A method for mediating biochar modification process based on microscopic images

Abstract

The present invention belongs to the field of dynamic regulation and machine vision technology, and proposes a biochar modification process mediation method based on microscopic images, which collects data from a rotary kiln in real time to obtain a monitoring map, pre-processes the monitoring map to form a monitoring map sequence, identifies the pore area ratio and the collapsed pore ratio from any monitoring map in the monitoring map sequence, calculates the mediation strategy value according to the pore area ratio and the collapsed pore ratio, and performs modification mediation according to the mediation strategy value. The possibility or risk of imbalance between the effective porosity and the degree of collapse phenomenon in the modification process is extracted according to the time series changes of the measured data, so as to identify the most extreme conditions of the imbalance phenomenon in the preset working period, thereby providing more sufficient optimization opportunities for increasing the specific surface area and porosity of biochar in the biochar modification function, bringing gains outside the conservative range to the performance of biochar, and ensuring the stability of the yield rate and functional utility of biochar.

Description

Technical Field

[0001] The invention belongs to the field of dynamic regulation and machine vision technology, and specifically relates to a biochar modification process mediation method based on microscopic images. Background Art

[0002] Biochar is a carbon-rich material produced by the thermal combustion of organic raw materials under a certain limited oxygen. It can remove heavy metals and organic pollutants by adsorption. Biochar is particularly important in the field of water treatment. Increasing the specific surface area and porosity through modification is an effective method to improve the adsorption capacity of biochar. In order to increase the specific surface area and porosity of biochar and improve the adsorption capacity, it is usually modified to obtain modified biochar. The rotary kiln, which can make full use of physical or chemical mechanisms for modification, is a commonly used modification equipment. It can remove impurities such as ash on the surface of biochar and increase the number of pores, thereby increasing the specific surface area and porosity of biochar. However, if the modification time is too long, excessive modification will lead to pore collapse and a decrease in the number of effective pores, which will reduce the adsorption performance of biochar. Therefore, a reasonable modification time study is crucial for the performance optimization of modified biochar.

[0003] The current process usually only relies on experimental methods to determine the modification time. This method mainly includes experimental observation and analysis of the specific surface area and porosity performance of biochar at different times to find the best modification time. However, the modification termination control method of the preset time period is usually set to a more conservative modification time in order to reduce the loss of adsorption performance of biochar caused by the outbreak of pore collapse. When several optional modification times are obtained, a shorter option is usually selected to adapt to the risk of pore collapse outbreak, which will greatly waste the optimization opportunity of increasing the specific surface area and porosity of biochar in the biochar modification function. Therefore, there is an urgent need for a biochar modification process mediation method based on microscopic images to provide a more scientific identification method for the outbreak position of pore collapse in time, and avoid the loss of porosity optimization opportunities caused by this traditional termination control method based on the minimum termination time length. Summary of the invention

[0004] The purpose of the present invention is to propose a biochar modification process mediation method based on microscopic images to solve one or more technical problems existing in the prior art and at least provide a beneficial option or create conditions.

[0005] In order to achieve the above object, according to one aspect of the present invention, a method for mediating a biochar modification process based on microscopic images is provided, the method comprising the following steps:

[0006] The invention relates to a method for obtaining a monitoring map by collecting data from a rotary kiln in real time; preprocessing the monitoring map to form a monitoring map sequence; identifying the pore area ratio and the collapsed pore ratio from any monitoring map in the monitoring map sequence; calculating a mediation strategy value according to the pore area ratio and the collapsed pore ratio; and mediating the rotary kiln according to the mediation strategy value.

[0007] Furthermore, a method for acquiring real-time data from a rotary kiln to obtain a monitoring diagram is as follows: a modifier is a weak acid solution, including at least one of an acetic acid solution, a formic acid solution or a citric acid solution, and the modifier is injected or sprayed onto the biochar; the biochar is placed in the rotary kiln after preheating, and the biochar is selected from two or more of tobacco stem biochar, straw biochar, shell biochar, wood biochar, pine cone biochar and rice husk biochar; data of the biochar is acquired from a sampling window area arranged on the side wall of the rotary kiln through an electron microscope to obtain a monitoring diagram; and the scanning electron microscope acquisition interval is set to 5 to 30 seconds per time.

[0008] Specifically, Hitachi SU3500 was used as the electron microscope; after starting the heating system of the kiln body, the heating rate was 8-12°C / min, and data collection began after the temperature reached 450-550°C, and the recording time was recorded as Ts; data collection utilized a scanning electron microscope (SEM) installed in the fixed kiln wall at the kiln head of the rotary kiln to capture the microscopic image of the biochar surface, i.e., the monitoring image, in real time during the heating process of the rotary kiln.

[0009] Furthermore, the method of preprocessing the monitoring image to form a monitoring image sequence is: the preprocessing includes denoising the monitoring image and histogram equalization adjustment, wherein the denoising is at least one of wide wavelet coefficient denoising or Gaussian filtering denoising, and the histogram equalization adopts RMSHE recursive equalization, and the monitoring image obtained in real time and preprocessed is written into a sequence recorded as a monitoring image sequence.

[0010] Among them, histogram equalization adjustment enhances the contrast of the image by adjusting the histogram of the image, which can improve the recognition of pores that are originally difficult to detect or pore details that are easily ignored.

[0011] Further, a method for identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence is: identifying the pore area and the collapsed area from the processed image respectively by a pre-trained CNN convolutional neural network model, wherein the collapsed area is a subset of the pore area; defining the ratio of the area of ​​the pore to the area of ​​the entire image as the pore area ratio;

[0012] The morphological opening method is used to segment the image from the pore area and obtain several pore units. If the pore unit contains a collapsed area, it is marked as collapsed. The ratio of the number of pore units marked as collapsed to the total number of pore units is recorded as the collapsed pore ratio.

[0013] The collapsed region is a subset of the pore region, which means that any collapsed region belongs to the pore region;

[0014] The method for obtaining the pre-trained CNN convolutional neural network model is as follows: prepare training data containing pore region annotations and select a suitable CNN architecture, including any one of YOLOv5, ResNet-101, VGG19 or Inception v4;

[0015] The training data is input into the convolutional neural network of the CNN architecture, and features are extracted through multi-layer convolution and pooling operations. The obtained features are used to predict the pore area through the fully connected layer or segmentation layer. Through back propagation and optimization algorithms, the Adam method is preferred, and the network weights are iteratively adjusted to gradually improve the model's ability to identify pore areas on the training set. Finally, the accuracy and generalization ability of the model are verified with the test set, and the pre-trained CNN convolutional neural network model is obtained.

[0016] Furthermore, a method for calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio is: recording the monitoring time scale as the monitoring point Ti, i.e., Ti as the serial number of the monitoring point, obtaining the pore area ratio and the collapsed pore ratio at the monitoring point; each monitoring point obtains the corresponding monitoring binary values, which are recorded as the pore area ratio Porat and the collapsed pore ratio Corat, respectively, wherein the binary value refers to the element in the binary group; the monitoring period defaults to the period between the start of the rotary kiln and the current moment; due to the insufficient sensitivity of the initial data, the starting point of the monitoring period can be set to 5-20 minutes after startup.

[0017] The average value and interquartile range of all collapsed pore ratios during the monitoring period are recorded as the collapse mean Av_Cor and the sensitive parameter Se_Par, respectively, where the interquartile range is the difference between the upper quartile and the lower quartile; for any monitoring point during the monitoring period, the positive collapse deviation is recorded as: Co_Dev + Ti =max{0,Co_Dev + Ti-1 +(Corat Ti -Av_Cor)}, the negative collapse deviation is: Co_Dev - Ti =max{0,Co_Dev - Ti-1 +(Av_Cor-Corat Ti )}, where when Ti = 1, that is, at the initial monitoring point within the monitoring period, the positive collapse deviation Co_Dev + and the collapse negative deviation Co_Dev - Set to 0, Co_Dev +Ti-1 Represents the positive deviation of the collapse of the Ti-1th monitoring point , Corat Ti Represents the collapsed pore ratio of the Ti-th monitoring point.

[0018] For any monitoring point, if its positive collapse deviation or negative collapse deviation is greater than half of the sensitive parameter, the monitoring point is recorded as the first collapse position, and the period between any first collapse position and the first first collapse position in the reverse time direction is defined as a sub-collapse interval; j is used as the corresponding serial number of the monitoring point between the first collapse position in the sub-collapse interval, and the volatility characteristic vector of the i0th monitoring point in the sub-collapse interval is: Vo_tor i0 =(|(Porat i0 -Porat i0-1 )(Corat i0 -Corat i0-1 )|,|(Porat i0+1 -Porat i0 )(Corat i0+1 -Corat i0 )|);

[0019] Porat i0-1 , Porat i0 , Porat i0+1 Corat represents the pore area ratio of the i0-1, i0, and i0+1 monitoring points between the first collapse position in the sub-collapse interval. i0-1 , Corat i0 , Corat i0+1 They represent the collapsed pore proportions corresponding to the i0-1th, i0th, and i0+1th monitoring points, respectively. The first collapsed position refers to the monitoring points that do not include the first collapsed position;

[0020] The average vector of each volatility characteristic vector in the sub-collapse interval is the volatility mean vector, the dot product of the volatility mean vector and itself is the volatility equilibrium value, and the dot product of each volatility characteristic vector in the sub-collapse interval is calculated as the volatility state value FLval. If the volatility state value of the monitoring point between the first collapse position is greater than the volatility equilibrium value, then this moment is called the second collapse position; the proportion of the second collapse position in the sub-collapse interval to all monitoring points is the collapse risk rate; for any sub-collapse interval, if its collapse risk rate is greater than 50%, it is defined as a high-level collapse interval; the collapse positive deviations and collapse negative deviations corresponding to all second collapse positions constitute the sequence LsDev + and lsDev - ; The mediation strategy value Meval is calculated based on the number of high-level collapse intervals, positive collapse deviation, negative collapse deviation and fluctuation state value.

[0021] Since the above-mentioned mediation strategy value is calculated after the identification of the collapsed pore ratio, and since the sensitive parameters determined by the collapsed pore ratio locate the key moment when the biochar modification process has strong floating characteristics, the real-time state of the biochar modification process can be effectively quantified. However, in data scenarios that require rapid response to process changes, the method of collapsing positive deviation and negative deviation is difficult to distinguish between normal fluctuations and the true first collapse position, resulting in excessive quantization in data segments where the floating characteristics are not obvious enough, especially in the case of the calculation of the cumulative sum of each number of base points, the problem of misjudgment of the first collapse position is more obvious, which will cause the best time for mediation to be missed, resulting in excessive modification and causing the quality of biochar to decrease. In order to solve this problem of identifying data differences and improve the accuracy of calculating mediation strategy values, the present invention proposes a more preferred solution.

[0022] Preferably, the method for calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio is: set a time period as the monitoring period D_TMF, D_TMF∈[10,30] minutes, perform monitoring at equal intervals within the current monitoring period, that is, obtain the pore area ratio and the collapsed pore ratio once, and record the monitoring time scale as the monitoring point;

[0023] The monitoring binary values ​​obtained at any monitoring point are the pore area ratio Porat and the collapsed pore ratio Corat;

[0024] The product of the pore area ratio and the collapsed pore ratio of each monitoring point is called the collapse parameter Colpa. The original time series set is composed of a pair of monitoring points and their corresponding collapse parameters. The diff function is used to perform high-order difference operations on the original time series set of non-stationary time series to obtain the modified time series set. For any monitoring point, the absolute value of the difference between the corresponding elements in its original time series set and the modified time series set is defined as the divergence deviation Rerro.

[0025] The purpose and principle of obtaining the divergence deviation is to obtain the difference change rate between the original time series set and the modified time series set, so as to capture the time points in the time dimension with non-stationary change characteristics and quantify the relative degree of such non-stationary changes; specifically, a larger divergence deviation indicates that the original time series set has a stronger evolution trend near the monitoring point, while a smaller divergence deviation indicates that the original time series set is closer to a stable state.

[0026] Since the original time series set is a non-stationary time series, the diff function is used to correct it to a stationary time series, that is, the corrected time series set, where the diff function is called by the statsmodels library in Python.

[0027] All divergence deviations form a sequence DpLs in chronological order. The monitoring points corresponding to the maximum values ​​in the sequence DpLs are taken as the second collapse position. The period between any second collapse position and the first second collapse position in the reverse time direction is recorded as a micro-collapse segment. The number of monitoring points in the micro-collapse segment is recorded as TNum. The synergy correlation measure Comse between any two monitoring points in the micro-collapse segment is calculated:

[0028] Comse(j1,j2)=|(Colpa j1 -AvgPa j1 )(Colpa j2 -AvgPa j2 )| / (Vd_pa j1 ×Vd_pa j2 );

[0029] Where j1 and j2 are the serial numbers corresponding to any two monitoring points in the micro-collapse segment, Comse(j1,j2) represents the synergy correlation measure of the j1th and j2th monitoring points, AvgPa j1 and AvgPa j2 are the average values ​​of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points. The starting monitoring point of the monitoring period refers to the monitoring point farthest from the current moment. Vd_pa j1 and Vd_pa j2 is the variance of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points, Colpa j1 and Colpa j2 are the collapse parameters of the j1th and j2th monitoring points in the micro-collapse segment, respectively;

[0030] The principle of calculating the synergistic correlation measure is to distinguish normal fluctuations from the true first collapse position by increasing the similarity calculation of the pore characteristic data between monitoring points, reduce the misjudgment of the outbreak position of pore collapse, and avoid the loss of optimization opportunities for determining porosity.

[0031] The number of inflection points in the sequence formed by the synergistic correlation measure of any monitoring point in the micro-collapse segment and all the monitoring points in the reverse time direction is KS_len, and the synergistic correlation measure corresponding to each inflection point and the corresponding collapse parameter are respectively used to form the synergistic characteristic matrix COtri and the collapse characteristic matrix Patri; wherein COtri and Patri are matrices with one row and ks_len columns, and their mathematical expression COtri=[Comse ks1 ,Comse ks2 , …, Comse ks_len ], the corresponding collapse parameters of each inflection point are obtained to form the collapse characteristic matrix Patri, Patri = [Colpaks1 , Colpa ks2 , …, Colpa ks_len ]; where ks1, ks2 and ks_len are the numbers of the inflection points obtained, Comse ks_len and Colpa ks_len They represent the serial number of the ks_lenth inflection point, and the same applies to ks1 and ks2;

[0032] If the number of inflection points in the sequence formed by the synergistic correlation measure between any monitoring point in the micro-collapse segment and its monitoring point in the reverse time direction is not zero, the collapse parameter corresponding to the monitoring point is updated to COtri×Patri T , otherwise the collapse parameter of the monitoring point is not updated; the proportion of monitoring points whose collapse parameters are updated in the micro-collapse section is recorded as the update measure Remse, and the collapsed pore ratio is arranged in descending order. When the ranking value of the collapsed pore ratio of a monitoring point accounts for a smaller proportion of the total number of monitoring points than the update measure, or the collapse parameter of the monitoring point is updated, it is recorded as a key scale; the sequence of pore area ratios corresponding to all key scales is recorded as Poseq; the ranking value of any collapsed pore ratio refers to the ranking value of the collapsed pore ratio after descending arrangement;

[0033] For any monitoring point within the monitoring period D_TMF, the corresponding mediation strategy value Meval is calculated:

[0034] Meval=avg[Poseq] ∑ j3=1 Tolen ( Hmlen j3 · avg j3 { Rerro j4 × Colpa j4 } ) ;

[0035] Where j3 is the cumulative variable, j4 is the serial number of the monitoring point in the micro-collapse segment, Tolen is the number of micro-collapse segments in the monitoring period, and Hmlen is the number of micro-collapse segments in the monitoring period. j3 is the ratio of the number of key scales in the j3rd micro-collapse segment, Colpa j3,j4 ,Rerro j3,j4 are the collapse parameter and divergence deviation of the j4th monitoring point in the j3rd micro-collapse segment, avg is the symbol of the average function, avg j3 {Rerro j4 ×Colpa j4} means that the product of Rerro and Colpa of each monitoring point of the j3-th micro-collapse segment is first obtained, and then the average value of each product is obtained.

[0036] Beneficial effects: Since the mediation strategy value is actually calculated based on the pore area ratio and the collapsed pore ratio, on the one hand, the pore area ratio can indicate the effective porosity in the biochar, and the collapsed pore ratio indicates the degree of its collapse phenomenon. The mediation strategy value calculated by the dynamic characteristics of the two in time series actually extracts the possibility or risk of imbalance in the changing relationship between the two during the modification process, thereby identifying the most extreme conditions of the imbalance phenomenon in the preset working period, providing mathematical support and basis for further optimization of the mediation strategy.

[0037] Furthermore, the method for mediating the rotary kiln according to the mediation strategy value is: record the preset mediation time length as FIN.TM, divide the FIN.TM time period of the modification process into two equal lengths as a dividing point, the time period after the dividing point is taken as the second half of the modification, and the value range of the set consisting of all mediation strategy values ​​in the second half of the modification is recorded as the mediation value range; when the running time of the modification process exceeds the mediation time length, the mediation strategy value obtained in real time is recorded as the real-time strategy; when the real-time strategy exceeds the mediation value range, the rotary kiln is stopped and the rotary kiln mediation information is sent to the client.

[0038] Preferably, all undefined variables in the present invention, if not clearly defined, can be manually set thresholds.

[0039] The present invention also provides a biochar modification process mediation system based on microscopic images, the biochar modification process mediation system based on microscopic images comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor implementing the steps in the biochar modification process mediation method based on microscopic images when executing the computer program, the biochar modification process mediation system based on microscopic images can be run in computing devices such as desktop computers, laptop computers, PDAs, and cloud data centers, and the executable systems may include, but are not limited to, processors, memories, and server clusters, and the processor executing the computer program runs in the following system units:

[0040] An image acquisition unit, used for acquiring real-time data from the rotary kiln to obtain a monitoring image;

[0041] A data preprocessing unit, used for preprocessing the monitoring graph to form a monitoring graph sequence;

[0042] An image feature extraction unit, used for identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence;

[0043] a misaligned boundary identification unit for calculating a mediation strategy value based on a pore area ratio and a collapsed pore ratio;

[0044] The mediation control unit is used for mediating the rotary kiln according to the mediation strategy value.

[0045] The beneficial effects of the present invention are as follows: the present invention provides a biochar modification process mediation method based on microscopic images, collects data on the biochar surface in the modification process through SEM images, extracts the possibility or risk of imbalance between the effective porosity and the degree of collapse in the modification process according to the time series changes of the image, thereby identifying the most extreme conditions of the imbalance phenomenon in a preset working period, and then optimizing the mediation strategy and providing more on the original basis. In other words, the time point at which the risk of explosive growth of pore collapse occurs when the minimum estimated modification time is exceeded is generally identified, thereby providing more sufficient optimization opportunities for increasing the specific surface area and porosity of biochar in the biochar modification function, bringing gains to the performance of biochar outside the conservative range, especially in the large-scale production of biochar modification treatment process. The improvement effect is more obvious, ensuring the stability of the yield rate and functional utility of biochar. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The above and other features of the present invention will become more obvious by describing in detail the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar elements. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0047] Figure 1 Shown is a flow chart of a method for mediating biochar modification processes based on microscopic images;

[0048] Figure 2 Shown is a structural diagram of a biochar modification process mediation system based on microscopic images. DETAILED DESCRIPTION

[0049] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0050] like Figure 1 The figure shows a flow chart of a method for mediating the biochar modification process based on microscopic images. Figure 1 A method for mediating a biochar modification process based on microscopic images according to an embodiment of the present invention is described, the method comprising the following steps:

[0051] Example 1

[0052] The invention relates to a method for obtaining a monitoring map by collecting data from a rotary kiln in real time; preprocessing the monitoring map to form a monitoring map sequence; identifying the pore area ratio and the collapsed pore ratio from any monitoring map in the monitoring map sequence; calculating a mediation strategy value according to the pore area ratio and the collapsed pore ratio; and mediating the rotary kiln according to the mediation strategy value.

[0053] Furthermore, a method for acquiring a monitoring graph by real-time data collection from a rotary kiln is as follows: a modifier is a mixed solution of an acetic acid solution and a formic acid solution, and the modifier is sprayed onto the biochar; the biochar is placed into the rotary kiln after preheating, and the biochar is a mixed biochar of straw biochar and rice husk biochar; data of the biochar is collected from a sampling window area arranged on the side wall of the rotary kiln through an electron microscope to acquire a monitoring graph; and the scanning electron microscope acquisition interval is set to 10 seconds per time.

[0054] Specifically, a Hitachi SU3500 electron microscope was used; after starting the heating system of the kiln body, data collection began after the temperature reached 450°C at a heating rate of 10°C / min, and the recording time was recorded as Ts; data collection utilized a scanning electron microscope installed in the fixed kiln wall at the kiln head of the rotary kiln to capture the microscopic image of the biochar surface, i.e., the monitoring image, in real time during the heating process of the rotary kiln.

[0055] Furthermore, the method of preprocessing the monitoring image to form a monitoring image sequence is: the preprocessing includes denoising the monitoring image and histogram equalization adjustment, wherein the denoising is good wavelet coefficient denoising, and the histogram equalization adopts RMSHE recursive equalization, and the real-time obtained and preprocessed monitoring image is written into a sequence recorded as a monitoring image sequence.

[0056] Further, a method for identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence is: identifying the pore area and the collapsed area from the processed image respectively by a pre-trained CNN convolutional neural network model, wherein the collapsed area is a subset of the pore area; defining the ratio of the area of ​​the pore to the area of ​​the entire image as the pore area ratio;

[0057] The morphological opening method is used to segment the image from the pore area and obtain several pore units. If the pore unit contains a collapsed area, it is marked as collapsed. The ratio of the number of pore units marked as collapsed to the total number of pore units is recorded as the collapsed pore ratio.

[0058] The pre-trained CNN convolutional neural network model is a YOLOv5 recognition model; the training data is input into the convolutional neural network of the CNN architecture, and features are extracted through multi-layer convolution and pooling operations, and the obtained features are used to predict the pore area through the fully connected layer or segmentation layer. Through back propagation and optimization algorithms, the Adam method is preferred, and the network weights are iteratively adjusted to gradually improve the model's ability to identify pore areas on the training set. Finally, the accuracy and generalization ability of the model are verified with the test set, and the pre-trained CNN convolutional neural network model is obtained.

[0059] Furthermore, the method for calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio is: the average value and interquartile range of all collapsed pore ratios during the monitoring period are recorded as the collapse mean Av_Cor and the sensitive parameter Se_Par respectively; for any monitoring point during the monitoring period, its collapse positive deviation is recorded as: Co_Dev + Ti =max{0,Co_Dev + Ti-1 +(Corat Ti -Av_Cor)}, the negative collapse deviation is: Co_Dev - Ti =max{0,Co_Dev - Ti-1 +(Av_Cor-Corat Ti )};

[0060] For any monitoring point, if its positive collapse deviation or negative collapse deviation is greater than half of the sensitive parameter, the monitoring point is recorded as the first collapse position, and the period between any first collapse position and the first first collapse position in the reverse time direction is defined as a sub-collapse interval; j is used as the corresponding serial number of the monitoring point between the first collapse position in the sub-collapse interval, and the volatility characteristic vector of the i0th monitoring point in the sub-collapse interval is: Vo_tor i0 =(|(Porat i0 -Porat i0-1 )(Corat i0 -Corat i0-1 )|,|(Porat i0+1 -Porat i0 )(Corat i0+1 -Corat i0 )|);

[0061] The average vector of each volatility characteristic vector in the sub-collapse interval is the volatility mean vector, the dot product of the volatility mean vector and itself is the volatility equilibrium value, and the dot product of each volatility characteristic vector in the sub-collapse interval is calculated as the volatility state value FLval. If the volatility state value of the monitoring point between the first collapse position is greater than the volatility equilibrium value, then this moment is called the second collapse position; the proportion of the second collapse position in the sub-collapse interval to all monitoring points is the collapse risk rate; for any sub-collapse interval, if its collapse risk rate is greater than 50%, it is defined as a high-level collapse interval; the collapse positive deviations and collapse negative deviations corresponding to all second collapse positions constitute the sequence LsDev + and lsDev - ; The mediation strategy value Meval is calculated based on the number of high-level collapse intervals, collapse positive deviation, collapse negative deviation and fluctuation state value:

[0062] ;

[0063] Where PNum is the number of high-level collapse intervals during the monitoring period, RNum is the number of all second-level collapse positions of the current monitoring point in the reverse time direction during the monitoring period, and Porat i1 , FLval i1 is the pore area ratio and fluctuation state value corresponding to the i1th second collapse position. The sequence of collapse positive deviation and collapse negative deviation corresponding to all second collapse positions is LsDev + ,lsDev - , max is the maximum value function, ln() is the logarithmic function with the natural constant e as the base.

[0064] Furthermore, the method for mediating the rotary kiln according to the mediation strategy value is: the preset mediation time length is recorded as FIN.TM, the value is 2 hours, and the FIN.TM time period of the modification process is divided into two equal lengths as a dividing point, the time period after the dividing point is taken as the second half of the modification, and the value range of the set composed of all mediation strategy values ​​in the second half of the modification is recorded as the mediation value range; when the running time of the modification process exceeds the mediation time length, the mediation strategy value obtained in real time is recorded as the real-time strategy; when the real-time strategy exceeds the mediation value range, the rotary kiln is stopped and the rotary kiln mediation information is sent to the client.

[0065] Example 2

[0066] This embodiment 2 replaces the method of calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio in embodiment 1, specifically: set a time period as the monitoring period D_TMF, take a value of 30 minutes, and monitor at equal intervals during the current monitoring period;

[0067] The product of the pore area ratio and the collapsed pore ratio of each monitoring point is called the collapse parameter Colpa. The original time series set is composed of a pair of monitoring points and their corresponding collapse parameters. The diff function is used to perform high-order difference operations on the original time series set of non-stationary time series to obtain the modified time series set. For any monitoring point, the absolute value of the difference between the corresponding elements in its original time series set and the modified time series set is defined as the divergence deviation Rerro.

[0068] All divergence deviations form a sequence DpLs in chronological order. The monitoring points corresponding to the maximum values ​​in the sequence DpLs are taken as the second collapse position. The period between any second collapse position and the first second collapse position in the reverse time direction is recorded as a micro-collapse segment. The number of monitoring points in the micro-collapse segment is recorded as TNum. The synergy correlation measure Comse between any two monitoring points in the micro-collapse segment is calculated:

[0069] Comse(j1,j2)=|(Colpa j1 -AvgPa j1 )(Colpa j2 -AvgPa j2 )| / (Vd_pa j1 ×Vd_pa j2 );

[0070] Where j1 and j2 are the serial numbers corresponding to any two monitoring points in the micro-collapse segment, Comse(j1,j2) represents the synergy correlation measure of the j1th and j2th monitoring points, AvgPa j1 and AvgPa j2 are the average values ​​of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points. The starting monitoring point of the monitoring period refers to the monitoring point farthest from the current moment. Vd_pa j1 and Vd_pa j2 is the variance of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points, Colpa j1 and Colpa j2 are the collapse parameters of the j1th and j2th monitoring points in the micro-collapse segment, respectively;

[0071] The number of inflection points in the sequence formed by the synergistic correlation measure between any monitoring point in the micro-collapse segment and all monitoring points in the reverse time direction is KS_len, and the synergistic correlation measure corresponding to each inflection point and the corresponding collapse parameter are respectively used to form the synergistic characteristic matrix COtri and the collapse characteristic matrix Patri;

[0072] If the number of inflection points in the sequence formed by the synergistic correlation measure between any monitoring point in the micro-collapse segment and its monitoring point in the reverse time direction is not zero, the collapse parameter corresponding to the monitoring point is updated to COtri×Patri T ; The proportion of monitoring points whose collapse parameters are updated in the micro-collapse section is recorded as the update measure Remse, and the collapsed pore proportions are arranged in descending order. When the ranking value of the collapsed pore proportion of a monitoring point accounts for a proportion of the total number of monitoring points that is less than the update measure, or the collapse parameter of the monitoring point is updated, it is recorded as a key scale; the sequence of pore area proportions corresponding to all key scales is recorded as Poseq; calculate the mediation strategy value Meval:

[0073] Meval=avg[Poseq] ∑ j3=1 Tolen ( Hmlen j3 · avg j3 { Rerro j4 × Colpa j4 } ) ;

[0074] Where j3 is the cumulative variable, j4 is the serial number of the monitoring point in the micro-collapse segment, Tolen is the number of micro-collapse segments in the monitoring period, and Hmlen is the number of micro-collapse segments in the monitoring period. j3 is the ratio of the number of key scales in the j3rd micro-collapse segment, Colpa j3,j4 ,Rerro j3,j4 are the collapse parameter and divergence deviation of the j4th monitoring point in the j3rd micro-collapse segment, avg is the symbol of the average function, avg j3 {Rerro j4 ×Colpa j4} means that the product of Rerro and Colpa of each monitoring point of the j3-th micro-collapse segment is first obtained, and then the average value of each product is obtained.

[0075] Comparative Example 1

[0076] The mediation time of Comparative Example 1 is set to be consistent with the mediation time preset in Example 1.

[0077] Comparative Example 2

[0078] The mediation time of Example 2 is set to 4 hours.

[0079] Biochar samples of the biochar modified for 1.5 hours in Example 1, Example 2, Comparative Example 1 and Comparative Example 2 were collected as execution process samples, and biochar samples at the end of modification were collected as execution end samples. Microscopic images were collected for each biochar sample, and 5 images were obtained as the corresponding sample image set. The collapsed pore area was identified from each image of the sample image set, and the average collapsed pore area corresponding to the biochar samples was calculated; then the specific surface area was calculated for each sample by combining the nitrogen adsorption-desorption method with the BET model to measure the specific surface area.

[0080] Define the first coverage collapse index MRCI, and the calculation method is MRCI=CPR-1.5×PRIG, and the second coverage collapse index MRCII, and the calculation method is MRCI=1.5×CPR-PRIG; where PRIG is the specific surface area growth ratio, and the calculation method is PRIG=(PCV_Fin-PCV_Pcc) / PCV_Pcc, PCV_Fin and PCV_Pcc represent the specific surface areas obtained from the sample at the end of execution and the sample during execution, respectively; CPR represents the collapse porosity, CPR=(Ar_Fin-Ar_Pcc) / Ar_Pcc, Ar_Fin and Ar_Pcc represent the collapse pore areas obtained from the sample at the end of execution and the sample during execution, respectively.

[0081] Table 1 Comparison of void volume-collapse gain efficiency

[0082]

[0083] Table 1 shows a comparison table of void volume-collapse gain efficiency. The partial pore volume gain efficiency in the comparison table is the ratio of the first covering collapse index to the time difference between the corresponding acquisition time of the execution end sample and the execution process sample, representing the average gain effect of the pore volume in the process from the acquisition of the execution process sample to the end of the execution. Therefore, the smaller the partial pore volume gain efficiency value, the more it implies that the pore volume gain effect is greater than the degree of induced pore collapse risk per unit time increase in this period; the partial collapse gain efficiency is the ratio of the second covering collapse index to the time difference between the corresponding acquisition time of the execution end sample and the execution process sample, representing the average gain effect of the pore collapse risk in the process from the acquisition of the execution process sample to the end of the execution. Therefore, the smaller the partial collapse gain efficiency value, the more it implies that the degree of induced pore collapse risk is greater than the pore volume gain effect per unit time increase in this period.

[0084] This comparison table shows that the partial pore volume gain efficiency of Example 1 is significantly higher than that of other groups. This is because Example 1 belongs to a conservative modification time node in the actual modification process of the rotary kiln modification. The time node itself is set to adapt to more uncertain situations. Therefore, it often belongs to a higher threshold state to adapt to the outbreak risk of pore collapse, that is, its setting sensitivity to pore collapse is significantly lower than that of Example 1, Example 2 and Comparative Example 2. Therefore, the pore volume gain effect of Example 1 during the execution process and the end of the execution period is greater than the risk of inducing pore collapse, which is significantly higher than that of the other three groups, and the partial collapse gain efficiency of Example 1 is the smallest among all groups, indicating that the risk of inducing pore collapse is significantly lower than the pore volume gain effect, which also indicates that the sample obtained at the end of the execution of Example 1 has the greatest development potential from the perspective of pore volume development. Since Examples 1 and 2 adopt the dynamic mediation method of the present method, the porosity gain effect is smaller than the risk of inducing pore collapse for each additional unit time during the execution process to the end of the execution, indicating that the porosity increase effect and the inducing collapse risk are better balanced, making the porosity gain effect stronger than both Comparative Examples 1 and 2. The mediation time used in Comparative Example 2 is the setting with the highest risk of pore collapse, indicating that the sensitivity to pore collapse phenomenon is the strongest under this experimental condition, so the partial collapse gain efficiency of Comparative Example 2 has the maximum value in each group. Based on the comparison process, it can be seen that by adopting the dynamic mediation method of the present method, the active carbon porosity gain of the rotary kiln in the mediation process can be effectively stimulated, while the risk of pore collapse can be controlled. Therefore, the completion of the biochar modification process can be improved, thereby greatly improving the quality and effect of the biochar.

[0085] The embodiment of the present invention provides a biochar modification process mediation system based on microscopic images, such as Figure 2 Shown is a structural diagram of a biochar modification process mediation system based on microscopic images of the present invention. The biochar modification process mediation system based on microscopic images of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned embodiment of a biochar modification process mediation method based on microscopic images are implemented.

[0086] The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to run in the following units of the system:

[0087] An image acquisition unit, used for acquiring real-time data from the rotary kiln to obtain a monitoring image;

[0088] A data preprocessing unit, used for preprocessing the monitoring graph to form a monitoring graph sequence;

[0089] An image feature extraction unit, used for identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence;

[0090] a misaligned boundary identification unit for calculating a mediation strategy value based on a pore area ratio and a collapsed pore ratio;

[0091] The mediation control unit is used for mediating the rotary kiln according to the mediation strategy value.

[0092] The biochar modification process mediation system based on microscopic images can be run on computing devices such as desktop computers, laptops, PDAs, and cloud servers. The biochar modification process mediation system based on microscopic images can be operated on systems including, but not limited to, processors and memories. Those skilled in the art will appreciate that the example is merely an example of a biochar modification process mediation system based on microscopic images, and does not constitute a limitation on a biochar modification process mediation system based on microscopic images. It may include more or fewer components than the example, or a combination of certain components, or different components. For example, the biochar modification process mediation system based on microscopic images may also include input and output devices, network access devices, buses, and the like.

[0093] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the operation system of the biochar modification process mediation system based on microscopic images, and various interfaces and lines are used to connect the various parts of the entire biochar modification process mediation system based on microscopic images.

[0094] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the biochar modification process mediation system based on microscopic images by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0095] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.

Claims

1. A method for mediating a biochar modification process based on microscopic images, characterized in that: The method comprises the following steps: acquiring monitoring images from a rotary kiln in real time; preprocessing the monitoring images to form a monitoring image sequence; identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence; calculating a mediation strategy value according to the pore area ratio and the collapsed pore ratio; and mediating the rotary kiln according to the mediation strategy value; The method for calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio is as follows: according to the horizontal comparison of each collapsed pore ratio during the monitoring period, the positive collapse deviation and the negative collapse deviation are obtained, and each first collapse position is identified by using the positive collapse deviation and the negative collapse deviation, and the volatility characteristic vector is calculated by the instantaneous difference between the pore area ratio and the collapsed pore ratio of the first collapse position, and the volatility characteristic vector is used to calculate the volatility state value, and the second collapse position is obtained by horizontal comparison of the volatility state value, and the mediation strategy value is calculated according to the number of high-position collapse intervals, the positive collapse deviation, the negative collapse deviation and the volatility state value; The method for collecting data from the rotary kiln in real time to obtain a monitoring diagram is as follows: the modifier is a weak acid solution, and the modifier is injected or sprayed onto the biochar; the biochar is placed in the rotary kiln after preheating, and the data of the biochar is collected from the sampling window area arranged on the side wall of the rotary kiln through an electron microscope to obtain a monitoring diagram; the scanning electron microscope collection interval is set to 5 to 30 seconds per time; The method for calculating the mediation strategy value according to the pore area ratio and the collapsed pore ratio is as follows: the corresponding monitoring binary values ​​obtained at the monitoring point are recorded as the pore area ratio Porat and the collapsed pore ratio Corat respectively; the average value and interquartile range of all collapsed pore ratios in the monitoring period are recorded as the collapse mean Av_Cor and the sensitive parameter Se_Par respectively; for any monitoring point in the monitoring period, the collapse positive deviation and collapse negative deviation are calculated according to the collapse mean; For any monitoring point, if its positive collapse deviation or negative collapse deviation is greater than half of the sensitive parameter, the monitoring point is recorded as the first collapse position, and the period between any first collapse position and the first first collapse position in the reverse time direction is defined as a sub-collapse interval; i0 is used as the corresponding serial number of the monitoring point between the first collapse position in the sub-collapse interval, and the volatility characteristic vector of the i0th monitoring point in the sub-collapse interval is: Vo_tor i0 =(|(Porat i0 -Porat i0-1 )(Corat i0 -Corat i0-1 )|,|(Porat i0+1 -Porat i0 )(Corat i0+1 -Corat i0 )|); Porat i0-1 , Porat i0 , Porat i0+1 Corat represents the pore area ratio of the i0-1, i0, and i0+1 monitoring points between the first collapse position in the sub-collapse interval. i0-1 , Corat i0 , Corat i0+1 They represent the collapsed pore proportions corresponding to the i0-1th, i0th, and i0+1th monitoring points, respectively. The first collapsed position refers to the monitoring points that do not include the first collapsed position; The average vector of each volatility characteristic vector in the sub-collapse interval is the volatility mean vector, the dot product of the volatility mean vector and itself is the volatility equilibrium value, and the dot product of each volatility characteristic vector in the sub-collapse interval is calculated as the volatility state value FLval. If the volatility state value of the monitoring point between the first collapse position is greater than the volatility equilibrium value, then the monitoring point is called the second collapse position; the proportion of the second collapse position in the sub-collapse interval to all monitoring points is the collapse risk rate; for any sub-collapse interval, if its collapse risk rate is greater than 50%, it is defined as a high-level collapse interval; the collapse positive deviations and collapse negative deviations corresponding to all second collapse positions constitute the sequence LsDev + and lsDev - ; The mediation strategy value is calculated based on the number of high-level collapse intervals, collapse positive deviation, collapse negative deviation and fluctuation state value; The method for mediating the rotary kiln according to the mediation strategy value is as follows: the preset mediation time length is recorded as FIN.TM, and the FIN.TM time period of the modification process is divided into two equal lengths as a dividing point, the time period after the dividing point is taken as the second half of the modification, and the value range of the set composed of all mediation strategy values ​​in the second half of the modification is recorded as the mediation value range; when the running time of the modification process exceeds the mediation time length, the mediation strategy value obtained in real time is recorded as the real-time strategy; when the real-time strategy exceeds the mediation value range, the rotary kiln is stopped, and the rotary kiln mediation information is sent to the client.

2. The method for mediating biochar modification process based on microscopic images according to claim 1, characterized in that: The method for preprocessing the monitoring image to form a monitoring image sequence is as follows: the preprocessing includes denoising the monitoring image and adjusting the histogram equalization, wherein the denoising is at least one of wide wavelet coefficient denoising or Gaussian filtering denoising, and the histogram equalization adopts RMSHE recursive equalization, and the monitoring image obtained in real time and preprocessed is written into a sequence recorded as a monitoring image sequence.

3. The method for mediating biochar modification process based on microscopic images according to claim 1, characterized in that: The method for identifying the pore area ratio and the collapsed pore ratio from any monitoring image in the monitoring image sequence is as follows: using a pre-trained CNN convolutional neural network model to respectively identify the pore area and the collapsed area from the processed image, wherein the collapsed area is a subset of the pore area; defining the ratio of the area of ​​the pore to the area of ​​the entire image as the pore area ratio; The morphological opening method is used to segment the image from the pore area and obtain several pore units. If the pore unit contains a collapsed area, it is marked as collapsed. The ratio of the number of pore units marked as collapsed to the total number of pore units is recorded as the collapsed pore ratio.

4. The method for mediating biochar modification process based on microscopic images according to claim 1, characterized in that: The method for calculating the mediation strategy value based on the pore area ratio and the collapsed pore ratio can be replaced by: set a time period as the monitoring period D_TMF, record the product of the pore area ratio and the collapsed pore ratio at any monitoring point as the collapse parameter Colpa, the two-tuple consisting of the monitoring point and its corresponding collapse parameter constitutes the original time series set, and use the diff function to perform high-order difference operations on the original time series set of non-stationary time series to obtain the modified time series set; for any monitoring point, the absolute value of the difference between the corresponding elements in its original time series set and the modified time series set is defined as the divergence deviation Rerro; All divergence deviations form a sequence DpLs in chronological order, and the monitoring points corresponding to the maximum values ​​in the sequence DpLs are taken as the second collapse position. The period between any second collapse position and the first second collapse position in the reverse time direction is recorded as a micro-collapse segment, and the number of monitoring points in the micro-collapse segment is recorded as TNum; Calculate the synergy correlation measure Comse between any two monitoring points in the micro-collapse segment: Comse(j1,j2)=|(Colpa j1 -AvgPa j1 )(Colpa j2 -AvgPa j2 )| / (Vd_pa j1 ×Vd_pa j2 ), where j1 and j2 are the serial numbers corresponding to any two monitoring points in the micro-collapse segment, Comse(j1,j2) represents the synergistic correlation measure of the j1th and j2th monitoring points, AvgPa j1 and AvgPa j2 are the average values ​​of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points. The starting monitoring point of the monitoring period refers to the monitoring point farthest from the current moment. Vd_pa j1 and Vd_pa j2 is the variance of the collapse parameters of all monitoring points between the starting monitoring point of the monitoring period and the j1th and j2th monitoring points, Colpa j1 and Colpa j2 are the collapse parameters of the j1th and j2th monitoring points in the micro-collapse segment, respectively; The number of inflection points in the sequence formed by the synergistic correlation measure between any monitoring point in the micro-collapse segment and all monitoring points in the reverse time direction is KS_len, and the synergistic correlation measure corresponding to each inflection point and the corresponding collapse parameter are respectively used to form the synergistic characteristic matrix COtri and the collapse characteristic matrix Patri; If the number of inflection points in the sequence formed by the synergistic correlation measure between any monitoring point in the micro-collapse segment and its monitoring point in the reverse time direction is not zero, the collapse parameter corresponding to the monitoring point is updated to COtri×Patri T , the proportion of monitoring points whose collapse parameters are updated in the micro-collapse section is recorded as the update measure Remse, and the collapsed pore proportions are arranged in descending order. When the proportion of the ranking value of the collapsed pore proportion of a monitoring point to the total number of monitoring points is less than the update measure, or the collapse parameter of the monitoring point is updated, it is recorded as a key scale; the sequence composed of the pore area proportions corresponding to all key scales is recorded as Poseq, and the current mediation strategy value is calculated according to the sequence Poseq, collapse parameter and divergence deviation.

5. A biochar modification process mediation system based on microscopic images, characterized in that: The biochar modification process mediation system based on microscopic images comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the biochar modification process mediation method based on microscopic images described in any one of claims 1 to 4 are implemented. The biochar modification process mediation system based on microscopic images runs on desktop computers, laptop computers, PDAs, and computing devices in cloud data centers.

Images