Intelligent inspection system of gas generator set

Abstract

The invention discloses an intelligent inspection system of a gas generator set, and particularly relates to the field of intelligent inspection, which comprises a multi-dimensional operation parameter real-time acquisition module, a data processing module, an equipment operation evaluation module, a dynamic adjustment module, a fault analysis module, a fault disposal and effect verification module and an equipment health trend prediction module. The multi-dimensional operation parameter real-time acquisition module is used for acquiring key operation parameters in real time in the operation process of the gas generator set by utilizing a multi-parameter sensor array; the data processing module is used for carrying out preprocessing and feature extraction on the collected key operation parameters; the equipment operation evaluation module is used for carrying out operation risk analysis according to the extracted equipment state characteristics and constructing a risk evaluation model; according to the method, comprehensive sensing of the running state of the gas generator set is achieved by deploying the multi-parameter sensor array, multiple dimension parameters are collected, deep feature extraction and analysis are carried out on each parameter, and the one-sidedness of traditional inspection is avoided.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection technology, and more specifically, to an intelligent inspection system for gas generator sets. Background Technology

[0002] In the operation and maintenance of gas generator sets, traditional inspection methods mainly rely on manual on-site inspections, which have many limitations.

[0003] Manual inspections are significantly affected by subjective factors, making it difficult to fully capture subtle anomalies in equipment operation, such as weak vibration changes in the early stages of bearing wear or slow abnormal temperature rise in windings. Furthermore, traditional inspection cycles are fixed and cannot respond in real-time to dynamic changes in equipment operating status. This makes it difficult to detect potential faults such as frequency fluctuations caused by carbon buildup in gas turbine blades or partial discharge in generator stator windings in a timely manner. These faults are often only detected when they have developed to the point of affecting normal unit operation, greatly increasing the risk of unplanned downtime. In addition, some existing automated inspection systems can only monitor conventional parameters such as temperature and pressure at single points, lacking systematic collection and analysis of key characteristic parameters such as vibration spectra and lubricating oil ferrograms. This prevents the construction of a comprehensive equipment health status assessment system, and the data processing methods are simple, often relying on threshold comparison methods, which are difficult to adapt to the nonlinear changes in equipment operating parameters under complex operating conditions. This results in high false alarm and false negative rates, seriously affecting the reliability and effectiveness of inspections.

[0004] Therefore, there is an urgent need for an intelligent inspection system for gas generator sets, which can achieve a more comprehensive perception and accurate assessment of the operating status of gas generator sets through the deployment of multi-dimensional sensor arrays and the integration of intelligent algorithms. Summary of the Invention

[0005] To overcome the aforementioned shortcomings of existing technologies, this invention provides an intelligent inspection system for gas generator sets. By extracting abnormal equipment features and analyzing fault risks during the inspection process, and making refined and dynamic adjustments accordingly, it maximizes both equipment operational stability and inspection efficiency. Simultaneously, in the fault location phase, multi-source data fusion technology is used to achieve precise fault source localization, effectively solving the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent inspection system for gas generator sets, comprising: Multi-dimensional operating parameter real-time acquisition module: Utilizes a multi-parameter sensor array to acquire key operating parameters in real time during the operation of the gas generator set; Data processing module: preprocesses and extracts features from the collected key operating parameters; Equipment operation assessment module: Based on the extracted equipment status characteristics, it performs operational risk analysis and constructs a risk assessment model; Dynamic adjustment module: Dynamically adjusts the inspection strategy based on the results of comprehensive risk analysis; Fault Analysis Module: When identifying abnormal risks in equipment, it uses multi-source data fusion to locate the fault; Fault handling and effectiveness verification module: Targeted handling is carried out based on the fault location and type identification results; Equipment health trend prediction module: Stores the parameters and analysis results collected from each inspection into the database to build equipment health records; uses historical data to establish a trend prediction model, uses the exponential smoothing method to predict the changing trends of key parameters, and formulates maintenance plans in advance based on the prediction results to achieve preventive maintenance of equipment.

[0007] Preferably, the key operating parameters specifically include vibration parameters, temperature parameters, oil parameters, gas parameters, and pressure parameters; the vibration parameters are obtained by acquiring vibration acceleration time-domain signals in the 10–5000 Hz frequency band using vibration acceleration sensors arranged in the bearing housing and gearbox; the temperature parameters are obtained by acquiring the surface temperature of the generator stator winding and gas turbine cylinder components using an infrared temperature sensor; the oil parameters are obtained by detecting the concentration of particles larger than 5 μm in the lubricating oil using an oil particle size sensor; and the gas parameters are obtained by monitoring the exhaust gas from the combustion chamber using a gas composition analyzer. CO concentration; the pressure parameter: the gas intake pressure is collected by a pressure sensor.

[0008] Preferably, the preprocessing and feature extraction include vibration signal processing, temperature feature extraction, and oil feature extraction; the vibration signal processing involves performing a fast Fourier transform on the vibration acceleration time-domain signal to obtain a frequency domain spectrum, and extracting the amplitudes A1, A2, and A3 of the 1x, 2x, and 3x harmonic components, as well as the energy proportion in the high-frequency band of 1000-5000Hz. ,in, The frequency domain signal amplitude is represented; the temperature feature extraction involves calculating the stator winding temperature change rate ΔT / Δt and the standard deviation σT of the cylinder block temperature field distribution, reflecting the uniformity of temperature distribution. ,in, This represents the temperature at the i-th measuring point. The average temperature is represented by , and n represents the number of temperature measurement points; the oil feature extraction involves calculating the rate of change of metal particle concentration in the lubricating oil ΔC / Δt and the skewness coefficient of the particle size distribution. ,in, Indicates particle size, 's' represents the average size, and 's' represents the standard deviation.

[0009] Preferably, the risk assessment model includes a vibration risk model, a temperature risk model, an oil fluid risk model, and a comprehensive risk model; the vibration risk model is used to calculate the vibration risk coefficient: the amplitude of the comprehensive frequency doubling component and the proportion of high-frequency energy , where , and respectively represent the safety thresholds of each parameter; the temperature risk model is used to calculate the temperature risk coefficient: combining the temperature change rate and the standard deviation of the temperature field , where and represent its safety threshold; the oil fluid risk model is used to calculate the oil fluid risk coefficient: based on the particle concentration change rate and the skewness coefficient , where represents the safety threshold; the comprehensive risk model is used to obtain the comprehensive risk index of the equipment by integrating each risk coefficient .

[0010] Preferably, the dynamic adjustment method is: when RI ≤ 0.4, it is judged as a low-risk state, and the conventional inspection mode is adopted, and the inspection cycle is set to 8h; when 0.4 < RI ≤ 0.7, it is judged as a medium-risk state, the encrypted inspection mode is started, the inspection cycle is shortened to 2h, and the oil fluid sampling frequency is increased; when RI > 0.7, it is judged as a high-risk state, real-time online monitoring is immediately triggered, and the fault warning process is started at the same time.

[0011] Preferably, the fault location includes vibration source location and fault type identification; the vibration source location: based on the spatial coordinates of each vibration sensor and the time difference of signal arrival, the time difference location method is used to calculate the vibration source coordinates , , where and represent the sensor coordinates, v represents the vibration wave propagation speed, represents the signal arrival time; the fault type identification: construct a fault feature library, match the extracted vibration frequency domain features, temperature distribution features with the fault feature library, and use the Euclidean distance to measure the similarity , where represents the measured feature value, represents the corresponding value in the fault feature library, and m represents the feature dimension.

[0012] Preferably, the targeted disposal includes the disposal of mechanical component loosening faults, the disposal of lubricating oil deterioration faults, and the recalculation of the comprehensive risk index; the disposal of mechanical component loosening faults: calculate the optimal tightening torque ; the disposal of lubricating oil deterioration faults: calculate the oil change cycle according to the oil fluid particle size and metal content , where Indicates the reference oil change interval. , The coefficient represents an empirical factor, C represents particle concentration, and S represents metal content; the comprehensive risk index is recalculated: after treatment, operating parameters are collected again, and the comprehensive risk index after treatment is calculated. ,like If the value is less than 0.4, the treatment is deemed effective; otherwise, the treatment plan is adjusted.

[0013] Preferably, the trend prediction model includes temperature trend prediction and vibration trend prediction; the temperature trend prediction: ,in , Represents the smoothing coefficient. The temperature at time t represents the vibration trend prediction. Where k represents the trend coefficient and n represents the number of historical data points.

[0014] The technical effects and advantages of this invention are as follows: 1. This invention achieves comprehensive perception of the operating status of gas generator sets by deploying a multi-parameter sensor array. The collected parameters cover multiple dimensions, including vibration, temperature, oil, and gas. Furthermore, it performs in-depth feature extraction and analysis on each parameter, avoiding the one-sidedness of traditional inspections. By constructing a comprehensive risk assessment model and dynamic inspection strategy, it achieves optimized allocation of inspection resources. In low-risk conditions, it reduces unnecessary inspection frequency, while in high-risk conditions, it increases monitoring density. On the one hand, it can maximize the satisfaction of equipment health monitoring needs, and on the other hand, it can avoid the interference of frequent inspections on unit operation, which is conducive to reducing operation and maintenance costs while ensuring safe equipment operation. 2. This invention employs multi-source data fusion and spatiotemporal positioning technology in the fault location stage. Based on the vibration time difference positioning method and fault feature matching algorithm, it achieves accurate location and type identification of fault sources, and is not limited to single parameter analysis. In some cases, it can complete fault location during unit operation, thereby ensuring the continuity of unit operation to a certain extent. At the same time, it can better control the development trend of faults, reduce losses caused by the expansion of faults, and also avoid the waste of resources caused by blind shutdown for maintenance, thus reducing maintenance costs. 3. This invention establishes an inspection data archiving and health trend prediction model, using historical data to scientifically predict the changing trends of key equipment parameters, thus achieving a shift from passive maintenance to proactive prevention. It can detect potential equipment failures in advance, rationally arrange maintenance plans, reduce unplanned downtime, and improve unit operating efficiency. Simultaneously, the maintenance strategies formulated based on the prediction results are more precise, avoiding over-maintenance or under-maintenance, achieving the goals of improving equipment reliability, reducing operating costs, and promoting environmental protection. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] As attached Figure 1 The intelligent inspection system for a gas generator set shown includes a multi-dimensional operating parameter real-time acquisition module, a data processing module, an equipment operation evaluation module, a dynamic adjustment module, a fault analysis module, a fault handling and effect verification module, and an equipment health trend prediction module.

[0018] The multi-dimensional operating parameter real-time acquisition module: uses a multi-parameter sensor array to collect key operating parameters in real time during the operation of the gas generator set; Specifically, the key operating parameters include vibration parameters, temperature parameters, oil parameters, gas parameters, and pressure parameters. The vibration parameters are acquired by using vibration acceleration sensors located in the bearing housing and gearbox to collect time-domain vibration acceleration signals within the 10–5000 Hz frequency band. The temperature parameters are obtained using infrared temperature sensors to acquire the surface temperature of the generator stator windings and gas turbine cylinder components. The oil parameters are detected by using an oil particle size sensor to measure the concentration of particles larger than 5 μm in the lubricating oil. The gas parameters are monitored using a gas composition analyzer to measure the gas composition in the combustion chamber exhaust. CO concentration; the pressure parameter: the gas intake pressure is collected by a pressure sensor.

[0019] The data processing module preprocesses and extracts features from the collected key operating parameters; Specifically, it should be noted that the preprocessing and feature extraction include vibration signal processing, temperature feature extraction, and oil feature extraction; the vibration signal processing involves performing a fast Fourier transform on the vibration acceleration time-domain signal to obtain a frequency domain spectrum, and extracting the amplitudes A1, A2, and A3 of the 1x, 2x, and 3x harmonic components, as well as the energy proportion in the high-frequency band of 1000-5000Hz. ,in, The frequency domain signal amplitude is represented; the temperature feature extraction involves calculating the stator winding temperature change rate ΔT / Δt and the standard deviation σT of the cylinder block temperature field distribution, reflecting the uniformity of temperature distribution. ,in, This represents the temperature at the i-th measuring point. represents the average temperature, and n represents the number of temperature measurement points; for the extraction of oil fluid characteristics: calculate the change rate ΔC / Δt of the metal particle concentration in the lubricating oil and the skewness coefficient of the particle size distribution , where represents the particle size represents the average size, and s represents the standard deviation

[0020] The equipment operation evaluation module: conducts operation risk analysis based on the extracted equipment status characteristics and constructs a risk assessment model Specifically, it should be noted that: the risk assessment model includes a vibration risk model, a temperature risk model, an oil fluid risk model, and a comprehensive risk model; the vibration risk model is used to calculate the vibration risk coefficient: combining the amplitude of the comprehensive multiple-frequency component and the proportion of high-frequency energy , where , and respectively represent the safety thresholds of each parameter; the temperature risk model is used to calculate the temperature risk coefficient: combining the temperature change rate and the standard deviation of the temperature field , where and represent its safety threshold; the oil fluid risk model is used to calculate the oil fluid risk coefficient: based on the change rate of particle concentration and the skewness coefficient , where represents the safety threshold; the comprehensive risk model is used to obtain the equipment comprehensive risk index by integrating each risk coefficient .

[0021] The dynamic adjustment module: dynamically adjusts the patrol strategy according to the result of the comprehensive risk analysis Specifically, it should be noted that: the dynamic adjustment method is as follows: when RI≤0.4, it is judged as a low-risk state, and the conventional patrol mode is adopted, and the patrol cycle is set to 8h; when 0.4 < RI≤0.7, it is judged as a medium-risk state, the encrypted patrol mode is started, the patrol cycle is shortened to 2h, and the oil fluid sampling frequency is increased; when RI>0.7, it is judged as a high-risk state, real-time online monitoring is immediately triggered, and the fault warning process is started

[0022] The fault analysis module: when identifying that the equipment has abnormal risks, uses multi-source data fusion for fault location Specifically, it should be noted that: the fault location includes vibration source location and fault type identification; the vibration source location: based on the spatial coordinates of each vibration sensor and the time difference of signal arrival, uses the time difference location method to calculate the vibration source coordinates , , where and The sensor coordinates are represented by v, and the vibration wave propagation speed is represented by v. Indicates the signal arrival time; the fault type identification: constructs a fault feature database, matches the extracted vibration frequency domain features and temperature distribution features with the fault feature database, and uses Euclidean distance to measure similarity. ,in, Represents the measured characteristic value. This represents the corresponding value in the fault feature library, and m represents the feature dimension.

[0023] The fault handling and effect verification module: performs targeted handling based on the fault location and type identification results; Specifically, it should be noted that the targeted measures include addressing loose mechanical components, addressing deteriorated lubricating oil, and recalculating the comprehensive risk index; the addressing of loose mechanical components involves calculating the optimal tightening torque. The handling of lubricating oil deterioration faults involves calculating the oil change cycle based on oil particle size and metal content. ,in, Indicates the reference oil change interval. , The coefficient represents an empirical factor, C represents particle concentration, and S represents metal content; the comprehensive risk index is recalculated: after treatment, operating parameters are collected again, and the comprehensive risk index after treatment is calculated. ,like If the value is less than 0.4, the treatment is deemed effective; otherwise, the treatment plan is adjusted.

[0024] The equipment health trend prediction module stores the parameters and analysis results collected during each inspection into a database to build an equipment health record; it uses historical data to establish a trend prediction model, uses exponential smoothing to predict the changing trends of key parameters, and formulates maintenance plans in advance based on the prediction results to achieve preventive maintenance of the equipment.

[0025] Specifically, it should be noted that the trend prediction model includes temperature trend prediction and vibration trend prediction; the temperature trend prediction is as follows: ,in , Represents the smoothing coefficient. The temperature at time t represents the vibration trend prediction. Where k represents the trend coefficient and n represents the number of historical data points.

[0026] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent inspection system for a gas generator set, characterized in that, Including: Multi-dimensional operating parameter real-time acquisition module: Utilize a multi-parameter sensor array to real-time acquire key operating parameters during the operation of a gas generator set; Data processing module: Preprocess and extract features from the acquired key operating parameters; Equipment operation evaluation module: Conduct operation risk analysis based on the extracted equipment status features and construct a risk assessment model; Dynamic adjustment module: Dynamically adjust the patrol inspection strategy according to the results of comprehensive risk analysis; Fault analysis module: When identifying abnormal risks in the equipment, use multi-source data fusion for fault location; Fault handling and effect verification module: Conduct targeted handling according to the fault location and type identification results; Equipment health trend prediction module: Store the parameters and analysis results collected during each patrol inspection into a database to construct an equipment health file; Establish a trend prediction model using historical data, adopt the exponential smoothing method to predict the change trend of key parameters, and formulate a maintenance plan in advance through the prediction results to achieve preventive maintenance of the equipment.

2. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The key operating parameters specifically include vibration parameters, temperature parameters, oil parameters, gas parameters, and pressure parameters. The vibration parameters are obtained by collecting time-domain vibration acceleration signals in the 10–5000 Hz frequency band using vibration acceleration sensors located in the bearing housing and gearbox. The temperature parameters are obtained by acquiring the surface temperature of the generator stator windings and gas turbine cylinder components using an infrared temperature sensor. The oil parameters are detected by using an oil particle size sensor to measure the concentration of particles larger than 5 μm in the lubricating oil. The gas parameters are monitored by using a gas composition analyzer to measure the gas composition in the combustion chamber exhaust. CO concentration; the pressure parameter: the gas intake pressure is collected by a pressure sensor.

3. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The preprocessing and feature extraction include vibration signal processing, temperature feature extraction, and oil feature extraction; the vibration signal processing involves performing a fast Fourier transform on the vibration acceleration time-domain signal to obtain a frequency domain spectrum, and extracting the amplitudes A1, A2, and A3 of the 1x, 2x, and 3x harmonic components, as well as the energy percentage in the high-frequency band of 1000-5000Hz. ,in, The frequency domain signal amplitude is represented; the temperature feature extraction involves calculating the stator winding temperature change rate ΔT / Δt and the standard deviation σT of the cylinder block temperature field distribution, reflecting the uniformity of temperature distribution. ,in, This represents the temperature at the i-th measuring point. The average temperature is represented by , and n represents the number of temperature measurement points; the oil feature extraction involves calculating the rate of change of metal particle concentration in the lubricating oil ΔC / Δt and the skewness coefficient of the particle size distribution. ,in, Indicates particle size, 's' represents the average size, and 's' represents the standard deviation.

4. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The risk assessment model includes a vibration risk model, a temperature risk model, an oil risk model, and a comprehensive risk model; the vibration risk model is used to calculate the vibration risk coefficient by combining the amplitude of the harmonic component and the proportion of high-frequency energy. ,in, , as well as These represent the safety threshold values ​​for each parameter; the temperature risk model is used to calculate the temperature risk coefficient, combining the rate of temperature change and the standard deviation of the temperature field. ,in, and This indicates its safety threshold; the oil risk model is used to calculate the oil risk coefficient: based on the particle concentration change rate and skewness coefficient. ,in, This represents the safety threshold; the comprehensive risk model is used to synthesize various risk coefficients to obtain the comprehensive equipment risk index. .

5. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The dynamic adjustment method is as follows: When RI ≤ 0.4, it is judged as a low-risk state, and the conventional patrol inspection mode is adopted, with the patrol inspection cycle set to 8h; when 0.4 < RI ≤ 0.7, it is judged as a medium-risk state, the encrypted patrol inspection mode is started, the patrol inspection cycle is shortened to 2h, and the oil sample frequency is increased; when RI > 0.7, it is judged as a high-risk state, real-time online monitoring is immediately triggered, and at the same time, the fault warning process is started.

6. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The fault location includes vibration source location and fault type identification; the vibration source location is calculated using the time difference of arrival method based on the spatial coordinates of each vibration sensor and the signal arrival time difference. , ,in, and The sensor coordinates are represented by v, and the vibration wave propagation speed is represented by v. Indicates the signal arrival time; the fault type identification: constructs a fault feature database, matches the extracted vibration frequency domain features and temperature distribution features with the fault feature database, and uses Euclidean distance to measure similarity. ,in, Represents the measured characteristic value. This represents the corresponding value in the fault feature library, and m represents the feature dimension.

7. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The targeted measures include addressing loose mechanical components, resolving lubricating oil deterioration, and recalculating the comprehensive risk index; the addressing of loose mechanical components involves calculating the optimal tightening torque. The handling of lubricating oil deterioration faults involves calculating the oil change cycle based on oil particle size and metal content. ,in, Indicates the reference oil change interval. , The coefficient represents an empirical factor, C represents particle concentration, and S represents metal content; the comprehensive risk index is recalculated: after treatment, operating parameters are collected again, and the comprehensive risk index after treatment is calculated. ,like If the value is less than 0.4, the treatment is deemed effective; otherwise, the treatment plan is adjusted.

8. The intelligent inspection system for a gas generator set according to claim 1, characterized in that: The trend prediction model includes temperature trend prediction and vibration trend prediction; the temperature prediction: ,in , Represents the smoothing coefficient. The temperature at time t represents the vibration trend prediction. Where k represents the trend coefficient and n represents the number of historical data points.

Images