PCB product signal integrity prediction method and device and storage medium

Abstract

The invention discloses a PCB product signal integrity prediction method and device and a storage medium, and belongs to the technical field of PCB manufacturing, and the method comprises the steps: inputting an obtained to-be-predicted PCB design parameter into a pre-trained and optimized prediction regression mathematical model, and outputting a signal integrity prediction result; the training and optimizing method of the predictive regression mathematical model comprises the following steps: determining dielectric loss according to the type of a PCB material; the copper surface roughness is determined according to the copper foil brownification type; training a prediction regression mathematical model by using the dielectric loss, the copper surface roughness and the frequency, the line width spacing, the dielectric thickness and the signal layer thickness which are input in advance; according to the method, big data processing, SI calculation and SI prediction are carried out on the production influence factors before and after the production process by utilizing the prediction regression mathematical model, and the SI risk is prevented, so that the high-risk material number is reduced, the cost loss is reduced, and meanwhile, the product quality is ensured.

Description

Technical Field

[0001] The present invention relates to a method, device and storage medium for predicting signal integrity of PCB products, belonging to the technical field of PCB board manufacturing. Background Art

[0002] With the progress of technology, high-frequency and high-speed communication products are more and more widely used, and the requirements for the quality stability of PCB boards are also getting higher and higher. However, there are many types of raw materials and processing technologies for producing PCB boards at present, and these factors will all affect the stability of signal integrity SI. Therefore, a more systematic and scientific SI performance prediction and evaluation method is needed.

[0003] At present, the method for evaluating the SI performance of PCB boards in the board factory is as follows:

[0004] 1. During the production process, through testing, evaluation, recording, establishing its own SI database, analyzing the data influencing factors, such as copper foil type, brownification type, dielectric type, line width, line spacing, copper thickness, test frequency, and obtaining the SI result by the method of assigning weights, which can be used as a reference for the next production. Although the above method has practical production significance on the one hand, the current production materials and production processes change rapidly, which cannot meet the existing production conditions. On the other hand, the weights of the influencing factors are fixed, lacking frontier and flexibility.

[0005] 2. At present, there is no very mature insertion loss (2D) simulation software in the industry. What people use more, such as SIPolar9000, analyzes by inputting copper foil type roughness, dielectric thickness, line width, line spacing, copper thickness, test frequency, and dielectric constant (DK) of the dielectric. However, if the copper foil roughness and DF data of SI Polar9000 use measured data for prediction, the accuracy is very poor and can only be adjusted by coefficient change, and the overall accuracy has defects.

[0006] 3. Manually evaluate the SI performance stability by professional SI engineers. This method is experience-oriented and lacks scientificity. Summary of the Invention

[0007] The purpose of the present invention is to overcome the deficiencies in the prior art, and provide a method, device and storage medium for predicting signal integrity of PCB products. By using a predictive regression mathematical model, before and after the production process, big data processing is carried out on the production influencing factors, SI is calculated and predicted, and SI risks are prevented, so as to reduce high-risk part numbers, reduce cost losses and ensure product quality at the same time.

[0008] To achieve the above object, the present invention is implemented by the following technical solutions:

[0009] In the first aspect, the present invention provides a method for predicting signal integrity of PCB products, including:

[0010] Input the obtained PCB design parameters to be predicted into a pre-trained and optimized predictive regression mathematical model, and output the signal integrity prediction result;

[0011] The training and optimization method of the predictive regression mathematical model includes:

[0012] Determine the dielectric loss according to the type of PCB material;

[0013] Determine the copper surface roughness according to the type of copper foil browning;

[0014] Use the dielectric loss, copper surface roughness, and the pre-input frequency, line width and spacing, dielectric thickness, and signal layer thickness to train the predictive regression mathematical model, and calculate the error between the output value and the true value of the predictive regression mathematical model;

[0015] Adjust the hyperparameters of the predictive regression mathematical model to reduce the error between the output value and the true value of the predictive regression mathematical model to a preset range, and complete the training and optimization of the predictive regression mathematical model.

[0016] Further, the predictive regression mathematical model adopts the LinghtGBM mathematical model.

[0017] Further, the hyperparameters of the predictive regression mathematical model include the number of cycles and the training data ratio.

[0018] Further, when calculating the error between the output value and the true value of the predictive regression mathematical model, the mean squared error MSE, or the R-squared error R 2 , or the mean absolute error MAE is used.

[0019] Further, the calculation formula of the mean squared error MSE is as follows: ; where y pred is the predicted value, y is the actual value of the sample, is the amount of data, is the th data in the database.

[0020] Further, the calculation formula of the R-squared error R 2 is as follows: ; where is the sample mean.

[0021] Further, the calculation formula of the mean absolute error MAE is as follows: .

[0022] In a second aspect, the present invention provides a signal integrity prediction device for a PCB product, comprising:

[0023] a prediction module configured to input the obtained PCB design parameters to be predicted into a pre-trained and optimized prediction regression mathematical model, and output a signal integrity prediction result;

[0024] wherein, the training and optimization method of the prediction regression mathematical model includes:

[0025] determining the dielectric loss according to the type of PCB material;

[0026] determining the copper surface roughness according to the type of copper foil browning;

[0027] using the dielectric loss, copper surface roughness, and pre-input frequency, line width spacing, dielectric thickness, and signal layer thickness to train the prediction regression mathematical model, and calculating the error between the output value and the true value of the prediction regression mathematical model;

[0028] adjusting the hyperparameters of the prediction regression mathematical model to reduce the error between the output value and the true value of the prediction regression mathematical model to a preset range, thereby completing the training and optimization of the prediction regression mathematical model.

[0029] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of any one of the foregoing methods are implemented.

[0030] In a fourth aspect, the present invention provides a computer device, comprising:

[0031] a memory for storing computer programs / instructions;

[0032] a processor for executing the computer programs / instructions to implement the steps of any one of the foregoing methods.

[0033] In a fifth aspect, the present invention provides a computer program product, comprising computer programs / instructions, and when the computer programs / instructions are executed by a processor, the steps of any one of the foregoing methods are implemented.

[0034] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0035] 1. The present invention provides a signal integrity prediction method, device, and storage medium for a PCB product. By using a prediction regression mathematical model, before and after the production process, big data processing is performed on production influencing factors, SI is calculated and predicted, SI risks are prevented, so as to reduce high-risk part numbers, reduce cost losses, and ensure product quality at the same time.

[0036] 2. The present invention bypasses expensive engineering software and can train a mathematical model for predicting SI using production data without the need to equip additional test software and hardware, and is convenient for multiple people to use simultaneously. The model trained by the present invention using big data has more flexible weights of influencing factors, making the prediction results more accurate and capable of handling more complex production materials and production processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a schematic diagram of the architecture design of the present invention;

[0038] Figure 2 is a schematic diagram of the mathematical model structure used in the present invention;

[0039] Figure 3 is a flowchart of the method for predicting the signal integrity of PCB products provided by an embodiment of the present invention;

[0040] Figure 4 is a schematic diagram of data collection provided by an embodiment of the present invention;

[0041] Figure 5 is a schematic diagram of the device hardware information provided by an embodiment of the present invention;

[0042] Figure 6 is a schematic diagram of function calls provided by an embodiment of the present invention;

[0043] Figure 7 is a schematic diagram of feature extraction provided by an embodiment of the present invention;

[0044] Figure 8 is a schematic diagram of dataset division provided by an embodiment of the present invention;

[0045] Figure 9 is a schematic diagram of model hyperparameter setting provided by an embodiment of the present invention;

[0046] Figure 10 is a schematic diagram of model training provided by an embodiment of the present invention;

[0047] Figure 11 is a schematic diagram of predicting SI values provided by an embodiment of the present invention;

[0048] Figure 12 is a schematic diagram of evaluating the accuracy of the model provided by an embodiment of the present invention;

[0049] Figure 13 is a schematic diagram of SI engineers analyzing data provided by an embodiment of the present invention;

[0050] Figure 14 is a schematic diagram of comparing three SI values provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0051] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and should not be used to limit the protection scope of the present invention.

[0052] Embodiment 1. This embodiment introduces a method for predicting the signal integrity of a PCB product, including:

[0053] Input the obtained PCB design parameters to be predicted into a pre-trained and optimized prediction regression mathematical model, and output the signal integrity prediction result;

[0054] The training and optimization method of the prediction regression mathematical model includes:

[0055] Determine the dielectric loss according to the type of PCB material;

[0056] Determine the copper surface roughness according to the type of copper foil browning;

[0057] Use the dielectric loss, copper surface roughness, and the pre-input frequency, line width spacing, dielectric thickness, and signal layer thickness to train the prediction regression mathematical model, and calculate the error between the output value and the true value of the prediction regression mathematical model;

[0058] Adjust the hyperparameters of the prediction regression mathematical model to reduce the error between the output value and the true value of the prediction regression mathematical model to a preset range, and complete the training and optimization of the prediction regression mathematical model.

[0059] As Figure 1 、 Figure 2 、 Figure 3 shown, the application process of the method for predicting the signal integrity of the PCB product provided in this embodiment specifically involves the following steps:

[0060] 1) Select the prediction regression mathematical model. This invention uses the LinghtGBM mathematical model;

[0061] 2) Confirm the dielectric loss according to the material type;

[0062] 3) Confirm the copper surface roughness Rz according to the type of copper foil browning;

[0063] 4) Combine the above parameters, input the frequency, line width spacing, dielectric thickness, and signal layer thickness, and confirm the number of model input parameters;

[0064] 5) Train the model, and calculate the error according to the model output result and the true value;

[0065] 6) Adjust the model hyperparameters, such as the number of cycles and the proportion of training prediction data, to reduce the GAP of the model to an acceptable range;

[0066] 7) Input the prediction data, and the model can output the prediction value according to the prediction input;

[0067] 8) The SI engineer analyzes the results, completes the material pre - review, anomaly analysis, and production process specification.

[0068] The following combines with a preferred embodiment to illustrate the content involved in the above - mentioned embodiments.

[0069] The PCB product signal integrity prediction method provided by this embodiment includes the following content:

[0070] I. Data collection: Grab the data set through the database. The data set label types include FREQ (frequency), ER1 (dielectric constant of the medium), TAND (loss of the medium), RZ_DRUM (roughness of the non - browned surface), RZ_MATTE (roughness of the browned surface), W1 (etching virtual image length), W2 (etching real image length), S1 (distance between differential lines), H1 (thickness of the core), H2 (PP thickness), T1 (copper thickness of the trace), IL (insertion loss). As Figure 4 shown.

[0071] II. Model construction:

[0072] 1. Device hardware information, as Figure 5 shown;

[0073] Processor: AMD Ryzen 5 5600G with Radeon Graphics 3.90 GHz;

[0074] Processor bit - length: 64 - bit operating system, x64 - type processor;

[0075] Memory: 16.0 GB;

[0076] 2. Function calls, as Figure 6 shown;

[0077] Call the numpy function library of python, call the pandas function library, call the mean squared error function from the sklearn function library, call the mean squared error function from the sklearn function library, call Lightgbm named lgb, call the squared absolute error;

[0078] 3. Extract high - level data features according to the data labels, as Figure 7 shown;

[0079] 4. Divide the data set. For all rows of the data set, the first eleven columns are the model inputs, and the label IL (insertion loss) is the predicted output value, as Figure 8 shown;

[0080] 5. The hyperparameters of the LightGBM model are set as follows: boost_type: gbdt, object: regression, metric: RMSE, num_leaves: 50, learning_rate: 0.05, feature_fraction: 1, bagging_fraction: 0.9, bagging_freq: 0.9, verbose: 0, num_boost_round: 1000, early_stopping_rounds: 50; as Figure 9 shown;

[0081] 6. Create a model, bring in the model hyperparameters and the dataset for training, and save the trained model, as Figure 10 shown;

[0082] 7. Train the model, bring in the prediction input data for prediction, as Figure 11 shown;

[0083] 8. According to the predicted output values, use the mean squared error MSE, R-squared error R 2 and mean absolute error MAE as the model evaluation metrics, as Figure 12 shown. The formulas are as follows:

[0084] ;

[0085] ;

[0086] ;

[0087] where, is the sample mean, y pred is the predicted value, y is the actual sample value, is the number of data, is the th data in the database.

[0088] 9. The SI engineer analyzes based on the predicted values and the true values, and calculates the proportion of errors within 5%, as Figure 13 shown.

[0089] III. Test Results:

[0090] Test Data 1: Record the simulation values and mathematical modeling prediction values in the SI Polar9000 software for different project numbers, and compare them with the measured values. After batch verification and repeatability verification, this model has relatively stable test accuracy. It improves our accuracy and has been put into actual use. As shown in Table 1 and Figure 14 shown.

[0091] Table 1 SI Value Data Material Test Frequency Mathematical Modeling Value 9000 Estimated Value Customer Specification Measured Value R-579Y(N) 26.56 GHZ -0.65 -0.66 -0.7 -0.65 H360(C) 12.8 GHZ -0.64 -0.62 -0.69 -0.61 EM-892K; EM-370(Z); MC24P 26 GHZ -0.79 -0.81 -0.84 -0.78 TU-933E 16 GHZ -0.92 -0.94 -0.99 -0.95 R-5795(N)(Megtron8(N)) 26.56 GHZ -0.65 -0.67 -0.7 -0.65 R-578Y(GT); R-579Y(N) 26.56 GHZ -0.67 -0.68 -0.72 -0.7 R-5795(N)(Megtron8(N)) 26.56 GHZ -0.65 -0.67 -0.7 -0.66 R-578Y(GT) 13.28 GHZ -0.51 -0.52 -0.55 -0.5 EM-890K; EM-370(Z) 26.56 GHZ -0.85 -0.87 -0.91 -0.82 EM-890K 26.56 GHZ -1.15 -1.17 -1.25 -1.15 TU-943HR; TU-865 26.56 GHZ -0.58 -0.6 -0.63 -0.59 R-578Y(GT); R-1755V 13.28 GHZ -0.56 -0.57 -0.6 -0.575 H360(C) 14.5 GHZ -0.65 -0.67 -0.75 -0.63 Synamic6(X) 14.5 GHZ -0.62 -0.595 -0.67 -0.61 IT-968G 14.5 GHZ -0.65 -0.62 -0.75 -0.65 NY6300(C) 14.5 GHZ -0.63 -0.68 -0.75 -0.62 NY6180L 14.1 GHZ -1.09 -1.13 -1.18 -1.04 H360(C) 14.5 GHZ -0.7 -0.74 -0.9 -0.69 EM-890K 26.56 GHZ -1.11 -1.13 -1.15 -1.1 TU-863+(Thunderclad1+) 12.5 GHZ -1.07 -1.04 1.15 -1.1 H360(C) 14.5 GHZ -0.72 -0.78 -0.9 -0.69 EM-891; EM827 13.28 GHZ -0.65 -0.66 -0.7 -0.65 EM-892K 26.56 GHZ -0.83 -0.84 -0.87 -0.83 NY6300(C) 12.5 GHZ -0.75 -0.8 -0.9 -0.73

[0092] Embodiment 2. This embodiment provides a PCB product signal integrity prediction device, including:

[0093] A prediction module, configured to input the obtained PCB design parameters to be predicted into a pre-trained and optimized prediction regression mathematical model, and output a signal integrity prediction result;

[0094] Among them, the training and optimization method of the prediction regression mathematical model includes:

[0095] Determine the dielectric loss according to the type of PCB material;

[0096] Determine the copper surface roughness according to the type of copper foil brownification;

[0097] Use the dielectric loss, copper surface roughness, and the pre-input frequency, line width spacing, dielectric thickness, and signal layer thickness to train the prediction regression mathematical model, and calculate the error between the output value and the true value of the prediction regression mathematical model;

[0098] Adjust the hyperparameters of the prediction regression mathematical model to reduce the error between the output value and the true value of the prediction regression mathematical model to a preset range, and complete the training and optimization of the prediction regression mathematical model.

[0099] For the specific function implementation of the above modules, refer to the relevant content in the method of Embodiment 1, which will not be elaborated here.

[0100] Embodiment 3. This embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the steps of the method described in any one of Embodiment 1.

[0101] Embodiment 4. This embodiment provides a computer device, including:

[0102] A memory, configured to store computer programs / instructions;

[0103] A processor, configured to execute the computer programs / instructions to implement the steps of the method described in any one of Embodiment 1.

[0104] Embodiment 5. This embodiment provides a computer program product, including computer programs / instructions, and when the computer programs / instructions are executed by a processor, they implement the steps of the method described in any one of Embodiment 1.

[0105] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and deformations can still be made, and these improvements and deformations should also be regarded as the protection scope of the present invention.

[0106] Those skilled in the art should understand that the embodiments of the present disclosure can be provided as a method, a system or a computer program product. Therefore, the present disclosure can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present disclosure can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0107] The present disclosure is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0108] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure rather than to limit the scope of its protection. Although the present disclosure has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that after reading the present disclosure, various changes, modifications or equivalent replacements can still be made to the specific implementation manners of the invention, but these changes, modifications or equivalent replacements are all within the scope of the claims of the present disclosure awaiting approval.

Claims

1. A PCB product signal integrity prediction method, characterized in that: include: The obtained PCB design parameters to be predicted are input into the pre-trained and optimized prediction regression mathematical model, and the signal integrity prediction results are output; The training and optimization method of the prediction regression mathematical model includes: Determine dielectric loss based on the type of PCB material; Determine the copper surface roughness according to the copper foil browning type; The prediction regression mathematical model is trained using dielectric loss, copper surface roughness, and pre-input frequency, line width spacing, dielectric thickness, and signal layer thickness, and the error between the output value of the prediction regression mathematical model and the true value is calculated; Adjust the hyperparameters of the predictive regression mathematical model so that the error between the output value of the predictive regression mathematical model and the true value is reduced to a preset range, completing the training and optimization of the predictive regression mathematical model.

2. The PCB product signal integrity prediction method according to claim 1, characterized in that: The prediction regression mathematical model adopts LinghtGBM mathematical model.

3. The PCB product signal integrity prediction method according to claim 1, characterized in that: The hyperparameters of the predictive regression mathematical model include the number of cycles and the proportion of training data.

4. The PCB product signal integrity prediction method according to claim 1, characterized in that: The error between the output value of the regression mathematical model and the true value is calculated using the mean square error (MSE) or the R square error (R). 2 , or MAE, the absolute error.

5. The PCB product signal integrity prediction method according to claim 4, characterized in that: The calculation formula of the mean square error MSE is as follows: ； Among them, y pred is the predicted value, y is the actual sample value, is the amount of data, For the database data.

6. The PCB product signal integrity prediction method according to claim 5, characterized in that: The R square error R 2 The calculation formula is as follows: ； in, is the sample mean.

7. The PCB product signal integrity prediction method according to claim 6, characterized in that: The calculation formula of the absolute error MAE is as follows: 。 8. A PCB product signal integrity prediction device, characterized in that: include: The prediction module is used to input the acquired PCB design parameters to be predicted into the pre-trained and optimized prediction regression mathematical model, and output the signal integrity prediction results; The training and optimization method of the prediction regression mathematical model includes: Determine dielectric loss based on the type of PCB material; Determine the copper surface roughness according to the copper foil browning type; The prediction regression mathematical model is trained using dielectric loss, copper surface roughness, and pre-input frequency, line width spacing, dielectric thickness, and signal layer thickness, and the error between the output value of the prediction regression mathematical model and the true value is calculated; Adjust the hyperparameters of the predictive regression mathematical model so that the error between the output value of the predictive regression mathematical model and the true value is reduced to a preset range, completing the training and optimization of the predictive regression mathematical model.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

10. A computer device, characterized in that: include: Memory, for storing computer programs / instructions; A processor, configured to execute the computer program / instructions to implement the steps of the method according to any one of claims 1 to 7.

Images