Method and system for determining ash fusibility characteristic curve
Patent Information
- Authority / Receiving Office
- GB · GB
- Patent Type
- Applications
- Current Assignee / Owner
- THE UNIV OF NOTTINGHAM NINGBO CHINA
- Filing Date
- 2022-12-13
- Publication Date
- 2026-06-17
AI Technical Summary
Traditional methods for determining ash fusibility characteristic curves in biomass boiler design suffer from poor repeatability and inefficiency, requiring manual observation and lengthy processes to obtain accurate melting temperatures for multiple ash samples.
A method utilizing a back-propagation (BP) neural network to construct and train on composition indexes of ash samples, obtaining actual fusibility characteristic curves through image analysis and edge detection, allowing for rapid prediction of melting characteristics across temperatures.
This approach significantly reduces the time and error in determining ash fusibility characteristic curves, providing accurate and repeatable melting temperature data for biomass boiler design by leveraging indirect measurements and artificial neural networks.
Abstract
Description
TECHNICAL FIELD
[0001] The present disclosure relates to the field of prediction of ash fusibility characteristic curves, particularly, to a method and system for determining an ash fusibility characteristic curve. BACKGROUND
[0002] Ash is a derivative of biomass fuel combustion and pyrolysis. In the design process of a biomass boiler, it is necessary to consider mixed combustion of various biomass. The ash produced by the combustion of different biomass fuels has different characteristic melting temperatures, which brings a significant challenge to the design of biomass boilers. The traditional ash fusion test is widely used to measure the ash fusibility characteristic temperature. In this method, the conical ash pellet is heated to more than 1,500-1,600 °C in the oxidation or reduction atmosphere. In addition, four characteristic melting temperatures are determined for each sample according to the shape change of the ash pellet: deformation temperature, softening temperature, hemispherical temperature, and flow temperature. However, the poor repeatability of traditional detection methods makes it difficult to provide an accurate prediction of characteristic melting temperature. Besides, the design of a biomass boiler requires the characteristic melting temperatures of a large number of ash samples and the melting characteristics of biomass ash samples at different temperatures. The present disclosure provides a method for determining an ash fusibility characteristic curve based on a back-propagation (BP) neural network, which can shorten the period of determining a fusibility characteristic curve of the ash samples and obtain the complete change process of the ash samples with the temperature. SUMMARY
[0003] An objective of the present disclosure is to provide a method and system for determining an ash fusibility characteristic curve, which can shorten the period of determining a fusibility characteristic curve of ash samples and obtain a complete change process of the ash samples with a temperature.
[0004] The present disclosure provides the following technical solutions to achieve the above objective:
[0005] In the first aspect, the present disclosure provides a method for determining an ash fusibility characteristic curve, including the following steps: 20 02 23
[0006] obtaining an actual ash fusibility characteristic curve;
[0007] calculating composition indexes of ash samples;
[0008] constructing a BP neural network;
[0009] training the BP neural network by using the composition indexes of the ash samples and the actual ash fusibility characteristic curve; and
[0010] determining a fusibility characteristic curve of unknown ash samples based on the trained BP neural network.
[0011] Optionally, a process of obtaining an actual ash fusibility characteristic curve includes the following sub-steps explicitly:
[0012] ashing samples;
[0013] conducting ash pelleting;
[0014] heating ash pellets, and taking a picture of every preset temperature range;
[0015] conducting edge detection on each picture taken using an image analysis method of a Sobel operator; and
[0016] according to detection results, determining the relative height of the ash pellets to obtain the actual ash fusibility characteristic curve.
[0017] Optionally, a process of ashing samples includes the following sub-step explicitly:
[0018] ashing a series of biomass samples at 500-600°C.
[0019] Optionally, a process of conducting ash pelleting includes the following sub-step explicitly:
[0020] taking 0.5-1 g of the ash samples, pressing the ash samples in a cylindrical mould to make the ash pellets, and pressing the mould with an automatic press to make the ash pellets to ensure that the samples have the same size.
[0021] Optionally, a process of heating the ash pellets and taking a picture of every preset temperature range includes the following sub-step explicitly:
[0022] heating the ash pellets from 150-200°C to 1,500-1,600°C at a heating rate of 5-10°C / min, and meanwhile, automatically taking a picture of the ash samples using a black and white closed-loop digital camera every 0.5-l°C.
[0023] Optionally, the composition indexes of the ash samples include a slag viscosity index, an ash caking index, a dolomite ratio, a ratio of a basic oxide to an acid oxide of ash, a slagging index, and an aluminium titanium oxide content.
[0024] Optionally, the BP neural network includes an input layer, a hidden layer, and an output layer.
[0025] Optionally, the BP neural network is trained explicitly in three stages: signals forward propagation, errors back propagation, weights &thresholds adjustment. 20 02 23
[0026] In the second aspect, based on the above method in the present disclosure, the present disclosure further provides a system for determining an ash fusibility characteristic curve, including:
[0027] an actual ash fusibility characteristic curve determination module configured to obtain an actual ash fusibility characteristic curve;
[0028] an ash sample composition index determination module configured to calculate composition indexes of ash samples;
[0029] a neural network construction module configured to construct a BP neural network;
[0030] a training module configured to train the BP neural network by using the composition indexes of the ash samples and the actual ash fusibility characteristic curve; and
[0031] a prediction module configured to determine a fusibility characteristic curve of unknown ash samples based on the trained BP neural network.
[0032] According to specific examples provided by the present disclosure, the present disclosure discloses the following technical effects:
[0033] The composition of the ash samples determines the melting characteristics. Generally, when the composition is fixed, the melting characteristics of the ash samples are also determined. The present disclosure uses an indirect measurement of these easily measured physical quantities combined with the artificial neural network to indirectly measure melting characteristic data of the ash samples, avoiding the tedious processing and analysis process of traditional measurement methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] To describe the examples of the present disclosure or the technical solutions in the related art more clearly, the accompanying drawings required in the examples are briefly introduced below. The accompanying drawings described below are only some examples of the present disclosure. Those of ordinary skill in the art may further obtain other accompanying drawings based on these accompanying drawings without creative labour.
[0035] FIG. 1 is a flow block diagram of a method for determining an ash fusibility characteristic curve in an example of the present disclosure; and
[0036] FIG. 2 is a schematic structural diagram of a BP neural network in the example of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0037] The technical solutions of the examples of the present disclosure are clearly and completely described below with reference to the accompanying drawings. Apparently, the 20 02 23 described examples are merely a part rather than all of the examples of the present disclosure. All other examples obtained by those of ordinary skill in the art based on the examples of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.
[0038] An objective of the present disclosure is to provide a method and system for determining an ash fusibility characteristic curve, which can shorten the period of determining a fusibility characteristic curve of ash samples and obtain a complete change process of the ash samples with a temperature.
[0039] To make the above-mentioned objective, features, and advantages of the present disclosure clearer and more comprehensible, the present disclosure will be further described in detail below in conjunction with the accompanying drawings and specific implementations.
[0040] FIG. 1 is a flow block diagram of a method for determining an ash fusibility characteristic curve in an example of the present disclosure. As shown in FIG. 1, the method in the present disclosure includes the following steps.
[0041] Step 1 : An actual ash fusibility characteristic curve was obtained.
[0042] Specifically, the step included the following sub-steps.
[0043] Step 1.1: Samples were ashed. A series of biomass samples were ashed at 500-600°C.
[0044] Step 1.2: Ash pelleting. 0.5-1 g of the ash samples were taken. The ash samples were pressed in a cylindrical mould to make the ash pellets, and the mould was pressed with an automatic press to make the ash pellets to ensure that the samples had the same size.
[0045] Step 1.3: Photographing. The ash pellets were heated from 150-200°C to 1,500-1,600°C at a heating rate of 5-10°C / min, and meanwhile, a picture of the ash samples was automatically taken using a black and white closed-loop digital camera every 0.5-l°C.
[0046] Step 1.4: Data collecting. Edge detection was conducted on each picture taken using an image analysis method of a Sobel operator. According to detection results, a relative height of the ash pellets was obtained to obtain the fusibility characteristic curve data of the ash samples. Finally, the obtained data were imported into a BPdata database as an input data set.
[0047] Step 2: Composition indexes of ash samples were calculated.
[0048] Specifically, the step included the following sub-steps.
[0049] Step 2.1: The composition of the sample was determined. The ash samples were pressed into discs with a diameter of 32-34 mm at 40-50 MPa for 30-50 s with a low-pressure polyethene rimmed substrate using a press. The chemical composition of the ash samples was analyzed by an X-ray fluorescence spectrometer.
[0050] Step 2.2: The composition indexes of the ash samples were calculated. The composition of the ash samples could affect the melting characteristics, and the ash fusibility characteristic curve was evaluated in the form of a metal oxide index. The composition indexes included a slag viscosity index, R250 (an ash caking index, an empirical parameter for determining ash caking), a dolomite ratio, Rb / a (a ratio of a basic oxide to an acid oxide of ash), a slagging index, and an aluminium titanium oxide content. The calculation formulas are as follows: Slag viscosity index =
[0051]
[0052]
[0053]
[0054] ___________SiO2___________ S1O2 + + CaO + MgO’ n _ S1O2+AI2O3 250 SiO2 + AI203 + Fe2O3 + CaO’ „ , CaO + MgO Dolomite ratio =---------------2----------, Fe2O3 + CaO + MgO + K20 + Na20 „ _ Fe2O3 + CaO + MgO + K20 + Na20 B / A— SiO2 +A12O3 + TiO2
[0055] Slagging index =Rb / aX (Na2O + K2O), and
[0056] Aluminium titanium oxide content = A12O3 + TiO2,
[0057] where SiO2 is the content of silicon dioxide in the ash sample, Fe2O3 is the content of ferric oxide in the ash sample, CaO is the content of calcium oxide in the ash sample, MgO is the 20 02 23 content of magnesium oxide in the ash sample, AI2O3 is the content of aluminum oxide in the ash sample, Na2O is the content of sodium oxide in the ash sample, K2O is the content of potassium oxide in the ash sample, and TiO2 is the content of titanium dioxide in the ash sample.
[0058] Finally, the obtained data were imported into a BPdata database as an input data set.
[0059] Step 3: A BP neural network was constructed.
[0060] The BP network had: an input layer, a hidden layer, and an output layer. The BP algorithm took the error square as the objective function and used the gradient descent method to calculate the minimum value of the objective function.
[0061] The BP neural network was trained in three stages: signals forward propagation, errors back propagation, weights &thresholds adjustment. In this way, the iteration was repeated, and stopped when the number of iterations was greater than the maximum number of iterations or less than the target error. The trained BP neural network model was obtained (see FIG. 2 for details).
[0062] Step 4: The BP neural network was trained using the ash samples’ composition indexes and the actual ash fusibility characteristic curve.
[0063] Step 5: A fusibility characteristic curve of unknown ash samples was determined based on the trained BP neural network.
[0064] By repeatedly training the neural network with sufficient data until the training error was less than the preset value, it was equivalent to obtaining a function expression of the fusibility characteristic curve and composition indexes of the ash samples. After the training of the neural network reaches the standard, the fusibility characteristic curve of the ash samples could be indirectly determined by measuring the slag viscosity index, the R250, the dolomite ratio, the ratio of base to acid, the slagging index, and the aluminium titanium oxide content of the ash samples.
[0065] Based on the above method in the present disclosure, the present disclosure further 20 02 23 provided a system for determining an ash fusibility characteristic curve, including an actual ash fusibility characteristic curve determination module, an ash sample composition index determination module, a neural network construction module, a training module, and a prediction module.
[0066] The actual ash fusibility characteristic curve determination module was configured to obtain an actual ash fusibility characteristic curve.
[0067] The ash sample composition index determination module was configured to calculate the composition indexes of ash samples.
[0068] The neural network construction module was configured to construct a BP neural network.
[0069] The training module was configured to train the BP neural network by using the composition indexes of the ash samples and the actual ash fusibility characteristic curve.
[0070] The prediction module was configured to determine a fusibility characteristic curve of unknown ash samples based on the trained BP neural network.
[0071] The above solutions in the present disclosure have the following beneficial effects:
[0072] The composition of the ash samples determines the melting characteristics. Generally, when the composition is fixed, the melting characteristics of the ash samples are also determined. The present disclosure uses the indirect measurement of these easily measured physical quantities combined with the artificial neural network to indirectly measure melting characteristic data of the ash samples, avoiding the tedious processing and analysis process of traditional measurement methods.
[0073] The traditional method uses the ash fusibility test to obtain the fusibility characteristic temperatures of the ash samples: deformation temperature, softening temperature, hemispherical temperature, and flow temperature. This method relies on manual observation and subjective judgment of each temperature point, which has large errors, poor repeatability and an inability to judge the melting characteristics of the ash samples at different temperatures. For the ash fusibility characteristic curve in step 1.4, the melting temperatures of the ash samples were obtained by the image analysis method, which made the results more accurate and repeatable, and the melting characteristics of the ash samples at different temperatures could be obtained. However, when processing a large number of biomass samples, the tedious operation steps lead to a prolonged project cycle. In addition, long-term continuous shooting causes high camera loss, which leads to data errors in the later stage of the project. The method for determining an ash fusibility characteristic curve based on the BP neural network algorithm can reduce the measurement times of the ash fusibility characteristic curve of the samples through steps 3, 4 and 5 and shorten the project cycle. The ash fusibility characteristic curve data of other samples is obtained by a small amount of accurate data to ensure the accuracy of the overall data.
[0074] Examples of the present specification are described progressively, each example focuses on the difference from other examples, and the same and similar parts between the examples may refer to each other. Since the system disclosed in an embodiment corresponds to the method disclosed in another embodiment, the description is relatively simple, and reference can be made to the method description.
[0075] Specific examples are used herein to explain the principles and implementations of the present disclosure. The foregoing description of the examples is merely intended to help understand the method of the present disclosure and its core ideas; besides, various modifications may be made by those of ordinary skill in the art to specific implementations and the scope of application in accordance with the ideas of the present disclosure. In conclusion, the content of the present specification shall not be construed as limitations to the present disclosure. 20 02 23
Claims
1. A method for determining an ash fusibility characteristic curve, comprising the following steps:obtaining an actual ash fusibility characteristic curve;calculating composition indexes of ash samples;constructing a back-propagation (BP) neural network;training the BP neural network by using the composition indexes of the ash samples and the actual ash fusibility characteristic curve; anddetermining a fusibility characteristic curve of unknown ash samples based on the trained BP neural network.
2. The method for determining an ash fusibility characteristic curve according to claim 1, wherein a process of obtaining an actual ash fusibility characteristic curve specifically comprises the following sub-steps explicitly:ashing samples;conducting ash pelleting;heating ash pellets, and taking a picture of every preset temperature range;conducting edge detection on each picture taken using an image analysis method of a sobel operator; andaccording to detection results, determining the relative height of the ash pellets to obtain the actual ash fusibility characteristic curve.
3. The method for determining an ash fusibility characteristic curve according to claim 2, wherein a process of ashing samples comprises the following sub-step explicitly:ashing a series of biomass samples at 500-600°C.
4. The method for determining an ash fusibility characteristic curve according to claim 2, wherein a process of conducting ash pelleting comprises the following sub-step explicitly :taking 0.5-1 g of the ash samples, pressing the ash samples in a cylindrical mould to make the ash pellets, and pressing the mould with an automatic press to make the ash pellets to ensure that the samples have the same size.
5. The method for determining an ash fusibility characteristic curve according to claim 2, wherein a process of heating the ash pellets and taking a picture of every preset temperaturerange comprises the following sub-step explicitly:heating the ash pellets from 150-200°C to 1,500-1,600°C at a heating rate of 5-10°C / min, and meanwhile, automatically taking a picture of the ash samples using a black and white closed-loop digital camera every 0.5-l°C.
6. The method for determining an ash fusibility characteristic curve according to claim 1, wherein the composition indexes of the ash samples comprise: a slag viscosity index, an ash caking index, a dolomite ratio, a ratio of a basic oxide to an acid oxide of ash, a slagging index, and an aluminium titanium oxide content.
7. The method for determining an ash fusibility characteristic curve according to claim 1, wherein the BP neural network comprises: an input layer, a hidden layer, and an output layer.
8. The method for determining an ash fusibility characteristic curve according to claim 1, wherein the BP neural network is specifically trained in three stages: signals forward propagation, errors back propagation, weights &thresholds adjustment.
9. A system for determining an ash fusibility characteristic curve, comprising:an actual ash fusibility characteristic curve determination module configured to obtain an actual ash fusibility characteristic curve;an ash sample composition index determination module configured to calculate composition indexes of ash samples;a neural network construction module configured to construct a BP neural network;a training module configured to train the BP neural network by using the composition indexes of the ash samples and the actual ash fusibility characteristic curve; anda prediction module configured to determine a fusibility characteristic curve of unknown ash samples based on the trained BP neural network.10