CN-121993364-A - Fan blade voiceprint defect identification method and system

CN121993364ACN 121993364 ACN121993364 ACN 121993364ACN-121993364-A

Abstract

The invention provides a method and a system for identifying voiceprint defects of a fan blade, and relates to the technical field of identification of voiceprint defects of fans. The method comprises the steps of collecting historical voiceprint identification data, classifying and dividing defect types to form historical voiceprint identification classified data, extracting characteristics aiming at the defect types according to the historical voiceprint identification classified data to form voiceprint defect identification characteristic data, obtaining target voiceprint data, and carrying out defect identification analysis by combining the voiceprint defect identification characteristic data to form target voiceprint defect identification data. The method can improve the analysis efficiency and fully ensure the reliability and accuracy of defect identification through a more intelligent and efficient voiceprint data processing and analyzing method.

Inventors

YAN GANG
WANG YIMIN
ZHAO JIANFANG
ZHANG MINGJIE
JIANG TINGGANG
GAO SHUAI
Zang Hongyi
YAN YI
DONG YUHUI
ZHANG CHONG

Assignees

华电山东新能源有限公司肥城分公司

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (10)

1. The method for identifying the voiceprint defects of the fan blade is characterized by comprising the following steps of: Collecting historical voiceprint identification data, and classifying and dividing defect types to form historical voiceprint identification classification data; according to the historical voiceprint recognition classification data, extracting characteristics aiming at defect types to form voiceprint defect recognition characteristic data; And acquiring target voiceprint data, and carrying out defect identification analysis by combining the voiceprint defect identification characteristic data to form target voiceprint defect identification data.
2. The method for identifying voiceprint defects of a fan blade according to claim 1, wherein the collecting historical voiceprint identification data, classifying based on defect types, and forming historical voiceprint identification classification data, comprises: extracting voiceprint data with normal monitoring results according to the historical voiceprint identification data to form historical normal voiceprint data; Extracting voiceprint data of different defect types according to the historical voiceprint identification data to form corresponding historical defect type voiceprint data; And collecting the historical normal voiceprint data and the historical defect type voiceprint data of different defect types to form the historical voiceprint identification classification data.
3. The method for identifying voiceprint defects of a fan blade according to claim 2, wherein the performing feature extraction for defect types according to the historical voiceprint identification classification data to form voiceprint defect identification feature data comprises: According to the historical normal voiceprint data, performing feature analysis aiming at the blade state to form normal voiceprint feature data; Performing defect characteristic analysis on the different historical defect type voiceprint data by combining the normal voiceprint characteristic data to form defect voiceprint characteristic data corresponding to different defect types; And collecting the normal voiceprint feature data and the different defective voiceprint feature data to form the voiceprint defect identification feature data.
4. The method for identifying a voiceprint defect of a fan blade according to claim 2, wherein the performing feature analysis for a blade state according to the historical normal voiceprint data to form normal voiceprint feature data comprises: acquiring a voiceprint normal time domain change function in each continuous time period according to the historical normal voiceprint data; obtaining normal state parameter change functions of corresponding different blade state parameters for different voiceprint normal time domain change functions; Establishing a corresponding period voiceprint-state normal relation according to different voiceprint normal time domain change functions and different corresponding normal state parameter change functions; and determining the voiceprint-state normal relation according to different voiceprint-state normal relation in the time period.
5. The method for identifying a voiceprint defect of a fan blade according to claim 4, wherein the performing defect feature analysis on the different historical defect type voiceprint data in combination with the normal voiceprint feature data to form defect voiceprint feature data corresponding to different defect types includes: Obtaining a voiceprint type defect time domain change function in each continuous time period for different historical defect type voiceprint data; Obtaining variation functions of different state parameters in corresponding time periods for different voiceprint type defect time domain variation functions, and determining corresponding voiceprint type defect time domain abnormal variation functions by combining the voiceprint-state normal relation; and carrying out correlation analysis aiming at defect parameters according to different voice print type defect time domain normalizing function to form voice print type defect parameter correlation characteristic data.
6. The method for identifying a voiceprint defect of a fan blade according to claim 5, wherein the performing correlation analysis for defect parameters according to different time domain de-normalization functions of the voiceprint type defect to form voiceprint type defect parameter correlation feature data includes: Obtaining defect characterization change functions of defect characterization parameters which change with time in corresponding time periods for different voiceprint type defect time domain normalizing change functions; Determining a correlation relation of the defect characterization of the voiceprint type of the corresponding time period according to the time domain normalizing function of the defect of the voiceprint type of the different voiceprint types and the defect characterization varying function of the corresponding different defect characterization parameters; And determining the correlation relation of the voiceprint type defects according to the correlation relation of the voiceprint type defects in different time periods.
7. The method for identifying a voiceprint defect of a fan blade according to claim 6, wherein the obtaining target voiceprint data and performing defect identification analysis in combination with the voiceprint defect identification feature data to form target voiceprint defect identification data comprises: Extracting a target voiceprint time domain change function and state parameter target change functions of different state parameters according to the target voiceprint data; performing recognition analysis of defect types according to the target voiceprint time domain change function and the voiceprint defect recognition characteristic data to form target defect recognition result information; And carrying out the confirmation analysis of the defect state according to the target defect identification result information to form defect state analysis result data.
8. The method for identifying a voiceprint defect of a fan blade according to claim 7, wherein the identifying and analyzing a defect type according to the target voiceprint time domain change function in combination with the voiceprint defect identification feature data to form target defect identification result information includes: According to different state parameter target change functions and combining the voiceprint-state normal relation, determining a target voiceprint-state normal relation function; comparing the target voiceprint time domain change function with the target voiceprint-state normal relationship function in the following manner: If the accumulated quantity of the difference between the target voiceprint time domain change function and the target voiceprint-state normal relation function in the whole target analysis period is not more than the voiceprint normal allowable deviation in the target analysis period, forming a defect-free identification result; If, in the target analysis period, the cumulative amount of the difference between the target voiceprint time-domain change function and the target voiceprint-state normal relationship function in the whole target analysis period exceeds a voiceprint normal allowable deviation, then: And comparing the difference between the target voiceprint time domain change function and the target voiceprint-state normal relation function with different voiceprint type defect time domain normalizing function, and determining the defect type corresponding to the minimum value of the difference between the target voiceprint time domain change function and the target voiceprint-state normal relation function and the accumulation of the voiceprint type defect time domain normalizing function in a target analysis period as the target defect type.
9. The method for identifying a voiceprint defect of a fan blade according to claim 8, wherein the performing a defect state validation analysis based on the target defect identification result information to form defect state analysis result data comprises: obtaining a corresponding voiceprint type defect correlation relation according to the determined defect type; And determining a target characterization parameter change function of the defect characterization parameter according to the target voiceprint time domain change function and the target voiceprint-state normal relation function and by combining the voiceprint type defect correlation relation.
10. A fan blade voiceprint defect identification system, which adopts the fan blade voiceprint defect identification method according to any one of claims 1 to 9, and is characterized by comprising: The data acquisition unit is used for acquiring historical voiceprint identification data and target voiceprint data; The classification unit is used for classifying the defect types of the historical voiceprint recognition data acquired by the data acquisition unit to form historical voiceprint recognition classification data; The characteristic extraction unit is used for extracting characteristics aiming at defect types from the historical voiceprint recognition classification data formed by the classification unit to form voiceprint defect recognition characteristic data; and the defect identification unit is used for carrying out defect identification on the target voiceprint data acquired by the data acquisition unit in combination with the voiceprint defect identification characteristic data formed by the characteristic extraction unit to form target voiceprint defect identification data.

Description

Fan blade voiceprint defect identification method and system Technical Field The invention relates to the technical field of fan voiceprint defect identification, in particular to a fan blade voiceprint defect identification method and system. Background The fan blade is the most important structural component of the fan, and because the fan blade is used for processing the motion state for a long time and continuously bearing the wind power, the probability of occurrence of faults is high, and once the faults are maintained, the fan blade is inconvenient, so that the fan blade needs to be monitored in advance to perform maintenance work before things happen, thereby improving the working efficiency of the fan and saving the operation cost. Of course, the fan blade can have operation defects before failure, and real-time monitoring of the defects can effectively develop prior maintenance work. At present, the defect monitoring of the fan blade mainly comprises manual inspection and intelligent data acquisition and analysis. The intelligent data acquisition and analysis mainly comprises the steps of acquiring voiceprint data of a fan blade, performing a series of processes including noise reduction, modal analysis and the like, and judging whether defects exist according to the processed result data. Therefore, designing the method and the system for identifying the voiceprint defects of the fan blade, and improving the analysis efficiency through a more intelligent and efficient voiceprint data processing and analysis method, can fully ensure the reliability and the accuracy of defect identification, and is a problem to be solved at present. Disclosure of Invention The invention aims to provide a fan blade voiceprint defect identification method, which is used for dividing voiceprint data aiming at different defect types and extracting corresponding defect characteristics by acquiring historical voiceprint identification data, and a perfect defect identification database is established by utilizing big data. And then the database for defect identification is utilized to monitor whether the defects exist in the data to be subjected to defect identification monitoring, on one hand, the characteristics of the defects of different types can be accurately obtained by utilizing big data, reliable and accurate basic comparison data are provided for defect identification, the accuracy of defect identification is fully ensured, and on the other hand, compared with the traditional means such as utilizing modal analysis, the method has stronger adaptability, can quickly and efficiently realize defect identification, and the real-time defect identification monitoring effect is greatly improved. The invention also aims to provide a fan blade voiceprint defect recognition system, which is combined with each other through different functional units to form a compact system capable of realizing defect recognition on voiceprint data. The data acquisition unit is responsible for continuously acquiring historical data to expand basic big data, the classification and division unit is used for completing data cleaning of the basic big data, the characteristic extraction unit is used for carrying out targeted characteristic extraction on the divided big data, defect identification on target voiceprint data is rapidly and accurately realized by means of the defect identification unit, analysis efficiency is greatly improved, and meanwhile reliability and accuracy of defect identification can be fully guaranteed. The invention provides a fan blade voiceprint defect identification method, which comprises the steps of collecting historical voiceprint identification data, classifying and dividing defect types to form historical voiceprint identification classified data, extracting characteristics aiming at the defect types according to the historical voiceprint identification classified data to form voiceprint defect identification characteristic data, acquiring target voiceprint data, and carrying out defect identification analysis by combining the voiceprint defect identification characteristic data to form target voiceprint defect identification data. According to the method, the historical voiceprint recognition data are acquired to carry out voiceprint data division and corresponding defect feature extraction aiming at different defect types, and a perfect defect recognition database is built by utilizing big data. And then the database for defect identification is utilized to monitor whether the defects exist in the data to be subjected to defect identification monitoring, on one hand, the characteristics of the defects of different types can be accurately obtained by utilizing big data, reliable and accurate basic comparison data are provided for defect identification, the accuracy of defect identification is fully ensured, and on the other hand, compared with the traditional means such as utilizing modal analysis, the method