MES (Manufacturing Execution System) for automatically adjusting server manufacturing process based on current network fault

Abstract

The invention relates to the technical field of fault analysis, in particular to an MES (Manufacturing Execution System) for automatically adjusting a server manufacturing process based on a current network fault, which comprises a fault data processing module, a fault trend prediction module, a risk component identification module, a propagation path tracking module and a parameter adjustment module. According to the method, continuous statistical process monitoring is carried out on multi-dimensional quality indexes such as the first-time power-on fault rate, the annual fault rate and the existing network batch problem of the server, the abnormal fluctuation trend of the quality indexes can be recognized in advance before an isolated fault event occurs, the system does not passively respond to the occurring fault any more, and the system reliability is improved. Through fusion analysis of environment and operation factors such as load and memory utilization rate during server operation and temperature and humidity of a data center where the server is located, complex combination conditions causing quality indexes to deviate from a normal range are deeply mined. And the specific probability and the time window of future fault occurrence are calculated in a quantization mode.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault analysis, and in particular to a MES system for automatically adjusting server manufacturing processes based on network faults. BACKGROUND

[0002] The technical field of fault analysis refers to the technical activities of identifying and locating abnormal conditions of servers and related equipment in the process of running in an information-based manufacturing system, including real-time collection of server fault data in the running environment, classification analysis of fault characteristics, establishment of fault traceability in combination with process links, and interaction between analysis conclusions and manufacturing execution systems.

[0003] Among them, the method of adjusting the manufacturing process based on the network fault refers to when the server fails in the network operation, the manufacturing enterprise usually reads and compares the server log file to obtain the fault code or event identifier, and then modifies or adjusts the manufacturing process in the MES system according to the parameter relationship set in the process standard library.

[0004] The prior art mainly relies on the interpretation of server log files after the occurrence of faults, and this method has obvious limitations when dealing with complex or systematic problems. The core deficiency is the lag of reaction and the one-sidedness of analysis, because the technology must wait for the appearance of a clear fault code or event identifier before starting the adjustment process, and it cannot predict the performance degradation or index anomaly in the embryonic state, resulting in the loss of the best opportunity to prevent problems. At the same time, it only adjusts based on the preset relationship between a single fault code and process parameters, ignoring the deep reasons that the fault may be triggered by multiple factors such as environmental factors, running load and specific batch components. For example, when the network server has batch faults due to a certain brand of memory, the existing technology may only adjust the soldering temperature parameter of a general motherboard according to the memory error code in the log, but this adjustment does not identify that the root cause of the fault is that the specific brand of memory is more prone to virtual welding in a high-temperature and high-load environment, so the process adjustment lacks pertinence and cannot eliminate hidden dangers, ultimately causing the same nature of the fault to occur repeatedly in different batches of products, causing greater quality costs. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art and to provide a MES system for automatically adjusting server manufacturing processes based on network faults.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a MES system for automatically adjusting server manufacturing processes based on network faults comprises:

[0007] a fault data processing module, which counts three indexes of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server, establishes an Xbar-R graphical sequence, screens abnormal indexes in the Xbar-R graphical sequence, and generates a fault early warning instruction;

[0008] a fault trend prediction module, which analyzes changes of abnormal indexes corresponding to the fault early warning instruction over time, predicts an occurrence probability and a time of a future fault of the server in the live network as a fault trend, and generates a fault trend prediction result;

[0009] a risk component identification module, which identifies a future fault trend of the server in the live network that will have a negative impact on the overall quality of the server in the live network according to the fault trend prediction result, captures corresponding key components, calculates a comprehensive failure rate of the components, and marks a quality risk component;

[0010] a propagation path tracking module, which tracks a fault propagation path of the quality risk component in real time, analyzes whether a fault of the quality risk component will have an impact on other components, and generates a fault propagation path graph;

[0011] a parameter adjustment module, which automatically adjusts a server manufacturing process parameter according to the fault trend prediction result of the server in the live network and the fault propagation path graph of the quality risk component, and generates a new process configuration table.

[0012] As a further scheme of the present application, the fault early warning instruction specifically is an abnormal value of a fault index, a fault occurrence early warning, and fault analysis information, the fault trend prediction result includes a fault occurrence probability, a fault occurrence time, and a fault type prediction, the quality risk component specifically is a component with a failure frequency exceeding a frequency threshold, a fault type related component, and a key performance component, the fault propagation path graph includes a propagation path, an impact range, and an impact component, and the new process configuration table specifically is an adjusted process parameter, an optimized production process, and an improved production process.

[0013] As a further scheme of the present application, the fault data processing module includes:

[0014] a historical index modeling submodule, which collects three index data of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server in a specified period, establishes an Xbar-R graphical sequence, calculates a center line and upper and lower control limits of the Xbar-R graphical sequence, and establishes an index historical data control limit;

[0015] a live network index monitoring submodule, which obtains current values of three indexes of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server in a current period, compares the current values of each index with the index historical data control limit, screens indexes all exceeding the control limit, and obtains abnormal indexes;

[0016] The fault early warning generation submodule generates a fault early warning instruction according to a preset instruction rule base for each abnormal index corresponding early warning level and disposal plan.

[0017] As a further scheme of the present application, the fault trend prediction module comprises:

[0018] The abnormal index tracing submodule analyzes the abnormal index corresponding to the fault early warning instruction, collects data of the abnormal index changing over time within a specified period, integrates environmental factors and operation factors affecting the index change, performs trend analysis and qualitative prediction, and establishes a set of index change influence factors.

[0019] The fault path construction submodule calculates the information gain rate of the environmental factors and operation factors in the set of index change influence factors by using the C4.5 decision tree algorithm, selects the factor with the highest gain rate as a decision node to recursively partition the data set, until all branches reach a preset stopping condition, connects all decision nodes and branches, and generates a fault trend path.

[0020] The fault trend quantification submodule locates the decision node at the end of the path according to the fault trend path, counts and calculates the occurrence frequency and time distribution of the fault mode represented by the decision node, quantifies the occurrence probability and time of the future fault of the online server as the fault trend, and generates a fault trend prediction result.

[0021] As a further scheme of the present application, the integration of the environmental factors and operation factors affecting the index change comprises collecting the temperature and humidity of the server room where the online server is located as the environmental factors, and collecting the central processor load rate, memory usage rate, and disk read / write rate of the online server as the operation factors.

[0022] As a further scheme of the present application, the stopping of the partitioning of the current decision node until all branches reach a preset stopping condition is specifically stopping the partitioning of the current decision node when any of the following conditions is met: all data samples in the current decision node belong to the same fault category, the number of data samples in the current decision node is lower than a minimum sample number threshold, or the depth of the decision tree reaches a maximum depth threshold.

[0023] As a further scheme of the present application, the risk component identification module comprises:

[0024] The negative trend screening submodule obtains the future trends of the first power-on failure rate, annualized failure rate, and online batch problem of the three indexes in the fault trend prediction result, identifies and screens the trends that will have a negative impact on the overall quality of the server according to a preset quality risk judgment benchmark, and obtains the future fault trend of the online server.

[0025] A key component positioning sub-module traces original fault data associated with each of the future fault trends of the existing network server, performs secondary qualitative analysis on the original fault data, captures hardware or software that commonly appears in multiple data records and has an impact on quality as a key component, and obtains a list of key components to be evaluated;

[0026] A component risk marking sub-module collects fault records corresponding to the key components in the list of key components to be evaluated, counts the occurrence frequency of the key components under each indicator, calculates the comprehensive failure rate of the components, compares the comprehensive failure rate with a risk component threshold, and marks all components that exceed the risk component threshold as quality risk components.

[0027] As a further scheme of the present application, the propagation path tracking module comprises:

[0028] A propagation characteristic calculation sub-module collects the deployment positions of each of the quality risk components in the existing network server, counts associated fault records, calculates the frequency and time interval of the impact of each quality risk component on other components after failure, quantifies the quality risk component propagation probability and propagation delay, and generates component failure propagation data;

[0029] A potential impact analysis sub-module dynamically analyzes the potential impact range of the fault according to the potential impact of the quality risk component propagation probability and propagation delay on other components in the component failure propagation data;

[0030] A propagation path visualization sub-module establishes a fault transmission relationship between each component based on the fault potential impact range, takes the quality risk component as a starting node, takes the affected other components as target nodes, takes the potential impact relationship as a connection path, and generates a fault propagation path graph.

[0031] As a further scheme of the present application, the parameter adjustment module comprises:

[0032] A process correlation analysis sub-module calls the fault trend prediction result and the fault propagation path graph, analyzes the fault cause of the quality risk component and the time point at which the fault occurs in the manufacturing process, identifies a key manufacturing process link that affects the quality of the server, and generates a process parameter item to be adjusted;

[0033] A parameter adjustment list generation sub-module queries a fault phenomenon and process adjustment table in an MES system according to the process parameter item to be adjusted, automatically matches and obtains a corresponding adjustment plan, and generates a process parameter adjustment list;

[0034] A process configuration updating sub-module updates existing server manufacturing process parameter settings in the MES system based on the process parameter adjustment list, integrates and formats all adjusted parameters, and generates a new process configuration table.

[0035] Compared with the prior art, the application has the advantages and positive effects that:

[0036] In the application, through continuous statistical process monitoring on the first power-on failure rate, annualized failure rate and multi-dimensional quality indicators of the server batch problems in the existing network, the abnormal fluctuation trend of the quality indicators can be identified in advance before the occurrence of isolated failure events. Instead of being passive in responding to the occurred failures, the system is changed to actively predict potential quality risks. Through the fusion analysis of the load, memory usage rate and environmental and operating factors such as temperature and humidity of the data center where the server is located during the server operation, the complex combination conditions causing the quality indicators to deviate from the normal range are deeply mined. The specific probability and time window of future failure occurrence are calculated in a quantitative manner, and the key hardware or software causing the problem is accurately located. Further, it is revealed how the failure of a single risk component gradually transmits and affects other associated components to form a panoramic cognition of the chain reaction of the failure. Ultimately, the complete analysis from trend warning to root cause positioning is directly associated to the specific process parameters in the manufacturing link to realize the automatic, accurate and closed-loop optimization of the production process configuration, and fundamentally solves the problem of blindness in process adjustment caused by the broken cause-effect chain. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is a system flowchart of the application;

[0038] Figure 2 is a flowchart of the fault data processing module of the application;

[0039] Figure 3 is a flowchart of the fault trend prediction module of the application;

[0040] Figure 4 is a flowchart of the risk component identification module of the application;

[0041] Figure 5 is a flowchart of the propagation path tracking module of the application;

[0042] Figure 6 is a flowchart of the parameter adjustment module of the application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical scheme and advantages of the application more clear and explicit, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0044] Please refer to Figure 1 A MES system for automatically adjusting server manufacturing process based on network failure includes:

[0045] The fault data processing module calculates the first power-on failure rate, annual failure rate and batch problem in the existing network of the server, establishes an Xbar-R graphical sequence, screens abnormal indexes in the Xbar-R graphical sequence, and generates a fault warning instruction;

[0046] The fault trend prediction module analyzes the change of the abnormal indexes corresponding to the fault warning instruction over time, predicts the occurrence probability and time of future faults of the server in the existing network as a fault trend, and generates a fault trend prediction result;

[0047] The risk component identification module identifies the future fault trend of the server in the existing network that will have a negative impact on the overall quality of the server in the existing network according to the fault trend prediction result, captures the corresponding key components, calculates the comprehensive failure rate of the components, and marks the quality risk components;

[0048] The propagation path tracking module tracks the fault propagation path of the quality risk components in real time, analyzes whether the faults of the quality risk components will affect other components, and generates a fault propagation path diagram;

[0049] The parameter adjustment module automatically adjusts the server manufacturing process parameters according to the fault trend prediction result of the server in the existing network and the fault propagation path diagram of the quality risk components, and generates a new process configuration table.

[0050] The fault warning instruction specifically refers to the abnormal value of the fault index, the fault occurrence warning, and the fault analysis information. The fault trend prediction result includes the fault occurrence probability, the fault occurrence time, and the fault type prediction. The quality risk component specifically refers to the component with a fault frequency exceeding a frequency threshold, the fault type related component, and the key performance component. The fault propagation path diagram includes the propagation path, the influence range, and the affected components. The new process configuration table specifically refers to the adjusted process parameters, the optimized production process, and the improved production process.

[0051] Please refer to Figure 2 , the fault data processing module includes:

[0052] The historical index modeling submodule collects the first power-on failure rate, annual failure rate, and batch problem in the existing network of the server in a specified period, establishes an Xbar-R graphical sequence, calculates the center line and upper and lower control limits of the Xbar-R graphical sequence, and establishes an index historical data control limit;

[0053] First, collect the first power-on failure rate, annualized failure rate, and batch problem in the network of the specified server model in the past consecutive production cycles. For example, for "A type" servers, collect the weekly data of the above three indicators from the 9th week of 2024 to the 8th week of 2025, a total of 52 weeks. The first power-on failure rate refers to the proportion of failed equipment in the total test equipment during the first power-on test stage after the server leaves the factory; the annualized failure rate is the annual failure rate calculated based on the number of failures within a certain period of server operation; the batch problem in the network refers to the number of concentrated failure events in the same batch and same model of servers due to the same reason. The collected data is used to establish the Xbar-R control chart sequence of each indicator. Specifically, divide the continuous 52-week data into 13 subgroups, each containing 4 consecutive weeks of data. Taking the annualized failure rate as an example, the first subgroup contains the annualized failure rate data from the 1st week to the 4th week, which are 0.52%, 0.55%, 0.53%, and 0.56% respectively. Calculate the mean Xbar1 of the first subgroup, which is (0.52+0.55+0.53+0.56) divided by 4, resulting in 0.54%. Calculate the range R1 of the first subgroup, which is the difference between the maximum value 0.56% and the minimum value 0.52% in the subgroup, resulting in 0.04%. Repeat the above calculation process to obtain the mean values (Xbar1 to Xbar13) and ranges (R1 to R13) of all 13 subgroups. Then, calculate the average Xbar-bar of all subgroup means as the center line of the Xbar control chart. Assuming that the calculated Xbar-bar is 0.58%. Similarly, calculate the average R-bar of all subgroup ranges, assuming the result is 0.05%. According to the preset control chart coefficient table, for a subgroup with a sample size of 4, the values of coefficients A2, D3, and D4 are 0.729, 0, and 2.282 respectively. These coefficients are standard values fitted based on a large amount of experimental data according to statistical process control theory, and their setting basis is to ensure that the control limit can cover the normal fluctuations of the process at a confidence level of 99.73%. The upper control limit UCLx of the Xbar chart is calculated by adding the A2 coefficient 0.729 to the R-bar value 0.05% of the center line 0.58%, resulting in 0.61645%. The lower control limit LCLx of the Xbar chart is calculated by subtracting the A2 coefficient 0.729 from the R-bar value 0.05% of the center line 0.58%, resulting in 0.54355%. The center line of the R chart is R-bar, which is 0.05%. The upper control limit UCLr of the R chart is calculated by multiplying the D4 coefficient 2.282 by the R-bar value 0.05%, resulting in 0.1141%. The lower control limit LCLr of the R chart is calculated by multiplying the D3 coefficient 0 by the R-bar value 0.05%, resulting in 0%.Through the above calculation, the historical data control limit finally established for the annual failure rate index is: the center line of the Xbar chart is 0.58%, the upper control limit is 0.61645%, and the lower control limit is 0.54355%; the center line of the R chart is 0.05%, the upper control limit is 0.1141%, and the lower control limit is 0%. The same process is applied to the first power-on failure rate and the online batch problem indicators to establish the historical data control limit of all indicators.

[0054] The online indicator monitoring submodule obtains the current values of the first power-on failure rate, the annual failure rate, and the online batch problem of the online server in the current period, compares the current values of each indicator with the historical data control limit of the indicator, filters all indicators that exceed the control limit, and obtains abnormal indicators;

[0055] First, the current values of the indicators of the "A type" server in the current period, i.e., the 9th week of 2025, are obtained. Specifically, through the production management system and the operation and maintenance monitoring platform, the first power-on failure rate of this week is obtained as 0.45%, the annual failure rate is 0.63%, and the online batch problem is 2. After obtaining the current values, the current values of each indicator are compared with the corresponding historical data control limit. For the annual failure rate indicator, its current value 0.63% is compared with the Xbar chart control limit established in the previous step. The upper control limit of the Xbar chart is 0.61645%, and the lower control limit is 0.54355%. Since the current value 0.63% is greater than the upper control limit 0.61645%, the indicator is determined to exceed the control limit. Then, the first power-on failure rate indicator is compared. Assuming that the upper control limit of the first power-on failure rate Xbar chart is 0.50% and the lower control limit is 0.30% through the same method. The measured value of the current first power-on failure rate is 0.45%, which is between 0.30% and 0.50%, so it does not exceed the control limit. Finally, the online batch problem indicator is compared. Assuming that the upper control limit of its Xbar chart is 1.5 and the lower control limit is 0.2. The current value is 2, which is greater than the upper control limit 1.5, so the indicator is also determined to exceed the control limit. After comparing all indicators, all indicators that exceed their corresponding historical data control limits are filtered out. In this embodiment, the current values of the annual failure rate and the online batch problem both exceed their respective upper control limits. Therefore, the final set of abnormal indicators is {annual failure rate, online batch problem}.

[0056] The fault warning generation submodule matches the warning level and the disposal plan corresponding to each abnormal indicator according to the preset instruction rule library for the filtered abnormal indicators, and generates a fault warning instruction;

[0057] After receiving the set of abnormal indicators {annual failure rate, online batch problem}, for each abnormal indicator in the set, a preset instruction rule library is matched according to the instruction rule library. The instruction rule library is established based on the review analysis of historical failure data and treatment measures, combined with the experience of domain experts. The establishment process of the rule library includes: statistical analysis of the correlation between the amplitude of different indicators exceeding the control limit and the severity level of the quality problems subsequently triggered, dividing the exceeding amplitude into different intervals, and assigning a warning level and a standardized treatment plan to each interval. For example, for the annual failure rate indicator, the rule library is set as follows: when the indicator value exceeds the upper control limit by an amplitude of 0% to 5%, it is defined as “level 3 warning”, and the corresponding treatment plan is “start the root cause analysis process”; when the amplitude is between 5% and 10%, it is defined as “level 2 warning”, and the treatment plan is “notify the quality supervisor of the production line, and temporarily suspend the shipment of related batches of products”; when the amplitude exceeds 10%, it is defined as “level 1 warning”, and the treatment plan is “immediately stop the line inspection, and report to the quality management department”. The threshold setting of the warning level is obtained by regression analysis of 300 historical quality events of different levels, to find the best segmentation point of the abnormal amplitude of the indicator and the level of the event impact (quantified and classified according to economic loss and customer impact), to ensure that the false negative rate of high-level warning is less than 1% and the false positive rate is less than 5%. In this embodiment, the upper control limit of the annual failure rate is 0.61645%, the current value is 0.63%, and the exceeding amplitude is (0.63-0.61645) / 0.61645, the calculation result is about 2.2%. According to the rule library, 2.2% falls within the interval “0% to 5%”, so the matched warning level is “level 3 warning”, and the treatment plan is “start the root cause analysis process”. For the online batch problem indicator, the upper control limit is 1.5, the current value is 2, and the exceeding amplitude is (2-1.5) / 1.5, the calculation result is 33.3%. Assuming that the rule library is set to exceed 20% as “level 1 warning”, the indicator matches to “level 1 warning”, and the treatment plan is “establish a cross-department problem solving team to track the problem batch materials and equipment”. Finally, the sub-module integrates all the matching results to generate two independent failure warning instructions: the first instruction content is “abnormal indicator: annual failure rate, warning level: level 3, treatment plan: start the root cause analysis process”; the second instruction content is “abnormal indicator: online batch problem, warning level: level 1, treatment plan: establish a cross-department problem solving team to track the problem batch materials and equipment”.

[0058] Please refer to Figure 3 , the fault trend prediction module includes:

[0059] The abnormal index tracing sub-module analyzes the abnormal index corresponding to the fault warning instruction, collects data of the abnormal index changing with time within a specified period, integrates environmental factors and operation factors affecting the index change, performs trend analysis and qualitative prediction, and establishes an index change influence factor set;

[0060] The integration of environmental factors and operation factors affecting the index change includes collecting the temperature and humidity of the server room where the live server is located as environmental factors, and collecting the central processor load rate, memory usage rate, and disk read / write rate of the live server as operation factors.

[0061] After receiving the fault warning instruction of “Abnormal index: annual failure rate, warning level: level three”, the analysis of the abnormal index of annual failure rate begins. First, the data points of this index changing with time within a specified period, for example, the last 8 weeks, are collected to form a time series data, with specific values of {0.58%, 0.57%, 0.59%, 0.60%, 0.61%, 0.61%, 0.62%, 0.63%}. At the same time, environmental factors and operation factors affecting the index change are integrated. The integration of environmental factors is specifically to collect the average weekly temperature and humidity of the server room of the three data centers where the “A type” server is mainly deployed within the past 8 weeks. For example, the temperature and humidity of the first week are collected as {22.1℃, 45%RH}, {23.0℃, 46%RH}, {22.5℃, 44%RH}, and the average values of the readings of the three data centers are calculated as the representative values of the environmental factors of the week. The integration of operation factors is specifically to collect the weekly average load rate of the central processor, the weekly average usage rate of the memory, and the weekly average read / write rate of the disk of the server deployed in the same data center within the same time period. For example, the load rate, memory usage rate, and disk read / write rate of the first week are collected as {65%, 70%, 150MB / s}, {68%, 72%, 145MB / s}, {66%, 69%, 155MB / s}, and the average values are calculated as the representative values of the operation factors of the week. The annual failure rate data, environmental factor data, and operation factor data of the past 8 weeks are correspondingly formed into a multi-dimensional data set. Through trend analysis of the data set, it is observed that the annual failure rate gradually increases from 0.58% to 0.63%, showing a trend of continuous small increase. At the same time, the preliminary qualitative prediction is that if no intervention is taken, the annual failure rate of the next week will most likely continue to be above 0.63%. Finally, all the collected and integrated data, including the annual failure rate time series, the server room temperature time series, the server room humidity time series, the central processor load rate time series, the memory usage rate time series, and the disk read / write rate time series, collectively constitute the index change influence factor set.

[0062] The fault path construction submodule adopts a C4.5 decision tree algorithm to calculate information gain rates of environmental factors and operation factors in the index change influence factor set, selects a factor with the highest gain rate as a decision node to recursively split the data set, until all branches reach a preset stop condition, connects all decision nodes and branches, and generates a fault trend path.

[0063] The stop condition is that when any of the following conditions is met, the splitting of the current decision node is stopped: all data samples in the current decision node belong to the same fault category, the number of data samples in the current decision node is less than a minimum sample threshold, or the depth of the decision tree reaches a maximum depth threshold.

[0064] The index change influence factor set is processed. The index change influence factor set is described in the previous module as follows: "All collected and integrated data, including annual failure rate time series, computer room temperature time series, computer room humidity time series, central processor load rate time series, memory usage rate time series, and disk read / write rate time series, constitute the index change influence factor set."

[0065] A simplified index change influence factor set containing only four records is used as an example. Each record contains two influence factors: "memory weekly average usage rate" and "computer room weekly average temperature", and a final classification label. The label is determined according to whether the weekly annual failure rate is higher than the historical average of 0.58%, and is classified as "high failure" or "normal".

[0066] The specific four records are as follows: the first record: memory usage rate 70%, computer room temperature 22℃, label "normal". The second record: memory usage rate 75%, computer room temperature 26℃, label "normal". The third record: memory usage rate 85%, computer room temperature 23℃, label "high failure". The fourth record: memory usage rate 90%, computer room temperature 27℃, label "high failure".

[0067] The node (Node) in the C4.5 decision tree algorithm is the basic unit of the decision tree, including the root node (RootNode) containing the complete index change influence factor set at the beginning, the decision node (DecisionNode) representing the judgment of a certain influence factor, and the leaf node (LeafNode) representing the final classification conclusion.

[0068] The optimal split point is determined by calculating the information gain rate of each candidate influence factor, and the information gain rate is defined by the following complete formula:

[0069] ;

[0070] where information entropy The calculation formula is: ;

[0071] in, This represents the set of factors influencing the changes in the current node's metrics. At the root node where the algorithm begins, it specifically refers to the set of all four records listed above. In subsequent decision nodes, it refers to the subset of records assigned to that node. These are the influencing factors to be evaluated. At the root node, the algorithm will alternately consider "weekly average memory utilization" and "weekly average data center temperature" as influencing factors. Substitute into the formula to evaluate which is the better segmentation criterion. This represents the total number of fault categories. In this scenario, the final classification labels are only "High Fault" and "Normal," therefore... The value of is always 2. This is the index for the fault category. It's a counter variable used to iterate through all faults. Categories. Here. This can represent the "high failure" category. It can represent the "normal" category. The set of influencing factors of indicator changes at the current node In the middle, the first The proportion of each fault category. For example, in the four records of the root node, "high fault" appeared twice, and "normal" also appeared twice, therefore... The value is 2 / 4. The value is also 2 / 4. It is a logarithmic operation with base 2. In information theory, it is used to measure the amount of information. The lower the probability of an event, the greater its information content. The set of factors influencing the changes in indicators at the current node Information entropy. It is used to measure the current set. The "impurity" or "disorder" of a set is determined by its entropy. If all records in the set belong to the same category (e.g., all are "high fault"), the entropy is 0, indicating the purest state. If the number of records in each category is equal, the entropy is the highest (1 in this example), indicating the most disordered state. To use the impact factor The number of subsets generated after dividing the data at a certain split point. For continuous values ​​like "memory usage," the algorithm attempts to find an optimal split point (e.g., >80%) to divide the data in half; the number of subsets generated in this case is... The value is 2. This is the index of the subset. It is a counter variable used to iterate through all subsets. A subset. For example, when splitting by "memory usage > 80%", The subset that can represent "Memory usage rate ≤ 80%", The subset that can represent "Memory usage rate > 80%". is the impact factor After the division, the first subset is generated. For example, when the division criterion is "Memory usage rate > 80%", the subset containing the first and second records, the subset containing the third and fourth records. is the total number of records contained in the impact factor set of the current node. At the root node, the value is 4. is the number of records contained in the first subset . Continuing the example above, the value is 2, because the subset has two records; the value is also 2. is the information entropy of the first subset . It is used to measure the "impurity" within each subset after division. A good division will significantly reduce the information entropy of the subset. is the information gain. It represents the numerator part of the entire formula . Its physical meaning is: after dividing according to the impact factor , how much does the "impurity" of the entire system decrease. The greater the information gain of an impact factor, the stronger its classification ability. is the split information. It represents the denominator part of the entire formula . This term is used to measure the "intrinsic cost" of dividing according to the impact factor . If a division produces more and more scattered subsets, the value will be larger, which can be used to punish impact factors with too many values and prevent overfitting. is the information gain rate. It is the ratio of information gain to split information, and is the final decision basis for C4.5 algorithm to select the optimal impact factor and division point.

[0072] Step 1: Calculate the initial information entropy Ent(D) of the root node:

[0073] Among the four records, there are 2 "high failure" and 2 "normal", ;

[0074] Step 2: Calculate the information gain rate of the impact factor "Memory usage rate" (take the division point > 80% as an example):

[0075] subset (memory usage rate < 80%): {1st, 2nd} -> 0 "high failure", 2 "normal".

[0076] subset (memory usage rate > 80%): {3rd, 4th} -> 2 "high failure", 0 "normal".

[0077] ;

[0078] ;

[0079] ;

[0080] ;

[0081] .

[0082] Step 3: Calculate the information gain rate of the impact factor "machine room temperature" (take the split point > 25℃ as an example):

[0083] subset (machine room temperature < 25℃): {1st, 3rd} -> 1 "high failure", 1 "normal".

[0084] subset (machine room temperature > 25℃): {2nd, 4th} -> 1 "high failure", 1 "normal".

[0085] ;

[0086] ;

[0087] ;

[0088] ;

[0089] .

[0090] Step 4: Select the optimal split point and determine whether to terminate:

[0091] After completing the calculation of all candidate impact factors and split points at the root node, the algorithm will compare the results. According to the aforementioned calculation:

[0092] The information gain rate using "memory usage rate > 80%" for splitting is 1.0.

[0093] The information gain rate using "machine room temperature > 25℃" for splitting is 0.

[0094] By comparison, the value is much larger than , which shows that the division with the criterion of "memory usage rate > 80%" can reduce the uncertainty of the system to the greatest extent. Therefore, the algorithm selects "memory usage rate > 80%" as the final division criterion of the root node.

[0095] Next, the algorithm will examine the state of the new nodes (i.e. subsets) generated by this division, which requires an explanation of the meaning of "completely 'pure'". "Pure" means that all data records in a node belong to the same classification label.

[0096] Examine the first new node (subset , which satisfies "memory usage rate ≤ 80%"): this node contains {1st, 2nd} records, and their classification labels are all "normal". Since there is no any heterogeneous (i.e. no "high failure" records) in this node, it is a "completely pure" node.

[0097] Examine the second new node (subset , which satisfies "memory usage rate > 80%"): this node contains {3rd, 4th} records, and their classification labels are all "high failure". Similarly, there is also no any heterogeneous (i.e. no "normal" records) in this node, so it is also a "completely pure" node.

[0098] Because the two new nodes after division have reached the state of "completely pure", it means that any data reaching these nodes can get a clear and accurate classification conclusion without any further division. In the decision tree, such pure nodes will become leaf nodes (LeafNode), representing the end of decision. Therefore, the recursive construction process of the algorithm naturally terminates on these branches. Finally, the algorithm successfully constructs a clear and unambiguous fault trend path from the root node to the "high failure" leaf node: "memory usage rate > 80%".

[0099] The fault trend quantification submodule, according to the fault trend path, locates to the decision node at the end of the path, and counts and calculates the occurrence frequency and time distribution of the fault mode represented by the decision node, quantifies the occurrence probability and time of the future failure of the server in the live network as the fault trend, and generates the fault trend prediction result;

[0100] According to the generated fault trend path, quantitative analysis is performed. From the fault trend path, the decision nodes at the end of the path are located, which represent specific fault modes. For example, locate an end node with a path of "memory usage rate > 80%" and "machine room temperature > 25℃". First, count and calculate the occurrence frequency of the fault mode represented by this decision node. Specifically, in the original index change influence factor set, filter out all data records that meet the conditions of "memory usage rate > 80%" and "machine room temperature > 25℃". Assume that 20 historical records are filtered out. Then, in the 20 records, count the number of records marked as "high failure", which is assumed to be 18. Thus, the occurrence probability of this fault mode is calculated as 18 divided by 20, i.e. 90%. Next, analyze the time distribution of the occurrence of the 18 "high failure" records. By extracting the time stamp corresponding to each record, the time regularity of the occurrence of these failures is analyzed. For example, it is found that the occurrence time of 15 records is concentrated between 48 hours and 72 hours after meeting the two conditions of "memory usage rate > 80%" and "machine room temperature > 25℃". Based on this time distribution, the time window of future failure occurrence can be quantified. Finally, the occurrence probability and occurrence time are combined to quantify the occurrence probability and time of future failure of the present network server. The output of the fault trend prediction result is: when the memory usage rate of the server is continuously higher than 80% and the temperature of the machine room is higher than 25℃, the probability of the server occurring annualized failure rate increase event within 48 to 72 hours in the future is 90%.

[0101] Please refer to Figure 4 The risk component identification module comprises:

[0102] The negative trend screening submodule obtains the future trends of the first power-on failure rate, the annualized failure rate, and the present network batch problem of the three indexes in the fault trend prediction result, identifies and screens the trends that will have a negative impact on the overall quality of the server according to the preset quality risk judgment benchmark, and obtains the future fault trend of the present network server.

[0103] The fault trend prediction result is obtained, that is, "when the memory usage of the server is continuously higher than 80% and the temperature of the machine room is higher than 25°C, the probability of the server experiencing an annualized failure rate increase event in the next 48 to 72 hours is 90%". The sub-module identifies and filters this future trend according to the preset quality risk judgment criteria. The quality risk judgment criteria are jointly formulated in combination with industry standards and enterprise internal quality red lines. For example, the criteria are set as follows: any indicator with a predicted failure probability exceeding 70% or a predicted trend leading to an indicator value exceeding the historical control limit upper line is determined as a trend that will have a negative impact on the overall quality of the server. The setting of the criteria value is based on a retrospective analysis of the past 100 major quality accidents, which found that more than 90% of the accidents had an abnormal probability prediction value of the associated indicators exceeding 70% before the accident. In this embodiment, the predicted probability is 90%, which is greater than the 70% threshold set in the quality risk judgment criteria. Therefore, this trend is identified and filtered out as a future fault trend of the server in the live network. The final output of the future fault trend of the server in the live network is consistent with the prediction result, that is, a high-probability annualized failure rate increase event caused by high memory usage and high machine room temperature.

[0104] The key component positioning sub-module traces back the original fault data associated with each future fault trend of the server in the live network, performs secondary qualitative analysis on the original fault data, captures hardware or software that commonly appears in multiple data records and has an impact on quality as key components, and obtains a list of key components to be evaluated;

[0105] Upon receiving the future failure trend of the live server, the original failure data associated with this trend is traced. Specifically, the 18 historical data records marked as "high failure" that met the conditions of "memory utilization > 80%" and "data center temperature > 25℃" in the aforementioned steps are traced. A secondary qualitative analysis is performed on the original failure work orders or equipment logs corresponding to these 18 records. The analysis process involves manually or through natural language processing technology extracting detailed descriptions of the failure phenomenon, error codes, and replaced components from each failure record. For example, analyzing record 1 reveals a description of "unexpected device restart, log shows memory ECC error, recovered after replacing B-brand memory module." Analyzing record 2 reveals a description of "system blue screen, dump file points to memory management unit, B-brand memory module replaced." Each of these 18 records is analyzed to identify hardware or software that appears in multiple data records and affects quality. The analysis reveals that 14 of the 18 records explicitly point to a replacement operation due to a "B-brand memory module" failure. In addition, four records mentioned voltage instability alarms for "Model C power supplies," but none of these power supplies were ultimately replaced. By comparison, the frequency of "Brand B memory modules" was significantly higher than other components, and they were the direct cause of the failure. Therefore, "Brand B memory modules" were identified as a critical component, and a list of critical components to be evaluated was obtained. This list currently only contains "Brand B memory modules."

[0106] The component risk marking submodule collects the failure records corresponding to the key components in the list of key components to be evaluated, counts the occurrence frequency of key components under each indicator, calculates the overall failure rate of the components, compares it with the risk component threshold, and marks all components that exceed the risk component threshold as quality risk components.

[0107] Obtain the "Brand B memory module" from the list of key components to be evaluated, and compile its corresponding 14 fault records. To assess its risk level, the component's overall failure rate needs to be calculated, as follows:

[0108] ;

[0109] in, : Represents the annualized failure rate of a component. It is a percentage value that predicts the probability of the component failing after one year of continuous operation, based on data within the observation period. It is a core indicator for assessing its long-term quality risk. This represents the total number of component failures. Within a specified observation period, it is the total number of failure events whose root cause is attributed to "Brand B memory modules" after failure analysis. The unit is "units" or "times". : Represents the total number of hours in a year, which is a constant value. : represents the total operating hours of the component. It refers to the sum of the actual operating hours of all servers installed with "B brand memory" in the same observation period. This value reflects the cumulative working time of the entire component group.

[0110] The calculation process of the annual failure rate of "B brand memory" is as follows:

[0111] Parameter setting: Setting: Through root cause analysis of the 18 highly relevant historical failure records located in step 8, it is finally confirmed that the direct cause of 14 failures is "B brand memory" itself defects. Therefore, the parameter is set to 14. Setting: By querying the configuration management database and device operating time monitoring system, it is learned that in the past year's observation period, there are 2500 servers installed with this "B brand memory" in the network. The cumulative total operating time of these servers is 17,520,000 hours. Setting: The parameter represents the total hours in a year, i.e. , whose value is 8760.

[0112] Substitute into the formula to calculate: Substitute the above parameters into the annual failure rate calculation formula:

[0113] ;

[0114] Then, compare the calculated annual failure rate of 0.7% with the risk component threshold. The risk component threshold is set according to the historical average failure rate of similar components and the supplier quality agreement, specifically set to 1.5 times the historical average annual failure rate. Through statistical analysis of the failure data of all brand memories in the past two years, it is concluded that the historical average annual failure rate is 0.15%, so the risk component threshold is set to . The coefficient of 1.5 is based on the risk management strategy, which defines components exceeding the average level by 50% as risk items that need attention. This strategy has been jointly reviewed and confirmed by the supply chain and quality department. Since the calculated annual failure rate of 0.7% exceeds the risk component threshold of 0.225%, "B brand memory" is marked as a quality risk component.

[0115] Please refer to Figure 5 , the propagation path tracking module includes:

[0116] The propagation characteristic calculation submodule collects the deployment location of each quality risk component in the server in the network, and counts the associated failure records, calculates the frequency and time interval of the impact of the failure of each quality risk component on other components, and quantifies the quality risk component propagation probability and propagation delay, and generates component failure propagation data;

[0117] Analyze the identified quality risk component "B brand memory stick". First, collect the deployment location information of "B brand memory stick" in the live server, for example, query the configuration management database to find that the memory stick is mainly installed in the DIMM_A1 and DIMM_B1 slots of "A type" server. At the same time, statistics of the fault records associated with the memory stick, that is, the 14 fault records determined previously. Deeply analyze the detailed logs of the 14 records, calculate the frequency and time interval of the impact of "B brand memory stick" failure on other components. Analysis shows that in the 14 memory failures, 5 failures occurred within 24 hours, and the server's central processing unit (CPU) also recorded a large number of L2 cache correction errors. This indicates that memory failure may cause data corruption, which in turn affects the normal work of CPU. Therefore, the failure impact frequency of "B brand memory stick" to CPU is calculated as 5 divided by 14, about 35.7%. For time interval, calculate the time difference between the time point of memory failure and the time point of CPU cache error first occurrence in the 5 associated events, and the average value is 10 hours. Quantize the two calculation results as quality risk component propagation probability and propagation delay. The final generated component failure propagation data is: the quality risk component "B brand memory stick" has a 35.7% probability of affecting the L2 cache of "central processing unit" within 10 hours after failure.

[0118] The potential impact analysis submodule dynamically analyzes the potential impact range of the failure according to the potential impact of the quality risk component propagation probability and the propagation delay of the component failure propagation data to other components;

[0119] Based on the obtained component failure propagation data, that is, "the quality risk component 'B brand memory stick' has a 35.7% probability of affecting the L2 cache of 'central processing unit' within 10 hours after failure", dynamic analysis is carried out. The starting point of the analysis is the failure of "B brand memory stick". Based on the 35.7% propagation probability, it can be determined that the central processing unit is a direct and high-probability potential impact object. Then, the analysis range is further expanded. Query the hardware knowledge base and the fault associated history database to analyze the subsequent problems that may be caused by the central processing unit L2 cache error. The analysis result shows that the persistent error of L2 cache has a high probability (for example, historical data shows more than 60%) to cause the operating system kernel to crash, thereby causing the server to crash. Therefore, the potential impact range of the failure is expanded from the hardware level CPU to the system level operating system stability. Through this layer-by-layer recursive correlation analysis, a series of chain reactions that may be triggered by the failure of a core risk component are dynamically depicted, thereby determining the potential impact range of the failure, that is, from "B brand memory stick" failure to "central processing unit" L2 cache error, to "operating system kernel crash".

[0120] The propagation path visualization submodule establishes a fault transmission relationship between components based on the potential impact range of the fault, takes the quality risk component as the starting node, takes the affected other components as the target node, takes the potential impact relationship as the connection path, and generates a fault propagation path diagram;

[0121] Based on the potential impact range of the fault, the fault transmission relationship between components is established. The quality risk component "B brand memory stick" is taken as the starting node of fault propagation. The affected other components or system states, i.e. "central processing unit" and "operating system kernel", are taken as the target node. The potential impact relationship analyzed is taken as the connection path between nodes. Specifically, a directed graph is drawn: a node representing "B brand memory stick" is created; a node representing "central processing unit" is created; and a node representing "operating system kernel" is created. A directed edge is drawn from the "B brand memory stick" node to the "central processing unit" node, and the propagation characteristic data, i.e. "propagation probability 35.7%, propagation delay 10 hours", is labeled on the edge. Then, a directed edge is drawn from the "central processing unit" node to the "operating system kernel" node, and the corresponding propagation probability is labeled according to historical data analysis, for example "propagation probability 60%". By connecting all related nodes and paths, a fault propagation path diagram is finally generated, which clearly shows how the fault spreads from one component to another.

[0122] Please refer to Figure 6 The parameter adjustment module includes:

[0123] The process correlation analysis submodule calls the fault trend prediction result and the fault propagation path diagram, analyzes the fault cause of the quality risk component and the time point of occurrence in the manufacturing process, identifies the key manufacturing process link affecting the quality of the server, and generates the to-be-adjusted process parameter item;

[0124] Call the failure trend prediction results (high memory usage and high temperature environment lead to the rise of annual failure rate) and the failure propagation path graph (B brand memory stick is the risk source). First, analyze the failure cause of "B brand memory stick". Through the failure analysis (FA) of 14 returned failure memory sticks, it is found that the root cause of failure of 12 of them is that a certain pin of the memory particle is loose. Then, analyze at which time point in the manufacturing process this loose problem occurs. By tracing the production batch record of this brand memory stick and comparing with the server manufacturing process, it is found that the surface mount technology (SMT) reflow soldering link of the server mainboard is the key step of memory stick welding. By analyzing the process parameters of reflow soldering, especially the temperature curve setting, it is found that when the ambient temperature is high, the existing reflow soldering temperature curve may cause uneven heating of some memory particles with large heat capacity, thereby causing loose risk. Therefore, the key manufacturing process link affecting the quality of the server is identified as "mainboard reflow soldering", and the "reflow soldering temperature curve" is taken as the process parameter item to be adjusted.

[0125] The parameter adjustment list generation submodule queries the fault phenomenon and process adjustment table in the MES system according to the process parameter item to be adjusted, automatically matches and obtains the corresponding adjustment plan, and generates a process parameter adjustment list;

[0126] According to the determined process parameter item to be adjusted "reflow soldering temperature curve", the pre-set fault phenomenon and process adjustment table in the manufacturing execution system (MES) is queried. The table is a knowledge base established based on long-term process experiments and production data accumulation. The query conditions are "fault phenomenon: memory particle loose" and "associated components: B brand memory stick". The system automatically matches one or more corresponding adjustment plans. For example, a plan is found that describes: "for the loose problem of B brand memory stick, it is suggested to increase the target temperature of the preheating zone of reflow soldering from 150°C to 155°C, and extend the time of the constant temperature zone by 3 seconds to improve the wetting property of the solder joint". After the system obtains this specific adjustment plan, it is formatted and finally generates a process parameter adjustment list. The list clearly lists the equipment number, parameter name, current value and suggested new value that need to be adjusted. For example: "equipment: reflow soldering furnace RF-02, parameter: preheating zone temperature, current value: 150°C, adjustment value: 155°C; parameter: constant temperature zone time, current value: 60 seconds, adjustment value: 63 seconds".

[0127] The process configuration update submodule updates the existing server manufacturing process parameter settings in the MES system based on the process parameter adjustment list, integrates and formats all the adjusted parameters, and generates a new process configuration table;

[0128] Upon receiving the list of process parameter adjustments, the operator or automated script updates the existing server manufacturing process parameter settings for "reflow oven RF-02" in the manufacturing execution system (MES) operating interface according to the content of the list. Specifically, the "preheat zone temperature" parameter is modified from 150°C to 155°C, the "soak zone time" parameter is modified from 60 seconds to 63 seconds, and the configuration is saved. After the update operation is completed, the system integrates and formats all the adjusted parameters together with other process parameters that are not modified. This formatting process includes recording the modification time, the operator, and the modification reason (associated with the fault warning event ID) of the parameters, and organizing them according to the standard process file format. Finally, a new process configuration table with an updated version number is generated. This new process configuration table will be used as the basis for the subsequent production of the server model by the reflow soldering equipment.

[0129] The above merely describes the preferred embodiment of the present application, but the present application is not limited to the above, and any modification or change made by those skilled in the art based on the above disclosure without departing from the technical solution of the present application shall fall within the protection scope of the present application.

Claims

1. A MES system for automatically adjusting server manufacturing processes based on existing network faults, characterized in that, The system comprises: a fault data processing module, which counts three indexes of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server, establishes an Xbar-R graphical sequence, screens abnormal indexes in the Xbar-R graphical sequence, and generates a fault early warning instruction; a fault trend prediction module, which analyzes changes of abnormal indexes corresponding to the fault early warning instruction over time, predicts an occurrence probability and a time of a future fault of the server as a fault trend, and generates a fault trend prediction result; a risk component identification module, which identifies a future fault trend of the server that will have a negative impact on the overall quality of the server according to the fault trend prediction result, captures corresponding key components, calculates a comprehensive failure rate of the components, and marks quality risk components; a propagation path tracking module, which tracks a fault propagation path of the quality risk components in real time, analyzes whether a fault of the quality risk components will have an impact on other components, and generates a fault propagation path diagram; a parameter adjustment module, which automatically adjusts server manufacturing process parameters according to the fault trend prediction result of the server and the fault propagation path diagram of the quality risk components, and generates a new process configuration table.

2. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 1, wherein, The fault early warning instruction specifically refers to an abnormal value of a fault index, a fault occurrence warning, and fault analysis information. The fault trend prediction result includes a fault occurrence probability, a fault occurrence time, and a fault type prediction. The quality risk components specifically refer to components with a fault frequency exceeding a frequency threshold, fault type related components, and key performance components. The fault propagation path diagram includes a propagation path, an impact range, and an impact component. The new process configuration table specifically refers to adjusted process parameters, optimized production processes, and improved production processes.

3. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 1, wherein, The fault data processing module comprises: a historical index modeling submodule, which collects three index data of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server in a specified period, establishes an Xbar-R graphical sequence, calculates a center line and upper and lower control limits of the Xbar-R graphical sequence, and establishes index historical data control limits; a live network index monitoring submodule, which obtains current values of three indexes of a first power-on failure rate, an annualized failure rate and a batch problem in a live network of a server in a current period, compares the current values of each index with the index historical data control limits, screens indexes that exceed the control limits, and obtains abnormal indexes; a fault early warning generation submodule, which, for the screened abnormal indexes, matches an early warning level and a disposal plan corresponding to each abnormal index according to a preset instruction rule library, and generates a fault early warning instruction.

4. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 3, wherein, The fault trend prediction module comprises: an abnormal index tracing submodule, which analyzes the abnormal indexes corresponding to the fault early warning instruction, collects data of the abnormal indexes changing over time in a specified period, integrates environmental factors and operating factors affecting the change of the indexes, performs trend analysis and qualitative prediction, and establishes an index change influence factor set; The fault path construction submodule adopts a C4.5 decision tree algorithm to calculate information gain rates of environmental factors and operation factors in the set of index change influence factors, selects a factor with the highest gain rate as a decision node to recursively segment a data set, until all branches reach a preset stop condition, connects all decision nodes and branches, and generates a fault trend path; The fault trend quantification submodule locates to a decision node at an end of the path according to the fault trend path, counts and calculates occurrence frequencies and time distributions of fault modes represented by the decision node, quantifies occurrence probabilities and times of future faults of the online server as fault trends, and generates a fault trend prediction result.

5. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 4, wherein, The environmental factors and operation factors that affect the index changes include collected temperature and humidity of a machine room where the online server is located as the environmental factors, and collected central processor load rate, memory usage rate, and disk read / write rate of the online server as the operation factors.

6. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 4, wherein, The stop condition for all branches includes any one of the following conditions: all data samples in the current decision node belong to the same fault category, the number of data samples in the current decision node is lower than a minimum sample number threshold, or the depth of the decision tree reaches a maximum depth threshold.

7. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 4, wherein, For calculating the information gain ratio using the formula: ； wherein, is a set of index change impact factors of a current node, is an impact factor to be evaluated, is a total number of failure categories, is an index of a failure category, is a proportion of an index change impact factor set of a current node to a first failure category, is information entropy of an index change impact factor set of a current node , and a calculation formula is , is a number of subsets generated after division using a division point of an impact factor , is an index of a subset, is a total number of records included in an index change impact factor set of a current node , is a number of records included in a first subset , is information gain, is split information.

8. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 4, wherein, The risk component identification module includes: The negative trend screening submodule obtains future trends of three indexes of a first power-on failure rate, an annualized failure rate, and an online batch problem in the fault trend prediction result, identifies and screens trends that have a negative impact on the overall quality of the server according to a preset quality risk judgment criterion, and obtains future fault trends of the online server; The key component positioning submodule traces original fault data associated with each of the future fault trends of the online server, performs secondary qualitative analysis on the original fault data, captures hardware or software that commonly appears in multiple data records and has an impact on quality as a key component, and obtains a list of key components to be evaluated; The component risk marking submodule collects fault records corresponding to the key components in the list of key components to be evaluated, counts occurrence frequencies of the key components under each index, calculates a comprehensive failure rate of the component, compares the comprehensive failure rate with a risk component threshold, and marks all components that exceed the risk component threshold as quality risk components.

9. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 8, wherein, The propagation path tracking module includes: The propagation characteristic calculation submodule collects deployment positions of each of the quality risk components in the online server, counts associated fault records, calculates frequencies and time intervals of influences of each of the quality risk components on other components after failure, quantifies a quality risk component propagation probability and a propagation delay, and generates component failure propagation data; The potential influence analysis submodule dynamically analyzes a fault potential influence range according to potential influences of the quality risk component propagation probability and the propagation delay of the quality risk component on other components in the component failure propagation data. The propagation path visualization submodule establishes a fault transmission relationship between each component based on the potential influence range of the fault, takes the quality risk component as a starting node, takes other affected components as target nodes, takes a potential influence relationship as a connection path, and generates a fault propagation path diagram.

10. The MES system for automatically adjusting a server manufacturing process based on live network failures of claim 9, wherein, The parameter adjustment module comprises: The process correlation analysis submodule calls the fault trend prediction result and the fault propagation path diagram, analyzes the fault cause of the quality risk component and the time point at which the fault occurs in the manufacturing process, identifies a key manufacturing process link that affects server quality, and generates a to-be-adjusted process parameter item; The parameter adjustment list generation submodule queries a fault phenomenon and process adjustment table in the MES system according to the to-be-adjusted process parameter item, automatically matches and obtains a corresponding adjustment plan, and generates a process parameter adjustment list; The process configuration update submodule updates an existing server manufacturing process parameter setting in the MES system based on the process parameter adjustment list, integrates and formats all adjusted parameters, and generates a new process configuration table.

Images