阅读说明：本技术 Pcb走线参数设定装置、方法及存储介质 (PCB routing parameter setting device and method and storage medium ) 是由 马光辉 薛凯熏 林尚毅 于 2019-11-05 设计创作，主要内容包括：一种PCB走线参数设定方法、装置及计算机可读存储介质,所述方法包括：获取多个用于进行PCB走线设计的走线参数组合；将多个走线参数组合输入至预设PCB走线仿真软件得到与每一走线参数组合对应的应变参数组合；将多个走线参数组合及多个应变参数组合作为训练样本对预设网络模型进行训练得到第一预测模型；及对第一预测模型进行训练与测试得到阻抗参数的阻抗预测模型,当将多个走线参数组合输入至阻抗预测模型时,阻抗预测模型所预测得到的阻抗值与通过预设PCB走线仿真软件所计算得到的阻抗值的平均误差小于预设误差。本发明基于机器学习算法训练可得到学习PCB走线仿真软件的预测模型,可减少PCB布线设计时程,提高设计效率。(A PCB routing parameter setting method, a device and a computer readable storage medium are provided, the method comprises the following steps: acquiring a plurality of routing parameter combinations for PCB routing design; inputting a plurality of routing parameter combinations into preset PCB routing simulation software to obtain a strain parameter combination corresponding to each routing parameter combination; training a preset network model by taking a plurality of routing parameter combinations and a plurality of strain parameter combinations as training samples to obtain a first prediction model; and training and testing the first prediction model to obtain an impedance prediction model of the impedance parameters, wherein when the plurality of wiring parameters are combined and input to the impedance prediction model, the average error between the impedance value predicted by the impedance prediction model and the impedance value calculated by the preset PCB wiring simulation software is smaller than the preset error. The invention can obtain a prediction model for learning PCB wiring simulation software based on machine learning algorithm training, can reduce PCB wiring design time course, and improves design efficiency.)

1. A PCB routing parameter setting method is characterized by comprising the following steps:

acquiring a plurality of routing parameter combinations for PCB routing design;

inputting a plurality of routing parameter combinations into preset PCB routing simulation software to obtain a strain parameter combination corresponding to each routing parameter combination, wherein the strain parameter combination at least comprises an impedance parameter;

a plurality of routing parameter combinations and a plurality of strain parameter combinations are combined to form a training sample, and a preset network model is trained to obtain a first prediction model; and

training and testing the first prediction model to obtain an impedance prediction model of the impedance parameter;

when a plurality of wiring parameter combinations are input into the impedance prediction model, the average error between the impedance value predicted by the impedance prediction model and the impedance value calculated by the preset PCB wiring simulation software is smaller than a preset error.

2. The method of claim 1, wherein the strain parameter combination further comprises an insertion loss parameter, a far-end crosstalk parameter, and a near-end crosstalk parameter.

3. The method of claim 2, wherein the step of obtaining a plurality of trace parameter combinations for PCB trace design further comprises:

determining routing parameters influencing impedance parameters, intervention loss parameters, far-end crosstalk parameters and near-end crosstalk parameters of the PCB; and

and obtaining a plurality of routing parameter combinations based on the determined routing parameters and the design range which can be implemented by the factory side.

4. The method of claim 1, wherein the step of training a predetermined network model to obtain a first prediction model by combining a plurality of trace parameter combinations and a plurality of strain parameter combinations as training samples comprises:

combining a plurality of routing parameter combinations and a plurality of strain parameter combinations to form a plurality of training samples, wherein each training sample comprises a routing parameter combination and a strain parameter combination corresponding to the routing parameter combination;

randomly dividing a plurality of training samples into a training set and a testing set, wherein the data volume of the training set is larger than that of the testing set;

training the preset network model by using the training set and testing the intermediate model obtained by training by using the test set; and

and when the test result of the intermediate model meets the preset requirement, taking the intermediate model obtained by training as the first prediction model.

5. The method of claim 4, wherein the method further comprises:

when the test result of the intermediate model does not meet the preset requirement, adjusting the parameters of the preset network model;

retraining the adjusted preset network model by using the training set and testing the retrained intermediate model by using the testing set; and

and when the test result of the intermediate model obtained by retraining meets the preset requirement, taking the intermediate model obtained by retraining as the first prediction model.

6. The method of claim 5, wherein the predetermined network model is a neural network model, and the step of adjusting the parameters of the predetermined network model comprises:

adjusting the total number of layers and/or the number of neurons in each layer of the neural network model.

7. The method of claim 2, wherein the method further comprises:

training and testing the first prediction model to obtain a prediction model of the intervention loss parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the intervention loss parameter, the average error between an intervention loss value predicted by the prediction model of the intervention loss parameter and an intervention loss value calculated by preset PCB wiring simulation software is smaller than the preset error;

training and testing the first prediction model to obtain a prediction model of the far-end crosstalk parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the far-end crosstalk parameter, the average error between a far-end crosstalk value predicted by the prediction model of the far-end crosstalk parameter and a far-end crosstalk value calculated by preset PCB wiring simulation software is smaller than the preset error; and

and training and testing the first prediction model to obtain a prediction model of the near-end crosstalk parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the near-end crosstalk parameter, the average error between a near-end crosstalk value predicted by the prediction model of the near-end crosstalk parameter and a near-end crosstalk value calculated by preset PCB wiring simulation software is smaller than the preset error.

8. The method of claim 2, wherein the method further comprises:

training and testing the first prediction model to obtain a prediction model for predicting the intervention loss parameter, the far-end crosstalk parameter and the near-end crosstalk parameter;

when a plurality of wiring parameters are combined and input into the prediction model, the average errors of the intervention loss value, the far-end crosstalk value and the near-end crosstalk value predicted by the prediction model and the intervention loss value, the far-end crosstalk value and the near-end crosstalk value calculated by the preset PCB wiring simulation software are all smaller than the preset error.

9. A PCB trace parameter setting device, the device comprising a processor and a memory, the memory having a plurality of computer programs stored thereon, wherein the processor is configured to implement the steps of the PCB trace parameter setting method according to any one of claims 1 to 8 when executing the computer programs stored in the memory.

10. A computer-readable storage medium storing instructions executable by one or more processors to perform the steps of the PCB trace parameter setting method according to any one of claims 1 to 8.

Technical Field

The present invention relates to the field of Printed Circuit Board (PCB) wiring technology, and in particular, to a PCB trace parameter setting device and method, and a computer readable storage medium.

Background

With the rapid development of information products, along with the generation of a large number of signals and images, the instruction cycle requirement of a processor is improved, the high-speed circuit is widely applied, and the PCB wiring design is driven to be more precise. The traditional PCB design mode is obviously insufficient for solving the problems of connection, disconnection and short circuit, and a new PCB design method and a new PCB design process are developed. On a high-speed circuit board, because the pulse rise and fall are fast, it is particularly necessary to maintain the signal transmission stability through impedance control, and how to perform impedance design fast and accurately becomes one of the issues of great attention for engineers.

Before designing a PCB, a design engineer usually needs to consider more than ten kinds of impedance design requirements, and according to past experience, for each impedance design, enumerate several sets of wiring parameter combinations that have opportunities to achieve, perform simulation calculation by using IMLC software provided by Intel, calculate corresponding response variables of each set of control parameter combination, such as impedance, insertion loss (insertion loss), Far-end crossTalk (FEXT), Near-end crossTalk (Near-end crossTalk, NEXT), and the like, extract the wiring parameter combinations that satisfy the impedance design, and after comprehensively evaluating the combinations, select the optimal control parameter combination, and apply the optimal control parameter combination to a factory end to develop a product. However, in the above calculation process of IMLC software, it takes about 5 minutes to calculate each set of control parameter combination, so that conservative estimation takes at least several hours to complete the design and optimization of a PCB. And for large-scale communication products, such as servers, the server is composed of a plurality of circuit boards with different sizes, and the design and the scheme fixing can be completed within a plurality of days, so that the cost of labor and time is high.

Disclosure of Invention

In view of the above, there is a need to provide a PCB trace parameter setting device, method and computer readable storage medium, which can obtain a prediction model for learning PCB trace simulation software based on machine learning algorithm training, reduce PCB wiring design time and improve design efficiency.

An embodiment of the present invention provides a method for setting PCB routing parameters, including:

acquiring a plurality of routing parameter combinations for PCB routing design;

inputting a plurality of routing parameter combinations into preset PCB routing simulation software to obtain a strain parameter combination corresponding to each routing parameter combination, wherein the strain parameter combination at least comprises an impedance parameter;

a plurality of routing parameter combinations and a plurality of strain parameter combinations are combined to form a training sample, and a preset network model is trained to obtain a first prediction model; and

training and testing the first prediction model to obtain an impedance prediction model of the impedance parameter;

when a plurality of wiring parameter combinations are input into the impedance prediction model, the average error between the impedance value predicted by the impedance prediction model and the impedance value calculated by the preset PCB wiring simulation software is smaller than a preset error.

Preferably, the strain parameter combination further comprises an insertion loss parameter, a far-end crosstalk parameter and a near-end crosstalk parameter.

Preferably, the step of obtaining a plurality of trace parameter combinations for PCB trace design further includes:

determining routing parameters influencing impedance parameters, intervention loss parameters, far-end crosstalk parameters and near-end crosstalk parameters of the PCB; and

and obtaining a plurality of routing parameter combinations based on the determined routing parameters and the design range which can be implemented by the factory side.

Preferably, the step of training a preset network model by using a plurality of routing parameter combinations and a plurality of strain parameter combinations as training samples to obtain a first prediction model includes:

combining a plurality of routing parameter combinations and a plurality of strain parameter combinations to form a plurality of training samples, wherein each training sample comprises a routing parameter combination and a strain parameter combination corresponding to the routing parameter combination;

randomly dividing a plurality of training samples into a training set and a testing set, wherein the data volume of the training set is larger than that of the testing set;

training the preset network model by using the training set and testing the intermediate model obtained by training by using the test set; and

and when the test result of the intermediate model meets the preset requirement, taking the intermediate model obtained by training as the first prediction model.

Preferably, the method further comprises:

when the test result of the intermediate model does not meet the preset requirement, adjusting the parameters of the preset network model;

retraining the adjusted preset network model by using the training set and testing the retrained intermediate model by using the testing set; and

and when the test result of the intermediate model obtained by retraining meets the preset requirement, taking the intermediate model obtained by retraining as the first prediction model.

Preferably, the preset network model is a neural network model, and the step of adjusting the parameters of the preset network model includes:

adjusting the total number of layers and/or the number of neurons in each layer of the neural network model.

Preferably, the method further comprises:

training and testing the first prediction model to obtain a prediction model of the intervention loss parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the intervention loss parameter, the average error between an intervention loss value predicted by the prediction model of the intervention loss parameter and an intervention loss value calculated by preset PCB wiring simulation software is smaller than the preset error;

training and testing the first prediction model to obtain a prediction model of the far-end crosstalk parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the far-end crosstalk parameter, the average error between a far-end crosstalk value predicted by the prediction model of the far-end crosstalk parameter and a far-end crosstalk value calculated by preset PCB wiring simulation software is smaller than the preset error; and

and training and testing the first prediction model to obtain a prediction model of the near-end crosstalk parameter, wherein when a plurality of wiring parameters are combined and input into the prediction model of the near-end crosstalk parameter, the average error between a near-end crosstalk value predicted by the prediction model of the near-end crosstalk parameter and a near-end crosstalk value calculated by preset PCB wiring simulation software is smaller than the preset error.

Preferably, the method further comprises:

training and testing the first prediction model to obtain a prediction model for predicting the intervention loss parameter, the far-end crosstalk parameter and the near-end crosstalk parameter;

when a plurality of wiring parameters are combined and input into the prediction model, the average errors of the intervention loss value, the far-end crosstalk value and the near-end crosstalk value predicted by the prediction model and the intervention loss value, the far-end crosstalk value and the near-end crosstalk value calculated by the preset PCB wiring simulation software are all smaller than the preset error.

An embodiment of the present invention provides a PCB trace parameter setting device, which includes a processor and a memory, wherein the memory stores a plurality of computer programs, and the processor is configured to implement the steps of the PCB trace parameter setting method when executing the computer programs stored in the memory.

An embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a plurality of instructions, and the instructions are executable by one or more processors to implement the steps of the PCB trace parameter setting method.

Compared with the prior art, the PCB wiring parameter setting device, the PCB wiring parameter setting method and the computer readable storage medium have the advantages that the prediction model for learning the PCB wiring simulation software is obtained based on the machine learning algorithm training, the strain parameter combination corresponding to the wiring parameter combination can be obtained through calculation through the prediction model, the calculation speed is higher compared with the PCB wiring simulation software, the PCB wiring design time course can be reduced, and the design efficiency is improved.

Drawings

Fig. 1 is a functional block diagram of a PCB trace parameter setting device according to an embodiment of the invention.

Fig. 2 is a functional block diagram of a PCB trace parameter setting procedure according to an embodiment of the invention.

Fig. 3 is a flowchart of a PCB trace parameter setting method according to an embodiment of the invention.

Description of the main elements

Memory device	10
		Processor with a memory having a plurality of memory cells	20
PCB wiring parameter setting program	30
		Acquisition module	101
Computing module	102
		First training module	103
Second training module	104
		PCB wiring parameter setting device	100

The following detailed description will further illustrate the invention in conjunction with the above-described figures.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. 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 invention.

It is further noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

Fig. 1 is a schematic diagram of a PCB trace parameter setting device according to a preferred embodiment of the present invention.

In one embodiment, since the pulse is rapidly lifted and lowered on the high-speed circuit board, the stability of signal transmission can be maintained by impedance control to minimize the phenomenon of signal bounce, and how to rapidly and accurately perform impedance design becomes a problem to be faced by PCB wiring personnel. During PCB design, PCB wiring personnel usually need to consider more than ten impedance design requirements, and then enumerate a plurality of wiring parameter combinations which are organically achieved aiming at each impedance design according to past experience. When designing a PCB, parameters such as insertion loss, far-end crosstalk and near-end crosstalk may need to be considered, at this time, PCB wiring personnel need to estimate wiring parameter combinations which may achieve target impedance, target insertion loss, target far-end crosstalk and target near-end crosstalk, then the wiring parameter combinations are input into IMLC software provided by an Inter company for analog calculation to obtain impedance, insertion loss, far-end crosstalk and near-end crosstalk corresponding to each wiring parameter combination, after comprehensively evaluating the wiring parameter combinations, the optimal wiring parameter combination is selected and applied to a factory end for PCB development. The PCB routing parameter setting device 100 can establish an artificial intelligence model for learning an IMLC software calculation mode, thereby reducing PCB routing development time and improving design efficiency.

In one embodiment, the PCB trace parameter setting apparatus 100 includes a memory 10, a processor 20, and a PCB trace parameter setting program 30 stored in the memory 10 and executable on the processor 20. When the processor 20 executes the PCB trace parameter setting program 30, steps in the PCB trace parameter setting method embodiment are implemented, for example, steps S300 to S306 shown in fig. 3. Alternatively, the processor 20 implements the functions of the modules in the PCB trace parameter setting program embodiment, such as the modules 101 to 104 in fig. 2, when executing the PCB trace parameter setting program 30.

The PCB trace parameterization program 30 may be partitioned into one or more modules that are stored in the memory 10 and executed by the processor 20 to accomplish the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used for describing the execution process of the PCB trace parameter setting program 30 in the PCB trace parameter setting apparatus 100. For example, the PCB trace parameter setting program 30 can be divided into the acquisition module 101, the calculation module 102, the first training module 103, and the second training module 104 in fig. 2. Specific functions of the modules refer to the functions of the modules in fig. 2 below.

Those skilled in the art will appreciate that the schematic diagram is only an example of the PCB trace parameter setting apparatus 100, and does not constitute a limitation of the PCB trace parameter setting apparatus 100, and may include more or less components than those shown in the drawings, or combine some components, or different components, for example, the PCB trace parameter setting apparatus 100 may further include a display device, a network access device, a bus, and the like.

The Processor 20 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor 20 may be any conventional processor or the like, and the processor 20 may be connected to the various parts of the PCB trace parameter setting device 100 using various interfaces and buses.

The memory 10 may be used to store the PCB trace parameter setting program 30 and/or the module, and the processor 20 implements various functions of the PCB trace parameter setting apparatus 100 by running or executing the computer program and/or the module stored in the memory 10 and calling the data stored in the memory 10. The memory 10 may include high speed random access memory and may also include non-volatile memory such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other non-volatile solid state storage device.

Fig. 2 is a functional block diagram of a PCB trace parameter setting procedure according to a preferred embodiment of the present invention.

Referring to fig. 2, the PCB trace parameter setting program 30 may include an obtaining module 101, a calculating module 102, a first training module 103, and a second training module 104. In one embodiment, the modules may be programmable software instructions stored in the memory 10 and called to be executed by the processor 20. It will be appreciated that in other embodiments, the modules may also be program instructions or firmware (firmware) that are resident in the processor 20.

The obtaining module 101 is configured to obtain a plurality of trace parameter combinations for PCB trace design.

In one embodiment, the trace parameter combination may be a combination including a plurality of strain parameters that may affect impedance, insertion loss, far-end crosstalk, near-end crosstalk, and the like of a PCB (not shown), and each trace parameter combination may include one or more trace parameters. The routing parameters can be line width, line-to-line width, insulation thickness, copper foil thickness, dielectric constant and the like.

In an embodiment, a routing parameter combination library may be established first, where the routing parameter combination library includes a plurality of routing parameter combinations, and the routing parameter combinations may be established in the following manner: the method includes the steps of setting up wiring parameters which affect the values of impedance, insertion loss, far-end crosstalk and near-end crosstalk, and then designing a plurality of sets of wiring parameter combinations by combining the wiring parameters and the design range which can be implemented by the factory end through an experimental design method, wherein the plurality of sets of wiring parameter combinations can contain the same or different wiring parameters. For example, the plurality of routing parameter combinations may include the same routing parameters, but different specific parameter values of the routing parameters. The obtaining module 101 may obtain a plurality of trace parameter combinations for PCB trace design from the trace parameter combination library.

The calculation module 102 is configured to input a plurality of trace parameter combinations to preset PCB trace simulation software to obtain a strain parameter combination corresponding to each trace parameter combination.

In one embodiment, the strain parameter combination may include one or more strain parameters, which may be impedance parameters, insertion loss parameters, far-end crosstalk parameters, near-end crosstalk parameters, and the like. The preset PCB routing simulation software may be IMLC software provided by Inter corporation, the IMLC software may obtain a strain parameter combination corresponding to the routing parameter combination according to the input routing parameter combination, and the preset PCB routing simulation software may also be other simulation software capable of implementing PCB design. The calculation module 102 can input each trace parameter combination to the preset PCB trace simulation software to obtain a strain parameter combination corresponding to each trace parameter combination.

The first training module 103 is configured to cooperate the plurality of routing parameter combinations and the plurality of strain parameter sets as a training sample to train a preset network model to obtain a first prediction model.

In an embodiment, a plurality of the trace parameter combinations and a plurality of the strain parameter combinations may construct a training sample library, and each training feature of the training sample library may include a trace parameter combination and a strain parameter combination corresponding to the trace parameter combination. The first training module 103 may randomly divide the plurality of training samples into a training set and a test set, where a data amount of the training set is preferably greater than a data amount of the test set, for example, 80% of training features may be divided into the training set, 20% of training features may be divided into the test set, and then the training set is used to train the preset network model and the test set is used to test the trained intermediate model. When the test result of the intermediate model meets the preset requirement, the first training module 103 may use the trained intermediate model as the first prediction model.

In an embodiment, the preset network model may be a machine learning model, such as a CNN or RNN neural network model, but is not limited thereto. When the first training module 103 trains the preset network model (or machine learning model) by using the training set, an intermediate model may be obtained, each test data of the test set is input to the intermediate model for testing, a verification result may be obtained, the first training module 103 may count the verification result of each test data, and further obtain a final test result of the intermediate model, when the test result of the intermediate model meets the preset requirement, it indicates that the intermediate model obtained by training meets the requirement, and the first training module 103 may use the intermediate model obtained by training as the first prediction model. The predetermined requirement may be, for example, that the model test accuracy is greater than a predetermined threshold.

For example, the strain parameter combination includes an impedance parameter, an insertion loss parameter, a far-end crosstalk parameter, and a near-end crosstalk parameter, and the preset threshold is 95%. Each test data in the test set is input to the intermediate model to obtain a set of impedance value, intervention loss value, far-end crosstalk value, and near-end crosstalk value, if the errors of the impedance value, the intervention loss value, the far-end crosstalk value, and the near-end crosstalk value predicted by the intermediate model and the impedance value, the intervention loss value, the far-end crosstalk value, and the near-end crosstalk value calculated by the preset PCB trace simulation software are all smaller than a preset error value (for example, 10%), it is determined that the test result of the test data passes, if the test passing rate of the test data in the test set is greater than 95%, it is indicated that the intermediate model obtained by training can meet the actual requirements, and the first training module 103 can use the intermediate model obtained by training as the first prediction model.

When the test result of the intermediate model does not meet the preset requirement, the first training module 103 may adjust the parameters of the preset network model, and then train the adjusted preset network model again by using the training set and test the intermediate model obtained by the retraining by using the test set. When the test result of the retrained intermediate model meets the preset requirement, the first training module 103 uses the retrained intermediate model as the first prediction model, and when the test result of the retrained intermediate model still does not meet the preset requirement, the model parameter adjustment step may be repeated until the test result of the retrained intermediate model meets the preset requirement.

In an embodiment, the first training module 103 may adjust the parameters of the preset network model (or machine learning model) by adjusting the total number of layers (e.g., the number of hidden layers) and/or the number of neurons per layer (or model complexity) of the neural network model.

In an embodiment, since the strain parameter combination includes a plurality of strain parameters, the first prediction model can obtain a strain parameter combination matching the trace parameter combination according to the input trace parameter combination. In the actual model training process, in order to make the prediction accuracy of each strain parameter meet the test requirement, the error value may be set to be relatively large, for example, when the error values of the predicted impedance value, the intervention loss value, the far-end crosstalk value and the near-end crosstalk value are all within 10%, the prediction is determined to be accurate, and the following manner can be used for continuously training the first prediction model, so that the optimal prediction model of each response variable, for example, the optimal prediction model of the impedance parameter, can be obtained.

The second training module 104 is configured to train and test the first prediction model to obtain an impedance prediction model of the impedance parameter.

In an embodiment, the second training module 104 may train the first prediction model by using part or all of the data of the training set, test the trained first prediction model by using part or all of the data of the test set, and if the test result meets the requirement, use the trained first prediction model as the impedance prediction model. When a plurality of wiring parameter combinations in the test set are input to the impedance prediction model, the average error between the impedance value predicted by the impedance prediction model and the impedance value calculated by the preset PCB wiring simulation software is smaller than a preset error. The preset error can be set according to actual requirements, for example, the preset error is 5%.

When the impedance prediction model is obtained through training, a pre-designed wiring parameter combination can be input into the impedance prediction model, and the impedance prediction model can output an impedance value corresponding to the wiring parameter combination. The impedance prediction model is high in calculation speed, only needs 0.02 second approximately, is increased by 15000 times approximately compared with the calculation speed of IMLC software, and greatly reduces the PCB wiring design time course.

It is understood that the second training module 104 may further train and test the first prediction model to obtain the prediction model of the interventional loss parameter. When a plurality of wiring parameters in a test set are combined and input to the intervention loss parameter prediction model, the average error between the intervention loss value predicted by the intervention loss parameter prediction model and the intervention loss value calculated by the preset PCB wiring simulation software is less than 5%. The second training module 104 can also train and test the first prediction model to obtain the prediction model of the far-end crosstalk parameter. When a plurality of wiring parameter combinations in the test set are input to the prediction model of the far-end crosstalk parameter, the average error between the far-end crosstalk value predicted by the prediction model of the far-end crosstalk parameter and the far-end crosstalk value calculated by the preset PCB wiring simulation software is less than 5%. The second training module 104 may also train and test the first prediction model to obtain the prediction model of the near-end crosstalk parameter. When a plurality of wiring parameters are combined and input to the near-end crosstalk parameter prediction model, the average error between the near-end crosstalk value predicted by the near-end crosstalk parameter prediction model and the near-end crosstalk value calculated by the preset PCB wiring simulation software is less than 5%.

In an embodiment, the second training module 104 may further train and test the first prediction model to obtain a prediction model for predicting the insertion loss parameter, the far-end crosstalk parameter and the near-end crosstalk parameter. When a plurality of wiring parameters are combined and input into the prediction model, the insertion loss value, the far-end crosstalk value and the near-end crosstalk value predicted by the prediction model and the average error of the insertion loss value, the far-end crosstalk value and the near-end crosstalk value calculated by the preset PCB wiring simulation software are all smaller than a preset error, for example, smaller than 5%.

Fig. 3 is a flowchart of a PCB trace parameter setting method according to an embodiment of the invention. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs.

Step S300, obtaining a plurality of trace parameter combinations for PCB trace design.

In one embodiment, the trace parameter combination may be a combination including a plurality of strain parameters that may affect the PCB, such as impedance, insertion loss, far-end crosstalk, near-end crosstalk, and the like, and each trace parameter combination may include one or more trace parameters. For example, the trace parameters may be line width, line width between lines, insulation thickness, copper foil thickness, dielectric constant, and the like.

In an embodiment, a trace parameter combination library may be established first, and then a plurality of trace parameter combinations for PCB trace design may be obtained from the trace parameter combination library, where the trace parameter combination library includes a plurality of trace parameter combinations, and the trace parameter combinations may be established in the following manner: the method includes the steps of setting up wiring parameters which affect the values of impedance, insertion loss, far-end crosstalk and near-end crosstalk, and then designing a plurality of sets of wiring parameter combinations by combining the wiring parameters and the design range which can be implemented by the factory end through an experimental design method, wherein the plurality of sets of wiring parameter combinations can contain the same or different wiring parameters. For example, the plurality of routing parameter combinations may include the same routing parameters, but different specific parameter values of the routing parameters.

Step S302, inputting a plurality of trace parameter combinations to a preset PCB trace simulation software to obtain a strain parameter combination corresponding to each trace parameter combination, where the strain parameter combination at least includes an impedance parameter.

In one embodiment, the strain parameter combination may include one or more strain parameters, which may be impedance parameters, insertion loss parameters, far-end crosstalk parameters, near-end crosstalk parameters, and the like. The preset PCB routing simulation software may be IMLC software provided by Inter corporation, the IMLC software may obtain a strain parameter combination corresponding to the routing parameter combination according to the input routing parameter combination, and the preset PCB routing simulation software may also be other simulation software capable of implementing PCB design. Each of the trace parameter combinations may be input to the preset PCB trace simulation software to obtain a strain parameter combination corresponding to each of the trace parameter combinations.

Step S304, training a preset network model by using the routing parameter combinations and the strain parameter combinations as training samples to obtain a first prediction model.

In an embodiment, a plurality of the trace parameter combinations and a plurality of the strain parameter combinations may construct a training sample library, and each training feature of the training sample library may include a trace parameter combination and a strain parameter combination corresponding to the trace parameter combination. The training samples may be randomly divided into a training set and a test set, where the data size of the training set is preferably greater than the data size of the test set, for example, 80% of the training features may be divided into the training set, 20% of the training features may be divided into the test set, and then the training set is used to train the preset network model and the test set is used to test the trained intermediate model. When the test result of the intermediate model meets the preset requirement, the intermediate model obtained by training can be used as the first prediction model.

In an embodiment, the preset network model may be a machine learning model, such as a CNN or RNN neural network model. When the training set trains the preset network model to obtain an intermediate model, each test data of the test set is input into the intermediate model to be tested to obtain a verification result, the verification result of each test data can be obtained through statistics, a final test result of the intermediate model can be obtained, when the test result of the intermediate model meets the preset requirement, the intermediate model obtained through training meets the requirement, and the intermediate model obtained through training can be used as the first prediction model. The predetermined requirement may be, for example, that the model test accuracy is greater than a predetermined threshold.

Step S306, training and testing the first prediction model to obtain an impedance prediction model of the impedance parameter.

In an embodiment, when a plurality of trace parameter combinations are input to the impedance prediction model, an average error between an impedance value predicted by the impedance prediction model and an impedance value calculated by the preset PCB trace simulation software is smaller than a preset error.

In an embodiment, the first prediction model may be trained by using part or all of the data of the training set, and then the trained first prediction model is tested by using part or all of the data of the test set, and if the test result meets the requirement, the trained first prediction model is used as the impedance prediction model. When a plurality of wiring parameter combinations in the test set are input to the impedance prediction model, the average error between the impedance value predicted by the impedance prediction model and the impedance value calculated by the preset PCB wiring simulation software is smaller than a preset error. The preset error can be set according to actual requirements, for example, the preset error is 5%.

When the impedance prediction model is obtained through training, a pre-designed wiring parameter combination can be input into the impedance prediction model, and the impedance prediction model can output an impedance value corresponding to the wiring parameter combination. The impedance prediction model is high in calculation speed, only needs 0.02 second approximately, is increased by 15000 times approximately compared with the calculation speed of IMLC software, and greatly reduces the PCB wiring design time course.

According to the PCB routing parameter setting device, the PCB routing parameter setting method and the computer readable storage medium, the prediction model for learning the PCB routing simulation software is obtained based on machine learning algorithm training, the strain parameter combination corresponding to the routing parameter combination can be obtained through calculation through the prediction model, compared with the PCB routing simulation software, the calculation speed is higher, the PCB routing design time course can be reduced, and the design efficiency is improved.

It will be apparent to those skilled in the art that other variations and modifications may be made in accordance with the invention and its spirit and scope in accordance with the practice of the invention disclosed herein.