CN111812149A - Adiabatic acceleration calorimetry method based on machine learning - Google Patents
Adiabatic acceleration calorimetry method based on machine learning Download PDFInfo
- Publication number
- CN111812149A CN111812149A CN202010698202.3A CN202010698202A CN111812149A CN 111812149 A CN111812149 A CN 111812149A CN 202010698202 A CN202010698202 A CN 202010698202A CN 111812149 A CN111812149 A CN 111812149A
- Authority
- CN
- China
- Prior art keywords
- adiabatic
- sample
- fitting
- temperature
- curve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N25/00—Investigating or analyzing materials by the use of thermal means
- G01N25/20—Investigating or analyzing materials by the use of thermal means by investigating the development of heat, i.e. calorimetry, e.g. by measuring specific heat, by measuring thermal conductivity
Abstract
The invention provides an adiabatic accelerated calorimetry method based on machine learning, which comprises the steps of obtaining a mathematical nonlinear exponential growth fitting formula for accurately describing an adiabatic heat release process of a sample through fitting of a mathematical nonlinear model according to temperature rise data of a heat release stage actually measured by the sample, enabling a temperature control system of an adiabatic accelerated calorimetry instrument to carry out temperature programming according to the obtained fitting formula, obtaining an experimental curve closer to an actual adiabatic temperature rise process of the sample, and further realizing more accurate adiabatic heat measurement of a thermal runaway process. The invention overcomes the measurement error caused by temperature tracking lag, and is beneficial to more fully and accurately carrying out later thermal and kinetic analysis and evaluation; the finally obtained adiabatic calorimetry experimental curve can effectively improve the accuracy of subsequent thermodynamic calculation of the sample, and has important guiding significance for guiding the implementation of reaction safety risk assessment and chemical safety evaluation.
Description
Technical Field
The invention provides an adiabatic accelerated calorimetry method based on machine learning, and belongs to the technical field of thermally safe accelerated calorimetry equipment for chemical and chemical process.
Background
In the chemical industry, the chemical production has the obvious characteristics of flammability, explosiveness, high temperature, high pressure, toxicity, harmfulness, corrosiveness, various dangerous chemicals, complex process, harsh process conditions and the like, most chemical reactions are exothermic reactions, the potential danger is high, and the importance of chemical safety production is self-evident; the thermal runaway of chemical reaction is one of the main causes of accidents, so the research and the recognition of the thermal danger of chemical reaction process and chemical materials are particularly critical.
The thermal risk of the process can be effectively evaluated by testing the thermodynamic, kinetic and thermal risk parameters of the reaction such as the heat release amount, the heat release rate, the adiabatic temperature rise, the apparent activation energy, the self-accelerating decomposition temperature and the like, and the parameters are obtained by a calorimetric device; at present, the calorimetric equipments commonly used mainly include thermogravimetric analyzers (TGA), Differential Scanning Calorimeters (DSC), adiabatic accelerated calorimeters (ARC), Reaction calorimeters (RC 1)e) The method comprises three operation modes of constant temperature, dynamic state and heat insulation; generally, the heat release speed of a chemical reaction system is far higher than the cooling and heat dissipation speed, and almost no heat exchange exists between the reaction system and the environment, so that the adiabatic mode is closer to the reaction mode when the actual working condition is out of control thermally, and the data such as time, temperature, pressure, adiabatic temperature rise, maximum temperature rise rate and the like measured in the reaction process in the mode are more suitable for simulating the working condition so as to more accurately analyze the reaction safety risk, and are very important for the conversion from laboratory scale to industrial production.
ARC supports three modes of isothermal (isothermal), scanning (scan), heat-wait-search (H-W-S), which simulates a potential thermal runaway reaction and quantifies the thermal risk of chemicals by maintaining the reaction system in an adiabatic environment; however, due to the thermal inertia problem of a thermal resistor or a thermocouple in a conventional ARC temperature sensor and the limitation of the performance of the device, when the thermal runaway reaction process of the chemical material is actually measured, the corresponding temperature rise rate cannot be achieved, the temperature of the furnace body cannot quickly follow the temperature of the sample, so that the heat generation of the sample is not completely used for the temperature rise of the sample, and extra heat loss exists; therefore, the internal environment of the reaction system cannot be guaranteed to be in an adiabatic state; thermal runaway is a chemical reaction phenomenon which is violent and has high harmfulness, and the reaction rate and the temperature rise rate are usually high, so that the influence is particularly obvious when rapid measurement is carried out, certain deviation exists in important parameters such as the measured heat release, adiabatic temperature rise, maximum temperature rise rate and the like, and the accurate measurement and analysis of the thermal runaway process are difficult to carry out.
In the past, certain development is carried out in the field of optimization and improvement of the performance of ARC equipment, Phi-TECII developed by HEL company improves the performance of ARC based on a pressure compensation technology, and reduces the thermal inertia of the equipment to a certain extent, but the experimental operation is complicated, the equipment failure rate is high, and the device is not convenient to widely popularize in experimental research; d-ARC that American Omnic Technology promoted adopts the power compensation technique to confirm that the reaction enthalpy that sample reaction produced is originated in the heating of outside furnace body, does not fully consider the heat exchange effect that exists between reaction process sample cell and the heating furnace body, and in the research of chemical reaction characteristic, the heat loss problem that the sample that surveys appears in the self-discharge process still exists, and this makes the reliability of thermodynamic analysis result reduce, awaits further solution.
In order to ensure that the thermal runaway reaction process of a sample is in a good adiabatic state, obtain more accurate thermodynamic data and enable the thermal risk assessment of chemicals or chemical reactions to be more in line with the actual situation, the invention provides an adiabatic accelerated calorimetry method based on machine learning, overcomes the defect of temperature tracking lag in the prior art, realizes dynamic testing of the adiabatic process, and provides guidance for research in the fields of reaction safety risk assessment, thermodynamics and the like of fine chemistry.
Disclosure of Invention
The invention provides an adiabatic acceleration calorimetry method based on machine learning, and aims to solve the problem that the temperature tracking process of the existing adiabatic acceleration calorimetry instrument in the thermal runaway stage of the adiabatic tracking sample to be detected has hysteresis.
The technical solution of the invention is as follows: a heat insulation acceleration calorimetric method based on machine learning comprises the steps of obtaining a mathematical nonlinear exponential growth fitting formula for accurately describing a heat insulation heat release process of a sample through mathematical nonlinear model fitting according to heat release stage temperature rise data measured actually by the sample, enabling a temperature control system of a heat insulation acceleration calorimeter to carry out temperature programming according to the obtained fitting formula, obtaining an experimental curve closer to the actual heat insulation temperature rise process of the sample, and further achieving more accurate heat insulation calorimetric measurement of a thermal runaway process.
The invention has the advantages that:
1) the method overcomes the measurement error caused by temperature tracking lag, is beneficial to more accurately carrying out the calculation of later thermodynamics and kinetic parameters, and is beneficial to more fully and accurately carrying out later thermal and kinetic analysis and evaluation;
2) the finally obtained adiabatic calorimetry experimental curve can effectively improve the accuracy of subsequent thermodynamic calculation of the sample, and has important guiding significance for guiding the implementation of reaction safety risk assessment and chemical safety evaluation.
Drawings
FIG. 1 is a mathematical second-order exponential growth fit function standard curve.
Fig. 2 is a flow chart of the experiment and analysis of the machine learning method.
FIG. 3 is a TBHP adiabatic temperature rise test curve E1。
FIG. 4 shows a fitting curve S for the adiabatic temperature rise of TBHP1。
FIG. 5 is a TBHP adiabatic temperature rise test curve E2。
FIG. 6 is a graph of temperature difference (Δ T ═ E)2-S1)。
Detailed Description
A heat insulation acceleration calorimetric method based on machine learning comprises the steps of obtaining a mathematical nonlinear exponential growth fitting formula for accurately describing a heat insulation heat release process of a sample through mathematical nonlinear model fitting according to heat release stage temperature rise data measured actually by the sample, enabling a temperature control system of a heat insulation acceleration calorimeter to carry out temperature programming according to the obtained fitting formula, obtaining an experimental curve closer to the actual heat insulation temperature rise process of the sample, and further achieving more accurate heat insulation calorimetric measurement of a thermal runaway process.
The mathematical nonlinear model employs a second-order exponential growth fit function.
The temperature rise data of the actually measured exothermic phase of the sample is obtained by the following steps:
(1) placing a substance to be measured with a certain mass specific heat capacity into the sample pool, inserting a thermocouple into the sample pool and installing the sample pool on the adiabatic acceleration calorimeter to ensure that the sample pool is positioned in the center of a hearth of the adiabatic acceleration calorimeter and the heating process is uniformly heated;
(2) enabling the adiabatic acceleration calorimeter to carry out a dynamic adiabatic calorimetry experiment in a standard heating-waiting-searching (H-W-S) mode through control software, measuring an adiabatic reaction process of a sample in real time according to a measuring module of a temperature control system of the adiabatic acceleration calorimeter, and detecting a thermal runaway occurrence process of the sample in real time according to a tracking state until the experiment is finished when detecting that the self-heating rate of a substance to be detected with known specific heat capacity is higher than a preset self-heat release rate; the preset self-heat release rate is a sample self-heat release rate value predicted by the control program, when the self-heat release rate of the sample exceeds the value, the adiabatic tracing state is entered, and the sample self-heat release rate value predicted by the control program is preferably 0.02 ℃/min.
The method comprises the following steps of obtaining a mathematical nonlinear exponential growth fitting formula for accurately describing the adiabatic heat release process of the heat-insulation material through mathematical nonlinear model fitting, and specifically comprises the following steps:
(1) in the adiabatic reaction experiment process, data are monitored and recorded through a data acquisition system, the data mainly comprise time, temperature and pressure, when the experiment is finished and the temperature and the pressure of a sample pool are reduced to room temperature, the data are extracted for offline calculation and analysis, and the specific analysis flow is as follows: accurately selecting a reaction interval based on data analysis software, fully considering the characteristics of thermal runaway behavior, selecting a second-order exponential growth fitting function to perform mathematical nonlinear fitting on a sample temperature-time curve, and continuously iterating until converging to obtain a fitting formula and a fitting curve which meet the precision requirement;
(2) repeated tests are carried out under the same experimental conditions, particularly, a temperature control system of the adiabatic accelerated calorimeter learns the adiabatic heat release temperature rise trend of a substance to be tested with known comparative heat capacity in advance based on the obtained fitting formula, temperature is raised according to the temperature control system, and then a corresponding experimental curve is obtained;
(3) comparing the temperature difference between the fitted curve and an experimental curve obtained by successive repeated tests, and when the difference requirement is met, wherein the difference requirement is that delta T is less than or equal to 1 ℃, judging that the retested experimental curve is a temperature rise curve which can most accurately describe the actual adiabatic heat release process of the sample; and when the difference value requirement is not met, repeating the experiment and the analysis steps, carrying out nonlinear fitting on the experimental data retested each time by adopting the same process, and enabling a temperature control system of the adiabatic acceleration calorimeter to carry out pre-learning according to a newly obtained fitting formula so as to obtain an experimental curve measured by specific programmed temperature rise until the temperature difference between the fitting curve and the successive experimental curve meets the requirement.
The method is suitable for exploring the adiabatic heat release process of the tert-butyl hydroperoxide (TBHP) standard solution according to the experiment of an adiabatic acceleration calorimeter, and realizes better thermal and kinetic analysis and evaluation.
Preferably, a TAC-500A type adiabatic acceleration calorimeter is used as an evaluation object, and the TAC-500A type adiabatic acceleration calorimeter comprises a hearth, a sample pool, a thermocouple, a pressure sensor, a pressure release valve, an exhaust fan and instrument control and data analysis software; the material of the sample pool is preferably any one of alloy, hastelloy and stainless steel, and can be selected according to the physicochemical characteristics and experimental requirements of a sample to be tested; the sample cell is an externally-inserted spherical container, and the bottom of the sample cell is used for containing a sample to be tested with certain quality and is tightly connected with the instrument through a fastener; the thermocouple is inserted from an inlet at the outer side of the sample cell, so that the accurate collection of the internal temperature of the sample cell is realized; the pressure sensor monitors the change of the pressure in the sample cell in the adiabatic reaction process in real time, the monitoring range is 0-200 bar, and when the pressure in the sample cell exceeds a set threshold value due to reaction heat release, the instrument can continuously give out an acousto-optic alarm to remind experimenters of safety; the instrument control software displays the temperature, the pressure and the experimental state in the whole testing process in a digital and image mode so as to know the experimental process in real time, the included data acquisition system realizes the acquisition and storage work of the temperature and pressure data, and the data analysis software is used for accurately selecting a reaction interval and solving theoretical thermodynamic parameters.
Further, the second-order exponential growth fitting function is selected to carry out mathematical nonlinear fitting on the temperature-time curve of the sample, iteration is continuously carried out until convergence is achieved, a fitting formula and a fitting curve meeting the precision requirement are obtained, and a correlation coefficient R is used2As an evaluation index, the larger the correlation coefficient is, the better the fitting effect of the obtained fitting formula is, and the accuracy of 99.90% is used as the critical criterion for function model selection, if R is2If the temperature is higher than 99.90%, the obtained fitting formula can well describe the temperature increase trend of the sample in the heat release stage in the heat insulation process; if R is2And if the temperature is lower than 99.90%, correcting the fitting parameters according to the obtained temperature rise data, iterating for multiple times until the fitting accuracy meets the requirement, and obtaining a corresponding adiabatic heat release fitting curve.
Furthermore, the second-order exponential growth fitting function does not contain any physicochemical parameter and is not influenced by the related properties of the sample to be detected; based on a mathematical nonlinear second-order exponential growth fitting function without any physicochemical parameter, experimental data are fitted by combining with preliminarily known adiabatic process heat release stage data of a measured sample to obtain a fitting curve and a fitting formula for describing the adiabatic heat release reaction.
The expression of the second order exponential growth fit function is as follows:
in the formula, x0Is the x offset; y is0Is the y offset; a. the1、A2Is the strength; t is t1、t2Is the magnitude of the increase; the fitting process takes the experimental time as an independent variable and the temperature of the sample pool as a dependent variable until the obtained fitting is publicThe formula converges and meets the accuracy requirement.
The invention provides an adiabatic accelerated calorimetry method based on machine learning, which is mainly applied to calorimetry test of a sample in an adiabatic exothermic stage in a heating-waiting-searching mode, wherein the heating-waiting-searching mode is a main operation mode of an adiabatic accelerated calorimeter and is most widely applied, so that the method can meet the requirements of different safety detection and risk assessment; the core of the invention lies in that the adiabatic acceleration calorimeter is used for developing a dynamic experiment of the adiabatic exothermic reaction according to the increasing trend of a fitting function based on machine learning, obtaining a temperature rise curve which is more in line with the reality and realizing the accurate evaluation of the adiabatic performance, and the finally obtained adiabatic exothermic experiment curve can effectively improve the accuracy of the subsequent thermodynamic calculation of the sample, has important guiding significance for the chemical reaction mechanism research, the guidance of the reaction safety risk evaluation and the chemical product safety evaluation, can effectively measure the adiabatic exothermic process of the sample and realize the more accurate adiabatic performance evaluation.
Compared with the existing measurement method of the adiabatic acceleration calorimeter, the invention has the following beneficial effects:
1) on the basis of repeated tests, a fitting formula can be continuously corrected, so that the machine learning effect of the instrument temperature control system at each time is closer to the temperature rise change of the actual thermal runaway process of the sample, the finally obtained experimental data overcomes the measurement error caused by temperature tracking lag, the calculation of later thermodynamics and kinetic parameters is facilitated to be more accurately carried out, and the later thermal and kinetic analysis and evaluation are facilitated to be more fully and accurately carried out;
2) according to the invention, the thermal runaway process of the sample can be accurately described through a function equation obtained by nonlinear fitting, and fitting parameters can be continuously corrected according to repeated experiments, so that the final adiabatic temperature rise experiment curve is closer to the actual adiabatic heat release temperature rise process of the sample;
3) according to the invention, the adiabatic accelerated calorimeter is enabled to learn the adiabatic temperature rise change trend of the sample in advance according to the obtained fitting formula by adopting a data integration mode, the accurate test of the test environment with higher temperature rise rate of the sample to be tested can be realized by continuously optimizing the fitting formula, and the accuracy of the subsequent thermodynamic research is improved; in addition, based on the advance learning of the fitting formula, the high heating rate test required by the adiabatic decomposition of the sample can be basically achieved.
Example 1
An adiabatic accelerated calorimetry method based on machine learning is based on a mathematical nonlinear exponential growth fitting formula without any physicochemical parameter, and is combined with preliminarily known adiabatic process heat release stage data of a measured sample to fit to obtain a fitting curve and a fitting formula for describing an adiabatic heat release reaction; and then carrying out repeated experiments, enabling a temperature control system of the instrument to carry out temperature programming according to a fitting formula so as to obtain an adiabatic reaction curve which is more in line with the actual situation, comparing the temperature difference between the fitting curve and an experimental curve obtained by carrying out repeated experiments by the corresponding fitting formula, and judging that the obtained experimental curve is most approximate to the actual adiabatic heat release process of the sample when the difference value is small enough (delta T is less than or equal to 1 ℃), thereby being beneficial to carrying out later-stage thermal and kinetic analysis and evaluation more fully and accurately.
The method comprises the following specific steps:
firstly, a sample to be tested with a certain mass specific heat capacity is placed in a sample pool of a testing furnace body of an adiabatic acceleration calorimeter, then a corrosion-resistant thermocouple is inserted into the sample pool, a dynamic adiabatic calorimetry experiment is carried out in a standard heating-waiting-searching (H-W-S) mode according to a measurement module of a temperature control system of the adiabatic acceleration calorimeter, and the adiabatic reaction process of the sample is measured in real time;
after the experiment is finished, selecting temperature data of a sample in an adiabatic process heat release stage, and performing off-line calculation on the temperature data, wherein a reaction system is an exothermic reaction, and the heat release rate shows an exponential growth trend along with the rise of the reaction temperature, so that a second-order exponential growth fitting function is selected to fit the temperature rise data, and the fitting result reaches a high-precision level by determining unknown parameters in a model; by a correlation coefficient (R)2) As an evaluation index, if R2Above 99.90%, it is determined that the resulting fit equation canThe temperature increase trend of the sample in the heat release stage of the adiabatic process is well described; if R is2And if the temperature is lower than 99.90%, correcting the fitting parameters according to the obtained temperature rise data, iterating for multiple times until the fitting accuracy meets the requirement, and obtaining a corresponding adiabatic heat release fitting curve.
The temperature control system of the instrument performs temperature programming according to the obtained exponential growth fitting formula meeting the precision requirement by a machine learning method, so that the temperature rise process of the furnace body is kept consistent with the temperature change of the heat release process of the sample as far as possible, no heat exchange is ensured between the furnace body and the sample pool, and the adiabatic temperature rise experiment curve more conforming to the actual situation is obtained.
Comparing the temperature difference between the fitting curve and an experimental curve obtained by carrying out repeated tests by using a fitting formula, judging that the obtained experimental curve is most approximate to the actual adiabatic heat release process of a sample when the difference value is small enough (delta T is less than or equal to 1 ℃), and carrying out mathematical exponential nonlinear fitting on experimental data obtained by repeated tests according to the same flow if the difference value is not small enough, so that the instrument is subjected to temperature programming according to the obtained fitting formula to obtain a corresponding experimental curve, and comparing the temperature difference between the fitting curve and the experimental curve obtained according to the temperature difference by using the same method; and (4) circulating the steps until the temperature difference meets the requirement, and obtaining an experimental temperature rise curve which is finally closest to the actual adiabatic heat release process of the sample.
The temperature measuring module of the temperature control system is connected with the thermocouple, and can accurately measure the temperature of the sample cell, the furnace cover, the furnace wall and the furnace bottom at the same time.
The instrument temperature control system carries out temperature programming according to the obtained fitting formula, so that the temperature change trend of the sample in the heat insulation and heat release process can be learned in advance, and no heat exchange between the furnace body and the sample pool in the test process can be ensured as far as possible.
The mathematical nonlinear exponential growth fitting formula used by the heat-release stage temperature rise data does not contain any physicochemical parameter and is not influenced by the related properties of the sample to be detected.
The mathematical nonlinear model used is a second-order Exponential Growth fitting function (Fit explicit Growth-second order), which is defined as follows:
in the formula, x0Is the x offset; y is0Is the y offset; a. the1、A2Is the strength; t is t1、t2Is the magnitude of the increase; in the fitting process, the experiment time is used as an independent variable, and the temperature of the sample pool is used as a dependent variable until the fitting equation is converged and meets the precision requirement.
The fitting formula can be continuously corrected and optimized through repeated experiments, and an experimental curve more in line with the actual adiabatic heat release process of the sample can be obtained according to the critical criterion of temperature difference between the fitting curve and the experimental curve obtained by carrying out specific temperature programming on the fitting formula; the instrument performs machine learning according to a fitting formula obtained each time, so that the temperature rise process of the furnace body is continuously close to the actual heat insulation and heat release process of the sample, and particularly, heat insulation tracking with better performance can be realized at a high temperature rise rate stage; the optimal fitting formula and fitting curve can be obtained through repeated experiments and multiple iterations, so that more accurate adiabatic heat release dynamic temperature rise test is realized.
Example 2
The embodiment provides an adiabatic accelerated calorimetry method based on machine learning, which is mainly applied to calorimetry tests in an adiabatic heat release stage of a sample in a heating-waiting-searching mode, wherein the heating-waiting-searching mode is a main operation mode of an adiabatic accelerated calorimeter and is most widely applied, so that the embodiment can meet requirements of different safety detection and risk assessment.
A method for adiabatic acceleration calorimetry based on machine learning is characterized in that a selected mathematical function model is used as a core, a chemical reaction system is exothermic in a thermal runaway process, the exothermic rate is gradually accelerated along with the rise of reaction temperature and is in an exponential growth trend, and meanwhile, the chemical reaction system has certain complexity, so that a second-order exponential growth fitting function is selected to describe the temperature change process of a sample more reasonably and comprehensively.
The exponential growth exponential function has a growth rate proportional to its function value, and the expression of the second-order exponential growth fitting function is as follows according to the definition of the exponential growth model (also called a markassis growth model):
in the formula, x0Is the x offset; y is0Is the y offset; a. the1、A2Is the strength; t is t1、t2Is the magnitude of the increase;
in the temperature rise data fitting process, the experimental time is used as an independent variable, the temperature of the sample pool is used as a dependent variable, and a fitting formula which accords with the adiabatic temperature rise trend of the sample can be obtained by determining fitting parameters and meeting the requirements of convergence and precision of a second-order exponential growth fitting function.
The second-order exponential growth fitting function does not contain any physicochemical parameter and is not influenced by the related properties of the sample to be tested, each sample can obtain a fitting formula for accurately describing the adiabatic temperature rise characteristic of the sample under a specific experimental condition, but the obtained fitting formula is not limited by the experimental conditions of the sample.
To further clarify the second-order exponential growth fitting function and its curve characteristics used in the present application, it is explained in detail with reference to fig. 1, based on the setting of the initial values of the fitting parameters (y)0＝0，x0＝0，A1＝1，t1＝1，A2＝2，t22) determining its standard curve and combining the function horizontal asymptote and the location at the characteristic point (x)0，y0+A1+A2) The tangent of (a) was analyzed by comparison.
The result is shown in FIG. 1, where the second-order exponential growth fitting function equation under the fitting parameter initial value condition is y1＝ex+2ex/2When the independent variable x goes to negative infinity, the function value is infinitely close to zero, and the horizontal asymptote of the function is y2When the value is 0, a feature point (x) is calculated0，y0+A1+A2) The slope of (a) is taken as the tangent equation y3＝2x+ 3。
As is clear from fig. 1, the second-order exponential growth fitting function grows faster and faster, and exhibits an "explosive" growth trend, which coincides with the temperature rise change during the adiabatic runaway process of the chemical reaction, indicating that the second-order exponential growth fitting function has high reliability and general applicability in this embodiment.
The method proposed in this example takes an adiabatic acceleration calorimeter model TAC-500A as an evaluation target, and it should be understood that the application of this example is not limited to this adiabatic acceleration calorimeter.
The experimental device of the TAC-500A type adiabatic acceleration calorimeter mainly comprises the following parts: a hearth, a sample cell and a thermocouple; the TAC-500A type adiabatic acceleration calorimeter also comprises a plurality of temperature control modes, including a constant temperature mode, a heating-waiting-searching mode and a scanning mode, wherein different heating modes are realized on a sample to be tested in the sample pool through the hearth, and the temperature control range is between room temperature and 500 ℃; the sample pool is prepared from various materials such as titanium alloy, hastelloy and stainless steel, and can be selected according to the physicochemical characteristics and experimental requirements of a sample to be tested; the thermocouple is inserted into the bottom of the sample cell from the outer side of the sample cell, is used for collecting the temperature in the sample cell, and has the characteristics of wide application range, high measurement precision and convenience in operation; and after the sample cell is installed on the instrument testing pipeline, the sample cell is placed into the hearth to set a subsequent experiment mode.
The TAC-500A type adiabatic acceleration calorimeter also comprises a pressure sensor and a pressure release valve: the pressure sensor is connected with the instrument testing pipeline and can monitor the pressure change of the reaction system in the sample cell in real time; the instrument operation process can guarantee the gas tightness of reaction system based on the relief valve to pressure detection inefficacy can lead to system pressure to rise suddenly when reaction exothermal process temperature sharply risees, so need slowly release system pressure through the relief valve earlier after the experiment, prevent that the sample cell dismantles the process and take place danger.
The TAC-500A type adiabatic acceleration calorimeter also comprises instrument control and data analysis software: the instrument control software is used for setting an experiment mode, corresponding temperature control parameters and experiment information, the change of the pressure and the temperature in the sample cell can be digitally and dynamically displayed in the experiment process, the experiment stage can be known in real time in an image form, and the included data acquisition system realizes the acquisition and storage work of the temperature and the pressure data in the whole process; and the data analysis software is used for accurately selecting a reaction interval and solving theoretical thermodynamic parameters.
In this embodiment, the basic principle and the implementation flow of the adiabatic acceleration calorimetry method based on machine learning are as follows:
(1) a sample to be tested with known mass specific heat capacity is placed in a sample pool, a thermocouple is inserted, and after the sample to be tested and a TAC-500A type adiabatic acceleration calorimeter are installed, temperature control parameters and experimental information are set through instrument control software so that the adiabatic acceleration calorimeter can run an experiment in a standard heating-waiting-searching mode; after the experiment is finished, performing mathematical nonlinear fitting on the temperature rise data of the sample adiabatic exothermic reaction by using a second-order exponential growth fitting function formula to obtain a fitting formula which is convergent and meets the precision requirement and a corresponding fitting curve, and respectively recording the experiment curve, the fitting curve and the fitting formula as E1、S1、F1；
(2) The repeated adiabatic calorimetry experiment is carried out on a sample to be tested under the same experiment conditions, and can be specifically described as follows: enabling the instrument temperature control system to obtain a fitting formula F by means of data integration1Learning the temperature variation trend of the sample in the heat release stage in advance, and carrying out specific temperature programming according to the temperature variation trend to obtain an adiabatic temperature rise experimental curve (E) actually measured by the sample2) (ii) a Fitting curve S if adiabatic temperature rise1And adiabatic temperature rise test curve E2The temperature difference between the two is small enough (delta T is less than or equal to 1 ℃), the fitting formula S is shown1Can well describe the actual adiabatic temperature rise change trend of the sample and judge the adiabatic temperature rise experiment curve E of the retest2The temperature rise change course of the real thermal runaway process of the sample can be better expressed, whether curve fitting and adiabatic calorimetry experiments are repeatedly carried out according to the process is judged, and the temperature control system of the instrument obtains a fitting formula S according to correction and optimization every time in the processiPerforming a pre-machine learning until an adiabatic temperature rise test curve E is obtained therefromi+1And fitting curve SiUntil the temperature difference between the two meets the error requirement; in summary, the proposed machine-based approachThe specific experimental and analytical flow of the adiabatic acceleration calorimetry method of the warewasher is shown in fig. 2.
The key technology of the embodiment is the accurate acquisition of the fitting function and the machine learning process of the fitting function obtained by correcting and optimizing each time by the instrument temperature control system.
When the thermal behavior of the sample in the adiabatic state is measured, the reaction process and the adiabatic temperature rise interval of the sample cannot be predicted due to the fact that the physicochemical characteristics of the sample cannot be well mastered; the reaction system has certain complexity, the sample is in a self-reaction heat release state in the heat insulation heat release stage, the temperature shows the trend of slowly increasing and then violently changing, the problem that the heat release temperature of the sample cannot be tracked timely by the heating furnace body exists, and data deviation is generated; therefore, the fitting function needs to have higher accuracy and can effectively describe the temperature rise characteristics of the sample; furthermore, when the instrument temperature control system implements a dynamic machine learning process on the fitting function, the instrument temperature control system needs to have good adaptability on the fitting function so as to quickly and accurately detect the adiabatic heat release process of the sample in the testing process, obtain more accurate calorimetric data and provide a data base for later thermodynamic parameter solution.
Example 3
The decomposition process of organic peroxides is typically an exothermic reaction, with slow to fast decomposition rates, which self-thermal decomposition occurs when the temperature rises to a certain extent, eventually leading to thermal runaway or more serious damage; in this example, a standard solution of t-butyl hydroperoxide (TBHP) is used as an example, and the adiabatic heat release process of TBHP is explored according to an adiabatic accelerated calorimeter experiment, so that better thermal and kinetic analysis and evaluation are realized.
The specific analysis flow is as follows: the adiabatic acceleration calorimeter is selected as a heating-waiting-searching (H-W-S) mode, the temperature of a starting interval is set to be 70 ℃, the constant temperature time of the starting interval is 40min, the step temperature rise rate is 10 ℃/min, the step temperature rise step length is 5 ℃, the temperature detection threshold value is 0.02 ℃/min, and specific initial parameters are shown in table 1; when the adiabatic acceleration calorimeter searches that the sample starts to release heat, the sample enters an adiabatic tracing state until the exothermic reaction is finished, and the sample enters a cooling state; the control software of the adiabatic acceleration calorimeter realizes real-time recording of temperature and pressure data in the reaction process, derives experimental temperature rise data to obtain a fitting formula meeting the precision requirement through fitting, enables the adiabatic acceleration calorimeter to simulate the heat release temperature rise process of a sample according to the obtained fitting formula, and calculates thermodynamic parameters based on the adiabatic data obtained through simulation.
TABLE 1 initial parameters of the experiment
FIG. 3 is the adiabatic temperature rise test curve E of TBHP1And detecting that the sample generates an exothermic reaction within the range of about 93.876-195.538 ℃, wherein the temperature of the sample rises slowly in the initial stage of the reaction, and the temperature change is more obvious along with the accumulation of heat.
Adiabatic temperature rise Experimental Curve E from FIG. 3TBHP1When the temperature rise rate is high, the data fluctuation condition appears, which indicates that the furnace body cannot realize excellent temperature tracking effect, so that the sample pool generates heat loss, the adiabatic decomposition process of the TBHP cannot be completely detected, and the adiabatic measurement result has certain deviation; therefore, mathematical nonlinear fitting is carried out according to the proposed exponential growth model, the experimental data is divided into two parts to obtain better fitting effect, and the obtained fitting formula F1The following formulae (3) and (4), corresponding to R299.99% and 99.90%, respectively, showed excellent fitting degree.
Drawing a temperature rise fitting curve S in a corresponding heat release time range through the obtained fitting function formulas (3) and (4)1The fitting result is shown in FIG. 4, and the fitting curve S of adiabatic temperature rise can be seen1And adiabatic temperature rise test curve E1With a high degree of consistency therebetween.
Instrument temperature control system rootThe adiabatic temperature rise test curve E obtained by performing a repeated test by performing programmed temperature rise through the advance machine learning of the adiabatic reaction temperature rise tendency of the sample according to the formulas (3) and (4)2As shown in FIG. 5, the corresponding temperature rise range of the heat release is 93.703-204.776 ℃; compared with the experimental curve E1In a word, E2The self-heat-release state of the sample is detected at the stage of high temperature rise rate, and the self-heat-release state is closer to the actual adiabatic reaction mode of the sample, so that the defect of the instrument in performance is effectively overcome based on the machine learning method, and timely adiabatic tracking is realized.
For curve S1And E2The temperature difference between the two is analyzed, the comparison result is shown in figure 6, and the maximum temperature difference value delta T is foundmax0.294 ℃, within the required error range, indicating excellent effectiveness and reliability of the method of the invention; therefore, the fitting formula is not modified to repeatedly perform the adiabatic test based on the machine learning of the temperature control system.
In conclusion, the proposed adiabatic acceleration calorimeter method based on machine learning overcomes the defects of the traditional adiabatic acceleration calorimeter in adiabatic tracking; the method obtains a second-order exponential growth fitting function capable of accurately describing the exothermic reaction process through the initial adiabatic heat measurement test of the sample, further enables an instrument temperature control system to learn the adiabatic temperature rise trend of the tested sample in advance based on the obtained fitting function so as to repeatedly carry out adiabatic experiments, obtains temperature rise data more conforming to the actual thermal runaway reaction process of the sample, is beneficial to improving the accuracy of subsequent thermodynamic parameter calculation, and comprehensively shows that the method has important guiding significance for effectively implementing reaction safety risk assessment and chemical safety evaluation.
Claims (10)
1. A heat insulation acceleration calorimetric method based on machine learning is characterized in that the method comprises the steps of obtaining a mathematical nonlinear exponential growth fitting formula for accurately describing a heat insulation heat release process of a sample through mathematical nonlinear model fitting according to temperature rise data of a heat release stage actually measured by the sample, enabling a temperature control system of a heat insulation acceleration calorimeter to carry out temperature programming according to the obtained fitting formula, obtaining an experimental curve closer to the actual heat insulation temperature rise process of the sample, and further achieving more accurate heat insulation calorimetric measurement of a thermal runaway process.
2. The method of claim 1, wherein the mathematical nonlinear model uses a second-order exponential growth fit function.
3. The machine learning-based adiabatic accelerated calorimetry method according to claim 1 or 2, wherein the measured exothermic phase temperature rise data of the sample is obtained by:
(1) placing a substance to be measured with a certain mass specific heat capacity in a sample pool, inserting a thermocouple in the sample pool, and then installing the sample pool on an adiabatic acceleration calorimeter to ensure that the sample pool is positioned in the center of a hearth of the adiabatic acceleration calorimeter and the heating process is uniformly heated;
(2) the method comprises the steps of enabling an adiabatic acceleration calorimeter to carry out a dynamic adiabatic calorimetry experiment in a standard heating-waiting-searching mode through control software, measuring an adiabatic reaction process of a sample in real time according to a measuring module of a temperature control system of the adiabatic acceleration calorimeter, and detecting a thermal runaway occurrence process of the sample in real time according to an adiabatic tracking state until the experiment is finished when a self-heating rate of a substance to be detected with known specific heat capacity is higher than a self-heat release rate preset by a program.
4. The adiabatic acceleration calorimetry method based on machine learning according to claim 2, wherein the mathematical nonlinear exponential growth fitting formula for accurately describing the adiabatic heat release history is obtained by fitting a mathematical nonlinear model, and the method specifically comprises the following steps:
(1) in the adiabatic reaction experiment process, data are monitored and recorded through a data acquisition system, the data comprise time, temperature and pressure, when the experiment is finished and the temperature and the pressure of a sample pool are reduced to room temperature, the data are extracted for offline calculation and analysis, and the specific analysis process is as follows: accurately selecting a reaction interval based on data analysis software, selecting a second-order exponential growth fitting function to perform mathematical nonlinear fitting on a sample temperature-time curve according to the characteristics of thermal runaway behavior, and continuously iterating until convergence to obtain a fitting formula and a fitting curve which meet the precision requirement;
(2) repeated tests are carried out under the same experimental conditions, particularly, a temperature control system of the adiabatic accelerated calorimeter learns the adiabatic heat release temperature rise trend of a substance to be tested with known comparative heat capacity in advance based on the obtained fitting formula, temperature is raised according to the temperature control system, and then a corresponding experimental curve is obtained;
(3) comparing the temperature difference between the fitted curve and an experimental curve obtained by successive repeated tests, and judging the retested experimental curve to be a temperature rise curve which can most accurately describe the actual adiabatic heat release process of the sample when the difference requirement is met; and when the difference value requirement is not met, repeating the experiment and the analysis steps, carrying out nonlinear fitting on the experimental data retested each time by adopting the same flow, and enabling a temperature control system of the adiabatic acceleration calorimeter to carry out pre-learning according to a newly obtained fitting formula so as to obtain an experimental curve measured by specific programmed temperature rise until the temperature difference between the fitting curve and the successively measured curve meets the requirement.
5. The adiabatic acceleration calorimetry method based on machine learning of claim 4, wherein the second-order exponential growth fitting function is selected to perform mathematical nonlinear fitting on the sample temperature-time curve, and the iteration is continued until convergence to obtain a fitting formula and a fitting curve which meet the accuracy requirement, and a correlation coefficient R is used2As an evaluation index, the larger the correlation coefficient is, the better the fitting effect of the obtained fitting formula is, and the accuracy of 99.90% is used as the critical criterion for function model selection, if R is2If the temperature is higher than 99.90%, the obtained fitting formula can well describe the temperature increase trend of the sample in the heat release stage in the heat insulation process; if R is2And if the temperature is lower than 99.90%, correcting the fitting parameters according to the obtained temperature rise data, iterating for multiple times until the fitting accuracy meets the requirement, and obtaining a corresponding adiabatic heat release fitting curve.
6. The method of claim 2, 4 or 5, wherein the second-order exponential growth fitting function does not contain any physicochemical parameter and is not affected by the related properties of the sample to be tested.
7. The method of claim 2, 4 or 5, wherein the second-order exponential growth fitting function is expressed as follows:
in the formula, x0Is the x offset; y is0Is the y offset; a. the1、A2Is the strength; t is t1、t2Is the magnitude of the increase; in the fitting process, the experiment time is used as an independent variable, and the temperature of the sample pool is used as a dependent variable until the obtained fitting formula is converged and meets the precision requirement.
8. The method of claim 1, 2, 4 or 5, wherein the method is suitable for exploring the adiabatic heat release process of a t-butyl hydroperoxide standard solution according to the adiabatic acceleration calorimeter experiment.
9. The machine learning-based adiabatic acceleration calorimetry method according to claim 3, wherein the programmed self-heat-release rate is a value of the self-heat-release rate of the sample predicted by the control program, and when the self-heat-release rate of the sample exceeds the value, the adiabatic tracing state is entered; the control program predicted a sample self-heat-release rate of 0.02 ℃/min by default.
10. The machine learning-based adiabatic acceleration calorimetry method according to claim 4, wherein the difference in temperature difference is required to be Δ T ≦ 1 ℃.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010698202.3A CN111812149B (en) | 2020-07-20 | 2020-07-20 | Adiabatic acceleration calorimetry method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010698202.3A CN111812149B (en) | 2020-07-20 | 2020-07-20 | Adiabatic acceleration calorimetry method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111812149A true CN111812149A (en) | 2020-10-23 |
| CN111812149B CN111812149B (en) | 2021-10-15 |
Family
ID=72865690
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010698202.3A Active CN111812149B (en) | 2020-07-20 | 2020-07-20 | Adiabatic acceleration calorimetry method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111812149B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112685901A (en) * | 2020-12-31 | 2021-04-20 | 中国原子能科学研究院 | Calorimetric calculation method |
| CN113030173A (en) * | 2021-04-06 | 2021-06-25 | 南京工业大学 | Adiabatic acceleration calorimeter based on surface temperature measurement in sample cell |
| CN113109391A (en) * | 2021-05-21 | 2021-07-13 | 中国计量大学 | Adiabatic reaction dynamic specific heat capacity measuring method based on compensation power modulation |
| CN113125047A (en) * | 2021-04-19 | 2021-07-16 | 福州大学 | Calorimetric device based on calibration calorimetry and measuring method |
| CN113340937A (en) * | 2021-05-10 | 2021-09-03 | 武汉深维鼎测科技有限公司 | Rapid heat value measuring method and system based on oxygen bomb calorimeter |
| CN113506924A (en) * | 2021-06-17 | 2021-10-15 | 重庆金康动力新能源有限公司 | Thermal runaway early warning method and system for battery pack |
| CN114218645A (en) * | 2021-12-16 | 2022-03-22 | 哈尔滨工业大学 | Box girder steam-curing thermal coupling field theoretical model with physical property space-time variation and calculation method |
| CN115273998A (en) * | 2022-08-17 | 2022-11-01 | 沈阳化工研究院有限公司 | Chemical reaction risk analysis method |
| CN115359610A (en) * | 2022-07-04 | 2022-11-18 | 珠海派诺科技股份有限公司 | Temperature measurement type electrical fire monitoring method, equipment and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107389723A (en) * | 2017-07-31 | 2017-11-24 | 中国计量大学 | A kind of adiabatic reaction starting point based on difference variation judges and temperature method for tracing |
| CN107729625A (en) * | 2017-09-25 | 2018-02-23 | 江苏英索纳智能科技有限公司 | The method and device that thermometric error caused by a kind of operation heating to equipment compensates |
| CN110082429A (en) * | 2019-04-19 | 2019-08-02 | 四川升拓检测技术股份有限公司 | A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning |
| CN110146825A (en) * | 2019-04-24 | 2019-08-20 | 天津力神电池股份有限公司 | A kind of method of quick evaluation and test lithium ion battery security |
| CN110715953A (en) * | 2019-09-18 | 2020-01-21 | 浙江大学 | System and method for testing heat-conducting property of film material based on machine learning |
| CN110736764A (en) * | 2019-10-16 | 2020-01-31 | 杭州仰仪科技有限公司 | lithium battery specific heat capacity measuring method and device based on differential adiabatic tracing |
| US20200063665A1 (en) * | 2018-01-25 | 2020-02-27 | Dalian University Of Technology | Aero-engine full flight envelope model adaptive modification method based on deep learning algorithm |
| WO2020070786A1 (en) * | 2018-10-02 | 2020-04-09 | 株式会社島津製作所 | Method for creating discriminator |
-
2020
- 2020-07-20 CN CN202010698202.3A patent/CN111812149B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107389723A (en) * | 2017-07-31 | 2017-11-24 | 中国计量大学 | A kind of adiabatic reaction starting point based on difference variation judges and temperature method for tracing |
| CN107729625A (en) * | 2017-09-25 | 2018-02-23 | 江苏英索纳智能科技有限公司 | The method and device that thermometric error caused by a kind of operation heating to equipment compensates |
| US20200063665A1 (en) * | 2018-01-25 | 2020-02-27 | Dalian University Of Technology | Aero-engine full flight envelope model adaptive modification method based on deep learning algorithm |
| WO2020070786A1 (en) * | 2018-10-02 | 2020-04-09 | 株式会社島津製作所 | Method for creating discriminator |
| CN110082429A (en) * | 2019-04-19 | 2019-08-02 | 四川升拓检测技术股份有限公司 | A kind of auxiliary judgement method of the tunnel-liner non-destructive testing of combination machine learning |
| CN110146825A (en) * | 2019-04-24 | 2019-08-20 | 天津力神电池股份有限公司 | A kind of method of quick evaluation and test lithium ion battery security |
| CN110715953A (en) * | 2019-09-18 | 2020-01-21 | 浙江大学 | System and method for testing heat-conducting property of film material based on machine learning |
| CN110736764A (en) * | 2019-10-16 | 2020-01-31 | 杭州仰仪科技有限公司 | lithium battery specific heat capacity measuring method and device based on differential adiabatic tracing |
Non-Patent Citations (2)
| Title |
|---|
| 丁炯 等: "基于动态特性补偿的绝热加速量热仪温度随动控制优化", 《传感技术学报》 * |
| 谢嘉林: "绝热反应量热仪功率补偿技术研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅰ辑》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112685901A (en) * | 2020-12-31 | 2021-04-20 | 中国原子能科学研究院 | Calorimetric calculation method |
| CN112685901B (en) * | 2020-12-31 | 2024-03-22 | 中国原子能科学研究院 | Calorimeter calculation method |
| CN113030173A (en) * | 2021-04-06 | 2021-06-25 | 南京工业大学 | Adiabatic acceleration calorimeter based on surface temperature measurement in sample cell |
| WO2022213574A1 (en) * | 2021-04-06 | 2022-10-13 | 南京工业大学 | Adiabatic accelerating rate calorimeter based on sample cell inner surface temperature measurement |
| CN113030173B (en) * | 2021-04-06 | 2021-09-14 | 南京工业大学 | Adiabatic acceleration calorimeter based on surface temperature measurement in sample cell |
| CN113125047A (en) * | 2021-04-19 | 2021-07-16 | 福州大学 | Calorimetric device based on calibration calorimetry and measuring method |
| CN113340937B (en) * | 2021-05-10 | 2022-05-13 | 武汉深维鼎测科技有限公司 | Rapid heat value measuring method and system based on oxygen bomb calorimeter |
| CN113340937A (en) * | 2021-05-10 | 2021-09-03 | 武汉深维鼎测科技有限公司 | Rapid heat value measuring method and system based on oxygen bomb calorimeter |
| CN113109391B (en) * | 2021-05-21 | 2022-03-04 | 中国计量大学 | Adiabatic reaction dynamic specific heat capacity measuring method based on compensation power modulation |
| WO2022241866A1 (en) * | 2021-05-21 | 2022-11-24 | 中国计量大学 | Method for measuring dynamic specific heat capacity of adiabatic reaction on basis of compensation power modulation |
| CN113109391A (en) * | 2021-05-21 | 2021-07-13 | 中国计量大学 | Adiabatic reaction dynamic specific heat capacity measuring method based on compensation power modulation |
| CN113506924A (en) * | 2021-06-17 | 2021-10-15 | 重庆金康动力新能源有限公司 | Thermal runaway early warning method and system for battery pack |
| CN113506924B (en) * | 2021-06-17 | 2023-03-21 | 重庆金康动力新能源有限公司 | Thermal runaway early warning method and system for battery pack |
| CN114218645A (en) * | 2021-12-16 | 2022-03-22 | 哈尔滨工业大学 | Box girder steam-curing thermal coupling field theoretical model with physical property space-time variation and calculation method |
| CN114218645B (en) * | 2021-12-16 | 2022-07-01 | 哈尔滨工业大学 | Box girder steam-curing thermodynamic coupling field theoretical model with physical property space-time variation and calculation method |
| CN115359610A (en) * | 2022-07-04 | 2022-11-18 | 珠海派诺科技股份有限公司 | Temperature measurement type electrical fire monitoring method, equipment and storage medium |
| CN115273998A (en) * | 2022-08-17 | 2022-11-01 | 沈阳化工研究院有限公司 | Chemical reaction risk analysis method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111812149B (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111812149B (en) | Adiabatic acceleration calorimetry method based on machine learning | |
| EP3575892B1 (en) | Model parameter value estimation device and estimation method, program, recording medium with program recorded thereto, and model parameter value estimation system | |
| Roduit et al. | The simulation of the thermal behavior of energetic materials based on DSC and HFC signals | |
| US4130016A (en) | Adiabatic calorimeter apparatus and method for measuring the energy change in a chemical reaction | |
| CN109974902B (en) | Adiabatic acceleration calorimeter with dynamic thermal inertia correction characteristic | |
| Brown | Introduction to thermal analysis: techniques and applications | |
| CN107389723B (en) | A kind of judgement of adiabatic reaction starting point and temperature method for tracing based on difference variation | |
| Kossoy et al. | Specific features of kinetics evaluation in calorimetric studies of runaway reactions | |
| CN113272748A (en) | Analysis system and analysis method | |
| CN110957011A (en) | On-line production parameter estimation method of continuous stirring reactor under unknown time-varying measurement noise | |
| CN110010205A (en) | A method of assessment self-catalysis substance thermal stability | |
| CN113030173A (en) | Adiabatic acceleration calorimeter based on surface temperature measurement in sample cell | |
| Xu et al. | Modeling of a power compensated adiabatic reaction system for temperature control design and simulation analyses | |
| CN109084834B (en) | Liquid level and temperature measuring device suitable for titanium sponge reduction reaction | |
| McIntosh et al. | Obtaining more, and better, information from simple ramped temperature screening tests | |
| Iwata et al. | Thermal risk evaluation of organic peroxide by automatic pressure tracking adiabatic calorimeter | |
| CN103413029A (en) | Moving horizon estimation method for multispeed sampling continuous agitating pot type reactor | |
| Laszczyk et al. | Modelling of hybrid CSTR plant: Heat transfer considerations | |
| Wilson et al. | Observability matrix condition number in design of measurement strategies | |
| CN112255259A (en) | Test method for testing thermal stability of substance under constant pressure state | |
| Chen et al. | Heat Dissipation Analysis and Correction of the Sample in Accelerating Rate Calorimetric Experiment | |
| Kossoy et al. | Methodological aspects of the application of adiabatic calorimetry for thermal safety investigation | |
| CN109387537A (en) | Directly measure the device and method of chemicals self-accelerating decomposition temperature | |
| CN117235956A (en) | Adiabatic acceleration calorimetric method with rapid reaction heat dissipation correction | |
| JP2006084181A (en) | Temperature reactivity coefficient separate measuring method of pressurized water reactor |
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 |