CN116304892B - Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system - Google Patents
Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system Download PDFInfo
- Publication number
- CN116304892B CN116304892B CN202310592049.XA CN202310592049A CN116304892B CN 116304892 B CN116304892 B CN 116304892B CN 202310592049 A CN202310592049 A CN 202310592049A CN 116304892 B CN116304892 B CN 116304892B
- Authority
- CN
- China
- Prior art keywords
- scale
- fault
- feature
- layer
- fluidized bed
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003745 diagnosis Methods 0.000 title claims abstract description 43
- 239000003245 coal Substances 0.000 title claims abstract description 40
- 238000002309 gasification Methods 0.000 title claims abstract description 26
- 239000002245 particle Substances 0.000 claims abstract description 105
- 238000000605 extraction Methods 0.000 claims abstract description 61
- 238000013507 mapping Methods 0.000 claims abstract description 20
- 238000000926 separation method Methods 0.000 claims abstract description 18
- 238000013135 deep learning Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 9
- 230000008859 change Effects 0.000 claims description 20
- 239000012530 fluid Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 13
- 238000005259 measurement Methods 0.000 claims description 12
- 238000009826 distribution Methods 0.000 claims description 9
- 238000001816 cooling Methods 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000005243 fluidization Methods 0.000 claims description 5
- 239000002817 coal dust Substances 0.000 claims description 3
- 238000004939 coking Methods 0.000 claims description 3
- 239000007800 oxidant agent Substances 0.000 claims description 3
- 230000001590 oxidative effect Effects 0.000 claims description 2
- 238000005507 spraying Methods 0.000 claims description 2
- 239000010410 layer Substances 0.000 description 67
- 239000007787 solid Substances 0.000 description 11
- 239000007789 gas Substances 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000005054 agglomeration Methods 0.000 description 4
- 230000002776 aggregation Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 239000003034 coal gas Substances 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 201000004569 Blindness Diseases 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000003325 tomography Methods 0.000 description 2
- 239000002918 waste heat Substances 0.000 description 2
- 102100032215 Cathepsin E Human genes 0.000 description 1
- 241000132179 Eurotium medium Species 0.000 description 1
- 101000869031 Homo sapiens Cathepsin E Proteins 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 239000011335 coal coke Substances 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000003344 environmental pollutant Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 231100000719 pollutant Toxicity 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000005514 two-phase flow Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/06—Investigating concentration of particle suspensions
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M10/00—Hydrodynamic testing; Arrangements in or on ship-testing tanks or water tunnels
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Biochemistry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Analytical Chemistry (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Dispersion Chemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Fluid Mechanics (AREA)
- Fluidized-Bed Combustion And Resonant Combustion (AREA)
Abstract
The invention discloses a method and a device for diagnosing multi-scale flow state faults of a circulating fluidized bed in a coal gasification system, wherein the method comprises the following steps: s01, acquiring a feature set of the particle concentration of the circulating fluidized bed in a plurality of fault modes, wherein the feature set of the particle concentration changes with time and space, so as to form a particle concentration fault feature set; s02, performing scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features; s03, performing multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning to obtain an association mapping relation between each scale and a fault mode, and training to form a fault diagnosis recognition module for realizing fault diagnosis of the multi-scale flow state of the circulating fluidized bed. The invention can realize the fault diagnosis of the multi-scale flow state of the circulating fluidized bed and has the advantages of simple realization method, low cost, high diagnosis efficiency, high diagnosis precision and the like.
Description
Technical Field
The invention relates to the technical field of coal gasification systems, in particular to a method and a device for diagnosing multi-scale flow state faults of a circulating fluidized bed in a coal gasification system.
Background
The coal is mainly burnt directly, so that the efficiency is low, and gas and solid pollutants generated in the process can cause adverse effects on the ecological environment. The coal gasification technology is one of the important ways of high-efficiency clean utilization of coal, and the gasification process is a thermochemical processing process of coal, namely a process of converting coal into combustible gas through chemical reaction at high temperature and high pressure by taking coal or coal coke as raw materials and taking oxygen and water vapor as gasifying agents. The coal gasification system mainly comprises a gasification furnace (comprising a wind distribution device, a lifting pipe, a cyclone separator, a vertical pipe, a U-shaped material returning device, a material returning inclined pipe and the like), a coal gas waste heat recovery system (an air preheater, a waste heat boiler and the like), a coal gas dust removal system, a coal gas cooling system (a heat exchanger, a cooler and the like) and an auxiliary system, wherein the auxiliary system comprises a coal feeding system, a wind supply system, an ash cooling system, a circulating water system and the like. The core unit of the coal gasification system is a fluidized bed, wherein the Circulating Fluidized Bed (CFB) has been widely used because of the advantages of strong heat and mass transfer, uniform temperature distribution, large equipment production capacity, uniform mixing of solid particles, wide residence time adjustment range of the solid particles in the bed, easy enlargement, etc.
The circulating fluidized bed system internally comprises a coal supply system, a wind supply system, a water supply system, a cooling system and other numerous structures, and the structures are numerous and quite complex, so that when faults occur, the faults are probably caused by various reasons, the circulating fluidized bed needs to work at high temperature and high pressure, and the faults cannot be found out immediately in normal cases. In addition, another important feature of the circulating fluidized bed is that the solid particles are flowing along a set circulation loop, any failure that causes the particles to circulate can cause an unintended shutdown of the entire plant. Therefore, in order to ensure a long-period stable operation of the circulating fluidized bed, it is necessary to perform state monitoring and intelligent fault diagnosis on the circulating process of the particles inside the circulating fluidized bed.
For intelligent fault monitoring of a circulating fluidized bed, in the prior art, a sensor is usually arranged in the circulating fluidized bed, and intelligent fault diagnosis is realized based on a shallow learning model by using data detected by the sensor, such as a learning model of an Artificial Neural Network (ANN), a Support Vector Machine (SVM), particle Swarm Optimization (PSO), a Bayesian network and the like. However, the above intelligent fault diagnosis method based on the shallow learning model has the following problems:
1. the flow characteristics of the multi-scale particles cannot be accurately characterized. The operation of circulating fluidized bed particles is a typical gas-solid reaction system, different scales can exist in particle flow characteristics in a bed, for example, single particle size, particle agglomeration size and equipment size can exist, and the particle flow characteristics can exist in time and space superposition, namely, the particle flow characteristics not only comprise time, but also comprise space characteristics, the data detected by a sensor are directly used for model construction, the multi-scale particle flow characteristics of the circulating fluidized bed cannot be characterized, the mutual mapping relation between the multi-scale particle flow characteristics and a fault mode cannot be characterized, the fault diagnosis precision is low, and the multi-scale particle flow characteristics of the circulating fluidized bed are difficult to extract, so that the fault diagnosis difficulty of the fluidized bed is increased.
2. The diagnostic accuracy and efficiency are low. The traditional intelligent fault diagnosis method based on the shallow learning model can only realize fault diagnosis and classification aiming at single fault source data learning, namely model learning and fault diagnosis can only be carried out on data of a certain device in the circulating fluidized bed at a time, and multi-scale flow state fault on-line diagnosis of the circulating fluidized bed can not be realized rapidly and accurately.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the technical problems existing in the prior art, the invention provides the method and the device for diagnosing the multi-scale flow state faults of the circulating fluidized bed in the coal gasification system, which have the advantages of simple implementation method, low cost, high diagnosis efficiency and high diagnosis precision.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
a method for diagnosing multi-scale flow state faults of a circulating fluidized bed in a coal gasification system comprises the following steps:
s01, acquiring a feature set of the particle concentration of the circulating fluidized bed in a plurality of fault modes, wherein the feature set of the particle concentration changes with time and space, so as to form a particle concentration fault feature set;
s02, performing scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features;
s03, performing multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning to obtain an association mapping relation between each scale and a fault mode, and training to form a fault diagnosis identification module for realizing fault diagnosis of the multi-scale flow state of the circulating fluidized bed.
Further, in the step S01, the fault mode includes a device fault and a condition fault, the device fault includes a fault of any one or more devices of a circulating fluidized bed, a fan plate hole, a wind supply system, a water supply system and a cooling system, and the condition fault includes a mismatch of a ratio of coal particles to an oxidant; the step S01 also comprises the steps of adjusting a gas valve, a gas supply system and a feeding system to perform circulating fault operation on the circulating fluidized bed, and/or simulating a coking fault forming process in the circulating fluidized bed by continuously increasing the proportion of coal dust and/or simulating the fault forming process by spraying high-pressure gas into a nozzle to break particles.
Further, the step S02 includes:
decomposing the raw particle concentration signal into detail signals n of different scale levels i I represents the scale level, represented by high frequency ripple, low level n i Corresponds to microscale, n is the intermediate level between high frequency and low frequency i Corresponding to a mesoscale, n of low frequency fluctuation and high frequency level i The corresponding macro scale;
and dividing the particle concentration into scales according to the correlation, the time delay and the spatial variation between the particle concentration measurement signals of two adjacent sections in the circulating fluidized bed so as to divide the scales by integrating the correlation, the time delay and the spatial variation.
Further, the following formula is adopted as a scale division criterion：
Wherein,,iat the level of the different dimensions,c i as the correlation coefficient(s),vis a space change, namely the volume of two points at adjacent positions,τin order for the time delay to be a time delay,c m at the maximum value of the correlation coefficient,τ m for the time delay of the mesoscale signal,v m is the maximum value of the spatial variation.
Further, the method also comprises the step of dividing the scales into criteriaMapping to probability framework sigmoid according to the following criterion after mapping>To describe the interdependence between three variables of correlation, delay and spatial variation;
wherein the method comprises the steps of、/>And->Respectively time delaysτ i Correlation coefficientc i And spatial variationv i Normalized parameters, namely:
c 0 、τ 0 、v 0 respectively minimum values of correlation coefficient, time delay and space change;
according to the mapped criterionIs scaled according to the size of (1), wherein when the mapped criterion is + ->Deviation from 1 is smaller than a preset threshold value and +.>Is greater than a preset first scale threshold s 1 The time is divided into mesoscales, when +.>At [ s ] 2 ,s 1 ]When in range, the micro-scale is divided into micro-scale, when +.>At [0.5, s 2 ]When in range, the macro scale is divided into s 2 Is the second scale threshold, and +.>The method comprises the following steps:
。
further, in the step S03, the information entropy of each feature in the feature set after separation is calculated, the calculated information entropy is input into the multi-layer self-encoder for multi-layer feature extraction, wherein when the information entropy of each feature is calculated, the probability distribution of each data point of the input time sequence falling into each fault interval is estimated, the fault interval is the value range of feature data under different pre-divided fault modes, and the shannon entropy probability on different scales is calculated according to the obtained probability distribution of each fault intervalP（x n ）。
Further, each layer of the multi-layer self-encoder is codedThe encoder adopts the reconstruction error between the decoding result and the input signal as an evaluation index, and the expression of the reconstruction error is as followsWhereinyIn order to decode the result of the decoding,xfor the input signal, H is a dimension, H includes a time dimension and a space dimension.
Further, the multi-layer self-encoder comprises a first-layer self-encoder, a second-layer self-encoder and a fault classifier, the first-layer self-encoder inputs information entropy obtained by calculating each feature in the separated feature set to perform first-layer encoding, a first-layer extraction feature is obtained, the first-layer encoding result is decoded and then is evaluated by using a reconstruction error between the first-layer decoding result and the input information entropy, the first-layer extraction feature is input into the encoder of the second-layer self-encoder to perform second-layer encoding, a second-layer extraction feature is obtained, the second-layer encoding result is decoded and then is evaluated by using a reconstruction error between the second-layer decoding result and the input of the second-layer encoder, the fault classifier comprises a feature extraction network and a classifier network, the first-layer extraction feature and the second-layer extraction feature are input into the feature extraction network, feature extraction is performed by using unsupervised learning, a multi-scale information entropy feature set is formed, and the abstract feature entropy is used for training the abstract feature set, and the fault classifier is obtained.
A circulating fluidized bed multi-scale flow regime fault diagnosis device in a coal gasification system, comprising:
the fault feature set acquisition module is used for acquiring feature sets of the particle concentration of the circulating fluidized bed under a plurality of fault modes, wherein the feature sets change with time and space to form a particle concentration fault feature set;
the scale separation module is used for carrying out scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features;
and the multi-level feature extraction module is used for carrying out multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning, obtaining the association mapping relation among all scales, and constructing and forming a fault diagnosis and identification module for realizing fault diagnosis of the multi-scale flow state of the circulating fluidized bed.
A computer apparatus comprising a processor and a memory for storing a computer program, the processor being for executing the computer program to perform a method as described above.
Compared with the prior art, the invention has the advantages that:
1. according to the method, the fault feature set is formed by measuring the concentration of the circulating particles which changes along with time and space under a plurality of fault modes in the circulating fluidized bed, so that the fault feature set simultaneously comprises the time and space features, then the scale separation is carried out on the fault feature set to obtain macro-scale features, medium-scale features and micro-scale features, the multi-scale particle flow features of the circulating fluidized bed under different fault states can be fully excavated, multi-scale feature extraction and data fusion are carried out on multi-scale circulating fluidized bed fault data by utilizing a deep-learning multi-layer self-encoder, the mapping relation between the fault modes and each scale can be fully represented, the accurate fault identification model of the circulating fluidized bed is built, the intelligent on-line monitoring and identification of the multi-scale fluid state fault of the circulating fluidized bed can be rapidly and accurately realized, the running stability of the circulating fluidized bed equipment is improved, the production efficiency of a coal gasification system and the conversion rate from solid coal to gaseous coal are further facilitated, and the occurrence rate of production accidents and the energy consumption rate are reduced.
2. According to the invention, three factors of correlation, time delay and space change are comprehensively considered to carry out scale division, so that the characteristics of different scales on the three factors of correlation, time delay and space change are effectively represented, the interdependence relationship among three variables can be effectively described, the fluid dynamics characteristic in gas-solid flow can be more specifically described, the accuracy of the scale division of the particle flow characteristics of the circulating fluidized bed is improved, and the accurate fluid dynamics analysis of a complex system is facilitated; further, through mapping the scale division criteria to the probability framework sigmoid, the interdependence relationship among three variables of correlation, time delay and space change can be quantitatively described, so that the scale division of the circulating fluidized bed particle flow characteristics can be rapidly and accurately realized.
3. The invention further adopts the multi-layer self-encoder with a three-layer structure, can effectively abstract and fuse the characteristic information of the multi-scale fault characteristic set layer by layer, can realize information complementation and restriction among the characteristics of each layer, so that more detailed and perfect abstract characteristics can be obtained, the mapping relation between the fault mode and each scale can be accurately and comprehensively represented, the construction precision of the fault identification model can be further improved, the self-perception and self-regulation characteristic of a decision network can be further realized, the blindness and subjectivity of manually determined parameters can be effectively avoided, the rapid analysis, comparison and processing of monitoring data can be realized, further, the fault point can be found timely, the diagnosis speed can be improved, the fault time can be shortened, and the damage caused by faults can be effectively reduced.
Drawings
Fig. 1 is a schematic diagram of an implementation flow chart of a method for diagnosing a multi-scale fluidization of a circulating fluidized bed in a coal gasification system according to the embodiment.
Fig. 2 is a detailed flow chart of the circulating fluidized bed fault diagnosis method in the embodiment.
Fig. 3 is a schematic diagram of a circulating fluidized bed fault feature collection implementation in a specific application embodiment of the present invention.
Fig. 4 is a schematic diagram of the structure of a multi-layer self-encoder in an embodiment of the present invention.
Detailed Description
The invention is further described below in connection with the drawings and the specific preferred embodiments, but the scope of protection of the invention is not limited thereby.
As used in this disclosure, the terms "a," "an," "the," and/or "the" are not intended to be limiting, but rather are to be construed as covering the singular and the plural, unless the context clearly dictates otherwise. The terms "first," "second," and the like, as used in this disclosure, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. Likewise, the word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items.
In consideration of multi-scale particle flow characteristics of a circulating fluidized bed and time and space characteristics of the particle flow characteristics, the method firstly measures circulating particle concentration which changes along with time and space under a plurality of fault modes in the circulating fluidized bed to form a fault characteristic set, so that the fault characteristic set simultaneously comprises time and space characteristics, then scale separation is carried out on the fault characteristic set to obtain macro-scale characteristics, medium-scale characteristics and micro-scale characteristics, multi-scale particle flow characteristics of the circulating fluidized bed under different fault states can be fully excavated, then characteristic extraction and data fusion are carried out on multi-scale circulating fluidized bed fault data by utilizing a deep-learning multi-layer self-encoder, and a mapping relation between the fault modes and macro-scale, medium-scale and micro-scale is obtained, thereby establishing a circulating fluidized bed fault identification model, realizing intelligent on-line monitoring and identification of the multi-scale fluid fault of the circulating fluidized bed rapidly and accurately, improving the running stability of circulating fluidized bed equipment, and being beneficial to improving the production efficiency of a solid coal gasification system and the conversion rate of the solid coal to the gaseous coal, and reducing the occurrence rate of production accidents and energy consumption rate.
As shown in fig. 1 to 4, the method for diagnosing a multi-scale flow state fault of a circulating fluidized bed in a coal gasification system according to the embodiment includes the following steps:
s01, constructing a fault feature set: and acquiring a feature set of the particle concentration of the circulating fluidized bed in a plurality of fault modes, wherein the feature set of the particle concentration changes with time and space, so as to form a particle concentration fault feature set.
In this embodiment, the failure modes include equipment failures and condition failures, and the equipment failures include structural failures of the circulating fluidized bed, the louver holes, the air supply system, the water supply system, the cooling system, and the like, and part failures of the inlet angle, the louver position, the air flow nozzle speed, and the like, such as louver hole blockage, louver mounting too high or tilting, CFB liner falling off, the air supply system, the water supply system, the cooling system, and the like. For equipment faults, the CFB equipment parts in various fault modes can be designed and processed in advance to replace corresponding parts in a CFB experiment platform so as to simulate the equipment faults. The circulating fluidized bed can be operated in a circulating fault mode by adjusting the air valve, the air supply system and the feeding system.
In this embodiment, the condition faults mainly refer to the ratio of pulverized coal to gasifying agent, and specifically include faults such as inconsistent ratio of coal particles to oxidizing agent. For the condition faults, the proportion of coal dust can be continuously increased to simulate the series fault forming processes such as coking and the like in the CFB. In addition, high-pressure gas can be sprayed into the nozzle, so that collision among particles is increased, the particles are broken, the fault forming process of the fault process is simulated, and further, the fault information characteristics of all scales can be effectively obtained.
In a specific application embodiment, as shown in fig. 3, a measurement platform (ERT) based on a resistance tomography is first constructed, and a data acquisition module is set at positions of P1-P5 (coal system, circulating fluidized bed, air supply system, water supply system, and cooling system) in the CFB, so as to acquire concentration information of fault particles in different fault modes such as structural faults, import-export condition faults and the like in the CFB, obtain circulating particle concentrations in multiple fault modes in the CFB, and construct a multi-scale data set of multiple fault modes. Further, the particle concentration of the P1-P4 section inside the CFB is measured in a certain time period (3-10 min) under different fault modes by utilizing a tomography system, and the image reconstruction is carried out by utilizing an image reconstruction algorithm, so that the characteristic set of the particle phase concentration along with time and space can be effectively obtained to be used as a fault data set.
Compared with the traditional fault diagnosis through the data set characteristics of the differential pressure signal source, the method can acquire comprehensive characteristic data of particle concentration of the circulating fluidized bed in a plurality of fault modes through the steps, and the particle physical characteristic sets simultaneously contain time and space characteristics, so that the time and space characteristics of the particle concentration characteristics in each fault state of the CFB can be effectively represented, and further the excavation of association relations of macroscopic scale, microscale and mesoscale is facilitated.
S02, scale separation: and performing scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features.
In order to describe the interactions and high-speed impact enhancement between particles and walls, to break up, refine, refocus, etc. the particles in the bed are formed in different sizes, the present embodiment divides the particle concentration characteristics into three dimensions, i.e. microscale for single particle size, mesoscale for particle agglomeration size, and macroscale for device size.
Step 2.1, decomposing the raw particle concentration signal into detail signals n of different scale levels by wavelet analysis i I represents the scale level, and dividing the particle concentration signal in the gas-solid two-phase flow into three scales is equivalent to dividing the detail signals of different frequency bands or levels into microscale, mesoscale and macroscale.
In particular in circulating fluidized beds, high frequency fluctuations in the microscale corresponding signal representing particle size, i.e. low level n i The method comprises the steps of carrying out a first treatment on the surface of the Mesoscale representing particle agglomeration corresponds to a moderate level of n i I.e. medium level n i The method comprises the steps of carrying out a first treatment on the surface of the Representing low-frequency fluctuations in the signal corresponding to the macro-scale of the device, i.e. high levels of n i The frequency is lower and the fluctuation is more gentle.
And 2.2, dividing the particle concentration into scales according to the correlation, the time delay and the spatial variation between the particle concentration measurement signals of two adjacent sections in the circulating fluidized bed, so as to divide the scales according to the comprehensive correlation, the time delay and the spatial variation.
For the operation of the fluidized bed particles, the parameters such as the particle size distribution, the particle concentration and the like of the particles change to different degrees every time the particles are circulated, so that the information of the particles comprises the characteristics of time and space. And because the particle concentration characteristics are divided into microscale, mesoscale and macroscale, namely microscale of single particle size formed by particles in the bed, mesoscale of particle agglomeration size and macroscale of equipment size, correlation, time delay and space change performance can be distinguished under different scales. Taking the agglomerate as an example, if the spatial distance between different points of adjacent cross-sections in the fluid bed is in the same order of magnitude as the length of the agglomerate, then the movement of the agglomerate can be measured through these two cross-section points. For a non-contact particle concentration measurement, its two adjacent cross-sectional measurement signals have the following characteristics in three dimensions: for macro scale, the correlation of two measurement signals is strong, the space is delayed, and no time delay exists; for the mesoscale, the correlation of two measurement signals is good, the space delay and the time delay are obvious; for microscale, the two measurement signals are completely uncorrelated, spatially delayed, with no delay. That is, the scale division of the signals needs to comprehensively consider three factors of correlation, time delay and space variation.
In the embodiment, three factors including correlation, time delay and space change are comprehensively considered to conduct scale division, the characteristics of different scales on the three factors including correlation, time delay and space change are effectively represented, the mutual dependence relationship among the three variables can be effectively described, compared with a traditional method for utilizing wavelet energy multi-scale division basis or average value, the fluid dynamics characteristic in gas-solid flow can be described in more detail, accuracy of scale division of the flow characteristics of the particles of the circulating fluidized bed is improved, and accurate fluid dynamics analysis of a complex system is facilitated.
In this embodiment, a correlation-delay-space equation (Correlation Time Space Equation, CTSE) is constructed specifically according to the following formula (1) as a scale division criterion。
（1）
Wherein,,iat the level of the different dimensions,cas the correlation coefficient(s),vis a space change, namely the volume of two points at adjacent positions,τin order for the time delay to be a time delay,c m at the maximum value of the correlation coefficient,τ m for the time delay of the mesoscale signal,v m is the maximum value of the spatial variation.The meaning of the equation is calculated on each scale levelτ i , c i , v i ) And%τ m , c m , v m ) I.e. each scale is measured using the spatial volume as a scale division criterion. Because the spatial variation is the volume between two points at the adjacent positions, the embodiment adopts a spatial volume mode to measure each scale, and can more fully excavate the spatial variation characteristics of different scales on the basis of excavating the correlation and time delay characteristics of different scales, thereby ensuring the accuracy of scale division.
The correlation coefficient is a statistical index describing the degree and direction of linear correlation between three variables. The two signals have similar fluctuation and no negative correlation, and the value range of the correlation coefficient is [0,1 ]]A closer to 1 indicates a higher similarity of the two signals, with minimum values of correlation coefficient, delay and spatial variation beingc 0 、τ 0 Andv 0 ，dfor the space distance between two adjacent points, the following calculation can be performedv m ：
（2）
Further normalize each parameter to make the value range of each parameter be [0,1 ]]Can obtain normalized time delayCorrelation coefficient->And spatial variation->：
（3）
The embodiment further uses normalized time delayCorrelation coefficient->And spatial variation->Metric dividing criterion->Mapping to probability framework sigmoid according to the following criterion after mapping>To enable quantitative description of the interdependence between the three variables of correlation, time delay and spatial variation, so that finally, according to the mapped criteriaCan be rapidly scaled.
（4）
Due to time delayCorrelation coefficient->And spatial variation->Are all greater than 0, therefore->The value range is [0.5,1]The present embodiment is specifically described as the post-mapping criterion +.>Deviation from 1 is smaller than a preset threshold value and +.>Is greater than a preset first scale threshold s 1 The time is divided into mesoscales, when +.>At [ s ] 1 ,1]When in range, the micro-scale is divided into micro-scale, when +.>At [ s ] 2 , s 1 ]When in range, the macro scale is divided into s 2 Is the second scale threshold, and +.>I.e. can be expressed as:
(5)
the following characteristics may exist due to time delays, correlation coefficients, and spatial variations at different scales: when (when)v i When the correlation coefficient and the delay value are very small, only the correlation coefficient and the delay value approach at the same timeτ m Andc m at each decomposition levelτ i ，c i ，v i ) The closer the point isτ m ，c m ) Horizontal leveliThe more likely it is that it will be of the mesoscale,approximately 1; microscale signalcThe value of (2) is the smallest, macro-scale signalτThe value of (2) is the smallest; slowly increasev i The correlation coefficient and the time delay value of the mesoscale approach at the same timeτ m Andc m ，/>approximately 1; microscale signalcThe value of (c) is also 0,τslowly increasing, ++>Slowly increasing, macro scale letterNumber of the signτ、The value of c is gradually increased, so that the sigmoid function image changes with the parameters, ++>The macro scale increases faster and the micro scale increases slower, the embodiment uses the characteristics to determine the specific range of each scale, and adopts the division rule as described in (5) to divide the scale, namely when->Is classified into micro-mesoscale when 1 is left or right, when +.>Microscale around 0.75, when +.>Macro scale is set near 0.5-0.75, and the range can be adjusted and set according to actual requirements.
（6）
According to the particle concentration fault feature set, the measured particle concentration is subjected to scale division by adopting the scale division mode based on correlation, time delay and space change, and the factors of three variables including correlation, time delay and space change are integrated, so that the particle flow feature scale division of the circulating fluidized bed can be rapidly and accurately realized.
S03, extracting multi-level features: and carrying out multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning to obtain an association mapping relation between each scale and a fault mode, and constructing and forming a fault diagnosis and identification module for realizing fault diagnosis of the multi-scale flow state of the circulating fluidized bed, namely, inputting real-time collected circulating fluidized bed operation data by using the fault diagnosis and identification module, namely, diagnosing whether faults occur or not and the fault type when the faults occur.
In this embodiment, information of each feature in the feature set after separation is calculated firstEntropy, inputting the calculated information entropy into a multi-layer self-encoder for multi-layer feature extraction, wherein when the information entropy of each feature is calculated, as the input is a sequence containing time and space, firstly estimating probability distribution of each data point of the input time sequence falling into C fault intervals, wherein the fault intervals are the value ranges of the feature data under different pre-divided fault modes, and then calculating shannon entropy probabilities on different scales according to the obtained probability distribution of each fault intervalP（x n ) To achieve entropy calculations involving time and space. Wherein shannon entropy probability P (x n ) Larger means that the time and space feature sets of the corresponding particles are better corresponding to a certain fault class, so that the smaller the amount of information contained is, whereas the smaller the probability is, the larger the amount of information is.
In this embodiment, information entropy is used as a basic physical quantity of information measurement, and since the input is multi-element information, a generalized information entropy calculation rule needs to be used, and a description form of the generalized information entropy is as follows:
（7）
an unsupervised learning-based self-Encoder is a deep learning-based feature engineering network model, and consists of an Encoder (Encoder) and a Decoder (Decoder). The self-encoder reconstructs the input data at the output end, takes the result of the intermediate hidden layer as the characteristic code, and indicates that the better the coding effect of the intermediate hidden layer on the input data is, the more complete the input data can be coded. As shown in fig. 4, the multi-layer self-encoder in this embodiment includes three layers of a first layer self-encoder, a second layer self-encoder and a fault classifier, the first layer self-encoder inputs information entropy obtained by calculating each feature in the separated feature set to perform first layer encoding to obtain a first layer extraction feature, decodes the first layer encoding result, performs evaluation on the first layer encoding result by using a reconstruction error between the first layer decoding result and the input information entropy to determine whether to relearn, inputs the first layer extraction feature into the encoder of the second layer self-encoder to perform second layer encoding to obtain a second layer extraction feature, decodes the second layer encoding result, performs evaluation on the second encoding result by using a reconstruction error between the second decoding result and the input of the second layer encoder to determine whether to relearn, the fault classifier includes a feature extraction network and a classifier network, inputs the first layer extraction feature and the second layer extraction feature into the feature extraction network to perform feature extraction by using unsupervised learning to form a multi-scale abstract entropy information abstraction feature set, and performs training of the fault abstract feature set by using the abstract entropy, and performs recognition of the fault abstract feature set. The fault classifier is characterized in that various known fault states correspond to an output layer of the feature extraction network, the output layer is subjected to classification training and error back propagation for a plurality of times, and finally a fault recognition model is constructed and formed, wherein the fault recognition model comprises association mapping relations between different scale features and different fault modes. After the state data of the circulating fluidized bed is acquired in real time, the state data is input into a fault identification model, and diagnosis of fault types and fault positions can be achieved.
According to the embodiment, the multi-layer self-encoder with the three-layer structure is adopted, the information entropy of the multi-scale fault feature set is input first to be encoded through the first-layer encoder, the first-layer feature extraction is achieved, the encoding result of the first-layer encoder is encoded through the second-layer encoder, the second-layer feature extraction is achieved, the first-layer extraction features and the second-layer extraction features are subjected to unsupervised learning to conduct feature extraction again, the multi-scale information entropy abstract feature set is formed, the layer-by-layer abstract fusion of macro-scale, medium-scale and micro-scale features can be achieved respectively, information complementation and restriction can be achieved among the features of all layers, compared with single-layer information, more detailed and perfect abstract features can be obtained, and therefore the mapping relation between a fault mode and all scales can be represented accurately and comprehensively, and the construction precision of a fault diagnosis model is further improved.
According to the embodiment, the multi-layer self-encoder based on deep learning is adopted, so that the characteristic information of the multi-scale fault characteristic set contained in the CFB unit can be effectively abstracted and fused, the self-sensing and decision network self-adjusting characteristics are realized, the blindness and subjectivity of manually determining parameters can be effectively avoided, the model can be used for realizing rapid analysis, comparison and processing of monitoring data, fault points can be found in time, the diagnosis speed is improved, the fault time is shortened, and the damage caused by faults is effectively reduced.
The encoding process in the self-encoder is to find a certain sparse representation of entropy values of various information, and in this embodiment, each layer of encoder of the multi-layer self-encoder further includes using a reconstruction error between the decoding result y and the input signal x as an evaluation index to determine whether to need to be relearned, and since the multi-element information is introduced, the reconstruction error also needs to be evaluated in a multi-scale space, and the actual fact is a redefined generalized distance, and if the dimension is H:
（8）
wherein the method comprises the steps ofyIn order to decode the result of the decoding,xfor the input signal, H is a dimension, H includes a time dimension and a space dimension.
Compared with the information entropy of the traditional time sequence, the embodiment can make the dimension of the information entropy richer and more diverse by increasing the space dimension, and further make the reconstruction error evaluation index more accurate.
The fault diagnosis of the invention includes fault location, fault degree prediction, and the like.
The multi-scale fluid state fault diagnosis device of the circulating fluid bed in the coal gasification system of the embodiment comprises:
the fault feature set acquisition module is used for acquiring feature sets of the particle concentration of the circulating fluidized bed under a plurality of fault modes, wherein the feature sets change with time and space to form a particle concentration fault feature set;
the scale separation module is used for carrying out scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features;
and the multi-level feature extraction module is used for carrying out multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning, obtaining the association mapping relation between each scale and the fault mode, and training to form a fault diagnosis and identification module for realizing fault diagnosis of the multi-scale flow state of the circulating fluidized bed.
In this embodiment, an ERT measurement platform is specifically adopted to collect a fault feature set, as shown in fig. 2, the scale separation module and the multi-level feature extraction module may be implemented by using a processor, that is, the processor performs scale separation and multi-level feature extraction according to the fault feature set to obtain an integrated fault identification model, and by using the model, online diagnosis and identification of fault types and degrees can be performed, and by using the integrated model, maintenance measures can be conveniently and rapidly formulated, so that equipment maintenance, fault prediction and optimization of a production line are conveniently implemented.
The embodiment also provides a computer device, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the processor is used for executing the computer program to execute the multi-scale fluid state fault diagnosis method of the circulating fluidized bed in the coal gasification system.
The foregoing is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Therefore, any simple modification, equivalent variation and modification of the above embodiments according to the technical substance of the present invention shall fall within the scope of the technical solution of the present invention.
Claims (8)
1. A method for diagnosing a multi-scale fluid state fault of a circulating fluid bed in a coal gasification system is characterized by comprising the following steps:
s01, acquiring a feature set of the particle concentration of the circulating fluidized bed in a plurality of fault modes, wherein the feature set of the particle concentration changes with time and space, so as to form a particle concentration fault feature set;
s02, performing scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features;
s03, performing multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning to obtain an association mapping relation between each scale and a fault mode, and training to form a fault diagnosis identification module for realizing fault diagnosis of a multi-scale flow state of the circulating fluidized bed;
the multi-layer self-encoder comprises a first-layer self-encoder, a second-layer self-encoder and a fault classifier, wherein the first-layer self-encoder inputs information entropy obtained by calculating each feature in the separated feature set to perform first-layer encoding to obtain a first-layer extraction feature, decodes the first-layer encoding result, performs feature extraction by using a reconstruction error between the first-layer decoding result and input information entropy, inputs the first-layer extraction feature into the encoder of the second-layer self-encoder to perform second-layer encoding to obtain a second-layer extraction feature, decodes the second-layer encoding result, performs feature extraction by using a reconstruction error between the second-layer decoding result and the input of the second-layer encoder to perform evaluation on the second-layer encoding result, and the fault classifier comprises a feature extraction network and a classifier network, inputs the first-layer extraction feature and the second-layer extraction feature into the feature extraction network to perform feature extraction by using unsupervised learning to form a multi-scale information abstract feature entropy, and performs feature extraction by using a multi-scale information abstract feature entropy, and performs a training model to obtain a fault recognition model;
in the step S02, the particle concentration is classified into various scales according to the correlation, time delay and spatial variation between the particle concentration measurement signals of two adjacent sections in the circulating fluidized bed, and the scales are classified according to the comprehensive correlation, time delay and spatial variation, and the following formula is adopted as a scale classification criterion：
Wherein,,iat the level of the different dimensions,cas the correlation coefficient(s),vis a space change, namely the volume of two points at adjacent positions,τin order for the time delay to be a time delay,c m at the maximum value of the correlation coefficient,τ m for the time delay of the mesoscale signal,v m is the maximum value of the spatial variation.
2. The method for diagnosing a multi-scale fluidization failure of a circulating fluidized bed in a coal gasification system according to claim 1, wherein in the step S01, the failure mode includes a failure of equipment and a failure of condition, the failure of equipment includes a failure of any one or more of a circulating fluidized bed, a fan plate hole, a wind supply system, a water supply system, and a cooling system, and the failure of condition includes a mismatch of a ratio of coal particles to an oxidant; the step S01 also comprises the steps of adjusting a gas valve, a gas supply system and a feeding system to perform circulating fault operation on the circulating fluidized bed, and/or simulating a coking fault forming process in the circulating fluidized bed by continuously increasing the proportion of coal dust and/or simulating the fault forming process by spraying high-pressure gas into a nozzle to break particles.
3. The method for diagnosing a multi-scale fluidization failure of a circulating fluidized bed in a coal gasification system according to claim 1, wherein in step S02, the raw particle concentration signal is decomposed into detail signals n of different scale levels i ，iRepresenting scale level, by high frequency fluctuation, low level n i Corresponds to microscale, n is the intermediate level between high frequency and low frequency i Corresponding to a mesoscale, n of low frequency fluctuation and high frequency level i Corresponds to the macro scale.
4. A method for diagnosing a multi-scale fluidization fault of a circulating fluidized bed in a coal gasification system according to claim 3, further comprising determining a scale division criterionMapping to a summary as followsRate frame sigmoid mapped criterionTo describe the interdependence between three variables of correlation, delay and spatial variation;
wherein the method comprises the steps of、/>And->Respectively time delaysτ i Correlation coefficientc i And spatial variationv i Normalized parameters, namely:
c 0 、τ 0 、v 0 respectively minimum values of correlation coefficient, time delay and space change;
according to the mapped criterionIs scaled by the size of (1), wherein when the deviation of the mapped criterion from 1 is smaller than a predetermined threshold value and +.>Is greater than a preset first scale threshold s 1 The time is divided into mesoscales, when +.>At [ s ] 2 ,s 1 ]When in range, the micro-scale is divided into micro-scale, when +.>At [0.5, s 2 ]When in range, the macro scale is divided into s 2 Is a second scale threshold value, andthe method comprises the following steps:
。
5. the method for diagnosing a circulating fluidized bed multi-scale fluid state fault in a coal gasification system according to any one of claims 1 to 4, wherein in step S03, the information entropy of each feature in the separated feature set is calculated, the calculated information entropy is input into the multi-layer self-encoder to perform multi-layer feature extraction, wherein when the information entropy of each feature is calculated, probability distribution that each data point of an input time sequence falls into each fault interval is estimated, the fault interval is a value range of feature data under different fault modes divided in advance, and shannon entropy probability on different scales is calculated according to the obtained probability distribution of each fault intervalP（x n ）。
6. The method for diagnosing a multi-scale fluidization fault of a circulating fluidized bed in a coal gasification system according to any one of claims 1 to 4, wherein a reconstruction error between a decoding result and an input signal is adopted as an evaluation index in each layer of encoder of the multi-layer self-encoder, and the expression of the reconstruction error is as followsWhereinyIn order to decode the result of the decoding,xfor the input signal, H is a dimension, H includes a time dimension and a space dimension.
7. A circulating fluidized bed multi-scale flow state fault diagnosis device in a coal gasification system, which is characterized by comprising:
the fault feature set acquisition module is used for acquiring feature sets of the particle concentration of the circulating fluidized bed under a plurality of fault modes, wherein the feature sets change with time and space to form a particle concentration fault feature set;
the scale separation module is used for carrying out scale separation on each particle concentration in the particle concentration fault feature set to obtain a separated feature set, wherein the separated feature set comprises macro-scale features, meso-scale features and micro-scale features;
the multi-level feature extraction module is used for carrying out multi-level feature extraction on the separated feature set by using a multi-level self-encoder based on deep learning to obtain an association mapping relation between each scale and a fault mode, and training to form a fault diagnosis and identification module for realizing fault diagnosis of a multi-scale flow state of the circulating fluidized bed;
the multi-layer self-encoder comprises a first-layer self-encoder, a second-layer self-encoder and a fault classifier, wherein the first-layer self-encoder inputs information entropy obtained by calculating each feature in the separated feature set to perform first-layer encoding to obtain a first-layer extraction feature, decodes the first-layer encoding result, performs feature extraction by using a reconstruction error between the first-layer decoding result and input information entropy, inputs the first-layer extraction feature into the encoder of the second-layer self-encoder to perform second-layer encoding to obtain a second-layer extraction feature, decodes the second-layer encoding result, performs feature extraction by using a reconstruction error between the second-layer decoding result and the input of the second-layer encoder to perform evaluation on the second-layer encoding result, the fault classifier comprises a feature extraction network and a classifier network, inputs the first-layer extraction feature and the second-layer extraction feature into the feature extraction network to perform feature extraction by using unsupervised learning to form a multi-scale information abstract feature entropy, and performs feature extraction by using a multi-scale information abstract feature entropy, and trains a fault recognition model to obtain the fault model;
in the scale separation module, according to circulating fluidizationThe correlation, time delay and space change between the particle concentration measurement signals of two adjacent sections in the bed are used for dividing the particle concentration into scales, the scales are divided by the comprehensive correlation, time delay and space change, and the following formula is adopted as a scale division criterion：
Wherein,,iat the level of the different dimensions,cas the correlation coefficient(s),vis a space change, namely the volume of two points at adjacent positions,τin order for the time delay to be a time delay,c m at the maximum value of the correlation coefficient,τ m for the time delay of the mesoscale signal,v m is the maximum value of the spatial variation.
8. A computer device comprising a processor and a memory for storing a computer program, characterized in that the processor is adapted to execute the computer program to perform the method according to any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310592049.XA CN116304892B (en) | 2023-05-24 | 2023-05-24 | Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310592049.XA CN116304892B (en) | 2023-05-24 | 2023-05-24 | Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116304892A CN116304892A (en) | 2023-06-23 |
| CN116304892B true CN116304892B (en) | 2023-08-01 |
Family
ID=86818984
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310592049.XA Active CN116304892B (en) | 2023-05-24 | 2023-05-24 | Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116304892B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108869145A (en) * | 2018-04-26 | 2018-11-23 | 中国水利水电科学研究院 | Pumping plant unit diagnostic method based on compound characteristics index and depth limit learning machine |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109102032A (en) * | 2018-09-03 | 2018-12-28 | 中国水利水电科学研究院 | A kind of pumping plant unit diagnostic method based on depth forest and oneself coding |
| CN112116029A (en) * | 2020-09-25 | 2020-12-22 | 天津工业大学 | Intelligent fault diagnosis method for gearbox with multi-scale structure and characteristic fusion |
| CN112696667A (en) * | 2020-12-31 | 2021-04-23 | 华电国际电力股份有限公司天津开发区分公司 | Bed temperature early warning system of circulating fluidized bed boiler unit |
| CN112879278B (en) * | 2021-01-11 | 2022-09-30 | 苏州欣皓信息技术有限公司 | Pump station unit fault diagnosis method based on noise signal A weighting analysis |
| EP4152210A1 (en) * | 2021-09-15 | 2023-03-22 | BAE SYSTEMS plc | System and method for training an autoencoder to detect anomalous system behaviour |
| CN114386452B (en) * | 2021-12-06 | 2024-04-05 | 西安交通大学 | Nuclear power circulating water pump sun gear fault detection method |
| CN114280935A (en) * | 2021-12-16 | 2022-04-05 | 北京工业大学 | Multi-stage fermentation process fault monitoring method based on semi-supervised FCM and SAE of information entropy |
| CN115047839B (en) * | 2022-08-17 | 2022-11-04 | 北京化工大学 | Fault monitoring method and system for industrial process of preparing olefin from methanol |
-
2023
- 2023-05-24 CN CN202310592049.XA patent/CN116304892B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108869145A (en) * | 2018-04-26 | 2018-11-23 | 中国水利水电科学研究院 | Pumping plant unit diagnostic method based on compound characteristics index and depth limit learning machine |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116304892A (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Adams et al. | Prediction of SOx–NOx emission from a coal-fired CFB power plant with machine learning: Plant data learned by deep neural network and least square support vector machine | |
| Liu et al. | Principal component analysis-based ensemble detector for incipient faults in dynamic processes | |
| CN102175345B (en) | Soft measurement method for fire box temperature of multi-nozzle opposed coal water slurry gasification furnace | |
| Wei et al. | Research on TE process fault diagnosis method based on DBN and dropout | |
| Hariri-Ardebili et al. | The role of artificial intelligence and digital technologies in dam engineering: Narrative review and outlook | |
| Zhang et al. | Farthest-nearest distance neighborhood and locality projections integrated with bootstrap for industrial process fault diagnosis | |
| CN116050665B (en) | Heat supply equipment fault prediction method | |
| Tang et al. | A deep learning model for measuring oxygen content of boiler flue gas | |
| CN107844659A (en) | The agent model modeling method of coal water slurry gasification process | |
| Wang et al. | Operation optimization of Shell coal gasification process based on convolutional neural network models | |
| Du et al. | Operating mode recognition based on fluctuation interval prediction for iron ore sintering process | |
| CN116520799A (en) | Complex industrial process fault detection method based on space-time variation diagram attention self-encoder | |
| Wang et al. | Smart systems engineering contributing to an intelligent carbon-neutral future: opportunities, challenges, and prospects | |
| CN116304892B (en) | Multi-scale flow state fault diagnosis method and device for circulating fluidized bed in coal gasification system | |
| Belmajdoub et al. | Efficient machine learning model to predict fineness, in a vertical raw meal of Morocco cement plant | |
| Sun et al. | A multimode structured prediction model based on dynamic attribution graph attention network for complex industrial processes | |
| Xiao et al. | Fault diagnosis of unseen modes in chemical processes based on labeling and class progressive adversarial learning | |
| Li et al. | Early warning of critical blockage in coal mills based on stacked denoising autoencoders | |
| Zhu et al. | A novel intelligent model integrating PLSR with RBF-kernel based extreme learning machine: Application to modelling petrochemical process | |
| Ren et al. | Spatial-temporal associations representation and application for process monitoring using graph convolution neural network | |
| Zhang et al. | Monitoring method for gasification process instability using BEE-RBFNN pattern recognition | |
| Movahed et al. | Modeling and optimization of NO emission for a steam power plant by data‐driven methods | |
| Verron et al. | Fault diagnosis with bayesian networks: Application to the tennessee eastman process | |
| Chi et al. | Calculation method of probability integration method parameters based on MIV-GP-BP model | |
| Lv et al. | A spatial–temporal variational graph attention autoencoder using interactive information for fault detection in complex industrial processes |
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 |