Modular manufacturing control method and system for high-density computing boards

Abstract

The application relates to the technical field of manufacturing control, in particular to a board card modular manufacturing control method and system facing high-density computing power, wherein the technical scheme comprehensively collects production data and historical use conditions of each module of the board card, provides rich and multi-dimensional data basis for subsequent analysis, respectively performs welding point abnormality prediction from a welding process and a use process, comprehensively considers influences of various factors on welding point quality through image analysis, multi-source data modeling and other means, avoids single index misjudgment, and improves accuracy and comprehensiveness of abnormality prediction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of manufacturing control, in particular to a board card modular manufacturing control method and system facing high-density computing power. BACKGROUND

[0002] In order to meet the increasing demand for computing power, the integration of board cards is becoming higher and higher, and modular manufacturing has become a mainstream trend. By combining modules with different functions, production efficiency can be improved, costs can be reduced, and subsequent maintenance and upgrades can be facilitated. However, this manufacturing method also brings a series of new challenges.

[0003] Firstly, existing abnormality prediction methods mostly only consider a single factor, such as temperature and pressure during welding, ignoring the dynamic changes of the welding points during use and the coupling effect of multiple physical fields. For example, in actual use, welding points are affected by mechanical vibration, thermal cycling, current load and other factors, and the interaction of these factors may cause problems such as fatigue fracture of welding points after a period of use, but existing prediction methods cannot accurately assess this potential risk.

[0004] Secondly, for the overall quality evaluation of board cards, there is currently a lack of comprehensive and scientific methods. Most evaluations are based on only partial key indicators or single-dimensional data, without fully considering the mutual relationship between welding quality and equipment operating status. For example, only welding process parameters are focused on while ignoring possible abnormal situations that may occur during long-term operation of the equipment, or only the electrical performance of the equipment is considered while ignoring the impact of welding point quality on overall performance. This one-sided evaluation method is likely to lead to misjudgment of the quality of board cards and cannot accurately reflect the true reliability and service life of board cards.

[0005] Finally, when board cards have faults or potential quality problems, how to reasonably allocate maintenance resources is the key to improving maintenance efficiency and reducing costs. Currently, maintenance decisions are mainly based on the severity of faults or the importance of equipment, but the degree of quality abnormality of welding points and their importance in board cards are not comprehensively considered, which may cause maintenance personnel to invest a lot of time and effort in the maintenance of some secondary welding points while ignoring the key welding points that have a greater impact on the overall performance of board cards, thereby failing to effectively avoid the occurrence of cascading failures and increasing the risk of system failure. In view of this, the present application proposes a board card modular manufacturing control method and system facing high-density computing power. SUMMARY

[0006] In order to overcome the defects and shortcomings proposed in the background art, the present application provides a board card modular manufacturing control method and system facing high-density computing power.

[0007] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0008] In a first aspect, the application provides a board card modular manufacturing control method for high-density computing power, comprising the following steps:

[0009] In step S1, production data of each module of the board card and historical use of each module of the corresponding board card are obtained, wherein the production data includes temperature and humidity, pressure conditions and image conditions of the welding points during the production process, and the historical use of each module of the corresponding board card is historical information transmission data of each welding point and corresponding use of mechanical operation conditions;

[0010] In step S2, welding point welding abnormality prediction is performed based on temperature and humidity, pressure conditions and image conditions of the welding points during the welding process;

[0011] In step S3, welding point use abnormality prediction is performed based on historical use of power transmission conditions of each welding point and corresponding use of mechanical operation conditions;

[0012] In step S4, quality prediction of the corresponding welding point is performed based on welding point welding abnormality prediction results and corresponding welding point use abnormality prediction results;

[0013] In step S5, maintenance judgment of the corresponding welding point is performed according to the quality prediction results of the corresponding welding point;

[0014] In step S6, maintenance of the corresponding welding head is performed according to the maintenance judgment results of the corresponding welding head.

[0015] In an implementation manner of the application, the production data is obtained through corresponding sensors, excessive or insufficient pressure of the welding head in contact with the welding point will cause welding quality to decrease, image data of the corresponding welding point is obtained through image sensors, three-dimensional images of the corresponding welding point are obtained to evaluate the quality of the welding point, welding quality is predicted through the corresponding welding point image and welding environment, historical use of each module of the corresponding board card includes current condition data of corresponding data transmission of each welding point of the corresponding board card in the corresponding use scene, transmission current flowing through the welding point will cause ablation effect to the welding point, resulting in damage to the welding point, and planned data amount transmission condition data and corresponding vibration condition data of the corresponding welding point position of the use equipment are obtained, including vibration amplitude and vibration frequency conditions, vibration of the equipment will have negative impact on the connection of the welding point, resulting in falling of the corresponding welding point, and under the combined influence of vibration and burning, the welding point connection is not firm and is easy to fall off.

[0016] In an implementation manner of the application, the welding point welding abnormality prediction in step S2 comprises the following specific steps:

[0017] S21, obtain three-dimensional image condition data of the welding point and standard three-dimensional image condition data corresponding to the welding point, and obtain the number condition and crack area condition of the surface cracks of the welding point, obtain a crack anomaly by comparing the total area condition of the cracks with a safety area of the corresponding cracks, obtain a dissimilarity degree of the two images of the three-dimensional image condition data of the welding point and the standard three-dimensional image condition data corresponding to the welding point, the dissimilarity degree is obtained by subtracting a similarity degree from 1, obtain a similarity degree of the two images, and obtain an image anomaly of the welding point by weighted summation of the dissimilarity degree and the crack anomaly;

[0018] S22, analyze the service life of the welding point according to the obtained image anomaly of the welding point, the temperature and humidity and pressure condition in the production process, the service life analysis of the welding point can be performed in a neural network manner, obtain the image anomaly of the welding point, the temperature and humidity and pressure condition in the production process, and the service life condition of the corresponding welding point under a standard test condition in a historical scene, construct a deep learning neural network model with the image anomaly of the welding point, the temperature and humidity and pressure condition in the production process as input and the service life condition of the welding point under the standard test condition as output, comprehensively analyze the production data such as the image anomaly, the temperature and humidity, and the pressure to avoid misjudgment of a single index and improve the comprehensiveness of quality evaluation, the neural network can capture a nonlinear relationship by training the neural network with historical data, the prediction accuracy is higher than that of a traditional statistical method, the temperature and humidity change may cause thermal stress or oxidation reaction, and the pressure fluctuation affects the welding contact resistance, which will change the microstructure of the welding point, there is a complex mapping relationship between the service life of the welding point and multiple factors, and the neural network can approximate the nonlinear function through hidden layer nodes;

[0019] S23, set a ratio of a corresponding service life safety value to the estimated service life condition as a welding point welding anomaly prediction result, wherein the corresponding service life safety value is a safety service life set during production of the corresponding board card, the process locates defects through image analysis, models and associates process parameters and service life through multiple source data, and finally outputs risks in the form of a safety ratio.

[0020] In an implementation manner of the present application, the welding point service anomaly prediction in the step S3 includes the following specific steps:

[0021] S31, obtain historical information transmission data of the corresponding welding point and corresponding mechanical operation conditions, and obtain data amount transmission condition data planned to be transmitted, by integrating the historical transmission data of the welding point and the mechanical operation records, long-term load patterns can be recognized, and the one-sidedness of relying only on instantaneous detection can be avoided, and the actual transmission amount is compared with the design expectation to determine whether the welding point is working long-term overload;

[0022] S32, historical current transmission data of the corresponding welding point is acquired, heat generation power of the corresponding welding point position is acquired based on the influence of current transmission and the resistance of the corresponding position welding point, a burn abnormality analysis result is obtained by comparing the heat generation power with a safety value corresponding to the heat generation power, real-time heat power is calculated by current and resistance, which directly reflects the local temperature rise risk of the welding point and is more sensitive than single temperature monitoring;

[0023] S33, operation of the corresponding mechanical is acquired, vibration abnormality analysis is performed on the corresponding welding point position of the use equipment through corresponding vibration data of the welding point position, wherein the vibration abnormality analysis manner is: a corresponding vibration abnormality analysis result is obtained by multiplying a comparison value of vibration frequency and a safety vibration frequency and a comparison value of vibration amplitude and a safety vibration amplitude, the safety threshold values of the two dimensions of frequency and amplitude are integrated, the vibration damage is quantified by the product, and the damage rate is doubled when the two cooperate;

[0024] S34, the welding point is analyzed abnormally by the equipment running through weighted summation of the burn abnormality analysis result and the vibration abnormality analysis result, heat power and vibration result are fused by weighting, and the failure risk of the welding point under real working conditions is more comprehensively evaluated;

[0025] S35, the importance of the welding point is acquired by a ratio of data transmission amount of the corresponding welding point to standard data transmission amount of the welding point, core welding points are identified by the data transmission amount ratio, and detection resources of high importance nodes are preferentially ensured.

[0026] In an implementation manner of the present application, the quality prediction of the corresponding welding point in the step S4 comprises the following specific contents:

[0027] S41, welding abnormality prediction results of the corresponding welding point and corresponding abnormality analysis results of the welding point by the equipment running are acquired;

[0028] S42, welding abnormality prediction results of the welding point and corresponding abnormality analysis results of the welding point by the equipment running are weighted and summed to obtain quality abnormalities of the corresponding welding point, the quality abnormalities of the welding point are quantified by weighted summation, and one-sided evaluation is avoided.

[0029] In an implementation manner of the present application, the maintenance judgment of the corresponding welding joint of the welding point in the step S5 comprises the following specific contents:

[0030] The quality anomaly of the corresponding welding point and the importance of the corresponding welding point are obtained, the importance of the corresponding welding point is standardized, and the standardized importance of the corresponding welding point is multiplied by the quality anomaly to obtain a repair value of the corresponding welding point position. The quality anomaly is compared with the corresponding quality anomaly standard value, and the welding point with the quality anomaly greater than or equal to the corresponding quality anomaly standard value is set as an abnormal welding point. The welding head at the position of the corresponding abnormal welding point is set as an abnormal welding head. The abnormal welding head is repaired according to the descending order of the repair value. A transmission instruction is transmitted to a maintenance port.

[0031] In a second aspect, the application further provides a board card modular manufacturing control system facing high-density computing power, comprising:

[0032] A data acquisition module acquires production data of each module of the board card and historical use of each module of the corresponding board card;

[0033] A welding anomaly prediction module predicts welding anomalies of welding points based on temperature, humidity, pressure during welding of the welding points, and image conditions of the welding points;

[0034] A welding point use anomaly prediction module predicts welding point use anomalies based on historical use of power transmission of each welding point and corresponding use of mechanical operation;

[0035] A quality prediction module predicts quality of corresponding welding points based on welding anomaly prediction results of the welding points and corresponding welding point use anomaly prediction results;

[0036] A maintenance judgment module judges maintenance of welding heads of corresponding welding points according to quality prediction results of the corresponding welding points;

[0037] A maintenance processing module maintains corresponding welding heads according to maintenance judgment results of the corresponding welding heads of the welding points;

[0038] A control module is configured to control operation of the data acquisition module, the welding anomaly prediction module, the welding point use anomaly prediction module, the quality prediction module, the maintenance judgment module, and the maintenance processing module.

[0039] In a third aspect, the application provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program that can be called by the processor, and the processor executes a board card modular manufacturing control method facing high-density computing power by calling the computer program stored in the memory.

[0040] In a fourth aspect, the application provides a computer readable storage medium storing instructions, which, when executed on a computer, cause the computer to execute a board card modular manufacturing control method facing high-density computing power.

[0041] Compared with the prior art, the application has the following advantages and beneficial effects:

[0042] The application comprehensively collects the production data and historical use of each module of the board card, provides rich and multi-dimensional data basis for subsequent analysis, respectively predicts the welding point anomaly from the welding process and the use process, avoids single index misjudgment by comprehensively considering the influence of various factors on the welding point quality through image analysis, multi-source data modeling and other means, and improves the accuracy and comprehensiveness of the anomaly prediction;

[0043] In the quality prediction link, the application quantitatively analyzes the welding anomaly and the analysis result of the equipment running anomaly, and then quantitatively analyzes the overall quality anomaly degree of the welding point, avoids one-sided evaluation, combines the quality anomaly and the importance of the welding point to obtain a repair value in the maintenance judgment, avoids relying on a single index, arranges the repair abnormal welding joint in descending order according to the repair value, realizes efficient allocation of maintenance resources, preferentially processes high-risk and high-importance welding points, avoids cascading failure, and reduces the system failure risk. BRIEF DESCRIPTION OF DRAWINGS

[0044] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments with reference to the attached drawings:

[0045] Figure 1 It is a schematic diagram of the overall process of the method embodiment 1 of the application;

[0046] Figure 2 It is a schematic diagram of the S2 step process of the method embodiment 1 of the application;

[0047] Figure 3 It is a schematic diagram of the S3 step process of the method embodiment 1 of the application;

[0048] Figure 4 It is a schematic diagram of the structure of the system embodiment 2 of the application. DETAILED DESCRIPTION

[0049] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the specific embodiments of the application will be described in detail below with reference to the drawings of the specification.

[0050] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be implemented in other ways different from the description, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited by the specific embodiments disclosed below.

[0051] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0052] Example 1

[0053] like Figures 1 to 3 As shown, this embodiment provides a modular manufacturing control method for high-density computing power boards, specifically including the following steps:

[0054] Step S1: Obtain the production data of each module of the board and the historical usage of each module of the corresponding board. The production data includes the temperature, humidity, pressure and solder joint images during the production process, while the historical usage of each module of the corresponding board includes the historical information transmission data of each solder joint and the corresponding mechanical operation.

[0055] In this embodiment, specifically, the production data in the data acquisition step is acquired through corresponding sensors. For example, the corresponding temperature and humidity data is acquired through a corresponding temperature and humidity sensor to obtain the temperature and humidity data at the solder joint location during the welding process. The pressure sensor acquires the pressure between the welding head and the solder joint during the welding process. Excessive or insufficient pressure between the welding head and the solder joint will cause a decrease in welding quality. The corresponding solder joint image data is acquired through an image sensor to obtain a three-dimensional image of the corresponding solder joint to evaluate the quality of the solder joint. The welding quality is estimated by combining the corresponding solder joint image and the welding environment. The historical usage data of each module of the corresponding board includes the corresponding... The data includes current data for each solder joint on the board in the corresponding application scenario. When the transmission current flows through the solder joint, it can cause ablation and damage to the solder joint. At the same time, the planned data volume requires transmission data and vibration data of the corresponding solder joint location of the equipment, including vibration amplitude and frequency. The vibration of the equipment can negatively affect the connection of the solder joint, causing the corresponding solder joint to fall off. Under the combined influence of vibration and ablation, the weak connection of the solder joint is prone to falling off. Therefore, this solution analyzes the abnormal situation of weak connection of the solder joint leading to falling off under the combined influence of vibration and ablation, and then evaluates whether the solder joint connection can meet the requirements.

[0056] Step S2: Based on the temperature, humidity, pressure and image of the weld joint during the welding process, predict welding anomalies.

[0057] In this embodiment, the weld joint welding anomaly prediction in step S2 specifically includes the following steps:

[0058] S21, obtain three-dimensional image condition data of the welding point and standard three-dimensional image condition data corresponding to the welding point, and obtain the number condition and crack area condition of the surface cracks of the welding point, obtain crack abnormalities through the comparison value of the total area condition of the cracks and the safe area of the corresponding cracks, obtain the degree of difference of the two images of the three-dimensional image condition data of the welding point and the standard three-dimensional image condition data corresponding to the welding point, the degree of difference is obtained by subtracting the degree of similarity from 1, the degree of similarity of the two images can be obtained by superimposing and averaging the degree of similarity of each pixel value, or can be obtained by dividing the volume occupied by the image of the three-dimensional image intersection by the volume occupied by the union of the two images, obtain the similarity of the two images, obtain the image abnormality of the welding point by weighted summation of the degree of difference and the crack abnormality, compare the actual total area of the welding point cracks with the safe area, which can intuitively quantify the potential defect risk of the welding point, avoid structural failure caused by accumulation of micro cracks, combine pixel similarity or volume overlap rate, quantitatively evaluate the deviation of the welding point and the standard model, identify abnormalities such as geometric deformation and material loss, based on the theory of fracture mechanics, the crack propagation rate is related to the stress intensity factor, and the total area exceeding the standard indicates that the fatigue life is reduced, the three-dimensional image difference reflects the morphology deviation of the welding point (such as virtual welding and pores), and the volume overlap rate (Jaccard similarity coefficient) is a general index for evaluating geometric consistency;

[0059] S22, according to the image abnormality of the obtained welding spot, the temperature and humidity and pressure in the production process, welding spot service life analysis is carried out, welding spot service life analysis can be carried out through the way of neural network, the image abnormality of welding spot in historical scene, the temperature and humidity and pressure in the production process, and the service life of corresponding welding spot under standard test condition are obtained, the deep learning neural network model with input of image abnormality of welding spot, temperature and humidity and pressure in production process and output of service life of welding spot under standard test condition is constructed according to the image abnormality of welding spot in historical scene, temperature and humidity, pressure condition in production process and service life of corresponding welding spot under standard test condition, the specific steps are: the historical data is divided into 9:1 training set and test set; 90% weight, bias training set is input into deep learning neural network model for training to obtain initial deep learning neural network model; 10% weight, bias test set is used to test the initial deep learning neural network model, and the output of the initial deep learning neural network model that meets the maximum service life condition accuracy is taken as the deep learning neural network model; the welding spot data obtained is input into the deep learning neural network model to obtain the service life condition, the production data such as image abnormality, temperature and humidity, pressure are comprehensively considered to avoid single index misjudgment, improve the comprehensiveness of quality evaluation, the neural network is trained through historical data, which can capture nonlinear relationship (such as the synergistic effect of temperature and humidity fluctuation on welding spot life), the prediction accuracy is higher than that of traditional statistical method, because temperature and humidity change may cause thermal stress or oxidation reaction, pressure fluctuation affects welding contact resistance, which will change the microstructure of welding spot (such as grain coarsening), there is a complex mapping relationship between welding spot life and multiple factors, and the neural network approximates this nonlinear function through hidden layer nodes;

[0060] The specific steps include: image abnormality information collection: collect image data of welding spots, which can come from industrial camera snapshot on production line, offline sampling shooting, etc. The image contains the appearance characteristics of the welding spot, such as whether the shape is regular, whether there are cracks, whether the surface is smooth, etc. These abnormal information may affect the service life of welding spot; production environment data collection: record the temperature and humidity and pressure in the production process, temperature and humidity sensor and pressure sensor can be installed in the key position of production equipment to collect data in real time. Different temperature and humidity and pressure conditions may cause internal structure difference of welding spot in the forming process, thereby affecting its service life; service life data collection: the service life of corresponding welding spot is obtained through standard test, standard test can simulate actual use scene, apply certain current, vibration, temperature change and other conditions to welding spot, record the time from the beginning of use to the failure of welding spot as the target output of the model;

[0061] Check if there are missing values in the collected data, if so, choose the appropriate processing method according to the characteristics of the data, for temperature and humidity and pressure data, the mean or median of the variable can be calculated to fill in the missing values; for image data, if part of the information is missing and cannot be repaired, consider deleting the sample directly; identify the outliers in the data, such as temperature and humidity or pressure data that appear obviously beyond the normal range, for such outliers, you can modify them according to the actual situation, if you can't modify them, delete them, divide the collected historical data into training set and test set according to the ratio of 9:1, the training set is used for the training of the model, so that the model can learn the rules in the data; the test set is used to evaluate the performance of the model on unseen data, to test the generalization ability of the model; for this kind of multi-type data input problem, use fully connected neural network, which consists of input layer, hidden layer and output layer, the number of neurons in the input layer is determined according to the number of input features, such as the sum of the dimensions of image features, temperature and humidity and pressure data, the hidden layer can be set to multiple layers, each layer contains a certain number of neurons, the neurons are connected through weights, which is used to extract deep features of the data. The output layer has only one neuron, which is used to predict the service life of the welding point; use activation function in the hidden layer, such as rectified linear unit, ReLU function can introduce non-linear factor, so that the model can learn more complex patterns, its calculation is simple, which can effectively alleviate the problem of gradient disappearance and improve the training efficiency of the model; choose mean square error as the loss function; mean square error measures the average square error between the predicted value and the true value of the model, which can reflect the accuracy of the model prediction, during the training process, the goal of the model is to minimize the value of the loss function, use Adam optimization algorithm to update the weights of the model, which can adaptively adjust the learning rate of each parameter, so that the model can converge faster during the training process, determine the number of training rounds and batch size; the number of training rounds indicates the number of times the model learns the entire training set, and the batch size indicates the number of samples used for training each time; appropriate training rounds and batch size can improve the training effect and efficiency of the model, and the test set is used to test the trained initial deep learning neural network model. Input the input data in the test set into the model to get the predicted results of the model, and then compare them with the true service life data to calculate evaluation indicators such as mean absolute error; mean absolute error reflects the average absolute deviation between the predicted value and the true value of the model, which can directly reflect the prediction accuracy of the model. In the process of multiple training, the hyperparameters of the model are adjusted, such as network structure, training rounds, batch size, etc., the performance of the model on the test set is observed, and the model with the smallest mean absolute error on the test set is selected as the final model, which has good accuracy and generalization ability in predicting the service life of the welding point.

[0062] S23, set the ratio of the corresponding service life safety value to the corresponding estimated service life condition as the solder joint welding abnormality prediction result, wherein the corresponding service life safety value is the safe service life set during production of the corresponding board card, the process locates defects through image analysis, models and associates process parameters and service life from multiple sources, and finally outputs risks in the form of a safety ratio;

[0063] Step S3, solder joint use abnormality prediction based on historical power transmission conditions of each solder joint and corresponding use mechanical operation conditions;

[0064] In this embodiment, specifically, the solder joint use abnormality prediction in step S3 includes the following specific steps:

[0065] S31, obtain historical information transmission data of the corresponding solder joint and corresponding use mechanical operation conditions, and obtain planned data volume transmission condition data, by integrating the historical transmission data of the solder joint and the mechanical operation record, long-term load patterns (such as cumulative damage caused by high-frequency transmission) can be identified, and the one-sidedness of relying only on instantaneous detection can be avoided, by comparing the actual transmission volume with the design expectation, it can be judged whether the solder joint is working long-term overload (such as current overload caused by data congestion), according to Joule's law, high-frequency or excess data transmission will increase the current, causing the solder joint to accumulate heat, accelerating material aging (such as tin whisker growth), and mechanical vibration may cause fretting wear, which will exacerbate solder joint crack propagation when combined with electrical load (such as vibration + current);

[0066] S32, obtain historical current transmission condition data of the corresponding solder joint, obtain generated heat power condition of the corresponding solder joint position based on the influence of current transmission and the resistance condition of the corresponding position solder joint, obtain burn abnormality analysis result through the comparison value of the generated heat power condition and the safety value corresponding to the generated heat power, calculate real-time heat power through current and resistance, which directly reflects the local temperature rise risk of the solder joint, and is more sensitive than single temperature monitoring (such as resistance change indicating microscopic cracks), the ratio of dynamic heat power to safety value can early warn burn or delamination caused by overload, the energy dissipation of current through the solder joint will increase the temperature, and local melting may occur if the heat dissipation is insufficient;

[0067] S33, obtain the corresponding use mechanical operation condition, and perform vibration abnormality analysis through the corresponding vibration condition data of the corresponding solder joint position of the use equipment, wherein the vibration abnormality analysis method is: multiply the comparison value of the vibration frequency and the safe vibration frequency, and the comparison value of the vibration amplitude and the safe vibration amplitude to obtain the corresponding vibration abnormality analysis result, the safety thresholds of the two dimensions of frequency and amplitude are comprehensively considered, the product quantifies the vibration hazard (high-frequency small amplitude or low-frequency large amplitude can be fatal), the product form converts different dimension parameters into dimensionless risk value, facilitating cross-device comparison, high-frequency vibration is easy to cause fatigue cracks, and large amplitude leads to plastic deformation; the damage rate is doubled when the two work together;

[0068] S34, the abnormal analysis result of the device operation on the welding point is obtained by weighted summation of the abnormal analysis result of the abnormal analysis result and the vibration abnormal analysis result, the weighted fusion of the thermal power (electrical heating field) and the vibration (mechanical field) result is obtained, the failure risk of the welding point under the real working condition is more comprehensively evaluated (such as "thermal-vibration" combined creep fracture), the weight is distributed according to different device characteristics (such as server focusing on current, industrial robot focusing on vibration), and diversified scenes are adapted;

[0069] S35, the importance of the welding point is obtained by the ratio of the data transmission amount of the corresponding welding point to the standard data transmission amount of the welding point, the core welding point is identified by the data transmission amount ratio (such as the mainboard power supply circuit), the detection resources of high importance nodes are preferentially guaranteed, the high importance welding point can shorten the detection period or increase the safety threshold (such as allowing lower thermal power ratio), the key welding point bears higher current density, and small defects may cause system-level failure;

[0070] Step S4, quality prediction of the corresponding welding point is performed based on the welding abnormality prediction result of the welding point and the corresponding welding point use abnormality prediction result;

[0071] In this embodiment, specifically, the quality prediction of the corresponding welding point in step S4 includes the following specific contents:

[0072] S41, the welding abnormality prediction result of the corresponding welding point and the corresponding device operation abnormality analysis result of the welding point are obtained;

[0073] S42, the welding abnormality prediction result of the welding point and the corresponding device operation abnormality analysis result of the welding point are weighted and summed to obtain the quality abnormality of the corresponding welding point, the welding abnormality weight is 0.6 and the device abnormality weight is 0.4 by weighted summation, the overall quality abnormality degree of the welding point is quantified, and one-sided evaluation is avoided;

[0074] Step S5, maintenance judgment of the corresponding welding point welding head is performed according to the quality prediction result of the corresponding welding point;

[0075] In this embodiment, specifically, the maintenance judgment of the corresponding welding point welding head in step S5 includes the following specific contents:

[0076] The quality anomaly of the corresponding welding point and the importance of the corresponding welding point are obtained, the importance of the corresponding welding point is standardized, and the standardized importance of the corresponding welding point is multiplied by the quality anomaly to obtain a repair value of the corresponding welding point position. The quality anomaly is compared with the corresponding quality anomaly standard value, and the welding point with a quality anomaly greater than or equal to the corresponding quality anomaly standard value is set as an abnormal welding point. The welding head at the position of the corresponding abnormal welding point is set as an abnormal welding head. The abnormal welding head is repaired according to the descending order of the repair value. A transmission instruction is transmitted to a maintenance port. The repair value integrates the fault severity (quality anomaly) and the position criticality (importance), guides the efficient allocation of maintenance resources, and the abnormal welding point with low importance may not need to be processed immediately (such as a decorative welding point of a non-load-bearing structure). The quality anomaly (technical index) and the importance of the welding point (reliability weight) are combined to avoid relying on a single index (such as only focusing on the failure rate while ignoring the key welding point). The importance is standardized (normalized to [0, 1]) to eliminate the dimensional differences between different welding points. If the welding point with a high repair value is not processed in time, it may cause a cascade failure (such as a high-current welding point that causes overheating, referring to the thermal failure chain reaction model). Rationality: If the quality anomaly is high but the importance is low (such as a redundant circuit welding point), the repair value is medium. If the quality anomaly is low but the importance is extremely high (such as a main control chip welding point), the repair value is still high.

[0077] Step S6, according to the maintenance judgment result of the corresponding welding head, the maintenance of the corresponding welding head is carried out.

[0078] In the embodiment, the maintenance port receives the maintenance instruction through a wired or wireless manner, and then automatically allocates maintenance personnel to maintain the corresponding welding head.

[0079] It should be noted that the weight parameter and the threshold value of the present application are obtained by historical experiments. A large amount of board production and use data is collected, including various abnormal conditions of the welding point, quality detection results, service life, etc. From the collected data, features related to weight determination are selected, such as temperature and humidity, pressure, image abnormalities during the welding process, current transmission, vibration conditions during use, etc. Machine learning algorithms such as linear regression, logistic regression, decision tree, etc. are used to train the model with the welding point quality as the target variable and the features as the independent variable. According to the trained model, the coefficients or importance scores of each feature are extracted, which are normalized and used as weights. For example, in the linear regression model, the coefficients of each feature can directly reflect its influence on the target variable. Normalizing these coefficients can obtain the weights. The production data of each module and the historical use of each module of the board are obtained. The historical data is imported into each step of the welding joint maintenance judgment analysis of the present embodiment, and the future welding joint maintenance time of the welding point is obtained. The use time of the welding point is sorted, and the analysis results and processing order results are imported into the matlab fitting software for data fitting to obtain the set parameter value that meets the highest welding point use time.

[0080] In the present embodiment, it should be noted that the present embodiment has the following advantages and advantages: the present embodiment comprehensively collects the production data and historical use of each module of the board, providing rich and multi-dimensional data basis for subsequent analysis. In terms of abnormal prediction, the welding point is predicted from the welding process and the use process. Through image analysis, multi-source data modeling, and other means, combined with physical theories such as fracture mechanics and Joule's law, the influence of various factors on the welding point quality is considered comprehensively, avoiding single index misjudgment, and improving the accuracy and comprehensiveness of abnormal prediction.

[0081] Embodiment 2

[0082] As Figure 4As shown, the embodiment provides a board card modular manufacturing control system for high-density computing power, which is used for the implementation of the board card modular manufacturing control method for high-density computing power in embodiment 1, and specifically includes: a data acquisition module, which acquires production data of each module of the board card and historical use of each module of the corresponding board card; a welding abnormality prediction module, which predicts welding abnormality of a welding point based on temperature and humidity, pressure during welding of the welding point, and image of the welding point; a welding point use abnormality prediction module, which predicts welding point use abnormality based on historical use of power transmission of each welding point and corresponding use of mechanical operation; a quality prediction module, which predicts quality of the corresponding welding point based on welding abnormality prediction result of the welding point and corresponding welding point use abnormality prediction result; a maintenance judgment module, which judges maintenance of a welding joint of the corresponding welding point based on the quality prediction result of the corresponding welding point; a maintenance processing module, which performs maintenance of the corresponding welding joint based on the maintenance judgment result of the corresponding welding joint of the welding point; and a control module, which is used for controlling operation of the data acquisition module, the welding abnormality prediction module, the welding point use abnormality prediction module, the quality prediction module, the maintenance judgment module, and the maintenance processing module, wherein the control module controls operation of the corresponding modules by using a PLC or a program existing technology control mode, and specific steps of each module of the embodiment of the system are the same as the specific steps of the method embodiment of embodiment 1, which will not be repeated here.

[0083] Embodiment 3

[0084] An electronic device according to an embodiment of the present application includes a processor and a memory, wherein the memory stores a computer program that can be invoked by the processor, and the processor executes a board card modular manufacturing control method for high-density computing power by invoking the computer program stored in the memory. It should be noted that all computer programs of the board card modular manufacturing control method for high-density computing power are implemented using C language.

[0085] Embodiment 4

[0086] The embodiment provides a computer readable storage medium, which stores an erasable computer program.

[0087] When the computer program runs on the computer device, the computer device executes the above-mentioned board card modular manufacturing control method for high-density computing power.

[0088] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions according to the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center through a wired network or / and a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, and the like, which includes one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0089] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed in the present application can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0090] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device, and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0091] In several embodiments provided in the present application, it should be understood that the disclosed system, device, and method can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of units is only one, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed units can be indirect coupling or communication connection through some interfaces, devices, or units, which can be electrical, mechanical, or other forms.

[0092] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0093] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present alone, or two or more units can be integrated in one unit.

[0094] In the description of the specification, the description referring to the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the present application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0095] The basic principles and main features of the present application and the advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for modular manufacturing control of a board card facing high-density computing power, characterized by, The method comprises the following steps: Obtain production data of each module of the board card and historical use of each module of the corresponding board card, wherein the production data comprises temperature and humidity, pressure condition and image condition of the welding spot in the production process, and the historical use of each module of the corresponding board card is historical information transmission data of each welding spot and corresponding use of mechanical operation condition; Based on the temperature and humidity, pressure condition and image condition of the welding spot in the welding process, welding spot welding abnormality prediction is performed; Based on the historical use of electrical power transmission of each welding spot and the corresponding use of mechanical operation condition, welding spot use abnormality prediction is performed; Based on the welding spot welding abnormality prediction result and the corresponding welding spot use abnormality prediction result, quality prediction of the corresponding welding spot is performed; According to the quality prediction result of the corresponding welding spot, maintenance judgment of the corresponding welding head of the welding spot is performed; and according to the maintenance judgment result of the corresponding welding head of the welding spot, maintenance of the corresponding welding head is performed; The welding spot welding abnormality prediction comprises the following specific steps: Obtain three-dimensional image condition data of the welding spot and standard three-dimensional image condition data of the corresponding welding spot, and obtain the number of surface cracks and crack area of the welding spot; the crack abnormality is obtained by comparing the total area of the cracks with the safety area of the corresponding cracks; the image abnormality of the welding spot is obtained by weighting and summing the difference degree of the three-dimensional image condition data of the welding spot and the standard three-dimensional image condition data of the corresponding welding spot and the crack abnormality; According to the welding spot image abnormality, the temperature and humidity and the pressure condition in the production process, welding spot service life analysis is performed, and the welding spot service life analysis is performed by means of neural network; The ratio of the corresponding service life safety value to the estimated service life condition is set as the welding spot welding abnormality prediction result; The welding spot use abnormality prediction comprises the following specific steps: Obtain historical information transmission data of the corresponding welding spot and corresponding use of mechanical operation condition, and obtain planned data amount transmission condition data; Obtain historical current transmission condition data of the corresponding welding spot, obtain generated heat power condition of the corresponding welding spot position based on the influence of current transmission and the resistance condition of the corresponding position welding spot, and obtain burning abnormality analysis result by comparing the generated heat power condition with the safety value of the corresponding generated heat power; Obtain the operation condition of the corresponding use of mechanical equipment, and perform vibration abnormality analysis on the corresponding vibration condition data of the corresponding welding spot position of the use equipment, wherein the vibration abnormality analysis method is: multiplying the comparison value of the vibration frequency and the safety vibration frequency with the comparison value of the vibration amplitude and the safety vibration amplitude to obtain the corresponding vibration abnormality analysis result; Weighted sum of the burning abnormality analysis result and the vibration abnormality analysis result is performed to obtain the abnormal analysis of the equipment operation on the welding spot; The ratio of the data transmission amount of the corresponding welding spot to the standard data transmission amount of the welding spot is obtained to obtain the importance of the welding spot.

2. The high-density compute oriented board card modular manufacturing control method of claim 1, wherein, The quality prediction of the corresponding welding spot comprises the following specific contents: Obtain welding spot welding abnormality prediction result of the corresponding welding spot and corresponding equipment operation abnormal analysis result of the welding spot. The welding abnormality prediction result of the welding spot and the abnormal analysis result of the corresponding equipment operation on the welding spot are weighted and summed to obtain the quality abnormality of the corresponding welding spot.

3. The high-density compute oriented board card modular manufacturing control method of claim 2, wherein, The maintenance judgment of the welding head of the corresponding welding spot includes the following specific contents: The quality abnormality of the corresponding welding spot and the importance of the corresponding welding spot are obtained, the importance of the corresponding welding spot is standardized, and then the quality abnormality is multiplied to obtain the repair value of the position of the corresponding welding spot. The quality abnormality is compared with the corresponding quality abnormality standard value, and the welding spot whose quality abnormality is greater than or equal to the corresponding quality abnormality standard value is set as an abnormal welding spot. The welding head at the position of the corresponding abnormal welding spot is set as an abnormal welding head. The abnormal welding head is repaired according to the descending order of the repair value. The transmission instruction is transmitted to the maintenance port.

4. The high-density compute oriented board card modular manufacturing control method of claim 3, wherein, The construction of the neural network includes the following specific steps: obtaining the image abnormality of the welding spot in the historical scene, the temperature and humidity and pressure conditions in the production process, and the service life of the corresponding welding spot under the standard test condition. A deep learning neural network model is constructed, with the image abnormality of the welding spot in the historical scene, the temperature and humidity and pressure conditions in the production process, and the service life of the corresponding welding spot under the standard test condition as the input, and the service life of the welding spot under the standard test condition as the output. The specific steps are as follows: divide the historical data into a 9:1 training set and a test set; input the 90% weight and bias training set into the deep learning neural network model for training to obtain an initial deep learning neural network model; test the initial deep learning neural network model using the 10% weight and bias test set, and output the initial deep learning neural network model output that meets the maximum service life condition accuracy as the deep learning neural network model.

5. The high-density compute oriented board card modular manufacturing control method of claim 4, wherein, The historical use of each module of the board card includes the current condition of each welding spot of the corresponding board card in the corresponding use scene, including the planned data amount transmission condition data and the corresponding vibration condition data of the corresponding welding spot position of the use equipment, including the vibration amplitude and vibration frequency condition.

6. A high-density computing oriented board card modular manufacturing control system for implementing the high-density computing oriented board card modular manufacturing control method of any one of claims 1-5, characterized by, The system includes: A data acquisition module acquires the production data of each module of the board card and the historical use of each module of the corresponding board card; A welding abnormality prediction module predicts welding abnormality of the welding spot based on the temperature and humidity, pressure conditions and image conditions of the welding spot during welding; A welding spot use abnormality prediction module predicts welding spot use abnormality based on the historical use of each welding spot and the corresponding use of mechanical operation; A quality prediction module predicts the quality of the corresponding welding spot based on the welding abnormality prediction result of the welding spot and the welding spot use abnormality prediction result; A maintenance judgment module judges the maintenance of the welding head of the corresponding welding spot according to the quality prediction result of the corresponding welding spot; A maintenance processing module maintains the corresponding welding head according to the maintenance judgment result of the corresponding welding head of the welding spot; A control module is configured to control the operation of the data acquisition module, the welding abnormality prediction module, the welding spot use abnormality prediction module, the quality prediction module, the maintenance judgment module and the maintenance processing module.

7. An electronic device comprising: A processor and a memory, wherein the memory has stored a computer program which can be invoked by the processor; characterized in that the processor, by invoking the computer program stored in the memory, executes the plate card modular manufacturing control method for high-density computing power as claimed in any one of claims 1-5.

Images