A vehicle-mounted non-inductive brushless fan operation state data processing method and system

Abstract

The present application relates to the technical field of vehicle-mounted fan monitoring, and particularly relates to a vehicle-mounted non-inductive brushless fan operation state data processing method and system. The method comprises: collecting a bus current signal under a preset starting sequence, obtaining a to-be-tested current sequence through processing, and obtaining a standard reference sequence; calculating a second-order difference feature of the standard reference sequence and constructing a weight model to obtain a weighted shortest path distance between the to-be-tested current sequence and the standard reference sequence; establishing a sliding window of historical qualified product distance values, calculating a dynamic batch reference value, and determining a relative deviation degree; extracting an energy feature of the to-be-tested current sequence, and using a multi-dimensional health assessment model to calculate a health confidence index and determine a fan state in combination with the relative deviation degree. That is, the scheme of the present application realizes full-process closed-loop intelligent detection through deep feature extraction and batch dynamic compensation, and solves the problem of extensive lag of the traditional threshold method.

Description

Technical Field

[0001] The present invention relates to the technical field of on-vehicle fan monitoring. More specifically, the present invention relates to a method and system for processing operation state data of an on-vehicle induction brushless fan. Background Art

[0002] The on-vehicle induction brushless fan is a core component in the automotive seat ventilation and power battery cooling systems. Such fans usually rely on the back electromotive force zero-crossing detection technology to achieve commutation control. Therefore, it is of great significance to determine through the effective value of the feedback current or the rotational speed, etc. in the functional test link. Existing conventional solutions mostly use a set fixed current range to perform static threshold determination, and if the collected value falls within the corresponding range, it is regarded as qualified. This kind of technology has established a determination logic for product screening by the range of macroscopic physical parameters. Although it can meet the basic function discrimination, it shows obvious extensiveness in the field of refined quality management.

[0003] For more stringent detection requirements, improved solutions attempt to collect the details of the commutation signal or combine the dynamic feedback information during the startup process. These improvement measures are based on the real-time observation of the evolution characteristics of the measured values, and determine the magnet steel consistency or the magnetic flux quality by correlating the variation rules of the physical waveforms. This processing logic aims to use the internal connection between the data distribution and component defects to achieve higher-precision monitoring. However, existing such technologies rely too much on a single determination criterion during specific implementation.

[0004] In addition, the above improvement methods and existing technologies expose obvious logical defects and deficiencies when processing data. First, limited by the existing sampling mechanism and fixed evaluation dimensions, the system is difficult to capture the millisecond-level microscopic distortion induced by bearing pitting or inter-turn micro-short circuit of the winding at the commutation moment. Even if the above hidden defects cause morphological distortion, as long as its macroscopic average current fluctuates within the range, the existing analysis strategy will ignore the structural abnormality of the waveform, and ultimately lead to sub-healthy products carrying the risk of early failure passing through the final market; second, the static threshold logic lacks a compensation mechanism for the data drift caused by raw material batches, often resulting in misjudgment and frequent occurrence of false non-conforming products. Finally, most of such current systems output binary results, and the data generated by them lack a continuous index of quantified health degree, thus it is difficult to support the higher-dimensional equipment monitoring and product predictive maintenance requirements. Summary of the Invention

[0005] The purpose of the present invention is to propose a method and system for processing operation state data of an on-vehicle induction brushless fan, so as to solve the problems of missed detection of microscopic waveform distortion, misjudgment due to batch process drift, and lack of quantified health assessment index in the test of on-vehicle induction brushless fans in the prior art; for this purpose, the present invention provides solutions in the following two aspects.

[0006] In a first aspect, the present invention provides a method for processing operating status data of an on-board non-sensory brushless fan, comprising: acquiring bus current signals of the on-board non-sensory brushless fan under a preset start-up sequence; discretizing and denoising the bus current signals to obtain a current sequence to be measured, and acquiring a predetermined standard reference sequence; calculating the second-order difference features of each data point in the standard reference sequence; constructing a attention weight model based on the second-order difference features; calculating the weighted shortest path distance between the current sequence to be measured and the standard reference sequence using the attention weight model; establishing a sliding window for storing historical qualified product distance values; calculating a dynamic batch reference value for batch drift based on historical data within the sliding window; determining the relative deviation of the current product based on the weighted shortest path distance and the dynamic batch reference value; extracting the energy features of the current sequence to be measured; combining the relative deviation with the energy features; calculating the fan's health confidence index using a multidimensional health assessment model; and determining the fan's operating status based on the health confidence index.

[0007] Thus, through the above steps, the present invention can perform in-depth processing of wind turbine operation data throughout the entire process, from high-frequency acquisition to weighted feature extraction, and then to batch compensation and final scoring, forming a closed-loop intelligent detection scheme that effectively solves the problems of the traditional threshold method being crude and lagging.

[0008] Preferably, obtaining the predetermined standard reference sequence includes: selecting a gold sample fan that has passed full performance testing and recording the current waveform of the gold sample fan during startup as a standard reference sequence; calculating the maximum second-order difference of all points in the standard reference sequence and storing the maximum value as a normalization factor for normalization of the second-order difference features in subsequent calculations.

[0009] Preferably, the matching weights of each data point in the attention weight model satisfy the following relationship:

[0010]

[0011] In the formula, Indicating the first in the standard reference sequence The weight of each data point; Indicates the sensitivity amplification factor; This represents the maximum value of the second difference of all data points in the standard reference sequence; Represents the first zero-resistance constant; This represents the normalized stretching factor; Indicating the first in the standard reference sequence The second difference values ​​of each data point.

[0012] Thus, by constructing a logarithmic weighting model based on second-order difference, the algorithm essentially equips data processing with a microscope. Even if the macroscopic mean of the current is normal, as long as there is a slight curvature change at the commutation point (such as a transient change in resistance caused by a minor defect in the bearing), the weight will amplify the distance difference at that point, thereby significantly improving the detection capability of latent defects.

[0013] Preferably, the dynamic batch reference value satisfies the following relationship:

[0014]

[0015] In the formula, Indicates the dynamic batch baseline value; Indicates a backtracking index. This represents the most recently stored data; Indicates the length of the first-in-first-out queue; This represents the weighted distance value of qualified products; This is the time decay coefficient.

[0016] In this way, by establishing a follow-up threshold, the system can automatically adapt to the overall parameter drift caused by batch fluctuations in raw materials. When the characteristics of the entire batch of fans deviate, the baseline value will also be adjusted accordingly, thereby ensuring the stability of the relative deviation and completely solving the pain point of batch false kills caused by material fluctuations.

[0017] Preferably, the step of establishing a sliding window for storing historical qualified product distance values ​​includes: maintaining a first-in-first-out queue with a preset length in system memory; when the production line starts and the data in the queue has not reached the preset length, using a preset fixed empirical benchmark as the judgment basis; when the queue is full, switching to using the dynamic batch benchmark value for judgment; whenever a new qualified product is judged, pushing its distance value to the head of the queue and removing the oldest data from the tail of the queue.

[0018] Preferably, the step of extracting the energy characteristics of the current sequence under test includes: calculating the sum of squares of the current amplitudes at each sampling point in the current sequence under test to obtain the energy integral of the current test sequence; and using the same weighted moving average logic as the dynamic batch reference value to maintain the average energy reference of the current batch in real time.

[0019] Preferably, the health confidence index satisfies the following relationship:

[0020]

[0021] In the formula, Indicates the health confidence index; Indicates morphological deviations. The squaring operation applies a nonlinear penalty to large deviations. Indicates the relative deviation; Indicates the relative energy deviation rate. , This represents the energy integral of the current test sequence. Indicates the average energy baseline of the previous batch; , These are the weighting coefficients; This is the third zero-prevention constant, used to prevent the denominator from being 0; This represents the arctangent function.

[0022] In this way, through the mapping effect of the arctangent function, the dimensionless deviation characteristics are transformed into intuitive health scores from 0 to 100. This not only distinguishes between qualified and unqualified products, but also identifies sub-healthy products, providing accurate data support for production line quality grading and predictive maintenance of equipment.

[0023] Preferably, determining the operating status of the fan based on the health confidence index includes: setting a first threshold and a second threshold, wherein the first threshold is greater than the second threshold; if the health confidence index is greater than or equal to the first threshold, the fan operating status is determined to be qualified; if the health confidence index is less than the first threshold but greater than or equal to the second threshold, the fan operating status is determined to be sub-healthy and a warning signal is triggered; if the health confidence index is less than the second threshold, the fan operating status is determined to be unqualified.

[0024] In this way, by setting a gradient-based dual-threshold judgment logic, the wind turbine status has been transformed from qualitative analysis to quantitative hierarchical management; by introducing a sub-health warning interval, an early fault detection mechanism has been effectively established, which can achieve predictive maintenance while ensuring the stability of wind turbine operation, thereby reducing operation and maintenance costs and extending the service life of wind turbines.

[0025] Preferably, the second-order difference feature is used to characterize the curvature of the waveform change, and is obtained by calculating the rate of difference between the current point and the adjacent points in the standard reference sequence. The magnitude of the second-order difference feature is positively correlated with the degree of waveform distortion during current commutation.

[0026] In the second aspect, a vehicle-mounted sensorless brushless fan operation status data processing system includes:

[0027] processor;

[0028] The memory stores computer instructions for processing the operating status data of the vehicle-mounted sensorless brushless fan. When the computer instructions are executed by the processor, the system performs the above-described method for processing the operating status data of the vehicle-mounted sensorless brushless fan.

[0029] The beneficial effects of this invention are as follows: This invention solves the technical problems of difficult fault identification and susceptibility to environmental deviation interference in vehicle-mounted non-contact brushless fans in complex environments by using a comprehensive approach of second-order differential weighting, sliding window benchmark, and multi-dimensional index fusion; through a quantitative health confidence evaluation system, it achieves refined monitoring and early warning of fan status, significantly improving the management level of the entire product life cycle. Attached Figure Description

[0030] Figure 1 This schematically illustrates a flowchart of the steps in a method for processing the operating status data of a vehicle-mounted sensorless brushless fan in this embodiment.

[0031] Figure 2 This diagram illustrates the drift of process parameters and adaptive clustering analysis for different production batches.

[0032] Figure 3 The diagram illustrates a comparison of the defect interception effectiveness distribution in the multidimensional feature space. Detailed Implementation

[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0034] like Figure 1 As shown in this embodiment, a method for processing the operating status data of a vehicle-mounted sensorless brushless fan includes the following steps:

[0035] Step S1: Collect the bus current signal of the vehicle-mounted sensorless brushless fan under the preset start sequence, discretize and denoise the bus current signal to obtain the current sequence to be measured, and obtain the predetermined standard reference sequence.

[0036] Specifically, the vehicle-mounted sensorless brushless fan is connected to the test fixture, which collects the bus current signal through a high-precision shunt resistor or Hall current sensor connected in series in the circuit.

[0037] In this embodiment, in order to capture the commutation details of the brushless motor at high speed, the sampling frequency is preferably set to 50kHz (i.e., 50,000 points are collected per second), and the sampling time is preferably set to 2 seconds to cover the complete start-up and acceleration process.

[0038] The acquired raw analog signal is converted from analog to digital to obtain a discrete current sequence. Due to electromagnetic interference in the industrial environment, a wavelet transform denoising algorithm is used to filter out high-frequency glitches to obtain a smooth current sequence. For example, a db4 wavelet basis is used to perform a three-level decomposition and reconstruction of the discrete current sequence to filter out high-frequency glitches and obtain a smooth current sequence. The current sequence is represented as follows: ,in This represents the current amplitude, measured in amperes (A).

[0039] During the production line initialization or model changeover phase, a "gold standard" fan, whose performance indicators have been confirmed to be excellent through full-performance laboratory testing, is selected. Its starting current waveform is collected under identical conditions and used as a standard reference sequence. The maximum second-order difference value of all points in the standard reference sequence is calculated and stored as a normalization factor for subsequent calculations to normalize the second-order difference characteristics.

[0040] For example, a microcurrent segment with a length of 0.1 milliseconds was acquired. The original current sequence was [0.50A, 0.52A, 0.95A, 0.55A, 0.51]A. Among them, 0.95A may be an instantaneous electromagnetic interference spike.

[0041] After wavelet denoising, the peak is smoothed, resulting in a smooth current sequence of [0.50A, 0.52A, 0.53A, 0.55A, 0.51A], thus restoring the true current change trend.

[0042] Thus, by combining high-frequency oversampling with wavelet denoising, the true steepness information of the current waveform during commutation is preserved, while environmental noise is eliminated, laying a data foundation for subsequent high-precision comparison.

[0043] Step S2: Calculate the second-order difference features of each data point in the standard reference sequence, construct an attention weight model based on the second-order difference features, and use the attention weight model to calculate the weighted shortest path distance between the current sequence under test and the standard reference sequence.

[0044] First, calculate the second-order difference features of each data point in the standard reference sequence:

[0045] The second-order difference feature is used to characterize the acceleration or curvature of data changes. Specifically, it is obtained by calculating the rate of change of difference between the current point and its immediate neighbors in the standard reference sequence. The magnitude of the second-order difference feature is positively correlated with the degree of waveform distortion during current commutation.

[0046] Data points The second-order difference value satisfies the expression:

[0047]

[0048] In the formula, Indicating the first in the standard reference sequence Second difference values ​​for each data point; , , These represent the first and second elements in the standard reference sequence. , , The current value at each point.

[0049] At the commutation point, the current usually changes abruptly, and the second-order differential value is relatively large; during the steady conduction phase, the second-order differential value is close to 0.

[0050] Then, construct the attention weight model:

[0051] To amplify the influence of the commutation point in distance calculation, a weight calculation formula is constructed:

[0052]

[0053] In the formula, Indicating the first in the standard reference sequence The weight of each data point; This represents the sensitivity amplification factor, with a preferred value of 4.0; Indicating the first in the standard reference sequence Second difference values ​​for each data point; This represents the maximum value of the second difference of all data points in the standard reference sequence; This represents the first zero-prevention constant, preferably 0.0001, used to prevent the denominator from being 0; This represents the normalization stretching factor, preferably 10, used to stretch the normalized values ​​to the linear sensitivity region of the logarithmic function.

[0054] Finally, calculate the weighted shortest path distance:

[0055] Specifically, the weighted shortest path distance is calculated using the Dynamic Time Warping (DTW) algorithm. The Euclidean distance at each step is multiplied by the corresponding weight.

[0056] For example, the values ​​of three points near a certain commutation point in the standard reference sequence are: 0.50A, 0.80A, and 0.50A.

[0057] The second difference of the intermediate data points is calculated as follows: A.

[0058] Assume the maximum value of the second difference of all data points in the standard reference sequence. According to the above weighting formula:

[0059] Among them, the normalization term is ; logarithmic terms are Weight .

[0060] Based on the above calculations, the error at the commutation point will be magnified by 10.6 times.

[0061] In contrast, if it is a stationary point, the second difference value is about 0.01, and the calculated weight is about 1.1.

[0062] In this way, by using nonlinear weighting based on second-order difference, the system intelligently focuses the detection on the commutation feature point of the motor, which significantly amplifies the mathematical distance of minute mechanical assembly defects or magnetic circuit anomalies, thereby achieving precise interception of latent defects.

[0063] Step S3: Establish a sliding window for storing historical qualified product distance values, calculate the dynamic batch reference value for batch drift based on the historical data in the sliding window, and determine the relative deviation of the current product based on the weighted shortest path distance and the dynamic batch reference value.

[0064] The purpose of this step is to eliminate overall data drift and misjudgment caused by batch differences in raw materials.

[0065] Specifically, establishing a sliding window for storing historical qualified product distance values ​​includes:

[0066] Maintain a first-in, first-out (FIFO) queue with a preset length in system memory. For example, a queue of length... First-in-first-out queue, preferred The queue only stores the weighted distance values ​​of products whose judgment result is "qualified". .

[0067] When the production line starts and the data in the queue has not reached the preset length, a preset fixed experience benchmark is used as the basis for judgment; when the queue is full, the judgment is switched to the dynamic batch benchmark value; whenever a new qualified product is judged, its distance value is pushed to the front of the queue and the oldest data at the back of the queue is removed.

[0068] In an optional embodiment, an exponentially weighted moving average algorithm is used to calculate the dynamic batch baseline value. The dynamic batch baseline value satisfies the expression:

[0069]

[0070] In the formula, Indicates the dynamic batch baseline value; Indicates a backtracking index. This represents the most recently stored data; Indicates the length of the first-in-first-out queue; This represents the weighted distance value of qualified products; The time decay coefficient is preferably 0.05. This formula assigns greater weight to more recent data, enabling the benchmark value to quickly follow changes in the production line.

[0071] In an optional embodiment, the relative deviation satisfies the expression:

[0072]

[0073] In the formula, Indicates the relative deviation; This represents the second zero-prevention constant, preferably with a value of 0.001; This represents the weighted shortest path distance between the measured current sequence and the standard reference sequence. This represents the dynamic batch baseline value.

[0074] For example, the production line replaced a batch of grease, which caused a slight increase in the resistance of the fan, and the overall distance value drifted from 10.0 to 12.0.

[0075] If a dynamic benchmark is not used and the fixed threshold is set to 11.0, all new batches of products will be misjudged as non-compliant.

[0076] Using this method, the most recent data in the sliding window gradually becomes around 12.0.

[0077] For example, the three most recent values ​​in the window are 12.0, 12.0, and 12.0. After calculation, Assuming the weighted shortest path distance between the measured current sequence of a normal product and the standard reference sequence is 12.1, the calculated relative deviation of the current product is... .

[0078] A relative deviation of 1.0 indicates that the product is normal relative to the current batch, and the system determines that it is qualified.

[0079] Combination Figure 2 As shown, in different batches A, B, C, and D, the "+" sign indicates the dynamically calculated batch center. Figure 2 It can be seen that the algorithm can keep up with batch drift and effectively identify outliers that deviate from the center of the current batch.

[0080] In this way, a "follow-up threshold" is established by using a sliding window calculation of streaming data, which enables the detection standard to adapt to fluctuations in raw materials and processes, ensuring the removal of outliers and avoiding batch false kills caused by overall batch differences.

[0081] Step S4: Extract the energy characteristics of the current sequence to be tested, combine the relative deviation with the energy characteristics, and use a multidimensional health assessment model to calculate the health confidence index of the wind turbine. Determine the operating status of the wind turbine based on the health confidence index.

[0082] The purpose of this step is to fuse multidimensional features and output an intuitive quantitative score.

[0083] First, energy features are extracted. Specifically, extracting the energy features of the current sequence under test includes: calculating the sum of squares of the current amplitudes at each sampling point in the current sequence under test to obtain the energy integral of the current test sequence;

[0084] The average energy benchmark for the current batch is maintained in real time using the same weighted moving average logic as the dynamic batch benchmark value.

[0085] Energy integral of the current test sequence , Indicates the first current in the sequence to be measured. The current amplitude at each data point.

[0086] Average energy benchmark for the current batch The maintenance method is the same as the moving average algorithm in S3.

[0087] Then, the health confidence index is calculated. The health confidence index satisfies the following expression:

[0088]

[0089] In the formula, Indicates the health confidence index; Indicates morphological deviations. The squaring operation is used to apply a nonlinear penalty to large deviations. Indicates the relative deviation; Indicates the relative energy deviation rate. ; , Indicates the weighting coefficient. 5.0 is preferred. 2.0 is preferred; This represents the third zero-prevention constant, used to prevent the denominator from being zero; This represents the arctangent function.

[0090] In an optional embodiment, determining the operating status of the fan based on the health confidence index includes:

[0091] A first threshold and a second threshold are set; wherein the first threshold is greater than the second threshold; if the health confidence index is greater than or equal to the first threshold, the fan operation status is determined to be qualified; if the health confidence index is less than the first threshold but greater than or equal to the second threshold, the fan operation status is determined to be sub-healthy and an early warning signal is triggered; if the health confidence index is less than the second threshold, the fan operation status is determined to be unqualified.

[0092] For example, the first threshold is 90 points and the second threshold is 70 points.

[0093] exist At that time, the fan's operating status is judged to be qualified; At that time, the wind turbine's operating status was determined to be sub-healthy and an early warning signal was triggered; If the fan is not in good condition, its operating status is deemed unqualified.

[0094] For example, a qualified product: (Slight waveform deviation) Deviation 0%.

[0095] Calculated: ; ; point.

[0096] A defective product: (Waveform distortion) deviation 0% .

[0097] Calculated: ; ; point.

[0098] Combination Figure 3 As shown, the parabolic boundary formed by the HCI score can more accurately eliminate products with high distortion risk compared to the traditional rectangular box.

[0099] In this way, by constructing a multidimensional health assessment model, complex waveform differences and energy fluctuations are mapped into a standardized health index of 0-100 points. This not only achieves high-precision automated judgment, but also identifies "sub-healthy" products that traditional methods cannot detect, providing a more refined data dimension for production line quality management.

[0100] This invention also provides a vehicle-mounted sensorless brushless fan operation status data processing system. The system includes a processor and a memory, the memory storing computer program instructions. When the processor executes the computer program instructions, it implements the vehicle-mounted sensorless brushless fan operation status data processing method described above according to this invention.

[0101] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and therefore will not be described in detail here.

[0102] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented by computer-readable / executable instructions stored or otherwise maintained on such a computer-readable medium.

[0103] In the description of this specification, "multiple" means at least two, such as two, three or more, etc., unless otherwise expressly and specifically defined.

[0104] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A method for processing operating status data of a vehicle-mounted sensorless brushless fan, characterized in that, include: The bus current signal of the vehicle-mounted sensorless brushless fan under a preset start sequence is collected, the bus current signal is discretized and denoised to obtain the current sequence to be measured, and a predetermined standard reference sequence is obtained. Calculate the second-order difference features of each data point in the standard reference sequence, construct an attention weight model based on the second-order difference features, and use the attention weight model to calculate the weighted shortest path distance between the current under test sequence and the standard reference sequence; the matching weight of each data point in the attention weight model is as follows: Indicating the first in the standard reference sequence The weight of each data point; Indicates the sensitivity amplification factor; This represents the maximum value of the second difference of all data points in the standard reference sequence; Represents the first zero-resistance constant; This represents the normalized stretching factor; Indicating the first in the standard reference sequence Second difference values ​​for each data point; A sliding window is established to store historical qualified product distance values. A dynamic batch baseline value is calculated based on the historical data within the sliding window to account for batch drift. The relative deviation of the current product is determined based on the weighted shortest path distance and the dynamic batch baseline value. for: Indicates a backtracking index. This represents the most recently stored data; Indicates the length of the first-in-first-out queue; This represents the weighted distance value of qualified products; This is the time decay coefficient; The energy characteristics of the measured current sequence are extracted, combined with the relative deviation and the energy characteristics, and a multidimensional health assessment model is used to calculate the health confidence index of the wind turbine. The operating status of the wind turbine is determined based on the health confidence index. for: Indicates morphological deviations. The squaring operation applies a nonlinear penalty to large deviations. Indicates the relative deviation; Indicates the relative energy deviation rate. , This represents the energy integral of the current test sequence. Indicates the average energy baseline of the previous batch; , These are the weighting coefficients; This is the third zero-prevention constant, used to prevent the denominator from being 0; This represents the arctangent function.

2. The method for processing operating status data of a vehicle-mounted, sensorless, brushless fan according to claim 1, characterized in that, The process of obtaining the predetermined standard reference sequence includes: selecting a gold sample fan that has passed full performance testing and recording the current waveform of the gold sample fan during the startup process as the standard reference sequence; Calculate the maximum second-order difference of all points in the standard reference sequence, and store this maximum value as a normalization factor for normalization of the second-order difference features in subsequent calculations.

3. The method for processing operating status data of a vehicle-mounted, sensorless, brushless fan according to claim 1, characterized in that, The establishment of a sliding window for storing historical qualified product distance values ​​includes: Maintain a first-in-first-out queue of a preset length in system memory; When the production line starts up and the data in the queue has not reached the preset length, the preset fixed empirical benchmark is used as the basis for judgment. Once the queue is full, the system switches to using the dynamic batch baseline value for judgment. Whenever a new qualified product is determined, its distance value is pushed to the front of the queue, and the oldest data at the back of the queue is removed.

4. The method for processing operating status data of a vehicle-mounted sensorless brushless fan according to claim 1, characterized in that, The extraction of energy features from the current sequence under test includes: Calculate the sum of squares of the current amplitudes at each sampling point in the current sequence to be tested, and obtain the energy integral of the current test sequence. The average energy benchmark for the current batch is maintained in real time using the same weighted moving average logic as the dynamic batch benchmark value.

5. The method for processing operating status data of a vehicle-mounted sensorless brushless fan according to claim 1, characterized in that, The step of determining the operating status of the fan based on the health confidence index includes: setting a first threshold and a second threshold, wherein the first threshold is greater than the second threshold; If the health confidence index is greater than or equal to the first threshold, the fan operation status is determined to be qualified. If the health confidence index is less than the first threshold and greater than or equal to the second threshold, the wind turbine is determined to be in a sub-healthy state and an early warning signal is triggered. If the health confidence index is less than the second threshold, the fan operation status is determined to be unqualified.

6. The method for processing operating status data of a vehicle-mounted sensorless brushless fan according to claim 1, characterized in that, The second-order difference feature is used to characterize the curvature of the waveform change. It is obtained by calculating the rate of difference between the current point and the adjacent points in the standard reference sequence. The magnitude of the second-order difference feature is positively correlated with the degree of waveform distortion during current commutation.

7. A vehicle-mounted sensorless brushless fan operation status data processing system, characterized in that, include: processor; A memory storing computer instructions for processing operating status data of an on-board sensorless brushless fan, wherein when the computer instructions are executed by the processor, the system performs the on-board sensorless brushless fan operating status data processing method according to any one of claims 1-6.

Images