A predictive detection method and system for gas turbines

Abstract

This invention provides a predictive detection method and system for gas turbines, belonging to the field of gas turbine technology. The predictive detection method for gas turbines provided by this invention includes the following steps: collecting electricity demand information data and gas turbine operating data to generate an adjustment plan; adjusting the gas turbine operating system according to the adjustment plan; determining a safety score based on the gas turbine operating data; and importing the gas turbine operating data into a simulation system in real time. This invention locates abnormal noises by sounding the audio source, thereby detecting the location of the abnormal noise from the sensor position. It determines the location of the abnormal noise based on the simulated image and operating sound, and changes the flashing position of a light to the location of the abnormal noise, alerting the operator that the abnormal noise has occurred at that point, thus shortening the sensor detection time and improving detection response capability.

Description

Technical Field

[0001] This invention belongs to the field of gas turbine technology and relates to predictive detection methods, specifically a predictive detection method and system for gas turbines. Background Technology

[0002] Gas turbine power generation is a highly efficient and clean method of power generation. It achieves energy conversion through a three-step cycle: air compression by a compressor, fuel combustion in a combustion chamber, and power expansion by a turbine. Its core advantages are rapid start-up, requiring only 10-30 minutes for cold start, making it suitable for grid peak shaving. Its combined cycle efficiency exceeds 60%, far surpassing that of coal-fired power units. It also boasts strong fuel adaptability, being compatible with low-carbon fuels such as natural gas and hydrogen. Operation and maintenance require a focus on the lubrication system (oil acid value / particle size monitoring) and vibration fault diagnosis.

[0003] With the widespread application of gas turbines in aerospace, power, and marine industries, playing a crucial role as core equipment in these fields, increasing attention has been paid to their healthy operational status. Gas turbines often operate for extended periods in harsh environments with high temperatures, high pressures, and high speeds. Furthermore, the significant load variations that occur with changes in their operating conditions make them susceptible to abnormal operating states. The complex structure of gas turbines further complicates the potential for such abnormalities. When a gas turbine malfunctions, it severely impacts the unit's performance, hinders normal operation, and leads to substantial economic losses or even major accidents. Therefore, conducting health status diagnostics and analysis of gas turbines is crucial for improving operational reliability, reducing abnormal conditions, minimizing maintenance costs, and mitigating economic losses caused by these abnormalities.

[0004] Existing technology CN110879151A discloses a remote monitoring and diagnostic system and method for gas turbines based on operational big data, including the following steps: selecting the gas turbine equipment to be monitored; collecting real-time operating data of the gas turbine from the monitored object using a real-time data acquisition unit; creating a personalized dynamic model for a specific gas turbine, and adjusting it by combining measurement data and the correlation decoupling analysis of the physical parameters based on the expected numerical deviation analysis, to reflect the baseline operating state of the gas turbine; using a real-time database management unit for data transmission, storage, and massive data management, and processing the gas turbine operating data through an optimized visualization interface; conducting gas turbine operating status monitoring and diagnosis, and performing analysis and judgment.

[0005] Existing technology CN119533949A discloses a multi-sensor-based gas turbine fault diagnosis system and control method, including a data acquisition device, a data core service system, and a monitoring center. The sensor nodes of the data acquisition device collect dynamic signals from multiple parts of the gas turbine. The data acquisition module of the data acquisition device connects to the sensor nodes, receives the dynamic signals, and transmits them to the data receiving module. The data receiving module transmits the dynamic signals to the data processing module. The data processing module extracts various feature parameters using a feature parameter extraction algorithm, performs standardized score calculations on each real-time acquired dynamic signal feature data point, and sets a single scoring threshold corresponding to each feature parameter. The standardized scores of each feature are fused to obtain a comprehensive anomaly scoring formula, and a comprehensive scoring threshold is set. When the feature data point of the current dynamic signal exceeds the comprehensive scoring threshold or the single scoring threshold corresponding to each feature parameter, the data processing module sends an anomaly signal to the alarm module.

[0006] Existing gas turbines are connected to generators to drive electricity generation. When there is a large demand for electricity, the gas turbines need to operate at high power, which can easily lead to mechanical problems. Existing gas turbine early warning systems only make predictions based on sensor data and trends. However, gas turbines are large and the sensor detection range is limited. Therefore, it takes time for the sensors to collect complete data. During this time, the internal components of the gas turbine may have changed from minor damage to severe damage. Thus, by the time it is determined that adjustment or shutdown is needed, the machine has already suffered irreversible damage. Furthermore, due to the limited sensor installation location and the high internal operating temperature of the gas turbine, it is impossible to install cameras, so it is impossible to confirm or predict what kind of problems will occur in the components in the first instance. Summary of the Invention

[0007] This invention addresses the problems of long sensor detection times and inability to obtain information about the internal operation of gas turbines, leading to untimely predictions, in existing technologies. It provides a predictive detection method and system for gas turbines. The predictive detection method includes the following steps: collecting electricity demand information and gas turbine operating data to generate an adjustment plan; adjusting the gas turbine operating system according to the adjustment plan; determining a safety score based on the gas turbine operating data; and importing the gas turbine operating data into a simulation system in real time. This invention locates abnormal noises by sound, starting from the sensor position and detecting the location of the abnormal noise. This determines the rotation angle of the core component, thus inferring where the abnormal noise occurs. The location of the abnormal noise is determined based on the simulated image and operating sound, and the flashing light position is changed to the location of the abnormal noise to alert the operator. By combining operating data with the location of the abnormal noise, future operational risks can be accurately predicted, thereby improving subsequent operational efficiency.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] On one hand, the present invention provides a predictive detection method for gas turbines, comprising the following steps:

[0010] Collect electricity demand information and gas turbine operation data to generate an adjustment plan, and adjust the gas turbine operation system according to the adjustment plan;

[0011] A safety table is set up, and a safety score is determined based on the gas turbine operating data.

[0012] The gas turbine operating data is imported into the simulation system in real time to generate a simulation image;

[0013] Collect the operating sound of the gas turbine and separate the operating sound of several components by noise reduction;

[0014] The simulated images are combined with the operating sounds of several components and historical data to provide early warning analysis of operating trends, thereby generating an analysis report.

[0015] The audio and volume of the sounds from the operation of several components are converted into lighting patterns, and the warning lights execute the lighting patterns in real time.

[0016] Based on the analysis report and changes in warning lights combined with historical data, the adjustment plan was revised again.

[0017] If the revised adjustment plan fails to meet the operational safety requirements, the analysis report and lighting change plan will be sent to personnel for remote control.

[0018] Preferably, the predictive detection method further includes setting up and detecting a cooling system and a lubrication system, wherein the cooling system and lubrication system are used to assist the operation of the gas turbine;

[0019] Among these methods, the operation status of the lubrication system in the gas turbine is observed, and the shaft motion of the gas turbine is photographed and uploaded to the simulation system to assist in the optimization and analysis report.

[0020] Preferably, the predictive detection method further includes setting a lubricating oil viscosity reference table and adjusting the lubricating oil in the lubrication system according to the lubricating oil viscosity reference table;

[0021] An image of the adjusted lubricating oil is captured, and the colors in the image are compared with the lubricating oil colors in the lubricating oil viscosity chart to determine the current lubricating oil viscosity.

[0022] Preferably, the prediction detection method further includes the following steps:

[0023] The lubricating oil image is pixelated to identify impurities within the lubricating oil and to estimate the quantity of impurities.

[0024] Based on the estimated amount of impurities, combined with simulated images and component operating sounds, the problematic components in the gas turbine were identified and their locations determined.

[0025] The result was determined by combining the location of the problematic component with historical data and the sound of the component operating.

[0026] Preferably, the predictive detection method further includes setting up warning lights around the gas turbine, and the number of rings of the warning lights is the same as the number of core components inside the gas turbine;

[0027] The method for changing the warning light includes, after determining the location of the problematic component, the warning light at that location flashes as a whole;

[0028] Preferably, the overall flashing time is at least 3 seconds;

[0029] The abnormal noise is detected at a specific location of the problematic component by the sound of the component's operation. The warning light is then activated to flash locally at the area where the abnormal noise occurs, and the sound is recorded and photographed.

[0030] Preferably, the local flickering time is at least 3 seconds;

[0031] Preferably, the historical data contains the sound of each component operating individually and the corresponding sound results;

[0032] Sound is collected at the location of each core component, and the sound at the location of a single core component in real time is obtained by separating the wavelength and frequency of the sound.

[0033] The sound at the location of a single core component during real-time operation is compared with the sound of the component operating individually in the historical data to obtain the corresponding sound result;

[0034] The audio and volume of the sound result are converted into the flashing amplitude of the light to warn the operator.

[0035] Preferably, pigments of different colors are added to lubricating oils of different grades;

[0036] The lubricating oil viscosity reference table includes different color images corresponding to lubricating oils of different viscosities;

[0037] The historical data stores solutions corresponding to the sound results, and the solutions include changing the viscosity of the lubricating oil.

[0038] The color of the lubricating oil is determined based on the captured image. When it is necessary to change the viscosity of the lubricating oil, the concentration ratio is obtained by adjusting the color based on the determined color.

[0039] A portion of the lubricating oil inside the gas turbine is discharged based on the concentration ratio, and new lubricating oil is added to form a lubricating oil with the appropriate viscosity.

[0040] Preferably, the peak electricity consumption month is determined by the historical data, and the peak electricity consumption day within the peak electricity consumption month is determined by collecting holiday information or weather information; the gas turbine is maintained in advance based on the peak electricity consumption day.

[0041] Preferably, the prediction detection method further includes combining the location of the abnormal noise area with the simulated image, and marking the abnormal noise sound that appears at each time point at the location of the abnormal noise area to generate an anomaly file;

[0042] The abnormal files and analysis reports are sent to personnel so that they can quickly analyze whether there are any operational risks if the gas turbine continues to operate.

[0043] On the other hand, the present invention provides a system for predictive detection of gas turbines, which uses the above-mentioned predictive detection method for detection and includes the following modules:

[0044] The prediction module collects peak electricity consumption and weather information to determine electricity demand, thereby generating adjustment plans for the maintenance of the gas turbine in advance;

[0045] The monitoring module detects mechanical, temperature, pressure, vibration, lubrication, and cooling data of the gas turbine during operation and aggregates them to generate simulated images. By collecting the sound of the gas turbine during operation, the module combines the sound with the simulated images to predict operational risks and generates new adjustment plans based on the operational risks and historical data.

[0046] The analysis module is used to separate and reduce noise during the operation of the gas turbine, obtaining the individual operating sounds of each component in the development process. By locating the individual operating sounds of each component and combining them with historical data, the system determines the location of the problem in each component and generates an analysis report by combining the simulation images.

[0047] The communication module transmits the analysis report to the human operator, who then assesses operational risks and further optimizes and adjusts the strategy.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The predictive detection method and system for this gas turbine locates abnormal noises by using the audio of the noise source, and then detects the location of the abnormal noise from the sensor position to determine the rotation angle of the core component. This allows the system to infer where the abnormal noise occurs in the core component. Based on the simulated image and the sound of operation, the system determines the location of the abnormal noise and changes the flashing position of the indicator light to the location of the abnormal noise to alert the operator that the abnormal noise has occurred. By combining the operating data with the location of the abnormal noise, the system can accurately predict future operational risks, thereby improving subsequent operational performance.

[0050] 2. The predictive detection method and system for this gas turbine uses a camera to capture images of lubricating oil with rapidly changing colors. Due to the friction between mechanical parts, the lubricating oil can be quickly stirred and mixed. By observing the color change trend in the images captured by the camera, it is possible to quickly determine whether the required viscosity grade of lubricating oil has been achieved, thus allowing the machine to adjust the lubricating oil without stopping the machine in emergency situations.

[0051] 3. The predictive detection method and system for this gas turbine uses flashing lights at the determined location of abnormal noise to allow experienced personnel to predict the operational problems that will occur at that location. This allows operators to make intuitive and immediate judgments, rather than waiting for sensors to detect for a long time before providing data. This cooperation with operators improves the predictive effect. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the architecture of the gas turbine predictive detection method of the present invention. Detailed Implementation

[0053] 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.

[0054] Example 1: A predictive detection method for gas turbines

[0055] Figure 1 A schematic diagram of the gas turbine predictive detection method of the present invention is shown. This embodiment provides a predictive detection method for gas turbines, including the following steps:

[0056] Collect electricity demand information and gas turbine operation data to generate adjustment plans, and adjust the gas turbine operation system according to the adjustment plans;

[0057] Establish an operational safety table and determine safety scores based on gas turbine operating data;

[0058] The gas turbine operating data is imported into the simulation system in real time to generate simulated images;

[0059] Collect the operating sound of the gas turbine and separate the operating sound of several components by noise reduction;

[0060] The simulation images are combined with the operating sounds of several components and historical data to provide early warning analysis of operating trends, thereby generating an analysis report.

[0061] The audio and volume of the sounds from the operation of several components are converted into lighting patterns, and the warning lights execute the lighting patterns in real time.

[0062] Based on the analysis report and changes in warning lights combined with historical data, the plan was revised and adjusted again.

[0063] If the revised adjustment plan fails to meet the operational safety requirements, the analysis report and lighting change plan will be sent to personnel for remote control.

[0064] Electricity demand information data is used to determine the peak electricity consumption months through historical data, and to collect holiday or weather information to determine the peak electricity consumption days within the peak electricity consumption months. Based on the peak electricity consumption days, gas turbines are maintained in advance.

[0065] Gas turbine operating data includes mechanical data, temperature data, pressure data, vibration data, lubrication data, and cooling data;

[0066] The above data was collected using sensors, and the sensors selected included:

[0067] Speed ​​sensors monitor turbine rotor speed to prevent overspeed faults; commonly used are magnetoelectric or eddy current sensors.

[0068] Temperature sensors, thermocouples, detect the temperature in high-temperature areas (>700℃) such as combustion chamber and turbine exhaust; resistance temperature detectors (RTDs) / NTC sensors monitor medium and low temperature areas such as intake air temperature and bearing temperature (NTC sensors have fast response and high temperature resistance).

[0069] Pressure sensor, static pressure monitoring: measures compressor intake / exhaust pressure, gas pressure, etc., commonly using high temperature pressure sensors (Kulite XTE-190 series, temperature resistant up to 273℃);

[0070] Dynamic pressure monitoring captures combustion chamber pressure pulsations (IMI dynamic pressure sensor, supports remote and local installation);

[0071] Vibration sensors and accelerometers detect rotor vibration, and displacement sensors monitor axial displacement to prevent mechanical imbalance or bearing failure.

[0072] Flame detectors monitor the flame status in the combustion chamber in real time to prevent fuel buildup and explosion caused by flameout;

[0073] Combustion dynamic sensor, dedicated dynamic pressure sensor (PCB OTIS series) is directly installed in the combustion chamber to monitor combustion instability and prevent component damage;

[0074] Arrayed eddy current sensor group for non-destructive testing of high-temperature blade cracks and creep holes (custom scanner);

[0075] The borescope measuring points are located at the compressor, turbine, and other positions for periodic internal inspection.

[0076] The safety table includes the threshold range set for the data collected by the aforementioned sensors. If the data collected by the aforementioned sensors is lower than the set threshold range, 6 points will be deducted.

[0077] If the overall score in the safety operation table is below 80 points, the location detected by the sensor is marked and included in the analysis report. Subsequently, mechanical data, temperature data, pressure data, vibration data, lubrication data, and cooling data are uploaded to the simulation system.

[0078] The gas turbine simulation system includes the gSIM full-condition simulation system and the T-MATS open-source toolkit, etc.

[0079] A simulated view of the internal operation of a gas turbine is obtained through a gas turbine simulation system, thereby generating a simulated image;

[0080] Several sound-receiving devices are installed on the outer surface of the gas turbine. The number of sound-receiving devices is the same as the number of core components of the gas turbine. For example, sound-receiving devices are installed at corresponding positions such as the power output end, cylinder, rotor, compressor, combustion chamber, burner, and turbine. The sound of the core components of the gas turbine running is obtained through the sound-receiving devices.

[0081] Since sound is composed of frequency, amplitude, wavelength, waveform, loudness, timbre, etc., and the historical data includes the sound of each component running and the corresponding running state, the sound collected by each sound receiving device is compared and analyzed with the sound of the component running in the historical data to determine the current running state of the component.

[0082] Furthermore, in order to quickly locate the corresponding component's operating sound from the chaotic sounds, the radio device is adjusted to only receive at least one or more of the component's frequency, amplitude, wavelength, waveform, loudness, timbre, and noise, thereby quickly confirming the component's operating status. For example,

[0083] The compressor operates in the range of 500-3000Hz. The discrete noise is caused by the periodic cutting of airflow by the blades. The energy is concentrated in the mid-to-high frequency range. When running at full load, it accounts for 35% of the noise of the whole machine. The peak frequency of a single-stage blade can reach more than 110dB.

[0084] The combustion chamber has low-frequency characteristics, with a pressure pulsation frequency of <200Hz and an amplitude related to fuel injection and flame stability. High-frequency characteristics include combustion instability noise covering 200-2000Hz, and occasional high-frequency knock noise with an instantaneous sound pressure level that can increase by 15-20dB.

[0085] Turbine flow stripping results in a wideband noise of 100Hz-8kHz, with energy peaks concentrated in 1-4kHz.

[0086] It should be noted that the historical data contains the operating sounds of the components at different power levels;

[0087] In summary, the operating status of a component can be quickly determined by combining the initial detection data obtained from the sensor with the sound obtained from the recording device.

[0088] The historical data of past operation trends is retrieved. The operation status, simulated image, and operation sound are used to filter the operation trends that match the operation status, simulated image, and operation sound from the historical data using a calculation formula. A similarity threshold is set, and the operation trends obtained by the calculation and filtering are not lower than the similarity threshold. An analysis report is then generated based on the operation trends obtained by the calculation and filtering.

[0089] It should be noted that the calculation and screening formula uses K-means and DBSCAN algorithms to select one of the several trends in the historical data that match the data, thus obtaining accurate prediction results;

[0090] Select one of the following data from the operating sound of several components: frequency data, amplitude data, wavelength data, waveform data, loudness data, and timbre data, and convert it into a lighting effect. For example, the higher the frequency data, the more frequently the light flashes.

[0091] The lights are arranged around the axis of the gas turbine, with a ring of lights around each core component. Since most of the core components of the gas turbine rotate around the axis, when a core component malfunctions, the sound of the core component is acquired in real time through a sound-receiving device. By judging whether the sound of the core component is moving away from or closer to the sound-receiving device, and combining the position detection of the sensor, the location of the abnormal noise in the core component can be inferred. The position of the flashing lights can then be controlled according to the location of the abnormal noise.

[0092] It should be noted that by locating the audio of the abnormal noise, the sensor position is used as the starting point to detect the location where the abnormal noise occurs, thereby determining the rotation angle of the core component, and thus inferring where the abnormal noise occurs in the core component.

[0093] Furthermore, the location of abnormal noises can be determined based on simulated images and operating sounds, and the flashing position of the lights can be changed to the location of the abnormal noises to alert the operators that the abnormal noises have occurred at this point. By combining operating data with the location of abnormal noises, future operational risks can be accurately predicted, thereby improving subsequent operational efficiency.

[0094] Example 2: A predictive detection method for gas turbines

[0095] This embodiment is an improvement based on Embodiment 1, according to... Figure 1 The system also includes a cooling system and a lubrication system, which are used to assist in the operation of the gas turbine.

[0096] Among these methods, the operation status of the lubrication system in the gas turbine is observed, and the shaft motion of the gas turbine is photographed and uploaded to the simulation system to assist in the optimization analysis report.

[0097] It also includes setting a lubricating oil viscosity reference table, and adjusting the lubricating oil in the lubrication system according to the lubricating oil viscosity reference table;

[0098] An image of the adjusted lubricating oil is captured, and the colors in the image are compared with the lubricating oil colors in the lubricating oil viscosity chart to determine the current lubricating oil viscosity.

[0099] One end of the drive shaft is connected to one end of the generator;

[0100] By photographing the lubricating oil running through the drive shaft in the gas turbine, the lubrication effect is observed. The running status of the drive shaft is then uploaded to a simulation system and combined with the drive shaft running data detected by sensors to obtain real drive shaft motion data.

[0101] If poor lubrication leads to bearing wear, the vibration sensor (accelerometer) will detect an abnormal increase in vibration amplitude. Combined with the detection of metal particles in the oil, the fault point can be located.

[0102] When electricity demand suddenly increases, requiring the gas turbine to operate at higher power, if the viscosity of the lubricating oil inside the gas turbine cannot meet the increased power requirements, it will cause mutual wear of the internal mechanical parts of the gas turbine. The usual solution is to shut down the gas turbine, wait for the internal lubricating oil to cool down, drain the used lubricating oil, and add a new grade of lubricating oil. However, this method takes a long time and cannot be used in emergencies. Therefore, in the existing methods, unless necessary, two lubricating oils with different viscosity grades can be mixed to adjust the viscosity and adapt to the current operating environment.

[0103] Different colored pigments are added to lubricating oils of different grades;

[0104] The lubricating oil viscosity chart includes different colored images corresponding to lubricating oils of different viscosities;

[0105] The historical data stores solutions corresponding to the sound results, including solutions such as changing the lubricating oil viscosity.

[0106] The color of the lubricating oil is determined based on the captured image. When it is necessary to change the viscosity of the lubricating oil, the concentration ratio is obtained by adjusting the color based on the determined color.

[0107] A portion of the lubricating oil inside the gas turbine is discharged based on the concentration ratio, and new lubricating oil is added to form a lubricating oil with the same viscosity.

[0108] Add an equal proportion of pigment to each type of lubricating oil, and combine them according to the color matching to create several mixed lubricating oil icons of different colors and viscosities:

[0109] Orange, made with 52% lemon yellow and 48% scarlet, viscosity grade VG 32;

[0110] Orange-red, with 8% red and 92% yellow, viscosity grade VG 46;

[0111] Rose pink, with 85% purple-red and 15% scarlet, viscosity grade VG 68;

[0112] When it is necessary to mix two or more lubricating oils of different viscosities, the proportions are determined according to the color mixing formula. For example, if a gas turbine currently has 3,000 liters of lemon-yellow lubricating oil and now requires a VG 32 viscosity grade lubricating oil, 48% of the total weight of lemon-yellow lubricating oil is discharged, and 48% of bright red lubricating oil is added to create an orange VG 32 viscosity grade lubricating oil. Compared to the relatively long time required for sensors to detect changes in viscosity grade, a color mixing method is used. The color of the mixed lubricating oil changes rapidly when captured by a camera, and due to the friction between the mechanical parts, the lubricating oil can be quickly stirred and blended. By observing the color change trend in the image captured by the camera, it is possible to quickly determine whether the required viscosity grade of lubricating oil has been achieved, allowing the machine to adjust the lubricating oil without stopping the machine in emergency situations.

[0113] It should be noted that the newly added lubricating oil needs to be heated to the same temperature as the lubricating oil inside the gas turbine.

[0114] Example 3: A predictive detection method for gas turbines

[0115] This embodiment is an improvement based on embodiment 2, according to... Figure 1 The image shown is a pixelated image of lubricating oil, which identifies impurities within the lubricating oil and estimates the quantity of impurities.

[0116] Based on the estimated amount of impurities, combined with simulated images and component operating sounds, the problematic components in the gas turbine were identified and their locations determined.

[0117] The result was determined by combining the location of the problematic component with historical data and the sound of the component operating.

[0118] Furthermore, when poor lubrication leads to bearing wear, the vibration sensor (accelerometer) will simultaneously detect an abnormal increase in vibration amplitude. Combined with oil metal particle detection, the fault point can be located. However, the installation position of the detection sensor is limited, meaning that oil metal particle detection still requires the machine to run for a relatively long time to detect where the parts are abnormally worn. Therefore, by capturing images of the lubricating oil, and considering that the lubricating oil is brightly colored while the wear particles from the worn parts are grayish-brown, the color difference between the two is quite obvious. This allows the images of the lubricating oil to be pixelated to find the grayish-brown pixels. By capturing 5 images within 10 seconds, the trend of the grayish-brown pixels can be quickly determined, thereby predicting whether the lubricating oil grade is correct. It should be noted that the lubricating oil area is the same in all 5 images.

[0119] Example 4: A predictive detection method for gas turbines

[0120] This embodiment is an improvement based on embodiment 3, according to... Figure 1 The demonstration also included warning lights surrounding the gas turbine, with the number of rings of warning lights matching the number of core components inside the gas turbine.

[0121] The method for changing the warning light includes, after determining the location of the problematic component, having the warning light at that location flash for at least 3 seconds.

[0122] The abnormal noise is identified by the sound of the component in operation, and a warning light is activated to flash locally at the area where the abnormal noise occurs for at least 3 seconds, and the sound is recorded.

[0123] The historical data contains the sounds of each component operating individually and the corresponding sound results.

[0124] Sound is collected at the location of each core component, and the sound at the location of a single core component in real time is obtained by separating the wavelength and frequency of the sound.

[0125] The sound at the location of a single core component during real-time operation is compared with the sound of the component running individually in historical data to obtain the corresponding sound results.

[0126] The audio and volume of the sound result are converted into the flashing amplitude of the light to warn the operator.

[0127] The location of the abnormal noise area at the specific development address is combined with the simulation image, and the abnormal noise sound at each time point is marked at the location of the abnormal noise area to generate an anomaly file;

[0128] The abnormal documents and analysis reports are sent to personnel so that they can quickly analyze whether there are any operational risks if the gas turbine continues to operate.

[0129] According to Example 1, by limiting various sound composition conditions, the sound receiving device is modified to directionally collect the operating sound of individual components in the gas turbine. For example, when the compressor noise in the gas turbine is maintained at 2000Hz and the peak frequency reaches 110dB, when the compressor moves to a certain position, the compressor noise increases to 2800Hz and the peak frequency increases to 140dB. At this time, it is determined that there is an abnormal noise when the compressor moves to a certain position. The abnormal noise sound at this time is compared with the individual operation of the corresponding component in the historical data. Based on the abnormal noise location, abnormal noise data, and previous abnormal noise sound results, the problem of the component is inferred, and it is determined whether adjustments need to be made, such as changing the lubricating oil viscosity, adjusting the operating power, adjusting the fuel and intake air ratio, etc.

[0130] Once the location of the faulty component causing the abnormal noise is determined, an abnormal noise threshold is set.

[0131] When the abnormal noise exceeds the threshold by 10%-15%, the light will flash 3 times per second.

[0132] When the abnormal noise exceeds the threshold by 16%-25%, the light will flash 6 times per second.

[0133] When the abnormal noise exceeds the threshold by 26%-35%, the light will flash 9 times per second.

[0134] Furthermore, the method of flashing the lights involves flashing a ring of lights around the outer surface of the gas turbine corresponding to the problematic component. This allows the operator to first identify which component is malfunctioning. Then, the lights are flashed locally at the identified location of the abnormal noise. This allows experienced personnel to predict the operational problems that will occur at that location, enabling the operator to make a direct and immediate judgment instead of waiting for the sensors to detect data over a long period of time. This collaboration with the operator improves the predictive effectiveness.

[0135] It should be noted that since the gas turbine operates 24 hours a day without stopping, although there are on-duty personnel at the operating site, these personnel may not have extensive experience. Therefore, it is necessary to take pictures and record the location of the flashing light, and then send the subsequent sensor data, simulated images, and abnormal noise data to experienced personnel. Compared with traditional sensor data, the method of taking pictures and recording the location of the flashing light, and comparing it with simulated images and abnormal noise data, can provide a more intuitive and clear assessment of machine problems.

[0136] Example 5: A system for predictive detection of gas turbines

[0137] A system applying a predictive detection method for a gas turbine includes:

[0138] The forecasting module collects peak electricity consumption and weather information to determine electricity demand, thereby generating adjustment plans for the maintenance of the gas turbine in advance.

[0139] The monitoring module detects mechanical data, temperature data, pressure data, vibration data, lubrication data, and cooling data of the gas turbine in operation, and summarizes them to generate simulated images;

[0140] By collecting the sounds of gas turbines during operation, combining the sounds with simulated images, operational risks are predicted, and new adjustment plans are generated based on the operational risks and historical data.

[0141] The analysis module is used to separate and reduce noise during the operation of the gas turbine, obtaining the individual operating sounds of each component in the development process.

[0142] By combining the sound localization of individual components with historical data, the location of the problem in a single component can be determined, and an analysis report can be generated by combining simulated images.

[0143] The communication module transmits the analysis report to the human operator, who then assesses operational risks and further optimizes and adjusts the strategy.

[0144] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0145] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A predictive detection method for gas turbines, characterized in that, Includes the following steps: Collect electricity demand information and gas turbine operation data to generate an adjustment plan, and adjust the gas turbine operation system according to the adjustment plan; A safety table is set up, and a safety score is determined based on the gas turbine operating data. The gas turbine operating data is imported into the simulation system in real time to generate a simulation image; Collect the operating sound of the gas turbine and separate the operating sound of several components by noise reduction; The simulated images are combined with the operating sounds of several components and historical data to provide early warning analysis of operating trends, thereby generating an analysis report. The audio and volume of the sounds from the operation of several components are converted into lighting patterns, and the warning lights execute the lighting patterns in real time. Based on the analysis report and changes in warning lights combined with historical data, the adjustment plan was revised again. If the revised adjustment plan fails to meet the operational safety requirements, the analysis report and lighting change plan will be sent to personnel for remote control.

2. The prediction and detection method according to claim 1, characterized in that, It also includes setting up and testing a cooling system and a lubrication system, which are used to assist the operation of the gas turbine; Among these methods, the operation status of the lubrication system in the gas turbine is observed, and the shaft motion of the gas turbine is photographed and uploaded to the simulation system to assist in the optimization and analysis report.

3. The predictive detection method according to claim 2, characterized in that, It also includes setting a lubricating oil viscosity reference table, and adjusting the lubricating oil in the lubrication system according to the lubricating oil viscosity reference table; An image of the adjusted lubricating oil is captured, and the colors in the image are compared with the lubricating oil colors in the lubricating oil viscosity chart to determine the current lubricating oil viscosity.

4. The predictive detection method according to claim 3, characterized in that, It also includes the following steps: Pixelated images of lubricating oil are used to identify impurities within the oil and estimate their quantity. Based on the estimated amount of impurities, combined with simulated images and component operating sounds, the problematic components in the gas turbine were identified and their locations determined. The result was determined by combining the location of the problematic component with historical data and the sound of the component operating.

5. The predictive detection method according to claim 4, characterized in that, It also includes setting up warning lights around the gas turbine, and the number of rings of the warning lights is the same as the number of core components inside the gas turbine; The method for changing the warning light includes, after determining the location of the problematic component, the warning light at that location flashes as a whole; The abnormal noise is detected at a specific location of the problematic component by the sound of the component's operation. The warning light is then activated to flash locally at the area where the abnormal noise occurs, and the sound is recorded and photographed.

6. The predictive detection method according to claim 5, characterized in that, The historical data contains the sounds of each component operating individually and the corresponding sound results. Sound is collected at the location of each core component, and the sound at the location of a single core component in real time is obtained by separating the wavelength and frequency of the sound. The sound at the location of a single core component during real-time operation is compared with the sound of the component operating individually in the historical data to obtain the corresponding sound result; The audio and volume of the sound result are converted into the flashing amplitude of the light to warn the operator.

7. The predictive detection method according to claim 6, characterized in that, Different colored pigments are added to lubricating oils of different grades; The lubricating oil viscosity reference table includes different color images corresponding to lubricating oils of different viscosities; The historical data stores solutions corresponding to the sound results, and the solutions include changing the viscosity of the lubricating oil. The color of the lubricating oil is determined based on the captured image. When it is necessary to change the viscosity of the lubricating oil, the concentration ratio is obtained by color adjustment based on the determined color. A portion of the lubricating oil inside the gas turbine is discharged based on the concentration ratio, and new lubricating oil is added to form a lubricating oil with the appropriate viscosity.

8. The predictive detection method according to claim 1, characterized in that, The historical data is used to determine the months with peak electricity consumption, and holiday or weather information is collected to determine the peak days within those months. Based on these peak days, the gas turbines are maintained in advance.

9. The predictive detection method according to claim 5, characterized in that, It also includes combining the location of the abnormal noise area with the simulation image, and marking the abnormal noise that appears at each time point at the location of the abnormal noise area to generate an anomaly file; The abnormal files and analysis reports are sent to personnel so that they can quickly analyze whether there are any operational risks if the gas turbine continues to operate.

10. A system for predictive detection of gas turbines, characterized in that, The detection is performed using the predictive detection method according to any one of claims 1-9, and includes the following modules: The prediction module collects peak electricity consumption and weather information to determine electricity demand, thereby generating adjustment plans for the maintenance of the gas turbine in advance; The monitoring module detects mechanical, temperature, pressure, vibration, lubrication, and cooling data of the gas turbine during operation and aggregates them to generate simulated images. By collecting the sound of the gas turbine during operation, the module combines the sound with the simulated images to predict operational risks and generates new adjustment plans based on the operational risks and historical data. The analysis module is used to separate and reduce noise during the operation of the gas turbine, obtaining the individual operating sounds of each component in the development process. By locating the individual operating sounds of each component and combining them with historical data, the system determines the location of the problem in each component and generates an analysis report by combining the simulation images. The communication module transmits the analysis report to the human operator, who then assesses operational risks and further optimizes and adjusts the strategy.

Images