Fan fault diagnosis system and method based on voiceprint recognition

Abstract

The invention belongs to the technical field of fan fault voiceprint recognition, and particularly relates to a fan fault diagnosis system and method based on voiceprint recognition. According to the method, the small change of the operation state of the fan can be reflected in time, rapid abnormal detection, early warning in advance and reduction of the risk of accidental shutdown or greater loss at the initial stage of a fault are facilitated, voiceprint information is combined with operation parameters and state data, and through multi-dimensional feature extraction and optimization processing, the operation efficiency of the fan is improved. The method greatly improves the accuracy and robustness of fault diagnosis, enables the maintenance work to be more targeted through the accurate fault recognition and residual life prediction, achieves preventive maintenance, reduces the accidental damage and downtime of equipment, reduces the operation and maintenance cost, timely discovers and processes potential faults, and improves the maintenance efficiency. It can be ensured that the draught fan continuously operates in an efficient and safe state, the accident risk is reduced, and the overall operation efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of voiceprint recognition of fan faults, and particularly relates to a fan fault diagnosis system and method based on voiceprint recognition. Background Art

[0002] With the rapid development of renewable energy worldwide, wind power, as a green and clean energy source, has gained widespread adoption worldwide. In wind power generation systems, wind turbines are core equipment, and their operating status directly impacts the overall system's power generation efficiency and safety. However, due to the long-term operation of wind turbines under complex operating conditions, internal components such as blades, bearings, and gears are prone to wear, fatigue, and other failures. This leads to frequent equipment failures, which can cause downtime, reduced efficiency, and even safety accidents. Therefore, achieving real-time monitoring of wind turbine status, fault warnings, and remaining life prediction has become a key technical issue in wind turbine operation and maintenance management.

[0003] Traditional methods for diagnosing fan faults primarily rely on technologies such as vibration monitoring, temperature detection, and infrared thermal imaging. While these methods can reflect the fan's operating status to a certain extent, they suffer from complex sensor layout, limited data collection, numerous detection blind spots, and sensitivity to environmental interference. In recent years, with the advancement of signal processing and artificial intelligence technologies, fault diagnosis methods based on voiceprint recognition have gradually attracted the attention of researchers. Voiceprint recognition technology utilizes the acoustic signals generated during fan operation and, by extracting key voiceprint features, can effectively capture internal fan fault information under non-contact conditions. It offers the advantages of low cost, strong real-time performance, and ease of implementation.

[0004] However, existing wind turbine fault diagnosis methods based on voiceprint recognition still face several challenges. First, the voiceprint information collected during wind turbine operation often contains a large amount of environmental noise, which directly affects the extraction of fault features and the accuracy of diagnosis. Second, the voiceprint characteristics of the wind turbine will change significantly under different operating modes. How to correct the voiceprint characteristics through operating mode compensation and improve the robustness of fault diagnosis is an urgent problem to be solved. In addition, the operating status of the wind turbine not only depends on the voiceprint signal, but is also closely related to the operating parameters and status of the wind turbine. How to realize the fusion processing of multi-dimensional data and comprehensively obtain accurate fault information and remaining life prediction is an important direction of current research. Summary of the Invention

[0005] The purpose of the present invention is to provide a fan fault diagnosis system and method based on voiceprint recognition, which can achieve all-round monitoring and fault diagnosis of the fan operating status through multi-level integration with fan operating parameters, status data and historical information.

[0006] The technical solutions adopted by the present invention are as follows: A fan fault diagnosis method based on voiceprint recognition, comprising: Obtain voiceprint information of the fan operation, obtain key voiceprint features based on the voiceprint information, obtain operation mode information of the fan, and obtain voiceprint feature pattern information based on the key voiceprint features and operation mode information; Obtaining the operating parameter data of the fan, and obtaining voiceprint feature parameter information based on the operating parameter data and key voiceprint features; Obtaining the fan's operating status feature data and voiceprint feature weight data, and obtaining optimized feature information based on the fan's operating status feature data and voiceprint feature weight data; The fan fault information is obtained based on the voiceprint feature pattern information, voiceprint feature parameter information and optimization feature information, and the predicted life of the fault source is obtained based on the fan fault information, and a diagnostic report is generated.

[0007] In a preferred embodiment, the steps of obtaining voiceprint information of the fan operation, obtaining key voiceprint features based on the voiceprint information, obtaining operation mode information of the fan, and obtaining voiceprint feature parameter information based on the key voiceprint features and the operation mode information include: Get the voiceprint information of the fan operation; Obtain key voiceprint features based on voiceprint information; Obtain the fan component voiceprint feature vector and the environment voiceprint feature vector based on the key voiceprint features; Get the fan's operating mode information; Acquire a mode compensation table, wherein the mode compensation table includes a plurality of operation mode information and an operation mode compensation value corresponding to each operation mode information; Obtaining target operating mode information according to the operating mode information, and obtaining a corresponding operating mode compensation value from a mode compensation table according to the target operating mode information; The voiceprint characteristic pattern value is obtained according to the voiceprint characteristic vector of the fan component, the environmental voiceprint characteristic vector and the operation mode compensation value, and is marked as voiceprint characteristic pattern information.

[0008] In a preferred embodiment, the steps of obtaining operating parameter data of the fan and obtaining voiceprint feature parameter information according to the operating parameter data and key voiceprint features include: Obtain the operating parameter data of the fan; Acquire corresponding multiple operating parameter information according to the operating parameter data; Obtain multiple operating parameter values ​​in each operating parameter information; Obtain the fan component voiceprint feature vector and the environment voiceprint feature vector based on the key voiceprint features; The voiceprint feature parameter value is obtained according to the multiple operating parameter values ​​in each operating parameter information, the fan component voiceprint feature vector and the environmental voiceprint feature vector, and is marked as voiceprint feature parameter information.

[0009] In a preferred embodiment, the steps of obtaining the operating state characteristic data and voiceprint characteristic weight data of the fan, and obtaining optimized characteristic information according to the operating state characteristic data and voiceprint characteristic weight data of the fan include: Obtain the operating status characteristics of the fan and the voiceprint feature weight data; Obtaining an operating status value of the fan according to the operating status characteristics of the fan; Obtaining a voiceprint feature weight value according to the voiceprint feature weight data; The optimized feature value is obtained according to the running state value and the voiceprint feature weight value, and marked as the optimized feature information.

[0010] In a preferred embodiment, the step of obtaining the operating status value of the fan according to the operating status characteristic data of the fan includes: Acquiring multiple operating state characteristic values ​​according to the operating state characteristic data of the fan; Obtain a state table, wherein the state table includes a plurality of operating state characteristic interval values ​​and an operating state proportion value corresponding to each operating state characteristic interval value; Obtain a target operating state characteristic interval value according to the operating state characteristic value, and obtain a corresponding operating state weight from the state table according to the target operating state characteristic interval value; The operating state value is obtained according to multiple operating state feature values ​​and corresponding operating state weights.

[0011] In a preferred embodiment, the step of obtaining the voiceprint feature weight value according to the voiceprint feature weight data includes: Obtaining the duration of voiceprint changes based on voiceprint feature weight data; Get the standard change duration; Obtain historical voiceprint feature weight value information based on voiceprint feature weight data; Acquire corresponding multiple historical voiceprint feature weight values ​​according to the historical voiceprint feature weight value information; The voiceprint feature weight value is obtained according to multiple historical voiceprint feature weight values, standard change duration, and change duration.

[0012] In a preferred embodiment, the steps of obtaining fan fault information based on voiceprint feature pattern information, voiceprint feature parameter information, and optimized feature information, obtaining a predicted life span of the fault source based on the fan fault information, and generating a diagnostic report include: Obtaining a corresponding voiceprint feature pattern value according to the voiceprint feature pattern information; Obtaining corresponding voiceprint feature parameter values ​​according to the voiceprint feature parameter information; Obtaining corresponding optimized feature values ​​according to the optimized feature information; Obtaining a comprehensive voiceprint feature value based on the voiceprint feature pattern value, the voiceprint feature parameter value, and the optimized feature value; Obtain a voiceprint fault table, wherein the voiceprint fault table includes multiple voiceprint feature comprehensive interval values ​​and fan fault information corresponding to each voiceprint feature comprehensive interval; Obtaining a target voiceprint feature comprehensive interval value based on the voiceprint feature comprehensive value; Obtain the corresponding fan fault information from the voiceprint table based on the comprehensive interval value of the target voiceprint feature Obtain the predicted life of the fault source based on the fan fault information and generate a diagnostic report.

[0013] In a preferred embodiment, the steps of obtaining the predicted life of the fault source based on the wind turbine fault information and generating a diagnostic report include: Obtaining fault source component information of the fan according to the fault information of the fan, obtaining operating state information of the fault source component according to the fault source component information, and obtaining a corresponding operating state vector of the fault source component according to the operating state information of the fault source component; Obtain the corresponding evaluation duration based on the operating status information of the fault source component, mark the time node when the wind turbine fault information is obtained as the end time of the evaluation period, and obtain the start time of the evaluation period based on the evaluation duration, and construct the evaluation period based on the start time and end time of the evaluation period; Obtaining historical sound information and component temperature information of the fault source component during the evaluation period, obtaining corresponding multiple historical sound vectors based on the historical sound information, and obtaining corresponding multiple component temperature values ​​based on the component temperature information; Get the sound weight and component temperature weight; Obtaining an operating condition value of a fault source component according to a plurality of historical sound vectors, a plurality of component temperature values, a sound weight, and a component temperature weight; Obtaining a life comparison table, wherein the life comparison table includes multiple operating condition interval values ​​and a predicted life of a fault source corresponding to each operating condition interval value; The target operating condition interval value is obtained according to the operating condition value, and the corresponding fault source predicted life is obtained from the life comparison table according to the target operating condition interval value, and a diagnosis report is generated.

[0014] The present invention also provides a fan fault diagnosis system based on voiceprint recognition, which is used in the above-mentioned fan fault diagnosis method based on voiceprint recognition, comprising: The voiceprint module is used to obtain the voiceprint information of the fan operation, obtain the key voiceprint features based on the voiceprint information, obtain the operation mode information of the fan, and obtain the voiceprint feature pattern information based on the key voiceprint features and the operation mode information; The characteristic parameter module is used to obtain the operating parameter data of the fan and obtain the voiceprint characteristic parameter information based on the operating parameter data and key voiceprint features; The optimization feature module is used to obtain the operating status feature data of the fan and the voiceprint feature weight data, and obtain the optimization feature information based on the operating status feature data of the fan and the voiceprint feature weight data; The evaluation module is used to obtain fan fault information based on voiceprint feature pattern information, voiceprint feature parameter information and optimization feature information, obtain the predicted life of the fault source based on the fan fault information, and generate a diagnostic report.

[0015] And, a fan fault diagnosis terminal based on voiceprint recognition, comprising: one or more processors; a storage device having one or more programs stored thereon; When one or more programs are executed by one or more processors, the one or more processors implement a fan fault diagnosis method based on voiceprint recognition.

[0016] The technical effects achieved by the present invention are: The present invention can promptly reflect minor changes in the operating status of the fan, help to quickly detect anomalies in the early stage of a fault, give early warning, and reduce the risk of unexpected shutdown or greater losses. It combines voiceprint information with operating parameters and status data, and through multi-dimensional feature extraction and optimization processing, it greatly improves the accuracy and robustness of fault diagnosis. Accurate fault identification and remaining life prediction make maintenance work more targeted, achieve preventive maintenance, reduce accidental damage and downtime of equipment, thereby reducing operation and maintenance costs, and timely discover and deal with potential faults, which can ensure that the fan continues to operate in an efficient and safe state, reduce accident risks, and improve overall operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of the method provided by the present invention; Figure 2 This is a system module diagram provided by the present invention. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in a preferred embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it constitute a separate or selective embodiment that is mutually exclusive of other embodiments.

[0021] Secondly, the present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention in detail, for the sake of convenience, the schematic diagrams are only examples and should not limit the scope of protection of the present invention.

[0022] Please see the attached Figure 1 As shown, a fan fault diagnosis method based on voiceprint recognition is provided, including: S1. Acquire voiceprint information of the fan operation, obtain key voiceprint features based on the voiceprint information, obtain operation mode information of the fan, and obtain voiceprint feature pattern information based on the key voiceprint features and the operation mode information; S2. Obtain operating parameter data of the fan, and obtain voiceprint feature parameter information based on the operating parameter data and key voiceprint features; S3. Obtaining the operating status feature data and voiceprint feature weight data of the fan, and obtaining optimized feature information based on the operating status feature data and voiceprint feature weight data of the fan; S4. Obtain fan fault information based on the voiceprint feature pattern information, voiceprint feature parameter information, and optimization feature information, obtain the predicted life of the fault source based on the fan fault information, and generate a diagnosis report.

[0023] As in the above steps S1 to S4, by collecting sound data during the operation of the fan, the key voiceprint features are extracted using the signal processing algorithm, and the operation mode information of the fan (such as low speed, high speed, etc.) is obtained at the same time. It is matched with the extracted key voiceprint features to obtain the voiceprint feature pattern information of the specific operation state of the fan, and the various parameter data of the fan during operation (such as temperature, vibration, pressure, etc.) are collected. Combined with the key voiceprint features obtained previously, the voiceprint feature parameter information reflecting the actual working conditions is extracted, and the optimized feature information is generated by obtaining the feature data of the fan operation state and the weight data of each voiceprint feature. The voiceprint feature pattern information, voiceprint feature parameter information and optimized feature information are used for comprehensive analysis to identify the potential fan. Fault, based on the fault information obtained by diagnosis, predicts the remaining life of the fault source and generates a detailed diagnostic report, which can timely reflect the slight changes in the operating status of the fan, help to quickly detect anomalies in the early stage of the fault, give early warning, and reduce the risk of unexpected shutdown or greater losses. The voiceprint information is combined with operating parameters and status data, and multi-dimensional feature extraction and optimization processing are used to greatly improve the accuracy and robustness of fault diagnosis. Accurate fault identification and remaining life prediction make maintenance work more targeted, realize preventive maintenance, reduce accidental damage and downtime of equipment, thereby reducing operation and maintenance costs, and timely discover and deal with potential faults, which can ensure that the fan continues to operate in an efficient and safe state, reduce accident risks, and improve overall operating efficiency.

[0024] In a preferred embodiment, the steps of obtaining voiceprint information of a fan operation, obtaining key voiceprint features based on the voiceprint information, obtaining operation mode information of the fan, and obtaining voiceprint feature parameter information based on the key voiceprint features and the operation mode information include: S101, obtaining voiceprint information of fan operation; S102, obtaining key voiceprint features based on the voiceprint information; S103, obtaining a fan component voiceprint feature vector and an environment voiceprint feature vector based on the key voiceprint features; S104, obtaining operation mode information of the fan; S105, obtaining a mode compensation table, wherein the mode compensation table includes a plurality of operation mode information and an operation mode compensation value corresponding to each operation mode information; S106. Obtain target operating mode information according to the operating mode information, and obtain a corresponding operating mode compensation value from a mode compensation table according to the target operating mode information; S107 , obtaining a voiceprint characteristic pattern value according to the fan component voiceprint characteristic vector, the environment voiceprint characteristic vector, and the operation mode compensation value, and marking it as voiceprint characteristic pattern information.

[0025] As in the above steps S101 to S107, the acoustic signals generated by the fan during operation are collected in real time by sensors installed in or around the fan to form original soundprint data. The key features reflecting the operating status of the fan, such as spectrum characteristics, energy distribution, amplitude changes, etc., are extracted from the original data using signal processing technology (such as time-frequency analysis, filtering, feature extraction algorithm, etc.). The key features are further processed to separate the inherent soundprints emitted by the mechanical components inside the fan and the parts reflecting the surrounding environmental noise, namely the fan component soundprint feature vector and the environmental soundprint feature vector, and collect the wind The current operating status information of the fan, such as low speed, high speed and other modes, reflects the operating mode under different working conditions. A mode compensation table is pre-built, which contains multiple operating modes and their corresponding compensation values. The purpose is to correct the soundprint characteristic deviation caused by the change of operating mode under different working conditions. The current mode is judged according to the actual operating mode, and the corresponding compensation value is matched. The soundprint feature vector of the fan component, the environmental soundprint feature vector and the operating mode compensation value are combined to calculate the soundprint feature mode value and mark it as the soundprint feature mode information. The calculation formula of the soundprint feature mode value is as follows: In the formula, c represents the soundprint feature mode value, F represents the soundprint feature vector of the fan component, H represents the environmental soundprint feature vector, and Y represents the operating mode compensation value. The influence of the operating conditions is corrected by mode compensation, so that the extracted soundprint features can more accurately reflect the actual operating status of the fan, thereby improving the accuracy of fault diagnosis. By using the operating mode information and mode compensation mechanism, the analysis strategy can be dynamically adjusted to adapt to the operating status of the fan under different working conditions (such as low load, high load, different speeds, etc.), ensuring the stability and consistency of the characteristic parameter information.

[0026] In a preferred embodiment, the steps of obtaining operating parameter data of the fan and obtaining voiceprint feature parameter information based on the operating parameter data and key voiceprint features include: S201, obtaining operating parameter data of the fan; S202, obtaining corresponding multiple operating parameter information according to the operating parameter data; S203, obtaining multiple operating parameter values ​​in each operating parameter information; S204, obtaining a fan component voiceprint feature vector and an environment voiceprint feature vector based on the key voiceprint features; S205 , obtaining voiceprint feature parameter values ​​according to the multiple operating parameter values ​​in each operating parameter information, the fan component voiceprint feature vector, and the environment voiceprint feature vector, and marking them as voiceprint feature parameter information.

[0027] As in the above steps S201 to S205, various physical parameters generated by the fan during operation, such as temperature, vibration, speed, pressure, load, etc., are collected in real time using sensors, monitoring systems and other equipment. Different categories of operating parameter information are extracted from the collected raw data to ensure that each key parameter has a detailed description and distinction. Each type of parameter is deeply decomposed to extract multiple values ​​or indicators. These data can more carefully reflect the operating status of the fan under different working conditions. The key voiceprint features extracted previously are used to model the voiceprints of various components inside the fan to generate voiceprint feature vectors of the fan components. At the same time, the noise generated by non-equipment in the environment is extracted to form an environmental voiceprint feature vector. The multiple parameter values ​​in each operating parameter information are combined with the voiceprint feature vectors of the fan components and the environment to calculate the voiceprint feature parameter value, mark and form the voiceprint feature parameter information. The calculation formula of the voiceprint feature parameter value is: , where w represents the voiceprint feature parameter value, F represents the fan component voiceprint feature vector, H represents the environment voiceprint feature vector, k represents the number of multiple operating parameter information, k=1,2,3…m, i represents the number of multiple operating parameter values ​​in each operating parameter information, where i=1,2,3…n, It is represented as the i-th operating parameter value in each operating parameter information. By organically combining the operating parameter data with the voiceprint characteristics, the operating status of the fan can be more comprehensively reflected, thereby accurately identifying potential abnormalities or faults in the equipment. It can adapt to changes in different working conditions (such as different loads, speeds, temperatures, etc.), making the voiceprint feature parameter information more adaptable and universal.

[0028] In a preferred embodiment, the steps of obtaining the operating state characteristic data and voiceprint characteristic weight data of the fan, and obtaining optimized characteristic information according to the operating state characteristic data and voiceprint characteristic weight data of the fan include: S301, obtaining the operating status characteristics and voiceprint feature weight data of the fan; S302, obtaining an operating status value of the fan according to the operating status characteristics of the fan; S303, obtaining a voiceprint feature weight value according to the voiceprint feature weight data; S304: Obtain an optimized feature value according to the operating state value and the voiceprint feature weight value, and mark it as optimized feature information.

[0029] As in the above steps S301 to S304, the operating status characteristic data of the fan (such as low-frequency vibration, fan heat radiation, speed stability, etc.) and the voiceprint feature weight data are collected. The operating status characteristic data is used to describe the current working condition of the fan, and the voiceprint feature weight data indicates the proportion and importance of each voiceprint feature in the overall analysis. According to the collected operating status characteristic data, one or more values ​​that can reflect the operating status of the fan, namely, the operating status value, are calculated. The voiceprint feature weight data is used to extract the voiceprint feature weight value. The operating status value is combined with the voiceprint feature weight value to calculate an optimized feature value. The calculation formula of the optimized feature value is y=Z*Q, where y represents the optimized feature value, Z represents the operating status value, and Q represents the voiceprint feature weight value. The operating status of the fan and the voiceprint feature weight are comprehensively considered, so that the final optimized feature information can more accurately reflect the actual status of the equipment, thereby improving the accuracy of fault detection and diagnosis.

[0030] In a preferred embodiment, the step of obtaining the operating status value of the wind turbine according to the operating status characteristic data of the wind turbine includes: S3021. Acquire multiple operating status characteristic values ​​according to the operating status characteristic data of the wind turbine; S3022. Obtain a state table, wherein the state table includes a plurality of operating state characteristic interval values ​​and an operating state proportion value corresponding to each operating state characteristic interval value; S3023. Obtain a target operating state characteristic interval value according to the operating state characteristic value, and obtain a corresponding operating state weight from the state table according to the target operating state characteristic interval value; S3024. Obtain an operating status value according to multiple operating status feature values ​​and corresponding operating status weights.

[0031] As in the above steps S3021 to S3024, by extracting the characteristic data of the fan operation state, multiple key operation state characteristic values ​​are extracted therefrom, such as low-frequency vibration, heat radiation of the fan rotating parts, speed stability, etc. The state table is a pre-established reference data, which contains multiple interval values ​​of operation state characteristics. Each interval value corresponds to a certain operation state proportion (or weight), reflecting the degree of influence or probability of occurrence of the interval in the overall state evaluation. According to the extracted operation state characteristic value, it is determined which target operation state characteristic interval the value falls into in the state table. Once the target interval is determined, the corresponding operation state weight can be obtained from the state table. Using multiple operation state characteristic values ​​and their corresponding weights, a comprehensive operation state value is calculated. The calculation formula of the operation state value is: , where Z represents the operating state value, i represents the number of multiple operating state characteristic values, h=1,2,3…j, Expressed as the hth operating state characteristic value, It is expressed as the operating status weight corresponding to the hth operating status eigenvalue. By extracting and assigning weights to multiple eigenvalues, the operating status of the wind turbine can be more comprehensively and finely characterized, providing quantitative indicators for equipment health assessment. By assigning weights to each eigenvalue, the interference of single feature data noise or abnormal fluctuations on the overall status assessment can be effectively reduced, making the final calculated operating status value more stable and reliable.

[0032] In a preferred embodiment, the step of obtaining the voiceprint feature weight value according to the voiceprint feature weight data includes: S3031. Obtaining the duration of the voiceprint change based on the voiceprint feature weight data; S3032. Obtain standard change duration; S3033. Obtain historical voiceprint feature weight value information based on the voiceprint feature weight data; S3034. Obtain corresponding multiple historical voiceprint feature weight values ​​according to the historical voiceprint feature weight value information; S3035. Obtain a voiceprint feature weight value according to multiple historical voiceprint feature weight values, a standard change duration, and a change duration.

[0033] As in steps S3031 to S3035 above, the duration of the voiceprint signal changing over time is extracted from the voiceprint feature weight data, and a standard change duration that is pre-set or obtained through statistical analysis is obtained. The standard duration represents the baseline value that the voiceprint feature should reach under normal or ideal working conditions. Based on the existing voiceprint feature weight data, historical data records, i.e., historical voiceprint feature weight value information, are obtained, and multiple specific historical voiceprint feature weight values ​​are extracted from them. The multiple historical voiceprint feature weight values ​​are combined with the standard change duration and the current actual change duration to calculate the current voiceprint feature weight value. The calculation formula for the voiceprint feature weight value is: , where Q represents the voiceprint feature weight value, Expressed as the duration of the change, It represents the standard change duration, r represents the number of multiple historical voiceprint feature weight values, r=2,3,4…q, Represented as the weight value of the rth historical voiceprint feature, It is expressed as the r-1th historical voiceprint feature weight value. By comparing the current change duration with the standard change duration, the weight of the voiceprint feature can be adaptively adjusted to better reflect the actual operating status and abnormal changes of the device. Using historical voiceprint feature weight data as a reference, long-term trends and fluctuation patterns can be captured, making the current weight calculation more forward-looking and robust, which helps reduce the risk of misjudgment due to single abnormal data.

[0034] In a preferred embodiment, the steps of obtaining fan fault information based on voiceprint feature pattern information, voiceprint feature parameter information, and optimized feature information, obtaining a predicted life span of the fault source based on the fan fault information, and generating a diagnostic report include: S401, obtaining a corresponding voiceprint feature pattern value according to voiceprint feature pattern information; S402: Obtain corresponding voiceprint feature parameter values ​​according to the voiceprint feature parameter information; S403, obtaining corresponding optimized feature values ​​according to the optimized feature information; S404, obtaining a comprehensive voiceprint feature value based on the voiceprint feature pattern value, the voiceprint feature parameter value, and the optimized feature value; S405: Obtain a voiceprint fault table, wherein the voiceprint fault table includes multiple voiceprint feature comprehensive interval values ​​and fan fault information corresponding to each voiceprint feature comprehensive interval; S406, obtaining a target voiceprint feature comprehensive interval value according to the voiceprint feature comprehensive value; S407: Obtain corresponding fan fault information from the voiceprint table based on the target voiceprint feature comprehensive interval value. S408. Obtain the predicted life of the fault source according to the fan fault information, and generate a diagnosis report.

[0035] As in the above steps S401 to S408, the corresponding voiceprint feature pattern value is extracted from the voiceprint feature pattern information, the corresponding voiceprint feature parameter value is obtained from the voiceprint feature parameter information, and the optimized feature value is extracted from the optimized feature information. The above three types of feature values ​​(voiceprint feature pattern value, voiceprint feature parameter value and optimized value) are fused and calculated to obtain a voiceprint feature comprehensive value. The calculation formula of the voiceprint feature comprehensive value is K=c*w*y, where K represents the voiceprint feature comprehensive value, c represents the voiceprint feature pattern value, w represents the voiceprint feature parameter value, and y represents the optimized feature value. A voiceprint fault table is pre-constructed, which divides a plurality of voiceprint feature comprehensive intervals into the table. Each interval corresponds to specific fan fault information. According to the calculated voiceprint feature comprehensive value , match the target's comprehensive voiceprint feature interval, and obtain the corresponding fan fault information from the voiceprint fault table based on the target interval value, realizing the mapping from feature data to fault category. According to the matched fan fault information, the preset life prediction model is used to calculate the predicted life of the fault source, and automatically generate a diagnostic report, providing detailed fault description, risk warning and maintenance suggestions. Through the fusion calculation of multi-dimensional feature values ​​(mode, parameters, optimization), it can more comprehensively and accurately reflect the actual status of the equipment, thereby improving the accuracy of fault identification and classification. Timely and accurate fault detection and life prediction can effectively prevent unexpected shutdowns and accidents during equipment operation, thereby ensuring the continuity and safety of fan operation, extending the service life of equipment and reducing maintenance costs.

[0036] In a preferred embodiment, the steps of obtaining the predicted life of the fault source based on the wind turbine fault information and generating a diagnostic report include: S4081. Obtain fault source component information of the fan according to the fan fault information, obtain operating state information of the fault source component according to the fault source component information, and obtain a corresponding operating state vector of the fault source component according to the fault source component operating state information; S4082. Obtain a corresponding evaluation duration based on the operating status information of the fault source component, mark the time node when the wind turbine fault information is obtained as the end time of the evaluation period, obtain the start time of the evaluation period based on the evaluation duration, and construct the evaluation period based on the start time and end time of the evaluation period; S4083. Obtain historical sound information and component temperature information of the fault source component within the evaluation period, obtain corresponding multiple historical sound vectors based on the historical sound information, and obtain corresponding multiple component temperature values ​​based on the component temperature information; S4084, obtaining sound weight and component temperature weight; S4085. Obtaining an operating condition value of the fault source component based on multiple historical sound vectors, multiple component temperature values, sound weights, and component temperature weights; S4086. Obtain a life comparison table, wherein the life comparison table includes multiple operating condition interval values ​​and the predicted life of the fault source corresponding to each operating condition interval value; S4087. Obtain a target operating condition interval value based on the operating condition value, obtain the corresponding fault source predicted life from the life comparison table based on the target operating condition interval value, and generate a diagnosis report.

[0037] As in the above steps S4081 to S4087, based on the fan fault information, the fault source component where the abnormality occurs is first determined, such as blades, bearings or gears, and for the determined fault source component, its operating status information is further collected, which includes key indicators such as vibration, noise, and temperature. The collected operating status information is converted into a quantitative vector, and an appropriate evaluation duration is determined based on the operating status information of the fault source component. This duration reflects the time window required to evaluate the dynamic changes in the component status. The time node when the fault information is obtained is used as the end time of the evaluation period, and the start time is obtained by reverse calculation based on the evaluation duration to form a complete evaluation period. During the evaluation period, historical sound information and component temperature information of the fault source component are obtained, and after processing the historical sound data, multiple sound vectors are extracted; at the same time, the temperature data is organized into multiple specific temperature values, and the corresponding weights of the sound data and the component temperature data are determined based on field experience or a preset model. The multiple historical sound vectors and temperature values ​​are weighted and fused according to the corresponding weights to calculate a comprehensive value (operating condition value) reflecting the current operating condition of the fault source component. The calculation formula of the operating condition value is: , where g is the power adjustment coefficient, Expressed as sound weight, It is represented by the component temperature weight, d is represented by the number of multiple historical sound vectors, d = 1, 2, 3...f, It is represented as the dth historical sound vector, a represents the number of the temperature values ​​of multiple components, a=1,2,3…b, It is represented as the temperature value of the ath component. The pre-established life comparison table defines multiple operating condition interval values ​​and specifies the corresponding fault source predicted life for each interval. This table is usually constructed based on historical data, experimental tests or statistical analysis. According to the calculated operating condition value, the target operating condition interval is determined, and the corresponding fault source predicted life is matched from the life comparison table. According to the predicted life of the fault source and related evaluation data, a detailed diagnostic report containing the fault component status, predicted life, risk warning and maintenance recommendations is generated to provide a reference basis for operation and maintenance decisions. The component's operating state vector and historical sound and temperature data can be used to dynamically evaluate the health status of the fault source component, and accurate remaining life prediction can be achieved through the operating condition value and the life comparison table. The comprehensive collection of multi-dimensional data (sound, temperature, operating status) and weighted operating condition values ​​provide solid data support for fault life prediction. The generated diagnostic report can help operation and maintenance personnel accurately judge the severity of the fault and future risks, thereby optimizing maintenance plans.

[0038] Please see the attached Figure 2 As shown, the present invention also provides a fan fault diagnosis system based on voiceprint recognition, which is used in the above-mentioned fan fault diagnosis method based on voiceprint recognition, comprising: The voiceprint module is used to obtain the voiceprint information of the fan operation, obtain the key voiceprint features based on the voiceprint information, obtain the operation mode information of the fan, and obtain the voiceprint feature pattern information based on the key voiceprint features and the operation mode information; The characteristic parameter module is used to obtain the operating parameter data of the fan and obtain the voiceprint characteristic parameter information based on the operating parameter data and key voiceprint features; The optimization feature module is used to obtain the operating status feature data of the fan and the voiceprint feature weight data, and obtain the optimization feature information based on the operating status feature data of the fan and the voiceprint feature weight data; The evaluation module is used to obtain fan fault information based on voiceprint feature pattern information, voiceprint feature parameter information and optimization feature information, obtain the predicted life of the fault source based on the fan fault information, and generate a diagnostic report.

[0039] As mentioned above, the voiceprint module collects the voiceprint information generated during the operation of the fan through sensors installed on or around the fan, and uses signal processing and feature extraction algorithms to extract key voiceprint features from the original sound, while obtaining the operating mode information of the fan, and using this information in combination with key voiceprint features to construct voiceprint feature pattern information that reflects the actual state of the fan under specific working conditions. The feature parameter module is responsible for obtaining various physical parameter data (such as temperature, vibration, pressure, etc.) during the operation of the fan in real time, and using the collected operating parameter data in combination with key voiceprint features to extract voiceprint feature parameter information from multiple angles and across fields, so that the actual operating state of the fan can be more comprehensively described, and the feature module is optimized to obtain feature data reflecting the operating state of the fan and pre-determined voiceprint features. The system combines the operating status data with the voiceprint feature weight to obtain optimized feature information. The evaluation module comprehensively utilizes the voiceprint feature pattern information, feature parameter information and optimized feature information to calculate a comprehensive fault feature value. According to the preset voiceprint fault table, the comprehensive feature value is mapped to the corresponding fan fault information, and the remaining life of the fault source is predicted based on the fault information. Finally, a diagnostic report containing fault description, predicted life, risk warning and maintenance suggestions is automatically generated to provide a basis for fan maintenance decision-making. It can collect and process voiceprint and operating parameter data in real time, and detect minor anomalies in fan operation in time, thereby achieving early warning and reducing the risk of sudden failures. By integrating voiceprint, operating parameters and status feature data, fault diagnosis is made more comprehensive and accurate.

[0040] And, a fan fault diagnosis terminal based on voiceprint recognition, comprising: one or more processors; a storage device having one or more programs stored thereon; When one or more programs are executed by one or more processors, the one or more processors implement a fan fault diagnosis method based on voiceprint recognition.

[0041] The foregoing is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained herein shall, unless otherwise specified or limited, be implemented in accordance with conventional means in the art.

Claims

1. A fan fault diagnosis method based on voiceprint recognition, characterized in that: include: Obtain voiceprint information of the fan operation, obtain key voiceprint features based on the voiceprint information, obtain operation mode information of the fan, and obtain voiceprint feature pattern information based on the key voiceprint features and operation mode information; Obtaining the operating parameter data of the fan, and obtaining voiceprint feature parameter information based on the operating parameter data and key voiceprint features; Obtaining the fan's operating status feature data and voiceprint feature weight data, and obtaining optimized feature information based on the fan's operating status feature data and voiceprint feature weight data; The fan fault information is obtained based on the voiceprint feature pattern information, voiceprint feature parameter information and optimization feature information, and the predicted life of the fault source is obtained based on the fan fault information, and a diagnostic report is generated.

2. The fan fault diagnosis method based on voiceprint recognition according to claim 1 is characterized in that: The steps of obtaining voiceprint information of the fan operation, obtaining key voiceprint features based on the voiceprint information, obtaining operation mode information of the fan, and obtaining voiceprint feature parameter information based on the key voiceprint features and the operation mode information include: Get the voiceprint information of the fan operation; Obtain key voiceprint features based on voiceprint information; Obtain the fan component voiceprint feature vector and the environment voiceprint feature vector based on the key voiceprint features; Get the fan's operating mode information; Acquire a mode compensation table, wherein the mode compensation table includes a plurality of operation mode information and an operation mode compensation value corresponding to each operation mode information; Obtaining target operating mode information according to the operating mode information, and obtaining a corresponding operating mode compensation value from a mode compensation table according to the target operating mode information; The voiceprint characteristic pattern value is obtained according to the voiceprint characteristic vector of the fan component, the environmental voiceprint characteristic vector and the operation mode compensation value, and is marked as voiceprint characteristic pattern information.

3. The fan fault diagnosis method based on voiceprint recognition according to claim 1 is characterized in that: The steps of obtaining the operating parameter data of the fan and obtaining voiceprint feature parameter information according to the operating parameter data and key voiceprint features include: Obtain the operating parameter data of the fan; Acquire corresponding multiple operating parameter information according to the operating parameter data; Obtain multiple operating parameter values ​​in each operating parameter information; Obtain the fan component voiceprint feature vector and the environment voiceprint feature vector based on the key voiceprint features; The voiceprint feature parameter value is obtained according to the multiple operating parameter values ​​in each operating parameter information, the fan component voiceprint feature vector and the environmental voiceprint feature vector, and is marked as voiceprint feature parameter information.

4. The fan fault diagnosis method based on voiceprint recognition according to claim 1 is characterized in that: The steps of obtaining the operating status characteristic data and voiceprint characteristic weight data of the fan, and obtaining optimized characteristic information according to the operating status characteristic data and voiceprint characteristic weight data of the fan include: Obtain the operating status characteristics of the fan and the voiceprint feature weight data; Obtaining an operating status value of the fan according to the operating status characteristics of the fan; Obtaining a voiceprint feature weight value according to the voiceprint feature weight data; The optimized feature value is obtained according to the running state value and the voiceprint feature weight value, and marked as the optimized feature information.

5. The method for diagnosing fan faults based on voiceprint recognition according to claim 4, characterized in that: The step of obtaining the operating status value of the fan according to the operating status characteristic data of the fan includes: Acquiring multiple operating state characteristic values ​​according to the operating state characteristic data of the fan; Obtain a state table, wherein the state table includes a plurality of operating state characteristic interval values ​​and an operating state proportion value corresponding to each operating state characteristic interval value; Obtain a target operating state characteristic interval value according to the operating state characteristic value, and obtain a corresponding operating state weight from the state table according to the target operating state characteristic interval value; The operating state value is obtained according to multiple operating state feature values ​​and corresponding operating state weights.

6. The fan fault diagnosis method based on voiceprint recognition according to claim 4 is characterized in that: The step of obtaining the voiceprint feature weight value according to the voiceprint feature weight data includes: Obtaining the duration of voiceprint changes based on voiceprint feature weight data; Get the standard change duration; Obtain historical voiceprint feature weight value information based on voiceprint feature weight data; Acquire corresponding multiple historical voiceprint feature weight values ​​according to the historical voiceprint feature weight value information; The voiceprint feature weight value is obtained according to multiple historical voiceprint feature weight values, standard change duration, and change duration.

7. The method for diagnosing fan faults based on voiceprint recognition according to claim 1, characterized in that: The steps of obtaining fan fault information according to voiceprint feature pattern information, voiceprint feature parameter information, and optimized feature information, obtaining a predicted life of a fault source according to the fan fault information, and generating a diagnostic report include: Obtaining a corresponding voiceprint feature pattern value according to the voiceprint feature pattern information; Obtaining corresponding voiceprint feature parameter values ​​according to the voiceprint feature parameter information; Obtaining corresponding optimized feature values ​​according to the optimized feature information; Obtaining a comprehensive voiceprint feature value based on the voiceprint feature pattern value, the voiceprint feature parameter value, and the optimized feature value; Obtain a voiceprint fault table, wherein the voiceprint fault table includes multiple voiceprint feature comprehensive interval values ​​and fan fault information corresponding to each voiceprint feature comprehensive interval; Obtaining a target voiceprint feature comprehensive interval value based on the voiceprint feature comprehensive value; Obtain the corresponding fan fault information from the voiceprint table based on the comprehensive interval value of the target voiceprint feature Obtain the predicted life of the fault source based on the fan fault information and generate a diagnostic report.

8. The method for diagnosing fan faults based on voiceprint recognition according to claim 7, characterized in that: The steps for obtaining the predicted life of the fault source based on the fan fault information and generating a diagnostic report include: Obtaining fault source component information of the fan according to the fault information of the fan, obtaining operating state information of the fault source component according to the fault source component information, and obtaining a corresponding operating state vector of the fault source component according to the operating state information of the fault source component; Obtain the corresponding evaluation duration based on the operating status information of the fault source component, mark the time node when the wind turbine fault information is obtained as the end time of the evaluation period, and obtain the start time of the evaluation period based on the evaluation duration, and construct the evaluation period based on the start time and end time of the evaluation period; Obtaining historical sound information and component temperature information of the fault source component during the evaluation period, obtaining corresponding multiple historical sound vectors based on the historical sound information, and obtaining corresponding multiple component temperature values ​​based on the component temperature information; Get the sound weight and component temperature weight; Obtaining an operating condition value of a fault source component according to a plurality of historical sound vectors, a plurality of component temperature values, a sound weight, and a component temperature weight; Obtaining a life comparison table, wherein the life comparison table includes multiple operating condition interval values ​​and a predicted life of a fault source corresponding to each operating condition interval value; The target operating condition interval value is obtained according to the operating condition value, and the corresponding fault source predicted life is obtained from the life comparison table according to the target operating condition interval value, and a diagnosis report is generated.

9. A fan fault diagnosis system based on voiceprint recognition, applied to the fan fault diagnosis method based on voiceprint recognition according to any one of claims 1 to 8, characterized in that: include: The voiceprint module is used to obtain the voiceprint information of the fan operation, obtain the key voiceprint features based on the voiceprint information, obtain the operation mode information of the fan, and obtain the voiceprint feature pattern information based on the key voiceprint features and the operation mode information; The characteristic parameter module is used to obtain the operating parameter data of the fan and obtain the voiceprint characteristic parameter information based on the operating parameter data and key voiceprint features; The optimization feature module is used to obtain the operating status feature data of the fan and the voiceprint feature weight data, and obtain the optimization feature information based on the operating status feature data of the fan and the voiceprint feature weight data; The evaluation module is used to obtain fan fault information based on voiceprint feature pattern information, voiceprint feature parameter information and optimization feature information, obtain the predicted life of the fault source based on the fan fault information, and generate a diagnostic report.

10. A fan fault diagnosis terminal based on voiceprint recognition, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When one or more programs are executed by one or more processors, the one or more processors implement the fan fault diagnosis method based on voiceprint recognition as described in any one of claims 1 to 8.

Images