A method for quality traceability throughout the entire production lifecycle of a DC brushless fan.

Abstract

This invention relates to a quality traceability method for the entire lifecycle of DC brushless fan production. The method includes constructing an operating condition spectrum library and establishing operating condition design anchor points; generating product benchmark files and writing them into a quality database; establishing a quality traceability chain and performing three-level association mapping; collecting manufacturing parameters and uploading them with dual tags combining process node numbers and operator employee numbers; constructing a health status index by combining the fan's real-time speed fluctuation rate and phase current harmonic distortion rate; achieving source location of quality anomalies through deviation analysis; and collecting failure component detection data and scrapping reasons, feeding them back to the operating condition spectrum library for dynamic revision and updating. This invention, by constructing a full lifecycle quality traceability system covering the entire process of design, manufacturing, operation, and recycling, achieves precise location and closed-loop optimization of quality problems in DC brushless fans, improving product quality management, fault diagnosis accuracy, and long-term operational capability.

Description

Technical Field

[0001] This invention relates to a method for quality traceability throughout the entire production lifecycle of a DC brushless fan. Background Technology

[0002] With the development of intelligent manufacturing and industrial internet technologies, product quality traceability systems have been widely applied in the manufacturing industry. In the production traceability of motor products, existing technologies generate sub-traceability codes for components by acquiring component information, coils by acquiring coil information, rotors by acquiring rotor information, and stators by acquiring stator information, ultimately generating a finished motor traceability code. This information is then stored in a cloud server, enabling information association and traceability of key components during the production process. Furthermore, technologies have disclosed full lifecycle forward and reverse quality traceability for discrete manufacturing products. By configuring traceability code rules and multi-dimensional traceability conditions, production barcodes are generated and bound as unique traceability identifiers for the product at each production stage, providing one-click traceability functionality.

[0003] However, existing technologies generally suffer from the following shortcomings in practical applications: They are mostly limited to data traceability in a single or limited stage, such as focusing only on production process traceability or usage phase monitoring, lacking a complete data chain integration across the entire lifecycle from product design, raw material procurement, manufacturing, factory testing, actual use to final disposal and recycling. This fragmented traceability approach makes it difficult to comprehensively trace the root causes of quality problems. Furthermore, they do not adequately consider the technical characteristics of specific product types, especially for products like brushless DC fans with unique technical parameters (such as PWM control characteristics, Hall sensor configuration, speed-torque characteristics, etc.). Existing technologies have failed to design dedicated traceability dimensions and analysis methods for these unique parameters, resulting in insufficient relevance and practicality of traceability data. When product quality anomalies occur, only historical data queries are available; they cannot provide root cause diagnosis of quality problems, fault development trend prediction, or targeted quality improvement suggestions based on multi-source data correlation analysis.

[0004] In summary, existing technologies generally suffer from defects such as fragmented traceability, lack of product-specificity, passive data recording, insufficient coupling between modules, and lack of closed-loop feedback mechanisms. This makes it impossible to achieve organic integration, intelligent analysis, and dynamic traceability of quality data at all stages of the entire life cycle from design to disposal, and makes it difficult to solve practical problems in quality definition, accurate fault prediction, and continuous quality improvement of DC brushless fans.

[0005] In view of the above-mentioned shortcomings, the present invention aims to create a quality traceability method for the entire life cycle of DC brushless fan production, so as to make it more valuable for industrial use. Summary of the Invention

[0006] To address the aforementioned technical problems, the purpose of this invention is to provide a method for quality traceability throughout the entire production lifecycle of a DC brushless fan.

[0007] The present invention provides a quality traceability method for the entire production lifecycle of a DC brushless fan, comprising constructing an operating condition spectrum library based on historical heat load, temperature and humidity fluctuations and heat dissipation interference data of server racks, and binding it with the PWM control parameters and bearing clearance characteristics of the fan to form operating condition design anchor points; collecting the fan's design parameters from the operating condition design anchor points, generating a product benchmark file containing design feature vectors and storing it in a quality database;

[0008] A quality traceability chain is built by scanning the incoming barcode in the quality database. Batch numbers, component test values ​​and performance test values ​​are extracted from the product benchmark files and written into the quality traceability chain through a three-level association mapping. Data including stator winding coil resistance value, rotor magnetization waveform and magnetic flux uniformity, assembly Hall sensor angle deviation and air gap coaxiality are collected and re-uploaded to the quality database after being doubly marked with process node number and operator number. The health status index is calculated by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine. The current health status index is compared and analyzed with the factory test benchmark and design parameter benchmark to locate the source node of quality deviation. When the wind turbine is scrapped and recycled, the scrapping cause and component test data of the source node are collected and fed back to the operating condition spectrum library.

[0009] Furthermore, the specific implementation process of constructing an operating condition spectrum library based on the historical thermal load, temperature and humidity fluctuations, and heat dissipation interference data of the server rack, and binding it with the PWM control parameters of the fan and the bearing clearance characteristics to form operating condition design anchor points includes: extracting historical thermal load time-series data from the operation and maintenance database of the target server rack; collecting temperature and humidity fluctuation sampling sequences output by the temperature and humidity sensors inside the rack and records of heat dissipation interference events caused by air duct obstruction and airflow short circuits; after normalization processing, classifying and storing the data in the operating condition spectrum library according to rack model and installation location; extracting the PWM duty cycle adjustment range, response frequency boundary, and target speed mapping relationship into a PWM control parameter set according to the fan speed control protocol; extracting the radial clearance tolerance zone and axial clearance tolerance zone as bearing clearance characteristics according to the bearing preload design document; and performing field-level association binding between the PWM control parameter set and bearing clearance characteristics, indexed by the rack thermal environment category code, and the corresponding operating condition records in the operating condition spectrum library to generate operating condition design anchor points.

[0010] Furthermore, the specific implementation process of collecting wind turbine design parameters from operating condition design anchor points, generating product benchmark files containing design feature vectors, and storing them in the quality database includes: using the index code of the operating condition design anchor points as the retrieval key, reading the PWM control parameter set, bearing clearance characteristics, and thermal environment classification information bound to the anchor points from the quality database; collecting design parameters such as rated speed, locked rotor torque, insulation class, stator slot type parameters, and magnetic circuit air gap target value according to the wind turbine design specification documents; quantizing and encoding by data type segments, and concatenating them into design feature vectors; using the product model code and design version number as the joint primary key, encapsulating the design feature vectors into a product benchmark file, recording the file generation timestamp, design supervisor's employee number, and associated operating condition anchor point code, and writing it into the quality database after format verification.

[0011] Furthermore, a quality traceability chain is constructed by scanning incoming materials in the quality database. The specific implementation process of extracting batch numbers, component test values, and performance test values ​​from the product benchmark file and writing them into the quality traceability chain through a three-level association mapping includes: reading the incoming material label using a QR code scanning device to extract the material code and supply batch number, generating an incoming material scanning record and writing it into the quality database as the head node of the quality traceability chain; performing structured parsing on the supplier quality report arriving with the batch to extract the magnetic material grade and magnetic induction intensity test value, copper wire purity and wire diameter tolerance test value, and insulating varnish withstand voltage and dielectric loss performance test value, forming an incoming material quality data package; and writing the values ​​of each field in the incoming material quality data package into the corresponding level nodes of the quality traceability chain according to the three-level association mapping relationship: first-level supplier-batch, second-level batch-material, and third-level material-test item.

[0012] Furthermore, the specific implementation process of collecting data including stator winding coil resistance, rotor magnetization waveform and magnetic flux density uniformity, assembly Hall sensor angular deviation, and air gap coaxiality, and then re-uploading it to the quality database after being doubly tagged with process node number and operator ID, includes: applying measurement excitation to each phase winding, collecting coil resistance values ​​and comparing them with design reference values; scanning the magnetic induction intensity distribution on the rotor surface, extracting the peak sequence of the magnetization waveform and the magnetic flux density uniformity coefficient; collecting the zero-crossing timing of the three-phase Hall sensor electrical signals, calculating the angular deviation between adjacent Hall signals, and detecting the coaxiality deviation of the stator and rotor air gap in the circumferential direction using a coaxiality measuring instrument; attaching the current process node number and the operator ID of the executing process as dual traceability markers, encapsulating them together with the collection timestamp into a structured data packet, and re-uploading it to the quality database via industrial Ethernet after data verification.

[0013] Furthermore, by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine to calculate the health status index, and comparing and analyzing the current health status index with the factory test benchmark and design parameter benchmark, the specific implementation process for locating the traceability node of quality deviation includes: collecting the wind turbine speed sequence, calculating the ratio of the standard deviation to the mean of the speed in the current period to obtain the speed fluctuation rate; collecting the three-phase current waveform of the wind turbine through a current transformer, extracting the amplitude of each harmonic component through fast Fourier transform, and calculating the harmonic distortion rate; using the speed fluctuation rate and harmonic distortion rate as input variables, synthesizing the health status index according to the predetermined weight coefficient, and comparing it with the health status index alarm threshold; performing deviation calculations on the current health status index, speed fluctuation rate, and harmonic distortion rate with the factory test benchmark value and design parameter benchmark value stored in the quality database item by item, and tracing back step by step in the quality traceability chain according to the deviation amplitude and the deviation occurrence sequence to locate the quality deviation to the corresponding incoming material batch node, winding process node, and magnetization process node, and outputting the traceability node number and deviation quantification result.

[0014] Furthermore, during the scrapping and recycling of wind turbines, the specific implementation process of collecting scrapping reasons and component testing data from traceability nodes and feeding them back to the operating condition spectrum library includes: In the scrapping and recycling stage, the product reference file code, operating condition category, and cumulative running time of the wind turbine are extracted by scanning the QR code on the wind turbine body; offline testing is performed on the stator winding, rotor magnet, bearing, and Hall sensor of the recycled wind turbine, and the deviation of the current measured parameters of each component from the factory reference parameters and the dominant failure mode code are recorded; the scrapping reason classification identifier, component failure detection data, and the operating condition spectrum library category code to which the wind turbine belongs during operation are associated and encapsulated, and written into the failure sample set of the operating condition spectrum library in the form of a feedback data packet; the operating condition spectrum library revises and updates the boundary values ​​of PWM control parameters and the bearing clearance characteristic tolerance range under each operating condition category based on the cumulative failure sample set.

[0015] By means of the above-described solution, the present invention has at least the following advantages: 1. This invention constructs a full lifecycle quality traceability system covering the entire process from design, incoming materials, manufacturing, testing, operation, and end-of-life recycling. It correlates and binds historical thermal load, temperature and humidity fluctuations, and heat dissipation interference events of server racks with the PWM control parameters and bearing clearance characteristics of DC brushless fans, forming design anchor points for operating conditions. Furthermore, it generates product benchmark files containing design feature vectors, achieving deep coupling between product design parameters and actual operating conditions. This establishes a continuous data chain from the design source to final failure and recycling, ensuring that the performance changes, structural matching relationships, and quality evolution trajectory of the fan under different operating conditions are fully preserved. This effectively solves the problems of isolated traceability data, fragmented lifecycle data, and the inability to form systematic quality correlations.

[0016] 2. This invention establishes a three-level correlation mapping quality traceability chain and combines it with real-time acquisition of key manufacturing parameters such as stator winding coil resistance, rotor magnetization waveform, magnetic flux density uniformity, Hall sensor angle deviation, and air gap coaxiality to achieve precise tracking of the core manufacturing process of DC brushless fans. Through a dual-marking mechanism of process node numbers and operator employee numbers, each piece of manufacturing data can be accurately linked to the corresponding production stage. A health status index is constructed by combining the fan's real-time speed fluctuation rate and phase current harmonic distortion rate, and multi-dimensional deviation analysis is performed against factory test benchmarks and design parameter benchmarks. This allows for step-by-step tracing back along the quality traceability chain to the incoming material batch node, winding process node, or magnetization process node, achieving precise location of quality anomalies.

[0017] 3. In the wind turbine scrapping and recycling stage, this invention obtains the operational condition category and cumulative operating time by scanning the product's QR code. It also performs offline testing on core components such as the stator winding, rotor magnet, bearings, and Hall sensors. Failure mode codes, parameter deviations, and scrapping reasons are fed back to the operational condition spectrum library, forming a failure sample set. Based on the accumulated failure data, the boundary values ​​of PWM control parameters and the bearing clearance characteristic tolerance range are dynamically revised, thereby establishing a closed-loop quality optimization mechanism between design, manufacturing, operation, and recycling. This allows for continuous optimization of wind turbine structural design and process parameters using historical failure data, improving operational adaptability and long-term operational reliability. Simultaneously, it provides data support for subsequent product design, supply chain quality control, and production process optimization, enhancing the stability, service life, and intelligent quality management level of DC brushless wind turbine products.

[0018] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show a certain embodiment of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This invention proposes a quality traceability method for the entire lifecycle of DC brushless fan production; Figure 2 This is a schematic diagram of the product benchmark file proposed in this invention; Figure 3 This is a schematic diagram of the three-level association mapping principle proposed in this invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0022] Reference Figures 1 to 3 This invention provides a method for quality traceability throughout the entire production lifecycle of a DC brushless fan, the technical solution of which is as follows: Example 1: Reference Figure 1 This embodiment proposes a quality traceability method for the entire lifecycle of DC brushless fan production, including: Based on the historical heat load, temperature and humidity fluctuations and heat dissipation interference data of the server rack, an operating condition spectrum library is constructed and bound to the PWM control parameters of the fan and the bearing clearance characteristics to form operating condition design anchor points; the design parameters of the fan are collected from the operating condition design anchor points to generate a product benchmark file containing design feature vectors and store it in the quality database. A quality traceability chain is built by scanning the incoming barcode in the quality database. Batch numbers, component test values ​​and performance test values ​​are extracted from the product benchmark files and written into the quality traceability chain through a three-level association mapping. Data including stator winding coil resistance value, rotor magnetization waveform and magnetic flux uniformity, assembly Hall sensor angle deviation and air gap coaxiality are collected and re-uploaded to the quality database after being doubly marked with process node number and operator number. The health status index is calculated by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine. The current health status index is compared and analyzed with the factory test benchmark and design parameter benchmark to locate the source node of quality deviation. When the wind turbine is scrapped and recycled, the scrapping cause and component test data of the source node are collected and fed back to the operating condition spectrum library.

[0023] Furthermore, based on the server rack's historical thermal load, rack temperature and humidity fluctuations, and heat dissipation interference data, an operating condition spectrum library is constructed and bound to the fan's PWM control parameters and bearing clearance characteristics. The specific implementation process for forming operating condition design anchor points includes: Historical thermal load time-series data is extracted from the target server rack's maintenance database. Temperature and humidity fluctuation sampling sequences from internal temperature and humidity sensors, as well as records of heat dissipation interference events caused by airflow obstruction and short circuits, are collected. After normalization, these data are categorized by rack model and installation location and stored in the operating condition spectrum library. Based on the fan speed control protocol, the mapping relationship between the PWM duty cycle adjustment range, response frequency boundary, and target speed is extracted as a PWM control parameter set. Based on the bearing preload design documents, radial clearance tolerance zones and axial clearance tolerance zones are extracted as bearing clearance characteristics. The PWM control parameter set and bearing clearance characteristics are indexed using the rack thermal environment category code and then linked at the field level with the corresponding operating condition records in the operating condition spectrum library to generate operating condition design anchor points.

[0024] Specifically, taking a standard 42U server rack (model DC-42U-A) in a data center as the target rack, the operation and maintenance database contains 730 consecutive days of heat load records, sampled at 5-minute intervals, totaling 210,240 heat load sampling points. Each sample value records the real-time total power consumption of the servers within the rack, measured in watts, ranging from 800W to 4200W. This historical heat load time-series data reflects the actual thermal power evolution of the rack under different service load cycles and serves as the foundational data source for constructing the operational condition spectrum library.

[0025] For collecting temperature and humidity fluctuation sequences, the cabinet is equipped with four platinum resistance temperature sensors and two capacitive humidity sensors, continuously outputting temperature and humidity data at a sampling interval of 10 seconds to form a temperature and humidity fluctuation sampling sequence. After collection, the historical heat load time-series data, temperature and humidity fluctuation sampling sequences, and records of heat dissipation interference events are uniformly normalized to eliminate interference from differences in dimensions and magnitudes on subsequent operating condition classification. The normalization process maps heat load values ​​to a dimensionless range of 0 to 1 based on the upper and lower limits of the physical range of each data type. Temperature values ​​are linearly normalized according to the sensor range (-20℃ to 80℃), humidity values ​​are synchronously normalized according to the range (0 to 100%RH), and heat dissipation interference levels are uniformly standardized using integer codes from level 1 to 5. Specifically, the thermal interference level is determined according to the following rules: Level 1 is defined as airflow obstruction lasting less than 5 minutes and not causing the rack outlet temperature to exceed the design limit by more than 2°C; Level 2 is defined as obstruction lasting 5 to 15 minutes or causing the outlet temperature to rise by 2°C to 5°C; Level 3 is defined as obstruction lasting 15 to 30 minutes or causing a temperature rise exceeding 5°C to 8°C; Level 4 is defined as obstruction lasting 30 to 60 minutes or causing a temperature rise exceeding 8°C to 12°C; and Level 5 is defined as obstruction lasting more than 60 minutes or causing a temperature rise exceeding 12°C. Airflow short-circuit events are determined according to the corresponding temperature rise standard. Thermal load normalization uses the maximum measured thermal load value from the historical data of each rack model as the upper limit and the minimum measured thermal load value as the lower limit, performing independent normalization for each rack model to avoid normalization mismatch between racks with different thermal power ranges.

[0026] After normalization, the data is categorized and stored in the operating condition genealogy database based on two dimensions: rack model and installation location. Taking rack model DC-42U-A, installed in row 3, column 5 of building A in the data center, as an example, all historical operating condition records for this rack are stored in the operating condition genealogy database using the triplet "rack model-row code-column code" as the classification key, forming a complete set of operating condition records for subsequent field-level binding operations. Based on the fan speed control protocol, the PWM control parameters are extracted into a set format. This PWM control parameter set includes three core parameter fields: PWM duty cycle adjustment range, and the mapping relationship between the response frequency boundary and the target speed. Based on the speed control protocol document of the DC brushless fan, the PWM duty cycle adjustment range is extracted to be 20% to 100%, corresponding to a fan speed range of 800 rpm to 7200 rpm; the response frequency boundary is 25 kHz, which is the upper limit of the PWM carrier frequency. Exceeding this boundary will cause a sharp increase in coil eddy current losses; the target speed mapping relationship is recorded in the form of a duty cycle-speed correspondence table, with a total of 18 standard operating points. Each operating point includes a duty cycle value (step size 5%) and the corresponding measured average value of the rated speed. The PWM control parameter set is stored in the form of a structured table, and the field width, data type and field definition are consistent with the field definition of the operating condition spectrum library database to ensure compatibility of subsequent field-level binding operations.

[0027] Based on the bearing preload design documents, the radial and axial clearance tolerance zones are extracted as bearing clearance characteristics. Exceeding the upper limit will cause the bearing to lose preload due to thermal expansion under high-temperature conditions, while falling below the lower limit will result in excessive bearing preload and abnormally high operating friction torque. The radial and axial clearance tolerance zones together constitute the bearing clearance characteristic field group, recorded in the bearing clearance characteristic data structure with micrometers as the unit of measurement. The design values ​​of these bearing clearance characteristics directly determine the shaft system stability of the fan in different thermal environments and are one of the key dimensions linking the operating condition design anchor point with the product baseline file.

[0028] Using the rack thermal environment category code as an index, the PWM control parameter set and bearing clearance characteristics are linked and bound at the field level to the corresponding operating condition records in the operating condition spectrum library to generate operating condition design anchor points. In specific implementation, the rack thermal environment category code is automatically assigned by the system based on three normalized statistical indicators: average heat load, temperature and humidity fluctuation amplitude, and heat dissipation interference frequency, through a clustering algorithm, forming a total of 6 thermal environment categories, coded from HEC-01 to HEC-06. The clustering algorithm adopts the K-means algorithm, which uses the normalized values ​​of the average heat load, temperature and humidity fluctuation amplitude, and heat dissipation interference frequency to form a three-dimensional feature vector. The number of categories K=6 is determined based on the silhouette coefficient evaluation results. Specifically, clustering is performed stepwise within the range of K from 2 to 10, and the silhouette coefficient is calculated. The K value that maximizes the silhouette coefficient is taken as the final number of categories. In this embodiment, the silhouette coefficient corresponding to K=6 is 0.712, which is higher than the results of adjacent K values. The specific calculation method for the normalized value of temperature and humidity fluctuation amplitude is as follows: Calculate the range within a 24-hour sliding window for each temperature sampling sequence, and take the average of the four temperature ranges to obtain the temperature fluctuation amplitude; calculate the humidity fluctuation amplitude for the two humidity sampling sequences in the same way, and take the average; normalize the temperature fluctuation amplitude to 100℃ (sensor range) and the humidity fluctuation amplitude to 100%RH, and then sum them with a weighted ratio of 0.7:0.3 to obtain the normalized value of temperature and humidity fluctuation amplitude. The normalized value of heat dissipation interference frequency is taken as the cumulative number of heat dissipation interference events of levels 1 to 5 within the entire acquisition period (548 days or 730 days), and linear normalization is performed with the historical maximum number of events for each cabinet model as the upper limit. K-means clustering adopts the K-means++ initialization strategy, with a convergence criterion of centroid displacement less than 0.001 for 10 consecutive iterations. The centroid displacement is calculated using Euclidean distance. Specifically, after each iteration, for each category, the square root of the sum of the squares of the differences between the old and new centroid coordinates in three dimensions—normalized mean value of heat load, normalized value of temperature and humidity fluctuation amplitude, and normalized value of heat dissipation interference frequency—is taken to obtain the centroid displacement for that category in this iteration. The maximum value among all K categories' centroid displacements is taken as the overall centroid displacement for this iteration. When the overall centroid displacement for 10 consecutive iterations is less than 0.001, the K-means clustering algorithm is considered to have converged. The silhouette coefficient is calculated as follows: For each sample point i, the average distance a(i) (intra-cluster aggregation degree) between it and all other sample points in the same category, and the average distance b(i) (inter-cluster separation degree) between it and all sample points in its nearest neighbor category are calculated. The silhouette coefficient s(i) of this sample point is calculated as (b(i) - a(i)) / max(a(i), b(i)). The average silhouette coefficient of all sample points is taken to obtain the overall silhouette coefficient corresponding to the current K value.Clustering is performed once every 100 new historical operating records of the cabinets to update the category centers. New cabinets are assigned to the nearest category by calculating the above three statistical characteristics based on their historical operation and maintenance data and comparing them with the existing six category centers using Euclidean distance nearest neighbor comparison.

[0029] The field-level association binding operation uses HEC-03 as an example: A total of 187 operating condition records with the thermal environment category code HEC-03 are retrieved from the operating condition spectrum library. The duty cycle adjustment range field (value from 20% to 100%), response frequency boundary field (value 25kHz), and target speed mapping table field in the PWM control parameter set are respectively bound to their corresponding fields in the operating condition spectrum library. Simultaneously, the radial clearance tolerance zone field (upper limit 18 micrometers, lower limit 7 micrometers) and the axial clearance tolerance zone field (upper limit 35 micrometers, lower limit 10 micrometers) are bound to the bearing clearance feature column under the HEC-03 category in the operating condition spectrum library. After binding, the system generates a unique operating condition design anchor point record for the HEC-03 category, assigns the anchor point code ANC-2024-HEC03-001, records the anchor point generation timestamp and operation account number, completes the creation of the operating condition design anchor point, and stores it in the quality database.

[0030] This embodiment structures and stores historical data on thermal load timing, temperature and humidity fluctuations, and heat dissipation interference under the actual operating environment of the server rack into a database. This data is then linked to the PWM control parameters of the DC brushless fan and the bearing clearance design parameters. This allows for a traceable data correspondence between process decisions made at the product design stage and actual usage conditions after installation. In product lifecycle analysis, the accuracy of fault location can be improved from the process level to the operating condition parameter coupling level, enabling cross-stage root cause tracing of quality anomalies and providing quantitative data for operating condition adaptability optimization in subsequent design iterations.

[0031] Furthermore, the specific implementation process of collecting wind turbine design parameters from operating condition design anchor points, generating a product benchmark file containing design feature vectors, and storing it in the quality database includes: Using the index code of the operating condition design anchor point as the retrieval key, the set of PWM control parameters, bearing clearance characteristics, and thermal environment classification information bound to the anchor point are read from the quality database. Based on the fan design specification document, design parameters such as rated speed, locked rotor torque, insulation class, stator slot parameters, and magnetic circuit air gap target value are collected. The parameters are segmented and quantized according to data type and concatenated into a design feature vector. Using the product model code and design version number as the joint primary key, the design feature vector is encapsulated into a product baseline file, recording the file generation timestamp, the design supervisor's employee number, and the associated operating condition anchor point code. After format verification, the file is written into the quality database.

[0032] Reference Figure 2Specifically, using the anchor point code ANC-2024-HEC03-001 as the search key, a search request is initiated to the quality database. The system locates the corresponding record in the working condition spectrum database association table based on this code and reads the following three sets of bound data: First, a set of PWM control parameters, including a PWM duty cycle adjustment range of 20% to 100%, a carrier response frequency upper limit of 25kHz, and a duty cycle-speed mapping table consisting of 18 standard operating points; Second, bearing clearance characteristics, including a radial clearance tolerance band upper limit of 18 micrometers and a lower limit of 7 micrometers, and an axial clearance tolerance band upper limit of 35 micrometers and a lower limit of 10 micrometers; Third, thermal environment classification information, recording that the thermal environment category to which this anchor point belongs is HEC-03 (high load continuous type), with a corresponding normalized value of 0.82 for the mean thermal load, a normalized value of 0.67 for the temperature fluctuation amplitude, and a normalized value of 0.54 for the frequency of heat dissipation interference. The above three sets of data serve as the operating condition constraint background for subsequent design parameter collection and feature vector construction, ensuring that the generated product benchmark profile is consistent with the thermal operating conditions of the target installation environment.

[0033] Based on the wind turbine design specifications, the system collects design parameters such as rated speed, locked torque, insulation class, stator slot parameters, and target value of magnetic circuit air gap. In practical implementation, the design specification document (version V2.3) of the DC brushless fan is used as the data source, and the following design parameters are read item by item: the rated speed is 5400 rpm, which corresponds to the measured average steady-state speed under a rated PWM duty cycle of 75%; the stall torque is 42 mN·m, which is the maximum static torque output by the motor under zero speed conditions, and is read from the value marked on the torque-current characteristic curve at the zero-speed operating point in the design specification; the insulation class is F, corresponding to a maximum allowable operating temperature of 155℃ for the winding; the stator slot parameters include 12 slots, slot width of 1.8 mm, slot depth of 7.2 mm, and winding pitch of 3 slots. The above four parameters together determine the stator magnetic circuit cross-sectional area and the winding winding process window; the target value of the magnetic circuit air gap is 0.35 mm, which is the design nominal value of the radial air gap after stator and rotor assembly. This value directly affects the air gap magnetic flux density distribution and torque output linearity of the motor. The design parameters are extracted field by field from the corresponding clauses of the design specification document and bound to the design version number V2.3 to prevent the parameter versions from being mixed due to design iterations.

[0034] After collecting the above design parameters, the system performs segmented quantization encoding on each parameter according to its data type and concatenates them into a design feature vector according to a fixed field order. The segmented quantization encoding rules are formulated based on the physical dimensions and accuracy requirements of the parameters: speed parameters are in rpm, retaining integers and encoded as 4-digit decimal numbers; a rated speed of 5400 rpm is encoded as "5400"; torque parameters are in mN·m, retaining one decimal place and encoded as 5-digit numbers; a stall torque of 42.0 mN·m is encoded as "042.0"; insulation class is converted using an alphanumeric mapping rule, with F-class mapped to the numerical code "05"; in stator slot type parameters, the number of slots is encoded as a 2-digit integer "12", and the slot width is encoded as a 4-digit millimeter value "1.8". 0”, slot depth is coded as “7.20” in 4 millimeters, winding pitch is coded as “03” in 2 integers; the target value of the magnetic circuit air gap is in millimeters, with 2 decimal places, and is coded as “0.35” in 4 digits; in the PWM parameters, the duty cycle lower limit is coded as “20”, the upper limit is coded as “100”, and the response frequency upper limit is coded as “25000”; in the bearing clearance characteristics, the radial clearance upper limit is coded as “018”, the lower limit is coded as “007”, the axial clearance upper limit is coded as “035”, and the lower limit is coded as “010”; the thermal environment category code is directly referenced as “HEC03”. The encoded segments are concatenated in a fixed order, using vertical bars as field separators to form a complete design feature vector, as shown in the example below: 5400|042.0|05|12|1.80|7.20|03|0.35|20|100|25000|018|007|035|010|HEC03. This design feature vector has a fixed length, and the field order strictly corresponds to the feature vector field definition table in the quality database, allowing it to be directly used for feature comparison and deviation analysis between subsequent product batches. The field definition table contains 16 fixed segments, namely: rated speed (4-digit integer), stall torque (5-digit number including 1 decimal place), insulation class (2-digit integer code, B=01, F=05, H=07), number of stator slots (2-digit integer), slot width (4-digit number including 2 decimal places), slot depth (4-digit number including 2 decimal places), winding pitch (2-digit integer), magnetic air gap (4-digit number including 2 decimal places), PWM duty cycle lower limit (2-digit integer), PWM duty cycle upper limit (3-digit integer), PWM response frequency upper limit (5-digit integer), bearing radial clearance upper limit (3-digit integer), radial clearance lower limit (3-digit integer), axial clearance upper limit (3-digit integer), axial clearance lower limit (3-digit integer), and thermal environment category code (5-digit alphanumeric code). All models share the same field definition table, and default fields are filled with all zeros.

[0035] Using the product model code and design version number as a joint primary key, the design feature vector is encapsulated into a product baseline file, and then written into the quality database with attached file metadata. In specific implementation, the product model code is set to "BLDC-60-24V", and the design version number is set to "V2.3". The two are concatenated to form the joint primary key "BLDC-60-24V_V2.3", which serves as the unique identifier of the product baseline file in the quality database. In addition to the design feature vector itself, the product benchmark file encapsulation content also includes the following three metadata fields: the file generation timestamp is recorded as "2025-03-15 09:47:22", recording the file writing time with second-level precision, providing a timeline basis for subsequent version audits and historical comparisons; the design responsible person's employee number is recorded as "ENG-20210332", which is associated with the corresponding engineer account in the enterprise's human resources database to ensure the traceability of the design parameter source; the associated operating condition anchor point code is recorded as "ANC-2024-HEC03-001", establishing a direct association channel between the product benchmark file and the operating condition genealogy library, enabling the health status analysis in the subsequent operation phase to be traced back to the corresponding thermal environment operating condition background.

[0036] After the file is packaged, the system performs format validation, verifying field by field whether the feature vector encoding length, data type, and value range conform to the constraints of the quality database field definition table. In this implementation, the validation involves 16 field rules. If any field fails the validation, a write rejection will be triggered and an error log will be generated. The file must be manually reviewed and corrected before being resubmitted. After the format validation passes, the product benchmark file is written to the benchmark file master table of the quality database in the form of a structured record, and a new related record is added to the working condition anchor point association index table simultaneously, completing the full import of the product benchmark file.

[0037] This embodiment uses the operating condition design anchor point index code as the retrieval starting point, and uniformly encodes and encapsulates thermal environment classification information, PWM control parameters, and bearing clearance characteristics with the fan body design parameters into a design feature vector. This allows the product benchmark file to naturally carry operating condition adaptation information, rather than isolated static parameter records. A composite primary key ensures the uniqueness of the file version, and the metadata binding, combined with timestamps and the design supervisor's ID, ensures that each version of the benchmark file has complete source verifiability. In subsequent health status deviation analysis during operation, the current measured indicators can be directly compared field-by-field with the design benchmark under the target operating condition. This refines the quality deviation location accuracy from the product model level to the specific operating condition-version-parameter field level, significantly improving the directionality and analysis efficiency of fault tracing.

[0038] Furthermore, the process of constructing a quality traceability chain by scanning incoming data in the quality database, extracting batch numbers, component test values, and performance test values ​​from the product benchmark file, and writing them into the quality traceability chain through a three-level association mapping includes: The incoming material label is read by a QR code scanning device to extract the material code and supply batch number, generate an inbound scanning record and write it into the quality database, serving as the head node of the quality traceability chain; the supplier quality report arriving with the batch is subjected to structured parsing to extract the magnetic material grade and magnetic induction intensity test value, copper wire purity and wire diameter tolerance test value, and insulating varnish withstand voltage and dielectric loss performance test value, forming an incoming material quality data package; according to the three-level association mapping relationship of first-level supplier-batch, second-level batch-material, and third-level material-test item, the values ​​of each field in the incoming material quality data package are written into the corresponding level node of the quality traceability chain.

[0039] Reference Figure 3 Specifically, the incoming material receiving station is equipped with an industrial-grade QR code scanner with a scanning accuracy of 0.1 mm bar width, supporting both QR Code and Data Matrix dual-code recognition. Operators scan the outer packaging labels of the arriving magnetic materials. After decoding, the scanner uploads the original code string to the quality database. The system parses the code string to extract the material code "MAT-NdFeB-N42-001" and the supply batch number "LOT-20240312-SH-007". The material code field identifies the incoming material as N42 grade neodymium iron boron sintered magnets. In the supply batch number field, "20240312" represents the arrival date, "SH" represents the supplier's registered region code, and "007" represents the 7th batch shipment number of that supplier in that month.

[0040] The system generates an inbound barcode scanning record based on the above two fields. The record includes five fields: material code, supply batch number, barcode scanning station number "WS-IQC-03", operator employee number "OPR-20190856", and inbound timestamp "2024-03-12 14:23:05". This record is written to the traceability chain master table of the quality database using the supply batch number "LOT-20240312-SH-007" as the primary key. This record becomes the head node of the batch's quality traceability chain, and all subsequent incoming material inspection data, manufacturing process data, and operational monitoring data are linked downwards from this head node. The system performs structured parsing on the supplier quality report accompanying the batch, extracting the magnetic material grade and magnetic induction intensity test values, copper wire purity and wire diameter tolerance test values, and insulating varnish withstand voltage and dielectric loss performance test values ​​to form an incoming material quality data package. In practice, this batch of goods was accompanied by three supplier quality reports, corresponding to the three categories of incoming materials: magnetic materials, enameled copper wire, and insulating varnish.

[0041] Based on the magnetic material quality report, the system performed structured analysis and extracted the following test data: the magnetic material grade is N42; the average remanence (Br) is 1.28T; the average coercivity (Hcb) is 923kA / m; the average maximum energy product of magnetic induction is 318kJ / m³; the batch sample size was 20 pieces, and the pass rate was 100%. The magnetic induction value reflects the magnetic field output capability of the magnet under standard magnetization conditions and is directly related to the uniformity of magnetic flux density after the fan rotor magnetization process. It is a core indicator for judging the batch consistency of magnetic materials.

[0042] According to the quality report of the enameled copper wire, the system analysis and extraction showed that the purity of the copper wire was 99.97% (mass fraction), the nominal wire diameter was 0.45 mm, the wire diameter tolerance range was 0.448 mm to 0.452 mm, and the tolerance band width was 4 micrometers, which meets the requirement of ±6 micrometers of the upper limit of wire diameter tolerance specified in the design specification. The number of wire coils sampled in this batch was 15, and all of them passed the wire diameter tolerance compliance inspection.

[0043] Regarding the quality report of the insulating varnish, the system analyzed and extracted the withstand voltage test value as 3800V (power frequency withstand voltage for 1 minute), which is higher than the minimum withstand voltage requirement of 3000V specified in the design specification; the dielectric loss test value is 0.0032, which is lower than the upper limit of dielectric loss of 0.005 specified in the design specification, indicating that the energy dissipation characteristics of this batch of insulating varnish under high-frequency alternating electric field meet the requirements of Class F insulation level. All structured analysis results of the above three types of incoming materials are summarized and packaged into an incoming material quality data package. The data package contains 17 test fields and includes three traceability metadata items: supplier code "SUP-CN-NdFeB-022", quality report number "QR-20240312-007", and report issuance date "2024-03-11".

[0044] The system uses a three-tiered mapping relationship: Level 1 supplier-batch, Level 2 batch-material, and Level 3 material-testing item. It then writes the values ​​of each field in the incoming material quality data package into the corresponding level nodes of the quality traceability chain.

[0045] At the Level 1 supplier-batch association layer, the system uses the supplier code "SUP-CN-NdFeB-022" as the parent node key and attaches the supply batch number "LOT-20240312-SH-007" as a child node to the supplier's historical batch list in the quality traceability chain. Simultaneously, it writes three batch-level attribute fields: quantity received, quality report number, and warehousing timestamp. This Level 1 node records the supplier's supply history, enabling the system to horizontally compare the incoming material quality trends of different batches from the same supplier across batches. In this embodiment, this supplier has accumulated 43 batch nodes since January 2023. The historical batch magnetic induction intensity detection average fluctuation range is 1.26T to 1.31T. The current batch's 1.28T is within the historical average range, and no batch anomaly warning has been triggered.

[0046] At the second-level batch-material association layer, the system uses the supply batch number "LOT-20240312-SH-007" as the parent node key, and attaches the three types of materials contained in this batch as child nodes with material codes "MAT-NdFeB-N42-001", "MAT-CW-0.45-001", and "MAT-VL-F-001" respectively. Each material node records four material-level attribute fields: material name, specifications, quantity received, and storage location number. The second-level node establishes a correspondence between batches and specific material categories, ensuring that materials used in any process during subsequent manufacturing can be traced back to the specific supply batch through the material node. This allows for precise identification of problematic incoming materials at the batch level when manufacturing quality anomalies occur.

[0047] At the Level 3 material-test item association layer, the system uses each material code as the parent node key and attaches all test field values ​​of the corresponding material in the incoming material quality data package as test item child nodes. Taking the material code "MAT-NdFeB-N42-001" as an example, its attached test item child nodes include three test item nodes: remanence (Br), coercivity (Hcb), and maximum energy product. The enameled copper wire material node has two test item nodes: copper wire purity and wire diameter tolerance. The insulating varnish material node has two test item nodes: withstand voltage and dielectric loss. In addition to recording the test values ​​and judgment results, each test item node also synchronously records the testing equipment number and the testing execution timestamp, forming a complete source label for the test data.

[0048] After all 17 detection fields are written, the system automatically triggers the traceability chain integrity verification to check whether the parent-child relationship between the first and third level nodes is completely closed. After confirming that there are no dangling nodes, the system marks the quality traceability chain status of this batch as "warehousing completed" and writes the traceability chain sealing timestamp "2024-03-12 15:41:18" to the quality database, thus completing the construction of the full quality traceability chain for this batch of incoming materials.

[0049] This embodiment constructs a three-level tree-like association structure with the chain head node as the anchor starting point, so that the incoming batch information, material category information and specific test item values ​​form a progressively converging association relationship in the same data structure. Quality anomalies at any level can be quickly located upwards to the supplier and batch, and downwards to the specific non-conforming test item, providing a continuous data-driven basis for supplier quality rating and dynamic adjustment of incoming material acceptance standards.

[0050] Furthermore, the specific implementation process of collecting data including stator winding coil resistance, rotor magnetization waveform and magnetic flux uniformity, assembly Hall sensor angular deviation, and air gap coaxiality, and then re-uploading this data to the quality database after double-marking it with process node number and operator employee number, includes: Measurement excitation is applied to each phase winding, and the coil resistance value is collected and compared with the design reference value. The magnetic induction intensity distribution on the rotor surface is scanned, and the peak sequence of the magnetization waveform and the magnetic flux density uniformity coefficient are extracted. The zero-crossing time sequence of the three-phase Hall sensor electrical signal is collected, the angular deviation between adjacent Hall signals is calculated, and the coaxiality deviation of the stator and rotor air gap in the circumferential direction is detected by a coaxiality measuring instrument. The current process node number and the operator number of the process are added as dual traceability markers, and together with the collection timestamp, they are encapsulated into a structured data packet. After data verification, the data is re-uploaded to the quality database via industrial Ethernet.

[0051] Specifically, the winding station is equipped with a four-wire precision resistance meter with a measurement accuracy of ±0.1 milliohms and a range covering 0 to 200 ohms. During measurement, a 10 mA DC excitation current is applied to the U-phase, V-phase, and W-phase windings respectively, and the DC resistance value of the coil is collected phase by phase. The coil resistance value reflects the comprehensive result of the amount of copper wire used in the winding, the number of turns, and the welding contact resistance. If the inconsistency of the three-phase resistance exceeds 0.10 ohms, it will directly lead to an imbalance of the three-phase current during motor operation, causing torque fluctuations. Therefore, the difference comparison results must all meet the tolerance constraints before proceeding to the next process.

[0052] After the rotor magnetization process is completed, scan the magnetic induction intensity distribution on the rotor surface, and extract the peak sequence of the magnetization waveform and the magnetic flux density uniformity coefficient. In a specific implementation, a rotary Gauss meter scanning frame is configured at the magnetization detection station. The distance between the probe and the rotor surface is fixed at 1.0 mm, the rotor rotates at a constant speed of 1 rpm, and the Gauss meter continuously collects the normal magnetic induction intensity on the rotor surface with an angular resolution of 360 sampling points per circumference, forming a 360-point magnetic induction intensity distribution sequence.

[0053] The calculation method of the magnetic flux density uniformity coefficient is as follows: Take the absolute value of all N-pole magnetic peak values on the rotor surface respectively, and calculate the N-pole uniformity coefficient U N =Minimum absolute value of N-pole peak / Maximum absolute value of N-pole peak; Take the absolute value of all S-pole magnetic peak values, and calculate the S-pole uniformity coefficient U S =Minimum absolute value of S-pole peak / Maximum absolute value of S-pole peak; The magnetic flux density uniformity coefficient is taken as U = (U N +U S ) / 2 to eliminate the systematic deviation effect between the N pole and the S pole. The minimum value of the N-pole peak of this workpiece is 479 mT, the maximum value is 482 mT, the minimum absolute value of the S-pole peak is 478 mT, the maximum value is 481 mT, and the comprehensive calculated magnetic flux density uniformity coefficient is 0.9938. The design specification stipulates that the lower limit of the magnetic flux density uniformity coefficient is 0.980, and the judgment result of this workpiece is qualified. When the magnetic flux density uniformity coefficient is lower than the lower limit, it indicates that there are significant differences in the magnetic field intensities output by different magnetic poles, which will cause periodic torque fluctuations during the operation of the motor. In the server cabinet heat dissipation scenario, it is manifested as unstable fan speed, affecting the heat control accuracy of the cabinet. Therefore, this coefficient is the core index for quality control of the magnetization process.

[0054] After the complete assembly process, the system acquires the zero-crossing timing of the three-phase Hall sensor electrical signals, calculates the angular deviation between adjacent Hall signals, and detects the coaxiality deviation of the stator and rotor air gap in the circumferential direction using a coaxiality measuring instrument. For Hall sensor angular deviation acquisition, a low-speed rotor drive platform is configured at the assembly and testing station to drive the rotor at a uniform speed of 50 rpm, simultaneously acquiring the digital level signals output by the U, V, and W phase Hall sensors, and recording the zero-crossing timestamp sequence of the rising and falling edges of each phase Hall signal in each revolution. When the Hall sensor angular deviation exceeds the allowable range, the motor controller will be unable to trigger commutation at the correct rotor position, leading to incorrect commutation timing, increased speed fluctuation rate, and even loss of synchronization. Therefore, angular deviation detection during the assembly stage is a key control point to ensure the stability of the fan operation. For air gap coaxiality detection, a laser coaxiality measuring instrument is configured at the station, with eight measurement points evenly distributed along the circumference of the stator and rotor assembly, spaced at 45° intervals, measuring the radial air gap value between the stator inner hole and the rotor outer circle point by point. When the air gap coaxiality exceeds the upper limit, the stator and rotor eccentricity will cause the rotor to generate unilateral magnetic pull during rotation, which will aggravate the radial load on the bearing and form a coupled effect with the bearing clearance characteristic design value, which is one of the main causes of accelerated bearing failure.

[0055] The system adds a dual traceability marker of process node number and operator number to all collected data from the above three processes, and encapsulates it into a structured data packet along with the collection timestamp. After data verification, it is uploaded to the quality database via industrial Ethernet. In specific implementation, the stator winding process corresponds to the process node number "WND-NODE-012", the operator number is "OPR-20180634", and the collection timestamp is "2024-03-15 08:52:31"; the rotor magnetization process corresponds to the process node number "MAG-NODE-007", the operator number is "OPR-20200217", and the collection timestamp is "2024-03-15 10:14:45"; the complete machine assembly process corresponds to the process node number "ASM-NODE-023", the operator number is "OPR-20210445", and the collection timestamp is "2024-03-15 13:38:09".

[0056] The structured data packet uses the workpiece number "BL240315-0047" as the primary key and the process node number as the grouping key. It encapsulates the detection field values, judgment results, process node numbers, operator numbers, and data collection timestamps of each process into a single process quality record. Three processes generate three process quality records, with a total of 41 fields in the data packet. After encapsulation, the system performs format verification on the data packet. The verification includes three categories of rules: field completeness (no missing fields), numerical range compliance (each measured value is within the sensor's range), and the validity of the process node number (confirmed by comparison with the process node registry in the quality database). Once all rules pass, the system uploads the data packet to the quality database via industrial Ethernet and writes the aforementioned record to the corresponding process node layer in the quality traceability chain.

[0057] This embodiment designs targeted parameter acquisition schemes for three core processes of DC brushless fans: stator winding, rotor magnetization, and overall assembly. It incorporates product-specific pre-failure signals such as inconsistent coil resistance, deteriorated magnetic flux density uniformity, and out-of-tolerance air gap coaxiality into the quality traceability chain, ensuring a high degree of match between the coverage dimensions of manufacturing quality data and the actual failure modes of the fan. A dual-marking mechanism of process node numbers and operator employee numbers ensures that each piece of testing data has dual traceability capabilities based on process location and personnel responsibility. When health status deviations occur in subsequent operation phases, the system can accurately locate the root cause of the deviation to specific process nodes and operation records along the traceability chain, providing data support for continuous improvement of production processes and optimization of operating procedures.

[0058] Furthermore, by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine to calculate the health status index, and comparing and analyzing the current health status index with the factory test benchmark and design parameter benchmark, the specific implementation process for locating the source node of quality deviation includes: The system collects the fan speed sequence and calculates the ratio of the standard deviation to the mean speed within the current period to obtain the speed fluctuation rate. It also collects the three-phase current waveforms of the fan through a current transformer, extracts the amplitude of each harmonic component using a fast Fourier transform, and calculates the harmonic distortion rate. Using the speed fluctuation rate and harmonic distortion rate as input variables, a health status index is synthesized based on predetermined weighting coefficients and compared with the health status index alarm threshold. The system then performs deviation calculations on the current health status index, speed fluctuation rate, and harmonic distortion rate against the factory test benchmark values ​​and design parameter benchmark values ​​stored in the quality database. Based on the deviation amplitude and the timing of deviation occurrence, the system traces back level by level in the quality traceability chain, locating the quality deviation to the corresponding incoming material batch node, winding process node, and magnetization process node, and outputs the traceability node number and deviation quantification result.

[0059] Specifically, when the wind turbine is operating under rated conditions, the system collects the speed sequence through the Hall sensor signal processing unit at a sampling frequency of once every 10 milliseconds, with each sampling window lasting 5 seconds. Each window contains 500 speed sampling points. Three-phase current waveforms are collected via current transformers installed on the three-phase power supply leads, and a Fast Fourier Transform is performed to extract the fundamental frequency and the amplitudes of each harmonic component. The specific formula for calculating the harmonic distortion rate is as follows: ; Where I1 is the effective value amplitude of the fundamental component, I n The effective value amplitude of the nth harmonic component is given, and N is the truncated harmonic order. In this embodiment, N=50. The amplitude of each harmonic component is extracted by performing a Fast Fourier Transform on the three-phase current waveforms in each analysis window. The number of FFT points is 512. Before the transformation, a Hanning window is applied to the sampling sequence to suppress spectral leakage.

[0060] Using speed fluctuation rate and harmonic distortion rate as input variables, a health status deviation index is synthesized based on predetermined weighting coefficients. These predetermined weighting coefficients are determined based on statistical results of historical failure samples, with a weighting coefficient of 0.4 for speed fluctuation rate and 0.6 for harmonic distortion rate. The synthesis rules are as follows: The measured value of speed fluctuation rate ω... r Calculate the normalized excess ratio R based on the design baseline value ω0. ω =(ω r ω0) / ω0; for the measured value of harmonic distortion rate T r Calculate the normalized excess ratio R based on its design baseline value T0. T = (T r T0) / T0; the normalized excess ratios of both are multiplied by their respective weights and then summed. The health status deviation index D = 0.4 × R ω +0.6×R TThe index is 0 when the wind turbine operating parameters are exactly the same as the factory standard, and increases with the degree of degradation, reaching its minimum under the design standard condition. The system compares the health status deviation index with the alarm threshold of 0.50. If the index exceeds the threshold, a quality deviation alarm is triggered and an alarm event record is generated. The alarm threshold is taken from the lower quartile of the health status deviation index distribution of confirmed scrapped samples in the quality database. Specifically, the screening criteria for confirmed scrapped samples are: the cumulative running time has exceeded the expected design life under the corresponding operating condition category, and the dominant failure mode code recorded in the quality database is not empty (i.e., the dominant failure cause has been identified through offline component testing). Samples scrapped due to accidental damage (such as external impact, liquid intrusion, and other abnormal operating conditions leading to non-life-related scrapping) are not included in the alarm threshold statistics to ensure that the threshold statistics only reflect normal degradation failure patterns. The statistical time window selected above, from 90 to 30 days before scrapping, is based on historical sample analysis results: within this period, the health status deviation index has shown a stable upward trend, which can represent the true degradation state of the sample as it approaches failure. However, some samples within 30 days before scrapping experience sudden failures, causing abnormal jumps in the index, and including them in the statistics would interfere with the representativeness of the threshold. The cumulative number of newly added failed samples that triggers the recalculation of the threshold is set to 50. This value is determined based on the statistical sample size stability requirement: historical data has verified that the lower quartile tends to stabilize after the sample size exceeds 100. Recalculating every 50 new samples can achieve a balance between ensuring the update frequency and computational resource consumption, while ensuring that the total sample size participating in the statistics is sufficient to support the convergence of quartile estimation each time. After obtaining the sample index set, the 25th percentile is taken after sorting from smallest to largest. For the HEC-03 and HEC-05 categories, this value is concentrated in the range of 0.48 to 0.52. Finally, 0.50 is uniformly taken as the alarm threshold.

[0061] After the wind turbine completes the overall assembly inspection and passes the quality verification of each process node, it enters the factory testing stage. The factory testing station applies the rated PWM duty cycle drive to the wind turbine, and after 60 seconds of steady-state operation, it collects the speed sequence and calculates the factory speed fluctuation rate benchmark value. Simultaneously, it collects the three-phase current waveform and calculates the factory harmonic distortion rate benchmark value. According to the aforementioned health status deviation index synthesis rules, it calculates the factory health status deviation index benchmark value. The three benchmark values, along with the factory testing timestamp and testing station number, are written into the quality database and stored in association with the corresponding workpiece number and product benchmark file code, serving as a reference benchmark for calculating health status deviations in subsequent operation stages.

[0062] Harmonic distortion rate reflects the energy proportion of each harmonic component relative to the fundamental frequency in the stator electrical circuit. The change in this ratio is mainly determined by the symmetry of the winding impedance distribution. If the copper wire diameter deviation of each phase exceeds the design tolerance zone, the number of winding turns is uneven, or the inter-turn insulation is locally damaged during the stator winding process, it will lead to an asymmetrical distribution of the three-phase winding impedance, causing an abnormal increase in the amplitude of each harmonic component relative to the fundamental frequency. Therefore, when the harmonic distortion rate deviates from the factory reference deviation rate beyond the alarm threshold, the system prioritizes listing the winding process node as the first backtracking direction. Speed ​​fluctuation rate reflects the uniformity of the rotor output torque in the time domain. If there are differences in the magnetization intensity between magnetic poles during the magnetization process, or if the magnetic flux density uniformity coefficient is lower than the design lower limit, the equivalent electromagnetic torque generated by different magnetic poles under the rated drive current will exhibit amplitude differences, causing periodic fluctuations in the drive speed during steady-state operation. Therefore, when the speed fluctuation rate deviates from the factory reference deviation rate beyond the alarm threshold, the system prioritizes listing the magnetization process node as the first backtracking direction. The above correspondence is determined based on the physical transmission mechanism of the electrical and mechanical systems of the DC brushless fan. When both types of deviations trigger alarms simultaneously, the system further determines the starting process direction of the dominant failure source based on the sequence of the first continuous exceedance of the two types of deviation indicators by the factory reference.

[0063] The specific rules for the system to determine the deviation index "first time exceeding" the factory benchmark are as follows: The time series of speed fluctuation rate in the historical health status index record is scanned using a sliding window continuous over-threshold mechanism. The continuous over-exceedance judgment condition is that the speed fluctuation rate exceeds the factory test benchmark value within 3 consecutive adjacent sampling windows (each sampling window lasts for 5 seconds, and the 3 windows cover a total of 15 consecutive seconds). The deviation rate of each window relative to the benchmark is not less than 5%. The earliest sampling window start timestamp that meets this condition is recorded as the time series node of the first time the speed fluctuation rate exceeds the benchmark. The first time series node of the harmonic distortion rate exceeds the benchmark is judged using the same mechanism. Similarly, the harmonic distortion rate must exceed the factory benchmark value within 3 consecutive sampling windows and the deviation rate of each window must not be less than 5%.

[0064] After determining the tracing process direction of each deviation index, the system calculates the proportion of deviations exceeding the standard for the same batch of workpieces under the corresponding process node in the quality traceability chain to determine whether the deviation belongs to a process-related systematic deviation or an individual workpiece's occasional deviation. The abnormal clustering alarm threshold for process nodes is set at a rate of no less than 75% for the number of workpieces exceeding the standard for the corresponding deviation index in the same batch. This threshold value is based on the statistical analysis of historical failure samples in the quality database: Statistical analysis of historical cases confirmed as process-related batch-related systematic deviations shows that the proportion of affected workpieces exceeding the standard in the same batch is concentrated between 78% and 96%, with an average of 87%; while in historical cases confirmed as individual workpiece occasional failures, the proportion of exceeding the standard in the same batch is concentrated between 12% and 43%, with an average of 26%. There is a significant gap between the two distributions around 75%. Using 75% as the clustering alarm threshold can effectively distinguish between process-related systematic deviations and random occasional deviations, preventing the tracing of local occasional exceedance errors within a batch to process-related root causes.

[0065] Further retrieval of 180 historical health status index records since the wind turbine was put into operation revealed that the timing point at which the speed fluctuation rate first exceeded the factory benchmark was 7 days earlier than the timing point at which the harmonic distortion rate abnormality occurred. This indicates that the speed system abnormality occurred before the electrical system abnormality, providing timing guidance for subsequent tracing. Based on the magnitude and timing of deviations, the system traces back level by level in the quality traceability chain. Statistics show that among the 23 workpieces in the same batch under the magnetization process node MAG-NODE-007, 82.6% had a rotational speed fluctuation rate deviation exceeding 100%, exceeding the abnormal clustering alarm threshold for process nodes. Therefore, the magnetization process node MAG-NODE-007 was determined to be the source node for the rotational speed fluctuation rate deviation. Similarly, among the workpieces in the same batch under the winding process node WND-NODE-012, 73.9% had a harmonic distortion rate deviation exceeding 70%, making WND-NODE-012 the source node for the harmonic distortion rate deviation. The incoming material batch node LOT-20240312-SH-007 was verified to have a magnetic induction intensity margin of 6.7%, and no incoming material abnormality alarm was triggered. The system ultimately outputs the traceability node numbers WND-NODE-012 and MAG-NODE-007, along with the corresponding deviation quantification results, and writes them into the quality database.

[0066] This embodiment combines multidimensional deviation analysis of speed fluctuation rate and harmonic distortion rate with step-by-step backtracking of the quality traceability chain to accurately locate abnormal health status during operation to specific manufacturing process nodes, realizing closed-loop diagnosis from operation perception to manufacturing root cause, significantly improving the accuracy of quality deviation location and engineering operability.

[0067] Furthermore, the specific implementation process of collecting the reasons for scrapping and component inspection data at the traceability nodes and feeding them back to the operating condition spectrum library during the wind turbine scrapping and recycling process includes: In the end-of-life recycling process, the product baseline file code, operating condition category, and cumulative running time of the wind turbine are extracted by scanning the QR code on the wind turbine body. Offline testing is performed on the stator winding, rotor magnet, bearing, and Hall sensor of the recycled wind turbine to record the deviation of the current measured parameters from the factory baseline parameters and the dominant failure mode code of each component. The classification identifier of the scrapping cause, the component failure detection data, and the category code of the operating condition spectrum library during the wind turbine's operation are associated and encapsulated, and written into the failure sample set of the operating condition spectrum library in the form of a feedback data packet. Based on the cumulative failure sample set, the operating condition spectrum library revises and updates the boundary values ​​of the PWM control parameters and the bearing clearance characteristic tolerance range under each operating condition category.

[0068] Specifically, in the end-of-life recycling process, operators scan the QR code on the wind turbine body. The system, based on the QR code decoding result, links to the corresponding traceability node in the quality database, extracting the wind turbine's product baseline file code BLDC-60-24V_V2.3, operating condition category code HEC-03, and cumulative operating time. Offline testing is then performed on the stator windings, rotor magnets, bearings, and Hall effect sensors of the recycled wind turbine. The stator winding test results show that the average resistance of the three-phase coils increased from the factory baseline value of 3.41 ohms to 4.18 ohms, a deviation of +0.77 ohms. The winding insulation resistance decreased from the factory baseline value of 500 megohms to 63 megohms, and the dominant failure mode code was recorded as "STA-INS-DEG" (insulation degradation). The rotor magnet test results showed that the magnetic flux density uniformity coefficient decreased from the factory standard value of 0.9935 to 0.9612, and the peak magnetic induction intensity decreased from the factory standard value of 481 millitriles to 446 millitriles, with a deviation rate of 7.3%. The dominant failure mode code was recorded as "ROT-MAG-DEM" (demagnetization decay). The bearing test results showed that the measured radial clearance increased from the factory standard value of 12 micrometers to 31 micrometers, exceeding the upper limit of the design tolerance zone by 18 micrometers, with a deviation of +13 micrometers. The measured axial clearance increased from the factory standard value of 22 micrometers to 48 micrometers, exceeding the upper limit of the design tolerance zone by 35 micrometers, with a deviation of +13 micrometers. The dominant failure mode code was recorded as "BRG-CLR-EXC" (clearance out of tolerance). The Hall sensor test results showed that the three-phase spacing angle deviation increased from the factory average value of 0.3° to 1.8°, exceeding the upper limit of the design allowable deviation by 1.5°. The dominant failure mode code was recorded as "HLL-ANG-EXC" (angle out of tolerance).

[0069] The scrapping cause classification identifier is determined based on the comprehensive judgment of the four dominant failure mode codes as "composite failure - insulation degradation dominant type", and the code is recorded as "FAIL-COMP-INS". The system associates and encapsulates the scrapping cause classification identifier, the four types of component failure detection data, the deviation of each component and the failure mode code with the wind turbine commissioning condition category code HEC-03 to form a feedback data packet, and writes it into the failure sample set under the HEC-03 category in the condition spectrum library with the workpiece number BL240315-0047 as the primary key. Based on the accumulated 142 failure samples, the condition spectrum library revises and updates the boundary values ​​of PWM control parameters and the bearing clearance characteristic tolerance range under the HEC-03 category. Specifically, the system counts 91 samples out of the 142 samples where bearing clearance exceeding the tolerance is the dominant failure cause, and the average cumulative running time of such samples is 27840 hours, which is lower than the expected design life value. Based on the failure sample distribution, the system revised and tightened the upper limit of the radial clearance tolerance zone for bearings under the HEC-03 category from the original design value of 18 micrometers to 15 micrometers, and the upper limit of the axial clearance tolerance zone from 35 micrometers to 29 micrometers. Simultaneously, the upper limit of the long-term operation of the PWM duty cycle under this operating condition category was lowered from 100% to 92%. These revisions were updated to the corresponding fields in the operating condition spectrum library and simultaneously pushed to the product baseline file for subsequent use as operating condition design anchor points in the design phase of new wind turbine models.

[0070] This embodiment establishes an adaptive quality optimization mechanism covering the entire product lifecycle by forming a closed-loop feedback between the measured failure data during the scrapping and recycling stage and the operating condition spectrum library. This enables the boundary values ​​of design parameters to be continuously iteratively corrected based on real failure samples, effectively improving the reliability and design life matching accuracy of subsequent products under the same operating conditions.

[0071] Example 2: This embodiment fully deploys the above-mentioned quality traceability method for the entire life cycle of DC brushless fan production on the system of a DC brushless fan production line, with the application scenario being convection cooling of high-density computer cabinets.

[0072] Furthermore, during the construction of the operational condition spectrum library and the generation of operational condition design anchor points, the system extracted historical operational data from the data center's operation and maintenance management platform for the 47U high-density rack (model HC-47U-B) compatible with the target model. The extraction period was 548 consecutive days, with a sampling interval of once every 3 minutes, resulting in 263,040 thermal load sampling points. The single-point thermal load value ranged from 1200 watts to 6800 watts, exceeding the thermal power range of the BLDC-60-24V model, reflecting the enhanced heat dissipation requirements under high-density computing scenarios. The rack is equipped with 6 platinum resistance temperature sensors and 3 capacitive humidity sensors. The measured temperature range was 22℃ to 61℃, with a maximum daily temperature fluctuation of 17℃, and the humidity range was 20%RH to 68%RH. Regarding thermal interference event recording, 79 airflow obstruction events and 44 airflow short-circuit events were recorded within 548 days, a higher frequency than in low-density rack scenarios. After normalization, the above data was categorized by cabinet model HC-47U-B and installation location and stored in the operating condition spectrum library. A duty cycle-speed mapping table was extracted based on the BLDC-80-48V fan speed control protocol, consisting of a PWM duty cycle adjustment range of 15% to 100%, a response frequency upper limit of 30 kHz, and 22 standard operating points. Furthermore, based on the preload design documents for the matching 7204 series angular contact ball bearings, the upper limit of radial clearance tolerance (16 micrometers) and the lower limit of 6 micrometers, and the upper limit of axial clearance tolerance (30 micrometers) and the lower limit of 8 micrometers were extracted as bearing clearance characteristics. Using the cabinet thermal environment category code HEC-05 (high-density enhanced heat dissipation type, normalized average thermal load value of 0.91) as the index, field-level association binding was performed to generate the operating condition design anchor point ANC-2024-HEC05-003. The anchor point generation timestamp and the design supervisor's employee number were recorded and written into the quality database.

[0073] Furthermore, during the product baseline profile generation stage, the system uses the anchor point code ANC-2024-HEC05-003 as the search key to read the PWM control parameter set, bearing clearance characteristics, and thermal environment classification information bound to the anchor point. Based on the BLDC-80-48V fan design specification document (version V1.6), it collects the rated speed of 6800 rpm, stall torque of 68 millinewtons, insulation class H (corresponding to a maximum allowable winding operating temperature of 180℃), number of stator slots (12), and slot width. Design parameters such as 2.1 mm, slot depth 8.6 mm, winding pitch 3 slots, and target value of magnetic circuit air gap 0.40 mm are segmented, quantized, encoded, and concatenated into a design feature vector. The product baseline file is packaged with the joint primary key BLDC-80-48V_V1.6, and the file generation timestamp, design supervisor's number ENG-20190478, and associated operating condition anchor code ANC-2024-HEC05-003 are added. After passing the format verification of 16 fields, the vector is written into the quality database.

[0074] Furthermore, in the stage of constructing the quality traceability chain, at the incoming material warehousing station, the outer packaging labels of the received silicon steel sheets and enameled copper wires are scanned, and the material codes MAT-SiSteel-35W270-001 and the supply batch number LOT-20240418-WH-012 are extracted, and the head node of the chain is generated and written into the quality database. The system performs structured parsing on the quality report of the supplier for each batch, extracts the average residual magnetism detection value of the magnetic material of 1.32 T, the average coercivity detection value of 948 kA / m, the purity of the copper wire of 99.96%, the measured range of the wire diameter tolerance of 0.537 mm to 0.542 mm (nominal 0.54 mm, tolerance ±0.008 mm), the withstand voltage detection value of the insulating paint of 4200 V, and the dielectric loss detection value of 0.0028, and forms an incoming material quality data packet. According to the three-level association mapping relationship of the first-level supplier-batch, the second-level batch-material, and the third-level material-test item, the incoming material quality data packet is written into the corresponding layer nodes of the quality traceability chain level by level, and the status of the traceability chain is marked as warehousing completed.

[0075] Furthermore, in the stage of collecting manufacturing process parameters, at the winding station, a precision resistor measuring instrument applies a measurement excitation of 10 mA to the U, V, and W three-phase windings respectively. The measured resistance values of each phase coil are 3.86 Ω, 3.85 Ω, and 3.87 Ω. The design reference value is 3.85 Ω, and the deviations of the three phases are all within the tolerance band of ±0.15 Ω, so it is judged to be qualified. The process node number is WND-NODE-031 and the operator number is OPR-20211063. After the magnetization process is completed, a rotary gaussmeter scans the surface of the rotor with an angular resolution of 3,600 points. The N-pole peak values of the 4 pairs of magnetic poles are 516 mT, 514 mT, 517 mT, and 515 mT respectively, and the magnetic flux density uniformity coefficient is 0.9942, which is higher than the lower limit of the design specification of 0.980, so it is judged to be qualified. The process node number is MAG-NODE-014 and the operator number is OPR-20200389. For the whole machine assembly process, the interval angle deviations of the three-phase Hall sensors are collected as +0.3°, -0.2°, and -0.1° respectively, all within the allowable range of ±1.5°; the laser coaxiality measuring instrument measures the average air gap value as 0.401 mm at 8 points in the circumferential direction, and the maximum air gap difference at 8 points is 0.004 mm, which is lower than the upper limit of the coaxiality deviation of 0.015 mm, so it is judged to be qualified. The process node number is ASM-NODE-041 and the operator number is OPR-20190726. The above 3 processes have a total of 41 detection fields, with double traceability marks of the additional process node number and the operator number and the acquisition timestamp. After data verification, they are uploaded to the quality database through the industrial Ethernet.

[0076] Furthermore, during the health status monitoring and source tracing phase of operation, the system continuously collects operational data from the BLDC-80-48V fans deployed in the HC-47U-B rack of the data center. Taking the data from the 210th day of continuous operation of the fan with workpiece number BL240420-0112 as an example, the average speed of 500 sampling points within a 5-second sampling window is 6763 revolutions per minute, the standard deviation is 112.8 revolutions per minute, the calculated speed fluctuation rate is 0.01668, and the factory test baseline value is 0.00802; the average harmonic distortion rate of the three-phase current waveform calculated by fast Fourier transform is 0.1147, and the factory test baseline value is 0.0658. Both the speed fluctuation rate and harmonic distortion rate are significantly higher than the factory baseline. The system synthesizes a health status index based on weighting coefficients of 0.40 and 0.60, with a calculated result of 0.768, exceeding the alarm threshold of 0.500, triggering a deviation alarm. The system traces back level by level along the quality traceability chain. Statistics show that among the 26 workpieces in the same batch under the magnetization process node MAG-NODE-014, 79.6% had a rotational speed fluctuation rate deviation exceeding 100%, thus identifying MAG-NODE-014 as the source node for rotational speed deviation. Similarly, under the winding process node WND-NODE-031, 71.2% of the workpieces had a harmonic distortion rate deviation exceeding 70%, identifying WND-NODE-031 as the source node for harmonic distortion rate deviation. The incoming material batch node LOT-20240418-WH-012 was verified, and no alarms were triggered for any remaining quantities. The system outputs the traceability node numbers WND-NODE-031 and MAG-NODE-014, along with the deviation quantification results, writes them to the quality database, and pushes a process review alarm.

[0077] Furthermore, during the closed-loop update phase of the scrapping and recycling and operating condition spectrum database, taking the scrapping and recycling of wind turbine BL240420-0112 after a cumulative operation of 28,512 hours as an example, the QR code on the machine body is scanned to link to the traceability node, extracting the product baseline file code BLDC-80-48V_V1.6, operating condition category HEC-05, and cumulative operating time of 28,512 hours. Offline testing shows that the stator winding insulation resistance decreased from the factory baseline value of 480 megohms to 58 megohms, the rotor magnet magnetic flux density uniformity coefficient decreased from 0.9942 to 0.9589, the bearing radial clearance increased from the factory baseline value of 11 micrometers to 34 micrometers (exceeding the upper tolerance limit of 16 micrometers), and the Hall sensor three-phase spacing angle deviation increased from 0.2° to 2.1° (exceeding the allowable upper limit of 1.5°). The dominant failure mode codes are recorded as insulation degradation, demagnetization attenuation, clearance deviation, and angle deviation, respectively. The system associates and encapsulates the scrap cause classification identifier FAIL-COMP-BRG (bearing failure-dominant type), component failure detection data, and operating condition category code HEC-05, and writes them into the failure sample set of the operating condition spectrum library. At the time of execution in this embodiment, the HEC-05 category had accumulated 168 failure samples. Based on the cumulative sample statistics, the operating condition spectrum library shows that 71.4% of the samples had bearing clearance exceeding tolerance as the dominant failure cause, corresponding to an average cumulative runtime of 27,200 hours, lower than the expected design life. The system tightens the upper limit of the radial clearance tolerance band for bearings under the HEC-05 category from 16 micrometers to 13 micrometers and the upper limit of the axial clearance tolerance band from 30 micrometers to 25 micrometers. Simultaneously, it lowers the long-term operating limit of the PWM duty cycle from 100% to 90%. The revised results are updated in the operating condition spectrum library and pushed to the product baseline file generation module for subsequent design calls.

[0078] This embodiment achieves continuous data chain traceability throughout the entire lifecycle of the DC brushless fan, from design, incoming materials, manufacturing, operation to scrapping and recycling, in scenarios with higher operating heat load intensity and significant differences in fan design parameters. It refines the accuracy of quality deviation tracing from the product batch level to specific process nodes and operator levels, and dynamically revises operating condition design parameters through closed-loop feedback of failure samples, effectively improving the product quality management level and long-term operational reliability of the DC brushless fan in complex thermal environments.

[0079] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for quality traceability throughout the entire production lifecycle of a DC brushless fan, characterized in that: Based on the historical heat load, temperature and humidity fluctuations and heat dissipation interference data of the server rack, an operating condition spectrum library is constructed and bound to the PWM control parameters of the fan and the bearing clearance characteristics to form operating condition design anchor points; the design parameters of the fan are collected from the operating condition design anchor points to generate a product benchmark file containing design feature vectors and store it in the quality database. A quality traceability chain is built by scanning the incoming barcode in the quality database. Batch numbers, component test values ​​and performance test values ​​are extracted from the product benchmark files and written into the quality traceability chain through a three-level association mapping. Data including stator winding coil resistance value, rotor magnetization waveform and magnetic flux uniformity, assembly Hall sensor angle deviation and air gap coaxiality are collected and re-uploaded to the quality database after being doubly marked with process node number and operator number. The health status index is calculated by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine. The current health status index is compared and analyzed with the factory test benchmark and design parameter benchmark to locate the source node of quality deviation. When the wind turbine is scrapped and recycled, the scrapping cause and component test data of the source node are collected and fed back to the operating condition spectrum library.

2. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The specific implementation process of constructing an operating condition spectrum library based on historical thermal load, temperature and humidity fluctuations, and heat dissipation interference data of server racks, and binding it with the PWM control parameters of fans and bearing clearance characteristics to form operating condition design anchor points includes: extracting historical thermal load time-series data from the target server rack's maintenance database; collecting temperature and humidity fluctuation sampling sequences output by internal temperature and humidity sensors and records of heat dissipation interference events caused by air duct obstruction and airflow short circuits; and storing these data in the operating condition spectrum library after normalization and categorization by rack model and installation location. Based on the fan speed control protocol, the mapping relationship between the PWM duty cycle adjustment range, response frequency boundary, and target speed is extracted as a PWM control parameter set. Based on the bearing preload design documents, the radial clearance tolerance zone and axial clearance tolerance zone are extracted as bearing clearance characteristics. The PWM control parameter set and bearing clearance characteristics are indexed by the rack thermal environment category code and bound to the corresponding operating condition records in the operating condition spectrum library at the field level to generate operating condition design anchor points.

3. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The specific implementation process of collecting wind turbine design parameters from operating condition design anchor points, generating a product benchmark file containing design feature vectors, and storing it in the quality database includes: using the index code of the operating condition design anchor point as the retrieval key, reading the PWM control parameter set, bearing clearance characteristics, and thermal environment classification information bound to the anchor point from the quality database; collecting design parameters such as rated speed, locked rotor torque, insulation class, stator slot parameters, and magnetic circuit air gap target value according to the wind turbine design specification document; quantizing and encoding the parameters by data type and concatenating them into a design feature vector; encapsulating the design feature vector into a product benchmark file using the product model code and design version number as the joint primary key, recording the file generation timestamp, the design supervisor's employee number, and the associated operating condition anchor point code, and writing it into the quality database after format verification.

4. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The process of constructing a quality traceability chain by scanning incoming materials in the quality database involves extracting batch numbers, component test values, and performance test values ​​from product benchmark files and writing them into the quality traceability chain through a three-level association mapping. Specifically, this includes: reading incoming material labels using a QR code scanner to extract material codes and supply batch numbers, generating incoming material scanning records, and writing them into the quality database as the head node of the quality traceability chain; performing structured parsing on supplier quality reports arriving with each batch to extract magnetic material grades and magnetic induction intensity test values, copper wire purity and wire diameter tolerance test values, and insulating varnish withstand voltage and dielectric loss performance test values, forming an incoming material quality data package; and writing the values ​​of each field in the incoming material quality data package into the corresponding level nodes of the quality traceability chain according to the three-level association mapping relationship: first-level supplier-batch, second-level batch-material, and third-level material-test item.

5. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The specific implementation process of collecting data including stator winding coil resistance, rotor magnetization waveform and magnetic flux density uniformity, assembly Hall sensor angular deviation, and air gap coaxiality, and then re-uploading it to the quality database after being doubly tagged with process node number and operator ID, includes: applying measurement excitation to each phase winding, collecting coil resistance values ​​and comparing them with design reference values; scanning the magnetic induction intensity distribution on the rotor surface, extracting the peak sequence of the magnetization waveform and the magnetic flux density uniformity coefficient; collecting the zero-crossing timing of the three-phase Hall sensor electrical signals, calculating the angular deviation between adjacent Hall signals, and detecting the coaxiality deviation of the stator and rotor air gap in the circumferential direction using a coaxiality measuring instrument; attaching the current process node number and the operator ID of the executing process as dual traceability markers, encapsulating them together with the collection timestamp into a structured data packet, and re-uploading it to the quality database via industrial Ethernet after data verification.

6. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The process of calculating a health status index by combining the real-time speed fluctuation rate and phase current harmonic distortion rate of the wind turbine, and comparing the current health status index with the factory test benchmark and design parameter benchmark to locate the traceability node of quality deviation includes: collecting the wind turbine speed sequence, calculating the ratio of the standard deviation to the mean of the speed in the current period to obtain the speed fluctuation rate; collecting the three-phase current waveform of the wind turbine through a current transformer, extracting the amplitude of each harmonic component through fast Fourier transform, and calculating the harmonic distortion rate; using the speed fluctuation rate and harmonic distortion rate as input variables, synthesizing the health status index according to the predetermined weight coefficient, and comparing it with the health status index alarm threshold; performing deviation calculations on the current health status index, speed fluctuation rate, and harmonic distortion rate with the factory test benchmark value and design parameter benchmark value stored in the quality database item by item, and tracing back step by step in the quality traceability chain according to the deviation amplitude and the deviation occurrence sequence to locate the quality deviation to the corresponding incoming material batch node, winding process node, and magnetization process node, and outputting the traceability node number and deviation quantification result.

7. The method for quality traceability throughout the entire production lifecycle of a DC brushless fan according to claim 1, characterized in that: The specific implementation process of collecting the scrapping reasons and component test data from the traceability nodes and feeding them back to the operating condition spectrum library during the scrapping and recycling of wind turbines includes: In the scrapping and recycling stage, the product reference file code, the type of operating condition, and the cumulative running time of the wind turbine are extracted by scanning the QR code on the wind turbine body; offline testing is performed on the stator winding, rotor magnet, bearing, and Hall sensor of the recycled wind turbine, and the deviation of the current measured parameters of each component from the factory reference parameters and the dominant failure mode code are recorded; the scrapping reason classification identifier, component failure test data, and the operating condition spectrum library category code to which the wind turbine belongs during operation are associated and encapsulated, and written into the failure sample set of the operating condition spectrum library in the form of a feedback data packet; the operating condition spectrum library revises and updates the boundary values ​​of PWM control parameters and the bearing clearance characteristic tolerance range under each operating condition category based on the cumulative failure sample set.

Images