Chip product after-sales information management method and system

Abstract

The invention provides an after-sales information management method and system for chip products, and the method comprises the steps: carrying out the text processing of various types of unstructured data in the after-sales information of the chip products, and obtaining structured after-sales basic data; carrying out mining analysis on the after-sales basic data based on an association rule mining algorithm, mining association relationships between different fault phenomena and chip parameters, and obtaining fault association rule data; performing fault cause inference on the fault association rule data by using a Bayesian network algorithm to obtain a fault cause inference result; and aiming at the fault reason inference result, combining historical maintenance cost data, adopting a cost-benefit analysis algorithm, evaluating costs and expected effects of different maintenance schemes, and determining an optimal maintenance strategy as a management result of the after-sales information of the chip product by calculating a cost-benefit ratio. According to the method, a reasonable maintenance scheme is made, so that resource waste is avoided, and enterprise benefit maximization is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method and a system for managing after-sales information of chip products. Background Art

[0002] The quality and stability of chip products are directly related to the overall performance and reliability of equipment. Therefore, after-sales information management is crucial for chip manufacturers and related enterprises.

[0003] However, the existing management of after-sales information of chip products faces many difficulties. In the fault diagnosis process based on after-sales information, traditional methods mainly rely on manual experience and lack scientific and accurate analysis means. Maintenance personnel often can only judge the cause of the fault based on their accumulated experience, and it is difficult to comprehensively analyze the internal relationship between the fault phenomenon and chip parameters. In the face of complex chip faults, this method is extremely prone to misdiagnosis, resulting in unreasonable maintenance plans, which not only increases the maintenance cost, but also prolongs the equipment downtime, reducing customer satisfaction and enterprise competitiveness.

[0004] Furthermore, when determining the maintenance strategy, the traditional mode does not fully consider the cost-benefit factors. Most of the maintenance plans are selected based on conventional processes without comprehensively weighing the cost input and expected maintenance effects of different plans, which will cause waste of resources and cannot maximize the enterprise benefits. Summary of the Invention

[0005] The main object of the present invention is to provide a method and a system for managing after-sales information of chip products, aiming to overcome the defect that the current situation cannot make a reasonable maintenance plan for after-sales information, resulting in waste of resources.

[0006] To achieve the above object, the present invention provides a method for managing after-sales information of chip products, including the following steps:

[0007] Performing text processing on various types of unstructured data in the after-sales information of chip products to obtain structured after-sales basic data;

[0008] Based on the association rule mining algorithm, mining and analyzing the after-sales basic data to mine the association relationship between different fault phenomena and chip parameters, and obtaining fault association rule data;

[0009] Using the Bayesian network algorithm to infer the cause of the fault for the fault association rule data, and obtaining a fault cause inference result;

[0010] For the fault cause inference result, combining with historical maintenance cost data, adopting a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the after-sales information of the chip product.

[0011] Furthermore, various types of unstructured data in after-sales information include user feedback text, chip operation logs, and maintenance record documents;

[0012] The chip parameters include chip model, operating environment, and operating duration.

[0013] Furthermore, the Bayesian network algorithm is used to infer the cause of the fault from the fault correlation rule data, and the fault cause inference result is obtained, including:

[0014] According to the input requirements of the Bayesian network algorithm, the fault correlation rule data is adaptively processed to obtain adapted data; the adaptive processing includes adjusting the data feature format and value range, discretizing continuous features, and uniformly encoding categorical features;

[0015] Evidence information is extracted from the adapted data, and the known fault phenomena and chip parameters are mapped to the corresponding nodes of the Bayesian network algorithm and marked as observed nodes to obtain the network input;

[0016] Using the inference algorithm in the Bayesian network algorithm, the posterior probability of the unobserved potential fault cause nodes is calculated based on the network input to obtain the posterior probability distribution;

[0017] The fault cause inference result is determined according to the posterior probability distribution.

[0018] Furthermore, for the fault cause inference result, combined with historical maintenance cost data, a cost-benefit analysis algorithm is used to evaluate the costs and expected effects of different maintenance plans, including:

[0019] Analyze the fault cause inference result to determine the maintenance operations to solve the fault; combine the maintenance operations to generate multiple different maintenance plans; among them, each maintenance plan corresponds to a complete maintenance process;

[0020] Extract the direct cost elements related to each maintenance plan from the historical maintenance cost data to obtain the direct cost of each maintenance plan;

[0021] Analyze the occurrence pattern of indirect costs in similar maintenance situations in historical data, and combine the severity of the current fault and the estimated maintenance duration to estimate the indirect cost of each maintenance plan. Add the direct cost and the indirect cost to obtain the total cost of each maintenance plan;

[0022] Based on historical maintenance data, count the number of successful repair cases for each maintenance plan for the same or similar fault causes, calculate the proportion of successful repair cases in the total maintenance cases, and predict the fault repair rate of each maintenance plan;

[0023] Analyze the changes in various performance indicators of chip products before and after repair in historical data, and calculate the change score of performance indicators using the weighted average method;

[0024] Perform a weighted calculation on the failure repair rate and the change score of performance indicators to obtain a quantitative index of the expected effect.

[0025] Further, by calculating the cost-benefit ratio, determine the optimal maintenance strategy, including:

[0026] Divide the quantitative index of the expected effect of each maintenance plan by its total cost to obtain the corresponding cost-benefit ratio;

[0027] Sort the cost-benefit ratios of all maintenance plans, and select the plan with the highest cost-benefit ratio as the optimal maintenance strategy.

[0028] Further, after determining the optimal maintenance strategy, it also includes:

[0029] Establish a mapping relationship between the after-sales information of the chip product and the optimal maintenance strategy, and add it to the data table corresponding to the chip product in the after-sales information management system;

[0030] Obtain the basic management password of the after-sales information management system;

[0031] Based on the after-sales information of the chip product and the optimal maintenance strategy, construct a data matrix;

[0032] Based on the basic management password, perform an adaptive adjustment on the elements of the data matrix to obtain an adjusted matrix;

[0033] Based on the adjusted matrix, obtain a data table management password to perform permission management on the data table.

[0034] Further, based on the basic management password, performing an adaptive adjustment on the elements of the data matrix to obtain an adjusted matrix includes:

[0035] Based on the basic management password, generate a two-row and two-column digital character matrix;

[0036] Starting from the upper left corner of the data matrix, superimpose the digital character matrix on the data matrix; when superimposing, if the element in the data matrix is a number, add the number at the overlapping position of the digital character matrix to it to obtain the adaptively adjusted element;

[0037] Move the digital character matrix successively, and adaptively adjust the corresponding matrix elements of the data matrix until the digital character matrix moves out of the data matrix to obtain an adjusted matrix; wherein, the moving step of the digital character matrix is to move two positions both downward and rightward.

[0038] Further, based on the adjusted matrix, obtain a data table management password, including:

[0039] Generate a plurality of basic numbers based on the basic management password;

[0040] Generate three different graphics based on each of the basic numbers using different generation rules;

[0041] Overlay the three graphics into the adjusted matrix according to a preset rule, and detect the intersection areas between any two graphics;

[0042] Extract the characters in the adjusted matrix located in all the intersection areas, and combine them to obtain combined characters as the data table management password.

[0043] The present invention also provides an after-sales information management system for a chip product, including:

[0044] A processing module for performing text processing on various unstructured data in the after-sales information of the chip product to obtain structured after-sales basic data;

[0045] A mining module for performing mining analysis on the after-sales basic data based on an association rule mining algorithm to mine the association relationship between different fault phenomena and chip parameters, and obtain fault association rule data;

[0046] An inference module for performing fault cause inference on the fault association rule data by using a Bayesian network algorithm to obtain a fault cause inference result;

[0047] A management module for, in view of the fault cause inference result, combining historical maintenance cost data, and using a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining an optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the after-sales information of the chip product.

[0048] The present invention also provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method described in any one of the above are implemented.

[0049] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0050] The after-sales information management method and system for chip products provided by the present invention include: performing text processing on various types of unstructured data in the after-sales information of chip products to obtain structured after-sales basic data; performing mining analysis on the after-sales basic data based on the association rule mining algorithm to mine the association relationship between different fault phenomena and chip parameters, and obtaining fault association rule data; using the Bayesian network algorithm to infer the fault cause for the fault association rule data to obtain a fault cause inference result; for the fault cause inference result, combining historical maintenance cost data, adopting a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the after-sales information of the chip products. In the present invention, by inferring the fault cause inference result based on the after-sales information, then combining historical maintenance cost data, adopting a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio, a reasonable maintenance plan is made, thereby avoiding resource waste and achieving the maximization of enterprise interests. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic diagram of the steps of the after-sales information management method for chip products in an embodiment of the present invention;

[0052] Figure 2 is a block diagram of the structure of the after-sales information management system for chip products in an embodiment of the present invention;

[0053] Figure 3 is a schematic block diagram of the structure of a computer device in an embodiment of the present invention.

[0054] The implementation, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] Referring to Figure 1 , an after-sales information management method for chip products is provided in an embodiment of the present invention, including the following steps:

[0057] Step S1, performing text processing on various types of unstructured data in the after-sales information of chip products to obtain structured after-sales basic data;

[0058] Step S2: Based on the association rule mining algorithm, conduct mining and analysis on the after-sales basic data to discover the association relationships between different fault phenomena and chip parameters, and obtain the fault association rule data;

[0059] Step S3: Use the Bayesian network algorithm to infer the fault causes for the fault association rule data, and obtain the fault cause inference results;

[0060] Step S4: For the fault cause inference results, in combination with the historical maintenance cost data, adopt the cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans. By calculating the cost-benefit ratio, determine the optimal maintenance strategy as the management result of the after-sales information of the chip product.

[0061] In this embodiment, as described in step S1 above, the after-sales information of the chip product includes various unstructured data. For example, the natural language description when the customer reports a fault may mention abnormal phenomena that occur during the use of the chip, such as "the chip gets severely hot and the computer frequently freezes"; the maintenance reports written by the maintenance personnel, which record the subjective feelings and operation situations during the maintenance process; and the test logs recorded by the test personnel, which contain scattered records during the performance tests of various chip items. These data do not have a fixed format and specification, making it difficult to directly analyze and process them.

[0062] Therefore, first, remove the noise information in the text, such as extra spaces, punctuation marks, special characters, etc., and correct spelling mistakes. For example, correct "the chip gets severely hot" in the customer feedback to "the chip gets severely hot". Then, split the long text into individual words or phrases for subsequent analysis. For Chinese text, special word segmentation tools such as Jieba can be used; for English text, simple word segmentation can be performed according to spaces. Secondly, determine the part of speech of each word, which helps to understand the semantic structure of the text. For example, distinguish nouns, verbs, adjectives, etc., to prepare for subsequent information extraction. Extract key information from the segmented and annotated text, such as fault phenomena, occurrence time, chip model, etc., and organize this information into a structured data format, such as a table form, where each row represents a piece of after-sales information and each column represents a specific attribute.

[0063] As described in step S2 above, based on the association rule mining algorithm, conduct mining and analysis on the after-sales basic data to discover the association relationships between different fault phenomena and chip parameters, and obtain the fault association rule data. Among them, association rule mining is a data mining technique used to discover potential association relationships between different items in a dataset. In the management of chip after-sales information, the focus is on the association between fault phenomena and chip parameters. For example, whether there is an association between the chip temperature being too high and the operation speed decreasing. Commonly used association rule mining algorithms include the Apriori algorithm, the FP-growth algorithm, etc.

[0064] Further process the structured after-sales basic data, and convert the data into a format suitable for processing by the association rule mining algorithm, such as a transaction dataset. Each transaction represents a piece of after-sales information and contains multiple items, such as fault phenomena and chip parameters.

[0065] Furthermore, set the minimum support and minimum confidence: Support represents the frequency of an association rule appearing in the dataset, and confidence represents the proportion of transactions that contain the consequent while also containing the antecedent. By setting appropriate minimum support and minimum confidence, meaningful association rules can be filtered out. For example, setting the minimum support to 0.1 means that the association rule must appear in at least 10% of the transactions; setting the minimum confidence to 0.8 means that in transactions containing the antecedent, at least 80% of the transactions also contain the consequent.

[0066] Use the selected association rule mining algorithm to mine association rules that meet the minimum support and minimum confidence in the preprocessed dataset. For example, mine an association rule such as "chip temperature > 80°C => operation speed decreases", indicating that when the chip temperature exceeds 80°C, there is a high probability that the operation speed will decrease. Evaluate the mined association rules. In addition to support and confidence, other metrics such as lift can also be considered to further filter out valuable association rules and obtain fault association rule data.

[0067] As described in step S3 above, the Bayesian network algorithm is a graphical model based on probabilistic inference. It uses a directed acyclic graph to represent the causal relationship between variables and describes the degree of dependence between variables through a conditional probability table. In inferring the cause of chip failures, the Bayesian network can infer possible causes of failure based on the association relationship between known fault phenomena and chip parameters.

[0068] Based on the fault association rule data and domain knowledge, determine the nodes and directed edges in the Bayesian network. Nodes represent fault phenomena, chip parameters, and possible causes of failure, and directed edges represent the causal relationship between variables. For example, "excessive chip temperature" may be caused by reasons such as "heat dissipation module failure" or "chip overclocking", and these nodes are connected by directed edges. Use the fault association rule data to estimate the conditional probability table for each node in the Bayesian network. The conditional probability table describes the probability that the node takes different values given the values of its parent nodes. For example, the probability of "excessive chip temperature" given "heat dissipation module failure" and "chip overclocking".

[0069] Input the actually observed fault phenomena and chip parameters as evidence into the Bayesian network, and use Bayesian inference algorithms (such as variable elimination algorithm, clique tree propagation algorithm) to calculate the posterior probability of each possible fault cause. The posterior probability represents the probability of a certain fault cause occurring given the known evidence. Select the fault cause with a higher posterior probability as the inference result.

[0070] As described in step S4 above, collect various cost data during previous chip repair processes, including component replacement costs, manual repair costs, equipment usage costs, etc., and at the same time record the repair effects corresponding to each repair plan, such as fault repair rate, chip performance recovery degree, etc. Organize and classify these data for subsequent analysis.

[0071] The above cost-benefit analysis is a method to evaluate the pros and cons of different solutions by comparing the costs and expected effects of different solutions. In chip repair, the cost-benefit ratio can be defined as the ratio of the expected effect to the cost. The higher the cost-benefit ratio, the better the effect that can be obtained with the input cost for the repair plan. According to the fault cause inference result, combined with the characteristics of the chip product and repair experience, formulate multiple possible repair plans, such as replacing faulty components, performing software upgrades, adjusting chip parameters, etc. For each repair plan, estimate its total required cost, including direct costs and indirect costs, in combination with historical repair cost data. Direct costs such as component procurement costs, repair personnel salaries, etc.; indirect costs such as equipment downtime losses caused by repair. Evaluate the expected effects of each repair plan, such as the possibility of fault repair, the degree of chip performance recovery, etc., according to historical repair data and domain knowledge. Quantification indicators can be used to represent the expected effects, such as fault repair rate, percentage of performance improvement, etc. Divide the quantification indicator of the expected effect of each repair plan by its total cost to obtain the corresponding cost-benefit ratio. Compare the cost-benefit ratios of different repair plans, and select the repair plan with the highest cost-benefit ratio as the optimal repair strategy, and this strategy is the management result of the after-sales information of the chip product.

[0072] In the embodiment of the present invention, by inferring the fault cause inference result based on after-sales information, then combining historical repair cost data, using the cost-benefit analysis algorithm, evaluating the costs and expected effects of different repair plans, determining the optimal repair strategy by calculating the cost-benefit ratio, and making a reasonable repair plan, thus avoiding resource waste and achieving the maximization of enterprise interests.

[0073] In one embodiment, various types of unstructured data in the after-sales information include user feedback text, chip operation logs, and repair record documents;

[0074] The chip parameters include chip model, operating environment, and operating duration.

[0075] In one embodiment, the Bayesian network algorithm is used to infer the cause of the fault from the fault association rule data, and the fault cause inference result is obtained, including:

[0076] According to the input requirements of the Bayesian network algorithm, the fault association rule data is adapted to obtain adapted data; the adaptation process includes adjusting the data feature format and value range, discretizing continuous features, and uniformly encoding categorical features;

[0077] Extract evidence information from the adapted data, map the known fault phenomena and chip parameters to the corresponding nodes of the Bayesian network algorithm, and mark them as observed nodes to obtain the network input;

[0078] Using the inference algorithm in the Bayesian network algorithm, calculate the posterior probability of the unobserved potential fault cause nodes based on the network input to obtain the posterior probability distribution;

[0079] Determine the fault cause inference result according to the posterior probability distribution.

[0080] In this embodiment, first, according to the input requirements of the Bayesian network algorithm, the fault association rule data is adapted to obtain adapted data. The Bayesian network algorithm usually has specific formats and requirements for input data, while the fault association rule data may come from different data sources, and there may be differences in its feature format, value range, etc., and it cannot be directly used for the calculation of the Bayesian network. Therefore, it is necessary to adapt these data to ensure that the data can be correctly recognized and processed by the Bayesian network algorithm.

[0081] Different data sources can use different data formats to represent the same information. For example, for the temperature parameter of the chip, some data may be recorded in integer form, and some may be recorded in floating-point form. The adaptation process needs to unify these data into one format so that the Bayesian network can perform effective calculations.

[0082] The value range of the data varies due to factors such as measurement equipment and measurement methods. For example, the voltage parameter of the chip may have different value ranges in different test environments. The adaptation process needs to unify these value ranges to avoid deviations in the calculation results of the Bayesian network due to differences in the value ranges.

[0083] The Bayesian network algorithm may face the problem of high computational complexity when dealing with continuous data. Therefore, it is usually necessary to discretize continuous features. For example, the continuous values of the chip temperature are divided into several intervals, such as "low temperature (<50°C)", "medium temperature (50°C - 80°C)", "high temperature (>80°C)". Discretization methods include equal-width discretization, equal-frequency discretization, etc., and appropriate methods can be selected according to the specific data characteristics and business requirements. For categorical features, such as the chip failure types (short circuit, open circuit, overheating, etc.), different data sources may adopt different coding methods. Adaptation processing requires unifying the coding of these categorical features. For example, digital coding is used, where "short circuit" is coded as 1, "open circuit" is coded as 2, "overheating" is coded as 3, etc., so that the Bayesian network can process these features.

[0084] In a Bayesian network, evidence information refers to known facts or observation results. Through these evidence information, other nodes in the network can be inferred. Extracting evidence information from the adapted data can provide the necessary input for the inference of the Bayesian network, thereby accurately inferring potential failure causes.

[0085] In the adapted data, find the known failure phenomena and chip parameters. For example, it is known that the chip has the failure phenomenon of "frequent crashes", and the current chip temperature is "high temperature (>80°C)". Corresponding these known failure phenomena and chip parameters to the corresponding nodes in the Bayesian network. A Bayesian network is a directed acyclic graph, where each node represents a variable, such as a failure phenomenon, a chip parameter, or a potential failure cause. For example, "frequent crashes" is corresponding to the node representing this failure phenomenon in the Bayesian network, and "high temperature (>80°C)" is corresponding to the node representing the chip temperature. Mark the corresponding nodes as observed nodes, indicating that the values of these nodes are known. By marking the observed nodes, the Bayesian network can infer other unobserved nodes based on this known information.

[0086] The inference algorithm of the Bayesian network is used to calculate the posterior probability of unobserved nodes based on the known evidence information (the values of observed nodes). The posterior probability represents the probability of an event occurring given the evidence. In the inference of chip failure causes, by calculating the posterior probability of potential failure cause nodes, the likelihood of each potential failure cause can be understood.

[0087] There are various inference algorithms for Bayesian networks, such as variable elimination algorithm, clique tree propagation algorithm, etc. Different inference algorithms are suitable for different types of Bayesian networks and data scales, and the appropriate inference algorithm can be selected according to the specific situation. Based on the network input (the values of the observed nodes), the inference algorithm will calculate the posterior probabilities of the unobserved potential failure cause nodes according to the structure and conditional probability table of the Bayesian network. The conditional probability table describes the conditional probabilities of each node under different values of its parent nodes. For example, in the case where the chip temperature is known to be "high temperature (>80°C)" and the "frequent system crashes" failure phenomenon occurs, the inference algorithm will calculate the posterior probabilities of potential failure cause nodes such as "heat dissipation module failure" and "chip aging" according to the structure and conditional probability table of the Bayesian network. Through calculation, the posterior probabilities of each potential failure cause node are obtained, and these probabilities form the posterior probability distribution. The posterior probability distribution reflects the likelihood of each potential failure cause occurring under the known evidence.

[0088] The posterior probability distribution gives the likelihood of each potential failure cause occurring. By analyzing the posterior probability distribution, the most likely failure cause can be determined.

[0089] Generally, the potential failure cause node with the highest posterior probability can be selected as the inference result. For example, if the posterior probability of the "heat dissipation module failure" node is the highest, then it can be inferred that the failure cause of the chip is likely to be the heat dissipation module failure. To avoid misjudgment, a posterior probability threshold can be set. Only when the posterior probability of a potential failure cause node exceeds this threshold, will it be used as the inference result. For example, set the threshold to 0.5. If the posterior probability of a potential failure cause node is greater than 0.5, then the failure cause corresponding to this node is considered a possible inference result; if the posterior probabilities of all potential failure cause nodes are less than the threshold, then it may be necessary to further collect evidence or adjust the structure and parameters of the Bayesian network. In some cases, there will be multiple potential failure cause nodes with relatively high posterior probabilities. At this time, the failure causes corresponding to these nodes can be comprehensively considered, and multiple possible failure cause inference results can be given, and these results can be sorted according to the size of the posterior probability to provide a reference for subsequent maintenance and troubleshooting work.

[0090] In one embodiment, for the failure cause inference result, combined with historical maintenance cost data, a cost-benefit analysis algorithm is used to evaluate the costs and expected effects of different maintenance plans, including:

[0091] Analyze the failure cause inference result to determine the maintenance operations to solve the failure; combine the maintenance operations to generate multiple different maintenance plans; where each maintenance plan corresponds to a complete maintenance process;

[0092] Extract the direct cost elements related to each maintenance plan from the historical maintenance cost data to obtain the direct cost of each maintenance plan;

[0093] Analyze the occurrence pattern of indirect costs in similar maintenance situations in historical data, and combine the severity of the current failure and the estimated maintenance duration to estimate the indirect cost of each maintenance plan. Add the direct cost and the indirect cost to obtain the total cost of each maintenance plan;

[0094] Based on historical maintenance data, count the number of successful repair cases after adopting each maintenance plan for the same or similar failure reasons, calculate the proportion of successful repair cases in the total maintenance cases, and predict the failure repair rate of each maintenance plan;

[0095] Analyze the changes in various performance indicators of the chip product before and after maintenance in historical data, and calculate the change score of the performance indicators using the weighted average method;

[0096] Perform a weighted calculation on the failure repair rate and the change score of the performance indicators to obtain a quantitative index of the expected effect.

[0097] In this embodiment, in-depth analysis of the failure cause inference result is the basis for formulating subsequent maintenance plans. By carefully studying the failure cause and combining the design principle, technical specifications of the chip product and past maintenance experience, various maintenance operations required to solve the current failure can be accurately determined. For example, if the failure cause is inferred to be a certain capacitor of the chip being damaged, the corresponding maintenance operation may be to replace the capacitor; if the failure is due to an error in the chip program, the maintenance operation may be to upgrade or reprogram the chip software.

[0098] Reasonably combine the determined maintenance operations to generate multiple different maintenance plans. Each maintenance plan represents a complete maintenance process, covering all links from fault diagnosis to maintenance implementation and finally to final inspection. Different combination methods can meet different maintenance needs and scenarios. For example, for a chip problem with multiple fault points, one maintenance plan may be to focus on dealing with the main fault points first and then the secondary fault points; while another plan may be to perform maintenance in the order of fault occurrence. By generating multiple maintenance plans, more options can be provided for subsequent evaluation to find the optimal plan.

[0099] Historical maintenance cost data is an important basis for calculating direct costs. These data record all the expenses incurred in the past when repairing similar chip failures. From these data, we can extract the direct cost elements related to each repair plan. Direct costs mainly include the cost of spare parts required for repair, the cost of using repair tools, and the labor costs of maintenance personnel, etc. For example, for a repair plan that requires replacing a certain spare part of a chip, we can find the purchase price of this spare part from the historical data; for a plan involving the use of specific repair tools, we can query the rental or usage cost of this tool; at the same time, according to the working hours and wage standards of maintenance personnel, we can calculate the labor costs. Summing up these direct cost elements, we can obtain the direct cost of each repair plan.

[0100] In addition to direct costs, some indirect costs will also be incurred during the repair process. By analyzing the occurrence patterns of indirect costs in similar repair situations in historical data, we can understand the relationships between these indirect costs and factors such as the severity of the failure and the estimated repair duration. For example, in the historical data, it may be found that the more severe the failure and the longer the repair time, the greater the production losses caused by equipment downtime. Combining the specific severity of the current failure and the estimated repair duration, we can more accurately estimate the indirect costs of each repair plan. Indirect costs usually include production losses during the equipment downtime caused by chip repair, additional transportation costs, quality inspection costs, etc.

[0101] Adding the direct cost of each repair plan and the estimated indirect cost, we can obtain the total cost of this repair plan. The total cost is an important indicator for evaluating the economy of the repair plan and plays a key role in the subsequent cost-benefit analysis.

[0102] Historical maintenance data provides rich information for predicting the failure repair rate. By counting the number of successful repair cases after adopting each repair plan for the same or similar failure reasons and calculating the proportion of these successful repair cases in the total number of repair cases, we can obtain the failure repair rate of each repair plan. The failure repair rate reflects the effectiveness of this repair plan in solving specific failures. For example, if a certain repair plan has successfully repaired 80 times in the past 100 repairs for the same failure, then the failure repair rate of this plan is 80%. The higher the failure repair rate, the more likely it is that this repair plan can successfully solve the current failure, and it is one of the important indicators for evaluating the expected effect of the repair plan.

[0103] Analyzing the changes in various performance indicators of chip products before and after repair in historical data can help understand the impact of each repair plan on chip performance. These performance indicators may include the computing speed, stability, power consumption, etc. of the chip. The method of calculating the change score of performance indicators using weighted average is to comprehensively consider the importance of each performance indicator. Different performance indicators may have different weights in the actual use of the chip. For example, for a chip mainly used for high-speed computing, the weight of the computing speed indicator may be relatively high; while for a chip with strict power consumption requirements, the weight of the power consumption indicator may be greater. By assigning corresponding weights to each performance indicator and then calculating the weighted average of the changes in performance indicators before and after repair, a comprehensive change score of performance indicators can be obtained, which can more comprehensively reflect the improvement effect of the repair plan on chip performance.

[0104] To more intuitively compare the expected effects of different repair plans, it is necessary to perform weighted calculation on the fault repair rate and the change score of performance indicators to obtain a quantified indicator of the expected effect. Appropriate weights are assigned to the fault repair rate and the change score of performance indicators respectively, and the determination of the weights can be carried out according to actual business needs and focuses. For example, if more attention is paid to whether the repair plan can successfully solve the fault, the weight of the fault repair rate can be set higher; if more attention is paid to the performance improvement after chip repair, the weight of the change score of performance indicators can be relatively increased. The quantified indicator of the expected effect obtained through weighted calculation can comprehensively reflect the expected effects of each repair plan in solving faults and improving chip performance, providing a strong basis for finally selecting the optimal repair plan.

[0105] In one embodiment, by calculating the cost-benefit ratio, the optimal repair strategy is determined, including:

[0106] Dividing the quantified indicator of the expected effect of each repair plan by its total cost to obtain the corresponding cost-benefit ratio;

[0107] Sorting the cost-benefit ratios of all repair plans and selecting the plan with the highest cost-benefit ratio as the optimal repair strategy.

[0108] In one embodiment, after determining the optimal repair strategy, it further includes:

[0109] Establishing a mapping relationship between the after-sales information of the chip product and the optimal repair strategy and adding it to the data table corresponding to the chip product in the after-sales information management system;

[0110] Obtaining the basic management password of the after-sales information management system;

[0111] Based on the after-sales information of the chip product and the optimal repair strategy, constructing a data matrix;

[0112] Based on the basic management password, adaptively adjust the elements of the data matrix to obtain an adjusted matrix;

[0113] Based on the adjusted matrix, obtain a data table management password to perform permission management on the data table.

[0114] In this embodiment, after determining the optimal maintenance strategy, a mapping relationship is established between the after-sales information of the chip product (such as fault phenomenon, inferred fault cause results, relevant chip parameters, etc.) and the optimal maintenance strategy. This mapping relationship clarifies the best maintenance plan to be adopted for specific chip after-sales information and is an important basis for after-sales information management. For example, when the after-sales information of the chip is "serious overheating and slow operation speed", corresponding to the optimal maintenance strategy of "replacing the heat dissipation module and performing software optimization", the two form a specific mapping.

[0115] Add the established mapping relationship to the data table corresponding to the chip product in the after-sales information management system. The data table is the core storage structure of the after-sales information management system and is used to organize and store various after-sales related information of the chip product. By adding the mapping relationship to the data table, it becomes more convenient to query, count, and analyze after-sales information subsequently, and it is also convenient to track and manage the maintenance situation of the chip product.

[0116] The basic management password is an important security credential for the after-sales information management system and is used for initial access and operation permission verification of the system. Obtaining the basic management password is a prerequisite for subsequent data processing and permission management. This password is usually set and stored by the system administrator and has relatively high security requirements. Obtain the basic management password through legal channels (such as administrator authorization, identity verification, etc.) to ensure that only authorized personnel can perform subsequent operations and safeguard the security and integrity of system data.

[0117] Based on the after-sales information of the chip product and the optimal maintenance strategy, select relevant data elements to construct a data matrix. These data elements can include the time of fault occurrence, fault type, specific steps of the maintenance strategy, estimated maintenance cost, etc. For example, organize the same type of faults occurring at different times and their corresponding maintenance strategy steps into the row and column elements of the data matrix.

[0118] Arrange the selected data elements in matrix form according to certain rules. The structure of the data matrix can be designed according to actual needs. For example, the after-sales information can be used as the rows of the matrix, and the information related to the maintenance strategy can be used as the columns. By constructing the data matrix, complex after-sales information and maintenance strategies can be presented in a structured manner, facilitating subsequent mathematical operations and processing.

[0119] The basic management password plays a crucial role in this process. It provides specific rules and algorithms for the adjustment of matrix elements. Different basic management passwords correspond to different adjustment methods, which is to increase the security and flexibility of data processing. For example, the basic management password can correspond to an encryption algorithm or transformation rule, and the elements of the data matrix are adjusted according to this rule.

[0120] According to the rules corresponding to the basic management password, each element in the data matrix is adaptively adjusted. The adjustment methods can include numerical transformation of elements (such as addition, subtraction, multiplication, and division operations), position swapping, etc. Through this adjustment, the elements of the data matrix better meet specific security and management requirements, and an adjusted matrix is obtained. The adjusted matrix not only contains the original after-sales information and maintenance strategy data but also incorporates password-based security transformations, improving the confidentiality and anti-attack ability of the data.

[0121] According to the elements and structure of the adjusted matrix, a data table management password is generated using a specific algorithm. This algorithm can be a hash algorithm, encryption algorithm, etc. By processing the adjusted matrix, its characteristic information is converted into a unique password. For example, a hash operation is performed on the elements of the adjusted matrix to obtain a hash value of a fixed length as the data table management password.

[0122] The data table management password is used for permission management of the data tables corresponding to chip products in the after-sales information management system. Only those with the correct data table management password can access, modify, delete, etc. the data tables. In this way, the security and confidentiality of the chip after-sales information and maintenance strategies stored in the data tables are ensured, preventing unauthorized access and data leakage. At the same time, it is also convenient for auditing and tracing the operations of the data.

[0123] In one embodiment, based on the basic management password, the matrix elements of the data matrix are adaptively adjusted to obtain an adjusted matrix, including:

[0124] Based on the basic management password, a two-row and two-column digital character matrix is generated;

[0125] Starting from the upper left corner of the data matrix, the digital character matrix is superimposed on the data matrix; when superimposing, if the element in the data matrix is a number, it is added to the number at the overlapping position of the digital character matrix to obtain the adaptively adjusted element;

[0126] The digital character matrix is moved sequentially, and the matrix elements of the data matrix are correspondingly and adaptively adjusted until the digital character matrix moves out of the data matrix to obtain the adjusted matrix; where the moving step of the digital character matrix is to move two positions down and two positions to the right.

[0127] In this embodiment, the basic management password is the key input for the entire adaptive adjustment process of matrix elements. It contains specific information or features and is used to generate a digital character matrix required for subsequent operations. The basic management password can be regarded as a kind of key, which is converted into a digital combination with specific rules through a specific algorithm.

[0128] The basic management password is processed using a preset algorithm to generate a digital character matrix with two rows and two columns. This algorithm can be based on methods such as the hash value of the password, encoding conversion, etc. For example, perform a certain mathematical operation on the characters of the basic management password and arrange the operation results in a 2×2 matrix according to certain rules. The generated digital character matrix is unique and random, and its content is closely related to the basic management password, thus providing specific adjustment rules for subsequent data matrix adjustment.

[0129] Select the upper left corner of the data matrix as the starting point for superposition. This is a clear positioning method, ensuring that the superposition operation of the digital character matrix and the data matrix has a unified starting position. Such a starting point selection facilitates subsequent operations and calculations, and also makes the entire adjustment process consistent and repeatable.

[0130] When the digital character matrix is superposed on the data matrix, the elements at the overlapping positions need to be processed. If the element at the corresponding position in the data matrix is a number, perform a summation operation with the number at the overlapping position of the digital character matrix. This summation operation is a simple and effective data adjustment method. By adding the numbers at the corresponding positions of the two matrices, the value of the element in the data matrix can be changed, thereby achieving a preliminary adjustment of the data matrix. For example, if the element in the upper left corner of the data matrix is 5 and the element at the corresponding position of the digital character matrix is 3, then the value of the element at this position after adjustment becomes 5 + 3 = 8.

[0131] The above digital character matrix moves according to a specific moving step size, that is, it moves two positions both downward and to the right. This fixed moving step size rule ensures that the movement of the digital character matrix on the data matrix is regular, enabling the adjustment operation to comprehensively and orderly cover all parts of the data matrix.

[0132] After each movement of the digital character matrix, perform an adaptive adjustment on the elements in the data matrix at the overlapping positions with the digital character matrix, that is, repeat the element summation operation in the above steps. As the digital character matrix moves continuously, the elements in the data matrix will be adjusted in sequence until the digital character matrix completely moves out of the data matrix. This process continues, ensuring that each element in the data matrix can be affected by the digital character matrix to a certain extent, thereby achieving a comprehensive adjustment of the entire data matrix.

[0133] After multiple movements and element adjustments of the digital character matrix, the elements in the data matrix have all completed the adaptive adjustment. At this time, the new matrix obtained is the adjustment matrix. The adjustment matrix contains the data processed by specific rules. These data not only retain part of the information of the original data matrix but also incorporate the adjustment information brought by the digital character matrix generated based on the basic management password, thus making the adjustment matrix have higher security and specific processing characteristics and can be used for subsequent data encryption, permission management, and other operations.

[0134] In one embodiment, based on the adjustment matrix, obtaining the data table management password includes:

[0135] Generating a plurality of basic numbers based on the basic management password;

[0136] Generating three different graphics based on each of the basic numbers using different generation rules;

[0137] Overlaying the three graphics onto the adjustment matrix according to a preset rule and detecting the intersection regions between any two graphics;

[0138] Extracting the characters in the adjustment matrix located in all the intersection regions and combining them to obtain combined characters as the data table management password.

[0139] In this embodiment, the basic management password is the core starting point of the entire password generation process and contains specific information. This password is composed of letters, numbers, special characters, etc., representing specific access permissions or security level settings. By processing it with a specific algorithm, the potential characteristic information in the password can be mined.

[0140] Using a preset encryption algorithm or mathematical transformation rule to perform operations on the basic management password. For example, a hash function can be used. Taking the basic management password as input, a series of complex calculations are performed to output a series of numbers. These numbers are the basic numbers, which are closely related to the basic management password. Any minor change in the basic management password may result in completely different generated basic numbers. These basic numbers will be used as an important basis for generating graphics in subsequent steps. It can also be to combine the numbers in the basic management password according to rules to obtain the above-mentioned multiple basic numbers.

[0141] Equipping the above basic numbers with multiple unique generation rules, which can be based on the generation principles of geometric graphics, the mapping relationships of mathematical functions, etc. For example, for the above basic numbers, they can be used as parameters for the radius of a circle, and the coordinate points of the circle can be calculated through a mathematical formula to generate a circular graphic; or a polygon can be generated according to a specific polygon vertex coordinate generation algorithm.

[0142] Due to the adoption of different generation rules, the three generated figures are unique in shape, size, and position. These figures may include common geometric figures such as circles, rectangles, triangles, etc., or may be irregular figures defined by complex mathematical functions. Their specific forms depend on the base numbers and the corresponding generation rules, and this diversity increases the complexity and security of the subsequent password generation process.

[0143] The preset rule stipulates how to place the three generated figures on the adjustment matrix. This rule can be based on the matrix coordinate system, for example, specifying that a certain vertex of the figure is aligned with a specific position of the matrix, or superimposing them according to a certain interval and arrangement. Through this preset rule, it is ensured that the relative position relationship between the figure and the adjustment matrix is predictable and repeatable.

[0144] After the three figures are superimposed on the adjustment matrix, it is necessary to detect the intersection area between any two figures. This involves analyzing and calculating the geometric features of the figures, such as determining the intersection area by judging whether the boundary coordinates of the figures overlap. For complex irregular figures, more advanced geometric algorithms, such as Boolean operations, etc., are required to accurately calculate the intersection area. These intersection areas are an important part of extracting key information subsequently.

[0145] After determining all the intersection areas, extract the characters located within these intersection areas from the adjustment matrix. These characters are the data at specific positions in the adjustment matrix, and they are screened out after the previous figure superposition and intersection area detection, and have specific meanings and correlations.

[0146] Combine the extracted characters in a certain order. This order can be fixed rules such as from left to right, from top to bottom, etc. The combined characters form a new string, and this string is the data table management password. Since this password is generated based on the basic management password, the generated figures, and the specific intersection areas of the adjustment matrix, it has high randomness and uniqueness, and can provide reliable access control and security protection for the data table.

[0147] In one embodiment, obtaining the data table management password based on the adjustment matrix specifically includes:

[0148] Divide the adjustment matrix into multiple sub - matrices according to the preset number of rows and columns to obtain a set of sub - matrices.

[0149] Adopt the matrix eigenvalue calculation method to calculate the eigenvalues for each sub - matrix to obtain a set of eigenvalues corresponding to each sub - matrix.

[0150] Perform a hash operation on the set of eigenvalues, convert each eigenvalue into a hash value of a fixed length, and concatenate the hash values of all sub - matrices in the order of their positions in the adjustment matrix to obtain a concatenated hash value.

[0151] Use the basic management password as the key and perform an encryption operation on the concatenated hash value using a symmetric encryption algorithm (such as the AES algorithm). During the encryption process, generate an encryption key and an initialization vector based on the basic management password, and perform block encryption on the concatenated hash value to obtain an encrypted hash value.

[0152] According to a predefined character mapping table, each byte in the encrypted hash value is mapped to a specific character in the character mapping table. Traverse each byte of the encrypted hash value and convert it to the corresponding character according to the character mapping table. The final character sequence is the data table management password.

[0153] In one embodiment, obtaining the data table management password based on the adjustment matrix specifically includes the following steps:

[0154] Extract data features from the adjustment matrix, calculate the statistical features of the matrix elements, including but not limited to mean, variance, skewness, and kurtosis. When calculating, divide the matrix by rows and columns respectively, calculate these statistical feature values of the elements in each row and each column, and obtain a set of feature vectors containing row features and column features.

[0155] Convert the set of feature vectors into binary encoding. Convert each statistical feature value into a binary sequence of a fixed length through a preset encoding rule. For example, if the statistical feature value is a floating - point number, its fractional part and integer part can be encoded separately and then combined into a complete binary sequence. Connect the binary sequences corresponding to all feature vectors to form a long binary string.

[0156] Obtain the current timestamp accurate to milliseconds and convert the timestamp into binary form. Use an encryption algorithm (such as RSA asymmetric encryption), with the basic management password as the private key, to perform an encryption and fusion operation on the binary string and the timestamp binary. During the encryption process, insert the timestamp binary into a specific position of the binary string, and then perform an encryption operation to obtain the encrypted binary data.

[0157] Perform a hash operation on the encrypted binary data using a secure hash algorithm to generate a hash digest of a fixed length. The above hash digest has uniqueness and irreversibility, and can effectively ensure the integrity and security of the data.

[0158] Based on the data features of the hash digest, use the cyclic redundancy check (CRC) algorithm to generate a check code. Combine the check code with the hash digest and form the final combined code according to the rules.

[0159] According to a preset character mapping table, the binary data in the combined code is converted into a printable character sequence, and the finally obtained character sequence is the data table management password. The above character mapping table can be customized to establish a one-to-one correspondence between the binary sequence and letters, numbers, and special characters.

[0160] Referring to Figure 2 , in another embodiment of the present invention, a post-sales information management system for chip products is further provided, including:

[0161] A processing module for performing text processing on various types of unstructured data in the post-sales information of chip products to obtain structured post-sales basic data;

[0162] A mining module for performing mining analysis on the post-sales basic data based on the association rule mining algorithm to mine the association relationship between different fault phenomena and chip parameters, and obtain fault association rule data;

[0163] An inference module for inferring the cause of the fault for the fault association rule data by using the Bayesian network algorithm to obtain a fault cause inference result;

[0164] A management module for, in view of the fault cause inference result, combining historical maintenance cost data, adopting a cost-benefit analysis algorithm, evaluating the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the post-sales information of the chip product.

[0165] In this embodiment, for the specific implementation of each module in the above system embodiment, please refer to that described in the above method embodiment and will not be elaborated here.

[0166] Referring to Figure 3 , in an embodiment of the present invention, a computer device is further provided. The computer device can be a server, and its internal structure can be as Figure 3 shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected through a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0167] Those skilled in the art can understand, Figure 3The structure shown is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0168] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0169] In summary, the present invention provides a method and system for managing after-sales information of a chip product in an embodiment, including: performing text processing on various types of unstructured data in the after-sales information of the chip product to obtain structured after-sales basic data; performing mining analysis on the after-sales basic data based on an association rule mining algorithm to mine the association relationship between different fault phenomena and chip parameters to obtain fault association rule data; using a Bayesian network algorithm to infer the cause of the fault for the fault association rule data to obtain a fault cause inference result; for the fault cause inference result, combining historical maintenance cost data, using a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the after-sales information of the chip product. In the present invention, by inferring the fault cause inference result based on the after-sales information, then combining the historical maintenance cost data, using a cost-benefit analysis algorithm to evaluate the costs and expected effects of different maintenance plans, and determining the optimal maintenance strategy by calculating the cost-benefit ratio, a reasonable maintenance plan is made, thereby avoiding waste of resources and achieving the maximization of enterprise interests.

[0170] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0171] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, device, article or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, device, article or method including that element.

[0172] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall equally be included in the patent protection scope of the present invention.

Claims

1. A method for after-sales information management of chip products, characterized in that: The following steps are involved: Perform text processing on various unstructured data in the after-sales information of chip products to obtain structured after-sales basic data; Based on the association rule mining algorithm, the basic after-sales data is mined and analyzed to find out the association between different fault phenomena and chip parameters, and obtain fault association rule data; Using the Bayesian network algorithm to infer the fault cause of the fault association rule data, and obtaining the fault cause inference result; Based on the inference results of the fault cause, combined with historical maintenance cost data, a cost-benefit analysis algorithm is used to evaluate the costs and expected effects of different maintenance plans. By calculating the cost-benefit ratio, the optimal maintenance strategy is determined as the management result of the after-sales information of the chip product.

2. The after-sales information management method of chip products according to claim 1, characterized in that: Various types of unstructured data in after-sales information include user feedback text, chip operation logs, and maintenance record documents; The chip parameters include chip model, operating environment, and operating time.

3. The after-sales information management method of chip products according to claim 1, characterized in that: The Bayesian network algorithm is used to infer the fault cause of the fault association rule data to obtain the fault cause inference result, including: According to the input requirements of the Bayesian network algorithm, the fault association rule data is adapted to obtain adapted data; the adaptation process includes adjusting the data feature format and value range, discretizing continuous features, and uniformly encoding categorized features; Extracting evidence information from the adaptation data, corresponding known fault phenomena and chip parameters to corresponding nodes of the Bayesian network algorithm, marking them as observed nodes, and obtaining network input; The inference algorithm in the Bayesian network algorithm is used to calculate the posterior probability of the unobserved potential fault cause node based on the network input to obtain the posterior probability distribution; The fault cause inference result is determined based on the posterior probability distribution.

4. The after-sales information management method of chip products according to claim 1, characterized in that: Based on the inference results of the fault cause, combined with historical maintenance cost data, a cost-benefit analysis algorithm is used to evaluate the costs and expected effects of different maintenance plans, including: Analyze the fault cause inference results to determine the maintenance operation to solve the fault; combine the maintenance operations to generate multiple different maintenance plans; each maintenance plan corresponds to a complete maintenance process; Extract the direct cost elements related to each maintenance plan from the historical maintenance cost data to obtain the direct cost of each maintenance plan; Analyze the indirect cost occurrence patterns of similar maintenance situations in historical data, estimate the indirect cost of each maintenance plan based on the severity of the current fault and the estimated maintenance duration, and add the direct cost to the indirect cost to get the total cost of each maintenance plan; Based on historical maintenance data, count the number of successful repair cases after using various maintenance solutions for the same or similar fault causes, calculate the proportion of successful repair cases in the total maintenance cases, and predict the fault repair rate of each maintenance solution; Analyze the changes in various performance indicators of chip products before and after maintenance in historical data, and use the weighted average method to calculate the change score of performance indicators; The fault repair rate and the change score of the performance index are weighted and calculated to obtain a quantitative index of the expected effect.

5. The after-sales information management method of chip products according to claim 4, characterized in that: Determine the optimal maintenance strategy by calculating the cost-benefit ratio, including: Divide the expected effect quantified index of each maintenance plan by its total cost to obtain the corresponding cost-effectiveness ratio; The cost-effectiveness ratios of all maintenance options are ranked, and the option with the highest cost-effectiveness ratio is selected as the optimal maintenance strategy.

6. The after-sales information management method of chip products according to claim 1, characterized in that: After determining the optimal maintenance strategy, it also includes: Establishing a mapping relationship between the after-sales information of the chip product and the optimal maintenance strategy, and adding the mapping relationship to a data table corresponding to the chip product in the after-sales information management system; Obtain the basic management password of the after-sales information management system; Building a data matrix based on the after-sales information of the chip product and the optimal maintenance strategy; Based on the basic management password, adaptively adjusting the matrix elements of the data matrix to obtain an adjusted matrix; Based on the adjustment matrix, a data table management password is obtained to perform authority management on the data table.

7. The after-sales information management method of chip products according to claim 6, characterized in that: Based on the basic management password, adaptively adjusting the matrix elements of the data matrix to obtain an adjusted matrix includes: Based on the basic management password, generate a two-row and two-column digital character matrix; Taking the upper left corner of the data matrix as the starting point, the digital character matrix is ​​superimposed on the data matrix; when superimposing, if the elements in the data matrix are numbers, the numbers at the overlapping positions of the digital character matrix are summed to obtain the adaptively adjusted elements; The digital character matrix is ​​moved in sequence, and the matrix elements of the data matrix are adaptively adjusted accordingly, until the digital character matrix is ​​moved out of the data matrix to obtain an adjusted matrix; wherein the moving step length of the digital character matrix is ​​two positions downward and rightward.

8. The after-sales information management method of chip products according to claim 6, characterized in that: Based on the adjustment matrix, a data table management password is obtained, including: Based on the basic management password, generate multiple basic numbers; Based on each of the basic numbers, three different graphics are generated using different generation rules; Superimposing three graphics into the adjustment matrix according to a preset rule, and detecting the intersection area between any two graphics; The characters in all the intersecting areas of the adjustment matrix are extracted and combined to obtain a combined character as the data table management password.

9. An after-sales information management system for chip products, characterized in that: include: A processing module is used to perform text processing on various types of unstructured data in the after-sales information of chip products to obtain structured after-sales basic data; The mining module is used to mine and analyze the basic after-sales data based on the association rule mining algorithm, to mine the association between different fault phenomena and chip parameters, and to obtain fault association rule data; An inference module, used to use a Bayesian network algorithm to infer the cause of the fault on the fault association rule data to obtain an inference result of the cause of the fault; The management module is used to infer the cause of the fault, combine historical maintenance cost data, use a cost-benefit analysis algorithm to evaluate the cost and expected effect of different maintenance plans, and determine the optimal maintenance strategy by calculating the cost-benefit ratio as the management result of the after-sales information of the chip product.

10. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images