Probe life detection method and probe station

Abstract

The present invention relates to a probe life detection method and a probe station, wherein the probe life detection method includes: training a probe life deep learning model; its specific training method includes: collecting needle mark data at different time sequences during the full life cycle of the probe under different force control parameters; processing each needle mark data to highlight the time sequence characteristics of the needle mark; putting each processed needle mark data into the model for training to obtain a probe life deep learning model; using the trained probe life deep learning model; it specifically includes: collecting needle mark images of the probe under the current force control parameters; processing the needle mark images to highlight the time sequence characteristics of the needle marks; the probe life deep learning model identifies the time sequence characteristics of the processed needle mark images to obtain the current use time of the probe. According to the technical solution of the present invention, it can improve the accuracy of probe life detection.

Description

Technical Field

[0001] The present invention relates to the technical field of probe life detection, and in particular to a probe life detection method and a probe station. Background Art

[0002] The probes on the probe station are important tools for detecting the electrical properties of wafers and are the bridge between the tester and the probe station. With the development of technology, the life of the probe has become the key to improving production efficiency and product quality. If the probe continues to be used after its life has expired, it will cause the probe to malfunction and affect production. At present, traditional probe life detection methods mainly rely on empirical rules and simple statistical models, which limits the accuracy and applicability of probe life detection. How to improve the accuracy of probe life detection is a technical problem that technicians in this field need to solve. Summary of the invention

[0003] In view of this, the present invention provides a probe life detection method and a probe station, and the main technical problem to be solved is: how to improve the accuracy of probe life detection.

[0004] In order to achieve the above object, the present invention mainly provides the following technical solutions:

[0005] An embodiment of the present invention provides a probe life detection method, which includes:

[0006] Training a probe life deep learning model; the specific training method includes: step S11: collecting needle mark data at different time sequences during the entire life cycle of the probe under different force control parameters; step S12: processing each of the needle mark data to highlight the time sequence characteristics of the needle mark; step S13: putting the processed needle mark data into the model for training to obtain a probe life deep learning model;

[0007] Use the trained probe life deep learning model; specifically:

[0008] Step S21: collecting an image of the needle mark of the probe under the current force control parameters;

[0009] Step S22: processing the acupuncture needle mark image to highlight the time sequence characteristics of the needle mark;

[0010] Step S23: The probe life deep learning model identifies the time series features of the processed acupuncture needle mark image to obtain the current usage time of the probe.

[0011] In some embodiments, using the trained probe lifetime deep learning model further comprises:

[0012] Step S24: Obtain the remaining life of the probe under the current force control parameters according to the current use time of the probe and the life threshold of the probe under the current force control parameters.

[0013] In some embodiments, the probe life detection method further comprises:

[0014] Step S25: changing the force control parameters of the probe, repeating the above steps S21 to S24, and obtaining the remaining life of the probe under different force control parameters;

[0015] Step S26: selecting the force control parameter under the maximum remaining life of the probe as the optimal force control parameter, and adjusting the current force control parameter of the probe to the optimal force control parameter.

[0016] In some embodiments, the processing of each of the acupuncture needle mark data in step S12 to highlight the time sequence characteristics of the needle marks is specifically:

[0017] The needle mark data are subjected to needle mark edge extraction, illumination homogenization processing, filtering and noise removal, and Fourier transformation in sequence to highlight the temporal characteristics of the needle marks.

[0018] In some embodiments, in step S13, the processed needle mark data of each acupuncture needle are placed in a model for training, specifically: the processed needle mark data of each acupuncture needle are placed in a D-CNN-LSTM model for training.

[0019] In some embodiments, in step S22, the acupuncture needle mark image is processed to highlight the temporal characteristics of the needle mark, specifically:

[0020] The needle mark data are subjected to illumination uniformization processing, dynamic threshold segmentation, pad extraction and grayscale value scaling in sequence to highlight the timing characteristics of the needle marks.

[0021] In some embodiments, the force control parameter of the probe includes at least one of the running speed of the probe, the running acceleration of the probe, and the running smoothing time of the probe.

[0022] The present invention also provides a probe station, which includes an image acquisition module, a data processing module and an identification module;

[0023] Wherein, the image acquisition module is used to acquire the needle mark image of the probe under the current force control parameters;

[0024] The data processing module is used to process the acupuncture needle mark image to highlight the time sequence characteristics of the needle mark;

[0025] The recognition module is used to use the probe life deep learning model to recognize the time series features of the processed acupuncture needle mark image to obtain the current usage time of the probe.

[0026] In some embodiments, the probe station further comprises a probe life module;

[0027] The probe life module is used to obtain the remaining life of the probe under the current force control parameters according to the current use time of the probe and the life threshold of the probe under the current force control parameters.

[0028] In some embodiments, the probe station further includes a force control parameter adjustment module;

[0029] Among them, the force control parameter adjustment module is used to change the force control parameters of the probe, and use the image acquisition module, the data processing module and the identification module to obtain the remaining life of the probe under different force control parameters; then select the force control parameter under the maximum remaining life of the probe as the optimal force control parameter, and adjust the current force control parameter of the probe to the optimal force control parameter.

[0030] By means of the above technical solution, the probe life detection method and the probe station of the present invention have at least the following beneficial effects:

[0031] The present invention uses the needle mark image to monitor the probe life in real time and adjust the force control parameters to maintain the optimal performance of the probe. This method not only improves the use efficiency of the probe, but also extends its life and effectively improves production efficiency. Compared with the traditional probe life prediction method, this method improves the accuracy and real-time performance of the probe life detection.

[0032] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0034] Figure 1 It is a flowchart of a probe life detection method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0037] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0038] like Figure 1 As shown, an embodiment of the present invention proposes a probe life detection method, which includes training a probe life deep learning model and using the trained probe life deep learning model.

[0039] Among them, the specific training method of the probe life deep learning model includes:

[0040] Step S11: collecting needle mark data at different time sequences during the entire life cycle of the probe under different force control parameters.

[0041] Among them, during the whole life cycle of the probe from brand new to scrapped, the needle marks of the probe under the same force control parameters are different. The needle marks of the same probe under different force control parameters are also different. In step S11, the needle mark data of the probe at different usage times during the whole life cycle under different force control parameters need to be collected. Among them, when the probe is in a brand new state, its usage time is 0. When the probe is in a scrapped state, its usage time reaches the maximum value, and the usage time of the probe at this time is the life threshold of the probe.

[0042] It should be noted here that the needle mark data of the probe at different time sequences are the needle mark data of the probe at different use times.

[0043] In step S11 , the force control parameters (speed, acceleration, smoothing time, etc.) of multiple probes in the machine can be recorded and the time-series needle mark images of the probes' full life cycle under the force control parameters can be collected.

[0044] Step S12: Processing the needle mark data of each acupuncture needle to highlight the time sequence characteristics of the needle mark.

[0045] In a specific application example, the processing of each needle mark data in the above step S12 to highlight the time sequence characteristics of the needle marks is specifically as follows:

[0046] The needle mark data of each acupuncture needle were subjected to needle mark edge extraction, illumination homogenization, filtering and noise removal, and Fourier transformation in sequence to highlight the temporal characteristics of the needle marks.

[0047] Step S13: putting the processed needle mark data of each needle puncture into the model for training to obtain a probe life deep learning model.

[0048] In a specific application example, in the above step S13, the processed needle mark data of each needle puncture is put into the model for training, specifically: the processed needle mark data of each needle puncture is put into the D-CNN-LSTM model for training. The model can extract the time series information in the needle mark image from the data and optimize the force control parameters to obtain the model weight file and the life threshold of the probe.

[0049] Among them, using the trained probe life deep learning model specifically includes:

[0050] Step S21: collecting an image of the needle mark of the probe under the current force control parameters.

[0051] Specifically, the probe station reads the coordinate information of the pad before the needle is inserted, and then moves to the corresponding position to collect the needle mark image.

[0052] Step S22: Processing the acupuncture needle mark image to highlight the temporal characteristics of the needle mark.

[0053] In a specific application example, in the above step S22, the acupuncture needle mark image is processed to highlight the temporal characteristics of the needle mark, specifically:

[0054] The data of each needle mark are processed in sequence by illumination homogenization, dynamic threshold segmentation, pad extraction and grayscale value scaling to highlight the timing characteristics of the needle marks.

[0055] Among them, the timing characteristics of the needle mark can reflect the current force control parameters of the probe and the use time of the probe under the current force control parameters.

[0056] Step S23: The probe life deep learning model identifies the temporal features of the processed pin mark image of the needle insertion to obtain the current usage time of the probe.

[0057] In some embodiments, using the trained probe life deep learning model further includes:

[0058] Step S24: Based on the current usage time of the probe and the life threshold under the current force control parameter of the probe, obtain the remaining life of the probe under the current force control parameter.

[0059] In the above example, after knowing the current usage time of the probe, according to the life threshold of the probe under the current force control parameter, subtract the current usage time of the probe from the life threshold of the probe under the current force control parameter to know the remaining life of the probe under the current force control parameter.

[0060] Among them, compared with the traditional probe life detection method, the present invention uses a probe life deep learning model to detect the life of the probe, with higher accuracy. In addition, real-time detection of the probe life can also be achieved, with higher real-time performance.

[0061] The foregoing probe life detection method may further include:

[0062] Step S25: Change the force control parameter of the probe, and repeat the above steps S21 to S24 to obtain the remaining life of the probe under different force control parameters.

[0063] Step S26: Select the force control parameter corresponding to the maximum remaining life of the probe as the optimal force control parameter, and adjust the current force control parameter of the probe to the above optimal force control parameter.

[0064] In the above example, by adjusting the force control parameter of the probe, the best performance of the probe can be maintained. This method not only improves the usage efficiency of the probe, but also extends its life, and effectively improves the production efficiency.

[0065] In a specific application example, the force control parameter of the foregoing probe includes at least one of the running speed of the probe, the running acceleration of the probe, and the running smoothing time of the probe.

[0066] The present invention also provides a probe station, which may include an image acquisition module, a data processing module, and an identification module. Among them, the image acquisition module is used to acquire the pin mark image of the needle insertion of the probe under the current force control parameter. The data processing module is used to process the pin mark image to highlight the temporal features of the pin mark. The identification module is used to use the probe life deep learning model to identify the temporal features of the processed pin mark image of the needle insertion to obtain the current usage time of the probe.

[0067] In some embodiments, the aforementioned probe station further includes a probe life module, which is used to obtain the remaining life of the probe under the current force control parameters according to the current use time of the probe and the life threshold of the probe under the current force control parameters.

[0068] In some embodiments, the probe station further includes a force control parameter adjustment module. The force control parameter adjustment module is used to change the force control parameter of the probe, and obtain the remaining life of the probe under different force control parameters by using the image acquisition module, the data processing module and the recognition module; then the force control parameter under the maximum remaining life of the probe is selected as the optimal force control parameter, and the current force control parameter of the probe is adjusted to the optimal force control parameter.

[0069] Among them, the present invention uses the needle mark image to monitor the probe life in real time and adjust the force control parameters to maintain the optimal performance of the probe. This method not only improves the use efficiency of the probe, but also extends its life and effectively improves production efficiency. Compared with the traditional probe life prediction method, this method improves the accuracy and real-time performance of the probe life detection.

[0070] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A probe life detection method, characterized in that: include: Train a deep learning model for probe lifetime; The specific training method includes: step S11: collecting needle mark data at different time sequences during the entire life cycle of the probe under different force control parameters; step S12: processing each needle mark data to highlight the time sequence characteristics of the needle mark; step S13: putting the processed needle mark data into the model for training to obtain a probe life deep learning model; In step S13, the processed data of each acupuncture mark are put into the model for training, specifically: the processed data of each acupuncture mark are put into the D-CNN-LSTM model for training; Use the trained probe life deep learning model; specifically: Step S21: collecting an image of the needle mark of the probe under the current force control parameters; Step S22: processing the acupuncture needle mark image to highlight the time sequence characteristics of the needle mark; Step S23: The probe life deep learning model identifies the time series features of the processed acupuncture needle mark image to obtain the current usage time of the probe.

2. The probe life detection method according to claim 1, characterized in that: The use of the trained probe life deep learning model also includes: Step S24: Obtain the remaining life of the probe under the current force control parameters according to the current use time of the probe and the life threshold of the probe under the current force control parameters.

3. The probe life detection method according to claim 2, characterized in that: Also includes: Step S25: changing the force control parameters of the probe, repeating the above steps S21 to S24, and obtaining the remaining life of the probe under different force control parameters; Step S26: selecting the force control parameter under the maximum remaining life of the probe as the optimal force control parameter, and adjusting the current force control parameter of the probe to the optimal force control parameter.

4. The probe life detection method according to any one of claims 1 to 3, characterized in that: The processing of the acupuncture needle mark data in step S12 to highlight the time sequence characteristics of the needle marks is specifically as follows: The needle mark data are subjected to needle mark edge extraction, illumination homogenization processing, filtering and noise removal, and Fourier transformation in sequence to highlight the temporal characteristics of the needle marks.

5. The probe life detection method according to any one of claims 1 to 3, characterized in that: In step S22, the acupuncture needle mark image is processed to highlight the temporal characteristics of the needle mark, specifically: The needle mark data are subjected to illumination uniformization processing, dynamic threshold segmentation, pad extraction and grayscale value scaling in sequence to highlight the timing characteristics of the needle marks.

6. The probe life detection method according to any one of claims 1 to 3, characterized in that: The force control parameter of the probe includes at least one of the running speed of the probe, the running acceleration of the probe, and the running smoothing time of the probe.

7. A probe station, characterized in that: It includes an image acquisition module, a data processing module and a recognition module; Wherein, the image acquisition module is used to acquire the needle mark image of the probe under the current force control parameters; The data processing module is used to process the acupuncture needle mark image to highlight the time sequence characteristics of the needle mark; The recognition module is used to use the probe life deep learning model to recognize the time series characteristics of the processed acupuncture needle mark image to obtain the current use time of the probe; The specific training method of the probe life deep learning model includes: step S11: collecting needle mark data at different time sequences during the entire life cycle of the probe under different force control parameters; step S12: processing each of the needle mark data to highlight the time sequence characteristics of the needle mark; step S13: putting the processed needle mark data into the model for training to obtain the probe life deep learning model; In step S13, the processed needle mark data of each acupuncture needle are put into the model for training, specifically: the processed needle mark data of each acupuncture needle are put into the D-CNN-LSTM model for training.

8. The probe station according to claim 7, characterized in that, Also includes a probe life module; The probe life module is used to obtain the remaining life of the probe under the current force control parameters according to the current use time of the probe and the life threshold of the probe under the current force control parameters.

9. The probe station according to claim 7, characterized in that: It also includes a force control parameter adjustment module; Among them, the force control parameter adjustment module is used to change the force control parameters of the probe, and use the image acquisition module, the data processing module and the identification module to obtain the remaining life of the probe under different force control parameters; then select the force control parameter under the maximum remaining life of the probe as the optimal force control parameter, and adjust the current force control parameter of the probe to the optimal force control parameter.

Images