PCBA (Printed Circuit Board Assembly) processing method and system for improving strain displacement precision of medical color ultrasound equipment

Abstract

The invention discloses a PCBA (Printed Circuit Board Assembly) processing method and system for improving strain displacement precision of medical color ultrasound equipment. The method comprises the following steps: acquiring PCBA processing multi-dimensional original data; constructing a frequency domain characteristic decomposition displacement inversion model to invert an initial strain displacement state; determining electromagnetic interference key parameters by using a PCBA anti-interference electromagnetic simulation algorithm; establishing a random forest ultrasonic feature classification model to identify a strain displacement deviation type; pCBA processing parameters are adjusted according to the identification result; and related data are uploaded to a color Doppler ultrasound PCBA full-process MES tracing platform. Wherein the steps S3-S5 refine the simulation, classification and parameter adjustment process through multiple steps to ensure the technical landing. The system is implemented based on a multi-unit cooperation and support method. According to the method, through multi-model fusion and full-process data tracing, processing precision influence factors are comprehensively covered, deviation is accurately corrected, PCBA strain displacement precision is effectively improved, the requirements for imaging quality and diagnosis accuracy of medical color ultrasound equipment are met, and the method adapts to compliance and safety standards of the medical industry.

Description

Technical Field

[0001] This invention relates to the field of medical color ultrasound equipment manufacturing technology, and in particular to a PCBA manufacturing method and system for improving the strain displacement accuracy of medical color ultrasound equipment. Background Technology

[0002] In medical color Doppler ultrasound equipment, the PCBA (Printed Circuit Board Assembly) is a core component, and its strain-displacement accuracy directly affects the ultrasound imaging quality and diagnostic accuracy. As medical diagnostics increasingly demand higher resolution and stability from color Doppler ultrasound equipment, controlling the processing precision of PCBAs has become a key direction for industry technological development. Currently, the PCBA processing involves multiple influencing factors, including substrate material, welding technology, and electromagnetic environment. Precision control is achieved through data acquisition, model analysis, and parameter adjustment throughout the entire processing flow. Simultaneously, the medical industry's stringent requirements for product traceability necessitate the construction of a comprehensive end-to-end data management system for PCBA processing. This ensures the controllability and traceability of each processing step, meets compliance and safety standards for medical equipment, and drives the development of medical color Doppler ultrasound equipment towards higher precision and more reliable performance.

[0003] Existing technologies have significant shortcomings in controlling strain displacement accuracy during PCBA fabrication for medical color Doppler ultrasound equipment. On one hand, current fabrication methods often rely on single models or algorithms for localized analysis, failing to organically combine multiple models such as displacement inversion, electromagnetic simulation, and feature classification. This results in an inability to comprehensively cover the combined impact of multi-dimensional factors on displacement accuracy during fabrication, leading to insufficient precision in identifying and correcting accuracy deviations, and making it difficult to meet the fabrication requirements of high-resolution color Doppler ultrasound equipment. On the other hand, existing technologies lack an integrated data traceability and monitoring mechanism for the entire PCBA fabrication process. Data acquisition, model execution results, and parameter adjustment records are stored in a scattered manner, making real-time data correlation and traceability impossible. This hinders the rapid identification of the root cause of accuracy deviations and fails to provide complete data support for subsequent fabrication optimization, thus restricting the continuous improvement of PCBA fabrication accuracy. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides a PCBA processing method and system for improving the strain displacement accuracy of medical color ultrasound equipment.

[0005] The technical solution adopted in this invention is a PCBA processing method for improving the strain displacement accuracy of medical color ultrasound equipment, comprising the following steps: S1, collecting multi-dimensional raw data related to strain displacement during the PCBA processing of medical color ultrasound equipment, including PCBA substrate material characteristic data, component soldering position data, temperature field data and vibration frequency data during processing; S2, based on the collected multi-dimensional raw data, constructing a frequency domain feature decomposition displacement inversion model, and performing frequency domain transformation and feature extraction on the multi-dimensional raw data to perform inversion analysis on the initial strain displacement state during PCBA processing; S3, using a PCBA anti-interference electromagnetic simulation algorithm, combined with the inverted initial strain displacement state data, simulating electromagnetic interference sources that may exist during PCBA processing. S4. Simulate and determine the calibration parameters that affect the strain displacement accuracy of PCBA; S5. Establish a random forest ultrasonic feature classification model, input the electromagnetic interference calibration parameters obtained from the simulation and the ultrasonic detection data during the PCBA processing into the model, and classify and identify different strain displacement deviation types during PCBA processing; S6. Based on the classification and identification results, adjust various parameters of the medical color ultrasound equipment PCBA processing, including substrate etching depth parameters, component welding pressure parameters, reflow soldering temperature curve parameters, and cooling rate parameters; S7. Upload the adjusted processing parameters, different model running data, and PCBA processing data to the color ultrasound PCBA full-process MES traceability platform to monitor and trace the PCBA processing process in real time, and complete the medical color ultrasound equipment PCBA processing.

[0006] Furthermore, the expression for the frequency domain eigenvalue decomposition shift inversion model in S2 is: ,in, The initial strain-displacement values ​​of the PCBA obtained from the inversion are... The number of dimensions in the original multi-dimensional data. For the first The original data of the dimension, It is a frequency domain transformation function. For the first Frequency domain feature weights of dimensional data For frequency domain characteristic adjustment coefficients, For the first Error values ​​of dimensional data, For the first Inversion coefficients of dimensional data The length of time for data collection. It is a time-frequency correlation function. For the first Feature frequencies of dimensional data This is the strain error correction function. For the first Standard deviation of dimensional data.

[0007] Furthermore, the expression for the PCBA anti-interference electromagnetic simulation algorithm in S3 is as follows: ,in, This refers to the electromagnetic interference intensity value during PCBA manufacturing. The number of electromagnetic interference sources, For the first The current intensity of each interference source, For the first The equivalent inductance of each interference source The permeability of free space, For the first The distance from each interference source to the sensitive area of ​​the PCBA The electromagnetic attenuation coefficient, The dielectric thickness between the interference source and the PCBA. The number of sensitive components on the PCBA For the first The coupling coefficient of each sensitive component, For the first The operating voltage of each sensitive component For the first The operating angular frequency of each sensitive component For the first The equivalent resistance of each sensitive component For the first The equivalent capacitance of each sensitive component.

[0008] Furthermore, the expression for the random forest ultrasonic feature classification model in S4 is: ,in, The results are for classifying strain displacement deviations. For the set of all deviation categories, The number of decision trees in the random forest. For the first The number of nodes in a decision tree. For the first The first decision tree The weight of each node, For the first The first decision tree Each node classifies the sample. The probability, The regularization coefficient is . The number of features detected by ultrasound. For the first Weights of each ultrasound feature, This is a correlation function between ultrasound features and deviation categories. For the first Raw data from ultrasound examination. For the first Each ultrasonic detection feature value.

[0009] Furthermore, the calculation expression for adjusting the PCBA processing parameters of the medical color ultrasound equipment in S5 is as follows: ,in, These are the adjusted PCBA processing parameter values. These are the initial processing parameter values. Number of strain displacement deviation types For the first The parameter adjustment coefficients corresponding to the different types of deviations For the first Deviation values ​​for different types of deviations The maximum deviation value among all deviation types. The number of factors influencing electromagnetic interference. For the first The parameter correction coefficients corresponding to the various electromagnetic interference factors For the first The interference intensity of various electromagnetic interference factors This represents the maximum interference intensity among all electromagnetic interference factors.

[0010] Furthermore, the data association expression for uploading data to the color Doppler ultrasound PCBA end-to-end MES traceability platform in S6 is as follows: For the associated traceability data in the MES traceability platform, This is the time data for PCBA manufacturing process. This is the original collected data. For data association matrix, For the number of data types, For the first The traceability weight of each data type For data hash function, For the first Types of data.

[0011] Further, S3 includes the following sub-steps: S31, extracting coordinate data, strain displacement rate of change data, and material characteristic parameter data of the PCBA sensitive area from the initial strain displacement state data obtained by S2, and converting these data into a data format recognizable by the PCBA anti-interference electromagnetic simulation algorithm to ensure that the data can accurately reflect the initial strain displacement state of the PCBA during the simulation process; S32, constructing a three-dimensional model consistent with the actual PCBA processing scenario in the simulation environment, including a PCBA substrate model, component model, and a model of the surrounding environment of the processing equipment, and simultaneously loading the extracted and converted initial strain displacement state data into the three-dimensional model. In the model, the simulation model is kept consistent with the actual processing state; S33, set the type parameters, intensity parameters, and location parameters of the electromagnetic interference source, where the interference source types include electrostatic interference source, radio frequency interference source, and electromagnetic radiation interference source. Add the corresponding interference source in the simulation model according to the set parameters to simulate the electromagnetic interference situation that may occur in the actual processing process; S34, run the PCBA anti-interference electromagnetic simulation algorithm to obtain the strain displacement change data of the PCBA sensitive area under the action of different interference sources. By comparing and analyzing the strain displacement data under the interference-free state and the interference state, determine the electromagnetic interference calibration parameters that have a significant impact on the strain displacement accuracy of the PCBA.

[0012] Further, S4 includes the following sub-steps: S41, collecting ultrasonic testing data corresponding to different strain displacement deviation types during the processing of medical color ultrasound equipment PCBA. These data include ultrasonic echo signal data, ultrasonic propagation time data, and ultrasonic amplitude data. The collected data is screened to remove obviously abnormal data and retain valid data for model training; S42, extracting features from the screened ultrasonic testing data. The extracted features include the time-domain peak feature, frequency-domain dominant frequency feature, and wavelet transform coefficient feature of the ultrasonic signal. The extracted features are associated with the corresponding strain displacement deviation type to construct the sample dataset required for model training; S43, dividing... The sample dataset consists of a training set, a validation set, and a test set. The training set is used to train the random forest ultrasonic feature classification model, the validation set is used to adjust the model's hyperparameters, and the test set is used to evaluate the model's classification performance. Initial hyperparameters are set for the number of decision trees, the depth of the trees, and the node splitting threshold in the random forest. S44. The random forest ultrasonic feature classification model is trained using the training set. During the training process, the model's hyperparameters are continuously adjusted using the validation set until the model's classification performance on the validation set meets the preset requirements. Finally, the performance of the trained model is tested using the test set to ensure that the model can accurately classify and identify different strain displacement deviation types during PCBA processing.

[0013] Further, step S5 includes the following sub-steps: S51. Obtain the strain displacement deviation type data and corresponding deviation degree data identified in S4, and simultaneously retrieve the electromagnetic interference calibration parameter data determined in S3. Integrate these data into a parameter adjustment analysis dataset to provide data support for subsequent processing parameter adjustments; S52. Analyze the correlation between different strain displacement deviation types and PCBA processing parameters, and clarify the main influencing processing parameters corresponding to each deviation type. For example, the substrate etching depth parameter mainly affects the PCBA circuit accuracy and thus affects the strain displacement, while the welding pressure parameter affects the bonding degree between the components and the substrate and thus affects the strain displacement; S53. Based on the correlation analysis results and combined with the influence law of electromagnetic interference calibration parameters on processing parameters, preliminarily determine the adjustment direction and adjustment range of each processing parameter. For example, when the electromagnetic interference intensity is large, the reflow soldering temperature curve parameters need to be adjusted appropriately to reduce the interference effect; S54. Based on the preliminarily determined adjustment direction and adjustment range, combined with the processing parameter adjustment calculation expression in S5, accurately calculate the adjustment value of each processing parameter, and complete the adjustment of various parameters of the medical color ultrasound equipment PCBA processing.

[0014] A PCBA fabrication system for improving strain displacement accuracy in medical color Doppler ultrasound equipment is disclosed. This system is applied to a PCBA fabrication method for improving strain displacement accuracy in medical color Doppler ultrasound equipment. The system includes: a multi-dimensional data acquisition unit connected to the PCBA fabrication equipment, which acquires data related to strain displacement during fabrication, including substrate material characteristics, component soldering positions, temperature field, and vibration frequency, and transmits the data to a data processing unit; a data processing unit connected to both the multi-dimensional data acquisition unit and a model building and running unit, which receives raw data, performs format conversion and preliminary processing, transmits the processed data to the model building and running unit, and simultaneously receives the model running results transmitted from the model building and running unit; and a model building and running unit connected to both the data processing unit and a parameter adjustment unit, which constructs and runs a frequency domain feature decomposition displacement inversion model, a PCBA anti-interference electromagnetic simulation algorithm, and a random forest ultrasound feature classification model based on the processed data. The model, simulation, and classification results are transmitted to the parameter adjustment unit. The parameter adjustment unit, connected to the model building and running unit, the MES traceability platform unit, and the PCBA processing control unit, calculates the processing parameter adjustment values ​​based on the model results and preset rules, transmitting these values ​​to the PCBA processing control unit and simultaneously to the MES traceability platform unit. The PCBA processing control unit, connected to the parameter adjustment unit and the multi-dimensional data acquisition unit, receives the adjustment values ​​to control the operation of the processing equipment, performs parameter control, and transmits the equipment operating status data to the multi-dimensional data acquisition unit. The MES traceability platform unit, connected to the parameter adjustment unit, model building and running unit, and multi-dimensional data acquisition unit, receives adjustment data, model running data, raw data, and processing process data, integrates, stores, and correlates them, enabling full-process data traceability and real-time monitoring.

[0015] Beneficial Effects: This invention proposes a PCBA processing method and system for improving the strain displacement accuracy of medical color Doppler ultrasound equipment. The method integrates a frequency domain feature decomposition displacement inversion model, a PCBA anti-interference electromagnetic simulation algorithm, and a random forest ultrasound feature classification model in a step-by-step manner. This achieves an organic combination of multiple models, comprehensively covering multi-dimensional influencing factors such as substrate material, welding process, and electromagnetic environment. It accurately inverts the initial strain displacement state, simulates the effects of electromagnetic interference, and identifies deviation types, solving the problems of existing technologies relying on single-model local analysis and inaccurate identification and correction of accuracy deviations. Simultaneously, through the color Doppler ultrasound PCBA full-process MES traceability platform, processing parameters are traced... The system enables real-time interconnection, storage, and monitoring of numerical, model, and process data, constructing an integrated end-to-end data traceability mechanism. This allows for rapid identification of the root causes of accuracy deviations and provides data support for subsequent optimization, overcoming the shortcomings of existing technologies such as scattered data storage and the inability to achieve real-time interconnection and traceability. Through the coordinated operation of six major units—multi-dimensional data acquisition, data processing, model construction and operation, parameter adjustment, PCBA processing control, and MES traceability platform—the system ensures that the entire processing process is controllable and traceable. Ultimately, this effectively improves the strain displacement accuracy of PCBA, meeting the imaging quality and diagnostic accuracy requirements of high-resolution color Doppler ultrasound equipment, while also complying with medical industry compliance and safety standards. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method steps of the present invention;

[0017] Figure 2 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, a PCBA manufacturing method for improving strain displacement accuracy in medical color ultrasound equipment includes the following steps:

[0020] S1. Collect multi-dimensional raw data related to strain and displacement during the processing of PCBA for medical color ultrasound equipment. The multi-dimensional raw data includes PCBA substrate material characteristics data, component welding position data, temperature field data and vibration frequency data during the processing.

[0021] Specifically, step S1 involves collecting multi-dimensional raw data related to strain and displacement during the PCBA manufacturing process of medical color Doppler ultrasound equipment. This step provides fundamental data support for subsequent model construction and parameter adjustment, directly impacting the accuracy of subsequent precision analysis. During implementation, a data acquisition network needs to be built using multiple types of sensors. Substrate material characteristic data is collected using a material analyzer, with specific parameters including the substrate's dielectric constant (range 4.2-4.8), coefficient of thermal expansion (range 13-17 ppm / ℃), and elastic modulus (range 18-22 GPa). Component soldering position data is collected using a high-precision visual positioning system, with positioning accuracy controlled within ±0.02 mm. The collected data includes the alignment deviation between component pins and pads, and the center coordinates of the soldering points. Temperature field data during the manufacturing process is also collected. Data is collected using distributed temperature sensors, with one sensor per square centimeter and a sampling frequency of 1 Hz. The sampling range covers the preheating zone (temperature range 80-120℃), reflow zone (temperature range 220-250℃), and cooling zone (temperature range 25-40℃) of PCBA processing. Vibration frequency data is collected using triaxial accelerometers at a sampling frequency of 1000 Hz. The collected parameters include the vibration frequency (mainly concentrated in the 50-200 Hz range) and vibration amplitude (controlled within 0.01-0.05 mm) during the processing. Data collection timestamps are recorded synchronously during the acquisition process to ensure time consistency across all dimensions of data. After collection, the data is stored in a local database using a structured data storage format for easy subsequent retrieval and processing.

[0022] S2. Based on the collected multi-dimensional raw data, a frequency domain feature decomposition displacement inversion model is constructed. By performing frequency domain transformation and feature extraction on the multi-dimensional raw data, the initial strain displacement state during PCBA processing is inverted and analyzed.

[0023] Specifically, step S2 constructs a frequency domain feature decomposition displacement inversion model based on the collected multi-dimensional raw data. Through frequency domain analysis, it achieves accurate inversion of the initial strain and displacement state of PCBA processing, providing an initial state benchmark for subsequent electromagnetic simulation and deviation identification. In implementation, the multi-dimensional raw data collected in S1 is first converted to the frequency domain using a Fast Fourier Transform (FFT). During the conversion, the number of data sampling points is set to 1024, and the frequency resolution is controlled at 0.1Hz. Then, frequency domain feature extraction is performed. For substrate material characteristic data, amplitude and phase spectrum features are extracted in the frequency domain; for welding position data, position deviation frequency features are extracted in the frequency domain; for temperature field data, temperature fluctuation frequency features are extracted in the frequency domain; and for vibration frequency data, the main vibration frequency features are extracted in the frequency domain. Next, a frequency domain feature decomposition displacement inversion model is constructed. The model input is the extracted multi-dimensional frequency domain features. The model structure is as follows: A multi-layer feedforward network with 5 layers was adopted. The number of nodes in the input layer was consistent with the frequency domain feature dimension, the number of nodes in the hidden layers were set to 64, 32, and 16 respectively, and the number of nodes in the output layer was 1 (corresponding to the initial strain displacement value). During model training, the gradient descent optimization algorithm was used with a learning rate of 0.001 and 1000 iterations, and the training error was controlled within 0.001. After training, the multi-dimensional raw data collected in S1 was input into the model, and the model output was the initial strain displacement state data during PCBA processing. The inversion accuracy was controlled within ±0.005mm, and this data will be used as the initial input parameters for the subsequent S3 electromagnetic simulation.

[0024] S3. Using the PCBA anti-interference electromagnetic simulation algorithm, combined with the initial strain displacement state data obtained by inversion, the electromagnetic interference sources that may exist in the PCBA processing are simulated to determine the calibration parameters that affect the accuracy of PCBA strain displacement.

[0025] Specifically, step S3 uses a PCBA anti-interference electromagnetic simulation algorithm, combined with the initial strain-displacement state data retrieved in step S2, to simulate the impact of electromagnetic interference on the PCBA strain-displacement accuracy, determine key interference parameters, and provide a basis for interference suppression for subsequent parameter adjustments. During implementation, firstly, a PCBA electromagnetic simulation environment is built using professional electromagnetic simulation software. The spatial mesh division accuracy of the simulation environment is set to 0.1mm, and the simulation frequency range covers the electromagnetic frequencies of the PCBA processing equipment and its surrounding environment (10kHz-1GHz). Then, the initial strain-displacement state data retrieved in step S2 (including strain-displacement values ​​and displacement distribution locations) is imported into the simulation environment, along with a 3D PCBA model. The model includes a substrate (100mm×80mm×1.6mm in size) and components (including chips, capacitors, resistors, etc., with component package types including QFP, SOP, and 0402). Next, electromagnetic interference source parameters are set, with interference source types including electrostatic interference sources (voltage range 1-10kV) and radio frequency interference sources. (Power range 0.1-1W), electromagnetic radiation interference source (field strength range 1-10V / m), the interference source is set within 0.5-2m around the PCBA processing equipment; run the PCBA anti-interference electromagnetic simulation algorithm, the simulation time is set to 10s, the simulation step size is set to 0.1s, and the strain displacement change data of the PCBA sensitive area (including chip pins and signal lines) is collected in real time; after the simulation is completed, the strain displacement data under no interference and interference conditions are compared and analyzed, the strain displacement deviation value corresponding to different interference sources is calculated, and the interference source parameters with deviation values ​​greater than 0.01mm are identified as key interference parameters, including interference source type, interference intensity, and interference location. These parameters will serve as important basis for S4 deviation classification and S5 parameter adjustment.

[0026] S4. Establish a random forest ultrasonic feature classification model, input the electromagnetic interference calibration parameters obtained from the simulation and the ultrasonic detection data in the PCBA processing process into the model, and classify and identify different strain displacement deviation types in the PCBA processing process.

[0027] Specifically, step S4 establishes a random forest ultrasonic feature classification model to accurately identify different strain-displacement deviation types during PCBA processing, providing a clear basis for parameter adjustment in S5 and ensuring the targeted nature of the adjustment. During implementation, common strain-displacement deviation types during PCBA processing are first collected, including substrate bending deviation, component soldering offset deviation, and circuit strain deviation, with at least 1000 samples collected for each type. Then, ultrasonic testing data corresponding to each deviation type is collected. A high-frequency ultrasonic testing instrument (frequency range 10-50MHz) is used to test the PCBA. During testing, the distance between the ultrasonic probe and the PCBA surface is controlled at 0.5mm, and the probe movement speed is set to 1mm / s. The collected ultrasonic data includes ultrasonic echo signal intensity, echo propagation time, and echo frequency. Feature processing is performed on the collected ultrasonic data to extract features such as peak intensity, peak occurrence time, signal spectral width, and signal energy of the ultrasonic echo signal. 20 features are extracted for each sample. A random forest ultrasonic feature classification model is then constructed. A forest ultrasonic feature classification model was developed, with 50 decision trees and a maximum depth of 10 layers per tree. The Gini coefficient was used as the node splitting criterion, and the training set to test set ratio was set to 7:3 during training. The model was trained using processed ultrasonic feature sample data, employing 5-fold cross-validation. The model's classification accuracy was required to reach over 95%. After training, the key electromagnetic interference parameters obtained from S3 simulation and the real-time collected PCBA ultrasonic detection data were input into the model. The model output was the strain displacement deviation type, including the deviation type name and deviation degree (divided into slight, moderate, and severe, with corresponding deviation value ranges of 0.005-0.01mm, 0.01-0.02mm, and over 0.02mm, respectively), providing a clear direction for subsequent S5 parameter adjustments.

[0028] S5. Based on the classification and identification results, adjust the various parameters of the PCBA processing for medical color ultrasound equipment. The processing parameters include substrate etching depth parameters, component welding pressure parameters, reflow soldering temperature curve parameters, and cooling rate parameters.

[0029] Specifically, step S5 adjusts various parameters of the PCBA processing for medical color ultrasound equipment based on the strain-displacement deviation types identified in S4. This precise parameter adjustment counteracts the effects of deviations, improving PCBA strain-displacement accuracy, and is the core execution step for achieving precision control. During implementation, first, the types and initial values ​​of the PCBA processing parameters to be adjusted are identified, including substrate etching depth parameters (initial value 0.05mm), component soldering pressure parameters (initial value 5N), reflow soldering temperature profile parameters (preheating zone temperature 80-120℃, constant temperature zone temperature 150-180℃, reflow zone temperature 220-250℃, cooling zone temperature 25-40℃, with durations of 60s, 90s, 30s, and 60s respectively), and cooling rate parameters (initial value 5℃ / s). Then, the adjustment direction is determined based on the deviation types identified in S4. For example, substrate bending deviation corresponds to adjusting the cooling rate parameter, soldering offset deviation corresponds to adjusting the soldering pressure parameter, and circuit strain deviation corresponds to adjusting the etching depth parameter. The adjustment range is determined based on the degree of deviation; slight deviations correspond to an adjustment range of 5%-10% of the initial value, moderate deviations... The adjustment range for minor deviations is 10%-20% of the initial value, and the adjustment range for severe deviations is 20%-30% of the initial value. Parameter adjustments are implemented in steps. After adjusting one parameter, other parameters remain unchanged. A small-batch trial production (10 pieces) is conducted using the PCBA processing equipment, and strain-displacement data of the trial-produced PCBAs is collected. The strain-displacement data of the PCBAs after trial production is compared with the target accuracy requirement (target strain-displacement deviation less than 0.005mm). If the requirement is not met, the parameters are adjusted again based on the trial production data, with the adjustment range being 50% of the previous adjustment range, until the strain-displacement deviation value of the trial-produced PCBAs meets the target accuracy requirement. Finally, the processing parameters that meet the accuracy requirements are determined and used as the standard parameters for subsequent batch production, ensuring that the strain-displacement accuracy of the batch-produced PCBAs is consistently up to standard.

[0030] S6. Upload the adjusted processing parameters, different model operation data and PCBA processing data to the color ultrasound PCBA full-process MES traceability platform to monitor and trace the PCBA processing process in real time and complete the PCBA processing of medical color ultrasound equipment.

[0031] Specifically, step S6 involves uploading the adjusted processing parameters, model operation data, and PCBA processing data to the color Doppler ultrasound PCBA full-process MES traceability platform to achieve real-time monitoring and data traceability of the PCBA processing process. This ensures the controllability of the processing process and provides data support for subsequent quality analysis and process optimization. During implementation, the data types and formats to be uploaded are first clearly defined. Adjusted processing parameters include the final determined etching depth, welding pressure, reflow soldering temperature profile, and cooling rate parameters, with numerical data format. Model operation data includes input data, output initial strain-displacement values, and model error values ​​for the S2 inversion model; interference source parameters and simulation result data for the S3 electromagnetic simulation algorithm; and input ultrasonic data and output deviation type data for the S4 classification model, with structured table format. PCBA processing data includes the operating status of the processing equipment (speed, temperature, pressure), processing time (processing time per PCBA), and quality inspection data (strain-displacement detection values, appearance inspection results), with real-time data stream format. Subsequently, a data transmission channel is established using industrial Ethernet, with a transmission rate set to 100Mbps and a data transmission delay controlled within 100ms. Data encryption technology is also employed. The encryption algorithm (AES-256) ensures secure data transmission. Various data are uploaded to the MES traceability platform according to a preset format. The platform stores the uploaded data in real time for a period of 3 years and establishes a data association mechanism, linking the unique identification code (using a QR code) of each PCBA with the corresponding processing parameters, model data, and process data to achieve "one item, one code" traceability. The platform monitors data changes in real time and has a data threshold alarm function. When processing parameters exceed the standard range or strain displacement detection values ​​exceed the target accuracy requirements, the platform automatically issues alarm signals (including audible and visual alarms and SMS alarms) to remind staff to handle the situation promptly. Staff can query the entire process data of any PCBA through the platform, including processing parameter adjustment records, model operation results, and quality inspection data, facilitating the tracing of the root cause of quality problems. Simultaneously, process optimization analysis based on historical data provides data support for subsequent PCBA processing technology improvements.

[0032] Preferably, the expression for the frequency domain eigenvalue decomposition shift inversion model in S2 is: ,in, The initial strain-displacement values ​​of the PCBA obtained from the inversion are... The number of dimensions in the original multi-dimensional data. For the first The original data of the dimension, It is a frequency domain transformation function. For the first Frequency domain feature weights of dimensional data For frequency domain characteristic adjustment coefficients, For the first Error values ​​of dimensional data, For the first Inversion coefficients of dimensional data The length of time for data collection. It is a time-frequency correlation function. For the first Feature frequencies of dimensional data This is the strain error correction function. For the first Standard deviation of dimensional data.

[0033] Specifically, the frequency domain feature decomposition displacement inversion model in step S2 ensures the accuracy and operability of the initial strain displacement inversion by clarifying the model expression and parameter meanings, providing reliable initial data for subsequent electromagnetic simulation. During implementation, the number of dimensions of the multi-dimensional original data in the model is determined according to the data type collected in S1, typically set to 4 (corresponding to substrate material, welding position, temperature field, and vibration frequency data); the frequency domain transformation function uses Fast Fourier Transform to ensure the accuracy of the conversion from time domain data to frequency domain data; the frequency domain feature weights of each dimension are set according to the degree of influence of different data on displacement inversion, with the weight of substrate material characteristic data ranging from 0.3 to 0.4, the weight of vibration frequency data ranging from 0.25 to 0.35, and the weights of welding position and temperature field data each ranging from 0.15 to 0.2; the frequency domain feature adjustment coefficient is set according to the stability of the processing environment, within a stable environment... The value is taken as 0.1-0.2 in normal environments and 0.2-0.3 in complex environments; the data acquisition time is determined according to the single processing cycle of PCBA, usually set to 300-600 seconds; the time-frequency correlation function is used to establish the mapping relationship between different time nodes and corresponding frequency characteristics to ensure the time correlation of frequency domain analysis; the strain error correction function combines the elastic modulus, thermal expansion coefficient and other parameters of PCBA material to compensate for possible errors in the inversion process; the initial strain displacement value calculated by this model must be controlled within ±0.005mm to ensure that it can accurately reflect the initial state of PCBA processing and provide accurate input for S3 electromagnetic simulation.

[0034] Preferably, the expression for the PCBA anti-interference electromagnetic simulation algorithm in S3 is: ,in, This refers to the electromagnetic interference intensity value during PCBA manufacturing. The number of electromagnetic interference sources, For the first The current intensity of each interference source, For the first The equivalent inductance of each interference source The permeability of free space, For the first The distance from each interference source to the sensitive area of ​​the PCBA The electromagnetic attenuation coefficient, The dielectric thickness between the interference source and the PCBA. The number of sensitive components on the PCBA For the first The coupling coefficient of each sensitive component, For the first The operating voltage of each sensitive component For the first The operating angular frequency of each sensitive component For the first The equivalent resistance of each sensitive component For the first The equivalent capacitance of each sensitive component.

[0035] Specifically, the PCBA anti-interference electromagnetic simulation algorithm in step S3 accurately quantifies the electromagnetic interference intensity, providing a quantitative basis for determining key interference parameters and ensuring the targeted nature of subsequent deviation identification and parameter adjustment. During implementation, the number of electromagnetic interference sources is determined based on the actual environment of the PCBA processing workshop, typically including interference sources from the equipment itself and surrounding equipment, with a quantity of 3-5. The current intensity of each interference source is determined according to its type: the current intensity of the equipment's own electrostatic interference source ranges from 0.1-0.5A, and the current intensity of the surrounding radio frequency equipment interference source ranges from 0.5-1A. The equivalent inductance is calculated based on the circuit structure of the interference source, with a value range of 10-100μH. The vacuum permeability uses a fixed value of 4π×10^-7H / m. The distance from the interference source to the PCBA sensitive area is determined through on-site measurement, with a value range of 0.1-0.5m. The electromagnetic attenuation coefficient is set according to the type of medium in the workshop (e.g., air, insulating materials), with a value set under air medium conditions. 0.05-0.1, 0.1-0.2 under insulating material medium; the number of sensitive components is determined according to the number of key components (such as chips and signal amplifiers) on the PCBA, usually 5-10; the coupling coefficient is set according to the distance between the component and the interference source and the shielding measures, with 0.3-0.5 for unshielded components and 0.1-0.2 for shielded components; the operating voltage, angular frequency, equivalent resistance and capacitance are determined according to the component datasheet, such as the chip operating voltage of 3.3-5V and the angular frequency of 10^6-10^8 rad / s; the electromagnetic interference intensity value calculated by this algorithm needs to be compared with the reference value under interference-free conditions. Interference sources with a deviation greater than 0.01V / m will be listed as key interference sources, providing data support for S4 deviation classification.

[0036] Preferably, the expression for the random forest ultrasonic feature classification model in S4 is: ,in, The results are for classifying strain displacement deviations. For the set of all deviation categories, The number of decision trees in the random forest. For the first The number of nodes in a decision tree. For the first The first decision tree The weight of each node, For the first The first decision tree Each node classifies the sample. The probability, The regularization coefficient is . The number of features detected by ultrasound. For the first Weights of each ultrasound feature, This is a correlation function between ultrasound features and deviation categories. For the first Raw data from ultrasound examination. For the first Each ultrasonic detection feature value.

[0037] Specifically, the random forest ultrasonic feature classification model in step S4 improves the accuracy and reliability of strain displacement deviation classification, providing clear guidance on deviation types for parameter adjustment in S5. During implementation, the number of decision trees in the random forest is determined based on the sample data volume; when the sample size is 1000-2000, the number of decision trees is set to 50-80. The number of nodes in each decision tree is determined based on the feature dimension; when the feature dimension is 20, the number of nodes is set to 30-50. Node weights are allocated according to feature importance: the weight for ultrasonic echo peak intensity is 0.2-0.3, the weight for propagation time is 0.15-0.25, and the weights for spectral width and energy are each 0.1-0.2. The sample decision probability is calculated using the decision tree node splitting rules to ensure a reasonable probability allocation for each sample at different nodes. The regularization coefficient is used to prevent model overfitting and is determined based on the training set error. The difference adjustment ranges from 0.01 to 0.1; the number of ultrasonic detection features is determined according to the detection accuracy requirements, usually 20-30 features are extracted, including time domain and frequency domain features; the feature weight is set according to the contribution of the feature to the deviation classification, such as the weight of the time domain peak feature corresponding to the substrate bending deviation is 0.3-0.4; the ultrasonic feature-deviation category association function is used to establish the correspondence between feature values ​​and deviation types, ensuring that different feature combinations can be accurately mapped to the corresponding deviation types; after the model training is completed, the classification accuracy should reach more than 95%, and the recognition accuracy of slight, moderate and severe deviations should not be less than 94%, 96% and 98% respectively, to ensure that accurate deviation type data is provided for S5 parameter adjustment.

[0038] Preferably, the calculation expression for adjusting the PCBA processing parameters of the medical color ultrasound equipment in S5 is as follows: ,in, These are the adjusted PCBA processing parameter values. These are the initial processing parameter values. Number of strain displacement deviation types For the first The parameter adjustment coefficients corresponding to the different types of deviations For the first Deviation values ​​for different types of deviations The maximum deviation value among all deviation types. The number of factors influencing electromagnetic interference. For the first The parameter correction coefficients corresponding to the various electromagnetic interference factors For the first The interference intensity of various electromagnetic interference factors This represents the maximum interference intensity among all electromagnetic interference factors.

[0039] Specifically, step S5 involves adjusting PCBA processing parameters to achieve precise quantitative adjustment of these parameters, effectively offsetting the impact of strain-displacement deviations and improving PCBA processing accuracy. During implementation, the initial processing parameter values ​​are determined based on PCBA design requirements and equipment specifications. The initial substrate etching depth is 0.04-0.06 mm, the initial soldering pressure is 4-6 N, and the initial cooling rate is 4-6 °C / s. The number of strain-displacement deviation types is determined based on common deviation types encountered in actual processing, typically 3-4 types (substrate bending, soldering offset, circuit strain, component detachment). The parameter adjustment coefficient for each deviation type is set according to the degree of deviation impact; the cooling rate adjustment coefficient for substrate bending deviation is 0.1-0.2, and the soldering pressure adjustment coefficient for soldering offset deviation is 0.15-0.25. Deviation values ​​are output through the S4 classification model, with minor deviations ranging from 0.005-0.01 mm and moderate deviations from 0.005-0.01 mm. 1-0.02mm, severe deviations exceed 0.02mm; the maximum deviation value is the maximum value among all deviation types, used to normalize the deviation value and ensure that the adjustment range calculation is reasonable; the number of electromagnetic interference influencing factors is set according to the number of key interference sources determined in S3, usually 2-3; the parameter correction coefficient corresponding to each interference factor is set according to the interference intensity, taking a value of 0.2-0.3 under strong interference and 0.1-0.2 under weak interference; the maximum interference intensity is the maximum value among all interference factors, used to normalize the interference intensity value; the adjusted processing parameters calculated by this expression need to be verified by small-batch trial production, with a trial production quantity of 10-20 pieces, and after successful verification, they are used as batch production parameters to ensure that the PCBA strain displacement deviation is less than 0.005mm.

[0040] Preferably, the data association expression for uploading data to the color Doppler ultrasound PCBA end-to-end MES traceability platform in S6 is: For the associated traceability data in the MES traceability platform, This is the time data for PCBA manufacturing process. This is the original collected data. For data association matrix, For the number of data types, For the first The traceability weight of each data type For data hash function, For the first Types of data.

[0041] Specifically, step S5 involves adjusting PCBA processing parameters to achieve precise quantitative adjustment of these parameters, effectively offsetting the impact of strain-displacement deviations and improving PCBA processing accuracy. During implementation, the initial processing parameter values ​​are determined based on PCBA design requirements and equipment specifications. The initial substrate etching depth is 0.04-0.06 mm, the initial soldering pressure is 4-6 N, and the initial cooling rate is 4-6 °C / s. The number of strain-displacement deviation types is determined based on common deviation types encountered in actual processing, typically 3-4 types (substrate bending, soldering offset, circuit strain, component detachment). The parameter adjustment coefficient for each deviation type is set according to the degree of deviation impact; the cooling rate adjustment coefficient for substrate bending deviation is 0.1-0.2, and the soldering pressure adjustment coefficient for soldering offset deviation is 0.15-0.25. Deviation values ​​are output through the S4 classification model, with minor deviations ranging from 0.005-0.01 mm and moderate deviations from 0.005-0.01 mm. 1-0.02mm, severe deviations exceed 0.02mm; the maximum deviation value is the maximum value among all deviation types, used to normalize the deviation value and ensure that the adjustment range calculation is reasonable; the number of electromagnetic interference influencing factors is set according to the number of key interference sources determined in S3, usually 2-3; the parameter correction coefficient corresponding to each interference factor is set according to the interference intensity, taking a value of 0.2-0.3 under strong interference and 0.1-0.2 under weak interference; the maximum interference intensity is the maximum value among all interference factors, used to normalize the interference intensity value; the adjusted processing parameters calculated by this expression need to be verified by small-batch trial production, with a trial production quantity of 10-20 pieces, and after successful verification, they are used as batch production parameters to ensure that the PCBA strain displacement deviation is less than 0.005mm.

[0042] Preferably, step S3 includes the following sub-steps: S31, extracting coordinate data, strain displacement rate of change data, and material characteristic parameter data of the PCBA sensitive area from the initial strain displacement state data obtained by inversion in S2, and converting these data into a data format recognizable by the PCBA anti-interference electromagnetic simulation algorithm to ensure that the data can accurately reflect the initial strain displacement state of the PCBA during the simulation process; S32, constructing a three-dimensional model consistent with the actual PCBA processing scenario in the simulation environment, including a PCBA substrate model, component model, and processing equipment surrounding environment model, and simultaneously loading the extracted and converted initial strain displacement state data into the three-dimensional model. In the simulation model, the simulation model is kept consistent with the actual processing state; S33, set the type parameters, intensity parameters, and location parameters of the electromagnetic interference source, where the interference source types include electrostatic interference source, radio frequency interference source, and electromagnetic radiation interference source. Add the corresponding interference source in the simulation model according to the set parameters to simulate the electromagnetic interference situation that may occur in the actual processing; S34, run the PCBA anti-interference electromagnetic simulation algorithm to obtain the strain displacement change data of the PCBA sensitive area under the action of different interference sources. By comparing and analyzing the strain displacement data under the interference-free state and the interference state, determine the electromagnetic interference calibration parameters that have a significant impact on the strain displacement accuracy of the PCBA.

[0043] Specifically, step S3 clarifies the specific implementation steps of the PCBA anti-interference electromagnetic simulation algorithm. Through a progressively layered operation, it ensures the accurate determination of key electromagnetic interference parameters, providing reliable data for subsequent deviation identification. During implementation, S31 first extracts key information from the initial strain-displacement state data inverted in S2, including coordinate data of the PCBA sensitive area (accuracy ±0.01mm), strain-displacement change rate data (sampling interval 0.1s), and material property parameter data (such as dielectric constant 4.2-4.8, elastic modulus 18-22GPa). This data is then converted into a binary format recognizable by the simulation algorithm. During the conversion process, the data error is controlled within 0.001, ensuring the accuracy of the data in the simulation. S32 constructs a 1:1 scale three-dimensional model in professional simulation software. The model includes the substrate (100mm×80mm×1.6mm), components (QFP / SOP / 0402 packages), and the surrounding environment of the processing equipment (such as the workbench and heat dissipation device). The data processed in S31 is then loaded into the model. The model is positioned to match the actual processing state with a matching degree of over 98%. S33 sets 3-5 types of interference source parameters: electrostatic interference source voltage 1-10kV, radio frequency interference source power 0.1-1W, and electromagnetic radiation interference source field strength 1-10V / m. The interference sources are distributed within a 0.5-2m range around the PCBA, set according to the actual workshop layout. S34 runs the simulation algorithm with a simulation duration of 10s and a step size of 0.1s, collecting strain and displacement data of sensitive areas in real time. The data is compared between the interference-free state (displacement deviation ≤0.002mm) and the interference state. Interference parameters (such as interference source type, intensity, and location) that cause displacement deviation >0.01mm are identified as key parameters. Through refined operations, simulation errors are eliminated to ensure no key interference parameters are missed, providing accurate interference data support for the deviation classification in S4.

[0044] Preferably, step S4 includes the following sub-steps: S41, collecting ultrasonic testing data corresponding to different strain displacement deviation types during the processing of PCBA for medical color Doppler ultrasound equipment. This data includes ultrasonic echo signal data, ultrasonic propagation time data, and ultrasonic amplitude data. The collected data is filtered to remove obviously abnormal data, retaining valid data for model training; S42, extracting features from the filtered ultrasonic testing data. The extracted features include the time-domain peak value feature, frequency-domain dominant frequency feature, and wavelet transform coefficient feature of the ultrasonic signal. The extracted features are associated with the corresponding strain displacement deviation type to construct the sample dataset required for model training; S43, dividing the sample... This dataset consists of a training set, a validation set, and a test set. The training set is used to train the random forest ultrasonic feature classification model, the validation set is used to adjust the model's hyperparameters, and the test set is used to evaluate the model's classification performance. Initial hyperparameters are set for the number of decision trees, the depth of the trees, and the node splitting threshold in the random forest. S44. The random forest ultrasonic feature classification model is trained using the training set. During the training process, the model's hyperparameters are continuously adjusted using the validation set until the model's classification performance on the validation set meets the preset requirements. Finally, the performance of the trained model is tested using the test set to ensure that the model can accurately classify and identify different strain displacement deviation types during PCBA processing.

[0045] Specifically, step S4 clarifies the complete implementation process of the random forest ultrasonic feature classification model from data preparation to model validation, ensuring the accuracy and reliability of the model's classification of deviation types. During implementation, S41 collects ultrasonic testing data for three common strain-displacement deviations (substrate bending, weld offset, and circuit strain), with no fewer than 1000 samples for each deviation type. The acquisition device is a 10-50MHz high-frequency ultrasonic testing instrument, with the probe 0.5mm away from the PCBA surface and moving at a speed of 1mm / s. The collected data includes echo signal intensity (range 0-5V), propagation time (range 0-10μs), and amplitude data (range 0-2V). During screening, data with a signal-to-noise ratio <30dB are removed, retaining at least 95% of the valid data. S42 extracts features from the valid data, extracting 20 features including time-domain peak value (accuracy 0.01V), frequency-domain dominant frequency (accuracy 1MHz), and wavelet transform coefficients (decomposition level 5). These features are then associated with deviation types through a data association table. Binding and constructing a sample dataset in CSV format for easy model reading; S43: Divide the training, validation, and test sets in a 7:2:1 ratio, set initial hyperparameters for the random forest, 50 decision trees, 10 tree depths, and a node splitting threshold of 0.001. Hyperparameter settings must balance model accuracy and training efficiency; S44: Train the model using the training set, adjusting hyperparameters every 100 iterations using the validation set until the model achieves a classification accuracy ≥95% on the validation set. Finally, test the model on the test set, requiring a recognition rate of ≥94% / 96% / 98% for mild / moderate / severe biases, respectively. Through standardized data processing and model training procedures, classification errors caused by data quality or improper parameter settings are avoided, ensuring the model can accurately identify bias types and providing a clear direction for parameter adjustment in S5.

[0046] Preferably, step S5 includes the following sub-steps: S51. Obtain the strain displacement deviation type data and corresponding deviation degree data identified in S4, and simultaneously retrieve the electromagnetic interference calibration parameter data determined in S3. Integrate these data into a parameter adjustment analysis dataset to provide data support for subsequent processing parameter adjustments; S52. Analyze the correlation between different strain displacement deviation types and PCBA processing parameters, and clarify the main influencing processing parameters corresponding to each deviation type. For example, the substrate etching depth parameter mainly affects the PCBA circuit accuracy and thus affects the strain displacement, while the welding pressure parameter affects the bonding degree between the components and the substrate and thus affects the strain displacement; S53. Based on the correlation analysis results and combined with the influence law of electromagnetic interference calibration parameters on processing parameters, preliminarily determine the adjustment direction and adjustment range of each processing parameter. For example, when the electromagnetic interference intensity is large, the reflow soldering temperature curve parameters need to be adjusted appropriately to reduce the interference effect; S54. Based on the preliminarily determined adjustment direction and adjustment range, combined with the processing parameter adjustment calculation expression in S5, accurately calculate the adjustment value of each processing parameter, and complete the adjustment of various parameters of the medical color ultrasound equipment PCBA processing.

[0047] Specifically, step S5 clarifies the specific implementation process for adjusting PCBA processing parameters. Through data integration, correlation analysis, range determination, and precise calculation, the processing parameters are precisely adjusted, effectively improving the strain displacement accuracy of PCBA. During implementation, S51 first acquires the deviation type data (such as substrate bending, welding offset) and deviation severity data (slight 0.005-0.01mm, moderate 0.01-0.02mm, severe >0.02mm) output by S4. Simultaneously, it retrieves the key electromagnetic interference parameters (such as interference source type and intensity) determined by S3. These data are then integrated into a parameter adjustment analysis dataset in Excel format. Data integrity must be 100% to avoid adjustment deviations due to missing data. S52 analyzes the correlation between deviation types and processing parameters. For example, substrate bending deviation is mainly related to the cooling rate (initial 5℃ / s), welding offset deviation is mainly related to the welding pressure (initial 5N), and line strain deviation is mainly related to the substrate etching depth (initial 0.05mm). The correlation analysis uses the Pearson correlation coefficient method, and the absolute value of the correlation coefficient must be >0.7 to ensure significant correlation. S53 determines the parameter adjustment direction based on the correlation. For example, if the substrate bending deviation requires a reduction in the cooling rate, or if the welding offset deviation requires an adjustment in the welding pressure, the adjustment range must be determined in conjunction with the electromagnetic interference intensity. Under strong interference (field strength > 5V / m), the adjustment range is 20%-30% of the initial value, and under weak interference (field strength ≤ 5V / m), it is 10%-20%. The adjustment range must balance accuracy and equipment operation safety. Based on the adjustment range, S54 combines the parameter adjustment calculation logic, substitutes the deviation level data and interference parameters, and accurately calculates the adjustment value. For example, with an initial cooling rate of 5℃ / s, slight deviation, and weak interference, the adjustment value is 5×(1-10%)=4.5℃ / s. After the calculation, it is necessary to verify whether the adjustment value is within the allowable operating range of the equipment (e.g., cooling rate 2-8℃ / s) to ensure that the parameters are executable. Through a systematic parameter adjustment process, the subjectivity and error of traditional experience-based adjustments are avoided, and quantitative and precise adjustment of processing parameters is achieved, effectively offsetting the influence of deviation and ensuring that the PCBA strain displacement accuracy meets the standards.

[0048] The frequency domain feature decomposition displacement inversion model in this invention is a mathematical model that uses multi-dimensional processing data to infer the initial strain displacement state of PCBA through frequency domain analysis. It transforms time domain data into frequency domain features to achieve high-precision inversion. The method first collects data on substrate material properties (dielectric constant 4.2-4.8, elastic modulus 18-22 GPa), welding position (positioning accuracy ±0.02 mm), temperature field (80-250℃), and vibration frequency (50-200 Hz) during PCBA processing using multiple sensors. Then, it performs a fast Fourier transform on the data (1024 sampling points, frequency resolution 0.1 Hz) to extract frequency domain features such as amplitude spectrum and phase spectrum. Subsequently, it constructs a multi-layer feedforward network model (5-layer structure, input layer matching feature dimension, hidden layer node number 64 / 32 / 16, output layer 1 node), trains it using gradient descent algorithm (learning rate 0.001, 1000 iterations, error ≤0.001), and completes the model construction by combining the strain error correction function (incorporating thermal expansion coefficient 13-17 ppm / ℃). Inputting the original data will output the initial strain displacement value (accuracy ±0.005 mm). This model provides a real and accurate initial strain and displacement benchmark for subsequent electromagnetic simulation, avoiding subsequent analysis deviations caused by misjudgment of the initial state. It breaks through the limitations of traditional single data inversion, solves the technical problem of difficulty in accurately quantifying the initial state of PCBA processing, and lays a reliable data foundation for the entire precision control process.

[0049] The PCBA anti-interference electromagnetic simulation algorithm is a simulation calculation method that quantifies the impact of electromagnetic interference on strain displacement accuracy during PCBA manufacturing and selects key interference parameters. It simulates and recreates the process of interference affecting the PCBA. The process involves first building a professional electromagnetic simulation environment (spatial mesh accuracy 0.1mm, frequency range 10kHz-1GHz), importing the PCBA 3D model (substrate 100mm×80mm×1.6mm, including QFP / SOP / 0402 packaged components) and the inverted initial strain displacement data; then setting 3-5 types of interference source parameters (electrostatic interference 1-10kV, radio frequency interference 0.1-1W, electromagnetic radiation interference 1-10V / m, interference source distance from PCBA 0.5-2m), and substituting vacuum magnetic permeability... The algorithm collects parameters such as the electromagnetic insulation coefficient (4π×10^-7H / m), electromagnetic attenuation coefficient (0.05-0.1 for air, 0.1-0.2 for insulating materials), and coupling coefficient of sensitive components (0.3-0.5 for unshielded components, 0.1-0.2 for shielded components). Then, it runs a 10-second simulation with a 0.1-second step size, collecting real-time data on strain and displacement changes in sensitive areas such as chip pins and signal lines. Finally, it compares the displacement data with and without interference, identifying interference parameters (type, intensity, and location) that cause displacement deviations >0.01mm as key parameters. This algorithm accurately locates the electromagnetic interference sources and key parameters affecting the strain and displacement accuracy of PCBAs, providing interference data support for subsequent deviation identification. It fills the technical gap of "invisible and difficult-to-quantify" electromagnetic interference in traditional PCBA processing, avoiding accuracy loss due to unknown interference and ensuring the stability and controllability of the PCBA processing environment.

[0050] The Random Forest Ultrasonic Feature Classification Model is a machine learning model based on ultrasonic testing data, which achieves accurate identification of PCBA strain displacement deviation types through the collaboration of multiple decision trees. It utilizes multi-feature fusion and ensemble learning to improve classification accuracy. The model first collects over 1000 samples of three types of deviations: substrate bending, welding offset, and circuit strain. Using a 10-50MHz high-frequency ultrasonic testing instrument (probe distance 0.5mm from PCBA, moving speed 1mm / s), it acquires echo signal intensity (0-5V), propagation time (0-10μs), and amplitude (0-2V) data. Data with a signal-to-noise ratio <30dB is filtered out (effective data percentage ≥95%). Then, 20 features (time domain peak value, frequency domain dominant frequency, wavelet coefficients, etc.) are extracted and classified according to a 7:1 ratio. The training, validation, and test sets were partitioned in a 2:1 ratio. A random forest with 50 decision trees (each tree 10 layers deep, node splitting using the Gini coefficient) was then constructed and trained using 5-fold cross-validation (regularization coefficient 0.01-0.1). Node weights were adjusted (echo peak feature weight 0.2-0.3) to ensure classification accuracy ≥95% (slight / moderate / severe deviation recognition rates ≥94% / 96% / 98%, respectively). Finally, electromagnetic interference parameters and real-time ultrasonic data were input, and the deviation type and degree were output. This model provides clear and precise guidance on deviation types for adjusting processing parameters, avoiding blind adjustments and replacing the inefficiency and high error rate of traditional manual identification. It achieves rapid and accurate determination of PCBA strain displacement deviation, improving the efficiency and reliability of precision control.

[0051] The color Doppler PCBA full-process MES traceability platform is an information management platform that integrates data from all stages of PCBA processing to achieve real-time monitoring and traceability. It constructs a closed-loop management system of "data acquisition-analysis-storage-traceability." It first establishes an industrial Ethernet transmission channel (100Mbps speed, latency ≤100ms, AES-256 encryption) to access data from multi-dimensional data acquisition units, model operation units, and processing control units. Then, it uploads adjusted processing parameters (etching depth 0.04-0.06mm, welding pressure 4-6N, etc.), model operation data (inversion results, interference intensity, classification results), and processing data (equipment speed, temperature, detection values) in a structured format. A "one item, one code" (QR code identification) association is established through a 6×6 data association matrix and stored in a distributed database (storage period 3-5 years). Simultaneously, it sets up a threshold alarm function (triggers audible and visual alarms + SMS alarms when parameters exceed range or displacement deviation >0.005mm), and develops a data query module (response time 1-3 seconds) to support querying full-process data by PCBA unique code. The platform enables real-time monitoring, end-to-end traceability, and rapid problem localization of processing data, providing data support for process optimization. It meets the stringent requirements of the medical industry for equipment compliance and traceability, solves the problems of traditional data being scattered and difficult to trace, ensures the stability of PCBA processing quality, and provides complete data basis for subsequent continuous process optimization.

[0052] like Figure 2As shown, a PCBA processing system for improving strain displacement accuracy in medical color ultrasound equipment is disclosed. This system is applied to a PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment. It includes: a multi-dimensional data acquisition unit connected to the PCBA processing equipment, which acquires data related to strain displacement during processing, including substrate material characteristics, component soldering positions, temperature field, and vibration frequency, and transmits the data to a data processing unit; a data processing unit connected to both the multi-dimensional data acquisition unit and the model building and running unit, receiving the raw data, performing format conversion and preliminary processing, transmitting the processed data to the model building and running unit, and simultaneously receiving the model running results transmitted from the model building and running unit; and a model building and running unit connected to both the data processing unit and the parameter adjustment unit, which constructs and runs a frequency domain feature decomposition displacement inversion model, a PCBA anti-interference electromagnetic simulation algorithm, and a random forest ultrasound feature decomposition model based on the processed data. The model classifies and transmits the inversion, simulation, and classification results to the parameter adjustment unit. The parameter adjustment unit, connected to the model building and running unit, the MES traceability platform unit, and the PCBA processing control unit, calculates the processing parameter adjustment values ​​based on the model results and preset rules, transmitting these values ​​to the PCBA processing control unit and simultaneously to the MES traceability platform unit. The PCBA processing control unit, connected to the parameter adjustment unit and the multi-dimensional data acquisition unit, receives the adjustment values ​​to control the operation of the processing equipment, performs parameter control, and transmits the equipment operating status data to the multi-dimensional data acquisition unit. The MES traceability platform unit, connected to the parameter adjustment unit, the model building and running unit, and the multi-dimensional data acquisition unit, receives the adjustment data, model running data, raw data, and processing process data, integrates, stores, and correlates them, enabling full-process data traceability and real-time monitoring.

[0053] A PCBA manufacturing method and system for improving strain displacement accuracy in medical color ultrasound equipment is disclosed. The method involves collecting multi-dimensional raw data, then using a frequency domain feature decomposition displacement inversion model to invert the initial strain displacement state. Subsequently, a PCBA anti-interference electromagnetic simulation algorithm is used to determine key electromagnetic interference parameters. Simultaneously, a random forest ultrasound feature classification model is employed to identify strain displacement deviation types. Finally, processing parameters are adjusted based on the analysis results, and the data is uploaded to a traceability platform. The entire process integrates multiple models with parameter adjustments and in-depth data traceability, comprehensively covering key factors affecting strain displacement accuracy, such as substrate material, welding pressure, and electromagnetic environment. This avoids the limitations of single-technology analysis and enables dynamic optimization of the processing through data closure, ensuring that each step precisely targets the accuracy improvement goal and effectively guaranteeing that the PCBA processing quality meets the requirements of high-resolution color ultrasound equipment.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A PCBA manufacturing method for improving strain displacement accuracy in medical color ultrasound equipment, characterized in that, Includes the following steps: S1. Collect multi-dimensional raw data related to strain and displacement during the processing of PCBA for medical color Doppler ultrasound equipment. This multi-dimensional raw data includes PCBA substrate material characteristics, component soldering positions, temperature field data, and vibration frequency data during processing. S2. Based on the collected multi-dimensional raw data, construct a frequency domain feature decomposition displacement inversion model. By performing frequency domain transformation and feature extraction on the multi-dimensional raw data, perform inversion analysis on the initial strain and displacement state during PCBA processing. S3. Using a PCBA anti-interference electromagnetic simulation algorithm, combined with the inverted initial strain and displacement state data, simulate potential electromagnetic interference sources during PCBA processing to determine the impact of electromagnetic interference on the PCBA strain and displacement accuracy. S4. Establish a random forest ultrasonic feature classification model, input the electromagnetic interference calibration parameters obtained from the simulation and the ultrasonic detection data during the PCBA processing into the model, and classify and identify different strain displacement deviation types during PCBA processing; S5. Based on the classification and identification results, adjust various parameters of the medical color ultrasound equipment PCBA processing, including substrate etching depth parameters, component welding pressure parameters, reflow soldering temperature curve parameters, and cooling rate parameters; S6. Upload the adjusted processing parameters, different model running data, and PCBA processing data to the color ultrasound PCBA full-process MES traceability platform to monitor and trace the PCBA processing process in real time, and complete the medical color ultrasound equipment PCBA processing.

2. The PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, The expression for the frequency domain eigenvalue decomposition shift inversion model in S2 is as follows: ,in, The initial strain-displacement values ​​of the PCBA obtained from the inversion are... The number of dimensions in the original multi-dimensional data. For the first The original data of the dimension, It is a frequency domain transformation function. For the first Frequency domain feature weights of dimensional data For frequency domain characteristic adjustment coefficients, For the first Error values ​​of dimensional data, For the first Inversion coefficients of dimensional data The length of time for data collection. It is a time-frequency correlation function. For the first Feature frequencies of dimensional data This is the strain error correction function. For the first Standard deviation of dimensional data.

3. The PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, The expression for the PCBA anti-interference electromagnetic simulation algorithm in S3 is as follows: ,in, This refers to the electromagnetic interference intensity value during PCBA manufacturing. The number of electromagnetic interference sources, For the first The current intensity of each interference source, For the first The equivalent inductance of each interference source The permeability of free space, For the first The distance from each interference source to the sensitive area of ​​the PCBA The electromagnetic attenuation coefficient, The dielectric thickness between the interference source and the PCBA. The number of sensitive components on the PCBA For the first The coupling coefficient of each sensitive component, For the first The operating voltage of each sensitive component For the first The operating angular frequency of each sensitive component For the first The equivalent resistance of each sensitive component For the first The equivalent capacitance of each sensitive component.

4. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, The expression for the random forest ultrasonic feature classification model in S4 is: ,in, The results are for classifying strain displacement deviations. For the set of all deviation categories, The number of decision trees in the random forest. For the first The number of nodes in a decision tree. For the first The first decision tree The weight of each node, For the first The first decision tree Each node classifies the sample. The probability, The regularization coefficient is . The number of features detected by ultrasound. For the first Weights of each ultrasound feature, This is a correlation function between ultrasound features and deviation categories. For the first Raw data from ultrasound examination. For the first Each ultrasonic detection feature value.

5. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, The calculation expression for adjusting the PCBA processing parameters of the medical color Doppler ultrasound equipment in S5 is as follows: ,in, These are the adjusted PCBA processing parameter values. These are the initial processing parameter values. Number of strain displacement deviation types For the first The parameter adjustment coefficients corresponding to the different types of deviations For the first Deviation values ​​for different types of deviations The maximum deviation value among all deviation types. The number of factors influencing electromagnetic interference. For the first The parameter correction coefficients corresponding to the various electromagnetic interference factors For the first The interference intensity of various electromagnetic interference factors This represents the maximum interference intensity among all electromagnetic interference factors.

6. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, The data association expression for uploading data to the color ultrasound PCBA full-process MES traceability platform in S6 is: For the associated traceability data in the MES traceability platform, This is the time data for PCBA manufacturing process. This is the original collected data. For data association matrix, For the number of data types, For the first The traceability weight of each data type For data hash function, For the first Types of data.

7. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, S3 includes the following steps: S31, extracting coordinate data, strain displacement rate of change data, and material characteristic parameter data of the PCBA sensitive area from the initial strain displacement state data obtained from S2, and converting these data into a data format recognizable by the PCBA anti-interference electromagnetic simulation algorithm to ensure that the data can accurately reflect the initial strain displacement state of the PCBA during the simulation process; S32, constructing a three-dimensional model consistent with the actual PCBA processing scene in the simulation environment, including a PCBA substrate model, component model, and processing equipment surrounding environment model, and loading the extracted and converted initial strain displacement state data into the three-dimensional model. To ensure the simulation model remains consistent with the actual processing state; S33, Set the type, intensity, and location parameters of the electromagnetic interference source, including electrostatic interference, radio frequency interference, and electromagnetic radiation interference sources. Add the corresponding interference sources to the simulation model according to the set parameters to simulate the electromagnetic interference that may occur during the actual processing; S34, Run the PCBA anti-interference electromagnetic simulation algorithm to obtain the strain displacement change data of the PCBA sensitive area under the action of different interference sources. By comparing and analyzing the strain displacement data under the interference-free state and the interference state, determine the electromagnetic interference calibration parameters that have a significant impact on the strain displacement accuracy of the PCBA.

8. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, S4 includes the following steps: S41, collecting ultrasound detection data corresponding to different strain displacement deviation types during the processing of PCBA for medical color Doppler ultrasound equipment. This data includes ultrasound echo signal data, ultrasound propagation time data, and ultrasound amplitude data. The collected data is filtered to remove obviously abnormal data, retaining valid data for model training; S42, extracting features from the filtered ultrasound detection data. The extracted features include the time-domain peak value feature, frequency-domain dominant frequency feature, and wavelet transform coefficient feature of the ultrasound signal. The extracted features are associated with the corresponding strain displacement deviation type to construct the sample dataset required for model training; S43, dividing the sample size... The dataset consists of a training set, a validation set, and a test set. The training set is used to train the random forest ultrasonic feature classification model, the validation set is used to adjust the model's hyperparameters, and the test set is used to evaluate the model's classification performance. Initial hyperparameters are set for the number of decision trees, the depth of the trees, and the node splitting threshold in the random forest. S44. The random forest ultrasonic feature classification model is trained using the training set. During training, the model's hyperparameters are continuously adjusted using the validation set until the model's classification performance on the validation set meets the preset requirements. Finally, the performance of the trained model is tested using the test set to ensure that the model can accurately classify and identify different strain displacement deviation types during PCBA processing.

9. A PCBA processing method for improving strain displacement accuracy in medical color ultrasound equipment according to claim 1, characterized in that, S5 includes the following sub-steps: S51. Obtain the strain displacement deviation type data and corresponding deviation degree data identified in S4, and simultaneously retrieve the electromagnetic interference calibration parameter data determined in S3. Integrate these data into a parameter adjustment analysis dataset to provide data support for subsequent processing parameter adjustments; S52. Analyze the correlation between different strain displacement deviation types and PCBA processing parameters, and clarify the main influencing processing parameters corresponding to each deviation type. For example, the substrate etching depth parameter mainly affects the PCBA circuit accuracy and thus affects the strain displacement, while the welding pressure parameter affects the bonding degree between the components and the substrate and thus affects the strain displacement; S53. Based on the correlation analysis results and combined with the influence law of electromagnetic interference calibration parameters on processing parameters, preliminarily determine the adjustment direction and adjustment range of each processing parameter. For example, when the electromagnetic interference intensity is large, the reflow soldering temperature curve parameters need to be adjusted appropriately to reduce the interference effect; S54. Based on the preliminarily determined adjustment direction and adjustment range, combined with the processing parameter adjustment calculation expression in S5, accurately calculate the adjustment value of each processing parameter, and complete the adjustment of various parameters of the medical color ultrasound equipment PCBA processing.

10. A PCBA processing system for improving strain displacement accuracy in medical color ultrasound equipment, characterized in that, This system is applied to a PCBA processing method for improving strain displacement accuracy in medical color Doppler ultrasound equipment as described in claim 1, comprising: a multi-dimensional data acquisition unit connected to the PCBA processing equipment, acquiring data related to strain displacement during processing, including substrate material characteristics, component welding positions, temperature field, and vibration frequency, and transmitting the data to a data processing unit; a data processing unit connected to both the multi-dimensional data acquisition unit and the model building and running unit, receiving raw data and performing format conversion and preliminary processing, transmitting the processed data to the model building and running unit, and simultaneously receiving the model running results transmitted therefrom; and a model building and running unit connected to both the data processing unit and the parameter adjustment unit, constructing and running a frequency domain feature decomposition displacement inversion model, a PCBA anti-interference electromagnetic simulation algorithm, and a random forest ultrasound feature classification model based on the processed data, and transmitting the inversion, simulation, and... The classification results are transmitted to the parameter adjustment unit. The parameter adjustment unit, connected to the model building and running unit, the MES traceability platform unit, and the PCBA processing control unit, calculates the processing parameter adjustment values ​​based on the model results and preset rules, and transmits these values ​​to the PCBA processing control unit and the MES traceability platform unit. The PCBA processing control unit, connected to the parameter adjustment unit and the multi-dimensional data acquisition unit, receives the adjustment values ​​to control the operation of the processing equipment and performs parameter control, while also transmitting the equipment operating status data to the multi-dimensional data acquisition unit. The MES traceability platform unit, connected to the parameter adjustment unit, the model building and running unit, and the multi-dimensional data acquisition unit, receives the adjustment data, model running data, raw data, and processing data, integrates, stores, and correlates them, enabling full-process data traceability and real-time monitoring.

Images