CN115539220A - A fault detection method for gas turbine blade channel temperature sensor - Google Patents
A fault detection method for gas turbine blade channel temperature sensor Download PDFInfo
- Publication number
- CN115539220A CN115539220A CN202211220744.5A CN202211220744A CN115539220A CN 115539220 A CN115539220 A CN 115539220A CN 202211220744 A CN202211220744 A CN 202211220744A CN 115539220 A CN115539220 A CN 115539220A
- Authority
- CN
- China
- Prior art keywords
- gas turbine
- data
- turbine blade
- temperature sensor
- blade channel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
- F02C—GAS-TURBINE PLANTS; AIR INTAKES FOR JET-PROPULSION PLANTS; CONTROLLING FUEL SUPPLY IN AIR-BREATHING JET-PROPULSION PLANTS
- F02C9/00—Controlling gas-turbine plants; Controlling fuel supply in air- breathing jet-propulsion plants
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F01—MACHINES OR ENGINES IN GENERAL; ENGINE PLANTS IN GENERAL; STEAM ENGINES
- F01K—STEAM ENGINE PLANTS; STEAM ACCUMULATORS; ENGINE PLANTS NOT OTHERWISE PROVIDED FOR; ENGINES USING SPECIAL WORKING FLUIDS OR CYCLES
- F01K23/00—Plants characterised by more than one engine delivering power external to the plant, the engines being driven by different fluids
- F01K23/02—Plants characterised by more than one engine delivering power external to the plant, the engines being driven by different fluids the engine cycles being thermally coupled
- F01K23/06—Plants characterised by more than one engine delivering power external to the plant, the engines being driven by different fluids the engine cycles being thermally coupled combustion heat from one cycle heating the fluid in another cycle
- F01K23/10—Plants characterised by more than one engine delivering power external to the plant, the engines being driven by different fluids the engine cycles being thermally coupled combustion heat from one cycle heating the fluid in another cycle with exhaust fluid of one cycle heating the fluid in another cycle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05D—INDEXING SCHEME FOR ASPECTS RELATING TO NON-POSITIVE-DISPLACEMENT MACHINES OR ENGINES, GAS-TURBINES OR JET-PROPULSION PLANTS
- F05D2260/00—Function
- F05D2260/80—Diagnostics
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05D—INDEXING SCHEME FOR ASPECTS RELATING TO NON-POSITIVE-DISPLACEMENT MACHINES OR ENGINES, GAS-TURBINES OR JET-PROPULSION PLANTS
- F05D2260/00—Function
- F05D2260/81—Modelling or simulation
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05D—INDEXING SCHEME FOR ASPECTS RELATING TO NON-POSITIVE-DISPLACEMENT MACHINES OR ENGINES, GAS-TURBINES OR JET-PROPULSION PLANTS
- F05D2270/00—Control
- F05D2270/30—Control parameters, e.g. input parameters
- F05D2270/303—Temperature
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Control Of Turbines (AREA)
Abstract
本发明属于燃气‑蒸汽联合循环发电机组状态故障诊断检测领域,具体涉及一种燃气轮机叶片通道温度传感器故障检测方法,包括对燃气轮机叶片通道温度进行系统全面的控制逻辑分析,获取影响燃气轮机叶片通道温度变化的主要热参数;从燃气‑蒸汽联合循环机组数据库中采集筛选包含机组运行全工况的无故障数据;基于建立的LightGBM模型和自编码器模型,一方面成功消除了大气温度和工况正常变化对燃气轮机叶片通道温度的影响,另一方面基于建立的传感器故障准则,也成功将传感器故障和燃气轮机部件故障导致的热参数偏离区别开来,实现了实时在线对燃气轮机叶片通道温度传感器故障检测,对准确判断燃气‑蒸汽联合循环机组运行状态具有重要意义。
The invention belongs to the field of gas-steam combined cycle generator set state fault diagnosis and detection, and specifically relates to a gas turbine blade passage temperature sensor fault detection method, which includes systematically and comprehensively controlling the logic analysis of the gas turbine blade passage temperature, and obtaining the temperature changes affecting the gas turbine blade passage The main thermal parameters; from the gas-steam combined cycle unit database to collect and filter the fault-free data including the whole working condition of the unit; based on the established LightGBM model and autoencoder model, on the one hand, it successfully eliminates the normal changes in atmospheric temperature and working conditions The impact on the temperature of the gas turbine blade passage, on the other hand, based on the established sensor failure criteria, the thermal parameter deviation caused by the sensor failure and the gas turbine component failure has also been successfully distinguished, and the real-time online fault detection of the gas turbine blade passage temperature sensor has been realized. It is of great significance to accurately judge the operating status of the gas-steam combined cycle unit.
Description
技术领域technical field
本发明属于燃气-蒸汽联合循环发电机组状态故障诊断检测领域,具体涉及一种燃气轮机叶片通道温度传感器故障检测方法。The invention belongs to the field of state fault diagnosis and detection of a gas-steam combined cycle generator set, and in particular relates to a fault detection method for a gas turbine blade channel temperature sensor.
背景技术Background technique
燃气-蒸汽联合循环机组是构建新型电力系统的重要发电模式,具有运行灵活、启停迅速、碳排放强度低、输出功率范围广等优点。近年来天然气发电装机量持续高速增长,2025年装机规模将会达到 1.5亿千瓦以上。燃气轮机作为燃气-蒸汽联合循环机组的核心装置,燃气轮机工作环境复杂、工况多变,随着运行时间增长,大大增加其失效风险。燃气轮机叶片通道温度作为燃气轮机的一个重要的热参数,对于准确判断燃气轮机状态是不可或缺的。故要求传感器能准确无误地获得真实的叶片通道温度。一旦传感器发生故障,就会产生错误的监测信号,导致集控运行人员对燃机运行状况做出误判,甚至可能引发非停事故。The gas-steam combined cycle unit is an important power generation mode for building a new type of power system. It has the advantages of flexible operation, rapid start and stop, low carbon emission intensity, and wide output power range. In recent years, the installed capacity of natural gas power generation has continued to grow rapidly, and the installed capacity will reach more than 150 million kilowatts in 2025. Gas turbine is the core device of the gas-steam combined cycle unit. The working environment of the gas turbine is complex and the working conditions are changeable. As the operating time increases, the risk of failure is greatly increased. As an important thermal parameter of the gas turbine, the temperature of the gas turbine blade passage is indispensable for accurately judging the state of the gas turbine. Therefore, the sensor is required to accurately obtain the real temperature of the blade passage. Once the sensor fails, a wrong monitoring signal will be generated, causing the centralized control operator to make a misjudgment of the operating status of the gas turbine, and may even cause a non-shutdown accident.
燃气轮机在正常状态运行时,其叶片通道温度随大气温度和工况的变化而改变,且变化范围幅度较大,因此要检测传感器的状态是否正常,首先要将叶片通道温度随大气温度和工况正常变化的影响进行处理和消除,其次还要消除燃机本体设备故障对叶片通道温度带来的影响。目前对于燃气轮机叶片通道温度变工况分析主要通过基于物理模型与仿真结合的方法,但现阶段我国重型燃机行业主要被国外大型整机制造企业垄断,缺乏核心关键技术,建立精确的数学模型十分困难,变工况计算准确性也就无法保证。When the gas turbine is running in a normal state, the temperature of the blade passage changes with the atmospheric temperature and working conditions, and the range of change is relatively large. The impact of normal changes should be dealt with and eliminated, and secondly, the impact of gas turbine body equipment failure on the blade passage temperature should be eliminated. At present, the analysis of the temperature variation of the gas turbine blade channel is mainly based on the method based on the combination of physical models and simulations. However, at this stage, my country's heavy-duty gas turbine industry is mainly monopolized by foreign large-scale complete machine manufacturers, lacking core key technologies, and it is very difficult to establish an accurate mathematical model. Difficulties, the accuracy of calculation under variable working conditions cannot be guaranteed.
发明内容Contents of the invention
本发明的目的是克服现有技术存在的缺陷,提供一种燃气轮机叶片通道温度传感器故障检测方法。The purpose of the present invention is to overcome the defects in the prior art and provide a fault detection method for a gas turbine blade channel temperature sensor.
为了实现上述目的,本发明采用如下技术方案是:In order to achieve the above object, the present invention adopts the following technical solutions:
一种燃气轮机叶片通道温度传感器故障检测方法,包括以下步骤:A gas turbine blade channel temperature sensor failure detection method, comprising the following steps:
步骤1、对燃气轮机叶片通道温度进行系统全面的控制逻辑分析,获取影响燃气轮机叶片通道温度变化的主要热参数;Step 1. Carry out a systematic and comprehensive control logic analysis on the temperature of the gas turbine blade passage, and obtain the main thermal parameters that affect the temperature change of the gas turbine blade passage;
步骤2、从燃气-蒸汽联合循环机组数据库中采集筛选包含机组运行全工况的无故障数据;Step 2. Collect and filter the fault-free data including the full operating conditions of the unit from the database of the gas-steam combined cycle unit;
步骤3、为了保证数据的准确性,考虑到步骤2中采集的测量数据会有可能存在空值、失真异常等情况,故对采集出的数据进行清洗和滤波处理;Step 3. In order to ensure the accuracy of the data, considering that the measurement data collected in step 2 may have null values, abnormal distortion, etc., the collected data is cleaned and filtered;
步骤4、基于LightGBM模型(Light Gradient Boosting Machine) 的方法,根据历史无故障全工况运行数据建立叶片通道温度控制理论值回归模型,并利用步骤3中的数据进行模型的训练和测试;Step 4. Based on the method of the LightGBM model (Light Gradient Boosting Machine), a regression model of the theoretical value of the blade channel temperature control is established according to the historical operation data of no faults and full working conditions, and the data in step 3 is used for training and testing of the model;
步骤5、基于自编码器的方法,根据历史无故障全工况运行数据建立叶片通道温度编码——解码状态还原回归模型,并利用步骤3中的数据进行模型的训练和测试;Step 5. Based on the autoencoder method, establish a blade channel temperature encoding-decoding state restoration regression model based on the historical fault-free full working condition operating data, and use the data in step 3 to train and test the model;
步骤6、基于步骤4中LightGBM模型、步骤5中自编码器型模型,对叶片通道温度残差进行相互分析,确定相应的阈值;Step 6. Based on the LightGBM model in step 4 and the autoencoder model in step 5, the temperature residuals of the blade channel are mutually analyzed to determine the corresponding threshold;
步骤7、建立燃气轮机叶片通道温度传感器故障判断准则;Step 7, establishing a fault judgment criterion for the gas turbine blade channel temperature sensor;
步骤8、利用燃气轮机实测的运行数据,基于步骤7的故障判断准则,实现实时在线对燃气轮机叶片通道温度传感器故障检测。Step 8. Using the actual measured operation data of the gas turbine and based on the fault judgment criterion in step 7, real-time online fault detection of the gas turbine blade channel temperature sensor is realized.
优选地,所述步骤1中影响燃气轮机叶片通道温度变化的主要热参数主要为:燃烧室壳体压比、压气机进口空气温度、燃料流量、燃烧室旁路阀开度、IGV开度、燃烧室压力、燃气轮机功率等;其中燃烧室壳体压比的计算公式为:Preferably, the main thermal parameters that affect the temperature change of the gas turbine blade channel in the step 1 are: pressure ratio of the combustion chamber shell, air temperature at the inlet of the compressor, fuel flow, opening of the combustion chamber bypass valve, IGV opening, combustion Chamber pressure, gas turbine power, etc.; the formula for calculating the pressure ratio of the combustion chamber shell is:
式中:PCS为压气机出口压力,P0为环境大气压力。In the formula: PCS is the outlet pressure of the compressor, and P0 is the atmospheric pressure of the environment.
优选地,所述步骤2中从燃气-蒸汽联合循环机组数据库中采集筛选包含机组运行全工况的无故障数据;采集时间周期为一年,数据采样频率为5s~1min;采集数据点包括:叶片通道温度#1~#20。Preferably, in the step 2, from the database of the gas-steam combined cycle unit, the fault-free data including the full operating conditions of the unit are collected and screened; the collection time period is one year, and the data sampling frequency is 5s to 1min; the collected data points include: Blade channel temperature #1~#20.
优选地,所述步骤3中为了保证数据的准确性,考虑到步骤2中采集的测量数据会有可能存在空值、失真异常等情况,故对采集出的数据进行清洗和滤波处理。Preferably, in step 3, in order to ensure the accuracy of the data, the collected data are cleaned and filtered considering that the measurement data collected in step 2 may have null values, abnormal distortion, etc.
优选地,所述清洗的数据包括空值数据和离群点数据,进一步还可以包括停机工况的数据,即机组负荷为0或接近0时段对应的数据;Preferably, the cleaned data includes null data and outlier data, and may further include data of shutdown conditions, that is, data corresponding to a time period when the unit load is 0 or close to 0;
空值数据为某一时刻一个或多个测点存在空值的数据,离群点数据为超出正常范围的数据;编写相应的代码对空值数据进行去除。The null value data is the data with null value at one or more measuring points at a certain moment, and the outlier data is the data beyond the normal range; write the corresponding code to remove the null value data.
优选地,,所述步骤4中LightGBM(Light Gradient Boosting Machine)模型是一个实现GBDT(Gradient Boosting Decision Tree) 算法的框架;LightGBM模型输入层共3维,包括燃气轮机有功功率、燃烧室壳体压比、压气机进口空气温度;输出层为20维,包括叶片通道温度#1~#20。LightGBM模型输出的叶片通道温度#1~#20称为理论控制叶片通道温度#1~#20,表示为TCTi i=1,2,3……20。Preferably, the LightGBM (Light Gradient Boosting Machine) model in the step 4 is a framework for implementing the GBDT (Gradient Boosting Decision Tree) algorithm; the LightGBM model input layer has a total of 3 dimensions, including gas turbine active power, combustion chamber shell pressure ratio , The inlet air temperature of the compressor; the output layer is 20-dimensional, including the blade channel temperature #1~#20. The blade passage temperatures #1 to #20 output by the LightGBM model are called theoretically controlled blade passage temperatures #1 to #20, expressed as TCT i i=1, 2, 3...20.
优选地,所述步骤5中自编码器(Auto-Encoder,AE),是一种无监督式学习模型;基于反向传播算法与最优化方法(如梯度下降法),利用输入数据T本身作为监督,来指导神经网络尝试学习一个映射关系,从而得到一个重构输出TR;借助于深层神经网络的非线性特征提取能力,自编码器可以获得良好的数据表示。Preferably, the self-encoder (Auto-Encoder, AE) in the step 5 is an unsupervised learning model; based on the backpropagation algorithm and optimization method (such as the gradient descent method), the input data T itself is used as Supervision, to guide the neural network to try to learn a mapping relationship, so as to obtain a reconstructed output TR ; with the help of the nonlinear feature extraction ability of the deep neural network, the autoencoder can obtain a good data representation.
优选地,所述步骤6中基于步骤4中LightGBM模型、步骤5中自编码器模型,对叶片通道温度残差进行相互分析,确定相应的阈值。Preferably, in the step 6, based on the LightGBM model in the step 4 and the autoencoder model in the step 5, the temperature residuals of the blade channel are mutually analyzed to determine the corresponding threshold.
优选地,所述步骤7建立燃气轮机叶片通道温度传感器故障判断准则,准则为如下:Preferably, said step 7 establishes a gas turbine blade channel temperature sensor fault judgment criterion, which is as follows:
ε1i≥thresholdLGB ε 1i ≥threshold LGB
ε2i≥thresholdAE ε 2i ≥threshold AE
ε3i≤thresholdLGB-AE ε 3i ≤threshold LGB-AE
与现有技术相比较,具备如下优点:Compared with the prior art, it has the following advantages:
本发明提供一种燃气轮机叶片通道温度传感器故障检测方法,针对影响燃气轮机叶片通道温度热参数因素众多,无法准确确定的问题,对燃气轮机叶片通道温度进行系统全面的控制逻辑分析,从机理角度提取影响燃气轮机叶片通道温度变化的主要热参数。在对大量历史数据清洗及预处理的基础上,对燃气轮机叶片通道温度建立了 LightGBM模型,利用其更低的内存消耗、支持分布式可以快速处理海量数据等优点,利用“机理+大数据”思想消除了大气温度和负荷工况正常变化对燃气轮机叶片通道温度的影响。另一方面对叶片通道温度建立了自编码器模型,从大数据的思想角度实现了叶片通道温度的无故障情况下的状态重构。并通过残差分析及建立的传感器故障准则,成功将传感器故障和燃气轮机部件故障导致的热参数偏离区别开来,实现了实时在线对燃气轮机叶片通道温度传感器故障检测,并可以精确定位到具体哪只热电偶故障,有利于工作人员获取到更加细致的故障信息,便于及时排除故障。The invention provides a gas turbine blade channel temperature sensor fault detection method. Aiming at the problem that there are many thermal parameters affecting the temperature of the gas turbine blade channel and cannot be accurately determined, a systematic and comprehensive control logic analysis is performed on the temperature of the gas turbine blade channel, and the factors affecting the gas turbine are extracted from the perspective of mechanism. The main thermal parameters of the blade channel temperature variation. On the basis of cleaning and preprocessing a large amount of historical data, a LightGBM model is established for the temperature of the gas turbine blade passage, taking advantage of its lower memory consumption, support for distributed and fast processing of massive data, and the idea of "mechanism + big data" The influence of normal changes in atmospheric temperature and load conditions on the temperature of the gas turbine blade passage is eliminated. On the other hand, an autoencoder model is established for the temperature of the blade channel, and the state reconstruction of the temperature of the blade channel without failure is realized from the perspective of big data. And through the residual analysis and the established sensor failure criteria, the thermal parameter deviation caused by the sensor failure and the gas turbine component failure was successfully distinguished, and the real-time online detection of the gas turbine blade passage temperature sensor failure was realized, and the specific one could be accurately located The thermocouple failure is conducive to the staff to obtain more detailed fault information, which is convenient for timely troubleshooting.
附图说明Description of drawings
图1为本发明燃气轮机叶片通道温度传感器故障检测方法的流程示意图;Fig. 1 is the schematic flow sheet of the gas turbine blade channel temperature sensor failure detection method of the present invention;
图2为燃气轮机工作流程及传感器测点分布示意图;Figure 2 is a schematic diagram of the working process of the gas turbine and the distribution of sensor measuring points;
图3为叶片通道温度控制理论值回归模型示意图;Fig. 3 is a schematic diagram of the regression model of the theoretical value of the blade channel temperature control;
图4为叶片通道温度编码——解码状态还原回归模型示意图。Figure 4 is a schematic diagram of the blade channel temperature encoding-decoding state restoration regression model.
具体实施方式detailed description
为使本发明的上述目的、特征和优点更加清晰和方便理解,下面集合附图和具体实施例对本发明作进一步的详细说明。In order to make the above objects, features and advantages of the present invention clearer and easier to understand, the present invention will be further described in detail below with reference to the drawings and specific embodiments.
步骤1、对燃气轮机叶片通道温度进行系统全面的控制逻辑分析,获取影响燃气轮机叶片通道温度变化的主要热参数。Step 1. Carry out systematic and comprehensive control logic analysis on the temperature of the gas turbine blade passage, and obtain the main thermal parameters that affect the temperature change of the gas turbine blade passage.
步骤2、从燃气-蒸汽联合循环机组数据库中采集筛选包含机组运行全工况的无故障数据。Step 2. Collect and filter the fault-free data including all working conditions of the unit from the database of the gas-steam combined cycle unit.
步骤3、为了保证数据的准确性,考虑到步骤2中采集的测量数据会有可能存在空值、失真异常等情况,故对采集出的数据进行清洗和滤波处理。Step 3. In order to ensure the accuracy of the data, considering that the measurement data collected in step 2 may have null values, abnormal distortion, etc., the collected data is cleaned and filtered.
步骤4、基于LightGBM(Light Gradient Boosting Machine) 的方法,根据历史无故障全工况运行数据建立叶片通道温度控制理论值回归模型,如图3所示,并利用步骤3中的数据进行模型的训练和测试。Step 4. Based on the LightGBM (Light Gradient Boosting Machine) method, a regression model of the theoretical value of the blade channel temperature control is established based on the historical operating data without faults in all working conditions, as shown in Figure 3, and the data in step 3 is used for model training and test.
步骤5、基于自编码器的方法,根据历史无故障全工况运行数据建立叶片通道温度编码——解码状态还原回归模型,如图4所示,并利用步骤3中的数据进行模型的训练和测试。Step 5. Based on the autoencoder method, establish a blade channel temperature encoding-decoding state restoration regression model according to the historical operation data of no faults and full working conditions, as shown in Figure 4, and use the data in step 3 for model training and test.
步骤6、基于步骤4中LightGBM模型、步骤5中自编码器型模型,对叶片通道温度残差进行相互分析,确定相应的阈值。Step 6. Based on the LightGBM model in step 4 and the autoencoder model in step 5, the temperature residuals of the blade channel are mutually analyzed to determine the corresponding threshold.
步骤7、建立燃气轮机叶片通道温度传感器故障判断准则。Step 7. Establish a fault judgment criterion for the gas turbine blade channel temperature sensor.
步骤8、利用燃气轮机实测的运行数据,基于步骤7的故障判断准则,实现实时在线对燃气轮机叶片通道温度传感器故障检测。Step 8. Using the actual measured operation data of the gas turbine and based on the fault judgment criterion in step 7, real-time online fault detection of the gas turbine blade channel temperature sensor is realized.
所述步骤1中影响燃气轮机叶片通道温度变化的主要热参数主要为:燃烧室壳体压比、压气机进口空气温度、燃料流量、燃烧室旁路阀开度、IGV开度、燃烧室压力、燃气轮机功率等,具体如表1所示。其中燃烧室壳体压比的计算公式为:The main thermal parameters that affect the temperature change of the gas turbine blade channel in the step 1 are: combustion chamber shell pressure ratio, compressor inlet air temperature, fuel flow, combustion chamber bypass valve opening, IGV opening, combustion chamber pressure, The power of the gas turbine is shown in Table 1. The formula for calculating the pressure ratio of the combustion chamber shell is:
式中:PCS为压气机出口压力,P0为环境大气压力。In the formula: PCS is the outlet pressure of the compressor, and P0 is the atmospheric pressure of the environment.
表1影响燃气轮机叶片通道温度变化的主要热参数Table 1 The main thermal parameters affecting the temperature change of the gas turbine blade passage
所述步骤2中从燃气-蒸汽联合循环机组数据库中采集筛选包含机组运行全工况的无故障数据。建议采集时间周期为一年,数据采样频率为5s~1min。采集数据点包括:叶片通道温度#1~#20(通常来说燃机叶片通道圆周式分布20个热电偶传感器进行温度测量),环境大气压力、压气机进口空气温度、压气机出口压力、燃料流量、燃烧室旁路阀开度、IGV开度、燃烧室压力、燃气轮机功率。燃气轮机工作流程及传感器测点分布示意图如图2所示。In the step 2, from the gas-steam combined cycle unit database, the fault-free data including all working conditions of the unit are collected and screened. The recommended collection time period is one year, and the data sampling frequency is 5s to 1min. The collected data points include: blade channel temperature #1~#20 (generally, 20 thermocouple sensors are distributed around the gas turbine blade channel for temperature measurement), ambient atmospheric pressure, compressor inlet air temperature, compressor outlet pressure, fuel Flow rate, combustion chamber bypass valve opening, IGV opening, combustion chamber pressure, gas turbine power. The schematic diagram of the working process of the gas turbine and the distribution of sensor measuring points is shown in Figure 2.
所述步骤3中为了保证数据的准确性,考虑到步骤2中采集的测量数据会有可能存在空值、失真异常等情况,故对采集出的数据进行清洗和滤波处理。清洗的数据可以包括空值数据和离群点数据,进一步还可以包括停机工况的数据,即机组负荷为0或接近0时段对应的数据。In the step 3, in order to ensure the accuracy of the data, the collected data are cleaned and filtered considering that the measurement data collected in the step 2 may have null values, abnormal distortion, etc. The cleaned data may include null data and outlier data, and may further include data of shutdown conditions, that is, data corresponding to periods when the unit load is 0 or close to 0.
空值数据为某一时刻一个或多个测点存在空值的数据,离群点数据为超出正常范围的数据。编写相应的代码对空值数据进行去除,采用箱线图法去除离群点数据。假设q1、q3为数据的第1四分位数、第 3四分位数,箱线图法可以表示为:Null data refers to data with null values at one or more measuring points at a certain moment, and outlier data refers to data beyond the normal range. Write the corresponding code to remove the null value data, and use the box plot method to remove the outlier data. Assuming that q 1 and q 3 are the first quartile and the third quartile of the data, the boxplot method can be expressed as:
xmax=q3+1.5×(q3-q1)x max =q 3 +1.5×(q 3 -q 1 )
xmin=q3-1.5×(q3-q1)x min =q 3 -1.5×(q 3 -q 1 )
其中,xmax表示数据中的异常极大值,xmin为数据中的异常极小值。若数据中存在小于异常极小值,大于异常极大值的数据则确定为离群点数据,将其去除。Among them, x max represents the abnormal maximum value in the data, and x min is the abnormal minimum value in the data. If the data is smaller than the abnormal minimum value and greater than the abnormal maximum value, it is determined as outlier data and removed.
接着,对数据进行滤波处理。Next, filter the data.
对数据进行滤波处理的方法可以采用粒子滤波法。粒子滤波 (ParticleFilters,PF)算法是解决非线性非高斯动态系统的参数估计和状态滤波问题的关键方法之一,并且对系统的过程噪声和量测噪声没有任何限制,其精度可以逼近最优估计。根据带有噪声的观测值,递归估计非线性系统状态的后验概率密度p(x0:k|z1:k)。其中, x0:k={x0,x1,…xk}表示k时刻系统所产生的状态序列, z1:k={z1,z2,…zk}表示观测值序列。The method of filtering the data can adopt the particle filter method. The particle filter (Particle Filters, PF) algorithm is one of the key methods to solve the parameter estimation and state filtering problems of nonlinear non-Gaussian dynamic systems, and there is no limit to the process noise and measurement noise of the system, and its accuracy can approach the optimal estimate . Recursively estimate the posterior probability density p(x 0:k |z 1:k ) of the state of the nonlinear system from the observations with noise. Among them, x 0:k ={x 0 , x 1 ,...x k } represents the state sequence generated by the system at time k, and z 1:k ={z 1 , z 2 ,...z k } represents the observation value sequence.
粒子滤波的基本思想是构造一个基于样本的后验概率密度函数,使用N个粒子构成的集合表示系统状态的后验概率密度 p(x0:k|z1:k)。The basic idea of particle filtering is to construct a sample-based posterior probability density function, using a set of N particles Represents the posterior probability density p(x 0:k |z 1:k ) of the system state.
其中,{x0:k,i=0,1,…,N}表示支持样本(粒子)集合,抽取自后验概率分布的状态空间。各样本粒子的权值为且满足基于选择合适的重要性采样密度函数,通过不断的权值更新,最终完成数据的滤波过程。Wherein, {x 0: k , i=0, 1, . . . , N} represents a set of supporting samples (particles), extracted from the state space of the posterior probability distribution. The weight of each sample particle is and satisfied Based on the selection of the appropriate importance sampling density function, the filtering process of the data is finally completed through continuous weight updating.
所述步骤4中LightGBM(Light Gradient Boosting Machine)模型是一个实现GBDT(Gradient Boosting Decision Tree)算法的框架,支持高效率的并行训练,并且具有更快的训练速度、更低的内存消耗、更好的准确率、支持分布式可以快速处理海量数据等优点。 LightGBM模型输入层共3维,包括燃气轮机有功功率、燃烧室壳体压比、压气机进口空气温度;输出层为20维,包括叶片通道温度#1~#20。 LightGBM模型输出的叶片通道温度#1~#20称为理论控制叶片通道温度#1~#20,表示为TCTi i=1,2,3……20。The LightGBM (Light Gradient Boosting Machine) model in the step 4 is a framework for implementing the GBDT (Gradient Boosting Decision Tree) algorithm, supports efficient parallel training, and has faster training speed, lower memory consumption, better Accuracy, support for distributed, can quickly process massive data and other advantages. The input layer of the LightGBM model has 3 dimensions, including the active power of the gas turbine, the pressure ratio of the combustion chamber shell, and the inlet air temperature of the compressor; the output layer is 20 dimensions, including the blade channel temperature #1~#20. The blade passage temperatures #1 to #20 output by the LightGBM model are called theoretically controlled blade passage temperatures #1 to #20, expressed as TCT i i=1, 2, 3...20.
所述步骤5中自编码器(Auto-Encoder,AE),是一种无监督式学习模型。基于反向传播算法与最优化方法(如梯度下降法),利用输入数据T本身作为监督,来指导神经网络尝试学习一个映射关系,从而得到一个重构输出TR。借助于深层神经网络的非线性特征提取能力,自编码器可以获得良好的数据表示。自编码器算法模型包含两个主要的部分:Encoder(编码器)和Decoder(解码器)。The autoencoder (Auto-Encoder, AE) in the step 5 is an unsupervised learning model. Based on the backpropagation algorithm and optimization method (such as the gradient descent method), the input data T itself is used as supervision to guide the neural network to try to learn a mapping relationship, so as to obtain a reconstructed output T R . With the help of the nonlinear feature extraction capabilities of deep neural networks, autoencoders can obtain good data representations. The self-encoder algorithm model consists of two main parts: Encoder (encoder) and Decoder (decoder).
编码器的作用是把高维输入T编码成低维的隐变量h,从而强迫神经网络学习最有信息量的特征,编码过程为:The role of the encoder is to encode the high-dimensional input T into a low-dimensional hidden variable h, thereby forcing the neural network to learn the most informative features. The encoding process is:
h=gθ1(T)=σ(W1T+b1)h=g θ1 (T)=σ(W 1 T+b 1 )
解码器的作用是把隐藏层的隐变量h还原到初始维度,最好的状态就是解码器的输出能够完美地或者近似恢复出原来的输入,即 TR≈TThe role of the decoder is to restore the hidden variable h of the hidden layer to the original dimension. The best state is that the output of the decoder can perfectly or approximately restore the original input, that is, T R ≈ T
TR=gθ2(h)=σ(W2h+b2)T R =g θ2 (h)=σ(W 2 h+b 2 )
自编码器模型输入层共20维,输入是叶片通道温度#1~#20;自编码器模型输出层共20维,同样是叶片通道温度#1~#20。自编码器模型输出的叶片通道温度#1~#20称为状态还原叶片通道温度#1~#20。表示为SRTi i=1,2,3……20The input layer of the autoencoder model has a total of 20 dimensions, and the input is the blade channel temperature #1 to #20; the output layer of the autoencoder model has a total of 20 dimensions, and the same is the blade channel temperature #1 to #20. The vane channel temperatures #1-#20 output from the encoder model are referred to as state-restored vane channel temperatures #1-#20. Expressed as SRT i i=1,2,3...20
所述步骤6中基于步骤4中LightGBM模型、步骤5中自编码器模型,对叶片通道温度残差进行相互分析,确定相应的阈值。具体残差公式为:In the step 6, based on the LightGBM model in the step 4 and the autoencoder model in the step 5, the temperature residuals of the blade channel are mutually analyzed to determine the corresponding threshold. The specific residual formula is:
ε1i=|TCTi-Ti|,i=1,2,3……20ε 1i =|TCT i -T i |, i=1,2,3...20
ε2i=|SRTi-Ti|,i=1,2,3……20ε 2i =|SRT i -T i |, i=1,2,3...20
ε3i=|TCTi-SRTi|,i=1,2,3……20ε 3i =|TCT i -SRT i |, i=1,2,3...20
式中:ε1i为LightGBM模型输出与叶片通道温度热电偶测量数据残差,ε2i为自编码器模型输出与叶片通道温度热电偶测量数据有残差,ε3i为LightGBM模型输出与自编码器模型输出数据残差,Ti为叶片通道温度热电偶测量数据值。其中LightGBM模型输出与叶片通道温度热电偶测量数据有残差阈值,这里称为thresholdLGB;自编码器模型输出与叶片通道温度热电偶测量数据有残差阈值,这里称为 thresholdAE;LightGBM模型输出与自编码器模型输出有残差阈值,这里称为thresholdLGB-AE。In the formula: ε 1i is the residual error between the LightGBM model output and the thermocouple measurement data of the blade channel temperature, ε 2i is the residual error between the autoencoder model output and the blade channel temperature thermocouple measurement data, ε 3i is the LightGBM model output and the autoencoder Model output data residual, T i is the measured data value of the blade channel temperature thermocouple. Wherein the LightGBM model output and the blade channel temperature thermocouple measurement data have a residual threshold, which is called threshold LGB here; the autoencoder model output and the blade channel temperature thermocouple measurement data have a residual threshold, which is called threshold AE here; the LightGBM model output There is a residual threshold with the output of the autoencoder model, which is called threshold LGB-AE here.
通过对ε1i、ε2i、ε3i时间序列进行统计分析,从而确定相应阈值。采用的是核密度估计分析法,核密度表达式如下:Through statistical analysis of ε 1i , ε 2i , ε 3i time series, the corresponding thresholds are determined. The kernel density estimation analysis method is adopted, and the kernel density expression is as follows:
式中:为估计的概率密度值,n为样本数,h为窗宽,K 为核函数。选定置信度为99.5%,采用高斯核函数,得到相应阈值。In the formula: is the estimated probability density value, n is the number of samples, h is the window width, and K is the kernel function. The selected confidence level is 99.5%, and the Gaussian kernel function is used to obtain the corresponding threshold.
所述步骤7建立燃气轮机叶片通道温度传感器故障判断准则,准则为如下:Said step 7 establishes the gas turbine blade passage temperature sensor fault judgment criterion, and the criterion is as follows:
ε1i≥thresholdLGB ε 1i ≥threshold LGB
ε2i≥thresholdAE ε 2i ≥threshold AE
ε3i≤thresholdLGB-AE ε 3i ≤threshold LGB-AE
以上所述是本发明实施例的具体实施方式,应当指出,对于本技术领域的普通技术人员来说,在不脱离本发明原理的前提下,还可以做出若干改进和润饰,这些改进、润饰也应视为本申请的保护范围。The above is the specific implementation of the embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made. These improvements and modifications It should also be regarded as the protection scope of the present application.
Claims (9)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211220744.5A CN115539220B (en) | 2022-10-08 | 2022-10-08 | Gas turbine blade channel temperature sensor fault detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211220744.5A CN115539220B (en) | 2022-10-08 | 2022-10-08 | Gas turbine blade channel temperature sensor fault detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115539220A true CN115539220A (en) | 2022-12-30 |
CN115539220B CN115539220B (en) | 2024-07-05 |
Family
ID=84731754
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211220744.5A Active CN115539220B (en) | 2022-10-08 | 2022-10-08 | Gas turbine blade channel temperature sensor fault detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115539220B (en) |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090228230A1 (en) * | 2008-03-06 | 2009-09-10 | General Electric Company | System and method for real-time detection of gas turbine or aircraft engine blade problems |
CN101694182A (en) * | 2009-09-29 | 2010-04-14 | 上海中科清洁能源技术发展中心 | On-line failure diagnosis, prediction and feedback control method of small/medium size gas turbine and device thereof |
KR20140037347A (en) * | 2012-09-17 | 2014-03-27 | 이봉우 | Engine analysis apparatus using oxygen sensor |
CN110513336A (en) * | 2019-09-05 | 2019-11-29 | 哈尔滨电气股份有限公司 | A method for determining the off-line washing time of gas turbines in power stations |
US20190379589A1 (en) * | 2018-06-12 | 2019-12-12 | Ciena Corporation | Pattern detection in time-series data |
US20200141653A1 (en) * | 2018-11-02 | 2020-05-07 | Honeywell International Inc. | Flame analytics system |
CN111441869A (en) * | 2020-03-29 | 2020-07-24 | 至玥腾风科技集团有限公司 | Method and system for starting micro gas turbine |
US20200387797A1 (en) * | 2018-06-12 | 2020-12-10 | Ciena Corporation | Unsupervised outlier detection in time-series data |
CN113834656A (en) * | 2021-08-27 | 2021-12-24 | 西安电子科技大学 | A bearing fault diagnosis method, system, equipment and terminal |
CN113919237A (en) * | 2021-11-02 | 2022-01-11 | 南京凯奥思数据技术有限公司 | Method for online working condition segmentation and fault diagnosis of fan equipment |
-
2022
- 2022-10-08 CN CN202211220744.5A patent/CN115539220B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090228230A1 (en) * | 2008-03-06 | 2009-09-10 | General Electric Company | System and method for real-time detection of gas turbine or aircraft engine blade problems |
CN101694182A (en) * | 2009-09-29 | 2010-04-14 | 上海中科清洁能源技术发展中心 | On-line failure diagnosis, prediction and feedback control method of small/medium size gas turbine and device thereof |
KR20140037347A (en) * | 2012-09-17 | 2014-03-27 | 이봉우 | Engine analysis apparatus using oxygen sensor |
US20190379589A1 (en) * | 2018-06-12 | 2019-12-12 | Ciena Corporation | Pattern detection in time-series data |
US20200387797A1 (en) * | 2018-06-12 | 2020-12-10 | Ciena Corporation | Unsupervised outlier detection in time-series data |
US20200141653A1 (en) * | 2018-11-02 | 2020-05-07 | Honeywell International Inc. | Flame analytics system |
CN110513336A (en) * | 2019-09-05 | 2019-11-29 | 哈尔滨电气股份有限公司 | A method for determining the off-line washing time of gas turbines in power stations |
CN111441869A (en) * | 2020-03-29 | 2020-07-24 | 至玥腾风科技集团有限公司 | Method and system for starting micro gas turbine |
CN113834656A (en) * | 2021-08-27 | 2021-12-24 | 西安电子科技大学 | A bearing fault diagnosis method, system, equipment and terminal |
CN113919237A (en) * | 2021-11-02 | 2022-01-11 | 南京凯奥思数据技术有限公司 | Method for online working condition segmentation and fault diagnosis of fan equipment |
Non-Patent Citations (2)
Title |
---|
于美玲;刘永文;: "基于非线性模型的燃气轮机热参数故障诊断及仿真", 热能动力工程, no. 06, 20 November 2011 (2011-11-20) * |
刘钰宸;安静;: "一种基于lightGBM框架改进的GBDT风力发电机叶片开裂预测方法", 应用技术学报, no. 01, 30 March 2020 (2020-03-30) * |
Also Published As
Publication number | Publication date |
---|---|
CN115539220B (en) | 2024-07-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109240244B (en) | Data-driven equipment running state health degree analysis method and system | |
CN111159844B (en) | A Method for Abnormal Detection of Gas Turbine Exhaust Temperature in Power Station | |
CN111539553A (en) | Wind turbine generator fault early warning method based on SVR algorithm and skewness | |
CN110513336B (en) | A method for determining off-line water washing time of gas turbine in power station | |
CN111780800B (en) | A sensor fault monitoring, isolation and reconstruction method and system | |
CN114397526A (en) | State holographic sensing data-driven power transformer fault prediction method and system | |
CN105867132B (en) | Combined cycle unit oxygen content in exhaust smoke On-line Measuring Method of the stable state tracking from correction | |
CN116292367B (en) | Power plant fan system abnormal condition detection method based on one-dimensional convolution | |
CN114757380A (en) | A thermal power plant fault early warning system, method, electronic device and storage medium | |
CN111160457B (en) | Scroll engine fault detection method based on soft-class extreme learning machine | |
CN117829342A (en) | Coal-fired power plant nitrogen oxide emission prediction method based on improved random forest algorithm | |
CN109469919B (en) | Power station air preheater ash blocking monitoring method based on weight clustering | |
CN106596110B (en) | The automatic analyzing and diagnosing method of turbine-generator units waterpower imbalance fault based on online data | |
CN115687864A (en) | XGboost-KDE-based fan main bearing fault early warning method | |
CN115906437A (en) | Fan state determination method, device, equipment and storage medium | |
CN117951633B (en) | Photovoltaic power generation equipment fault diagnosis method and system | |
CN115539220B (en) | Gas turbine blade channel temperature sensor fault detection method | |
CN118332313A (en) | An online detection method for faults of sensors used to monitor the operating status of a combined cycle unit | |
CN118897513A (en) | An intelligent ash conveying energy-saving control system and method for a thermal power plant based on big data analysis | |
CN112130468A (en) | Device and method for simulating temperature of outer surface of high-pressure rotor of nuclear turbine | |
CN117369405A (en) | Automatic control loop fault diagnosis method and system based on clustering and migration learning | |
CN116241420A (en) | Fan blade icing fault diagnosis method, system, equipment and storage medium | |
CN114021283A (en) | Gas turbine compressor detection and water washing optimization method | |
Xuefeng et al. | Study on Multi-Sensor Parameter Warning Based on Gaussian Mixed Model and Nonlinear State Estimation Technique | |
CN112508350A (en) | Vibration early warning method and system for steam turbine generator unit |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |