阅读说明：本技术 介质滤波器的自动调谐系统、自动调谐方法及存储装置 (Automatic tuning system, automatic tuning method and storage device of dielectric filter ) 是由 金粲 于 2020-12-14 设计创作，主要内容包括：本申请提供了一种介质滤波器的自动调谐系统、自动调谐方法及存储装置,所述自动调谐系统包括：信号采集装置,用于采集当前介质滤波器的当前散射参数；图像获取设备,用于获得介质滤波器的图像数据,并将图像数据传输至控制装置；控制装置,用于接收信号采集装置采集的当前散射参数,若当前散射参数不满足预设要求,则接收图像获取设备传输的图像数据,并利用当前散射参数和训练后的学习模型获得打磨参数,以及根据打磨参数和图像数据得到打磨指令；打磨装置,与控制装置耦接,用于接收打磨指令,并根据打磨指令打磨当前介质滤波器。通过上述方式,本申请能够采用机器的方式实现自动调谐,以降低成本且提高生产效率。(The application provides an automatic tuning system, an automatic tuning method and a storage device of a dielectric filter, wherein the automatic tuning system comprises: the signal acquisition device is used for acquiring the current scattering parameters of the current dielectric filter; an image acquisition device for acquiring image data of the dielectric filter and transmitting the image data to the control apparatus; the control device is used for receiving the current scattering parameters acquired by the signal acquisition device, receiving image data transmitted by the image acquisition equipment if the current scattering parameters do not meet preset requirements, acquiring polishing parameters by using the current scattering parameters and the trained learning model, and acquiring polishing instructions according to the polishing parameters and the image data; and the polishing device is coupled with the control device and used for receiving the polishing instruction and polishing the current dielectric filter according to the polishing instruction. By the mode, automatic tuning can be achieved by adopting a machine mode, so that cost is reduced, and production efficiency is improved.)

1. An automatic tuning system for a dielectric filter, comprising:

the signal acquisition device is used for acquiring the current scattering parameters of the current dielectric filter;

the image acquisition equipment is used for acquiring the image data of the dielectric filter and transmitting the image data to the control device;

the control device is coupled with the signal acquisition device and the image acquisition equipment and used for receiving the current scattering parameters acquired by the signal acquisition device, receiving image data transmitted by the image acquisition equipment if the current scattering parameters do not meet preset requirements, obtaining polishing parameters by using the current scattering parameters and a trained learning model, and obtaining a polishing instruction according to the polishing parameters and the image data, wherein the polishing parameters comprise polishing positions and polishing amounts on the current dielectric filter, and the polishing instruction comprises polishing tracks and polishing amounts;

and the polishing device is coupled with the control device and used for receiving the polishing instruction and polishing the current dielectric filter according to the polishing instruction.

2. The automatic tuning system of claim 1, wherein the control means comprises:

the data processor is coupled with the signal acquisition device and the image acquisition equipment and is used for receiving the current scattering parameters acquired by the signal acquisition device, receiving image data transmitted by the image acquisition equipment when the current scattering parameters do not meet preset requirements, obtaining polishing parameters by using the current scattering parameters and a trained learning model, and obtaining initial spatial positions of polishing point positions according to the polishing parameters and the image data;

and the motion controller is coupled with the data processor and used for planning a track according to the initial spatial position of each polishing point and the polishing position on the current dielectric filter to obtain a polishing track.

3. The automatic tuning system of claim 1, wherein the current dielectric filter comprises a plurality of coupling elements, and wherein the data processor, when obtaining the polishing parameters using the current scattering parameters and the trained learning model, is specifically configured to:

obtaining a current coupling matrix according to the current scattering parameters;

comparing the current coupling matrix with a theoretical coupling matrix to obtain the detuning amount of each coupling unit;

and inputting the detuning amount of each coupling unit into the trained learning model, and calculating to obtain the polishing position and polishing amount on the current dielectric filter.

4. The automatic tuning system of claim 3, wherein prior to inputting the detuning amount of each of the coupling elements into the trained learning model, the data processor is further configured to:

judging whether the detuning amount of each coupling unit is smaller than a preset threshold value;

if the detuning amount of at least one coupling unit is smaller than a preset threshold value, stopping automatically tuning the dielectric filter;

otherwise, entering the step of inputting the detuning quantity of each coupling unit into the trained learning model.

5. The automatic tuning system of claim 2, wherein the sharpening device comprises a positioning mechanism and a sharpening mechanism,

the positioning mechanism is coupled with the motion controller and used for receiving the polishing track and controlling the polishing mechanism to move according to the polishing track;

the polishing mechanism is coupled with the data processor and the positioning mechanism and used for following the polishing track to move and polishing the current dielectric filter according to the polishing amount.

6. The automatic tuning system of any one of claims 1-5,

the signal acquisition device acquires the scattering parameters of the current dielectric filter after being polished each time in real time and transmits the acquired scattering parameters to the control device;

the control device is also used for training the learning model according to the scattering parameters transmitted by the signal acquisition device and the grinding amount of the scattering parameters acquired each time correspondingly.

7. A method for automatically tuning a dielectric filter, comprising:

obtaining a current scattering parameter and image data of a current dielectric filter;

judging whether the current scattering parameter meets a preset requirement or not;

if not, obtaining polishing parameters by using the current scattering parameters and the trained learning model, wherein the polishing parameters comprise polishing positions and polishing amounts on the current dielectric filter; obtaining a polishing instruction according to the polishing parameter and the image data, wherein the polishing instruction comprises a polishing track and the polishing amount; issuing the polishing instruction to enable a polishing device to polish the current dielectric filter according to the polishing instruction;

and if so, completing the automatic tuning process of the dielectric filter.

8. The auto-tuning method of claim 7, wherein the step of deriving a buffing instruction from the buffing parameters and the image data comprises:

obtaining an initial spatial position of each polishing point location according to the polishing parameters and the image data;

and planning a track according to the initial space position of each polishing point and the polishing position on the current dielectric filter to obtain a polishing track.

9. The auto-tuning method of claim 7, wherein the dielectric filter comprises a plurality of coupling elements, and the step of obtaining the polishing parameters using the current scattering parameters and the trained learning model comprises:

obtaining a current coupling matrix according to the current scattering parameters;

comparing the current coupling matrix with a theoretical coupling matrix to obtain the detuning amount of each coupling unit;

and inputting the detuning amount of each coupling unit into the trained learning model, and calculating to obtain the polishing position and polishing amount on the current dielectric filter.

10. The automatic tuning method according to claim 9, wherein the step of inputting the detuning amount of each coupling element into the trained learning model further comprises:

judging whether the detuning amount of each coupling unit is smaller than a preset threshold value;

if the detuning amount of at least one coupling unit is smaller than a preset threshold value, stopping automatically tuning the dielectric filter;

otherwise, entering the step of inputting the detuning quantity of each coupling unit into the trained learning model.

11. The auto-tuning method of claim 9, wherein after the auto-tuning process of the dielectric filter is completed, further comprising:

updating a training set according to the scattering parameters of the dielectric filter before and after polishing each time and the polishing parameters of the corresponding scattering parameters acquired each time;

and training the learning model by using the updated training set.

12. An apparatus having storage functionality, characterized in that program instructions are stored which can be executed to implement the steps in the auto-tuning method according to any of claims 7-11.

Technical Field

The application belongs to the technical field of dielectric filters, and particularly relates to an automatic tuning system, an automatic tuning method and a storage device of a dielectric filter.

Background

In the 5G era, dielectric filters are becoming smaller and lighter. The dielectric filter is expected to become a market mainstream scheme in the 5G era due to the characteristics of small volume, large Q value, low insertion loss, good stability, high bearing power and the like. The dielectric filter generally uses ceramic powder as raw materials, and in the production process, because the performance of the dielectric filter is difficult to meet the specification requirement due to errors in dry pressing of the ceramic powder, processing of a dielectric substrate and outer layer electroplating, tuning is carried out, namely, an outer coating of the dielectric filter is ground, so that the performance of the dielectric filter meets the index requirement.

Compared with the traditional cavity dielectric filter which is tuned by adjusting the resonant rod, the tuning of the dielectric filter can be realized only by grinding. Due to the irreversibility of ceramic grinding, when the dielectric filter is over-adjusted, the dielectric filter is basically directly scrapped.

At present, the dielectric filter tuning is mainly to polish the specified position of the plating layer of the dielectric filter by workers according to an auxiliary tuning algorithm. When tuning work is carried out, workers need to control the polishing amount according to experience, and considerable time is needed for accumulating the experience, so that the labor cost of old workers is high, the training time of new workers is long, and the difficulty in mass production of the dielectric filter is high.

Disclosure of Invention

The application provides an automatic tuning system, an automatic tuning method and a storage device of a dielectric filter, which can realize automatic tuning, reduce tuning cost and improve production efficiency.

In order to solve the technical problem, the application adopts a technical scheme that: there is provided an automatic tuning system of a dielectric filter, comprising: the signal acquisition device is used for acquiring the current scattering parameters of the current dielectric filter; the image acquisition equipment is used for acquiring the image data of the dielectric filter and transmitting the image data to the control device; the control device is coupled with the signal acquisition device and the image acquisition equipment and used for receiving the current scattering parameters acquired by the signal acquisition device, receiving image data transmitted by the image acquisition equipment if the current scattering parameters do not meet preset requirements, obtaining polishing parameters by using the current scattering parameters and a trained learning model, and obtaining a polishing instruction according to the polishing parameters and the image data, wherein the polishing parameters comprise polishing positions and polishing amounts on the current dielectric filter, and the polishing instruction comprises polishing tracks and polishing amounts; and the polishing device is coupled with the control device and used for receiving the polishing instruction and polishing the current dielectric filter according to the polishing instruction.

In order to solve the above technical problem, another technical solution adopted by the present application is: there is provided an automatic tuning method of a dielectric filter, comprising: obtaining a current scattering parameter and image data of a current dielectric filter; judging whether the current scattering parameter meets a preset requirement or not; if not, obtaining polishing parameters by using the current scattering parameters and the trained learning model, wherein the polishing parameters comprise polishing positions and polishing amounts on the current dielectric filter; obtaining a polishing instruction according to the polishing parameter and the image data, wherein the polishing instruction comprises a polishing track and the polishing amount; issuing the polishing instruction to enable a polishing device to polish the current dielectric filter according to the polishing instruction; and if so, completing the automatic tuning process of the dielectric filter.

In order to solve the above technical problem, another technical solution adopted by the present application is: there is provided a device having a memory function, storing program instructions executable to implement the steps in the auto-tuning method of any of the above embodiments.

Being different from the prior art situation, the beneficial effect of this application is: the application can predict the current polishing parameters of the current scattering parameters of the corresponding dielectric filter by utilizing the trained learning model, and the subsequent polishing device can automatically polish according to the polishing instructions corresponding to the current polishing parameters. The method can realize the mechanical automation of the tuning process of the dielectric filter, and reduce the production cost; compared with manual polishing, the polishing machine can work for 24 hours, so that the yield is improved, and the productivity is increased; and the dielectric filter processed by the automatic system has better consistency, and the standardized control of the dielectric filter is convenient.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:

fig. 1 is a schematic structural diagram of an embodiment of an automatic tuning system of a dielectric filter according to the present application;

FIG. 2 is a schematic flow chart diagram illustrating an embodiment of a method for automatically tuning a dielectric filter according to the present application;

FIG. 3 is a flowchart illustrating an embodiment corresponding to step S105 in FIG. 2;

FIG. 4 is a flowchart illustrating an embodiment of the auto-tuning method after step S103 in FIG. 2;

FIG. 5 is a flowchart illustrating an embodiment corresponding to step S301 in FIG. 4;

FIG. 6 is a schematic flow chart illustrating another embodiment corresponding to step S301 in FIG. 4;

fig. 7 is a schematic diagram of a framework of an embodiment of the apparatus with storage function according to the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Referring to fig. 1, fig. 1 is a schematic structural diagram of an embodiment of an automatic tuning system of a dielectric filter according to the present invention. The auto-tuning system includes a signal acquisition device 30, an image acquisition device 32, a control device 34, and a polishing device 36.

In particular, the information collecting device 30 is used for collecting the current scattering parameter of the current dielectric filter, specifically, the scattering parameter, i.e. the S parameter, is an important parameter in the microwave transmission, which describes the frequency domain characteristics of the transmission channel, such as the reflection, crosstalk, loss, etc. of the signal. In this embodiment, the signal acquisition device 30 may include a vector network analyzer, and the vector network analyzer may be used to easily acquire and obtain the current scattering parameters.

The image acquisition device 32 is used for acquiring image data of the dielectric filter and transmitting the image data to the control device 34; in the present embodiment, the image acquisition device 32 includes a camera.

The control device 34 is coupled to the signal acquisition device 30 and the image acquisition device 32, and configured to receive the current scattering parameter acquired by the signal acquisition device 30, receive image data transmitted by the image acquisition device 32 if the current scattering parameter does not meet a preset requirement, obtain a polishing parameter by using the current scattering parameter and a trained learning model, and obtain a polishing instruction according to the polishing parameter and the image data, where the polishing parameter includes a polishing position and a polishing amount on a current dielectric filter, and the polishing instruction includes a polishing track and a polishing amount. Generally speaking, a plurality of blind holes are arranged on one side surface of the dielectric filter, the blind holes are preset polishing positions, and the polishing amount is the polishing depth. The grinding track can be a grinding track planned by the grinding mechanism moving from the initial spatial position of the grinding point to the grinding position on the dielectric filter and further grinding a certain depth at the grinding position.

In this embodiment, the control device 34 may be differentiated according to functions, and includes a data processor 340 and a motion controller 342 coupled to each other; the data processor 340 is coupled to the signal acquisition device 30 and the image acquisition device 32, and configured to receive a current scattering parameter acquired by the signal acquisition device 30, and if the current scattering parameter does not meet a preset requirement (for example, the preset requirement includes a range of the scattering parameter), receive image data transmitted by the image acquisition device 32, obtain a polishing parameter by using the current scattering parameter and a trained learning model, and obtain an initial spatial position of each polishing point according to the polishing parameter and the image data; for example, the initial spatial position may be a position of each polishing point on the current dielectric filter under a world coordinate system before the current polishing; before formal polishing, the polishing mechanism can move to the polishing point from the initial position of the polishing mechanism, and then a polishing process is carried out from the polishing point; the motion controller 342 is configured to perform trajectory planning according to the initial spatial position of each polishing point and the polishing position on the current dielectric filter, so as to obtain a polishing trajectory; for example, the trajectory planning may be performed by an interpolation algorithm or the like. This design makes the function of the control device 34 clearer, and improves the processing efficiency of the system.

In addition, in the present embodiment, the control device 34 is further configured to stop the auto-tuning system when the current scattering parameter meets a preset requirement. By means of the mode, the polishing efficiency can be effectively improved, and the polishing cost is reduced.

In addition, in order to improve the accuracy of the learning model, the signal acquisition device 30 may also acquire the scattering parameters of the current dielectric filter after each polishing, and transmit the acquired scattering parameters to the control device 34; the control device 34 is further configured to train a learning model according to the scattering parameters transmitted by the signal acquisition device 30 and the polishing amount of the scattering parameters corresponding to each acquisition. And the process of specifically training the learning model will be described in detail later.

The polishing device 36 is coupled to the control device 34 for receiving the polishing command and polishing the present dielectric filter according to the polishing command. In this embodiment, the grinding device 36 includes a positioning mechanism and a grinding mechanism. Wherein, the positioning mechanism is coupled with the motion controller 342, and is used for receiving the polishing track and controlling the motion of the polishing mechanism according to the polishing track; the polishing mechanism is coupled to the data processor 340 and the positioning mechanism, and includes a polishing head or a laser for following the polishing track and polishing the current dielectric filter according to the polishing amount. Preferably, the polishing head, laser, etc. of the polishing mechanism may be located right above the polishing position of the dielectric filter at the time of polishing. The positioning structure may include a robot and a plurality of clamps coupled to the robot; the clamp is used for clamping a polishing head or a laser, the robot is used for moving according to a polishing track, and the clamp is driven by the robot to move. Of course, in other embodiments, the positioning mechanism may be configured to receive the polishing track and control the current motion of the dielectric filter according to the polishing track, and the position of the polishing mechanism may be fixed.

Referring again to fig. 1, the auto-tuning system provided by the present application may further include a memory 38, coupled to the control device 34, and storing therein at least one relevant parameter in the learning model, where the parameter may be a weight corresponding to each node in the learning model; wherein one dielectric filter type corresponds to one learning model. The control device 34 can obtain the trained learning model by directly calling the relevant parameters stored in the memory 38.

Of course, the memory 38 may also be used to store scattering parameters of the dielectric filter before and after each polishing and the corresponding polishing amount; the control device 34 is also used to update the learning model with the data stored in the memory after the polishing of the dielectric filter is completed.

Through the system, the current polishing parameters of the current scattering parameters of the corresponding dielectric filter can be obtained by the trained learning model, and the subsequent polishing device can automatically polish according to the polishing instructions corresponding to the current polishing parameters. The method can realize the mechanical automation of the tuning process of the dielectric filter, and reduce the production cost; compared with manual polishing, the polishing machine can work for 24 hours, so that the yield is improved, and the productivity is increased; and the dielectric filter processed by the automatic system has better consistency, and the standardized control of the dielectric filter is convenient.

The automatic tuning method of the self-tuning system is further explained below from the perspective of the method. Referring to fig. 2, fig. 2 is a schematic flow chart of an embodiment of an automatic tuning method for a dielectric filter according to the present application, where an execution subject of the automatic tuning method may be a control device in the automatic tuning system, and the automatic tuning method specifically includes:

s101: and obtaining the current scattering parameters of the current dielectric filter and image data.

The specific implementation manner of the step S101 may be: obtaining a current scattering parameter of a current dielectric filter measured by a vector network analyzer; obtaining an image of one side of the dielectric filter, which is acquired by the image acquisition equipment and provided with a plurality of blind holes; for example, the image pickup apparatus picks up an image from directly above the side of the dielectric filter provided with the plurality of blind holes. The plurality of blind holes can be subsequently polished.

S102: and judging whether the current scattering parameters meet preset requirements or not.

Specifically, the method before step S102 may further include: preset requirements, i.e. tuning requirements, corresponding to the type of the present dielectric filter are obtained, which preset requirements may include a range of scattering parameters. It should be noted that the types of the dielectric filters include differences in dimension, that is, two dielectric filters that are three coupling units, and even if the two dielectric filters are different only in size, the two dielectric filters are of different types, and the requirements of the different types of dielectric filters on scattering parameters are different.

S103: and if the preset requirement is met, completing the automatic tuning process of the dielectric filter.

S104: and if the preset requirements are not met, obtaining polishing parameters by using the current scattering parameters and the trained learning model, wherein the polishing parameters comprise the polishing position and the polishing amount on the current dielectric filter.

Specifically, the learning model may be a CNN neural network model, an RNN neural network model, or the like, and different types of the dielectric filter may correspond to different learning models. Of course, in other embodiments, the algorithm used by the learning model may also be a conventional machine learning algorithm, such as an XGBoost algorithm or a random forest algorithm, which can also perform better learning. In this embodiment, the polishing parameters may include a plurality of polishing positions and a plurality of polishing amounts, and the polishing positions and the polishing amounts correspond to each other one to one.

Generally, a dielectric filter includes a plurality of coupling units, e.g., 3, 5, etc. The specific implementation process of step S104 may be: and inputting the characteristic parameters of each coupling unit extracted from the current scattering parameters into the trained learning model to obtain the current polishing parameters corresponding to each coupling unit.

In an application scenario, the characteristic parameter may be a detuning amount, and the step S104 specifically includes: obtaining a current coupling matrix according to the current scattering parameters, wherein the specific process can be referred to in the prior art and is not detailed herein; comparing the current coupling matrix with the theoretical coupling matrix to obtain the detuning amount of each coupling unit; and inputting the detuning amount of each coupling unit into the trained learning model to obtain the current polishing parameter corresponding to each coupling unit, namely calculating to obtain the polishing position and polishing amount on the current dielectric filter. Generally speaking, the detuning amount and the polishing amount of the same coupling unit are not in a linear corresponding relationship, and different coupling units can influence each other, so that the detuning amount and the polishing amount are in a complex nonlinear relationship; by means of the good fitting characteristic of the learning model structure to the nonlinear relation, the accuracy of the calculated polishing parameters can be improved.

S105: and obtaining a polishing instruction according to the polishing parameter and the image data, wherein the polishing instruction comprises a polishing track and a polishing amount.

Specifically, referring to fig. 3, fig. 3 is a flowchart illustrating an embodiment corresponding to step S105 in fig. 2, where step S105 specifically includes:

s201: and obtaining the initial spatial position of each polishing point according to the polishing parameters and the image data.

Specifically, in this embodiment, the polishing point positions may be the polishing positions on the dielectric filter polished last time; for example, when the dielectric filter is polished at the last time, polishing operation is performed on a certain area at the bottom of a certain blind hole on the dielectric filter, and the polishing point position is a certain area at the bottom of the blind hole polished at the last time; or obtaining the position to be polished, namely the polishing point position, through image analysis according to the image data. The specific implementation process of the step S201 may be: obtaining image coordinates of each polishing point of the current dielectric filter on the image before polishing, and converting the image coordinates into intermediate coordinates under a camera coordinate system by utilizing a transformation relation between the image coordinate system and the camera coordinate system; and converting the intermediate coordinate into a space coordinate under the world coordinate system by utilizing a transformation relation between the camera coordinate system and the world coordinate system, wherein the space coordinate of each grinding point under the world coordinate system is the initial space position of each grinding point.

S202: and planning a track according to the initial space position of each polishing point and the polishing position on the current dielectric filter to obtain a polishing track.

Specifically, the polishing track may be a track in which the position of the dielectric filter is unchanged, and the polishing mechanism moves from an initial spatial position of the polishing point to the polishing position; alternatively, the polishing track may be a track in which the position of the polishing mechanism is unchanged and the dielectric filter moves to a predetermined position.

S203: and forming a grinding instruction by the grinding track and the grinding amount.

S106: and issuing a polishing instruction so that the polishing device polishes the current dielectric filter according to the polishing instruction.

By the method, the current polishing parameters of the corresponding dielectric filter can be predicted by using the trained learning model, and the subsequent polishing device can automatically polish according to the polishing instruction corresponding to the current polishing parameters. The method can realize the mechanical automation of the tuning process of the dielectric filter, and reduce the production cost; compared with manual polishing, the polishing machine can work for 24 hours, so that the yield is improved, and the productivity is increased; and the dielectric filter processed by the automatic system has better consistency, and the standardized control of the dielectric filter is convenient.

Further, before the step S104, the method further includes: judging whether the detuning amount of each coupling unit is smaller than a preset threshold value, wherein the threshold value can be a numerical value close to 0; if the detuning amount of at least one coupling unit is smaller than a preset threshold value, stopping automatically tuning the dielectric filter; otherwise, the process proceeds to step S104. Generally speaking, the detuning amount of each coupling unit should be gradually reduced in the polishing process, and if the current scattering parameter of the dielectric filter does not meet the preset requirement, but the detuning amount of at least one coupling unit is reduced to the threshold value, it indicates that the current dielectric filter is scrapped, and the tuning is not successful no matter how the dielectric filter is polished, and the subsequent polishing process of the dielectric filter is meaningless. Through increasing above-mentioned judgement step, can improve the efficiency of polishing, reduce the cost of polishing.

Further, as for the learning model, the obtaining process of the learning model utilized for the first time in the above steps S101 to S106 may be: obtaining characteristic parameters in scattering parameters of the dielectric filter in the process of manual successful polishing; forming an initial training set by using a plurality of groups of characteristic parameters and corresponding manual polishing parameters; and training the initial learning model by using the initial training set so as to obtain the trained learning model.

Further, in order to make the learning model more accurate, the learning model may be retrained again after the automatic tuning process of the dielectric filter is completed, please refer to fig. 4, and fig. 4 is a flowchart illustrating an embodiment of the automatic tuning method after step S103 in fig. 2. The automatic tuning method may further include:

s301: and updating the training set according to the scattering parameters of the dielectric filter before and after polishing each time and the polishing parameters corresponding to the scattering parameters acquired each time.

In an embodiment, please refer to fig. 5, fig. 5 is a flowchart illustrating an embodiment corresponding to step S301 in fig. 4, where step S301 specifically includes:

s401: and obtaining the detuning amount of each coupling unit in the dielectric filter before polishing each time, the detuning amount after the automatic tuning process is completed, and the polishing amount of each polishing.

S402: and taking the difference between the detuning amount of each coupling unit before each grinding and the detuning amount after the automatic tuning process is completed and the sum of all the grinding amounts from each grinding to the completion of the automatic tuning process as a set of training data.

S403: the training set is updated with the sets of training data.

For example, assuming that a dielectric filter to be debugged includes 3 coupling units, the polishing positions are all blind holes on the corresponding coupling units, and after polishing is successful, the detuning amount corresponding to each coupling unit is 1; as shown in table 1 below, table 1 is a table of data associated with a dielectric filter polishing process.

Table 1: related data table in grinding process of dielectric filter

Through the above data, the following two sets of training data can be obtained, wherein the input of the training data corresponding to the first grinding is: 10-1, 10-1 and 10-1, and the output is 2+3, 2+3 and 2+3, namely the input is 9, 9 and 9, and the output is 5, 5 and 5; the input of the training data corresponding to the second grinding is 5-1, 5-1 and 5-1, and the output is 3, 3 and 3, namely the input is 4, 4 and 4, and the output is 3, 3 and 3.

In another embodiment, referring to fig. 6, fig. 6 is a schematic flowchart illustrating another embodiment corresponding to step S301 in fig. 4, where step S301 specifically includes:

s501: and obtaining the detuning amount of each coupling unit in the dielectric filter before polishing and the polishing amount of each coupling unit in the dielectric filter.

S502: and taking the difference value of the detuning amount of each coupling unit before and after each grinding and the corresponding grinding amount of each grinding as a set of training data.

S503: the training set is updated with the sets of training data.

For example, still taking the data in table 1 as an example, the input of the training data corresponding to the first grinding is: 10-5, 10-5 and 10-5, the output is 2, 2 and 2, namely the input is 5, 5 and 5, and the output is 2, 2 and 2; the input of the training data corresponding to the second grinding is 5-1, 5-1 and 5-1, and the output is 3, 3 and 3, namely the input is 4, 4 and 4, and the output is 3, 3 and 3.

Generally, the detuning amount of the coupling element after polishing is smaller and smaller, but a failure of the polishing device may occur, which results in unsuccessful polishing and no reduction in the detuning amount of the coupling element after polishing. In this case, if the data is formed into training data, the parameters of the learning model may be affected. Therefore, before the step S502, the method may further include: judging whether the detuning amount of each coupling unit after polishing is smaller than the detuning amount before polishing; if not, the group data is removed. By the method, the accuracy of the learning model can be further improved.

S302: and training the learning model by using the updated training set.

Referring to fig. 7, fig. 7 is a schematic diagram of a framework of an embodiment of a device with a storage function according to the present application. The memory-enabled device 20 has stored thereon program instructions 200 that are capable of implementing all of the methods described above. The program instructions 200 may be stored in the storage device in the form of a software product, and include several instructions to enable a computer device (which may be a personal computer, a server, or a network device) or a processor (processor) to execute all or part of the steps of the method according to the embodiments of the present application. The aforementioned storage device includes: various media capable of storing program codes, such as a usb disk, a mobile hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, or terminal devices, such as a computer, a server, a mobile phone, and a tablet.

The above description is only an example of the present application and is not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings, or which are directly or indirectly applied to other related technical fields, are intended to be included within the scope of the present application.