BOM data management method and system based on PLM

Abstract

The invention relates to the technical field of data processing, in particular to a PLM-based BOM data management method and system, and the method comprises the steps: obtaining product BOM basic data, complementing missing parameters needed by MTBF calculation, and obtaining a calculation-ready BOM; splitting function fragments for the calculation-ready BOM according to a hierarchical structure, deploying MTBF calculation nodes, and calling the function fragments for calculation to obtain a dynamic reliability calculation result; a dynamic reliability calculation result is analyzed through a decision tree algorithm, BOM potential fault points are identified, risks are quantified, an optimization scheme is generated by combining PLM with a component reliability knowledge base, and an optimized reliability calculation result is obtained; and based on an optimized reliability calculation result, integrating key value-added contents through an intelligent enhancement system, normalizing a report format according to bid invitation requirements, and generating a standardized reliability prediction report. According to the scheme, a BOM full-process closed-loop mechanism with deep cooperation of the PLM and the MTBF is constructed, and the reliability and the accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a PLM-based BOM data management method and system. Background Technology

[0002] With the rapid increase in the company's market bidding needs, the reliability department's role in producing reliability forecast reports has become increasingly important. The accuracy and professionalism of these reports not only directly reflect the quality of the company's product design but also serve as a key persuasive factor in competing fiercely with rivals. To improve the efficiency and accuracy of this process, the company introduced MTBF software, aiming to optimize the reliability forecasting process for relevant projects through software tools.

[0003] However, a significant challenge in using this software for reliability forecasting is the need to manually import the BOM (Bill of Materials) of specific components into the software for calculation. This process is not only massive in scale but also makes it difficult to ensure the accuracy and timely updates of the input information, resulting in a substantial investment of manpower without achieving the desired output.

[0004] For companies that have already developed PLM (Product Lifecycle Management) projects, the traditional manual import method is clearly no longer sufficient to meet the growing business demands. Therefore, finding a more efficient, accurate, and reliable forecasting method that can seamlessly integrate with PLM projects is particularly urgent. Summary of the Invention

[0005] This invention significantly improves the efficiency and accuracy of reliability prediction by constructing a fully automated closed-loop mechanism for the entire BOM process that deeply collaborates with PLM and MTBF, avoids human error, meets the professional requirements of bidding reports, and provides data support for product reliability optimization.

[0006] The technical solution proposed in this invention is: a PLM-based BOM data management method, the method comprising: Obtain the product's BOM basic data, combine it with the PLM's built-in component reliability knowledge base to complete the missing parameters required for MTBF calculation, and after format regularization and integrity verification, obtain the calculation-ready BOM. Functional segments are decomposed based on the hierarchical structure of the computation-ready BOM, and MTBF computing nodes are deployed. Segmented calculations are performed based on parallel function calls. PLM is used to synchronize the actual operating data of the entire product lifecycle to the MTBF computing nodes, dynamically adjust the reliability parameters of components, and obtain dynamic reliability calculation results by integrating the calculation results through the PLM collaborative hub and verifying consistency. The dynamic reliability calculation results are analyzed by decision tree algorithm, potential failure points in BOM are identified and risks are quantified, and failure and risk information is pushed to PLM system. PLM combines component reliability knowledge base to generate optimization scheme, update calculation ready BOM and synchronize to MTBF calculation node for secondary calculation to obtain optimized reliability calculation results. Based on the optimized reliability calculation results, key value-added content is integrated through the intelligent enhancement system, and the report format is standardized according to the bidding requirements to generate a standardized reliability prediction report.

[0007] Preferably, the specific process for obtaining the computation-ready BOM is as follows: By connecting to the product design module through the standardized interface of the PLM system, basic BOM data is collected to form the original BOM dataset. The PLM's built-in component reliability knowledge base is invoked. Based on the component identifiers in the original BOM dataset, parameter matching is performed to extract the core parameters required for MTBF calculation. For missing parameters that are not matched, they are completed according to the parameter completion rules for similar components to obtain a complete parameter BOM. According to the parameter format requirements preset by MTBF, the fields of the complete parameter BOM are standardized and organized to unify the parameter units and data types; The system sequentially performs checks for missing core parameters in MTBF calculation, compliance of parameter format, and rationality of BOM hierarchical structure, removing invalid data and abnormal levels. Once the checks pass, a calculation-ready BOM is generated.

[0008] Preferably, the specific process of splitting the computation-ready BOM into functional fragments according to the hierarchical structure and deploying MTBF compute nodes to call the functional fragments for parallel computation is as follows: Analyze the hierarchical structure of the computation-ready BOM, identify the functional boundaries and independent computing units of each module, and determine the granularity of the partitioning. Prioritize the segments based on the importance of their functionalities; Each fragment is labeled with a unique identifier and an upstream and downstream computation dependency identifier. The computation result of the preceding fragment of a fragment with a dependency relationship becomes the input of the subsequent fragment. Based on the number of shards and computational complexity, corresponding MTBF computing nodes are dynamically deployed. Each shard is allocated to an idle computing node through a load balancing algorithm. The MTBF computing nodes call the shard data in priority order to perform parallel computation.

[0009] Preferably, the specific process of PLM synchronizing actual operating data throughout the product's lifecycle to the MTBF computing node to dynamically adjust component reliability parameters is as follows: By connecting the PLM system to R&D testing equipment, production monitoring system and on-site operation data acquisition terminal, the actual working condition data of the entire life cycle is collected. The operating condition data is time-series normalized and outlier removed at preset time intervals to generate a standardized operating condition dataset, which is then synchronized to each MTBF computing node in real time through a message queue mechanism. Extract component information corresponding to each slice using each MTBF computing node, and query and match the corresponding industry standard correction factor based on operating condition data. The calculation parameters are dynamically updated based on the correction factor and the original reliability parameters of the components using a preset parameter adjustment formula.

[0010] Preferably, the specific process for obtaining the dynamic reliability calculation results is as follows: After completing the sharding calculation using each MTBF computing node, the calculation results, along with the sharding identifier and calculation timestamp, are uploaded to the PLM collaboration hub. The calculation results are aligned according to the segmentation hierarchy through the PLM collaboration hub, the fusion weights are assigned based on the importance of module functions, and the preliminary reliability results of the whole machine are calculated through a weighted summation algorithm. Set a consistency check threshold, compare the deviation of each shard calculation result with the historical calculation data of the same type of shard, and identify the common deviation of the calculation results of different nodes to identify abnormal results that exceed the limit; For the fragments corresponding to abnormal results, recalculation is triggered, and the corrected fragment results are merged again until all results meet the consistency requirements, and the dynamic reliability calculation results are output.

[0011] Preferably, the specific process for identifying potential failure points in the BOM and quantifying the risks is as follows: Load the dynamic reliability calculation results and combine them with PLM historical fault data and industry fault database to construct a fault feature vector library; The decision tree algorithm is used to perform feature matching on outlier parameters and unreliable items in the calculation results to identify potential failure points; A risk quantification model is constructed from two dimensions: the probability of failure and the degree of failure impact. The risk value of each failure point is calculated and the risk level is classified. Generate a fault risk analysis report that includes the location of the fault point, characteristic description, and risk level.

[0012] Preferably, the specific process for obtaining the optimized reliability calculation result is as follows: Push the failure risk analysis report to the PLM system and use PLM to associate the corresponding computationally ready BOM node with the component reliability knowledge base. Based on the evaluation criteria, a multi-dimensional set of optimization solutions is generated, including component replacement, BOM structure adjustment, and parameter threshold correction. The optimal solution is selected through cost and reliability balance analysis, the feasibility of the solution is verified, and the final optimized solution is generated after confirming that there are no supply chain bottlenecks or structural conflicts. The corresponding fields of the compute-ready BOM are automatically updated based on the final optimization scheme, BOM version update logs are generated synchronously, and the updated compute-ready BOM is pushed to the MTBF compute node. MTBF compute nodes load the updated compute-ready BOM, reuse actual operating data and parameter adjustment rules throughout the product lifecycle, and perform secondary reliability calculations. The validity of the secondary calculation results is verified. After confirming that the results meet the reliability improvement expectations and there are no data anomalies, the optimized reliability calculation results are output.

[0013] Preferably, the specific process for obtaining the standardized reliability prediction report is as follows: Extract three types of value-added content from the PLM system: fault optimization and source tracing records, reliability benchmarking data of similar products, and endorsement of historical project verification results. The extracted value-added content is categorized, integrated, and logically organized, and arranged according to the chapter structure required by the tendering party. A standardized report template was used to unify the font and format, and the value-added content was organically integrated with the optimized reliability calculation results to generate a draft report; Embedded one-click traceability module, linking key data in the report with the PLM system's raw data, calculation logs, and optimization records; The initial draft report undergoes compliance verification, and a standardized reliability prediction report is generated after the verification is passed.

[0014] The present invention also provides a PLM-based BOM data management system, the system being used to execute the aforementioned PLM-based BOM data management method.

[0015] The present invention also provides a computer-readable storage medium storing a computer program, which is executed by a processor to implement the PLM-based BOM data management method described above.

[0016] The beneficial effects of this invention are: 1. By embedding a parameter completion module before PLM, designing an adaptive sharding engine and adaptive mapping engine at the BOM level, an integrated mechanism for BOM data preparation, computing architecture, and software adaptation is built. This completely solves the problems of high error and low efficiency in traditional manual BOM import, improving BOM data preprocessing efficiency by more than 60%. Distributed sharding and parallel computing significantly shorten the reliability prediction time for complex products, perfectly adapting to the timeliness requirements of batch bidding projects. The adaptive mapping capability can automatically adapt to different versions / types of MTBF software, eliminating the need for repeated manual interface configuration, significantly reducing the complexity and trial-and-error costs of cross-system adaptation, and providing efficient and stable data and architecture support for subsequent full-process optimization.

[0017] 2. Relying on the PLM full lifecycle operating condition data acquisition and operating condition parameter correction engine, the reliability calculation parameters are dynamically bound to the actual operating conditions of the product, replacing the traditional fixed parameter calculation. This reduces the deviation between the reliability prediction results and the actual operating effect to within ±5%, significantly improving the reference value of the data. The bidirectional data flow mechanism between PLM and MTBF allows the calculation results to trigger BOM optimization suggestions in reverse. After confirmation, the BOM is automatically updated and a second calculation verification is performed, forming a closed loop. This effectively reduces the product reliability risk caused by insufficient operating condition adaptation and promotes the dynamic iterative upgrade of BOM design.

[0018] 3. Based on PLM historical fault data and industry fault database, it accurately identifies potential faults at the component and module levels, avoids testing blind spots in advance, and reduces the cost of rework due to product failures in the later stages; the fault-design linkage optimization engine transforms fault points into targeted optimization solutions, and verifies the effect through secondary calculations after updating the BOM, ensuring that the testing effectiveness of the system continues to improve during long-term use; the intelligent enhancement system for bidding reports automatically integrates value-added content such as fault optimization tracing, competitor benchmarking analysis, and historical data endorsement, and generates standardized reports according to bidding requirements without additional manual processing. At the same time, the one-click tracing function can generate a core data verification report within 10 seconds, significantly improving the differentiated competitiveness in bidding review and adapting to the efficient delivery needs of batch bidding projects. Attached Figure Description

[0019] Figure 1 A flowchart of a PLM-based BOM data management method; Figure 2 This is a flowchart illustrating the management process of a PLM-based BOM data management method. Detailed Implementation

[0020] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0021] It is understood that the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0022] like Figure 1 and Figure 2 As shown, this invention discloses a PLM-based BOM data management method, the method comprising: The process begins by acquiring basic product BOM data and supplementing it with missing parameters for MTBF calculation using the component reliability knowledge base built into the PLM (Product Lifecycle Management) system. After format standardization and integrity verification, a computationally ready BOM is obtained. This BOM is then divided into functional segments based on its hierarchical structure. MTBF calculation nodes are deployed to perform parallel calculations on these functional segments. The PLM synchronizes actual operating data throughout the product's lifecycle to the MTBF calculation nodes to dynamically adjust component reliability parameters. The calculation results are then integrated and their consistency verified through the PLM's collaborative hub to obtain dynamic reliability calculation results. These results are analyzed using a decision tree algorithm to identify potential BOM failure points and quantify risks. Failure and risk information is pushed to the PLM system. The PLM, combined with the component reliability knowledge base, generates optimization schemes, updates the computationally ready BOM, and synchronizes it to the MTBF calculation nodes for secondary calculations, resulting in optimized reliability calculation results. Based on these optimized reliability calculation results, key value-added content is integrated through an intelligent enhancement system, and the report format is standardized according to bidding requirements to generate a standardized reliability prediction report.

[0023] Specifically, the process for obtaining the ready-to-use BOM is as follows: By connecting to the product design module through the standardized interface of the PLM system, basic BOM data is collected to form the original BOM dataset. The PLM's built-in component reliability knowledge base is invoked. Based on the component identifiers in the original BOM dataset, parameter matching is performed to extract the core parameters required for MTBF calculation. For missing parameters that are not matched, they are completed according to the parameter completion rules for similar components to obtain a complete parameter BOM. According to the parameter format requirements preset by MTBF, the fields of the complete parameter BOM are standardized and organized to unify the parameter units and data types; The system sequentially performs checks for missing core parameters in MTBF calculation, compliance of parameter format, and rationality of BOM hierarchical structure, removing invalid data and abnormal levels. Once the checks pass, a calculation-ready BOM is generated.

[0024] Among them, the PLM system is an information management system for the entire product lifecycle, covering the entire process of product design, R&D, production, operation and maintenance. Its standardized interface can adopt the Open Platform Communications Unified Architecture (OPCUA) interface to ensure efficient data communication with the product design module. The BOM basic data includes core data such as product part number, specifications, supplier information, hierarchical relationship (complete machine-module-component-part), and functional attributes, which are the basis for subsequent parameter completion and calculation.

[0025] In detail, the component reliability knowledge base built into PLM contains historical test data of components (such as reliability verification results in past projects), reliability parameters provided by suppliers (such as indicator data in factory inspection reports), industry standard parameters (such as reliability benchmark values ​​published by the state or industry), and failure statistics of similar products, providing authoritative data support for parameter matching and completion. The parameter completion rules for similar components can adopt interpolation completion method or mean completion method based on functional characteristics and application scenarios. For example, when a certain resistor component is missing a reliability parameter, it can be completed based on the mean parameter of resistors with the same power and precision in the knowledge base, ensuring the rationality of the completed parameter.

[0026] Specifically, field standardization and normalization are based on the parameter identification protocol of MTBF computing nodes, and perform full-dimensional standardization processing from three dimensions: field naming, parameter units, and data types. This ensures that computing nodes can directly parse and call the data. Field naming standardization adopts a unified naming rule of "parameter category - core attribute - measurement dimension" (such as "component - working life - hours" and "module - failure probability - percentage"). Parameter unit unification is enforced according to the default unit system of the MTBF calculation model, clarifying the conversion logic and precision retention standards for various parameters. Time-related parameters are uniformly converted to "hours (h)". Minutes (min) are converted to "1h = 60min", seconds (s) are converted to "1h = 3600s", and both are rounded to two decimal places. Probability parameters are uniformly converted to decimal form (unitless). Percentages (%) are converted to "decimal = percentage / 100", and parts per million (ppm) are converted to "decimal = ppm / 1,000,000", and both are rounded to six decimal places. For stress parameters (such as voltage and temperature), voltage is uniformly converted to "volts (V)", current to "amperes (A)", temperature to "degrees per Celsius (°C)", and power to "watts (W)", all rounded to one decimal place. Dimensional parameters are uniformly converted to... Millimeters (mm), centimeters (cm) are converted according to "1mm = 0.1cm", and meters (m) are converted according to "1mm = 0.001m", retaining two decimal places; data type standardization clearly defines the target data type and conversion rules for various parameters to ensure data parsing compatibility of computing nodes. Numerical parameters are uniformly converted to 64-bit floating-point numbers (Double), character values ​​(such as "10.5V" and "50%)) are first stripped of unit characters before conversion, and abnormal characters (such as letters and symbols) are directly removed and treated as missing values. Date and time parameters are uniformly converted to "YYYY-MM-DDHH:MM:SS". The format (e.g., "2024-05-2014:30:00") and string-type dates (e.g., "May 20, 2024") are extracted by regular expression matching and then reconstructed. Boolean parameters are uniformly converted to "1 (yes) / 0 (no)". Character expressions (e.g., "qualified / unqualified" "yes / no") are converted according to mapping rules. Data without explicit mapping is marked as invalid data. Enumerated parameters (e.g., component type, fault level) are converted to integer codes according to the PLM preset enumeration dictionary (e.g., "resistor=1, capacitor=2, inductor=3"). Enumerated values ​​not in the dictionary are marked as "999 (unknown)".

[0027] Completeness verification is achieved through a three-level verification process, which includes core parameter missing verification, parameter format compliance verification, and BOM hierarchy structure rationality verification. Core parameter missing verification requires checking three categories of parameters in the list of critical parameters necessary for MTBF calculation (based on industry standards and calculation model requirements such as GJB / Z299C, MIL-HDBK-217F): basic component parameters (part number, specification model, supplier code, material type, rated parameters (rated voltage / current / power)), reliability characteristic parameters (basic failure rate, operating stress level, environmental adaptability level, life test data), and hierarchical relationship parameters (parent module code, child component code, assembly relationship identifier). This involves traversing the complete BOM line by line for verification, performing mandatory field non-empty checks according to the above list, and verifying any missing critical parameter. If a parameter is missing, it is judged as "validation failed". BOM lines with missing key parameters are directly removed, and a removal log (including part number, missing parameter name, and removal time) is recorded. A completion reminder is also pushed to the PLM system. Parameter format compliance verification includes type compliance verification, value range verification, and format compliance verification. Type compliance verification verifies that the actual data type of each parameter is consistent with the standardization requirements (e.g., "basic failure rate" must be numeric, not character). Inconsistency is considered invalid. Value range verification is based on industry standards, component specifications, and calculation model constraints, clearly defining the valid value range for each parameter. The basic failure rate must satisfy 0 < numeric value ≤ 0.1 (unit: ...). If the value exceeds the specified range, it is considered abnormal. The working stress level must be between 1 and 5 (level 1 is the lowest and level 5 is the highest). Any integer value outside this range is considered abnormal. The environmental adaptability level must be AF (level A is the best and level F is the worst). Any value outside this enumeration range is considered abnormal. The rated voltage is set according to the component type (e.g., capacitor 1V-1000V, resistor 0V-500V). Any value exceeding this range is considered abnormal. Format compliance verification requires string parameters (e.g., part number, supplier code) to conform to preset regular expression rules, and date and time parameters to match the format "YYYY-MM-DDHH:MM:SS". Any non-compliance is considered abnormal. The handling mechanism is that parameters with abnormal type / format are marked as "format error", and parameters with abnormal value range are marked as "value exceeds limit". The original data is retained and an alert is pushed. The user needs to confirm whether to correct it. Before correction, it will not be included in the calculation. The BOM hierarchy structure rationality verification includes subordinate relationship uniqueness verification, hierarchy depth compliance verification, module code uniqueness verification, and assembly relationship validity verification. Subordinate relationship uniqueness verification is performed by comparing the parent module code with the child component. The system verifies the association between part codes, checking for conflicts such as "the same sub-component / part being associated with multiple parent module codes" (e.g., part A being associated with both module 1 and module 2 codes). If a conflict exists, it is considered an anomaly. The system also verifies the compliance of the BOM hierarchy, specifying a maximum of 5 levels (Complete Machine → First-Level Module → Second-Level Module → Component → Part). BOM lines exceeding the hierarchy depth are considered anomalies. The system verifies the uniqueness of module codes, checking for duplicate module codes within the same hierarchy (e.g., two second-level modules with the same code under the same first-level module). Duplicate codes are considered anomalies. The system verifies the assembly logic matching between parent modules and sub-components / parts (e.g., "Power Module" can only be associated with electrical components such as "Capacitors, Resistors, and Chips," and cannot be associated with non-electrical components such as "Mechanical Housing"). If they do not match, they are considered anomalies. The handling mechanism is to directly remove BOM lines with conflicting subordinate relationships or duplicate module codes, and mark BOM lines with excessive hierarchy depth or invalid assembly relationships as "structural anomalies" and send a warning. Users can adjust the hierarchy or assembly relationship and then re-verify.

[0028] This step automates the collection of BOM data through the standardized interface of the PLM system, avoiding errors from manual input; it uses a component reliability knowledge base to intelligently complete parameters, solving the problems of low efficiency and error-proneness of traditional manual completion; and it ensures the integrity and compliance of BOM data through multi-dimensional verification, providing a high-quality data foundation for subsequent distributed computing.

[0029] Specifically, the process of splitting the compute-ready BOM into functional fragments according to the hierarchical structure and deploying MTBF compute nodes to call the functional fragments for parallel computation is as follows: Analyze the hierarchical structure of the computation-ready BOM, identify the functional boundaries and independent computing units of each module, and determine the granularity of the partitioning. Prioritize the segments based on the importance of their functionalities; Each fragment is labeled with a unique identifier and an upstream and downstream computation dependency identifier. The computation result of the preceding fragment of a fragment with a dependency relationship becomes the input of the subsequent fragment. Based on the number of shards and computational complexity, corresponding MTBF computing nodes are dynamically deployed. Each shard is allocated to an idle computing node through a load balancing algorithm. The MTBF computing nodes call the shard data in priority order to perform parallel computation.

[0030] In detail, when splitting a computationally ready BOM into functional segments based on its hierarchical structure, it is necessary to first analyze the complete hierarchical relationship between the whole machine, module, component, and part, and then determine the granularity of the segmentation by combining the product's complexity metrics. The specific details are as follows: For applications with more than 500 components and ≥8 modules (e.g., complex electronic devices, industrial control systems), the components should be split into "first-level modules - second-level modules," such as "power module - AC / DC conversion sub-module" or "control module - main control chip sub-module." Each sub-module should contain 50-100 components, and each sub-module must correspond to an independent functional unit (e.g., power supply function, signal processing function) to ensure complete and independent computing capabilities and avoid dependency confusion caused by cross-functional splitting. For applications with 301-500 components and 5-7 modules (e.g., smart home appliances, small industrial equipment), the components should be split into "first-level modules - core components," such as "refrigeration module - compressor assembly" or "display module - screen driver assembly." Each sub-module should contain 80-120 components, balancing computing efficiency and module interrelationships. Core components must include the key functional components of the module, and non-core auxiliary components can be combined into 1-2 sub-modules. For products with a total number of components ≤300 and a number of modules <5 (such as simple mechanical products and basic electronic components), the components are split into "core components - sub-components", such as "transmission components - gear components" and "actuation components - cylinder components". The number of components in a single segment should not exceed 150. If the number of components in a core component is ≤50, it can be merged with functionally related sub-components into a single segment to avoid scheduling redundancy caused by too many segments. If a module has a "high concentration of components" (the number of components in a single module is >200 and accounts for more than 30% of the total number of components in the product), regardless of the overall complexity of the product, the module is split separately and further subdivided into "functional sub-units" to ensure that the computational load of a single segment is balanced (the difference between the maximum and minimum computational load is ≤30%). If the product has shared components across modules (such as shared power supplies and general interfaces), the shared components are treated as "shared segments" and marked as the preceding segments of all dependent modules, and their computation is performed first.

[0031] The module functional importance score is determined using the analytic hierarchy process (AHP). The evaluation dimensions include functional criticality (whether it is essential for the core operation of the product), fault impact scope (whether the fault will cause the entire machine to stop), and cost ratio (the proportion of the module's cost to the total product cost). Priorities are assigned based on the weighted scores of the three factors (with weights of 0.5, 0.3, and 0.2, respectively). Core functional modules with a functional importance score ≥ 80 (such as the main control module and power supply module) are assigned to the first priority level. Auxiliary functional modules with a functional importance score of 50-79 (such as communication modules and interface modules) are assigned to the second priority level. Non-critical modules with a functional importance score < 50 (such as heat dissipation modules and decorative parts) are assigned to the third priority level, ensuring that computing resources are allocated to critical modules first. Each fragment is labeled with a unique identifier and upstream / downstream computation dependency identifiers. The unique identifier adopts a structured format of "product code-level code-module code-fragment number" (e.g., "PROD-2024-001-MOD-003-SL-002", representing the second fragment of product 001, level 3, and module 003 of product 2024). Upstream / downstream computation dependency identifiers are associated through "preceding fragment unique identifier-subsequent fragment unique identifier-dependency type", which clarifies that the computation result of the preceding fragment is a necessary input for the subsequent fragment to perform the computation, thus avoiding dependency conflicts.

[0032] The detailed process for obtaining the module function importance score is shown below: The functional criticality dimension has a weight of 0.5. The score for this dimension is solely determined by the hierarchical role of modules in the BOM (Bill of Materials) as preset in the PLM system and not manually modifiable. Core-level modules (essential functional carriers for the entire machine's operation, such as the main control module and power supply module) receive 100 points; important-level modules (critical auxiliary modules supporting core functions, such as communication modules and interface modules) receive 80 points; auxiliary-level modules (providing only additional functions, such as heat dissipation modules and decorative parts) receive 50 points; and redundant-level modules (which can be omitted or replaced by other compatible modules, such as redundant interfaces and spare decorative parts) receive 20 points. The fault impact scope dimension has a weight of 0.3. The score for this dimension is determined by the fault impact level as defined in GJB / Z9001A or industry-standard fault level classifications. A fatal fault level (Level I, where module failure directly renders the entire machine completely inoperable with no emergency alternative) receives 100 points; a severe fault level receives 20 points. Fault Level II (module failure causing core function failure, the whole machine can only run non-core auxiliary functions) scores 80 points; General Fault Level III (module failure only affects its own function and does not affect other modules, the whole machine's core functions are normal) scores 50 points; Minor Fault Level IV (module failure does not affect the operation of any function and can be restored with only routine maintenance) scores 20 points. The cost ratio dimension has a weight of 0.2. The cost ratio dimension score is an absolute objective indicator based on the module's full life cycle cost (purchase cost and 5-year maintenance cost) directly output by the financial system and accurate to two decimal places. ≥20% of the total product cost scores 100 points, 10%-19% scores 80 points, 5%-9% scores 50 points, and <5% scores 20 points. The total score for module function importance is calculated as follows: Total score for module function importance = Functional criticality dimension score × 0.5 + Fault impact scope dimension score × 0.3 + Cost ratio dimension score × 0.2.

[0033] The deployment of MTBF compute nodes follows a dynamic adaptation logic: When the number of shards is ≤3, deploy 1 primary compute node and 1 standby node; when the number of shards is 3 < ≤6, deploy 2 primary compute nodes and 1 standby node; when the number of shards is >6, determine the number of primary nodes by rounding up from "number of shards / 3", and the number of standby nodes is fixed at 2. Compute nodes must meet the following hardware configuration requirements: CPU frequency ≥3.0GHz, memory ≥16GB, hard disk storage space ≥500GB, support parallel computing frameworks (such as Spark, Hadoop), and can be flexibly deployed on local physical servers or cloud servers such as Alibaba Cloud and Huawei Cloud. The deployment mode can be manually switched through the PLM system or automatically selected according to the number of shards. The load balancing algorithm dynamically adapts to the characteristics of each shard: when computational load is balanced (the difference in the number of components in each shard is ≤20%), a round-robin method is used, allocating shards to master nodes in order of priority; when computational load is unbalanced (there are shards with a difference in the number of components >30%), a minimum connection method is used, allocating shards with high computational load to the node with the fewest current connections; for computationally intensive shards (containing complex reliability model parameters), a CPU load priority method is used, monitoring the CPU utilization of each node in real time and allocating shards to nodes with a utilization of <60%. A node failover mechanism is also configured, with the PLM system monitoring the running status of each node in real time (checking connection status and computation progress every 10 seconds). When a master node experiences a disconnection or computation timeout (a single computation exceeding 30 minutes), the shards it is responsible for are automatically migrated to a backup node. During the migration process, completed computation progress is preserved to avoid duplicate computations and ensure uninterrupted parallel computing.

[0034] During the process of PLM synchronizing real-world operating data throughout the product lifecycle to the MTBF computing node, data acquisition covers three core stages: R&D, production, and field operation. During the R&D and testing phase, industrial sensors supporting RS485 / Ethernet interfaces were used to collect test temperatures (measurement range -40℃). 125℃, accuracy ±0.5℃), relative humidity (measurement range 0%) 100%RH, accuracy ±3%RH), operating voltage (measurement range 0V) 220V, accuracy ±0.1V), with a sampling frequency of once per second, data is transmitted to the PLM system in real time; during the production stage, the data acquisition terminal of the Manufacturing Execution System (MES) collects production load rate (accuracy ±1%), equipment start-up and shutdown frequency (unit: times / hour), and processing parameters (such as welding temperature, assembly torque, etc., with accuracy configured according to process type), with a sampling frequency of once per minute, and is synchronized after being summarized by production batch; during the field operation stage, the Internet of Things (IoT) terminal (supporting 4G / 5G / NB-IoT networks) collects ambient temperature, cumulative equipment usage time (unit: hours), and maintenance cycle (unit: days), with a sampling frequency of once every 5 minutes, and supports offline storage (data is temporarily stored in the terminal when offline, and is automatically re-transmitted after reconnection).

[0035] The preprocessing of operating data follows the workflow of "time series normalization - outlier removal - standardization": Time sequence normalization follows a unified format of UTC timestamps "YYYY-MM-DDHH:MM:SS.SSS". Data with missing timestamps is supplemented according to the collection order, and the latest record is retained for data with duplicate timestamps. Outlier removal uses a combination of two algorithms. First, global outliers are screened using the 3σ criterion. The mean μ and standard deviation σ of single-type operating condition data are calculated, and extreme data exceeding the range [μ-3σ, μ+3σ] are removed. Then, local outliers are identified using the Local Outlier Factor (LOF) algorithm. The neighborhood radius is set to 5 data points and the minimum sample size is 10. When the LOF value of a data point is greater than 1.5, it is determined to be a local outlier and removed. After preprocessing, a standardized operating condition dataset is generated. The data format is uniformly JSON, containing fields such as "operating condition type, collection timestamp, value, data source, and accuracy level".

[0036] Data synchronization uses a message queue mechanism to achieve real-time transmission, supporting both MQTT and JMS protocols for selection as needed. When using the MQTT protocol, configure the QoS (Quality of Service) level to 2 (ensuring messages are delivered only once), set the message retention time to 7 days, and the topic format to "PLM / Operating Condition Data / Product Code / Module Code". The subscriber (MTBF compute node) subscribes to the required data precisely according to the topic to avoid irrelevant data transmission. When using the JMS protocol, use the point-to-point (P2P) queue mode, name the queue according to "Product Code-Operating Condition Type", configure a 30-second message retry mechanism (maximum of 3 retries), and if the retry fails, record the exception log and push alarm information to the PLM system to ensure data transmission reliability.

[0037] After receiving the standardized operating condition dataset, each MTBF computing node extracts the component information corresponding to the current segment (by associating the component code in the BOM with the segment's unique identifier). Based on a three-dimensional mapping table of "operating condition type - component type - numerical range," it queries and matches the corresponding industry standard correction factor. The correction factor is based on authoritative reliability standards such as GJB / Z299C and MIL-HDBK-217F. For example, under temperature conditions, the correction factor for a common resistor is 1.0 at 25℃, 1.8 at 55℃, and 2.5 at 85℃; under load rate conditions, the correction factor is 1.0 when the motor load rate is ≤50%, and 1.0 when the load rate is ≤50%. The value is 1.3 when the reliability is 80% and 1.7 when it is greater than 80%. Parameter adjustment is performed using the preset formula "Updated reliability parameter = Original reliability parameter × Industry standard correction factor". The original reliability parameter is extracted from the ready-to-use BOM (e.g., the component base failure rate is 0.002). The adjusted parameters overwrite the original parameters in real time and are used for subsequent reliability calculations to ensure that the calculated parameters accurately match the actual operating conditions.

[0038] The specific calculation process for each segment reliability metric (with MTBF value as the core) is as follows: Each segment reliability index specifically refers to the segment core MTBF value (unit: hours, calculation result rounded to 2 decimal places), which is a specific numerical indicator used for weighted calculation of overall system reliability. The calculation process is as follows: First, the MTBF compute node parses the BOM structure of the current segment. Through component assembly relationship identifiers and functional attributes, it clarifies the connection topology of all components within the segment, accurately determining whether it is a series structure, parallel structure, or hybrid structure. A series structure means that all components are necessary units for the function to be realized, and the segment can only maintain its core function if all components work normally at the same time. A parallel structure means that the components are redundantly designed or functionally complementary, and the segment can be guaranteed not to fail if at least one component works normally. A hybrid structure means that there is a combination of local series and local parallel connections within the segment, which is handled according to the principle of "first splitting the local series / parallel sub-units according to the functional boundaries, and then integrating them according to the connection relationship between the sub-units".

[0039] Secondly, based on the adjusted component failure rate (i.e., the failure rate after operating condition correction, unit: ), calculate the overall failure rate of the fragmentation (unit: The failure rate of a cascaded structure is the arithmetic sum of the adjusted failure rates of all components, i.e., the adjusted failure rate of each component is accumulated (e.g., component 1 has an adjusted failure rate of 0.001). Component 2 is 0.0005 Component 3 is 0.0003 The total fragment failure rate is directly obtained as (0.001 + 0.0005 + 0.0003 = 0.0018). The failure rate calculation for the parallel structure is performed in three steps. The first step is to determine the preset operating time (uniformly set at 1000 hours, a standard industry testing cycle, conforming to the time reference requirements for reliability verification in the GJB / Z299C standard). The second step is to calculate the reliability of each component (Reliability = 1 - Adjusted failure rate × Preset operating time, applicable to low failure rate scenarios where adjusted failure rate × preset operating time ≤ 0.1, such as a component with an adjusted failure rate of 0.0005). The reliability is calculated as 1 - 0.0005 × 1000 = 0.5. The third step is to calculate the product of the failure probabilities of all components (failure probability = 1 - reliability; for example, if the reliability of three parallel components are 0.5, 0.6, and 0.7 respectively, the corresponding failure probabilities are 0.5, 0.4, and 0.3, and the product = 0.5 × 0.4 × 0.3 = 0.06). Then, the component reliability is obtained by using the formula "overall component reliability = 1 - product of all component failure probabilities" (1 - 0.06 = 0.94). Finally, the component failure rate is calculated using the formula "component failure rate = -ln(overall component reliability) / preset working time", where ln is the natural logarithm function. The calculation result is rounded to 6 decimal places (e.g., -ln(0.21) / 1000 ≈ 0.001567). For the segmented failure of the hybrid structure, the failure rate of each local sub-unit is calculated first according to the calculation rules for series and parallel structures (e.g., the failure rate of local series sub-unit 1 is 0.0008). The failure rate of the local parallel subunit 2 is 0.0006. The failure rate of the local cascaded subunit 3 is 0.0004. Then, based on the connection relationship between the sub-units, further integration is performed. If the sub-units are connected in series, the failure rate is accumulated (e.g., if sub-unit 1 and sub-unit 2 are connected in series, the combined failure rate = 0.0008 + 0.0006 = 0.0014). If the sub-units are connected in parallel, the calculation follows the logic of "reliability product - failure rate conversion" (e.g., if the merged sub-unit and sub-unit 3 are connected in parallel, first calculate their reliability: merged sub-unit reliability = 1 - 0.0014 × 1000 = 0.86, sub-unit 3 reliability = 1 - 0.0004 × 1000 = 0.6, then calculate the product = 0.86 × 0.6 = 0.516, finally calculate the failure rate = -ln(0.516) / 1000 ≈ 0.000659). This yields the total fragment failure rate. Finally, the total fragment failure rate is substituted into the core conversion formula "Reliability index of each fragment (MTBF value) = 1 / Total fragment failure rate" to obtain a specific value (e.g., total fragment failure rate 0.0018). MTBF value = 1 / 0.0018 ≈ 555.56 hours), and simultaneously calculate auxiliary numerical indicators such as the failure probability of the segment (failure probability = 1 - (1 - total failure rate of the segment × preset working time), such as 1 - (1 - 0.0018 × 1000) = 0.82), mean repair time (take the average value of failure repair data of similar products in the industry, such as 2.5 hours), to form a complete set of segment reliability numerical indicators.

[0040] The dynamic reliability calculation results are obtained through a closed-loop process of "sharded calculation - result upload - fusion verification - anomaly handling": After each MTBF computing node completes the fragment calculation, it packages and uploads the calculation results (including fragment reliability indicators, component-level reliability data, and calculation process logs), fragment unique identifiers, and calculation completion timestamps (accurate to milliseconds) to the PLM Collaboration Hub. The upload uses HTTPS protocol for encrypted transmission to prevent data leakage. As the core processing module, the PLM Collaboration Hub has a built-in fragment association mapping library (storing the correspondence between fragments and the whole machine and modules) and a calculation log storage unit (retaining calculation data for nearly one year). After receiving the fragment results, it first performs data alignment, mapping the calculation results of each fragment to the corresponding module position in the whole machine according to the fragment hierarchical encoding and parent module encoding, ensuring that the results are consistent with the product structure.

[0041] The results were fused using a weighted summation algorithm, with the fusion weights determined based on the functional importance of the modules: the weight of the slice corresponding to the first-priority module was set to 0.2. 0.3, the shard weight corresponding to the secondary priority module is set to 0.1. The shard weight corresponding to the level 3 priority module is set to 0.05. 0.08, the sum of the weights of all segments is 1.0, the weight values ​​are determined by the analytic hierarchy process, and the preliminary reliability result of the whole machine is calculated by the formula "preliminary reliability of the whole machine = Σ (reliability index of each segment × corresponding weight)".

[0042] Consistency checks should set dual thresholds: First, there's the deviation threshold between the sharding results and historical data. This threshold is set to 1.5 times the standard deviation of the same sharding calculation results from the last three similar products (default is no more than ±5%), used to verify the stability of the current calculation. Second, there's the deviation threshold between different nodes of the same source shard. If the same shard is calculated through multiple nodes (e.g., repeated calculations after failover), the threshold is set to ±3%, used to verify the consistency of node calculations. During verification, the deviation between the sharding results and historical data is first compared, then the node deviations of the same source shards are compared. Any deviation exceeding the corresponding threshold is considered an abnormal result.

[0043] Exception handling follows a hierarchical mechanism of "automatic retry - manual intervention": For segments that exceed limits, the PLM coordination hub automatically triggers recalculation, prioritizing allocation to backup nodes, with a maximum of two recalculations. If the recalculation result still exceeds limits, the system pushes detailed information about the abnormal segment (including calculated data, deviation values, and historical comparison data) to the PLM system's alarm module and freezes the corresponding BOM node. The cause of the anomaly (e.g., incorrect BOM data, abnormal operating condition data, incorrect calculation model parameters) must be manually verified and corrected before recalculation can be initiated. After all segment results pass consistency verification, the PLM coordination hub integrates the initial overall reliability results with detailed data from each module and segment, outputting the final dynamic reliability calculation results. The result format includes the overall MTBF value, reliability indicators for each module, reliability data for key components, and a calculation verification report, providing complete data support for subsequent fault prediction.

[0044] This step, through a refined BOM sharding strategy and dynamic node deployment, improves the reliability prediction efficiency of complex products by more than 50%, and shortens the time by more than 60% compared to traditional single-node calculations. The accurate collection of full lifecycle operating condition data and dynamic parameter adjustment solve the problem of the disconnect between traditional standardized parameters and actual operating scenarios, reducing the deviation between reliability calculation results and actual conditions to within ±5%. The hierarchical integration and dual consistency verification of the PLM collaboration hub ensure the accuracy and stability of the overall calculation results, laying a high-quality data foundation for subsequent fault identification and BOM optimization.

[0045] Specifically, the process of identifying potential failure points in the BOM and quantifying the risks is as follows: Load the dynamic reliability calculation results and combine them with PLM historical fault data and industry fault database to construct a fault feature vector library; The decision tree algorithm is used to perform feature matching on outlier parameters and unreliable items in the calculation results to identify potential failure points; A risk quantification model is constructed from two dimensions: the probability of failure and the degree of failure impact. The risk value of each failure point is calculated and the risk level is classified. Generate a fault risk analysis report that includes the location of the fault point, characteristic description, and risk level.

[0046] Among them, the fault feature vector library is a dataset that transforms fault features (such as component parameter drift range and module reliability failure threshold) into vector form, covering typical features of component-level faults (such as capacitor aging and resistor failure) and module-level faults (such as unstable power supply of power module and signal interruption of communication module); PLM historical fault data includes fault cases and maintenance records recorded in past projects, and the industry fault database is statistical data on similar product faults released by authoritative third-party institutions to ensure the comprehensiveness of fault features.

[0047] In detail, the specific implementation process of the decision tree algorithm is as follows: First, data preprocessing is performed, extracting key feature parameters from the dynamic reliability calculation results, including the failure rate after component adjustment, the deviation of the segment MTBF value from the industry standard value, the duration of parameter anomalies, and the fit of operating conditions (the degree of fit between actual operating conditions and component rated operating conditions). At the same time, typical fault labels (such as "component-level aging fault", "component-level short-circuit fault", "module-level power supply fault", "module-level communication fault", etc.) are extracted from the fault feature vector library. The feature parameters are matched with the fault labels to construct the algorithm training dataset. The dataset is divided into training and test sets in an 8:2 ratio to ensure balanced data distribution. Second, feature selection is performed. All feature parameters are sorted based on the information gain ratio criterion, and features that contribute highly to fault identification are prioritized. The selection rule is that features with an information gain ratio ≥ 0.3 are included in the modeling (such as "failure rate deviation" and "abnormal duration" with information gain ratios usually higher than 0.5, which are prioritized for retention; features with an information gain ratio lower than 0.2 for "operating condition fit" are removed). Finally, 3-5 core features are determined for decision tree construction. Next, a binary decision tree model is constructed. The core features in the training set are used as the basis for splitting. The root node selects the feature with the highest information gain ratio (such as "failure rate deviation amplitude > 30%)". The internal nodes are split into binary categories according to feature thresholds (such as "abnormal duration ≥ 5 minutes" and "MTBF deviation amplitude ≤ -20%)). The information gain ratio of the corresponding feature is calculated for each split node to ensure that the purity of the fault label of the child node is maximized after splitting. The leaf nodes are clearly labeled with the fault type or "no fault". For example, when a data satisfies "failure rate deviation amplitude > 30%" and "abnormal duration ≥ 5 minutes", the corresponding leaf node is labeled "component-level aging fault". When the data satisfies "MTBF deviation amplitude within ±5%" and "no parameter abnormal continuous record", the leaf node is labeled "no fault". Then, pruning optimization was carried out, using a combination of pre-pruning and post-pruning: in the pre-pruning stage, the maximum depth of the decision tree was set to 5 layers and the minimum number of samples per leaf node was 10 to avoid model overfitting; in the post-pruning stage, the model accuracy was verified through the test set, and branches that did not contribute significantly to the classification results were deleted (such as the fault identification accuracy of a certain branch being <60%), ultimately ensuring that the model's test set accuracy was ≥92%.Finally, fault determination is performed. The BOM data to be identified (including reliability parameters of components and modules, and anomaly records) is input into the trained and optimized decision tree model. The model is traversed layer by layer according to the node splitting rules. The complete path from the root node to the leaf node is the fault determination result. The matching confidence calculation and determination rules for the leaf node fault type are as follows: Each node is split based on core fault features (such as failure rate deviation, parameter drift amplitude, etc.) using a hard threshold. The splitting threshold is automatically determined by minimizing the information gain ratio or Gini impurity during the training phase. If the data to be identified meets the threshold condition, it traverses along the left branch; otherwise, it traverses along the right branch. If there are no second-level internal nodes, it directly traverses from the first-level internal nodes to the leaf node, ensuring that each data point corresponds to only one unique decision path. Matching confidence = root node feature importance weight × node conformity + first-level internal node feature importance weight × node conformity + second-level internal node feature importance weight × node conformity + leaf node category probability × 0.1, where... The node conformity is a binary judgment result (1.0 if the split threshold is met, 0 if not, and the feature importance weight of the level is set to 0 when there is no corresponding node). The feature importance weight is automatically generated based on the amount of impurity reduction (MDI index) when the node splits. The root node usually has the highest weight due to its greatest contribution to the split. The weight of internal nodes decreases with each level. The sum of the weights of all levels is normalized to 0.9. The leaf node category probability is the proportion of the corresponding fault type in the samples falling into that leaf node in the training set (statistically calculated by majority voting). The category probability of a leaf node without a fault label is 0. All parameters are standardized in the range of 0-1, and the final matching confidence (range 0-1) is obtained after weighted summation. If a leaf node is labeled with a specific fault type and the matching confidence is ≥85%, it is determined to be a potential fault point of the corresponding level (e.g., when traversing to the "module-level power supply fault" leaf node, it is determined to be a module-level potential fault point). If a leaf node is labeled "no fault" or the matching confidence is <85%, it is determined to be a potential fault point.

[0048] Specifically, the calculation logic of the risk quantification model is "Risk Value = Failure Occurrence Probability × Failure Impact". Details are as follows: The failure occurrence probability is derived from statistical analysis of PLM historical failure data over the past 5 years. The statistical scope is clearly defined as "natural failures caused by non-human factors," excluding failures caused by misoperation, external force damage, or other reasons not related to the product itself. The calculation formula is: Failure Occurrence Probability (Annual Probability) ≈ Failure Rate ( ) × annual standard operating hours (8760 hours), of which the failure rate ( The core calculation indicator is the failure rate (FRR). The formula is: "Number of natural failures of similar components / modules in the past 5 years ÷ Total runtime of similar components / modules in the past 5 years". This formula conforms to the definition of failure rate in reliability engineering (unit: 1 / h). Both sides of the equation have consistent dimensions (the number of failures is a dimensionless count, the total runtime is in hours, and the result has a dimension of 1 / h). The total runtime is defined as the cumulative operating hours of the component / module across all similar products (i.e., total quantity × average runtime of a single component). There is no need to introduce an additional average operating density (the original indicator leads to dimensional redundancy and confuses the definitions of failure rate and failure probability). The annual probability uses an engineering approximation algorithm (applicable to...). (For low failure rate scenarios where ×8760≤0.1), based on the exponential distribution characteristic, when the failure rate is low, the probability of failure ≈ failure rate × time, simplifying calculations while ensuring accuracy. Example: A certain type of capacitor has been used in 10 projects over the past 5 years, with a total quantity of 500 units and a cumulative runtime of 5 × 8760. The average number of natural failures per hour (i.e., each capacitor operates for an average of 10,000 hours) is 8; the failure rate is ( ) = 8 ÷ 5 × =1.6× Annual failure probability = 1.6 × ×8760≈0.014016 (i.e., 1.4016%); the impact of the fault is calculated based on the module's functional importance score, using the formula "Fault Impact = Total Module Functional Importance Score ÷ 100", ensuring consistency with the module priority assessment logic above. For example, if the core module (main control module) has a total functional importance score of 90, its fault impact = 90 ÷ 100 = 0.9; the important module (communication module) has a score of 75, with an impact of 0.75; the auxiliary module (heat dissipation module) has a score of 50, with an impact of 0.5; and the redundant module (backup interface) has a score of 20, with an impact of 0.2. The final risk value is the product of the fault occurrence probability (annual probability) and the fault impact, rounded to 6 decimal places. The risk levels are divided into: The risk levels are categorized into high, medium, and low, with thresholds set based on industry risk management standards and product reliability requirements: a risk value ≥ 0.008 (i.e., 0.8%) is considered high risk, as such failures may cause the core functions of the entire machine to fail, requiring immediate optimization; a risk value ≤ 0.003 (0.3%) ≤ risk value < 0.008 (0.8%) is considered medium risk, requiring inclusion in the next round of optimization plans and rectification within one month; a risk value < 0.003 (0.3%) is considered low risk, requiring no immediate optimization but continuous monitoring of its operational status, with risk values ​​updated quarterly. For example, if the probability of a core module failure is 1.4016% and the impact level is 0.9, the risk value = 0.014016 × 0.9 ≈ 0.012614, which is considered high risk and requires priority optimization.

[0049] Specifically, the process of obtaining the optimized reliability calculation results is as follows: The failure risk analysis report is pushed to the PLM system, and the PLM system associates the corresponding computationally ready BOM node with the component reliability knowledge base. Based on the evaluation criteria, a multi-dimensional set of optimization solutions is generated, including component replacement, BOM structure adjustment, and parameter threshold correction. The optimal solution is selected through cost and reliability balance analysis, the feasibility of the solution is verified, and the final optimized solution is generated after confirming that there are no supply chain bottlenecks or structural conflicts. The corresponding fields of the compute-ready BOM are automatically updated based on the final optimization scheme, BOM version update logs are generated synchronously, and the updated compute-ready BOM is pushed to the MTBF compute node. MTBF compute nodes load the updated compute-ready BOM, reuse actual operating data and parameter adjustment rules throughout the product lifecycle, and perform secondary reliability calculations. The validity of the secondary calculation results is verified. After confirming that the results meet the reliability improvement expectations and there are no data anomalies, the optimized reliability calculation results are output.

[0050] The evaluation criteria are "reliability improvement priority > supply chain feasibility priority > cost control priority", that is, the solution with the greatest reliability improvement is selected first, then the supply chain of the components required for the solution is ensured to be stable (no risk of shortage), and finally the optimization cost is controlled within the budget. The optimization solution set specifically includes: component replacement (such as replacing low reliability resistors with high reliability models), BOM structure adjustment (such as adding redundant design to core modules), and parameter threshold correction (such as adjusting the operating voltage threshold of components to reduce the probability of failure).

[0051] In detail, cost and reliability balance analysis refers to calculating the ratio of "reliability improvement / optimization cost" for each optimization scheme and selecting the scheme with the largest ratio as the optimal scheme; implementation feasibility verification includes supply chain verification (checking supplier inventory and delivery cycle) and structural conflict verification (confirming that the optimized BOM structure has no logical conflicts or installation interference); BOM version update log records parameter changes before and after optimization, optimization basis and operators to ensure data traceability.

[0052] Specifically, the secondary reliability calculation reuses the actual operating data and parameter adjustment rules of the entire life cycle, without the need for re-collection and setting, thus improving calculation efficiency. The standard for validity verification is: the reliability index after optimization is ≥10% higher than that before optimization (meeting the expected improvement), and the calculation results have no abnormal parameters, deviations, or other issues. If the verification passes, the optimized reliability calculation results are output; if it fails, the re-optimization scheme is returned.

[0053] This step identifies potential fault points in advance, avoiding rework costs that occur when faults are exposed during product testing or operation. By generating targeted optimization solutions through the PLM system and performing secondary calculations, a closed loop of "calculation-fault identification-optimization-recalculation" is formed, significantly improving product reliability.

[0054] Specifically, the process for obtaining a standardized reliability prediction report is as follows: Extract three types of value-added content from the PLM system: fault optimization and source tracing records, reliability benchmarking data of similar products, and endorsement of historical project verification results. The extracted value-added content is categorized, integrated, and logically organized, and arranged according to the chapter structure required by the tendering party. A standardized report template was used to unify the font and format, and the value-added content was organically integrated with the optimized reliability calculation results to generate a draft report; Embedded one-click traceability module, linking key data in the report with the PLM system's raw data, calculation logs, and optimization records; The initial draft report undergoes compliance verification, and a standardized reliability prediction report is generated after the verification is passed.

[0055] Among them, the value-added content is the core competitive element in the bidding review: the fault optimization and tracing record includes the original fault information, optimization measures, reliability improvement, etc., to prove the product reliability can be optimized; the reliability benchmarking data of similar products includes the industry average, the best value of competitors and the advantages of this product, highlighting the product competitiveness; the endorsement of historical project verification results refers to the actual operational reliability data of the same type of components or similar products in past projects, enhancing the credibility of the report.

[0056] In detail, the chapter structure required by the tendering party typically includes an abstract, BOM data preprocessing instructions, reliability calculation process, fault optimization instructions, benchmarking analysis, and commitment statement. The intelligent enhancement system arranges the value-added content and calculation results according to this structure to ensure clear logic and highlight key points. The standardized report template uses a unified font (such as SimSun, size 4), page margins (such as 2.5cm top and bottom, 3cm left and right), and data visualization format (such as displaying reliability indicators using bar charts and comparing before and after optimization using line charts), which meets the format requirements of the tendering party.

[0057] Specifically, the one-click traceability function module is implemented through the association mapping of "report key data - PLM data source". For example, clicking on "optimized reliability index" in the report can directly jump to the corresponding secondary calculation log and optimization plan document in the PLM system. The associated data sources include BOM raw data, operating condition collection records, calculation process logs, fault analysis reports, optimization plan documents, etc., to ensure that the authenticity of the data can be quickly verified in the bidding review.

[0058] In detail, compliance verification includes format verification and data integrity verification: format verification verifies whether the font, structure, chart style, etc. of the report meet the bidding requirements; data integrity verification verifies whether the report contains all required fields (such as calculation parameters, optimization measures, benchmarking data, etc.) without any missing or omitted information; if the verification fails, the intelligent enhancement system will push specific rectification items (such as "missing benchmarking data for similar products" or "chart format does not meet requirements"), and the report will be regenerated after rectification until the verification passes.

[0059] The processes described above with reference to the flowcharts in the embodiments disclosed in this invention can be implemented as computer software programs. The embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wire segments, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical fibers, RF, etc., or any suitable combination thereof.

[0060] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0061] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and explained in the embodiments. Without departing from the stated principles, the implementation of the present invention may have any variations or modifications.

Claims

1. A PLM-based BOM data management method, characterized in that, The method includes: Obtain the product's BOM basic data, combine it with the PLM's built-in component reliability knowledge base to complete the missing parameters required for MTBF calculation, and after format regularization and integrity verification, obtain the calculation-ready BOM. Functional segments are decomposed based on the hierarchical structure of the computation-ready BOM, and MTBF computing nodes are deployed. Segmented calculations are performed based on parallel function calls. PLM is used to synchronize the actual operating data of the entire product lifecycle to the MTBF computing nodes, dynamically adjust the reliability parameters of components, and obtain dynamic reliability calculation results by integrating the calculation results through the PLM collaborative hub and verifying consistency. The dynamic reliability calculation results are analyzed by decision tree algorithm, potential failure points in BOM are identified and risks are quantified, and failure and risk information is pushed to PLM system. PLM combines component reliability knowledge base to generate optimization scheme, update calculation ready BOM and synchronize to MTBF calculation node for secondary calculation to obtain optimized reliability calculation results. Based on the optimized reliability calculation results, key value-added content is integrated through the intelligent enhancement system, and the report format is standardized according to the bidding requirements to generate a standardized reliability prediction report.

2. The PLM-based BOM data management method according to claim 1, characterized in that, The specific process for obtaining the computation-ready BOM is as follows: By connecting to the product design module through the standardized interface of the PLM system, basic BOM data is collected to form the original BOM dataset. The PLM's built-in component reliability knowledge base is invoked. Based on the component identifiers in the original BOM dataset, parameter matching is performed to extract the core parameters required for MTBF calculation. For missing parameters that are not matched, they are completed according to the parameter completion rules for similar components to obtain a complete parameter BOM. According to the parameter format requirements preset by MTBF, the fields of the complete parameter BOM are standardized and organized to unify the parameter units and data types; The system sequentially performs checks for missing core parameters in MTBF calculation, compliance of parameter format, and rationality of BOM hierarchical structure, removing invalid data and abnormal levels. Once the checks pass, a calculation-ready BOM is generated.

3. The PLM-based BOM data management method according to claim 2, characterized in that, The specific process of splitting the computation-ready BOM into functional fragments according to the hierarchical structure and deploying MTBF compute nodes to call the functional fragments for parallel computation is as follows: Analyze the hierarchical structure of the computation-ready BOM, identify the functional boundaries and independent computing units of each module, and determine the granularity of the partitioning. Prioritize the segments based on the importance of their functionalities; Each fragment is labeled with a unique identifier and an upstream and downstream computation dependency identifier. The computation result of the preceding fragment of a fragment with a dependency relationship becomes the input of the subsequent fragment. Based on the number of shards and computational complexity, corresponding MTBF computing nodes are dynamically deployed. Each shard is allocated to an idle computing node through a load balancing algorithm. The MTBF computing nodes call the shard data in priority order to perform parallel computation.

4. The PLM-based BOM data management method according to claim 3, characterized in that, The specific process of PLM synchronizing actual operating data throughout the product's entire lifecycle to the MTBF computing node to dynamically adjust component reliability parameters is as follows: By connecting the PLM system to R&D testing equipment, production monitoring system and on-site operation data acquisition terminal, the actual working condition data of the entire life cycle is collected. The operating condition data is time-series normalized and outlier removed at preset time intervals to generate a standardized operating condition dataset, which is then synchronized to each MTBF computing node in real time through a message queue mechanism. Extract component information corresponding to each slice using each MTBF computing node, and query and match the corresponding industry standard correction factor based on operating condition data. The calculation parameters are dynamically updated based on the correction factor and the original reliability parameters of the components using a preset parameter adjustment formula.

5. A PLM-based BOM data management method according to claim 4, characterized in that, The specific process for obtaining the dynamic reliability calculation results is as follows: After completing the sharding calculation using each MTBF computing node, the calculation results, along with the sharding identifier and calculation timestamp, are uploaded to the PLM collaboration hub. The calculation results are aligned according to the segmentation hierarchy through the PLM collaboration hub, the fusion weights are assigned based on the importance of module functions, and the preliminary reliability results of the whole machine are calculated through a weighted summation algorithm. Set a consistency check threshold, compare the deviation of each shard calculation result with the historical calculation data of the same type of shard, and identify the common deviation of the calculation results of different nodes to identify abnormal results that exceed the limit; For the fragments corresponding to abnormal results, recalculation is triggered, and the corrected fragment results are merged again until all results meet the consistency requirements, and the dynamic reliability calculation results are output.

6. The PLM-based BOM data management method according to claim 5, characterized in that, The specific process for identifying potential failure points in the BOM and quantifying the risks is as follows: Load the dynamic reliability calculation results and combine them with PLM historical fault data and industry fault database to construct a fault feature vector library; The decision tree algorithm is used to perform feature matching on outlier parameters and unreliable items in the calculation results to identify potential failure points; A risk quantification model is constructed from two dimensions: the probability of failure and the degree of failure impact. The risk value of each failure point is calculated and the risk level is classified. Generate a fault risk analysis report that includes the location of the fault point, characteristic description, and risk level.

7. A PLM-based BOM data management method according to claim 6, characterized in that, The specific process for obtaining the optimized reliability calculation results is as follows: Push the failure risk analysis report to the PLM system and use PLM to associate the corresponding computationally ready BOM node with the component reliability knowledge base. Based on the evaluation criteria, a multi-dimensional set of optimization solutions is generated, including component replacement, BOM structure adjustment, and parameter threshold correction. The optimal solution is selected through cost and reliability balance analysis, the feasibility of the solution is verified, and the final optimized solution is generated after confirming that there are no supply chain bottlenecks or structural conflicts. The corresponding fields of the compute-ready BOM are automatically updated based on the final optimization scheme, BOM version update logs are generated synchronously, and the updated compute-ready BOM is pushed to the MTBF compute node. MTBF compute nodes load the updated compute-ready BOM, reuse actual operating data and parameter adjustment rules throughout the product lifecycle, and perform secondary reliability calculations. The validity of the secondary calculation results is verified. After confirming that the results meet the reliability improvement expectations and there are no data anomalies, the optimized reliability calculation results are output.

8. A PLM-based BOM data management method according to claim 7, characterized in that, The specific process for obtaining the standardized reliability prediction report is as follows: Extract three types of value-added content from the PLM system: fault optimization and source tracing records, reliability benchmarking data of similar products, and endorsement of historical project verification results. The extracted value-added content is categorized, integrated, and logically organized, and arranged according to the chapter structure required by the tendering party. A standardized report template was used to unify the font and format, and the value-added content was organically integrated with the optimized reliability calculation results to generate a draft report; Embedded one-click traceability module, linking key data in the report with the PLM system's raw data, calculation logs, and optimization records; The initial draft report undergoes compliance verification, and a standardized reliability prediction report is generated after the verification is passed.

9. A PLM-based BOM data management system, characterized in that, The system is used to execute a PLM-based BOM data management method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement a PLM-based BOM data management method as described in any one of claims 1-8.

Images