Additive molecule screening design method for performance optimization of ester-based insulating oil

Abstract

The invention discloses an additive molecule screening design method for optimizing the performance of ester-based insulating oil, and relates to the technical field of electrical insulating materials. The invention provides an additive molecule design method combined with a deep molecule generation model, aiming at solving the problems that the breakdown voltage performance is insufficient due to the fact that ester-based insulating oil is easy to generate rapid streamer under high-strength lightning impulse, and time and labor are consumed when an appropriate additive is found through an experiment. The method comprises the following steps: acquiring molecular data with target electronic attributes; taking the fragment of the molecular structure as the input of the model, and designing a new additive molecular structure by utilizing the molecular generation model; performing molecular screening according to target characteristics; and the performance of the screened molecules is verified through streamer discharge simulation. According to the method, generative design of ester-based insulating oil additive molecules is realized by using the deep autoregressive neural network, the molecular screening efficiency is remarkably improved, and a new technical route is provided for improving the breakdown voltage performance of the ester-based insulating oil.

Description

Technical Field

[0001] The invention relates to an additive molecule screening design method for optimizing the performance of ester-based insulating oil, and belongs to the technical field of electrical insulation materials. Background Art

[0002] With the rapid development of modern power systems towards ultra-high voltage (UHV) and extra-high voltage (EHV), power equipment has put forward increasingly stringent requirements on the electrical properties of insulating materials. As a vital insulating medium inside power transformers, the performance of insulating oil is directly related to the safe operation and stability of transformers. Due to the advantages of ester-based insulating oil such as easy degradation, high ignition point, and anti-oxidation, the solution of replacing the traditional insulating medium (mineral oil) in power transformers with ester-based insulating oil has been accepted by more and more researchers, and this solution has achieved certain application effects in engineering applications.

[0003] However, the insulation performance of ester-based insulating oil shows obvious deficiencies when facing high-intensity lightning impulse voltage and severe transient overvoltage. Especially under the conditions of non-uniform electric field and long oil gap, its breakdown voltage is significantly lower than that of mineral oil, which seriously restricts the large-scale promotion of ester-based insulating oil in large power transformers. In order to solve this key problem, it is first necessary to combine the existing streamer discharge research of ester-based insulating oil to deeply understand the lightning impulse breakdown mechanism of ester-based insulating oil.

[0004] At present, according to some related researches in experiments and simulations, it can be found that ester-based insulating oil is more likely to produce fast streamers during streamer discharge under the conditions of non-uniform electric field and long oil gap. These fast streamers have been confirmed to be the key factors leading to insulation failure of power transformers. In order to overcome this problem, it is particularly important to find methods and mechanisms to inhibit the generation of fast streamers of ester-based insulating oil.

[0005] On this basis, some studies have further found that the photoionization mechanism is an important reason for the transformation of slow streamers to fast streamers. Studies have found that the photoionization process can be effectively suppressed by adding molecular additives with low ionization energy (IP) or low first excited state (S1) energy. Such molecules can preferentially absorb photon energy, thereby suppressing the photoionization process of surrounding ester-based insulating oil molecules and ultimately curbing the formation of fast streamers. Early studies have confirmed that the addition of ultraviolet light absorbers can indeed improve the breakdown voltage performance of ester-based insulating oil, which points the way to solving the problem. Therefore, there is an urgent need to find a more efficient way to suppress the photoionization mechanism during streamer discharge, thereby improving the lightning impulse breakdown voltage of ester-based insulating oil.

[0006] However, among the numerous possible additives, how to efficiently find the optimal additive molecules and determine which molecular structural features and functional groups are most effective in suppressing the photoionization process remains a major challenge.

[0007] Traditional molecular screening methods mainly rely on trial and error. Researchers need to synthesize and test different candidate molecules one by one. This method is not only time-consuming, labor-intensive and costly, but more importantly, it is difficult to systematically explore the vast molecular space. Even experienced researchers find it difficult to accurately predict which molecular structures will exhibit ideal low ionization energy or low excited state properties based on intuition alone. With the vigorous development of machine learning and artificial intelligence technologies in the field of materials science, deep molecular generation models provide an innovative solution to this problem.

[0008] This deep learning-based molecular generation model has unique advantages: first, it can automatically learn and understand the complex relationship between molecular structure and performance, especially the relationship between the geometric configuration and electronic structure of the molecule and its target characteristics; second, the model can generate candidate molecules with low IP or low S1 energy according to the set target attributes, greatly improving the efficiency of molecular design; more importantly, by analyzing the large amount of molecular data generated by the model, we can identify the structural features and functional groups that play a key role in the target performance, thereby providing more in-depth theoretical guidance for the synthesis of additives. This data-driven approach not only significantly accelerates the material development cycle, but also helps us establish a systematic understanding of the quantitative-structural property relationship of molecules, providing strong technical support for the development of a new generation of high-performance ester-based insulating oil additives. Summary of the invention

[0009] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for screening and designing additive molecules for optimizing the performance of ester-based insulating oil. The method uses machine learning technology to realize the screening and design of ester-based insulating oil additive molecules, thereby accelerating the material design process and improving the breakdown voltage of ester-based insulating oil.

[0010] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0011] The present invention provides an additive molecule screening and design method for optimizing the performance of ester-based insulating oil, comprising:

[0012] Acquire molecular fragment data with target characteristics;

[0013] Taking molecular fragment data as input, multiple complete additive molecular structures are output based on the molecular generation model;

[0014] Screening multiple complete additive molecular structures according to target characteristics to obtain a screened additive molecular structure;

[0015] The molecular structure of the screened additives was verified by streamer discharge simulation analysis to obtain the molecular structure of the additive for optimizing the performance of ester-based insulating oil.

[0016] Furthermore, the target characteristics include low first excited state energy and low ionization energy; the molecular fragment data includes structural information and attribute information of the molecular fragments, the structural information includes the atom type and atomic position contained in the molecular fragments, and the attribute information includes structural attributes and electronic attributes.

[0017] Furthermore, the molecular generation model uses a deep autoregressive neural network, which includes a first processing route and a second processing route, the first processing route is used to process the structure information to obtain an atomic feature vector, and the second processing route is used to process the attribute information to obtain an attribute feature vector;

[0018] The output ends of the first processing route and the second processing route are both connected to the input end of the atom type prediction layer, the atom type prediction layer is used to predict the next atom type according to the atom feature vector and the attribute feature vector to obtain the next atom type probability distribution function, the output end of the atom type prediction layer is connected to the atom type sampling layer and the atom type embedding layer in sequence, the atom type sampling layer is used to randomly sample according to the next atom type probability distribution function to obtain the next atom type, and the atom type embedding layer is used to convert the next atom type into an atom type feature vector;

[0019] The atomic feature vector and the atomic type feature vector are sequentially multiplied and added and then input into the atomic distance prediction layer, the atomic distance prediction layer is used to predict the next atomic position according to the atomic feature vector and the atomic type feature vector to obtain the next atomic distance probability distribution function, the output end of the atomic distance prediction layer is connected to the input end of the atomic position sampling layer, the atomic position sampling layer is used to randomly sample the next atomic position according to the next atomic distance probability distribution function to generate the next molecular fragment, and the next molecular fragment is input into the first processing route and the second processing route for continuous iteration until the complete additive molecular structure is obtained.

[0020] Further, the first processing route includes an embedding layer, an atom pair distance weighting layer and a feature processing layer connected in sequence, the embedding layer is used to extract atomic feature vectors from structural information, the atom pair distance weighting layer is used to perform weighted interaction on atomic feature vectors according to interatomic distances, and the feature processing layer is used to process the relationship between atoms and their surroundings, and includes a plurality of feature interaction layers and feature mixing layers connected in sequence;

[0021] The second processing route includes an attribute embedding module and a multi-layer perceptron connected in sequence, wherein the attribute embedding module is used to convert structural attributes and electronic attributes into structural attribute feature vectors and electronic attribute feature vectors respectively, and the multi-layer perceptron is used to integrate the structural attribute feature vectors and the electronic attribute feature vectors to obtain the attribute feature vectors.

[0022] Furthermore, the expression of the atom pair distance weighted layer is:

[0023] in, represents the atomic characteristics of the ith atom, represents the atomic characteristics of the jth atom adjacent to the ith atom, represents the distance between atom i and its neighboring atom j, represents the distance cutoff function, represents the radial basis function, Represents network filtering;

[0024] The expression of the feature processing layer is: ;

[0025] in, represents the atomic feature of the i-th atom after feature processing, , are weight matrices, , are all bias terms, represents the activation function, ;

[0026] The expression of the attribute embedding module is:

[0027] ;

[0028] in, represents the structural attribute feature vector, represents multi-layer perceptron processing, represents the concentration weight of atom type a in the existing additive molecule fragment, Embedding vector representing atom type a; represents the electronic property eigenvector, represents the electronic properties, represents the minimum value of electronic properties, represents the interval parameter of the radial basis function;

[0029] The expression of the multi-layer perceptron is: ;

[0030] in, represents the attribute feature vector;

[0031] Furthermore, the expression of the atom type prediction layer is:

[0032] ;

[0033] in, represents the weight factor of the j-th candidate atom type, represents the sum of atomic features in the generated molecular fragments, represents the probability distribution function of the next atom type;

[0034] The expression of the atomic position prediction layer is:

[0035]

[0036] in, represents the weight factor of the jth candidate atomic position, represents the atomic number of the next atom, express The embedding vector of represents the attribute feature vector, represents the probability distribution function of the next atom position.

[0037] Furthermore, the molecular generation model further includes an attention module, and an input end of the attention module is connected to an output end of the feature processing layer;

[0038] The attention module includes a multi-head attention unit, a self-attention unit and a cross-attention unit connected in sequence, wherein the multi-head attention unit is used to project the atomic feature head to multiple representation subspaces, the self-attention unit is used to enable each atom to obtain information of other atoms, and the cross-attention unit is used to associate information of different feature spaces to achieve multi-dimensional fusion of atomic features;

[0039] The expression of the attention module is: ;

[0040] in, represents the atomic feature of the i-th atom after feature processing, represents the atomic features of the i-th atom after being processed by the attention module, represents the query matrix, represents the key matrix, represents the value matrix, Represents the feature dimension.

[0041] Furthermore, the method further includes pre-training the molecular generation model, wherein the pre-training method includes: a. Obtain multiple known complete molecular structures; b. Split multiple known complete molecular structures into multiple molecular fragments and form them into training set data; c. Use the training set data as input to train the molecular generation model, output the predicted complete molecular structure, calculate the loss function, and use the loss function to adjust the model parameters; d. Repeat steps c to d until the loss function drops to a preset range or reaches a preset number of iterations, and obtain a pre-trained molecular generation model; The loss function uses the cross entropy of type prediction and distance prediction for predicting the complete molecular structure.

[0042] Furthermore, the screening of multiple additive molecular structures according to the target characteristics to obtain the screened additive molecular structures includes: The energy prediction model is used to predict the first excited state energy and ionization energy of multiple additive molecular structures to obtain prediction results; Coarsely screen multiple additive molecular structures based on the prediction results to retain effective additive molecular structures that meet the target properties; Perform quantum chemical calculations on the first excited state energy and ionization energy of the effective additive molecular structure that meets the target characteristics to obtain the calculation results; The retained additive molecular structures are finely screened according to the calculation results, and the additive molecular structures with target properties within a preset range are retained and used as the screened additive molecular structures.

[0043] Furthermore, the streamer discharge simulation analysis and verification of the screened additive molecular structure is performed to obtain the additive molecular structure for optimizing the performance of the ester-based insulating oil, including: The streamer discharge simulation analysis is performed to verify the molecular structure of the screened additives, and the streamer discharge simulation analysis results are obtained; According to the results of streamer discharge simulation analysis, the molecular structure with the best effect of slowing down the streamer velocity is retained as the molecular structure of the additive for optimizing the performance of ester-based insulating oil.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention uses machine learning technology to realize the automatic screening of additive molecules, which speeds up the material development process, avoids the tedious process of traditional experimental screening, and reduces the development cost. Through systematic and data-driven molecular design, it can effectively ensure that the obtained additives have excellent electrical properties, providing a green solution for material design that is more efficient and more practical.

[0046] The additive molecules screened by the method of the present invention can effectively improve the breakdown voltage of ester-based insulating oil, shorten the development cycle of ester-based insulating oil, and provide a more sustainable material selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic flow chart of a method for screening and designing additive molecules for optimizing the performance of ester-based insulating oil in one embodiment of the present invention;

[0048] Figure 2 A schematic diagram of the structure of a molecular generation model in an embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of a partially complete molecular structure used for pre-training in Example 1 of the present invention;

[0050] Figure 4 A schematic diagram of a partially complete additive molecular structure output by the molecular generation model in Example 1 of the present invention;

[0051] Figure 5 This is a schematic diagram comparing the simulation results of COMSOL simulation to verify the development of streamer discharge in Example 1 of the present invention. The left side is the simulation effect of the photoionization model after adding certain additive molecules, and the right side is the simulation effect of the original photoionization model. DETAILED DESCRIPTION

[0052] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0053] Example 1

[0054] like Figure 1 As shown, an embodiment of the present invention provides an additive molecule screening and design method for optimizing the performance of ester-based insulating oil, comprising the following steps:

[0055] First, obtain the existing additive molecular fragment data

[0056] In this example, quantum chemical property data of molecules are collected from databases such as QM8 and VERDE materials DB; published literature is obtained through platforms such as Web of Science, Google Scholar and IEEE, covering the latest research results related to the energy and ionization potential (IP) of the first excited state (S1) of molecules. In addition, quantum chemical calculation tools such as Gaussian and ORCA are used for molecular structure optimization and energy calculation; experimental methods such as ultraviolet-visible spectroscopy (UV-Vis) and photoelectron spectroscopy (PES) are used to obtain directly measured S1 and IP data.

[0057] The Atom Simulation Environment (ASE) database is established through a multi-channel data collection process. The Atom Simulation Environment database is specifically used to store key information of molecular structures, such as atom types, atomic coordinates, and related physical and electronic properties, providing basic input data for molecular generation models.

[0058] The existing additive molecular fragment data includes structural information and property information of the existing molecular fragments. The structural information includes the atom types and atomic positions contained in the molecular fragments, and the property information includes structural properties and electronic properties.

[0059] Next, build a molecular generation model

[0060] In this embodiment, the molecular generation model adopts a deep autoregressive neural network, which can gradually predict the atomic type and atomic position of each atom in the molecule and continuously iterate to generate a complete molecular structure.

[0061] like Figure 2 As shown, the molecular generation model includes two routes. The first processing route is used to process the structure information to obtain the atomic feature vector, and the second processing route is used to process the attribute information to obtain the attribute feature vector. Specifically:

[0062] The first processing route includes an embedding layer, an atom pair distance weighting layer and a feature processing layer connected in sequence. The embedding layer is used to extract the atomic feature vector from the structural information. Then, the atom pair distance weighting layer performs weighted interaction on the atomic feature vector. The processing expression is:

[0063] in, represents the atomic characteristics of the ith atom, represents the atomic characteristics of the jth atom adjacent to the ith atom, represents the distance between atom i and its neighboring atom j, represents the distance cutoff function, represents the radial basis function, Represents network filtering.

[0064] The feature processing layer is used to process the relationship between atoms and their surrounding environment. It includes multiple feature interaction layers and feature mixing layers connected in sequence. Its processing expression is:

[0065] in, represents the atomic feature of the i-th atom after feature processing, , are weight matrices, , are biased, represents the activation function, .

[0066] The atomic feature of the i-th atom after feature processing is the final atomic feature of the i-th atom. The atomic features of all atoms are combined to form .

[0067] The second processing route includes an attribute embedding module and a multi-layer perceptron (MLP) connected in sequence. The attribute embedding module is used to process the structural attribute feature vector and the electronic attribute feature vector of the structural attribute and the electronic attribute respectively. The expressions for processing the structural attribute and the electronic attribute are:

[0068]

[0069] in, represents the structural attribute feature vector, represents multi-layer perceptron processing, represents the concentration weight of atom type a in the existing additive molecule fragment, Embedding vector representing atom type a; represents the electronic property eigenvector, represents the electronic properties, represents the minimum value of electronic properties, Represents the spacing parameter of the radial basis function.

[0070] Next, the multi-layer perceptron is used to integrate the structural attribute feature vector and the electronic attribute feature vector to obtain the attribute feature vector, and its processing expression is:

[0071] in, Represents the attribute feature vector.

[0072] The output end of the first processing route is divided into two paths. In order to further improve the model's ability to characterize atomic features, an attention module is also provided in this embodiment. The attention module is arranged after the output end of the first processing route, that is, both paths need to pass through the attention module before starting.

[0073] The attention module includes multi-head attention units, self-attention units, and cross-attention units connected in sequence. The multi-head attention unit achieves parallel attention to different aspects of the molecular structure by projecting atomic features into multiple representation subspaces. This mechanism is particularly suitable for processing atomic environments with complex structural features. The self-attention unit builds a global context perception layer so that each atom can obtain information about all other atoms in the molecule, significantly enhancing the model's ability to understand the relationship between atoms. The cross-attention unit achieves multi-dimensional fusion of atomic feature representation by associating information from different feature spaces, enhancing the model's ability to represent the local environment of atoms.

[0074] The introduction of these attention mechanisms not only improves the model’s ability to capture the local chemical environment of atoms, but also maintains a global perception of the relationships between atoms during the attention processing. The processing expression of the attention module is:

[0075] in, represents the atomic feature of the i-th atom after feature processing, represents the atomic features of the i-th atom after being processed by the attention module, represents the query matrix, represents the key matrix, represents the value matrix, Represents the feature dimension.

[0076] The attention module calculates the query-key similarity and dynamically assigns feature importance weights, thus achieving selective enhancement of atomic features.

[0077] The output end of the first processing route inputs the atomic feature vector and the attribute feature vector obtained by the second processing route into the atom type prediction layer. The atom type prediction layer is used to predict the next atom type to obtain the next atom type probability distribution function, and its processing expression is:

[0078]

[0079] in, represents the weight factor of the j-th candidate atom type, represents the sum of atomic features in the generated molecular fragments, Represents the next atom type probability distribution function.

[0080] The next atom type probability distribution function includes multiple candidate atoms and their probabilities. The output end of the atom type prediction layer is connected to the input end of the atom type sampling layer. The atom type sampling layer randomly samples an atom type from multiple candidate atoms as the next atom type according to the next atom type probability distribution function. In this embodiment, the random sampling is not necessarily the atom type corresponding to the maximum probability, but the atom type with the maximum probability is most likely to be sampled, which makes the selection of the next atom type more diverse.

[0081] After sampling by the atom type sampling layer, it is input into the atom type embedding layer. The atom type embedding layer is used to convert the next atom type obtained by sampling into an atom type feature vector to facilitate subsequent model processing.

[0082] The output end of the first processing route inputs the atomic feature vector and the atomic type feature vector into the atomic distance prediction layer after point multiplication and addition. The atomic distance prediction layer is used to predict the next atomic position to obtain the next atomic distance probability distribution function, and its processing expression is:

[0083]

[0084] in, represents the weight factor of the jth candidate atomic position, represents the atomic number of the next atom, express The embedding vector of represents the attribute feature vector, represents the probability distribution function of the next atom position.

[0085] The output of the atomic distance prediction layer is connected to the input of the atomic position sampling layer, and the atomic position sampling layer randomly samples the next atomic position according to the probability distribution function of the next atomic distance. At this point, the atomic type and atomic position of the next atom have been determined, and the next molecular fragment with an additional atom can be generated. The next molecular fragment is input into the first processing route and the second processing route again, and it is continuously iterated until a complete molecular structure is generated.

[0086] Then, the constructed molecular generation model needs to be pre-trained. The pre-training method includes: a. Obtain multiple known complete molecular structures; b. Split multiple known complete molecular structures into multiple molecular fragments, and form them into training set data; c. Use the training set data as input to train the molecular generation model, output the predicted complete molecular structure, and calculate the loss function; d. Repeat steps c~d until the loss function drops to a preset range or reaches a preset number of iterations. The loss function uses the cross entropy of the type prediction and distance prediction of the complete molecular structure to obtain a pre-trained molecular generation model.

[0087] Next, the existing additive molecular fragment data is used as input, and multiple additive molecular structures are obtained based on the output of the pre-trained molecular generation model.

[0088] A plurality of additive molecular structures are screened according to target characteristics. In this embodiment, the molecular structure is an additive for optimizing the performance of ester-based insulating oil, so the target characteristics that the molecular structure needs to meet are low first excited state energy and low ionization energy.

[0089] Screening is divided into coarse screening and fine screening. In the coarse screening stage, the energy prediction model is used to predict the first excited state and ionization energy of multiple additive molecular structures. The energy prediction model can quickly predict the first excited state energy and ionization energy of the molecule based on the structural characteristics, atomic type and position of the molecule. The prediction results are used to evaluate whether the molecular structure of the additive meets the target characteristics and eliminate invalid molecules (complete molecules are not necessarily valid molecules, but valid molecules are definitely complete molecules). Coarse screening can greatly reduce the number of molecules that need to be analyzed in depth, providing an efficient basis for subsequent fine screening.

[0090] The introduction of predictive models makes the virtual screening process more automated and faster, no longer relying on a large number of experimental measurements. Through the evaluation of predictive models, molecules with potential ideal properties can be preliminarily screened, improving the efficiency of the entire molecule generation and optimization process.

[0091] In the fine screening process, quantum chemical calculations are performed on the molecules that meet the target characteristics obtained through the coarse screening, and their properties are evaluated with higher precision, especially recalculating key electronic properties such as the first excited state energy and ionization energy. In this embodiment, the quantum chemical calculations use appropriate functionals and basis sets in the Gaussian software, and through more precise quantification methods, it is ensured that the molecules that are ultimately retained can meet strict performance requirements in practical applications.

[0092] In the fine screening process, quantum chemical calculations can not only accurately measure the electronic properties of molecules, but also reveal their stability and reactivity under different conditions, thus providing reliable data support for the final selection of molecules. This rigorous screening method provides practical verification for low S1 energy or low IP molecules generated by the model, which helps to further narrow the candidate range and lay a solid foundation for subsequent experimental verification and application.

[0093] Finally, the molecules retained by the screening are subjected to simulation analysis and verification of streamer discharge. In this embodiment, COMSOL simulation verification is adopted. Specifically, the photoionization model is used to simulate the performance of the molecule as an ester-based insulating oil additive in a streamer environment, and its influence on the streamer velocity is evaluated to verify the potential role of the additive molecule under actual discharge conditions, so as to confirm whether it can effectively reduce the streamer velocity, thereby improving the electrical properties of the ester-based insulating oil. Finally, the molecular structure with the best effect of slowing down the streamer velocity is the additive molecular structure used to optimize the performance of the ester-based insulating oil.

[0094] After determining the specific molecular structure of the additive, we enter the stage of synthesis and preparation of the additive. First, we select a suitable chemical synthesis route based on the structural characteristics of the molecule, and control the synthesis conditions (such as temperature, pressure, catalyst, etc.) to ensure the purity and structural integrity of the molecule.

[0095] After the synthesis is completed, the molecule is purified and verified as necessary, including conventional analytical methods such as nuclear magnetic resonance (NMR) and mass spectrometry (MS) to confirm the accuracy of the molecular structure. The final prepared molecule will be used as an additive in subsequent ester-based insulating oil blending experiments to provide basic materials for improving the electrical properties of ester-based insulating oil.

[0096] The synthesized additives are added to the ester-based insulating oil matrix at different concentrations and mixed thoroughly to ensure that the additives are evenly distributed in the oil. The mixed samples are preliminarily tested to observe their basic physical properties and stability to ensure that the uniformity and basic compatibility of the mixture meet expectations.

[0097] Then, the standard breakdown voltage test equipment, impulse voltage generator, was used to measure the breakdown voltage of ester-based insulating oil samples mixed with additives of different types and concentrations. In the experiment, the impulse voltage generator applied positive and negative voltage pulses to the samples, and tested them under extreme environments such as high temperature, high voltage and simulated lightning impulses, and recorded the breakdown voltage value of each sample.

[0098] After all tests are completed, the experimental data of various additives are recorded in detail, including key parameters such as the concentration, mixing ratio, and breakdown voltage value of each additive. By comparing and analyzing the performance data of different additives, the additive combination with the best effect of improving the breakdown voltage of ester-based insulating oil is selected, and special attention is paid to its environmental protection, such as whether it is non-toxic, easily soluble in ester-based insulating oil, and cost-controlled.

[0099] The following is a detailed explanation in conjunction with specific implementation data.

[0100] Figure 3 are some known complete molecular structures used to train the molecular generation model in this embodiment, Figure 4 The molecular structures of some additives were obtained through screening in this example. Figure 5 This is the simulation effect diagram for COMSOL simulation verification. Figure 5 The left side shows the simulation effect of the photoionization model after adding a certain additive molecule, and the right side shows the simulation effect of the original model. From the comparison results, it can be clearly observed that after adding a certain molecular additive, the development length of the streamer branch is significantly shortened, which further shows that in ester-based insulating oil, adding additives with low ionization energy or low first excited state has a certain effect on suppressing the generation of fast streamers in ester-based insulating oil.

[0101] The effect of adding the final determined additive molecules to the ester-based insulating oil FR3 on improving the breakdown voltage of the ester-based insulating oil is shown in Table 1.

[0102]

[0103] In lightning impulse breakdown tests, the "gap" (also called "oil gap" or "gap") usually refers to the distance between the test electrodes. In the standard pin-plate electrode configuration, this specifically refers to the vertical distance between the pin-tip electrode and the plate electrode. As can be seen from Table 1, the breakdown voltage of the insulating oil with additives is higher than that of the insulating oil without additives in most cases, which effectively proves that the final additive molecule can effectively improve the breakdown voltage performance of ester-based insulating oil.

[0104] Embodiment 2:

[0105] Based on Example 1, the difference between this embodiment and Example 1 is only that the attention module is not included.

[0106] Embodiment 3:

[0107] Based on Example 1, the difference between this embodiment and Example 1 is that the attention module only includes multi-head attention units.

[0108] Embodiment 4:

[0109] Based on Example 1, the difference between this embodiment and Example 1 is that the attention module only includes cross-attention units.

[0110] Embodiment 5:

[0111] Based on Example 1, the difference between this embodiment and Example 1 is that the attention module only includes a self-attention unit.

[0112] The 1000 molecular structures generated by the molecular generation models of Examples 2 to 5 were compared, and the results are shown in Table 2.

[0113] It can be seen from Table 1 that after the introduction of different attention mechanisms, the effectiveness, uniqueness and true molecule generation rate of the molecular structures generated by the model are improved to varying degrees.

[0114] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for screening and designing additive molecules for optimizing the performance of ester-based insulating oil, characterized in that: include: Acquire molecular fragment data with target characteristics; Taking molecular fragment data as input, multiple complete additive molecular structures are output based on the molecular generation model; Screening multiple complete additive molecular structures according to target characteristics to obtain a screened additive molecular structure; The molecular structure of the screened additives was verified by streamer discharge simulation analysis to obtain the molecular structure of the additive for optimizing the performance of ester-based insulating oil.

2. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 1, characterized in that: The target properties include low first excited state energy and low ionization energy; The molecular fragment data includes structural information and attribute information of the molecular fragment, wherein the structural information includes the atomic type and atomic position of the atoms contained in the molecular fragment, and the attribute information includes structural attributes and electronic attributes.

3. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 1, characterized in that: The molecular generation model adopts a deep autoregressive neural network, which includes a first processing route and a second processing route, wherein the first processing route is used to process the structure information to obtain an atomic feature vector, and the second processing route is used to process the attribute information to obtain an attribute feature vector; The output ends of the first processing route and the second processing route are both connected to the input end of the atom type prediction layer, the atom type prediction layer is used to predict the next atom type according to the atom feature vector and the attribute feature vector to obtain the next atom type probability distribution function, the output end of the atom type prediction layer is connected to the atom type sampling layer and the atom type embedding layer in sequence, the atom type sampling layer is used to randomly sample according to the next atom type probability distribution function to obtain the next atom type, and the atom type embedding layer is used to convert the next atom type into an atom type feature vector; The atomic feature vector and the atomic type feature vector are sequentially multiplied and added and then input into the atomic distance prediction layer, the atomic distance prediction layer is used to predict the next atomic position according to the atomic feature vector and the atomic type feature vector to obtain the next atomic distance probability distribution function, the output end of the atomic distance prediction layer is connected to the input end of the atomic position sampling layer, the atomic position sampling layer is used to randomly sample the next atomic position according to the next atomic distance probability distribution function to generate the next molecular fragment, and the next molecular fragment is input into the first processing route and the second processing route for continuous iteration until the complete additive molecular structure is obtained.

4. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 3, characterized in that: The first processing route includes an embedding layer, an atom pair distance weighting layer and a feature processing layer connected in sequence, wherein the embedding layer is used to extract an atomic feature vector from structural information, the atom pair distance weighting layer is used to perform weighted interaction on the atomic feature vector according to the distance between atoms, and the feature processing layer is used to process the relationship between atoms and their surrounding environment, and includes a plurality of feature interaction layers and feature mixing layers connected in sequence; The second processing route includes an attribute embedding module and a multi-layer perceptron connected in sequence, wherein the attribute embedding module is used to convert structural attributes and electronic attributes into structural attribute feature vectors and electronic attribute feature vectors respectively, and the multi-layer perceptron is used to integrate the structural attribute feature vectors and the electronic attribute feature vectors to obtain the attribute feature vectors.

5. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 4, characterized in that: The expression of the atom pair distance weighted layer is: ； in, represents the atomic characteristics of the ith atom, represents the atomic characteristics of the jth atom adjacent to the ith atom, represents the distance between atom i and its neighboring atom j, represents the distance cutoff function, represents the radial basis function, Represents network filtering; The expression of the feature processing layer is: ； in, represents the atomic feature of the i-th atom after feature processing, , are weight matrices, , are all bias terms, represents the activation function, ; The expression of the attribute embedding module is: ； ； in, represents the structural attribute feature vector, represents multi-layer perceptron processing, represents the concentration weight of atom type a in the existing additive molecule fragment, Embedding vector representing atom type a; represents the electronic property eigenvector, represents the electronic properties, represents the minimum value of electronic properties, represents the interval parameter of the radial basis function; The expression of the multi-layer perceptron is: ； in, Represents the attribute feature vector.

6. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 3, characterized in that: The expression of the atom type prediction layer is: ； ； in, represents the weight factor of the j-th candidate atom type, represents the sum of atomic features in the generated molecular fragments, represents the attribute feature vector, represents the probability distribution function of the next atom type; The expression of the atomic position prediction layer is: ； ； in, represents the weight factor of the jth candidate atomic position, represents the atomic number of the next atom, express The embedding vector of represents the attribute feature vector, represents the probability distribution function of the next atom position.

7. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 4, characterized in that: The molecule generation model further includes an attention module, wherein an input end of the attention module is connected to an output end of the feature processing layer; The attention module includes a multi-head attention unit, a self-attention unit and a cross-attention unit connected in sequence, wherein the multi-head attention unit is used to project the atomic feature head to multiple representation subspaces, the self-attention unit is used to enable each atom to obtain information of other atoms, and the cross-attention unit is used to associate information of different feature spaces to achieve multi-dimensional fusion of atomic features; The expression of the attention module is: ； in, represents the atomic feature of the i-th atom after feature processing, represents the atomic features of the i-th atom after being processed by the attention module, represents the query matrix, represents the key matrix, represents the value matrix, Represents the feature dimension.

8. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 1, characterized in that: The method also includes pre-training the molecule generation model, wherein the pre-training method includes: a. Obtain multiple known complete molecular structures; b. Split multiple known complete molecular structures into multiple molecular fragments and form them into training set data; c. Use the training set data as input to train the molecular generation model, output the predicted complete molecular structure, calculate the loss function, and use the loss function to adjust the model weight parameters; d. Repeat steps c to d until the loss function drops to a preset range or reaches a preset number of iterations, and obtain a pre-trained molecular generation model; The loss function uses the cross entropy of type prediction and distance prediction for predicting the complete molecular structure.

9. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 1, characterized in that: The method further comprises screening a plurality of additive molecular structures according to target characteristics to obtain the screened additive molecular structures, including: The energy prediction model is used to predict the first excited state energy and ionization energy of multiple additive molecular structures to obtain prediction results; Coarsely screen multiple additive molecular structures based on the prediction results to retain effective additive molecular structures that meet the target properties; Perform quantum chemical calculations on the first excited state energy and ionization energy of the effective additive molecular structure that meets the target characteristics to obtain the calculation results; The retained additive molecular structures are finely screened according to the calculation results, and the additive molecular structures with target properties within a preset range are retained and used as the screened additive molecular structures.

10. The additive molecule screening and design method for optimizing the performance of ester-based insulating oil according to claim 1, characterized in that: The streamer discharge simulation analysis and verification of the screened additive molecular structure is performed to obtain the additive molecular structure for optimizing the performance of the ester-based insulating oil, including: The streamer discharge simulation analysis is performed to verify the molecular structure of the screened additives, and the streamer discharge simulation analysis results are obtained; According to the results of streamer discharge simulation analysis, the molecular structure with the best effect of slowing down the streamer velocity is retained as the molecular structure of the additive for optimizing the performance of ester-based insulating oil.

Images