Two-way intelligent design method for high-performance all-organic composites

Abstract

A bidirectional intelligent design method of high-performance all-organic composite material, reinforcement learning trains parameters between organic semiconductor and polymer matrix to obtain design parameters of optimal all-organic composite material performance, which provides guidance for designing high-performance all-organic composite material, and the beneficial effects are: through reinforcement learning training, high energy storage density, high charge and discharge efficiency, and high breakdown field strength all-organic composite material are obtained. Reinforcement learning trains parameters between organic semiconductor and polymer matrix to obtain design parameters of optimal all-organic composite material performance, which provides guidance for designing high-performance all-organic composite material.

Description

Technical Field

[0001] This invention relates to the field of high energy density thin-film capacitors, and in particular to a bidirectional intelligent design method for high-performance all-organic composite materials. Background Technology

[0002] All-organic composite materials, due to their advantages such as lightweight, good processability, and strong chemical stability, have shown broad application potential in many fields, especially in high-tech fields such as energy storage, electronic and optoelectronic devices, automobiles, and aerospace. All-organic composite materials typically consist of a polymer matrix and organic fillers. By optimizing the interactions between the components in the composite material, its mechanical properties, thermal stability, and electrical properties can be significantly improved.

[0003] However, traditional materials design methods often rely on experience and trial and error, which is not only time-consuming and costly, but also makes it difficult to accurately predict the performance of materials under specific conditions. Summary of the Invention

[0004] The purpose of this invention is to solve the above-mentioned problems by designing a two-way intelligent design method for high-performance all-organic composite materials. The specific design scheme is as follows: A bidirectional intelligent design method for high-performance all-organic composite materials utilizes reinforcement learning to continuously train the parameters between the organic semiconductor and the polymer matrix, thereby obtaining the optimal design parameters for the performance of the all-organic composite material. This provides guidance for the design of high-performance all-organic composite materials. The design steps are as follows: Step 1. Initialize the parameters of the organic semiconductor material and the polymer matrix; Step 2. Calculate the potential distribution inside the all-organic composite material using a bipolar injection charge conduction model; Step 3. Use high-throughput phase-field simulation to depict the breakdown evolution of all-organic composite materials; Step 4. Construct a database of the breakdown properties of all-organic composite materials; Step 5. Construct a nonlinear expression for the feature data and output; Step 6. Normalize the feature data; Step 7. Divide the feature data into training and test sets; Step 8. Initialize the weights and thresholds of the training set; Step 9. Train the training set data using a reinforcement learning network; Step 10. Verify the nonlinear expression obtained from the training set and the test set using reinforcement learning; Step 11. Determine the optimal nonlinear expression based on the training results of reinforcement learning; Step 12.1. If the training result in step 11 does not yield the optimal nonlinear expression, return to step 9. Step 12.2. Use the optimal nonlinear expression obtained in step 11 to predict the performance of organic semiconductor-doped all-organic composites; Step 13. Guide the preparation of an all-organic composite material with excellent energy storage density, high charge and discharge efficiency, and high breakdown field strength in the experiment.

[0005] In step 1, the parameters of the polymer matrix include the energy density, energy conversion efficiency, and breakdown field strength of the all-organic composite material. All-organic composite materials are prepared by doping a certain proportion of organic molecular semiconductors into a polymer matrix to improve the energy density, energy conversion efficiency, and breakdown field strength of the polymer matrix, thereby achieving high-performance all-organic composite materials. In the field of polar dielectric energy storage, energy storage density, energy conversion efficiency, and breakdown field strength of all-organic composite dielectric materials are commonly used to analyze the material performance.

[0006] In step 2, trap depth can be introduced by doping organic molecular semiconductors with extremely low content and strong electron affinity to capture freely moving charge carriers within the polymer.

[0007] By creating a database of dielectric constants for organic molecular semiconductors, the trapping radii of different organic semiconductors in dielectric polymers can be analyzed. Compared to dielectric polymers, organic molecular semiconductors exhibit stronger electron affinity; electrons injected by electrodes and thermally excited can be trapped by the strong electrostatic attraction of organic molecular semiconductors. The formation of molecular traps of a certain depth between the organic molecular semiconductor and the polymer can prevent the escape of trapped electrons, thereby suppressing conductivity loss under high temperatures and high electric fields.

[0008] The electric field distribution is calculated by analyzing the potential distribution within the all-organic composite material. Based on an improved bipolar charge injection and transport model, the calculated electric field distribution is used in a high-throughput phase-field model of electrothermal breakdown to simulate the evolution of the breakdown stage. This invention utilizes a high-throughput phase-field simulation method to analyze the dielectric breakdown mechanism in polymer / organic molecular semiconductor all-organic polymer composite materials at high temperatures.

[0009] In step 2, the electric field distribution is calculated by calculating the potential distribution inside the all-organic composite material. Based on the improved bipolar charge injection and transport model, the calculated electric field distribution is used in the high-throughput phase field model of electrothermal breakdown to simulate the evolution of the breakdown stage. This invention uses the high-throughput phase field simulation method to analyze the dielectric breakdown mechanism in polymer / organic molecular semiconductor all-organic polymer composite materials at high temperature.

[0010] In step 4, by coupling the bipolar injection charge conduction model and phase field simulation, a database of the breakdown performance of all-organic composite materials is obtained, which includes organic molecular semiconductors with different contents, different types of organic molecular semiconductors, and the trap depth formed by organic molecular semiconductors and polymer matrix.

[0011] In step 5, an intelligent algorithm is used to obtain the all-organic composite material with optimal performance, and to obtain nonlinear expressions for the doping content of organic molecular semiconductors, the type of organic molecular semiconductors, and the depth of the formed traps on the energy storage density, breakdown field strength, and charge-discharge efficiency of the all-organic composite material.

[0012] This invention uses reinforcement learning to obtain optimal material parameter design, resulting in a high-performance all-organic composite material. In step 9, The reinforcement learning agent is trained with the goal of maximizing cumulative rewards. t The environmental state at any given time, To reinforce the actions that the learning agent can perform, Indicates that the intelligent agent is in t Actions performed at all times It is the reward value for reinforcement learning. Indicates that the intelligent agent is in t The reward value obtained from performing actions at any time. To reinforce learning, the state transition probability represents the action performed by the agent. From state Transition to the next state The probability, The reinforcement learning agent is trained to maximize cumulative reward, where the cumulative reward at time t is represented as: (18) in, for t The reward value obtained at each moment; As a discount factor, The importance of immediate feedback diminishes over time. Indicates that the intelligent agent is in t +1 is the reward value obtained by performing the action at time 1; Indicates that the intelligent agent is in t +2 The reward value obtained by performing the action at time 2; Indicates that the intelligent agent is in t +3 The reward value gained from performing the action at time 3; Indicates that the intelligent agent is in t + k +1 is the reward value obtained by performing the action at time 1. State value function The state value is defined as follows to evaluate whether a state is good or bad: (19) in, This indicates that the agent acts according to the policy. The cumulative reward obtained from state s until the end of the interaction. In each action of reinforcement learning, this model requires a change from the set. Get a current state s With action a Update from arrive The state of the model, the model's predicted value for the test sample, and the label the model chooses for the next step are determined by... and Decide, Optimal reward value in reinforcement learning The definition is as follows: (20) in, The optimal reward value in the composite material design scenario of this invention; These are policy variables representing the environmental state. for t Reward value at each time step. Behavioral value function The following definition is used to evaluate the quality of an action: (twenty one) in, For the agent according to the policy In state s Perform a series of actions a The accumulated rewards obtained until the interaction ends; for t The action performed at any given moment; for t time s Actions performed in a state a The reward value obtained, The maximum reward corresponds to for: (twenty two) in, For the agent according to the policy In state s Perform a series of actions a The maximum cumulative reward obtained until the interaction ends. Update using Bellman's formula Q value: (twenty three) in, For strategy i +1 in state s Perform a series of actions a Update the cumulative rewards obtained through interaction; For strategy i In state s +1 performs a series of actions a +1, the cumulative reward received upon completion of the update interaction; For strategy i In state s +1 performs a series of actions a +1, update the maximum cumulative reward value obtained after the interaction ends. Among them, when Gradually converge to the optimal behavior-value function to obtain the optimal policy: (twenty four) in, The optimal strategy; arg() is the inverse trigonometric function; To perform an action under strategy a a The maximum cumulative reward value obtained at that time High energy density, high charge / discharge efficiency, and high breakdown field strength of all-organic composite materials were obtained through reinforcement learning training. Reinforcement learning continuously trains the parameters between the organic semiconductor and the polymer matrix to obtain the optimal design parameters for the performance of the all-organic composite material, providing guidance for the design of high-performance all-organic composite materials.

[0013] The bidirectional intelligent design method for high-performance all-organic composite materials obtained through the above-described technical solution of the present invention has the following advantages: High energy density, high charge / discharge efficiency, and high breakdown field strength of all-organic composite materials were obtained through reinforcement learning training. Reinforcement learning continuously trains the parameters between the organic semiconductor and the polymer matrix to obtain the optimal design parameters for the performance of the all-organic composite material, providing guidance for the design of high-performance all-organic composite materials. Attached Figure Description

[0014] Figure 1 This is a flowchart of the bidirectional intelligent design method for high-performance all-organic composite materials described in this invention; Figure 2 This is a data-driven quantum deep neural network structure diagram of the bidirectional intelligent design method for high-performance all-organic composite materials described in this invention; Figure 3 This is a diagram illustrating the reinforcement learning optimization process of the bidirectional intelligent design method for high-performance all-organic composite materials described in this invention. Detailed Implementation

[0015] The present invention will now be described in detail with reference to the accompanying drawings.

[0016] A bidirectional intelligent design method for high-performance all-organic composite materials utilizes reinforcement learning to continuously train the parameters between the organic semiconductor and the polymer matrix, thereby obtaining the optimal design parameters for the performance of the all-organic composite material. This provides guidance for the design of high-performance all-organic composite materials. The design steps are as follows: Step 1. Initialize the parameters of the organic semiconductor material and the polymer matrix; Step 2. Calculate the potential distribution inside the all-organic composite material using a bipolar injection charge conduction model; Step 3. Use high-throughput phase-field simulation to depict the breakdown evolution of all-organic composite materials; Step 4. Construct a database of the breakdown properties of all-organic composite materials; Step 5. Construct a nonlinear expression for the feature data and output; Step 6. Normalize the feature data; Step 7. Divide the feature data into training and test sets; Step 8. Initialize the weights and thresholds of the training set; Step 9. Train the training set data using a reinforcement learning network; Step 10. Verify the nonlinear expression obtained from the training set and the test set using reinforcement learning; Step 11. Determine the optimal nonlinear expression based on the training results of reinforcement learning; Step 12.1. If the training result in step 11 does not yield the optimal nonlinear expression, return to step 9. Step 12.2. Use the optimal nonlinear expression obtained in step 11 to predict the performance of organic semiconductor-doped all-organic composites; Step 13. Guide the preparation of an all-organic composite material with excellent energy storage density, high charge and discharge efficiency, and high breakdown field strength in the experiment.

[0017] In step 1, the parameters of the polymer matrix include the energy density, energy conversion efficiency, and breakdown field strength of the all-organic composite material. All-organic composite materials are prepared by doping a certain proportion of organic molecular semiconductors into a polymer matrix to improve the energy density, energy conversion efficiency, and breakdown field strength of the polymer matrix, thereby achieving high-performance all-organic composite materials. In the field of polar dielectric energy storage, energy storage density, energy conversion efficiency, and breakdown field strength of all-organic composite dielectric materials are commonly used to analyze the material performance.

[0018] In step 2, trap depth can be introduced by doping organic molecular semiconductors with extremely low content and strong electron affinity to capture freely moving charge carriers within the polymer.

[0019] By creating a database of dielectric constants for organic molecular semiconductors, the trapping radii of different organic semiconductors in dielectric polymers can be analyzed. Compared to dielectric polymers, organic molecular semiconductors exhibit stronger electron affinity; electrons injected by electrodes and thermally excited can be trapped by the strong electrostatic attraction of organic molecular semiconductors. The formation of molecular traps of a certain depth between the organic molecular semiconductor and the polymer can prevent the escape of trapped electrons, thereby suppressing conductivity loss under high temperatures and high electric fields.

[0020] The electric field distribution is calculated by analyzing the potential distribution within the all-organic composite material. Based on an improved bipolar charge injection and transport model, the calculated electric field distribution is used in a high-throughput phase-field model of electrothermal breakdown to simulate the evolution of the breakdown stage. This invention utilizes a high-throughput phase-field simulation method to analyze the dielectric breakdown mechanism in polymer / organic molecular semiconductor all-organic polymer composite materials at high temperatures.

[0021] In step 2, the electric field distribution is calculated by calculating the potential distribution inside the all-organic composite material. Based on the improved bipolar charge injection and transport model, the calculated electric field distribution is used in the high-throughput phase field model of electrothermal breakdown to simulate the evolution of the breakdown stage. This invention uses the high-throughput phase field simulation method to analyze the dielectric breakdown mechanism in polymer / organic molecular semiconductor all-organic polymer composite materials at high temperature.

[0022] In step 4, by coupling the bipolar injection charge conduction model and phase field simulation, a database of the breakdown performance of all-organic composite materials is obtained, which includes organic molecular semiconductors with different contents, different types of organic molecular semiconductors, and the trap depth formed by organic molecular semiconductors and polymer matrix.

[0023] In step 5, an intelligent algorithm is used to obtain the all-organic composite material with optimal performance, and to obtain nonlinear expressions for the doping content of organic molecular semiconductors, the type of organic molecular semiconductors, and the depth of the formed traps on the energy storage density, breakdown field strength, and charge-discharge efficiency of the all-organic composite material.

[0024] This invention uses reinforcement learning to obtain optimal material parameter design, resulting in a high-performance all-organic composite material. In step 9, The reinforcement learning agent is trained with the goal of maximizing cumulative rewards. t The environmental state at any given time, To reinforce the actions that the learning agent can perform, Indicates that the intelligent agent is in t Actions performed at all times It is the reward value for reinforcement learning. Indicates that the intelligent agent is in t The reward value obtained from performing actions at any time. To reinforce learning, the state transition probability represents the action performed by the agent. From state Transition to the next state The probability, The reinforcement learning agent is trained to maximize cumulative reward, where the cumulative reward at time t is represented as: (18) in, for t The reward value obtained at each moment; As a discount factor, The importance of immediate feedback diminishes over time. Indicates that the intelligent agent is in t +1 is the reward value obtained by performing the action at time 1; Indicates that the intelligent agent is in t +2 The reward value obtained by performing the action at time 2; Indicates that the intelligent agent is in t +3 The reward value gained from performing the action at time 3; Indicates that the intelligent agent is in t + k +1 is the reward value obtained by performing the action at time 1. State value function The state value is defined as follows to evaluate whether a state is good or bad: (19) in, This indicates that the agent acts according to the policy. The cumulative reward obtained from state s until the end of the interaction. In each action of reinforcement learning, this model requires a change from the set. Get a current state s With action a Update from arrive The state of the model, the model's predicted value for the test sample, and the label the model chooses for the next step are determined by... and Decide, Optimal reward value in reinforcement learning The definition is as follows: (20) in, The optimal reward value in the composite material design scenario of this invention; These are policy variables representing the environmental state. for t Reward value at each time step. Behavioral value function The following definition is used to evaluate the quality of an action: (twenty one) in, For the agent according to the policy In state s Perform a series of actions a The accumulated rewards obtained until the interaction ends; for t The action performed at any given moment; for t time s Actions performed in a state a The reward value obtained, The maximum reward corresponds to for: (twenty two) in, For the agent according to the policy In state s Perform a series of actions a The maximum cumulative reward obtained until the interaction ends. Update using Bellman's formula Q value: (twenty three) in, For strategy i +1 in state s Perform a series of actions a Update the cumulative rewards obtained through interaction; For strategy i In state s +1 performs a series of actions a +1, the cumulative reward received upon completion of the update interaction; For strategy i In state s +1 performs a series of actions a +1, update the maximum cumulative reward value obtained after the interaction ends. Among them, when Gradually converge to the optimal behavior-value function to obtain the optimal policy: (twenty four) in, The optimal strategy; arg() is the inverse trigonometric function; To perform an action under strategy a a The maximum cumulative reward value obtained at that time High energy density, high charge / discharge efficiency, and high breakdown field strength of all-organic composite materials were obtained through reinforcement learning training. Reinforcement learning continuously trains the parameters between the organic semiconductor and the polymer matrix to obtain the optimal design parameters for the performance of the all-organic composite material, providing guidance for the design of high-performance all-organic composite materials.

[0025] Figure 1 This is a flowchart of the method of the present invention. First, the material properties of the organic semiconductor material and the polymer matrix are parameterized. A bipolar injection charge conduction model is used to calculate the potential distribution inside the all-organic composite material, providing input for the next step of high-throughput phase-field simulation. The calculated potential distribution and material parameters are used to construct a high-throughput phase-field model to simulate the internal breakdown evolution of the all-organic composite material, obtaining a database of the breakdown performance of the organic semiconductor material, the polymer matrix material, and the all-organic composite material. A nonlinear expression for the feature data and output is constructed. The feature data is normalized and divided into training and testing sets. Initial weights and thresholds are set for the reinforcement learning network. The network is continuously trained on the data to obtain the nonlinear expression after reinforcement learning training. The training results of the reinforcement learning network are verified using the test set to determine the optimal nonlinear expression. The nonlinear expression guides the material parameters of the organic semiconductor dopant and the polymer matrix during the preparation of the all-organic composite, and predicts the energy storage density, charge-discharge efficiency, and breakdown field strength of the all-organic composite.

[0026] Figure 3 This is a diagram illustrating the reinforcement learning optimization process of the method described in this invention. Reinforcement learning utilizes an agent to continuously optimize its structure based on the objective function during online training of feature data, obtaining the optimal objective function and outputting the optimal function's action and nonlinear expression. Reinforcement learning continuously optimizes based on its own parameters and input feature data, resulting in an output with both optimized and globally optimal performance. By selecting the nonlinear expression and feature parameters under the globally optimal reinforcement learning training result, reinforcement learning training yields a fully organic composite material with high energy density, high charge / discharge efficiency, and high breakdown field strength. By continuously training the parameters between the organic semiconductor and the polymer matrix, the design parameters for the optimal performance of the fully organic composite material are obtained, providing guidance for designing high-performance fully organic composite materials.

[0027] Example 1

[0028] The energy storage density of all-organic composite materials can be expressed by the linear dielectric equation as follows: (1) in, The energy storage density of the all-organic composite material; The dielectric constant of vacuum is 8.85 × 10⁻⁶. -12F / m; The relative permittivity of the all-organic composite material; This represents the breakdown strength of the all-organic composite material.

[0029] Example 2 A bipolar charge injection and conduction model based on the Schottky injection mechanism was used to study the charge conduction process inside the all-organic composite material. The leakage current density equation at the interface between the metal electrode and the polymer / organic molecular semiconductor all-organic polymer is as follows: (2) in, Richardson's coefficient, For temperature; An intrinsic injection barrier between the metal electrode and the all-organic composite material; Boltzmann's constant; e It is the elementary charge; E The electric field strength at the anode and cathode; is the relative permittivity of the all-organic composite material.

[0030] Because doped organic molecular semiconductors possess strong electron affinity, they can form trap depths at the polymer-organic molecular semiconductor interface to capture freely moving charge carriers in the polymer matrix. Inspired by Langevin theory, carrier recombination is an inevitable phenomenon within polymers. "Coulomb binding" measures the strength of the attraction exerted on charges confined within the trap. Simply put, when the trap radius is longer, charge carriers are more likely to recombine at the recombination center or trap. The Coulomb trap radius is: (3) in, R c The radius of the coulomb capture; It is the relative permittivity in vacuum; is the relative permittivity of organic molecular semiconductors; T For temperature.

[0031] Example 3 The depth of the molecular trap formed by the polymer matrix and the organic molecular semiconductor is: (4) in, The depth of the molecular trap; The electron affinity of organic molecular semiconductors; It represents the electron affinity of the polymer matrix.

[0032] The electron transport process within a polymer / organic molecular semiconductor all-organic composite material, considering the molecular traps and Coulomb trapping radii formed by the introduction of organic molecular semiconductors, can be represented as follows: (5) in, The density of free electrons; The mobility of free electrons; The diffusion factor of electrons; The radius of the coulomb capture; The recombination coefficient of free electrons and free holes; The recombination coefficient between free electrons and trapped holes; The density of captured holes; The electron capture coefficient; The density of the trapped electrons; This represents the maximum trap density of electrons; The electron escape coefficient; The depth of the electronic trap.

[0033] Hole transport within polymer / organic molecular semiconductor all-organic composite materials is represented as: (6) in, The density of free holes; The free hole mobility; The diffusion factor of holes; The recombination coefficient of trapped electrons and free holes; The hole capture coefficient; The maximum trap density of holes; The escape coefficient of the hole; The depth of the hole trap.

[0034] The trapped electrons in the polymer / organic molecular semiconductor all-organic composite material are represented as: (7) in, is the recombination coefficient of the trapped electrons and trapped holes.

[0035] The trapped holes in the polymer / organic molecular semiconductor all-organic composite material are represented as: (8) After calculating the charge transport within the all-organic composite material, the potential distribution can be obtained from the distributions of free and trapped charge carrier densities as follows: (9) in, For the Laplace operator; The relative permittivity of the all-organic composite material; It represents the electric potential.

[0036] Example 4 A continuous phase transition variable was introduced into the phase-field simulation. To describe the electrothermal evolution of the breakdown stage in polymer / organic molecular semiconductor all-organic dielectric composite materials, where and These represent the non-breakdown and breakdown stages, respectively. When organic molecular semiconductors are doped into dielectric polymers, the free energy contributed by the synergistic effects of phase separation, interfaces, thermal fields, and electric fields in a non-homogeneous medium is... Represented as: (10) in, The free energy density for phase separation; The gradient energy density; It is the electrostatic energy density; This is the thermal energy density.

[0037] The free energy density of phase separation is: (11) in, This is a phase separation barrier.

[0038] The gradient energy density is: (12) in, The gradient energy coefficient.

[0039] The electrostatic energy density is: (13) The thermal energy density is: (14) in, Electrical conductivity; The period during which the applied electric field is applied.

[0040] In this invention, the electrical conductivity of the all-organic composite material is: (15) in, q Unit charge; This represents the migration rate.

[0041] The formula for heat energy is: (16) The breakdown stage evolution of polymer / organic molecular semiconductor all-organic composite materials was simulated using a modified Allen-Cahn equation. The modified Allen-Cahn equation is expressed as: (17) in, This is a kinetic coefficient related to the interface migration rate, with a value of 1. ; Let be the Heaviside unit step function for electrostatic energy and Joule thermal energy.

[0042] Example 5 Figure 2 This is a structural diagram of the reinforcement learning method of this invention. The reinforcement learning structure includes 3 fingerprint features, 4 interactions, and 12 prototype functions. Three feature parameters of organic semiconductor materials are used as fingerprint features, and four interactions between the three fingerprint features are constructed using random combinations. The prototype functions employ... By continuously optimizing using reinforcement learning, the best-performing nonlinear expression is obtained.

[0043] The above technical solutions only embody the preferred technical solutions of the present invention. Any modifications that may be made by those skilled in the art to certain parts thereof embody the principles of the present invention and fall within the protection scope of the present invention.

Claims

1. A two-way intelligent design method for high-performance all-organic composite materials, wherein reinforcement learning continuously trains the parameters between the organic semiconductor and the polymer matrix to obtain the design parameters for the optimal performance of the all-organic composite material, providing guidance for the design of high-performance all-organic composite materials, characterized in that, The design steps are as follows: Step 1. Initialize the parameters of the organic semiconductor material and the polymer matrix; Step 2. Calculate the potential distribution inside the all-organic composite material using a bipolar injection charge conduction model; Step 3. Use high-throughput phase-field simulation to depict the breakdown evolution of all-organic composite materials; Step 4. Construct a database of the breakdown properties of all-organic composite materials; Step 5. Construct a nonlinear expression for the feature data and output; Step 6. Normalize the feature data; Step 7. Divide the feature data into training and test sets; Step 8. Initialize the weights and thresholds of the training set; Step 9. Train the training set data using a reinforcement learning network; Step 10. Verify the nonlinear expression obtained from the training set and the test set using reinforcement learning; Step 11. Determine the optimal nonlinear expression based on the training results of reinforcement learning; Step 12.

1. If the training result in step 11 does not yield the optimal nonlinear expression, return to step 9. Step 12.

2. Use the optimal nonlinear expression obtained in step 11 to predict the performance of organic semiconductor-doped all-organic composites; Step 13. Guide the preparation of an all-organic composite material with excellent energy storage density, breakdown field strength, and charge / discharge efficiency in the experiment. In step 4, by coupling the bipolar injection charge conduction model and phase-field simulation, the energy storage density, breakdown field strength, and charge-discharge efficiency of the all-organic composite material are obtained for different contents of organic molecular semiconductors, different types of organic molecular semiconductors, and the trap depth formed by organic molecular semiconductors and polymer matrix, and summarized into a breakdown performance database. In step 5, nonlinear expressions are constructed for the doping content of the organic molecular semiconductor, the type of organic molecular semiconductor, the depth of the formed trap, and the energy storage density, breakdown field strength, and charge / discharge efficiency of the all-organic composite material. This invention uses reinforcement learning to obtain optimal material parameter design, resulting in a high-performance all-organic composite material. In step 9, The reinforcement learning agent is trained with the goal of maximizing cumulative rewards. t The environmental state at any given time, To reinforce the actions that the learning agent can perform, Indicates that the intelligent agent is in t Actions performed at all times It is the reward value for reinforcement learning. Indicates that the intelligent agent is in t The reward value obtained from performing actions at any time. To reinforce learning, the state transition probability represents the action performed by the agent. From state Transition to the next state The probability, The reinforcement learning agent is trained to maximize cumulative reward, where the cumulative reward at time t is represented as: (18) in, for t The reward value obtained at each moment; As a discount factor, The importance of immediate feedback diminishes over time. Indicates that the intelligent agent is in t +1 is the reward value obtained by performing the action at time 1; Indicates that the intelligent agent is in t +2 The reward value obtained by performing the action at time 2; Indicates that the intelligent agent is in t +3 The reward value gained from performing the action at time 3; Indicates that the intelligent agent is in t + k +1 is the reward value obtained by performing the action at time 1. State value function The state value is defined as follows to evaluate whether a state is good or bad: (19) in, This indicates that the agent acts according to the policy. The cumulative reward obtained from state s until the end of the interaction. In each action of reinforcement learning, the model requires obtaining a current state from the set S. s Update from arrive The state of the model, the model's predicted value for the test sample, and the label the model chooses for the next step are determined by... and Decide, Optimal reward value in reinforcement learning The definition is as follows: (20) in, This represents the optimal reward value in the composite material design scenario of this invention. These are policy variables representing the environmental state. for t Reward value at each time step. Behavioral value function Used to evaluate the quality of an action, defined as follows: (21) in, For the agent according to the policy In state s Perform a series of actions a The cumulative rewards obtained until the interaction ends; for t The action performed at any given moment; for t time s Actions performed in a state a The reward value obtained, The maximum reward corresponds to for: (22) in, For the agent according to the policy In state s Perform a series of actions a The maximum cumulative reward obtained until the interaction ends. Update using Bellman's formula Q value: (23) in, For strategy i +1 in state s Perform a series of actions a Update the cumulative rewards obtained through interaction; For strategy i In state s +1 performs a series of actions a +1, the cumulative reward received upon completion of the update interaction; For strategy i In state s +1 performs a series of actions a +1, update the maximum cumulative reward value obtained after the interaction ends. Among them, when Gradually converge to the optimal behavior-value function to obtain the optimal policy: (24) in, The optimal strategy; arg() is the inverse trigonometric function; To perform an action under strategy a a The maximum cumulative reward value obtained at that time.

2. The bidirectional intelligent design method for high-performance all-organic composite materials according to claim 1, characterized in that, In step 1, the parameters of the polymer matrix include the energy storage density, breakdown field strength, and charge / discharge efficiency of the all-organic composite material.

3. The bidirectional intelligent design method for high-performance all-organic composite materials according to claim 1, characterized in that, In step 2, free-moving charge carriers within the polymer are captured by doping organic molecular semiconductors.

4. The bidirectional intelligent design method for high-performance all-organic composite materials according to claim 1, characterized in that, By calculating the potential distribution inside the all-organic composite material, the electric field distribution is calculated. Based on the improved bipolar charge injection and transport model, the calculated electric field distribution is used in the high-throughput phase field model of electrothermal breakdown to simulate the evolution of the breakdown stage. The high-throughput phase field simulation method is used to analyze the dielectric breakdown mechanism in polymer / organic molecular semiconductor all-organic polymer composite materials at high temperature.

5. The two-way intelligent design method for high-performance all-organic composite materials according to claim 1, characterized in that, In step 2, the calculated electric field distribution, based on the improved bipolar charge injection and transport model, is used in the high-flux phase field model of electrothermal breakdown to simulate the evolution of the breakdown stage.

Images