CN-122020368-A - Charging module hierarchical diagnosis method, system and device based on big data

CN122020368ACN 122020368 ACN122020368 ACN 122020368ACN-122020368-A

Abstract

The application discloses a charging module grading diagnosis method, system and device based on big data, which are used for improving the accuracy of a charging module. The method comprises the steps of obtaining multi-mode big data of a plurality of charging modules in real time, dividing the multi-mode big data into a plurality of sub-data modules according to a preset operation sample level, obtaining a complete proportion, a noise level and a variation coefficient, calculating a data quality score based on the complete proportion and the noise level, determining fault sensitivity based on the variation coefficient, calculating the data quality score and the fault sensitivity through a weighted fusion algorithm to obtain comprehensive priority, extracting multi-scale feature vectors through a preset multi-scale feature extraction model, comparing the fluctuation amplitude of the multi-scale feature vectors with the fluctuation envelope of a preset fluctuation benchmark to obtain a deviation value, judging whether the deviation value is higher than a preset health degree threshold, if so, calling a preset dynamic twin model to simulate, and outputting fault evolution trend and a diagnosis report.

Inventors

SU XIN
Bi Shuowei
LI XIONG

Assignees

深圳易能电科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260119

Claims (10)

1. A big data based hierarchical diagnosis method for a charging module, comprising: acquiring multi-mode big data of a plurality of charging modules in real time, wherein the multi-mode big data represent each dimension data of each charging module in an operating state; dividing the multi-mode big data into a plurality of sub-data modules according to a preset operation sample level, and obtaining the complete proportion, the noise level and the variation coefficient of each sub-data module; Calculating a data quality score of each sub-data module based on the complete proportion and the noise level, and determining fault sensitivity of each sub-data module based on the variation coefficient; weighting calculation is carried out on the data quality scores and the fault sensitivities through a preset weighting fusion algorithm, so that the comprehensive priority of each sub-data module is obtained; Inputting the corresponding sub-data module into a preset multi-scale feature extraction model according to the high-low order of the comprehensive priority, and extracting multi-scale feature vectors in the sub-data module through the preset multi-scale feature extraction model; Comparing the fluctuation amplitude of the multi-scale feature vector on time sequence with the fluctuation envelope of a preset fluctuation reference, and obtaining a corresponding deviation value of the current running state of the charging module according to the proportion that the fluctuation amplitude exceeds the upper envelope boundary or the lower envelope boundary of the fluctuation envelope; Judging whether the deviation value is higher than a preset health degree threshold value or not; If yes, a preset dynamic twin model is called to simulate the multi-scale feature vector, and a fault evolution trend and a diagnosis report are output.
2. The method according to claim 1, wherein the deriving the corresponding deviation value of the current operation state of the charging module from the proportion of the fluctuation amplitude exceeding the upper envelope boundary or the lower envelope boundary of the fluctuation envelope includes: Determining an overrun frequency of the fluctuation amplitude based on a number of times the fluctuation amplitude exceeds an upper envelope boundary or a lower envelope boundary of the fluctuation envelope; Determining an overrun strength of the fluctuation amplitude based on each time the fluctuation amplitude exceeds a difference of an upper envelope boundary or a lower envelope boundary of the fluctuation envelope; and fusing and calculating the overrun frequency, the overrun intensity and a preset weight coefficient to obtain a deviation value of the current running state of the charging module.
3. The charge module hierarchical diagnosis method according to claim 2, characterized in that, after the corresponding deviation value of the current operation state of the charge module is obtained from the ratio of the fluctuation amplitude exceeding the upper envelope boundary or the lower envelope boundary of the fluctuation envelope, the charge module hierarchical diagnosis method further comprises: Dynamically updating the fluctuation envelope of the preset fluctuation reference according to the overrun frequency and the overrun intensity; the dynamically updating the fluctuation envelope of the preset fluctuation reference according to the overrun frequency and the overrun intensity comprises the following steps: calculating recommended adjustment amounts of the upper envelope boundary and the lower envelope boundary respectively based on the overrun frequency and the overrun intensity and in combination with a preset adjustment coefficient; Judging whether the recommended adjustment amount is larger than a preset maximum allowable single adjustment threshold value or not; and if not, dynamically updating the upper envelope boundary and the lower envelope boundary according to the recommended adjustment quantity.
4. The charge module hierarchical diagnosis method according to claim 3, characterized in that after said determining whether the recommended adjustment amount is greater than a preset maximum allowable single adjustment threshold, the charge module hierarchical diagnosis method further comprises: if yes, the preset maximum allowable single adjustment threshold is used as the recommended adjustment quantity to dynamically update the upper envelope boundary and the lower envelope boundary.
5. The hierarchical diagnosis method of charging modules according to claim 1, wherein the dividing the multi-modal big data into a plurality of sub-data modules according to a preset operation sample level comprises: Constructing a cross-modal joint feature template of each level based on the preset operation sample level, wherein the cross-modal joint feature template comprises cross constraint relations between each modal feature template and different modes; Synchronizing time slicing is carried out on the multi-modal big data, a plurality of cross-modal data units in the same time window are generated, and single-modal characteristics and inter-modal cross-characteristics in each cross-modal data unit are extracted; matching the single-mode characteristic and the inter-mode cross characteristic with each mode characteristic template and the cross constraint relation respectively, and calculating the matching degree; Dividing the cross-modal data units into corresponding preset operation sample grades according to the matching degree, and respectively aggregating the cross-modal data units in different grades to obtain a plurality of sub-data modules.
6. The charge module hierarchical diagnosis method according to claim 5, characterized in that after the matching degree is calculated, the charge module hierarchical diagnosis method further comprises: Sorting all the cross-modal data units from high to low according to the matching degree; Detecting whether the difference value between the matching degree corresponding to each cross-modal data unit after sequencing and the matching degree corresponding to the previous cross-modal data unit is larger than or equal to a preset minimum distinguishing margin; If yes, executing the step of dividing the cross-mode data unit to the corresponding preset operation sample level according to the matching degree.
7. The charging module hierarchical diagnosis method according to any one of claims 1 to 6, characterized in that the preset multi-scale feature extraction model includes a variation modal decomposition unit, a multi-dimensional calculation unit, and an attention fusion unit; The extracting the multi-scale feature vector in the sub-data module through the preset multi-scale feature extraction model comprises the following steps: Carrying out signal decomposition on the time sequence signals in the sub-data module through the variation mode decomposition unit to obtain a plurality of eigenvalue components with different center frequencies; respectively calculating time domain statistics, frequency domain energy spectrum and time-frequency domain joint distribution of each intrinsic mode component through the multi-dimensional calculation unit; And fusing the time domain statistics, the frequency domain energy spectrum and the time-frequency domain joint distribution through the attention fusion unit to generate the multi-scale feature vector.
8. A big data based hierarchical diagnostic system for a charging module, comprising: the first acquisition unit is used for acquiring multi-mode big data of a plurality of charging modules in real time, wherein the multi-mode big data represent each dimension data of each charging module in an operating state; The second acquisition unit is used for dividing the multi-mode big data into a plurality of sub-data modules according to a preset operation sample level and acquiring the complete proportion, the noise level and the variation coefficient of each sub-data module; The determining unit is used for calculating the data quality score of each sub-data module based on the complete proportion and the noise level and determining the fault sensitivity of each sub-data module based on the variation coefficient; The calculation unit is used for carrying out weighted calculation on the data quality scores and the fault sensitivities through a preset weighted fusion algorithm to obtain the comprehensive priority of each sub-data module; the extraction unit is used for inputting the corresponding sub-data module into a preset multi-scale feature extraction model according to the high-low order of the comprehensive priority, and extracting multi-scale feature vectors in the sub-data module through the preset multi-scale feature extraction model; The comparison unit is used for comparing the fluctuation amplitude of the multi-scale feature vector on time sequence with the fluctuation envelope of a preset fluctuation reference, and obtaining a corresponding deviation value of the current running state of the charging module according to the proportion that the fluctuation amplitude exceeds the upper envelope boundary or the lower envelope boundary of the fluctuation envelope; The judging unit is used for judging whether the deviation value is higher than a preset health degree threshold value or not; And the simulation unit is used for calling a preset dynamic twin model to simulate the multi-scale feature vector if the fault evolution trend and the diagnosis report are generated.
9. A big data based hierarchical diagnostic device for a charging module, comprising: a processor, a memory, an input-output unit, and a bus; The processor is connected with the memory, the input/output unit and the bus; The memory holds a program that the processor invokes to execute the charge module hierarchical diagnosis method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a program which, when executed on a computer, performs the charge module hierarchical diagnosis method according to any one of claims 1 to 7.

Description

Charging module hierarchical diagnosis method, system and device based on big data Technical Field The application relates to the technical field of charging, in particular to a charging module hierarchical diagnosis method, system and device based on big data. Background With the large-scale deployment of the charging module in the scenes of electric vehicle charging stations, battery energy management, industrial energy storage and the like, the running stability of the charging module directly determines the charging experience of users, the safety life of equipment and the energy utilization efficiency. Because the charging module is in the high-frequency charging and discharging environment for a long time, hidden faults such as battery capacity attenuation, poor circuit contact, heat radiation fan clamping stagnation and the like easily occur, and if the hidden faults are not accurately identified in time, the hidden faults can be gradually developed into safety accidents such as charging interruption, equipment overload and even fire. In the prior art, multi-mode big data are generally collected, and then the fault diagnosis of the charging module is realized by combining signal processing and machine learning technologies. Specifically, the collected multi-mode big data is preprocessed in the modes of filtering, missing value interpolation, normalization and the like, and then representative features of each mode in the multi-mode big data are extracted based on a time domain, a frequency domain and a time-frequency domain method. Finally, diagnostic model is adopted to carry out diagnostic analysis on the representative characteristics, and a diagnostic result is obtained. However, in the actual data collection, there may be factors of unstable working environment, for example, in the collection process, the segments are lost due to communication interruption or the conditions such as integrity, noise level, etc. of various collected big data are different due to strong electromagnetic interference on site. In the prior art, only big data with different states are preprocessed, so that the characteristics of each mode can be diluted, and the accuracy and reliability of diagnosis are seriously affected. Disclosure of Invention In order to solve the technical problems, the application provides a charging module hierarchical diagnosis method, a charging module hierarchical diagnosis system and a charging module hierarchical diagnosis device based on big data. The following describes the technical scheme provided in the present application: The first aspect of the application provides a charging module hierarchical diagnosis method based on big data, which comprises the following steps: acquiring multi-mode big data of a plurality of charging modules in real time, wherein the multi-mode big data represent each dimension data of each charging module in an operating state; dividing the multi-mode big data into a plurality of sub-data modules according to a preset operation sample level, and obtaining the complete proportion, the noise level and the variation coefficient of each sub-data module; Calculating a data quality score of each sub-data module based on the complete proportion and the noise level, and determining fault sensitivity of each sub-data module based on the variation coefficient; weighting calculation is carried out on the data quality scores and the fault sensitivities through a preset weighting fusion algorithm, so that the comprehensive priority of each sub-data module is obtained; Inputting the corresponding sub-data module into a preset multi-scale feature extraction model according to the high-low order of the comprehensive priority, and extracting multi-scale feature vectors in the sub-data module through the preset multi-scale feature extraction model; Comparing the fluctuation amplitude of the multi-scale feature vector on time sequence with the fluctuation envelope of a preset fluctuation reference, and obtaining a corresponding deviation value of the current running state of the charging module according to the proportion that the fluctuation amplitude exceeds the upper envelope boundary or the lower envelope boundary of the fluctuation envelope; Judging whether the deviation value is higher than a preset health degree threshold value or not; If yes, a preset dynamic twin model is called to simulate the multi-scale feature vector, and a fault evolution trend and a diagnosis report are output. Optionally, the obtaining the deviation value of the current running state of the charging module according to the proportion that the fluctuation amplitude exceeds the upper envelope boundary or the lower envelope boundary of the fluctuation envelope includes: Determining an overrun frequency of the fluctuation amplitude based on a number of times the fluctuation amplitude exceeds an upper envelope boundary or a lower envelope boundary of the fluctuation envelope; Determining an overrun s