Carbon fiber composite material in-situ graded cracking method based on microwave bipolar heating

Abstract

The invention belongs to the field of material cracking, and particularly relates to a microwave bipolar heating-based in-situ graded cracking method for a carbon fiber composite material, which comprises the following steps of: initializing a feeding speed, and combining a preset three-dimensional model of a to-be-cracked material with a graded inversion function space; a multi-stage cracking control sequence containing parameters such as the feeding rate, the preheating temperature and the dual-frequency microwave radiation power is obtained through inversion, wherein a hierarchical inversion function space comprises a hierarchical hierarchical cracking feedback control node chain constructed through an associated fuzzy control algorithm. And generating a control instruction based on the control sequence and the hierarchical linkage control model, monitoring the splitting efficiency and the energy consumption conversion efficiency in real time, and feeding back to the node chain to adjust the control sequence. The node chain construction relates to flow splitting, parameter node set establishment, correlation analysis, feedback factor construction and the like, and finally, through simulation training optimization, splitting efficiency maximization and energy consumption and pollution minimization are realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of remote sensing monitoring, and particularly relates to a carbon fiber composite material in-situ hierarchical pyrolysis method and system based on microwave bipolar heating. BACKGROUND

[0002] With the wide application of advanced composite materials, especially carbon fiber reinforced composite materials in the fields of aerospace, automobiles, wind power and the like, the amount of waste thereof increases sharply. Traditional recycling technologies face severe challenges. For example, the water vapor method, the natural gas heating pyrolysis method or the chemical solvent pyrolysis method generally have problems of high energy consumption and low efficiency, need to maintain high temperature for a long time, have huge energy consumption and long processing period; the chemical method needs to use strong acid and strong base, generates secondary pollution, and has high post-processing cost; the high cost and performance loss lead to a large amount of waste composite materials being directly incinerated or landfilled, causing waste of valuable carbon fiber resources and causing serious environmental pollution, and the incineration process releases toxic gases such as dioxin and CO; the existing technical bottlenecks are mainly as follows: the traditional heating mode has slow heat conduction, which easily leads to uneven heating of the material, local overheating or incomplete pyrolysis; and the existing microwave pyrolysis equipment mostly uses a single frequency, has insufficient penetration depth, is only suitable for thin layer materials, and cannot effectively process composite materials with different thicknesses or densities; therefore, the application provides a carbon fiber composite material in-situ hierarchical pyrolysis method based on microwave bipolar heating. SUMMARY

[0003] In view of the deficiencies of the prior art, the application provides a carbon fiber composite material in-situ hierarchical pyrolysis method based on microwave bipolar heating. The method initializes a feeding speed, combines a preset three-dimensional model of a material to be pyrolyzed and a hierarchical inversion function space, and inversely obtains a multi-stage pyrolysis control sequence including parameters such as a feeding rate, a preheating temperature and a dual-frequency microwave radiation power, wherein the hierarchical inversion function space includes a hierarchical pyrolysis feedback control node chain constructed by an associated fuzzy control algorithm. Control instructions are generated based on the control sequence and a hierarchical linkage control model, and the pyrolysis efficiency and the energy consumption conversion efficiency are monitored in real time and fed back to the node chain to adjust the control sequence. The node chain construction involves process pyrolysis, parameter node set establishment, correlation degree analysis and feedback factor construction, and finally realizes maximization of pyrolysis efficiency and minimization of energy consumption and pollution through simulation training optimization, thereby improving the precision of pyrolysis control.

[0004] To achieve the above object, the application provides the following technical scheme:

[0005] The carbon fiber composite material in-situ hierarchical pyrolysis method based on microwave bipolar heating comprises the following steps:

[0006] Initialize the feeding speed, and based on a preset three-dimensional model of the material to be cracked and a preset hierarchical inversion function space, a multi-stage cracking control sequence is obtained by inversion; the multi-stage cracking control sequence includes a feeding rate, optimal temperatures and residence time lengths corresponding to first-stage preheating and second-stage preheating, an initial splitting preheating temperature, a double-frequency microwave radiation power, a hierarchical cracking frequency interval and temperature interval, a cracking efficiency of the material to be cracked, an energy consumption conversion efficiency, and a waste heat recovery utilization rate; the hierarchical inversion function space includes a hierarchical hierarchical cracking feedback control node chain; the hierarchical hierarchical cracking feedback control node chain is obtained by combining a correlation fuzzy control algorithm including a self-correlation function according to the multi-stage cracking control sequence;

[0007] Based on the multi-stage cracking control sequence and a preset hierarchical linkage control model, a multi-stage separation control instruction is generated, and the real-time cracking efficiency and energy consumption conversion efficiency of the material to be cracked are monitored and evaluated and fed back to each control node in the hierarchical hierarchical cracking feedback control node chain to adjust the multi-stage cracking control sequence in real time.

[0008] Specifically, the construction steps of the hierarchical hierarchical cracking feedback control node chain include:

[0009] Obtain the cracking process information corresponding to the carbon fiber composite material, and obtain the single cracking process sequence, the cracking process parameters corresponding to each cracking process, the control parameter set, and the overall cracking process sequence information through a preset process cracking model;

[0010] Based on the cracking process parameters and control parameters corresponding to each cracking process and the process cracking model, secondary cracking is performed to obtain a single cracking control substep sequence under each single cracking process and a cracking process parameter subsequence corresponding to the single cracking control substep;

[0011] Based on all single cracking processes and monitoring control and evaluation feedback processes, a cracking control node set in the hierarchical hierarchical cracking feedback control node chain and monitoring control nodes and feedback regulation nodes are constructed, and based on the overall cracking process sequence information and the monitoring and evaluation feedback processes, a cracking process sequence connection and a feedback connection in the hierarchical hierarchical cracking feedback control node chain are constructed.

[0012] Specifically, the construction steps of the hierarchical hierarchical cracking feedback control node chain further include:

[0013] Based on all the cracking control node set and monitoring nodes and feedback regulation nodes, a transverse cracking feedback control node chain is obtained by a blockchain algorithm based on the cracking process sequence connection and the feedback connection;

[0014] Based on the single cracking control substep sequence under each single cracking process, a longitudinal parameter node set corresponding to each single cracking process is constructed;

[0015] Based on the control sequence and time of the single cracking control sub-step sequence under the single cracking process, a longitudinal parameter connection is constructed.

[0016] Specifically, the construction step of the hierarchical and hierarchical cracking feedback control node chain further includes:

[0017] Based on the corresponding longitudinal parameter node set of all single cracking processes and the corresponding longitudinal parameter connection, the longitudinal parameter information chain corresponding to each single cracking process is constructed through a tree database, and the cracking process and control parameters corresponding to each sub-step are saved in the sub-node of the corresponding longitudinal parameter information chain;

[0018] Based on the cracking process control parameters saved by all longitudinal parameter information chains and the association algorithm, the association degree of the corresponding parameter nodes between each cracking control node is obtained;

[0019] Based on the association degree of the corresponding parameter nodes between each cracking control node and the radial kernel function, a bidirectional mapping function and an associated connection between the parameter nodes are constructed.

[0020] Specifically, the construction step of the hierarchical and hierarchical cracking feedback control node chain further includes:

[0021] Based on the bidirectional mapping function and the corresponding associated connection, the control parameter node network is constructed by a graph algorithm combining all parameter nodes;

[0022] Based on the parameter sequence corresponding to each cracking control node, the contribution degree of each parameter node to the cracking efficiency and energy conversion efficiency under each cracking control node is obtained by a factor analysis algorithm combining the set cracking efficiency and energy conversion efficiency result evaluation variables;

[0023] Based on the contribution degree of each parameter node to the cracking efficiency and energy conversion efficiency, the main control variable and auxiliary control variable sequence corresponding to each cracking control node are obtained.

[0024] Specifically, the construction step of the hierarchical and hierarchical cracking feedback control node chain further includes:

[0025] Based on the contribution degree of each auxiliary control variable in the two cracking control nodes to the corresponding main control variable and the association degree of the parameter nodes between the two cracking control nodes, an inverse association mapping weight between the two cracking control nodes is constructed.

[0026] Based on the inverse association mapping weight between the two cracking control nodes and the main control variable of the corresponding two cracking control nodes, a main control inverse function is constructed through a convolution kernel function, and the main control inverse function is embedded into the cracking process sequence connection between the corresponding cracking control nodes;

[0027] Based on the contribution degree product of each main control variable to the cracking efficiency and energy conversion efficiency, a feedback regulation factor is constructed and built into the feedback connection between each cracking control node and feedback regulation node, to obtain a hierarchical and graded cracking feedback control node chain.

[0028] Specifically, the construction step of the hierarchical and graded cracking feedback control node chain further includes:

[0029] Based on the hierarchical and graded cracking feedback control node chain and the depth control parameter search model configured by the control parameter node network, a simulation algorithm, a preset optimal feedback control function, and cracking efficiency threshold and energy conversion efficiency threshold are combined for training, to obtain a trained hierarchical and graded cracking feedback control node chain.

[0030] The optimal feedback control function is constructed according to the feedback regulation factor in the feedback column, combined with the main control variable, the splitting efficiency and the energy conversion efficiency, combined with the linear function, and the maximum value of the splitting efficiency and the energy conversion efficiency.

[0031] Specifically, the cracking control node set includes an analytical modeling node, a feeding node, a first preheating node, a second preheating node, a frequency splitting cracking node, an ionization combustion control node, and a preheating recovery regulation node.

[0032] The working process of the hierarchical and graded cracking feedback control node chain includes:

[0033] Through the analytical modeling node, a three-dimensional density model of the target object to be cracked is constructed according to the built-in three-dimensional modeling algorithm, combined with the geometric parameter information of the target object to be cracked, the density of each material in the composite material, and the optimal hierarchical cracking parameter; the optimal hierarchical cracking parameter is the double-frequency frequency value corresponding to the density of each density material in the composite material, which makes the cracking efficiency reach the maximum at the optimal second preheating target temperature.

[0034] Based on the three-dimensional density model of the target to be cracked, the first preheating temperature interval and the second temperature interval are combined, and through the pre-trained deep analysis model, the feeding speed between the feeding node and the first preheating node, the density and the composite density of each material, and the first optimal preheating temperature parameter, the first residence time length, the first preheating target temperature, the first inverse correlation mapping between the first preheating target temperature between the first preheating node and the second preheating node, the density and the composite density of each material, the feeding rate, the second residence time length, the second preheating temperature parameter, the preset second preheating target temperature, and the second inverse correlation mapping between the second preheating target temperature between the second preheating node and the frequency splitting cracking node, the corresponding third inverse correlation mapping between the second preheating target temperature, the optimal cracking parameter and the third residence time length of the frequency splitting cracking node are obtained. The feeding rate corresponding to the feeding node, the residence time length and the heating rate corresponding to the first preheating node and the second preheating node, and the cracking double-frequency frequency value and the third residence time length corresponding to the frequency splitting cracking node are obtained.

[0035] Specifically, the working process of the hierarchical and hierarchical cracking feedback control node chain includes:

[0036] According to the obtained feeding rate, the first fuzzy control sub-model built in the feeding node is combined to obtain the feeding control instruction for feeding, and based on the residence time length and the heating rate corresponding to the first preheating node and the second preheating node, the second fuzzy control sub-model is used for first preheating and second preheating, and the waste heat conversion efficiency is monitored in real time based on the preset preheating conversion curve;

[0037] When the corresponding second preheating target temperature is reached, the double-frequency cracking control is performed through the third fuzzy sub-model combined with the cracking double-frequency frequency value and the third residence time length, and the cracking efficiency is monitored in real time;

[0038] Based on the deviation value of the real-time monitored waste heat conversion efficiency and cracking efficiency and the corresponding waste heat conversion efficiency threshold and cracking efficiency threshold, the fourth inverse correlation mapping constructed by the energy conversion efficiency between the frequency splitting cracking node and the ionization combustion control node, the natural gas input quantity and the ionization control coefficient is combined with the fourth fuzzy sub-model to generate the ionization control instruction to ionize the generated waste gas.

[0039] Specifically, the working process of the hierarchical and hierarchical cracking feedback control node chain includes:

[0040] According to the energy conversion efficiency between the ionization combustion control node and the preheating recovery adjustment node, the feeding rate, the preheating heating rate and the corresponding first or second residence time length of the preheating stage, the fifth inverse correlation mapping and the sixth inverse correlation mapping are constructed respectively, and the fifth fuzzy sub-model is combined to obtain the compensation control instruction for the first or second preheating node. The preheating heating rate of the first or second preheating node is compensated according to the density and the composite density of each material of the target to be cracked;

[0041] Meanwhile, the feeding node, the first preheating node, the second preheating node, the frequency splitting cracking node, the ionization combustion control node and the preheating recovery adjustment node are monitored based on the monitoring control node, and the monitoring parameters are input to the feedback control node to evaluate the cracking efficiency and the energy conversion efficiency, so as to obtain the cracking efficiency evaluation value and the energy conversion efficiency evaluation value;

[0042] Based on the cracking efficiency evaluation value and the energy conversion efficiency evaluation value and the cracking efficiency threshold value and the energy conversion efficiency threshold value, the cracking efficiency deviation value and the energy conversion efficiency deviation value are obtained.

[0043] Based on the cracking efficiency deviation value and the energy conversion efficiency deviation value, the feedback is fed back to the hierarchical cracking feedback control node chain through the feedback connection, the optimal deviation control parameter is searched synchronously through the main control inversion function combined with the control parameter node network and the configured deep control parameter search model, and the optimal deviation control parameter obtained is fed back to the corresponding cracking control node, and the optimal main control variable corresponding to the adjacent two cracking control nodes is taken as a constraint, and the fuzzy control model of the corresponding cracking control node is combined to generate a cracking deviation adjustment instruction, so that the cracking efficiency and the energy conversion efficiency meet the corresponding condition threshold value in real time.

[0044] Compared with the prior art, the present application has the following advantages:

[0045] The present application is aimed at the deficiencies of the prior art, and realizes intelligent collaborative control of the whole cracking process through the construction of a hierarchical cracking feedback control node chain and a parameter node network, and the deep integration of blockchain algorithms, tree databases and graph algorithms. The core value lies in: based on the two-dimensional structure of the horizontal node chain and the vertical parameter information chain, the inversion correlation mapping weight constructed by combining factor analysis contribution and convolution kernel function is used to realize the cross-level precise collaboration of key process parameters such as feeding rate, preheating residence time and double-frequency cracking parameters; relying on the main control inversion function and the deep parameter search model, the preheating temperature, cracking frequency and other main control variables are dynamically adjusted by real-time monitoring of the cracking efficiency and energy conversion efficiency deviation; the preheating compensation instruction is generated by using the waste heat conversion monitoring data to realize the dynamic coupling of waste heat recovery and preheating parameters; based on the three-dimensional density model, the inversion correlation mapping network is used to adaptively fluctuate the material density, and finally a whole-chain intelligent control system of "parameter collaboration-real-time feedback-dynamic compensation-global optimization" is formed, which significantly improves the resource utilization efficiency and process adaptive ability, and improves the control of the conversion time node of each process flow, further improving the cracking efficiency and energy conversion efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 The present application is aimed at the deficiencies of the prior art, and realizes intelligent collaborative control of the whole cracking process through the construction of a hierarchical cracking feedback control node chain and a parameter node network, and the deep integration of blockchain algorithms, tree databases and graph algorithms. The core value lies in: based on the two-dimensional structure of the horizontal node chain and the vertical parameter information chain, the inversion correlation mapping weight constructed by combining factor analysis contribution and convolution kernel function is used to realize the cross-level precise collaboration of key process parameters such as feeding rate, preheating residence time and double-frequency cracking parameters; relying on the main control inversion function and the deep parameter search model, the preheating temperature, cracking frequency and other main control variables are dynamically adjusted by real-time monitoring of the cracking efficiency and energy conversion efficiency deviation; the preheating compensation instruction is generated by using the waste heat conversion monitoring data to realize the dynamic coupling of waste heat recovery and preheating parameters; based on the three-dimensional density model, the inversion correlation mapping network is used to adaptively fluctuate the material density, and finally a whole-chain intelligent control system of "parameter collaboration-real-time feedback-dynamic compensation-global optimization" is formed, which significantly improves the resource utilization efficiency and process adaptive ability, and improves the control of the conversion time node of each process flow, further improving the cracking efficiency and energy conversion efficiency.

[0047] Figure 2 Workflow diagram of the hierarchical and staged cracking feedback control node chain of the present application. DETAILED DESCRIPTION

[0048] Example 1

[0049] Referring to Figure 1 An embodiment of the present application provides a microwave double-pole heating-based in-situ staged cracking method for carbon fiber composite materials, which comprises the following steps:

[0050] S1, initializing a feeding speed, based on a preset three-dimensional model of the material to be cracked and a preset staged inversion function space, a multi-stage cracking control sequence is obtained by inversion; the multi-stage cracking control sequence comprises a feeding rate, optimal temperatures and residence time lengths corresponding to first-stage preheating and second-stage preheating, an initial splitting preheating temperature, a double-frequency microwave radiation power, a staged cracking frequency interval and a temperature interval, a cracking efficiency of the material to be cracked, an energy consumption conversion efficiency, and a waste heat recovery utilization rate; the staged inversion function space comprises a hierarchical and staged cracking feedback control node chain; the hierarchical and staged cracking feedback control node chain is obtained by combining a correlation fuzzy control algorithm with a self-correlation function according to the multi-stage cracking control sequence;

[0051] S2, based on the multi-stage cracking control sequence and a preset staged linkage control model, a multi-stage separation control instruction is generated, and the real-time cracking efficiency and energy consumption state of the material to be cracked are monitored and evaluated and fed back to each control node in the hierarchical and staged cracking feedback control node chain for real-time adjustment of the multi-stage cracking control sequence.

[0052] Further, it needs to be further explained in this embodiment that the in-situ staged cracking process of the carbon fiber composite material is as follows:

[0053] First, the material is transported through a high-temperature-resistant stainless steel mesh belt, and the mesh belt speed is controlled by a variable frequency motor, ranging from 0.3

[0054] -1.5 m / min, which can adapt to different sizes of waste materials such as sheets and blocks; it should be noted that in the feeding area of this embodiment, a laser thickness gauge and an infrared imager are provided to scan the thickness and density distribution of the material in real time, and then generate a three-dimensional data model. The system automatically adjusts the preheating zone temperature, microwave power and cracking time according to the scanning data, so as to avoid over-burning or insufficient cracking of the material.

[0055] Second, after the material enters the preheating zone, it is heated in stages. The first preheating stage uses natural gas combustion tail gas to preliminarily heat the material, reducing its water content; the second preheating stage uses high-temperature flue gas after waste gas combustion for deep preheating, making the material surface temperature uniformly rise to 300°C; in this embodiment, it should be noted that the waste gas generated during the cracking process enters the combustion chamber after ionization, and the high-temperature gas after combustion passes through a finned heat exchanger to transfer heat to the preheating zone, achieving heat recovery and further reducing natural gas consumption; it should be noted that the first preheating stage uses natural gas combustion tail gas, and the temperature range is preferably 200-250°C; the second preheating stage uses high-temperature flue gas after waste gas combustion, and the temperature is preferably 350°C.

[0056] Third, the material enters the microwave cavity, which adopts a multi-layer composite structure, with a stainless steel shielding layer on the outside and quartz glass lining on the inside. Nitrogen gas is filled into the cavity to prevent fiber oxidation. The microwave heating system is equipped with a double-frequency microwave source for alternative radiation, with a power of 10-100kW adjustable. At the same time, the material surface and internal temperature are monitored by optical fiber temperature sensors and infrared thermal imagers, and the microwave power and frequency ratio are dynamically adjusted to ensure complete cracking of the resin in the shortest time. It should be further noted that the corresponding frequency of the double-frequency microwave source is determined according to the corresponding material thickness and material properties. For example, the frequency of 915MHz has a penetration depth of 30-50cm, suitable for thick materials, and the frequency of 2450MHz has high energy density, suitable for thin materials or resin-rich layers.

[0057] Fourth, the waste gas generated by cracking first enters the ionization reactor and is ionized into active ions under a high-voltage electric field, making large molecular organic matter crack into CO2 and H2. The ionized gas is mixed with natural gas and completely oxidized in the combustion chamber to eliminate harmful substances such as dioxins. The combustion tail gas is recovered for heat by a waste heat boiler, converted into steam, or directly heated to the preheating zone. The final exhaust gas meets the "Integrated Emission Standard of Air Pollutants" after bag dust removal and activated carbon adsorption.

[0058] Fifth, the cracked carbon fibers pass through a sealed cooling channel and are sprayed with liquid nitrogen mist to reduce the fiber temperature from 600°C to below 50°C within 10 seconds to inhibit oxidation. The cooled fibers are physically sorted, first separated from residual resin particles by a multi-layer stainless steel vibrating screen, then precisely separated by electrostatic separation under a high-voltage electric field due to the difference in electrical conductivity between carbon fibers and metal impurities, and finally separated into different lengths by airflow separation using a centrifugal fan to adjust airflow speed, achieving classification and collection of short fibers and long fibers.

[0059] Further, the construction steps of the layered and hierarchical cracking feedback control node chain in this embodiment include:

[0060] Obtain the corresponding pyrolysis process information of the carbon fiber composite material, and obtain the single pyrolysis process sequence, the pyrolysis process parameters corresponding to each pyrolysis process, the control parameter set and the overall pyrolysis process sequence information through the preset process pyrolysis model;

[0061] It should be noted that the process pyrolysis model in this embodiment is preferably constructed by combining a directed acyclic graph with a critical path method;

[0062] Firstly, it should be noted that in this embodiment, after obtaining the pyrolysis process information of the carbon fiber composite material, a directed acyclic graph (DAG) is constructed using an adjacency list data structure, and the physical decomposition, pyrolysis and other process stages are defined as single node objects, including but not limited to parsing modeling nodes, feeding nodes, first preheating nodes, second preheating nodes, frequency splitting pyrolysis nodes, ionization combustion control nodes, preheating recovery adjustment nodes, etc. Each node contains process stage name, material input / output state, and equipment resource occupation attribute; secondly, based on the material state transition relationship (exemplarily, solid state→semi-solid state→gaseous state), a directed edge is established, and the edge attribute includes state transition condition, time consumption parameter. The earliest start time and the latest start time of all nodes are calculated by the critical path method, and the critical process chain with a total time difference of zero is identified; combined with the material thickness distribution matrix scanned by the loading area laser thickness meter and the density gray image generated by the infrared imager, the threshold segmentation algorithm is used to divide the material density into n intervals, and each interval corresponds to the sub-heating stage of the preheating area node. Exemplarily, the region with a density greater than 1.8 g / cm³ triggers the third preheating stage, automatically generates a sub-node and associates the temperature parameter interval.

[0063] Secondly, it should be noted that in the process parameter channel, when the COMSOLMultiphysics coupled simulation model is used in this embodiment, a multi-physical field collaborative solution strategy is adopted, the electromagnetic field distribution is solved first to determine the microwave energy injection mode, and then the temperature field distribution is calculated by coupling the heat conduction equation. Through the grid self-adaptive encryption technology, the grid is refined in the material dielectric constant mutation area to improve the calculation precision of the temperature gradient. When the material dielectric property database is used, a nonlinear mapping model of material density, temperature and microwave absorption rate is established by using a support vector machine algorithm; in the control parameter channel, when the PID rule library is called, a fuzzy inference system is used to real-time set the PID parameters, and the input variables include material thickness, density distribution and current temperature deviation; when the frequency of the dual-frequency microwave source is determined based on the three-dimensional data model, a decision tree algorithm is used for frequency selection. Exemplarily, thick materials select 915MHz, medium thickness materials use a mixed frequency mode, and thin materials use 2450MHz; at the same time, when the residence time tolerance interval is solved by the heat conduction equation, a finite difference method is used for numerical solution, and combined with the mesh belt speed PID control algorithm, dynamic matching of feeding speed and temperature control is realized.

[0064] Based on the combination of the cracking process parameters corresponding to each cracking flow and the control parameters, the secondary cracking is performed on the flow cracking model to obtain a single cracking control sub-step sequence and a cracking process parameter sub-sequence corresponding to the single cracking control sub-step under each single cracking flow;

[0065] Third, when inputting the parameter set into the recursive decomposition engine, a hidden Markov model (HMM) is used to construct a state transition matrix, and the Baum-Welch algorithm is used for parameter estimation to determine the causal strength between parameters. When using singular value decomposition (SVD) to extract the main causal chain, set the eigenvalue contribution rate threshold to retain the main influencing factors. When generating the "heating-constant temperature-cooling" sub-step sequence, the dynamic time warping (DTW) algorithm is used for pattern recognition of the temperature curve, and when calculating the linear dependence between parameters, the Bonferroni correction is performed on the Pearson correlation coefficient to reduce false positives caused by multiple comparisons. In the preheating segmented heating, based on the waste gas combustion heat recovery situation to determine the heating slope, the recoverable heat is calculated by the energy balance equation, and the influence of waste heat fluctuation on the heating process is predicted by the neural network to dynamically adjust the power compensation parameters of the PID output.

[0066] Fourth, based on the results of the first, second, and third steps, the DAG topology and atomic sub-steps are mapped to a low-dimensional vector space using graph embedding technology. In particular, when analyzing timing conflicts using a constraint programming engine (CP-SAT), hard constraints are converted to linear inequality constraints, and soft constraints are converted to penalty terms in the objective function. The big M method is used to handle logical constraints, such as "complete preheating before entering microwave cracking" which is converted to a 0-1 integer programming problem. It should be further noted that when labeling parallel control domains, the resource-constrained project scheduling algorithm (RCPSP) is used to consider resource constraints such as device quantity and energy supply. When setting up the waste gas treatment parallel node, the Petri net modeling technology is used to verify the safety and feasibility of the parallel process through reachability analysis, and the process chain logic tree containing accurate timing constraints and parallel relationships is generated.

[0067] Based on all single cracking flows and monitoring control and evaluation feedback flows, the cracking control node set in the hierarchical and graded cracking feedback control node chain and the monitoring control node and feedback regulation node are constructed, and based on the overall cracking process flow sequence information and monitoring and evaluation feedback flows, the cracking flow sequence connection and feedback connection in the hierarchical and graded cracking feedback control node chain are constructed.

[0068] Based on all the cracking control node set and monitoring node and feedback regulation node combined with cracking flow sequence connection and feedback connection, the transverse cracking feedback control node chain is obtained through the blockchain algorithm.

[0069] It needs to be further explained that when the embodiment applies the blockchain algorithm to construct the horizontal chain, the consortium chain architecture is adopted, each control node participates in block verification as a consensus node, a special smart contract is designed to realize automatic chaining and verification of parameter data, data format verification, timestamp verification and permission control logic are embedded in the contract, PBFT (practical Byzantine fault tolerance) consensus algorithm is adopted to ensure data consistency on the chain, and the block generation interval is set to 100 ms to meet the real-time control requirement. When storing process parameters, wavelet compression algorithm is used for continuous parameters to reduce storage overhead, and multiple copies of redundant storage are used for key parameters such as temperature and power. A cross-node query protocol is designed to support efficient data retrieval based on time window and parameter type, and provide real-time correlation analysis capability for parameter adjustment.

[0070] Based on the single cracking control sub-step sequence under each single cracking process, a vertical parameter node set corresponding to each single cracking process is constructed.

[0071] It needs to be further explained that when the embodiment constructs the vertical parameter node set, a parameter object is created for each sub-step, including a process parameter sub-sequence and timestamp information, and a time series database is used to store the parameter node. Secondly, when constructing the vertical parameter connection, time series similarity matching algorithms such as Euclidean distance and Manhattan distance are used to identify the time sequence dependency between parameters. In the microwave cavity processing, the sliding window technology is used to extract the time pattern of the power parameter during the alternating radiation process of the dual-frequency microwave source, and the dynamic Bayesian network is used to establish a probability model of the time sequence and parameter change, ensuring the continuity and correlation of the parameters in the time dimension.

[0072] Based on the control sequence and time corresponding to the single cracking control sub-step sequence under the single cracking process, the vertical parameter connection is constructed.

[0073] It needs to be further explained that when the embodiment integrates the vertical parameter node set through the tree database, B+ tree index structure is used to optimize parameter query efficiency, and the fan-out coefficient of non-leaf nodes is set to 100 to reduce disk I / O. In addition, the embodiment creates a unique identifier for each sub-step and uses a hierarchical coding scheme to represent the hierarchical relationship of the parameters, such as the ionization reactor parameter of the exhaust gas treatment process coded as “EP-IR-001”. A parameter version control mechanism is implemented, a new version number is generated each time the parameter is updated, supporting historical parameter backtracking and comparison, and a tree traversal algorithm is designed to support parameter retrieval in different dimensions, such as time dimension and process stage dimension, providing efficient data access capability for subsequent parameter correlation analysis.

[0074] Based on the corresponding longitudinal parameter node set and the corresponding longitudinal parameter connection of all single cracking processes, the longitudinal parameter information chain corresponding to each single cracking process is constructed through the tree database, and the cracking process and control parameter corresponding to each sub-step are saved in the sub-node of the corresponding longitudinal parameter information chain;

[0075] Based on the cracking process control parameters saved in all longitudinal parameter information chains and the association algorithm, the association degree of the corresponding parameter nodes between each cracking control node is obtained;

[0076] It should be further explained that in the application of the Apriori algorithm to calculate the association degree of the parameter nodes, the minimum support threshold is set to 0.6 and the minimum confidence threshold is set to 0.7. In addition, when the continuous parameters are discretized in this embodiment, the equal frequency binning method is used to divide the parameter value range into 10 intervals. When analyzing the association between the preheating temperature and the microwave power parameters, the material density is considered as a adjusting variable, the conditional association rule mining technology is used to generate association rules such as “IF material density > 1.5 g / cm³ AND preheating temperature = 250°C THEN microwave power = 800W [support = 0.65, confidence = 0.82]”, and the Lift value calculation and chi-square test are performed on the generated association rules to ensure the statistical significance of the association relationship.

[0077] Based on the association degree of the corresponding parameter nodes between each cracking control node and the radial kernel function, the bidirectional mapping function and the association connection between the parameter nodes are constructed.

[0078] It should be further explained that in this embodiment, the radial basis function network (RBFN) is used to construct the bidirectional mapping function according to the association degree, and the Gaussian kernel function is selected as the activation function. It should be noted that in this embodiment, the number of hidden layer nodes is set to 3 times the parameter dimension, the center parameter is initialized by the K-means algorithm, and the weight parameter is determined by the least squares method. In addition, in the dual-frequency microwave heating, the mapping between the material thickness, density and the dual-frequency frequency, residence time length, power parameters is established, the regularization term is introduced to prevent overfitting, and the leave-one-out cross-validation is used to evaluate the generalization ability of the model. In addition, an incremental learning mechanism is designed, when new parameter samples appear, the online gradient descent algorithm is used to update the network parameters to maintain the adaptability of the mapping function.

[0079] Based on the bidirectional mapping function and the corresponding association connection and all parameter nodes, the control parameter node network is constructed by graph algorithm;

[0080] It needs to be further explained that when the control parameter node network is constructed by using the graph algorithm in this embodiment, a weighted directed graph model is adopted, the nodes represent parameters, and the edges represent mapping functions and correlation degrees. It needs to be further explained that when the shortest path between nodes is calculated in this embodiment, the Dijkstra algorithm is adopted, and the edge weight is set as the reciprocal of the correlation degree. In addition, when the node importance evaluation mechanism is implemented in this embodiment, the PageRank algorithm is adopted to calculate the influence score of the parameter node, and the key control parameter is identified. To support dynamic adjustment of parameters, a graph editing operation interface is designed in this embodiment to allow adding, deleting nodes and edges, and updating mapping function parameters in runtime, so as to ensure that the control parameter node network can adapt to changes in process conditions.

[0081] Based on the parameter sequence corresponding to each cracking control node, the contribution degree of each parameter node to the cracking efficiency and energy conversion efficiency under each cracking control node is obtained by using the factor analysis algorithm in combination with the set cracking efficiency and energy conversion efficiency evaluation variables.

[0082] It needs to be further explained that when the parameter contribution degree is calculated by using the factor analysis algorithm in this embodiment, KMO test and Bartlett sphere test are first performed to ensure that the data is suitable for factor analysis, the cumulative variance contribution rate threshold of the extracted factor is set to 85%, the maximum variance rotation method is adopted to obtain the orthogonal factor solution, in the microwave cracking node analysis, a correlation matrix containing parameters such as double-frequency frequency, power and residence time is constructed, the common factor is determined by principal component analysis, the load value of each parameter on each factor is calculated, the square of the load value is taken as the contribution degree of the parameter to the factor, the cracking efficiency and energy conversion efficiency evaluation variables are combined, the path coefficient of the parameter contribution degree is verified by using the structural equation model, and the key parameter and its influence path are determined.

[0083] Based on the contribution degree of each parameter node to the cracking efficiency and energy conversion efficiency, the main control variable and auxiliary control variable sequence corresponding to each cracking control node are obtained.

[0084] It needs to be further explained that when the main and auxiliary control variables are determined according to the parameter contribution degree in this embodiment, the contribution degree threshold is set to 0.6, the parameters with a contribution degree greater than the threshold are taken as the main control variables, and the rest are taken as the auxiliary control variables. For example, in the preheating node, the contribution degree of the preheating temperature to the cracking efficiency is 0.82 and the contribution degree of the residence time is 0.45, so the preheating temperature is set as the main control variable and the residence time is set as the auxiliary control variable. Different control strategies are designed for the main and auxiliary control variables, the main control variable adopts PID cascade control, the auxiliary control variable adopts feedforward compensation control, and the main and auxiliary control variable coordination mechanism is established. When the main control variable approaches the constraint boundary, the auxiliary control variable is automatically adjusted to maintain system stability.

[0085] Based on the contribution of each auxiliary control variable in the two-split control node to the corresponding main control variable and the correlation degree of the parameter nodes between the two-split control nodes, an inverse correlation mapping weight between the two-split control nodes is constructed.

[0086] It needs to be further explained that in the present embodiment, the relative importance of the contribution of the auxiliary control variable to the main control variable and the correlation degree between the nodes is determined by the analytic hierarchy process (AHP) when constructing the inverse correlation mapping weight. When constructing the judgment matrix, the comparison scale is determined by combining expert evaluation and historical data statistics. In addition, when calculating the inverse correlation mapping weight between the preheating node and the microwave cracking node, the direct influence path of the preheating temperature on the microwave power and the indirect influence path through the material state change are considered, the fuzzy analytic hierarchy process is used to handle the uncertainty, the consistency of the judgment matrix is tested, and the rationality of the weight distribution is ensured.

[0087] Based on the inverse correlation mapping weight between the two-split control nodes and the main control variable of the corresponding two-split control nodes, a main control inverse function is constructed by a convolution kernel function, and the main control inverse function is embedded into the cracking process sequence connection between the corresponding two-split control nodes.

[0088] It needs to be further explained that in the present embodiment, the convolutional neural network is used to construct the main control inverse function combining the inverse correlation mapping weight and the main control variable. A one-dimensional convolution layer is designed to capture the time sequence dependence between parameters, the convolution kernel size is set to 3 and the step size is set to 1, the dimensionality is reduced by the pooling operation to extract features, the inverse correlation mapping weight is used as the initial parameter of the convolution kernel, and the network parameters are optimized by the back propagation algorithm. In the present embodiment, the product of the contribution of the main control variable to the cracking efficiency and the energy conversion efficiency is normalized, and the Sigmoid function is used to map it to the interval [0, 1]. In addition, the main control inverse function and the feedback control factor are integrated into the node connection to realize the closed-loop control of the parameters. When the cracking efficiency is monitored to decrease, the main control variable of the associated node is automatically adjusted.

[0089] Based on the product of the contribution of each main control variable to the cracking efficiency and the energy conversion efficiency, a feedback control factor is constructed and embedded into the feedback connection between each cracking control node and the feedback control node to obtain a hierarchical cracking feedback control node chain.

[0090] Based on the hierarchical cracking feedback control node chain and the deep control parameter search model configured by the control parameter node network, a simulation algorithm, a preset optimal feedback control function, and a cracking efficiency threshold and an energy conversion efficiency threshold are trained to obtain a trained hierarchical cracking feedback control node chain.

[0091] The optimal feedback control function is constructed according to the feedback regulatory factors in the feedback column, the combination of the main control variable, the splitting efficiency and the energy conversion efficiency, and the linear function, and the maximum value of the splitting efficiency and the energy conversion efficiency.

[0092] It should be further explained that when the deep control parameter search model is used to train the node chain, a deep reinforcement learning algorithm is adopted, and a proximal policy optimization algorithm is selected as a basic framework. It should be further explained that the state space designed in the embodiment includes the main control variable, the auxiliary control variable and the evaluation index of each node, and the action space is the parameter adjustment amplitude. When setting the reward function, the preset optimal feedback control function is taken as the target, the splitting efficiency and the energy conversion efficiency are taken as the main evaluation indexes, the experience replay mechanism is adopted to improve the sample utilization rate, the batch size is set to 64, and the training period is 10000 steps. In addition, when simulating the splitting process of different material densities and thicknesses, the Latin hypercube sampling method is adopted to generate training samples, cover the typical distribution interval of the material characteristics, and continuously adjust the node chain parameters through training, so that the optimal feedback control function obtains the maximum value of the splitting efficiency and the energy conversion efficiency, and finally the intelligent control node chain which can adapt to the actual splitting demand is obtained.

[0093] The process realizes the optimization of the whole process of carbon fiber composite pyrolysis through multidimensional technology integration. The core innovation lies in the construction of a hierarchical feedback control system and a multi-physical field coordination mechanism, forming a complete closed loop from material property perception to intelligent process parameter regulation. First, the directed acyclic graph and critical path method are used to analyze the pyrolysis process topology, mapping macro-process stages such as physical decomposition and pyrolysis into independent nodes, and establishing directed connections based on material state transition relationships to form an initial process skeleton. Combined with the material thickness, density three-dimensional data model generated by the laser thickness gauge and infrared imager in the feeding area, the node definition and edge relationship are dynamically adjusted, for example, the heating stage and temperature interval of the preheating zone node are divided according to the material density distribution, which deeply couples the process topology and material properties, and accurately defines the operation unit boundary for subsequent parameter extraction, avoiding cross-process interference. Second, the double-channel parameter extraction mechanism calculates the temperature field distribution through COMSOL Multiphysics coupled simulation model, generates a dedicated microwave absorption rate based on the material dielectric property database, and provides accurate process parameters for microwave pyrolysis. The control parameter channel calls the PID rule base built based on the distributed database and index algorithm, dynamically selects the frequency of the double-frequency microwave source according to the material thickness, and derives the temperature adjustment rate and power step function, while solving the residence time tolerance interval based on the heat conduction equation, realizing the coordinated optimization of control parameters and process parameters. This double-channel design enables the system to adjust the pyrolysis strategy according to the real-time material properties. For example, for composite materials with uneven density, the high-density area is identified through the three-dimensional data model, and the microwave power and residence time are increased accordingly to ensure the completeness of pyrolysis. Third, the recursive decomposition engine converts the parameter set into a sequence of atomic-level sub-steps such as "heating-constant temperature-cooling", extracts the main causal chain through the state transition matrix and singular value decomposition, calculates the linear dependence between parameters based on the Pearson correlation coefficient, and generates a sub-sequence of process parameters.In the preheating zone heating process, the mechanism determines the temperature parameters of the first preheating stage (200-250°C) and the second preheating stage (350°C), as well as the waste gas combustion waste heat recovery, to accurately determine the temperature rise slope and constant temperature time, and form a closely related sub-step sequence; the constraint programming engine (CP-SAT) uses DAG topology as the skeleton, inserts atomic sub-steps as leaf nodes, analyzes timing conflicts and marks parallel control domains, such as allowing preheating and feeding to be executed concurrently, while handling the parallel processes of waste gas ionization combustion and preheating zone heat recovery, generating a process chain logic tree containing timing constraints and parallel relationships to ensure process execution efficiency; fourth, the horizontal feedback chain constructed by the blockchain algorithm realizes the block storage and association of process parameters and control parameters at each node, ensuring that data is tamper-proof and traceable; for example, the web speed of the feeding node and the temperature parameters of the preheating node are real-time chained to form a traceable parameter flow, enabling the system to provide reliable basis for parameter adjustment by real-time feedback of parameter correlation between nodes; the vertical parameter information chain constructed by the tree database stores sub-step parameters in chronological order, and the bidirectional mapping function constructed by the radial kernel function realizes the dynamic association between parameter nodes. In the microwave cavity processing, the mechanism establishes real-time linkage between temperature monitoring and power adjustment according to the time sequence and power parameters of the alternating radiation of the dual-frequency microwave source, ensuring the continuity and correlation of parameters in the time dimension; fifth, the contribution of each parameter to the cracking efficiency and energy conversion efficiency is calculated by the factor analysis algorithm to determine the sequence of main control variables and auxiliary control variables, and to construct the inversion correlation mapping weight and main control inversion function, realizing intelligent mapping and collaborative adjustment between parameters; when the cracking efficiency fluctuates, the system automatically adjusts the preheating rate and microwave radiation power, combined with the ionization control coefficient compensation of the waste gas treatment link, to form a cross-node parameter collaborative optimization; the deep control parameter search model optimizes feedback control factors through simulation training, enabling the system to seek the optimal balance between cracking efficiency and energy conversion efficiency, for example, when the cracking efficiency approaches the threshold, the energy consumption parameter sensitivity is automatically reduced to prioritize cracking quality; In summary, the embodiment realizes the intelligentization, precision and high efficiency of the carbon fiber composite material cracking process through the complete technical chain of "material property perception - process topology construction - parameter dynamic extraction - sub-step decomposition - timing optimization - feedback control". From macro process scheduling to atomic level parameter control, each technical module cooperates and progresses layer by layer, forming a multi-dimensional and multi-level optimization closed loop, significantly improving the system's adaptability to different material properties, energy utilization efficiency and process stability, providing an innovative technical solution for the green recycling and reuse of carbon fiber composites.

[0094] Further, the cracking control node set in the embodiment includes an analysis modeling node, a feeding node, a first preheating node, a second preheating node, a frequency splitting cracking node, an ionization combustion control node, and a preheating recovery adjustment node. It should be further explained that the feeding node, the first preheating node, the second preheating node, the frequency splitting cracking node, the ionization combustion control node, and the preheating recovery adjustment node in the embodiment are all configured with a fuzzy control submodel, which is constructed and trained by combining a fuzzy control algorithm with the cracking process parameters of the corresponding node and the interactive control parameters of the adjacent node.

[0095] It should be explained that the analysis modeling node in the embodiment constructs a three-dimensional density model by a three-dimensional modeling algorithm, in combination with the geometric parameters, material density, and optimal layered cracking parameters of the target material to be cracked, to provide material characteristic basic data for the subsequent process, wherein the optimal layered cracking parameters are used to determine the double-frequency frequency value to maximize the cracking efficiency. The feeding node receives a feeding rate instruction, generates a feeding control instruction in combination with the built-in first fuzzy control submodel, controls the speed of the high-temperature-resistant stainless steel mesh belt through a variable frequency motor, adapts to different sizes of waste, and adjusts the feeding speed and preheating parameters in cooperation with the first preheating node through the first inversion correlation mapping. The first preheating node uses natural gas combustion tail gas for preliminary heating, adjusts the residence time and the heating rate according to the material density and the composite density in combination with the second fuzzy control submodel, and reduces the water content; the second preheating node deeply preheats to the target temperature through waste gas combustion high-temperature flue gas, optimizes the parameters based on the fuzzy control submodel at the same time, and monitors the waste heat conversion efficiency. The frequency splitting cracking node alternately radiates the double-frequency microwave source through the third fuzzy submodel according to the cracking double-frequency frequency value and the third residence time length after the material reaches the second preheating target temperature, dynamically adjusts the power and frequency ratio to ensure that the resin is completely decomposed and the cracking efficiency is monitored in real time. The ionization combustion control node generates ionization control instructions through the fourth inversion correlation mapping and the fourth fuzzy submodel based on the energy conversion efficiency and other parameters transmitted by the frequency splitting cracking node, burns the waste gas after ionization, and eliminates harmful substances. The preheating recovery adjustment node compensates the heating rate of the preheating node according to the energy conversion efficiency, feeding rate, and other parameters through the fifth, sixth inversion correlation mapping and the fifth fuzzy submodel, to realize heat energy recycling. The monitoring control node collects the working parameters of each node in real time, feeds back to the feedback control node to evaluate the cracking and energy consumption efficiency, generates a deviation value, and then optimizes the parameters through the main control inversion function and the deep control parameter search model, to drive the fuzzy control submodel of each node to generate adjustment instructions, forming a closed-loop control.

[0096] Further, please refer to Figure 2 The working process of the layered and hierarchical cracking feedback control node chain includes:

[0097] The three-dimensional density model of the target to be cracked is constructed by the analysis modeling node according to the built-in three-dimensional modeling algorithm in combination with the geometric parameter information of the target to be cracked, the density of each material in the composite material and the optimal delamination cracking parameter and the composite density; the optimal delamination cracking parameter is the double-frequency frequency value corresponding to the density of each density material in the composite material at the optimal second preheating target temperature, which makes the cracking efficiency reach the maximum;

[0098] Based on the three-dimensional density model of the target to be cracked, the first preheating temperature interval and the second temperature interval are preset, and the first inversion correlation mapping between the feeding speed between the feeding node and the first preheating node, the density of each material and the composite density and the first optimal preheating temperature parameter, the first residence time length, the preset first preheating target temperature, the first preheating target temperature between the first preheating node and the second preheating node, the density of each material and the composite density, the feeding rate, the second residence time length, the second preheating temperature parameter, the preset second preheating target temperature, and the third inversion correlation mapping corresponding to the second preheating target temperature, the optimal cracking parameter and the third residence time length between the second preheating node and the frequency splitting cracking node are obtained. The feeding rate corresponding to the feeding node, the residence time length and the heating rate corresponding to the first preheating node and the second preheating node, and the cracking double-frequency frequency value and the third residence time length corresponding to the frequency splitting cracking node.

[0099] It needs to be further explained that after the analysis modeling node in the embodiment receives the geometric parameter information of the target to be cracked, first, the material surface point cloud data is obtained by a three-dimensional laser scanner, an initial three-dimensional grid model is constructed by using a Delaunay triangulation algorithm, and the internal density distribution data of the material is obtained by combining an X-ray density scanning technology. The discrete density point data is interpolated into the three-dimensional grid by using the Kriging interpolation method to form an initial density field. Then, the optimal delamination cracking parameter database is called, which stores the corresponding relationship of the double-frequency frequency value of different materials at the optimal second preheating target temperature, and the frequency combination that makes the cracking efficiency maximum is found by using a particle swarm optimization algorithm; for carbon fiber composite materials, the system will automatically match the corresponding frequency response model according to the resin matrix type (such as epoxy resin, phenolic resin), and finally generate a three-dimensional density model containing density distribution and optimal cracking parameter; the model uses octree data structure for space division, and subdivides in the area with large density gradient to improve the parameter resolution and provide accurate material property basis for the subsequent process.

[0100] It needs to be further explained that the analysis modeling node in the embodiment preferably receives the geometric parameter information of the target material to be cracked, starts a three-dimensional laser scanner to scan the surface of the material, acquires point cloud data with a sampling interval of 0.1 mm, constructs an initial three-dimensional grid model by using a Delaunay triangulation algorithm, avoids long and narrow triangular elements by maximizing the minimum angle criterion, and simultaneously starts an X-ray density scanner to perform tomographic scanning on the material with a layer thickness of 5 mm to acquire internal density distribution data. The discrete density points are spatially interpolated by using a Kriging interpolation method, the spatial correlation is calculated by selecting a Gaussian variation function, the variation function parameters are fitted by using a least squares method, the density data is interpolated to each node of the three-dimensional grid to form an initial density field; secondly, when calling the optimal delamination parameter database, the database uses a MongoDB architecture, and composite indexes are established according to material types, density intervals and second preheating temperatures; for carbon fiber composites, the system first identifies the resin matrix type, analyzes the resin functional group characteristic peak by Fourier transform infrared spectroscopy, matches the frequency response model corresponding to the epoxy resin or phenolic resin in the database, finds the optimal double-frequency combination by using a particle swarm optimization algorithm, sets the particle number to 30, the iteration number to 50, and the inertia weight to linearly decrease from 0.9 to 0.4, the learning factor , the objective function is a nonlinear mapping of the delamination efficiency and the frequency combination, the particle swarm position is initialized by using a genetic algorithm, and the frequency ratio that maximizes the delamination efficiency is quickly converged to, thirdly, the three-dimensional density model in the embodiment uses an octree data structure for storage, the root node represents the entire material space, each time the node is divided into 8 sub-nodes, the density gradient threshold is set to 0.2 g / cm³, when the density standard deviation in the node exceeds the threshold, the node is further divided, the maximum depth is 7 layers, and it is ensured that the region with a large density gradient has a higher parameter resolution; the model output includes the density value of each voxel, the corresponding optimal double-frequency and the second preheating target temperature, and is stored in an HDF5 format to support fast random access; it needs to be further explained that the thermal resistance network model in the embodiment is a classical method in heat transfer that abstracts a complex thermal system into an equivalent model of “thermal resistance-heat source” by analogy with an electric circuit network, the core of which is to quantify heat transfer processes such as heat conduction, convection and radiation into thermal resistance elements (such as conduction thermal resistance), and to build the topological relationship of nodes and edges by graph theory, and to solve the heat flow and temperature difference by establishing a linear equation system based on Kirchhoff's law. The model can be physically discretized by using the lumped parameter method or hierarchical modeling, and is often coupled with the finite element method or modeled by using special software such as FloTHERM

[0101] According to the obtained feeding rate, a first fuzzy control sub-model built in the feeding node is combined to obtain a feeding control instruction for feeding, while first preheating and second preheating are performed based on the combination of the residence time length and the temperature rising rate corresponding to the first preheating node and the second preheating node and a second fuzzy control sub-model, and the waste heat conversion efficiency is monitored in real time based on a preset preheating conversion curve;

[0102] It needs to be further explained that the deep analysis model in the embodiment adopts an 18-layer ResNet architecture, the input layer is a density distribution matrix of 128x128x3, contains 5 residual blocks, each residual block is composed of two 3x3 convolution layers, adopts Batch Normalization and ReLU activation function, in the first inversion correlation mapping, a density sensitive region mask is generated through a spatial attention mechanism to calculate the weight value of each pixel, and the region with a weight value greater than 0.7 is identified as a key region. Secondly, when outputting the first optimal preheating temperature parameter, the Monte Carlo dropout technique is used for uncertainty evaluation, 20% of the neurons are randomly discarded during forward propagation of the model, 50 times of sampling are performed, the mean and standard deviation of the temperature prediction value are calculated, and a 95% confidence interval is constructed. It needs to be further explained that the second inversion correlation mapping in the embodiment introduces historical batch data as a priori to construct a Bayesian optimization framework based on a Gaussian process, sets the acquisition function as expected improvement (EI), selects the point with the maximum EI value in the two-dimensional parameter space of material density and second preheating temperature for evaluation each time, combines the heat resistance network model to calculate the residence time compensation, and the heat resistance network regards the material as a multi-layer flat plate, divides the heat resistance layer according to the density interval, and then obtains the temperature conduction time to determine the residence time compensation value; thirdly, the third inversion correlation mapping adopts an 8-layer time sequence convolution network, the convolution kernel size is 3, and the expansion factor is 1, 2, 4, 8, 16, 32, 64 and 128, which captures the time lag effect of the second preheating temperature change on the cracking parameters, calculates the dynamic time warping (DTW) distance of the current parameters and historical successful cases when outputting the double-frequency frequency value, uses the Euclidean distance as the measurement, accelerates the calculation through Sakoe-Chiba bandwidth limitation, and selects the frequency combination corresponding to the historical case with the minimum distance as the output to ensure the rationality of the parameters.

[0103] When the corresponding second preheating target temperature is reached, the double-frequency cracking control is performed through the combination of the double-frequency cracking frequency value and the third residence time length and the third fuzzy sub-model, and the cracking efficiency is monitored in real time;

[0104] It needs to be further explained that the first fuzzy control sub-model of the feeding node in the embodiment adopts a Mamdani type fuzzy reasoning system, the input variable feeding rate instruction is divided into 5 fuzzy sets {very low, low, medium, high, very high}, the material density fluctuation is divided into 5 fuzzy sets {very small, small, medium, large, very large}, and the output variable variable frequency motor control voltage is divided into 5 fuzzy sets {very low, low, medium, high, very high}. It needs to be further explained that the membership function in the embodiment adopts Gaussian distribution, the historical data is clustered through K-means clustering algorithm, the center and standard deviation of each fuzzy set are determined, the fuzzy rule base contains 25 rules, such as “if the feeding rate instruction is high and the material density fluctuation is large, then the output voltage is high”, the rule weight is determined through expert experience and historical control effect, and the defuzzification in the embodiment adopts the barycenter method to output the control signal which is smoothed by the Kalman filter. It needs to be further explained that the second fuzzy control sub-model of the first and second preheating nodes in the embodiment increases a fuzzy compensation link on the basis of traditional PID control, the input variables are waste heat conversion efficiency deviation and temperature deviation, each is divided into 7 fuzzy sets {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}, and the output is a temperature rise rate compensation amount, which is divided into 7 fuzzy sets {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}. The fuzzy rule adopts Mamdani product operation, such as “if the waste heat conversion efficiency deviation is negative large and the temperature deviation is negative large, then the temperature rise rate compensation amount is positive large”, the rule antecedent matching adopts the maximum membership degree method, and the consequent aggregation adopts the maximum value method. It needs to be further explained that the third fuzzy sub-model of the frequency splitting cracking node in the embodiment designs a two-dimensional fuzzy controller, the inputs are material temperature and cracking degree, and the output is power ratio. The embodiment adopts a product reasoning machine, single-point fuzzification, and defuzzification adopts a weighted average method, so as to ensure smooth switching of the microwave frequency, and the power transition time at the frequency switching time is set to 0.5 seconds, so as to avoid energy mutation.

[0105] Based on the deviation values of the real-time monitored waste heat conversion efficiency and cracking efficiency and the corresponding waste heat conversion efficiency threshold and cracking efficiency threshold, the ionization control instruction is generated by the fourth inverse correlation mapping constructed by the energy conversion efficiency, the natural gas input quantity and the ionization control coefficient between the frequency splitting cracking node and the ionization combustion control node in combination with the fourth fuzzy sub-model, so as to ionize the generated waste gas.

[0106] According to the fifth inverse correlation mapping and the sixth inverse correlation mapping respectively constructed by the energy conversion efficiency, the feeding rate, the preheating temperature rise rate and the corresponding first or second residence time length of the preheating recovery node between the ionization combustion control node and the preheating recovery node in combination with the fifth fuzzy sub-model, the compensation control instruction for the first or second preheating node is obtained, and the preheating temperature rise rate of the first or second preheating node is compensated according to the density and the composite density of each material of the target material to be cracked.

[0107] It needs to be further explained that the fourth fuzzy sub-model input variables of the ionization combustion control node in the embodiment are energy conversion efficiency, natural gas input and ionization control coefficient, which are processed by Z-score standardization, the fuzzy rule base is generated by combining expert knowledge and historical data, the decision tree algorithm is used to classify the historical data, the rules are extracted and the weights are optimized; when the cracking efficiency decreases, the system adjusts the ionization control coefficient through the fourth inverse correlation mapping, the inverse model adopts multilayer perception, the input is cracking efficiency deviation, energy conversion efficiency and natural gas input, the output is ionization voltage adjustment amount, the hidden layer contains 2 layers, the number of neurons is 10 and 5 respectively, and the activation function is ReLU. The ionization process in the embodiment adopts a pulse power source composed of IGBT modules, the main circuit is a full-bridge topology, the driving signal is generated by DSP chip TMS320F28379D, the frequency range is 1-10 kHz adjustable, the pulse width adjustment range is 1-100 μs, the output voltage is 0-30 kV adjustable, the ionization current is monitored in real time through a sampling resistor, and the hysteresis control is used to keep the current stable, and the hysteresis width is set to 5%. It needs to be further explained that the fifth fuzzy sub-model of the preheating recovery adjustment node in the embodiment adopts a feedforward-feedback control structure, the feedforward part calculates the theoretical compensation amount based on the energy balance equation, calculates the recoverable energy according to the heat generated by ionization combustion, determines the theoretical heating amount of the preheating zone, and the feedback part corrects the deviation through a fuzzy controller, the input is the deviation and the change rate of the actual temperature and the target temperature, and the output is the heating power compensation amount. The fuzzy set division is the same as that of the second fuzzy control sub-model.

[0108] Meanwhile, based on the monitoring control node, the working parameters of the feeding node, the first preheating node, the second preheating node, the frequency division cracking node, the ionization combustion control node and the preheating recovery adjustment node are monitored, and the monitoring parameters are input to the feedback regulation node for cracking efficiency and energy conversion efficiency evaluation, to obtain cracking efficiency evaluation value and energy conversion efficiency evaluation value;

[0109] Based on the cracking efficiency evaluation value and the energy conversion efficiency evaluation value and the cracking efficiency threshold value and the energy conversion efficiency threshold value, the cracking efficiency deviation value and the energy conversion efficiency deviation value are obtained;

[0110] It needs to be further explained that the monitoring control node in the embodiment deploys a multi-sensor network, including an infrared thermal imager, a gas analyzer, a laser range finder, etc.; it needs to be noted that the infrared thermal imager generates a three-dimensional temperature distribution cloud map through a temperature field reconstruction algorithm; it needs to be noted that the gas analyzer monitors the concentrations of CO, CO2, NOx and other gases in real time, with a sampling frequency of 1 Hz. After receiving the monitoring data, the feedback control node first performs data fusion, fuses multi-sensor data using Kalman filtering to improve state estimation accuracy. Then, the preset evaluation function is used to calculate the cracking efficiency and energy conversion efficiency. The cracking efficiency is obtained by comparing the concentration of the target component in the gas product with the theoretical value, and the energy conversion efficiency is calculated based on the ratio of input energy to effective cracking energy. After obtaining the deviation value, the main control inversion function searches for parameters through the control parameter node network, and uses a genetic algorithm to find the optimal solution in the parameter space. The crossover probability is set to 0.8, the mutation probability is set to 0.05, and the optimal parameter combination is output after 50 iterations.

[0111] Based on the cracking efficiency deviation value and the energy conversion efficiency deviation value, the feedback is connected to the hierarchical and graded cracking feedback control node chain, the optimal deviation control parameter is searched synchronously through the main control inversion function combined with the control parameter node network and the configured deep control parameter search model, and the optimal deviation control parameter obtained is fed back to the corresponding cracking control node. At the same time, the optimal main control variable corresponding to the adjacent two cracking control nodes is used as a constraint, and a fuzzy control model of the corresponding cracking control node is used to generate a cracking deviation adjustment instruction, so that the cracking efficiency and the energy conversion efficiency meet the corresponding condition threshold in real time.

[0112] It needs to be further explained that after the feedback control node outputs the optimal deviation control parameter, the system generates a control instruction through a model predictive control (MPC) framework with the optimal main control variable of the adjacent node as a constraint. In the feeding node, the MPC controller predicts the material position in the future 10 time steps, adjusts the feeding speed according to the temperature constraint of the first preheating node, and the control time domain is 5 steps and the prediction time domain is 10 steps. For the preheating node, a nonlinear MPC algorithm is used to consider the nonlinear characteristics in the heat transfer process, and a sequential quadratic programming (SQP) method is used to solve the optimization problem. In the frequency division cracking node, the MPC controller dynamically adjusts the power distribution of the dual frequency, while meeting the temperature and time constraints of the adjacent nodes, combined with the material density distribution and the microwave electromagnetic field simulation model. After the control instruction is issued to each node, the fuzzy control sub-model optimizes the instruction again, considers the dynamic characteristics of the actuator and environmental disturbances, and realizes parameter fine-tuning through an adaptive fuzzy PID control algorithm to ensure that the system can operate stably under various working conditions, so that the cracking efficiency and the energy conversion efficiency are maintained within the threshold range.

[0113] The embodiment realizes intelligent and accurate regulation and control of the whole process of carbon fiber composite material pyrolysis by constructing a hierarchical and graded pyrolysis feedback control node chain. The technical means and effects present multi-dimensional collaborative optimization characteristics. Specifically, the analysis modeling node adopts a combination of three-dimensional laser scanning and X-ray density scanning and a Kriging interpolation method to construct a three-dimensional model containing density distribution and optimal pyrolysis parameters. Exemplarily, the area with a density greater than 1.8 g / cm³ automatically triggers a three-stage preheating stage, so that the material property analysis accuracy is improved, and an accurate basis is provided for subsequent parameter inversion. The parameter inversion link adjusts the preheating temperature and residence time dynamically according to the material density distribution by using a ResNet deep model combined with a thermal resistance network. For example, the thermal resistance network is used to calculate the heat conduction time of different density layers to realize accurate compensation of the residence time and avoid pyrolysis deviation caused by material unevenness. In addition, the fuzzy control sub-model in the embodiment designs a dedicated control strategy for the characteristics of each node. The feeding node adjusts the belt speed according to the density fluctuation to adapt to waste materials of different sizes. The preheating node introduces waste heat conversion efficiency feedback to dynamically adjust the heating rate. For example, when the waste heat efficiency is low, the heating power is automatically increased. The frequency division pyrolysis node dynamically adjusts the power ratio of the dual-frequency power by a two-dimensional fuzzy controller to ensure complete decomposition of the resin. The waste gas treatment link ionizes the waste gas by using an IGBT pulse power supply and recovers heat energy by using feedforward-feedback control. For example, the heat energy of the combustion waste gas is accurately introduced into the preheating zone by calculating the theoretical compensation amount by using an energy balance equation to realize efficient use of energy. In addition, the monitoring feedback node in the embodiment improves data accuracy by using multi-sensor fusion and Kalman filtering and optimizes parameter search by using a genetic algorithm. For example, in parameter deviation adjustment, the optimal control combination is found by 50 generations of genetic iteration. The closed-loop control adopts an MPC framework to predict and optimize. For example, the feeding node predicts the material position in the next 10 steps according to the preheating temperature constraint and dynamically adjusts the speed. At the same time, the frequency division pyrolysis node optimizes the dual-frequency power distribution in combination with a microwave electromagnetic field model to ensure temperature uniformity. In summary, the embodiment realizes collaborative optimization of pyrolysis efficiency and energy consumption through the technical closed loop of material modeling, parameter inversion, fuzzy control, waste heat recovery, and monitoring feedback. For example, when the pyrolysis efficiency is found to be decreased, the system adjusts the preheating heating rate by using a main control inversion function and triggers ionization voltage compensation at the same time to form cross-node collaborative adjustment. The combination of the three-dimensional density model and the thermal resistance network improves the temperature control accuracy in the preheating stage. The adaptive switching of the dual-frequency microwave ensures energy matching of materials of different thicknesses. The parameter tracing system constructed by the blockchain and the tree database provides historical data support for process optimization, and finally realizes the efficiency, intelligence, and greenness of the carbon fiber composite material pyrolysis process.

[0114] Embodiment 2

[0115] An electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements a microwave dual-pole heating based in-situ graded pyrolysis method of carbon fiber composite material when executing the computer program.

[0116] A computer readable storage medium having stored thereon computer instructions that, when executed, perform a microwave bipolar heating based in-situ staged pyrolysis method of carbon fiber composites.

[0117] The embodiments of the present application are described above with reference to the drawings, but the present application is not limited to the specific embodiments described above, and the specific embodiments described above are merely illustrative, but not restrictive, and a person of ordinary skill in the art can make changes, modifications, replacements and variations to the above embodiments without departing from the purpose of the present application and the scope protected by the claims, and these are all within the protection of the present application.

Claims

1. A method for in-situ fractionated pyrolysis of carbon fiber composites based on microwave bipolar heating, characterized in that, The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method.

2. The microwave dual polar heating based in-situ staged pyrolysis of carbon fiber composite material process as claimed in claim 1 wherein, The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method.

3. The microwave dual polar heating based in-situ staged pyrolysis process of carbon fiber composite material as claimed in claim 2, wherein, The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method.

4. The microwave dual polar heating based in-situ fractional pyrolysis of carbon fiber composite material as claimed in claim 3 wherein, The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyrolysis method. The application relates to a hierarchical and staged pyrolysis feedback control node chain and a hierarchical and staged pyro Based on the correlation degree of the corresponding parameter nodes between each cracking control node, a bidirectional mapping function and an associated connection between the parameter nodes are constructed by combining a radial kernel function.

5. The microwave dual polar heating based in-situ staged pyrolysis process of carbon fiber composite material as claimed in claim 4 wherein, The construction steps of the hierarchical and hierarchical cracking feedback control node chain further include: Based on the bidirectional mapping function and the corresponding associated connection, a control parameter node network is constructed by combining all parameter nodes through a graph algorithm; Based on the corresponding parameter sequence of each cracking control node, the contribution of each parameter node to the cracking efficiency and energy conversion efficiency under each cracking control node is obtained by combining the set cracking efficiency and energy conversion efficiency result evaluation variables through a factor analysis algorithm; Based on the contribution of each parameter node to the cracking efficiency and energy conversion efficiency, the main control variable and auxiliary control variable sequence corresponding to each cracking control node are obtained.

6. The microwave dual polar heating based in-situ staged pyrolysis process of carbon fiber composite material as claimed in claim 5 wherein, The construction steps of the hierarchical and hierarchical cracking feedback control node chain further include: Based on the contribution of each auxiliary control variable in each cracking control node to the corresponding main control variable and the correlation degree of the parameter nodes between the two cracking control nodes, an inverse correlation mapping weight between the two cracking control nodes is constructed; Based on the inverse correlation mapping weight between the two cracking control nodes and the main control variable of the corresponding two cracking control nodes, a main control inversion function is constructed by combining a convolution kernel function, and the main control inversion function is embedded into the cracking process sequence connection between the corresponding cracking control nodes; Based on the contribution product of each main control variable to the cracking efficiency and energy conversion efficiency, a feedback regulation factor is constructed and embedded into the feedback connection between each cracking control node and the feedback regulation node, thereby obtaining a hierarchical and hierarchical cracking feedback control node chain.

7. The microwave-based dual-stage heating process for in-situ fractionation and pyrolysis of carbon fiber composites as claimed in claim 6, wherein, The construction steps of the hierarchical and hierarchical cracking feedback control node chain further include: Based on the hierarchical and hierarchical cracking feedback control node chain and the deep control parameter search model configured by the control parameter node network, a simulation algorithm, a preset optimal feedback control function, and cracking efficiency threshold and energy conversion efficiency threshold are trained to obtain a trained hierarchical and hierarchical cracking feedback control node chain. The optimal feedback control function is constructed by combining the feedback regulation factor in the feedback column, the main control variable, the splitting efficiency and the energy conversion efficiency, and the maximum value of the splitting efficiency and the energy conversion efficiency.

8. The microwave-based dual-stage heating process for in-situ fractionation and pyrolysis of carbon fiber composites as claimed in claim 7, wherein, The cracking control node set includes an analytical modeling node, a feeding node, a first preheating node, a second preheating node, a frequency splitting cracking node, an ionization combustion control node, and a preheating recovery control node. The working process of the hierarchical and hierarchical cracking feedback control node chain includes: A three-dimensional density model of the target object to be cracked is constructed by the analytical modeling node based on the built-in three-dimensional modeling algorithm, the geometric parameter information of the target object to be cracked, the density of each material in the composite material, and the optimal hierarchical cracking parameters; the optimal hierarchical cracking parameters are the double-frequency frequency values corresponding to the density of each density material in the composite material that make the cracking efficiency reach the maximum at the optimal second preheating target temperature. Based on the three-dimensional density model of the target to be cracked, the first preheating temperature interval and the second temperature interval are combined, and through the pre-trained deep analysis model, the feeding speed between the feeding node and the first preheating node, the density and composite density of each material, and the first optimal preheating temperature parameter, the first residence time length, the first preheating target temperature, the first inverse correlation mapping between the first preheating node and the second preheating node, the first preheating target temperature, the density and composite density of each material, the feeding rate, the second residence time length, the second preheating temperature parameter, the preset second preheating target temperature, and the second inverse correlation mapping between the second preheating node and the frequency splitting cracking node, the second preheating target temperature, the optimal cracking parameter, and the third residence time length are obtained. The corresponding third inverse correlation mapping, the feeding rate corresponding to the feeding node, the residence time length and the heating rate corresponding to the first preheating node and the second preheating node, and the cracking dual-frequency frequency value and the third residence time length corresponding to the frequency splitting cracking node are obtained.

9. The microwave-based dual-stage heating process for in-situ fractionation and pyrolysis of carbon fiber composites as claimed in claim 8, wherein, The working process of the hierarchical and hierarchical cracking feedback control node chain includes: According to the obtained feeding rate, the first fuzzy control sub-model built in the feeding node is combined to obtain the feeding control instruction for feeding, and based on the residence time length and the heating rate corresponding to the first preheating node and the second preheating node, the second fuzzy control sub-model is combined to perform first preheating and second preheating, and the waste heat conversion efficiency is monitored in real time based on the preset preheating conversion curve; When the corresponding second preheating target temperature is reached, the dual-frequency cracking control is performed through the third fuzzy sub-model combined with the cracking dual-frequency frequency value and the third residence time length, and the cracking efficiency is monitored in real time; Based on the deviation value of the real-time monitored waste heat conversion efficiency and cracking efficiency and the corresponding waste heat conversion efficiency threshold and cracking efficiency threshold, the fourth inverse correlation mapping constructed by the energy conversion efficiency between the frequency splitting cracking node and the ionization combustion control node, the natural gas input quantity and the ionization control coefficient is combined with the fourth fuzzy sub-model to generate the ionization control instruction to ionize the generated waste gas.

10. The microwave-based dual-stage heating process for in-situ fractionation and pyrolysis of carbon fiber composites as claimed in claim 9, wherein, The working process of the hierarchical and hierarchical cracking feedback control node chain includes: According to the fifth inverse correlation mapping and the sixth inverse correlation mapping respectively constructed by the energy conversion efficiency between the ionization combustion control node and the preheating recovery adjustment node, the feeding rate, the preheating heating rate and the corresponding first or second residence time length in the preheating stage, and the fifth fuzzy sub-model, the compensation control instruction for the first or second preheating node is obtained. The preheating heating rate of the first or second preheating node is compensated according to the density and composite density of each material of the target to be cracked; At the same time, based on the monitoring control node, the working parameters of the feeding node, the first preheating node, the second preheating node, the frequency splitting cracking node, the ionization combustion control node, and the preheating recovery adjustment node are monitored, and the monitoring parameters are input to the feedback control node for cracking efficiency and energy conversion efficiency evaluation, and the cracking efficiency evaluation value and the energy conversion efficiency evaluation value are obtained. Based on the cracking efficiency evaluation value and the energy consumption conversion efficiency evaluation value and the cracking efficiency threshold value and the energy consumption conversion efficiency threshold value, a cracking efficiency deviation value and an energy consumption conversion efficiency deviation value are obtained; Based on the cracking efficiency deviation value and the energy consumption conversion efficiency deviation value, feedback is fed back to the hierarchical and graded cracking feedback control node chain through a feedback connection, optimal deviation control parameter synchronous search is performed through a master control inversion function combined with a control parameter node network and a configured depth control parameter search model, and the obtained optimal deviation control parameter is fed back to a corresponding cracking control node. Meanwhile, the optimal master control variable corresponding to the adjacent two cracking control nodes is taken as a constraint, a cracking deviation adjustment instruction is generated combined with the fuzzy control model of the corresponding cracking control node, so that the cracking efficiency and the energy consumption conversion efficiency meet the corresponding condition threshold value in real time.

Images