CN114550838A - Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace - Google Patents

Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace Download PDF

Info

Publication number
CN114550838A
CN114550838A CN202210176584.2A CN202210176584A CN114550838A CN 114550838 A CN114550838 A CN 114550838A CN 202210176584 A CN202210176584 A CN 202210176584A CN 114550838 A CN114550838 A CN 114550838A
Authority
CN
China
Prior art keywords
temperature
data
layer
organic waste
gasification furnace
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.)
Pending
Application number
CN202210176584.2A
Other languages
Chinese (zh)
Inventor
杨彬
雷乐成
王焕旭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202210176584.2A priority Critical patent/CN114550838A/en
Publication of CN114550838A publication Critical patent/CN114550838A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/10—Analysis or design of chemical reactions, syntheses or processes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • 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
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E50/00—Technologies for the production of fuel of non-fossil origin
    • Y02E50/30—Fuel from waste, e.g. synthetic alcohol or diesel

Abstract

The invention relates to a soft measurement method for the temperature in a multi-component organic waste high-temperature gasification furnace, which selects process variables measured when a plurality of gasification furnaces normally operate, performs data preprocessing, selects BP neural networks with 3 layers, trains the BP neural networks by adopting a levenberg-marquardt algorithm, selects the neural networks with the minimum result error among the neural networks with different hidden layer node numbers as target neural networks, and utilizes the obtained neural networks to realize the soft measurement of the temperature in the multi-component organic waste high-temperature gasification furnace. The method effectively overcomes the defects of uncertain and unstable properties of the traditional temperature physical measurement and the components of the raw materials entering the furnace, can accurately predict the temperature in the multi-component organic waste high-temperature gasification furnace, and has important application value for temperature regulation and stable operation in the waste treatment process.

Description

Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace
Technical Field
The invention belongs to the field of energy industry and chemical engineering, and particularly relates to a soft measurement method for the temperature in a high-temperature gasification furnace in the process of treating multi-component organic waste.
Background
The industrial organic waste mainly comprises solid waste and liquid waste such as rectification residues, waste activated carbon, waste catalysts, industrial sludge, waste tires, waste printed circuit boards, waste organic solvents and the like, and the multi-component organic waste treatment is to cooperatively treat various organic wastes and other raw materials or fuels by using a gasification furnace, so that the harmless treatment and resource utilization of the organic wastes are realized while the normal production requirements of enterprises are met and the product quality and the environmental safety are ensured, meanwhile, the consumption of fossil fuels is reduced, and the emission reduction of carbon is realized. In a gasification furnace, the mixed slurry composed of multi-component organic wastes is pyrolyzed and gasified at high temperature to generate CO2、CO、H2The predominant syngas, inorganic, will melt and solidify into a glassy slag. However, in the actual treatment and utilization process, the organic waste and the coal are prepared into multi-component mixed slurry, and because a large amount of low-boiling-point volatile organic matters with unstable properties are contained in the multi-component mixed slurry, the organic matters are inevitably lost in quality and components after being dried at a certain temperature, so that the conventional all-element analysis and industrial analysis cannot be carried out.
At present, the temperature in the high-temperature gasification furnace for processing the multi-component organic waste is mainly monitored in real time by a thermocouple, and signals are transmitted to a central control room, so that an operation basis is provided for operators. However, on one hand, the temperature in the gasification furnace is as high as 1200-1400 ℃, the thermocouple is easy to fail after running for a period of time at the high temperature, and on the other hand, the thermocouple hole is gradually covered by ash in the gasification furnace to prevent the thermocouple from correctly measuring the temperature in the furnace, so that the furnace temperature value measured by the thermocouple gradually loses the reliability, and the production requirement of long-period running cannot be met. Under the actual normal condition, the service life of the thermocouple installed before the system is started is about one week, and the thermocouple cannot be replaced on site after the gasification device system normally operates. Therefore, thermocouples lose the important role of monitoring furnace temperature and guiding production operation, and can only be used for observing temperature changes during start-up and early operation of the system. Therefore, the soft measurement model is established by taking the temperature in the gasification furnace as a research object, and the method has important application value for temperature regulation and stable operation in the multi-component organic waste treatment process.
In the thesis of non-contact temperature measurement technology of the coal gasifier wall (Wangkui, door expensive, chemical automation and instrument, 2014.41: 349-.
In CN105930585A, the inventor completes the simulation of flow field and temperature field of Shell gasification furnace based on CFD method, and provides complete furnace information for the design and industrial production of Shell gasification furnace. However, the CFD calculation needs to fully consider the chemical reactions occurring in the gasifier, and the multi-component organic waste treatment process has complicated and variable chemical reactions occurring in the gasifier due to the complicated sources of organic wastes, including antibiotic residues, distillation residues, waste activated carbon, high-salt wastewater, waste organic solvents, etc., besides the coal gasification series reactions in which activated carbon participates, the pyrolysis reactions of various organic matters, and HCl and H caused by the introduction of Cl element and S element in high-salt wastewater2S formation reaction, etc., and the specific situation of the reaction in the gasifier cannot be determined, so the CFD calculation is not applicable to the problems to be solved by the present invention. Compared with CFD calculation, the neural network has the advantages that the whole reaction process does not need to be excessively known, and accurate calculation results can be obtained only by providing enough data for training, so that the neural network is selected as the implementation method of the soft temperature measurement in the multi-component organic waste high-temperature gasification furnace.
CN102175345A inventor selects dry base coal and enters a furnaceCentral oxygen flow, oxygen-coal ratio, outlet synthesis gas hydrogen H2CO and methane CH4The content is used as an input variable of a soft measurement model, the hearth temperature soft measurement of the multi-nozzle opposed coal water slurry gasification furnace is realized based on the BP artificial neural network, the relative error of the result is within the range of 5%, and the prediction precision is high. However, unlike the coal water slurry gasification process of the above patent, the multi-component organic waste treatment process of the present invention has the following difficulties:
1) in the process of gasifying the coal water slurry, the main components of the raw material coal water slurry are only water and coal dust, and the condition of main reactants in the gasification reaction can be determined by determining the amount of coal entering a furnace, so that the inventor selects dry-based coal as one of input variables; however, in the process of co-processing multi-component organic wastes, organic wastes such as distillation residues, antibiotic residues and organic solvents are contained in the slurry, and although the proportion of the organic wastes in the slurry is known, the composition of the organic wastes is unstable, so that one input variable cannot be selected to determine the condition of reactants in the process of processing the multi-component organic wastes.
2) In the coal water slurry gasification process, the oxygen-coal ratio is a very important parameter, because the primary purpose of controlling the oxygen-coal ratio is to control the oxygen-carbon ratio, and the oxygen-carbon ratio can directly influence the components of produced gas and the amount of effective gas; in the process of the cooperative treatment of the multi-component organic waste, because various organic waste components are complex and unstable, and even if the oxygen-material ratio of materials obtained by matching different batches is controlled to be the same, the oxygen-material ratio can also be different, the oxygen-material ratio is not suitable for being used as an input variable for the temperature soft measurement of the high-temperature gasification furnace for the cooperative treatment of the multi-component organic waste.
3) At present, most element analyzers require a sample to be dry, otherwise the instrument can be damaged or the measurement is inaccurate, in the process of treating the multi-component organic waste, a plurality of organic wastes are in a solid-liquid mixed state, such as antibiotic dregs, rectification residues, waste organic solvents and the like, and parts of liquid organic matters in the organic wastes are lost after drying treatment, so that the slurry cannot be subjected to full-element analysis, and the oxygen-carbon ratio or the element composition of the slurry cannot be used as input variables for the temperature soft measurement of the high-temperature gasifier for treating the multi-component organic waste.
Disclosure of Invention
In view of the above research background, the present invention establishes a soft measurement method for the temperature in a multi-component organic waste high-temperature gasification furnace, comprising the following steps:
step 1, acquiring data obtained in the operation process of an actual multi-component organic waste high-temperature gasification furnace, and selecting characteristic variables and target variables; the characteristic variables comprise gasifier cooling water inlet flow, cooling water outlet flow, cooling water inlet temperature, cooling water outlet temperature, cooling water pressure and synthetic gas temperature discharged from a chilling chamber, and the target variable is the temperature in the gasifier;
step 2, preprocessing the characteristic variable data acquired in the step 1 and the corresponding target variable data, removing abnormal data points caused by measurement errors and abnormal working conditions by adopting a box-line graph method, and then normalizing the data to remove the influence of the magnitude of the data per se;
step 3, establishing a BP neural network model for measuring the temperature in the gasification furnace;
step 4, training the neural network model established in the step 3 by using the data processed in the step 2 (namely the processed characteristic variable data and the corresponding target variable data), so as to determine the optimal number of hidden layer nodes and network weight of the BP neural network model, and obtain a soft measurement model capable of accurately measuring the temperature in the multi-component organic waste high-temperature gasification furnace; and measuring the temperature in the multi-component organic waste high-temperature gasification furnace by using a soft measurement model.
As a preferable scheme of the invention, the box diagram method in the step 2 comprises the following operations:
a group of data is evenly divided into 4 equal parts after being arranged from small to large, the numerical value of the first division point is called as a lower quartile and is marked as Q1The second division point is called the median and is denoted as Q2The third division point is called the upper quartile and is denoted as Q3(ii) a Lower quartile Q1And upper quartile Q3Is called the interquartile range IQR, i.e. the calculationThe formula of IQR is:
IQR＝Q3-Q1
the minimum of normal data, the lower edge, is:
lower edge ═ Q1-1.5QIR
The maximum of the normal data, the upper edge, is:
upper edge of Q3+1.5QIR
Data larger than the upper edge or smaller than the lower edge is regarded as an abnormal data point, and is deleted.
As a preferred embodiment of the present invention, the data normalization in step 2 is performed by:
converting the value of the measurement data into the range of [ -1,1], using a normalization function of:
Figure BDA0003520502450000041
wherein, XiIs input data, XminIs the minimum value of the input data, XmaxIs the maximum value of the input data, XnewIs the normalized data.
As a preferred embodiment of the present invention, the step 3 specifically comprises the following steps:
step 3.1, setting the number of layers of the neural network into three layers;
step 3.2, the first layer is an input layer, and the number of nodes of the first layer is equal to the number of characteristic variables and is 6; the second layer is an implicit layer, and the number of nodes of the second layer is determined by the following empirical formula:
Figure BDA0003520502450000042
h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, a is an adjusting constant between 1 and 10, a is determined by a subsequent training result, and h is 3+ a; the third layer is an output layer, the number of nodes is 1, namely the target of soft measurement: the temperature in the gasifier;
step 3.3, the transfer function from the input layer to the hidden layer is a hyperbolic tangent S-type function with the formula
Figure BDA0003520502450000043
The transfer function from the hidden layer to the output layer is a linear transfer function, which is expressed by
f(n)＝n
Step 3.4, selecting a levenberg-marquardt algorithm as a training algorithm;
step 3.5, recording the node of the BP neural network input layer as xiWith hidden layer neuron node pjThe output layer neuron node is y, and the weight value between the input layer neuron and the hidden layer neuron is wijThe weight value between hidden layer neuron and output layer neuron is vj(ii) a When the expected value of the output data is z, the calculation formula related to the BP neural network is as follows:
the output of hidden layer neurons:
Figure BDA0003520502450000051
output of output layer neurons:
Figure BDA0003520502450000052
error value of output result:
Figure BDA0003520502450000053
step 3.6, in order to reduce the error E arrival of the output result, the weight value w between neurons needs to be adjustedijAnd vj(ii) a Since the negative gradient is the direction in which the function value is reduced most quickly, a learning rate η is set, the weight is adjusted by η units along the direction of the negative gradient, the learning rate η is set to 0.01, and the adjustment is as follows:
Figure BDA0003520502450000054
the adjustment sequence of the BP neural network is as follows:
1) firstly adjusting the weight v from hidden layer to output layerjThe iteration formula for adjusting the weight from the hidden layer to the output layer is as follows:
vj(t+1)＝vj(t)+ηy(1-y)pj
2) readjusting the weight w from the input layer to the hidden layerijThe iterative formula from the input layer to the weight adjustment of the hidden layer is as follows:
wij(t+1)＝wij(t)+ηδjxi
wherein deltaj＝(z-y)f′(uj)vj，u＝∑iwijxi；
The error value E is gradually reduced by continuously adjusting the weight value between the neurons until the convergence requirement is reached.
As a preferred embodiment of the present invention, the step 4 specifically comprises the following steps:
training each neural network with different number of hidden layer neurons by using measurement data, wherein the number h of hidden layer neurons of each neural network is 3+ a, and a is 1-10 respectively; and comparing error values E of the output results of the neural networks, wherein the neural network with the minimum error is used as a soft measurement model of the temperature in the multi-component organic waste high-temperature gasification furnace.
Compared with the prior art, the invention is characterized in that:
1) the invention selects the neural network as the realization method of the temperature soft measurement in the multi-component organic waste high-temperature gasification furnace. Because the multi-component organic waste synergistic treatment process is complex in hazardous waste source, chemical reactions in the gasifier are complex and changeable, the neural network does not need to have excessive understanding of the whole reaction process, and accurate calculation results can be obtained as long as enough data are provided for training.
2) The neural network structure selected by the invention is a three-layer BP neural network. Generally speaking, the three-layer neural network can approximate any nonlinear function, the input variables of the invention are 6, and the three-layer neural network structure can ensure the rapid convergence of the neural network while realizing high-accuracy soft measurement, thereby reducing the time cost for debugging the neural network. BP (back propagation), namely an error back propagation algorithm, realizes automatic adjustment of the weight of the neuron through two processes of forward propagation of information and back propagation of errors, and can help the calculation result of the neural network to achieve higher accuracy.
3) The invention selects some process variables which can reflect the temperature condition in the gasification furnace in the running process of the gasification furnace as input variables of the neural network, wherein the process variables comprise the inlet flow of cooling water, the outlet flow of cooling water, the inlet temperature of cooling water, the outlet temperature of cooling water, the pressure of cooling water and the temperature of synthetic gas discharged from a chilling chamber, and finally soft measurement of the temperature in the multi-component organic waste high-temperature gasification furnace is realized. The proposed method has high accuracy, and compared with the actual value (accurate data when the thermocouple is not failed) measured by the thermocouple in the production process, the error is within the allowable range in industrial production.
4) The invention selects six input variables of cooling water inlet flow, cooling water outlet flow, cooling water inlet temperature, cooling water outlet temperature, cooling water pressure and synthesis gas temperature discharged from the chilling chamber, and has the advantages that the data are obtained by automatic instrument test in the actual production line, compared with the data of slurry concentration, heat value and the like which are measured in a laboratory after sampling is needed, the data of the selected input variables are convenient to obtain, the data frequency is high, and simultaneously, the sampling error caused by uneven distribution of slurry components is avoided.
Drawings
FIG. 1 is a process schematic diagram of a multi-component organic waste co-processing process;
FIG. 2 is a flow chart of model building and training of the present invention;
FIG. 3 is a structural topology of a neural network used in the present invention;
FIG. 4 is a comparison of soft measurement model prediction results with real data;
FIG. 5 is a graph of the relative percentage error of the predicted result versus the true result for the soft measurement model;
FIG. 6 is a comparison of the predicted results of the comparative example soft measurement model with real data;
FIG. 7 is a graph of the relative percentage error of the predicted results and the actual results of the comparative example soft measurement model.
Detailed Description
The invention will be further illustrated and described with reference to specific embodiments. The technical features of the embodiments of the present invention can be combined correspondingly without mutual conflict.
As shown in figure 1, the invention is a process schematic diagram of a multi-component organic waste co-processing process, the flow of a cooling water inlet, the flow of a cooling water outlet, the temperature of the cooling water inlet, the temperature of the cooling water outlet, the pressure of cooling water and the temperature of synthesis gas discharged from a chilling chamber are selected as input variables, selected characteristic variable data are obtained by automatic testing of instruments in an actual production line, compared with the data of slurry concentration, heat value and the like which are measured in a laboratory after sampling is needed, the selected input variable data are convenient to obtain and high in data frequency, and meanwhile, the sampling error caused by uneven distribution of slurry components is avoided.
As shown in FIG. 2, the soft measurement method for the temperature in the multi-component organic waste high-temperature gasification furnace provided by the invention comprises the following steps:
step 1, acquiring data obtained in the actual operation process of the multi-component organic waste high-temperature gasification furnace, summarizing a reaction principle and an operation experience, and selecting characteristic variables and target variables of a soft measurement model. The selected characteristic variables comprise cooling water inlet flow, cooling water outlet flow, cooling water inlet temperature, cooling water outlet temperature, cooling water pressure and chilling chamber synthesis gas temperature, and the target variable is the temperature in the gasifier.
And 2, preprocessing the characteristic variable data acquired in the step 1 and the corresponding target variable data, removing abnormal data points caused by measurement errors and abnormal working conditions by adopting a box-line graph method, and then normalizing the data to remove the influence of the magnitude of the data on the model. The box diagram method comprises the following specific operations:
a group of data is evenly divided into 4 equal parts after being arranged from small to large, the numerical value of the first division point is called as a lower quartile and is marked as Q1The second division point is called the median and is denoted as Q2The third division point is called the upper quartile and is denoted as Q3. Lower quartile Q1And upper quartile Q3The difference in (A) is called the interquartile range IQR, i.e., the formula for calculating IQR is:
IQR＝Q3-Q1
the minimum values of normal data (lower edge) are:
lower edge ═ Q1-1.5QIR
The maximum (upper edge) of the normal data is:
upper edge ═ Q3+1.5QIR
Data that is larger than the upper edge or smaller than the lower edge is considered an outlier data point and needs to be discarded.
The specific operation of normalization is as follows:
converting the value of the measurement data into the range of [ -1,1], using a normalization function of:
Figure BDA0003520502450000081
wherein, XiIs input data, XminIs the minimum value of the input data, XmaxIs the maximum value of the input data, XnewIs the normalized data.
And 3, randomly dividing the data normalized in the step 2 into 70%: 15%: and 15% of the three groups, which are respectively used as a training set, a testing set and a verification set.
And 4, establishing a BP neural network model for measuring the temperature in the gasification furnace. FIG. 3 shows a structural topology of the BP neural network of the present invention; the specific establishment steps of the BP neural network model for measuring the temperature in the gasifier are as follows:
and 4.1, setting the number of the neural network layers as three layers.
Step 4.2, the first layer is an input layer, and the number of nodes of the input layer is equal to the number of characteristic variables and is 6; the second layer is an implicit layer, and the number of nodes of the second layer is determined by the following empirical formula:
Figure BDA0003520502450000082
h is the number of nodes of the hidden layer, m is the number of nodes of the input layer, n is the number of nodes of the output layer, a is an adjusting constant between 1 and 10, and a is determined by a subsequent training result, wherein h is 3+ a. The third layer is an output layer, the number of nodes is 1, namely the target of soft measurement: the temperature in the gasifier.
Step 4.3, the transfer function from the input layer to the hidden layer is a hyperbolic tangent S-type function with the formula
Figure BDA0003520502450000083
The transfer function from the hidden layer to the output layer is a linear transfer function of the formula
f(n)＝n
Step 4.4, training the BP network by adopting a levenberg-marquardt algorithm, wherein the target error is 1 multiplied by 10-7The maximum number of iterations is 10000, and the learning rate is 0.01.
Step 4.5, recording the node of the BP neural network input layer as xiWith hidden layer neuron node pjThe output layer neuron node is y, and the weight value between the input layer neuron and the hidden layer neuron is wijWith a threshold value of θ and a weight between hidden layer neurons and output layer neurons of vj. When the expected value of the output data is t, the calculation formula related to the BP neural network is as follows:
the output of hidden layer neurons:
Figure BDA0003520502450000091
output of output layer neurons:
Figure BDA0003520502450000092
error value of output result:
Figure BDA0003520502450000093
step 4.6, to reduce the error E arrival of the output result, the weight values w between neurons need to be adjustedijAnd vj. Because the negative gradient is the direction that the function value is reduced most quickly, a learning rate eta is set, the weight value is adjusted by eta units along the direction of the negative gradient each time, the learning rate eta of the invention is set to be 0.01, and the adjustment each time is as follows:
Figure BDA0003520502450000094
the adjustment sequence of the BP neural network is as follows:
1) firstly adjusting the weight v from hidden layer to output layerjThe iteration formula for adjusting the weight from the hidden layer to the output layer is as follows:
vj(t+1)＝vj(t)+ηy(1-y)pj
2) readjusting the weight w from the input layer to the hidden layerijThe iterative formula from the input layer to the weight adjustment of the hidden layer is as follows:
wij(t+1)＝wij(t)+ηδjxi
wherein deltaj＝(z-y)f′(uj)vj，u＝∑iwijxi；
This adjustment method is called a gradient descent method, and the error value E is gradually reduced by continuously adjusting the weight values between neurons until the convergence requirement is reached.
And 5, analyzing the neural network model established in the step 4 by using the processed training data (each training data in the invention comprises a group of characteristic variable data and corresponding target variable data), so as to determine the optimal number of hidden layer nodes and network weight of the BP neural network, and obtaining the soft measurement model capable of accurately measuring the temperature in the multi-component organic waste high-temperature gasification furnace. The specific method comprises the following steps: the neural networks with different hidden layer neuron numbers h & lt 3+ a (a & lt 1-10 & gt) are respectively trained by using measurement data (namely, the neural networks with the hidden layer neuron numbers 4, 5, 6, 7, 8, 9, 10, 11, 12 and 13 respectively), and compared with error values E of output results of the neural networks, the neural network with the minimum error is the target model of the invention.
Experiments show that the soft measurement method for the temperature in the multi-component organic waste high-temperature gasification furnace is effective and feasible:
the method is characterized in that near 5-day data of a Texaco high-temperature gasification furnace which is acquired from 1 day to 6 days of 4 months in 2021 and is subjected to synergistic treatment on a certain unit of multi-component organic waste is taken as original data, and the data is divided into 70% randomly after normalization: 15%: and 15% of three groups are respectively used as a training set, a testing set and a verification set, and the characteristic variables are selected from cooling water inlet flow, cooling water outlet flow, cooling water inlet temperature, cooling water outlet temperature, cooling water pressure and chilling chamber synthesis gas temperature. After the neural network model is trained by using the data of the training set and the test set, the data of the verification set is input into the neural network, and the comparison between the obtained simulation result and the actual value measured by the thermocouple in the production process is shown in fig. 4, wherein the dotted line data in the graph is the simulation result of the neural network, and the solid line data is the actual value. Compared with the real furnace temperature measured by the thermocouple, the soft measurement model simulation result has the average relative percentage error of 0.71%, the average absolute error of 9.04 ℃, the root mean square error of 12.62, and as shown in fig. 5, the relative percentage error is 4.25% at most, and in industrial production, the relative percentage error is within the error allowable range.
For comparison, the data of the 5 days are still used as the original data to carry out the same normalization and grouping treatment, the characteristic variables are selected from slurry flow, central oxygen flow, oxygen-material ratio and synthetic gas H discharged from a chilling chamber2、CH4CO content, gasifier temperature simulation results obtained after network training and actual values measured by thermocouplesFor example, as shown in fig. 6, the data in the dotted line is the simulation result of the neural network, and the data in the solid line is the real value. Compared with the actual furnace temperature measured by the thermocouple, the soft measurement model simulation result has the average relative percentage error of 0.957%, the average absolute error of 12.25 ℃ and the root mean square error of 17.25, and as shown in FIG. 7, the maximum relative percentage error is 5.26%.
It can be seen that the soft measurement of the temperature in the multi-component organic waste high-temperature gasifier can achieve higher accuracy, the problem that the selection of characteristic variables is limited due to the fact that the components of reaction raw materials in the multi-component organic waste synergistic treatment process are complex and unfixed and element analysis cannot be carried out is solved, the process variables of gasifier operation including the relevant data of circulating cooling water and the temperature of synthesis gas are successfully used as input variables of a neural network, and finally the soft measurement of the temperature in the multi-component organic waste high-temperature gasifier is achieved.
In conclusion, the soft measurement method for the temperature in the multi-component organic waste high-temperature gasification furnace can accurately and effectively simulate and estimate the temperature in the multi-component organic waste high-temperature gasification furnace, and can assist in smooth operation of the multi-component organic waste co-processing process.

Claims (5)

1. A soft measurement method for the temperature in a multi-component organic waste high-temperature gasification furnace is characterized by comprising the following steps:
step 1, acquiring data obtained in the operation process of an actual multi-component organic waste high-temperature gasification furnace, and selecting characteristic variables and target variables; the characteristic variables comprise gasifier cooling water inlet flow, cooling water outlet flow, cooling water inlet temperature, cooling water outlet temperature, cooling water pressure and synthetic gas temperature discharged from a chilling chamber, and the target variable is the temperature in the gasifier;
step 2, preprocessing the characteristic variable data acquired in the step 1 and the corresponding target variable data, removing abnormal data points caused by measurement errors and abnormal working conditions by adopting a box-line graph method, and then normalizing the data to remove the influence of the magnitude of the data per se;
step 3, establishing a BP neural network model for measuring the temperature in the gasification furnace;
step 4, training the neural network model established in the step 3 by using the data processed in the step 2, and determining the optimal number of hidden layer nodes and network weight of the BP neural network model to obtain a soft measurement model capable of accurately measuring the temperature in the multi-component organic waste high-temperature gasifier; and (3) measuring the temperature in the multi-component organic waste high-temperature gasification furnace by using a soft measurement model.
2. The method for soft measurement of the temperature in a multi-component organic waste high-temperature gasification furnace according to claim 1, wherein the box plot method of step 2 is operated as follows:
a group of data is evenly divided into 4 equal parts after being arranged from small to large, the numerical value of the first division point is called as a lower quartile and is marked as Q1The second division point is called the median and is denoted as Q2The third division point is called the upper quartile and is denoted as Q3(ii) a Lower quartile Q1And upper quartile Q3The difference in (A) is called the interquartile range IQR, i.e., the formula for calculating IQR is:
IQR＝Q3-Q1
the minimum of normal data, the lower edge, is:
lower edge ═ Q1-1.5QIR
The maximum of the normal data, the upper edge, is:
upper edge ═ Q3+1.5QIR
Data that is larger than the upper edge or smaller than the lower edge is considered an outlier data point, which is discarded.
3. The method for soft measurement of the temperature in the multi-component organic waste high-temperature gasification furnace according to claim 1, wherein the step 2 is performed by normalizing the data by:
converting the value of the measurement data into the range of [ -1,1], using a normalization function of:
Figure FDA0003520502440000021
wherein, XiIs input data, XminIs the minimum value of the input data, XmaxIs the maximum value of the input data, XnewIs the normalized data.
4. The method for soft measurement of the temperature in the multi-component organic waste high-temperature gasification furnace according to claim 1, wherein the step 3 comprises the following specific steps:
step 3.1, setting the number of layers of the neural network into three layers;
step 3.2, the first layer is an input layer, and the number of nodes of the input layer is equal to the number of characteristic variables and is 6; the second layer is an implicit layer, and the number of nodes of the second layer is determined by the following empirical formula:
Figure FDA0003520502440000022
h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, a is an adjusting constant between 1 and 10, a is determined by a subsequent training result, and h is 3+ a; the third layer is an output layer, the number of nodes is 1, namely the target of soft measurement: the temperature in the gasifier;
step 3.3, the transfer function from the input layer to the hidden layer is a hyperbolic tangent S-type function with the formula
Figure FDA0003520502440000023
The transfer function from the hidden layer to the output layer is a linear transfer function of the formula f (n) n
Step 3.4, selecting a levenberg-marquardt algorithm as a training algorithm;
step 3.5, recording the node of the BP neural network input layer as xiWith hidden layer neuron node pjOutput layer neuron nodeY, the weight value between input layer neurons and hidden layer neurons is wijThe weight value between the hidden layer neuron and the output layer neuron is vj(ii) a When the expected value of the output data is z, the calculation formula related to the BP neural network is as follows:
the output of hidden layer neurons:
Figure FDA0003520502440000024
output of output layer neurons:
Figure FDA0003520502440000025
error value of output result:
Figure FDA0003520502440000031
step 3.6, to reduce the error E arrival of the output result, the weight values w between neurons need to be adjustedijAnd vj(ii) a Since the negative gradient is the direction in which the function value is reduced most quickly, a learning rate η is set, the weight is adjusted by η units along the direction of the negative gradient, the learning rate η is set to 0.01, and the adjustment is as follows:
Figure FDA0003520502440000032
the adjustment sequence of the BP neural network is as follows:
1) firstly adjusting the weight v from hidden layer to output layerjThe iteration formula for adjusting the weight from the hidden layer to the output layer is as follows:
vj(t+1)＝vj(t)+ηy(1-y)pj
2) readjusting the weight w from the input layer to the hidden layerijIterative algorithm for adjusting weights from input layer to hidden layerThe formula is as follows:
wij(t+1)＝wij(t)+ηδjxi
wherein deltaj＝(z-y)f′(uj)vj，u＝∑iwijxi；
The error value E is gradually reduced by continuously adjusting the weight value between the neurons until the convergence requirement is reached.
5. The method for soft measurement of the temperature in the multi-component organic waste high-temperature gasification furnace according to claim 1, wherein the step 4 comprises the following specific steps:
training each neural network with different number of hidden layer neurons by using measured data, wherein the number h of the hidden layer neurons of each neural network is 3+ a, and a is 1-10 respectively; and comparing error values E of the output results of the neural networks, wherein the neural network with the minimum error is used as a soft measurement model of the temperature in the multi-component organic waste high-temperature gasification furnace.
CN202210176584.2A 2022-02-25 2022-02-25 Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace Pending CN114550838A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210176584.2A CN114550838A (en) 2022-02-25 2022-02-25 Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210176584.2A CN114550838A (en) 2022-02-25 2022-02-25 Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace

Publications (1)

Publication Number Publication Date
CN114550838A true CN114550838A (en) 2022-05-27

Family

ID=81679209

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210176584.2A Pending CN114550838A (en) 2022-02-25 2022-02-25 Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace

Country Status (1)

Country Link
CN (1) CN114550838A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115330829A (en) * 2022-10-14 2022-11-11 冰轮智慧新能源技术(山东)有限公司 Method for identifying gasification reaction abnormity of straw gasification furnace

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115330829A (en) * 2022-10-14 2022-11-11 冰轮智慧新能源技术(山东)有限公司 Method for identifying gasification reaction abnormity of straw gasification furnace

Similar Documents

Publication Publication Date Title
WO2021159585A1 (en) Dioxin emission concentration prediction method
CN112183709B (en) Method for predicting and early warning excessive dioxin in waste incineration gas
CN110807554B (en) Generation method and system based on wind power/photovoltaic classical scene set
CN101067828A (en) Coke oven fire path temperature integrated moulding and soft measuring method
CN114550838A (en) Soft measurement method for temperature in multi-component organic waste high-temperature gasification furnace
CN109187914B (en) Coal quality characteristic-based method for predicting NOx generation amount of coal-fired power plant
CN115392629A (en) Coal gangue heat productivity soft measurement method based on improved SVR
Özveren et al. Investigation of steam gasification in thermogravimetric analysis by means of evolved gas analysis and machine learning
Beltramo et al. Evaluation of the linear and non-linear prediction models optimized with metaheuristics Application to anaerobic digestion processes
CN115034129A (en) NOx emission concentration soft measurement method for thermal power plant denitration device
CN111462835B (en) Dioxin emission concentration soft measurement method based on depth forest regression algorithm
CN111985682B (en) Furnace temperature prediction method of coal water slurry gasification furnace based on neural network
Dixon et al. Experience with data mining for the anaerobic wastewater treatment process
CN115762664A (en) Method for predicting gas production component in high-temperature gasification process of multi-component organic waste
CN110824099B (en) Method for predicting reaction performance in solid fuel chemical chain process based on GBRT
Zain et al. Development of a neural network Predictive Emission Monitoring System for flue gas measurement
CN110598185A (en) Coke oven coking unit consumption influence factor sequencing method and system
CN113536601B (en) Method for improving estimation accuracy of carbon content of catalyst of continuous reforming device
He et al. Prediction of MSWI furnace temperature based on TS fuzzy neural network
An et al. Research on a soft-measurement model of gasification temperature based on recurrent neural network
CN111222540A (en) Abnormal judgment dioxin detection method based on unsupervised learning of clustering
Wang A new variable selection method for soft sensor based on deep learning
Komarysta et al. OPTIMIZING BIOGAS PRODUCTION USING ARTIFICIAL NEURAL NETWORK.
CN115859770B (en) On-line monitoring system and device for domestic sewage fecal coliform based on GA and FFNN
CN112131775A (en) Solid oxide fuel cell performance reasoning and optimizing method

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