimensional force sensor, which can sense the vertical pressure, horizontal thrust, and other multi-dimensional forces during the bristle implantation process in real time, ensuring that the bristle implantation depth and force are uniform and consistent, achieving adaptive precision control.
[0003] However, existing methods for detecting irregularly shaped bristle toothbrush implantation are all "post-inspection" methods. By the time problems are discovered, defective toothbrushes have already been produced, and the overall process is mostly open-loop or semi-closed-loop, making it difficult to adaptively adjust the bristle implantation process based on real-time feedback. Summary of the Invention
[0004] The purpose of this invention is to provide a system and method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, so as to solve at least one of the above-mentioned deficiencies in the prior art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, comprising the following steps: S1. Collect six-dimensional force time-series waveform data of hair implantation under different working conditions for each hair implantation. The six-dimensional force sensor can be set at the connection between the hair implantation needle seat and the robotic arm, or at the root of the hair implantation needle holder, to capture the changes in vertical pressure, horizontal thrust, lateral tension, and rotational torque around the three axes in real time and accurately during hair implantation. Different working conditions include different brush batches, brush moisture content, brush head injection shrinkage rate, hair implantation needle wear degree, workshop temperature, humidity, hair implantation pressure, and needle insertion speed.
[0006] S2. Collect the index detection data corresponding to each six-dimensional force-time waveform data, and generate corresponding bristle implantation index detection data. The index detection data includes bristle pull-out force, bristle density, bristle fold flatness, and bristle shape, etc. Among them, bristle pull-out force refers to the maximum force value for pulling the implanted bristles out of the brush head. It is the most critical index in the bristle implantation process, directly reflecting the fixation firmness. Insufficient pull-out force will cause the bristles to fall off during use, posing a risk of accidental ingestion. Bristle density reflects whether the bristles are evenly distributed on the surface of the brush head. Uneven density will lead to cleaning problems. Inconsistent force can lead to dirt and grime buildup in some areas and reduced cleaning effectiveness in others. During implantation, the bristles are folded in half; the flatness of this fold directly affects the bristle filling within the holes and the final appearance of the implanted bristles. Uneven folding results in uneven force distribution on some bristles within the holes, affecting the consistency of the extraction force and the bristle bundle shape. Bristle bundle shape refers to the regularity of the shape of individual bristle bundles after implantation, i.e., whether the bristles are neatly clustered and aligned. Poor bundle shape (such as splayed, crooked, or scattered bristles) directly affects the toothbrush's appearance and user experience.
[0007] S3. Using the six-dimensional force time-series waveform data of hair implantation as input and the corresponding hair implantation index detection data as output, train the first machine learning model to obtain the hair implantation index prediction model. S4. Collect data on the changes in the parameters of the hair implantation equipment and the changes in the six-dimensional force-time waveform characteristics of hair implantation for each hair implantation relative to the previous hair implantation. S5. Using the six-dimensional force time-series waveform feature change data of hair implantation as input and the corresponding hair implantation equipment parameter change data as output, train the second machine learning model to obtain the hair implantation equipment parameter change prediction model. S6. Collect real-time six-dimensional force time-series waveform data of hair implantation and input it into the hair implantation index prediction model to obtain the current hair implantation index prediction data. S7. Determine whether the current predicted data for hair implantation indicators meets the set standard data for hair implantation indicators; S8. If yes, the corresponding indicators for hair implantation meet the standards, and the current process ends and returns to S6.
[0008] Furthermore, the method also includes the following steps: S8. If not, then construct an index loss function based on the output of the hair implantation index prediction model and the index error of the corresponding set hair implantation index standard data. In the hair implantation index prediction model, search for the hair implantation six-dimensional force time series waveform data that minimizes the index error function through backpropagation, and generate hair implantation six-dimensional force time series waveform prediction data. Furthermore, during the backpropagation process, optimization algorithms such as Adam, RMSprop, and PGD can be used to solve the problem. At the same time, constraints are added during the solution process to ensure that the solved hair implantation six-dimensional force time series waveform prediction data is within the normal range. S9. Input the error value between the predicted six-dimensional force time-series waveform data of hair implantation and the real-time six-dimensional force time-series waveform data of hair implantation into the prediction model of the parameter change of hair implantation equipment to obtain the predicted parameter change data of hair implantation equipment and generate the parameter correction data of hair implantation equipment. S10. Correct the hair implantation parameters of the hair implantation equipment according to the hair implantation equipment parameter correction data.
[0009] Furthermore, the six-dimensional force-time waveform data of hair implantation acquired in S1 is divided into multiple sub-waveforms, including: The six-dimensional force timing waveform data of hair implantation from the time the needle tip contacts the brush bristles to the time before it enters the hair implantation hole is divided into six-dimensional force timing sub-waveform data of the hair implantation needle entry period. The six-dimensional force time-series waveform data corresponding to the bristle pressing into the hole is divided into six-dimensional force time-series sub-waveform data of the bristle implantation period. The six-dimensional force-time waveform data corresponding to the period from needle withdrawal from the implantation hole to complete detachment is divided into six-dimensional force-time sub-waveform data for the needle withdrawal period. During the subsequent training of the first machine learning model, each sub-waveform data is preprocessed separately to generate a six-dimensional force-time vector.
[0010] Furthermore, S3 includes the following steps: S3.1. After preprocessing the six-dimensional force time-series waveform data of the hair-implanting, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of hair-implanting. The preprocessing includes zero-drift calibration: automatically acquiring and subtracting the no-load reference value before each hair implantation; noise suppression: using wavelet thresholding for noise reduction or Savitzky-Golay filtering to smooth the waveform, etc. Feature extraction can include the maximum, minimum, mean, variance, kurtosis, skewness, peak time, rise / fall slope, extraction of main frequency components and energy distribution through FFT transformation, and waveform morphology features (area, envelope, and number of zero crossings, etc.).
[0011] S3.2 Normalize the continuous values in the hair implantation index detection data, and normalize the discrete category indexes after one-hot encoding to generate the corresponding hair implantation index detection normalized data; among them, for continuous values (such as plucking force, hair polishing pass rate, bundle type score, etc.), Min-Max normalization to [0,1] or Z-score standardization can be used, and for discrete categories (such as hair deviation, missing hair, hair sticking, etc.), one-hot encoding can be used.
[0012] S3.3 Collect the six-dimensional force temporal feature vector of hair implantation and the normalized data of hair implantation index detection for each hair implantation, and form a first sample data pair. Collect all first sample data pairs to generate the first sample set. S3.4. Using the six-dimensional temporal feature vector of hair implantation force in the first sample set as input and the normalized data of the corresponding hair implantation index detection as output, train the first machine learning model to obtain the hair implantation index prediction model. Specifically, the first machine learning model can be a dual-stream fusion architecture of a convolutional neural network model and a long short-term memory network model, using a one-dimensional convolutional neural network to extract local pattern features and a long short-term memory network model to capture temporal dependencies, and outputting multi-task predictions after fusion; or a temporal convolutional network, Transformer Encoder, or other models can be used.
[0013] During training, the sample set is divided into a training set, a validation set, and a test set. The training set can use a large amount of historical waveform data (e.g., ≥5000 data points, covering normal and various abnormal operating conditions). K-fold cross-validation is used on the validation set to evaluate the model's generalization ability and prevent overfitting. R0 is used to evaluate the model's generalization ability. 2 Regression metrics such as RMSE and MAE are used to evaluate the predictive accuracy of continuous metrics, while accuracy, precision, and recall are used to evaluate the predictive performance of categorized metrics (such as defect identification). Before deployment in real-world applications, the model's performance on new batches of products is validated using an independent test set.
[0014] Furthermore, S4 includes the following steps: S4.1 Collect the parameter data of the hair implantation equipment each time hair implantation is performed, and calculate the change value of the current hair implantation equipment parameter data relative to the previous hair implantation equipment parameter data to obtain the hair implantation equipment parameter change data; wherein, the hair implantation equipment parameters may include hair implantation pressure setting value, hair implantation speed, needle insertion depth, XYZ coordinate offset of hair implantation needle, robotic arm posture adjustment amount, brush head clamping force, and multi-axis linkage parameters for irregular holes, etc. S4.2. After preprocessing the six-dimensional force time-series waveform data of each hair implantation, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of hair implantation. S4.3 Calculate the change value of the current six-dimensional force temporal feature vector of tufting relative to the previous six-dimensional force temporal feature vector of tufting, and obtain the temporal feature change data of the six-dimensional force of tufting.
[0015] Furthermore, S5 includes the following steps: S5.1 Normalize the parameter variation data of the hair implantation equipment to obtain normalized parameter variation data of the hair implantation equipment, and record the normalization parameters to obtain normalized parameter data of the equipment parameters; for example, when using Z-score for normalization, record the mean and standard deviation of each dimension parameter; when using Min-Max normalization, record the minimum, maximum, lower bound and upper bound of each dimension parameter. S5.2. Using the six-dimensional force time-series waveform feature change data of hair implantation as input and the normalized data of the corresponding hair implantation equipment parameter change as output, train the second machine learning model. S5.3. Based on the normalized parameter data of the equipment parameters, the output of the trained second machine learning model is denormalized, and the trained second machine learning model and the denormalization process are combined and encapsulated to obtain the feather implantation equipment parameter change prediction model.
[0016] Furthermore, S7 includes the following steps: S7.1 For indicators with continuous values in the current hair transplantation index prediction data, determine whether they all meet the corresponding set lower specification limits, such as hair tufting force ≥15N, adult hair type ≥40%, tuft type score ≥85 points, etc. S7.2 For indicators that are discrete categories in the current hair grafting indicator prediction data, determine whether the probability of defects in all categories is less than the set indicator category defect probability threshold. S7.3 Determine whether both S7.1 and S7.2 are true.
[0017] A detection and analysis system for the bristle implantation process of an irregularly shaped toothbrush is used to perform a detection and analysis method for the bristle implantation process of an irregularly shaped toothbrush, including a bristle implantation device, a six-dimensional force sensor, a storage module, and a smart chip. The bristle implantation device is used to implant bristles into a brush head and includes the following structure: The frame, electrical control cabinet, and feeding / transfer mechanism (such as robotic arm and turntable) are responsible for the precise positioning, transfer, and unloading of the brush handles; Multi-axis robotic arm: connects to the machine head and is responsible for spatial positioning; Hair implantation machine head: includes drive unit (motor, cam mechanism, etc.); Hair implantation needle holder: connects the machine head and hair implantation needle, and is the main deployment location of force sensors; Hair implantation needle holder: directly grasps the brush bristles, and its clamping force and movement directly affect the hair implantation quality; used to perform needle insertion, hair implantation and needle retraction operations.
[0018] The bristle box and bristle feeding mechanism are responsible for storing and precisely feeding the bristles to the machine head.
[0019] The six-dimensional force sensor is used to monitor the six-dimensional force timing waveform during each hair implantation by the hair implantation device, generate corresponding hair implantation six-dimensional force timing waveform data, and store it in the storage module. The storage module is also used to store computer programs; The smart chip is used to run the computer program to execute a method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, and to correct the bristle implantation parameters of the bristle implantation equipment based on feedback.
[0020] Beneficial effects: Compared with the prior art, the present invention provides a detection and analysis system and method for the processing of irregularly shaped bristles in toothbrushes. By collecting six-dimensional force time-series waveform data during bristle implantation under different working conditions and the corresponding detection index data after bristle implantation, a bristle implantation index prediction model is trained. This achieves the effect of predicting the corresponding index to determine whether the bristle implantation is qualified based on the real-time monitored six-dimensional force time-series waveform data of the support mold.
[0021] By analyzing the relationship between the six-dimensional force-time waveform characteristics of hair implantation and the parameter changes of hair implantation equipment, a prediction model for the parameter changes of hair implantation equipment is constructed. An index loss function is constructed by using the output of the hair implantation index prediction model and the index error of the corresponding set standard data for hair implantation index. Backpropagation is performed in the hair implantation equipment parameter change prediction model to minimize the index loss function, thereby realizing the prediction of the optimal correction amount of hair implantation equipment parameters. This allows for adaptive adjustment of the hair implantation equipment to ensure the hair implantation effect. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.
[0023] Figure 1 This is a flowchart illustrating the method steps provided in an embodiment of the present invention; Figure 2 This is a system structure block diagram provided for an embodiment of the present invention. Detailed Implementation
[0024] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0025] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0026] Exemplary embodiments will be described more fully below with reference to the accompanying drawings; however, these exemplary embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will enable those skilled in the art to fully understand the scope of this disclosure.
[0027] Where there is no conflict, the various embodiments of this disclosure and the features thereof in the embodiments may be combined with each other.
[0028] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.
[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. As used herein, the singular forms “a” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the said feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded.
[0030] The embodiments described herein can be described with reference to plan views and / or cross-sectional views using the ideal schematic diagrams of this disclosure. Therefore, the example illustrations can be modified according to manufacturing techniques and / or tolerances. Therefore, the embodiments are not limited to those shown in the drawings, but include modifications to configurations formed based on manufacturing processes. Therefore, the areas illustrated in the drawings are schematic in nature, and the shapes of the areas shown in the figures illustrate specific shapes of areas of an element, but are not intended to be limiting.
[0031] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the same meaning as they have in the context of the relevant art and this disclosure.
[0032] Please see Figure 1 A method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, comprising the following steps: S1. Collect six-dimensional force time-series waveform data of hair implantation under different working conditions for each hair implantation. The six-dimensional force sensor can be set at the connection between the hair implantation needle seat and the robotic arm, or at the root of the hair implantation needle holder, to capture the changes in vertical pressure, horizontal thrust, lateral tension, and rotational torque around the three axes in real time and accurately during hair implantation. Different working conditions include different brush batches, brush moisture content, brush head injection shrinkage rate, hair implantation needle wear degree, workshop temperature, humidity, hair implantation pressure, and needle insertion speed.
[0033] Furthermore, the six-dimensional force-time waveform data of the tufted hair can be divided into multiple sub-waveforms, including: The six-dimensional force timing waveform data of hair implantation from the time the needle tip contacts the brush bristles to the time before it enters the hair implantation hole is divided into six-dimensional force timing sub-waveform data of the hair implantation needle entry period. The six-dimensional force time-series waveform data corresponding to the bristle pressing into the hole is divided into six-dimensional force time-series sub-waveform data of the bristle implantation period. The six-dimensional force-time waveform data corresponding to the period from needle withdrawal from the implantation hole to complete detachment is divided into six-dimensional force-time sub-waveform data for the needle withdrawal period. During the subsequent training of the first machine learning model, each sub-waveform data is preprocessed separately to generate a six-dimensional force-time vector.
[0034] S2. Collect the index detection data corresponding to each six-dimensional force-time waveform data, and generate corresponding bristle implantation index detection data. The index detection data includes bristle pull-out force, bristle density, bristle fold flatness, and bristle shape, etc. Among them, bristle pull-out force refers to the maximum force value for pulling the implanted bristles out of the brush head. It is the most critical index in the bristle implantation process, directly reflecting the fixation firmness. Insufficient pull-out force will cause the bristles to fall off during use, posing a risk of accidental ingestion. Bristle density reflects whether the bristles are evenly distributed on the surface of the brush head. Uneven density will lead to cleaning problems. Inconsistent force can lead to dirt and grime buildup in some areas and reduced cleaning effectiveness in others. During implantation, the bristles are folded in half; the flatness of this fold directly affects the bristle filling within the holes and the final appearance of the implanted bristles. Uneven folding results in uneven force distribution on some bristles within the holes, affecting the consistency of the extraction force and the bristle bundle shape. Bristle bundle shape refers to the regularity of the shape of individual bristle bundles after implantation, i.e., whether the bristles are neatly clustered and aligned. Poor bundle shape (such as splayed, crooked, or scattered bristles) directly affects the toothbrush's appearance and user experience.
[0035] S3. Using the six-dimensional force-time waveform data of hair implantation as input and the corresponding hair implantation index detection data as output, train the first machine learning model to obtain the hair implantation index prediction model, including the following steps: S3.1. After preprocessing the six-dimensional force time-series waveform data of tufted hair, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of tufted hair. The preprocessing includes zero-drift calibration: automatically acquiring and subtracting the no-load reference value before each hair implantation; noise suppression: using wavelet thresholding for noise reduction or Savitzky-Golay filtering to smooth the waveform, etc. Feature extraction can include the maximum, minimum, mean, variance, kurtosis, skewness, peak time, rise / fall slope, extraction of main frequency components and energy distribution through FFT transformation, and waveform morphology features (area, envelope, and number of zero crossings, etc.).
[0036] S3.2 Normalize the continuous values in the hair grafting index detection data, and normalize the discrete category indexes after one-hot encoding to generate the corresponding hair grafting index detection normalized data; among them, for continuous values (such as plucking force, hair grinding qualification rate, bundle type score, etc.), Min-Max normalization to [0,1] or Z-score standardization can be used, and for discrete categories (such as hair deviation, missing hair, hair sticking, etc.), one-hot encoding can be used.
[0037] S3.3 Collect the six-dimensional force temporal feature vector of hair implantation and the normalized data of hair implantation index detection for each hair implantation, and form a first sample data pair. Collect all first sample data pairs to generate the first sample set. S3.4. Using the six-dimensional temporal feature vector of hair implantation force in the first sample set as input and the normalized data of the corresponding hair implantation index detection as output, train the first machine learning model to obtain the hair implantation index prediction model. Specifically, the first machine learning model can be a dual-stream fusion architecture of a convolutional neural network model and a long short-term memory network model, using a one-dimensional convolutional neural network to extract local pattern features and a long short-term memory network model to capture temporal dependencies, and outputting multi-task predictions after fusion; or a temporal convolutional network, Transformer Encoder, or other models can be used.
[0038] During training, the sample set is divided into a training set, a validation set, and a test set. The training set can use a large amount of historical waveform data (e.g., ≥5000 data points, covering normal and various abnormal operating conditions). K-fold cross-validation is used on the validation set to evaluate the model's generalization ability and prevent overfitting. R0 is used to evaluate the model's generalization ability. 2 Regression metrics such as RMSE and MAE are used to evaluate the predictive accuracy of continuous metrics, while accuracy, precision, and recall are used to evaluate the predictive performance of categorized metrics (such as defect identification). Before deployment in real-world applications, the model's performance on new batches of products is validated using an independent test set.
[0039] S4. Collect data on the changes in hair implantation equipment parameters and the changes in the six-dimensional force-time waveform characteristics of hair implantation for each hair implantation relative to the previous hair implantation, including the following steps: S4.1 Collect the parameter data of the hair implantation equipment each time hair implantation is performed, and calculate the change value of the current hair implantation equipment parameter data relative to the previous hair implantation equipment parameter data to obtain the hair implantation equipment parameter change data; wherein, the hair implantation equipment parameters may include hair implantation pressure setting value, hair implantation speed, needle insertion depth, XYZ coordinate offset of hair implantation needle, robotic arm posture adjustment amount, brush head clamping force, and multi-axis linkage parameters for irregular holes, etc. S4.2. After preprocessing the six-dimensional force time-series waveform data of each hair implantation, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of hair implantation. S4.3 Calculate the change value of the current six-dimensional force temporal feature vector of tufting relative to the previous six-dimensional force temporal feature vector of tufting, and obtain the temporal feature change data of the six-dimensional force of tufting.
[0040] S5. Using the six-dimensional force-time waveform feature change data of hair grafting as input and the corresponding hair grafting equipment parameter change data as output, train the second machine learning model to obtain the hair grafting equipment parameter change prediction model, including the following steps: S5.1 Normalize the parameter variation data of the tufting equipment to obtain normalized parameter variation data, and record the normalization parameters to obtain normalized parameter data of the equipment parameters; for example, when using Z-score for normalization, record the mean and standard deviation of each dimension parameter; when using Min-Max normalization, record the minimum, maximum, lower bound, and upper bound of each dimension parameter. S5.2. Using the six-dimensional force time-series waveform feature change data of hair implantation as input and the normalized data of the corresponding hair implantation equipment parameter change as output, train the second machine learning model. S5.3. Based on the normalized parameter data of the equipment parameters, the output of the trained second machine learning model is denormalized, and the trained second machine learning model and the denormalized data are combined and encapsulated to obtain the prediction model of the change of the feather implantation equipment parameters.
[0041] S6. Collect real-time six-dimensional force time-series waveform data for hair implantation and input it into the hair implantation index prediction model to obtain the current hair implantation index prediction data.
[0042] S7. Determine whether the current predicted data for hair transplantation indicators meets the set standard data for hair transplantation indicators, including the following steps: S7.1 For indicators with continuous values in the current hair transplantation index prediction data, determine whether they all meet the corresponding set lower specification limits, such as hair tufting force ≥15N, adult hair type ≥40%, tuft type score ≥85 points, etc. S7.2 For indicators that are discrete categories in the current hair grafting indicator prediction data, determine whether the probability of defects in all categories is less than the set indicator category defect probability threshold. S7.3 Determine whether both S7.1 and S7.2 are true.
[0043] S8. If yes, the corresponding indicators for hair implantation meet the standards, and the current process ends and returns to S6. If not, then construct an index loss function using the output of the tufting index prediction model and the index error of the corresponding tufting index standard data. In the tufting index prediction model, search for the tufting six-dimensional force time series waveform data that minimizes the index error function through backpropagation, and generate tufting six-dimensional force time series waveform prediction data. Furthermore, during the backpropagation process, optimization algorithms such as Adam, RMSprop, and PGD can be used to solve the problem. At the same time, constraints are added during the solution process to ensure that the solved tufting six-dimensional force time series waveform prediction data is within the normal range. S9. Input the error value between the predicted data of the six-dimensional force time-series waveform of hair implantation and the real-time data of the six-dimensional force time-series waveform of hair implantation into the prediction model of the parameter change of hair implantation equipment to obtain the predicted data of the parameter change of hair implantation equipment and generate the correction data of the parameters of hair implantation equipment. S10. Correct the hair implantation parameters of the hair implantation equipment based on the hair implantation equipment parameter correction data.
[0044] Please refer to Figure 2 A detection and analysis system for the bristle implantation process of irregularly shaped toothbrushes, comprising bristle implantation equipment, a six-dimensional force sensor, a storage module, and a smart chip; Bristle implantation equipment is used to implant bristles into a brush head and includes the following structure: The frame, electrical control cabinet, and feeding / transfer mechanism (such as robotic arm and turntable) are responsible for the precise positioning, transfer, and unloading of the brush handles; Multi-axis robotic arm: connects to the machine head and is responsible for spatial positioning; Hair implantation machine head: includes drive unit (motor, cam mechanism, etc.); Hair implantation needle holder: connects the machine head and hair implantation needle, and is the main deployment location of force sensors; Hair implantation needle holder: directly grasps the brush bristles, and its clamping force and movement directly affect the hair implantation quality; used to perform needle insertion, hair implantation and needle retraction operations.
[0045] The bristle box and bristle feeding mechanism are responsible for storing and precisely feeding the bristles to the machine head.
[0046] The six-dimensional force sensor is used to monitor the six-dimensional force timing waveform of the hair implantation equipment during each hair implantation, generate the corresponding hair implantation six-dimensional force timing waveform data, and store it in the storage module; The storage module is also used to store computer programs; The smart chip is used to run a computer program to execute a method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, so as to correct the bristle implantation parameters of the bristle implantation equipment based on feedback.
[0047] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. A method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes, characterized in that, Includes the following steps: S1. Acquire the six-dimensional force timing waveform data of hair implantation under different working conditions for each hair implantation using a six-dimensional force sensor; S2. Collect the index detection data corresponding to each six-dimensional force time-series waveform data, and generate the corresponding hair implantation index detection data; S3. Using the six-dimensional force time-series waveform data of hair implantation as input and the corresponding hair implantation index detection data as output, train the first machine learning model to obtain the hair implantation index prediction model. S4. Collect data on the changes in the parameters of the hair implantation equipment and the changes in the six-dimensional force-time waveform characteristics of hair implantation for each hair implantation relative to the previous hair implantation. S5. Using the six-dimensional force time-series waveform feature change data of hair implantation as input and the corresponding hair implantation equipment parameter change data as output, train the second machine learning model to obtain the hair implantation equipment parameter change prediction model. S6. Collect real-time six-dimensional force time-series waveform data of hair implantation and input it into the hair implantation index prediction model to obtain the current hair implantation index prediction data. S7. Determine whether the current predicted data for hair implantation indicators meets the set standard data for hair implantation indicators; S8. If yes, the corresponding indicators for hair implantation meet the standards, and the current process ends and returns to S6.
2. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, The method further includes the following steps: S8. If not, construct an index loss function based on the output of the hair implantation index prediction model and the index error of the corresponding set hair implantation index standard data. In the hair implantation index prediction model, search for the hair implantation six-dimensional force time series waveform data that minimizes the index error function through backpropagation, and generate hair implantation six-dimensional force time series waveform prediction data. S9. Input the error value between the predicted six-dimensional force time-series waveform data of hair implantation and the real-time six-dimensional force time-series waveform data of hair implantation into the prediction model of the parameter change of hair implantation equipment to obtain the predicted parameter change data of hair implantation equipment and generate the parameter correction data of hair implantation equipment. S10. Correct the hair implantation parameters of the hair implantation equipment according to the hair implantation equipment parameter correction data.
3. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, The six-dimensional force-time waveform data of hair implantation acquired in S1 is divided into multiple sub-waveforms, including: The six-dimensional force timing waveform data of hair implantation from the time the needle tip contacts the brush bristles to the time before it enters the hair implantation hole is divided into six-dimensional force timing sub-waveform data of the hair implantation needle entry period. The six-dimensional force time-series waveform data corresponding to the bristle pressing into the hole is divided into six-dimensional force time-series sub-waveform data of the bristle implantation period. The six-dimensional force time-series waveform data corresponding to the period from needle withdrawal from the hair implantation hole to complete detachment is divided into six-dimensional force time-series sub-waveform data of the hair implantation needle withdrawal period.
4. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, S3 includes the following steps: S3.
1. After preprocessing the six-dimensional force time-series waveform data of the hair-implanting, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of hair-implanting. S3.2 Normalize the continuous values in the hair implantation index detection data, and normalize the discrete category indexes after one-hot encoding to generate the corresponding hair implantation index detection normalized data. S3.3 Collect the six-dimensional force temporal feature vector of hair implantation and the normalized data of hair implantation index detection for each hair implantation, and form a first sample data pair. Collect all first sample data pairs to generate the first sample set. S3.
4. Using the six-dimensional force time-series feature vector of hair implantation in the first sample set as input and the normalized data of the corresponding hair implantation index detection as output, train the first machine learning model to obtain the hair implantation index prediction model.
5. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, S4 includes the following steps: S4.1 Collect the parameter data of the hair implantation equipment each time hair implantation is performed, and calculate the change value of the current hair implantation equipment parameter data relative to the previous hair implantation equipment parameter data to obtain the hair implantation equipment parameter change data. S4.
2. After preprocessing the six-dimensional force time-series waveform data of each hair implantation, feature extraction is performed and then normalization is performed to generate the corresponding six-dimensional force time-series feature vector of hair implantation. S4.3 Calculate the change value of the current six-dimensional force temporal feature vector of tufting relative to the previous six-dimensional force temporal feature vector of tufting, and obtain the temporal feature change data of the six-dimensional force of tufting.
6. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, S5 includes the following steps: S5.1 Normalize the parameter change data of the hair implantation equipment to obtain normalized parameter change data of the hair implantation equipment, and record the normalization processing parameters to obtain normalized parameter data of the equipment parameters. S5.
2. Using the six-dimensional force time-series waveform feature change data of hair implantation as input and the normalized data of the corresponding hair implantation equipment parameter change as output, train the second machine learning model. S5.
3. Based on the normalized parameter data of the equipment parameters, the output of the trained second machine learning model is denormalized, and the trained second machine learning model and the denormalization process are combined and encapsulated to obtain the feather implantation equipment parameter change prediction model.
7. The method for detecting and analyzing the bristle implantation process of an irregularly shaped toothbrush according to claim 1, characterized in that, S7 includes the following steps: S7.1 For indicators with continuous values in the current flocking indicator prediction data, determine whether they all meet the corresponding set lower specification limit; S7.2 For indicators that are discrete categories in the current hair grafting indicator prediction data, determine whether the probability of defects in all categories is less than the set indicator category defect probability threshold. S7.3 Determine whether both S7.1 and S7.2 are true.
8. A detection and analysis system for irregularly shaped bristle toothbrush bristle implantation processing, used to execute the detection and analysis method for irregularly shaped bristle toothbrush bristle implantation processing as described in any one of claims 1-7, characterized in that, Includes hair implantation equipment, a six-dimensional force sensor, a storage module, and a smart chip; The bristle implantation device is used to implant bristles into the brush head; The six-dimensional force sensor is used to monitor the six-dimensional force timing waveform during each hair implantation by the hair implantation device, generate corresponding hair implantation six-dimensional force timing waveform data, and store it in the storage module. The storage module is also used to store computer programs; The smart chip is used to run the computer program and execute the method for detecting and analyzing the bristle implantation process of irregularly shaped toothbrushes as described in any one of claims 1-7.