Performance analysis method and system for metallized film flexible capacitor

Abstract

The invention provides a performance analysis method and system for a metalized film flexible capacitor, and relates to the technical field of artificial intelligence. The method comprises the following steps: firstly, acquiring a target deposition thickness sequence formed by a target capacitor in a coating process; secondly, first deposition thickness data and a comparison deposition thickness sequence are determined from the target deposition thickness sequence; then, performing prediction processing based on the first deposition thickness data and coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence; and finally, performing comparative analysis on the predicted deposition thickness sequence and the comparative deposition thickness sequence to obtain metallized film performance data of the target capacitor. Based on the above content, the problem that the reliability of capacitor performance analysis is relatively low in the prior art can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a performance analysis method and system for a metallized thin film flexible capacitor. Background Art

[0002] The key to manufacturing a metallized thin film flexible capacitor, especially for DC support capacitors, is to improve its energy partitioning design to improve its stability and safety. The specific design involves the microstructure design of the base film and the metal layer, the edge structure design of the metallized film, the microstructure coating control method, etc. Among them, precise microstructure coating control is one of the core technologies to improve the performance of metallized thin film capacitors. For example, the uniformity of the metal layer on the film can avoid the problem of local electric field concentration and reduce the occurrence of breakdown. In this way, the performance of the capacitor can be determined by analyzing the uniformity of the metal layer. For example, if the determined uniformity is better, it can reflect better performance. However, in the prior art, the uniformity of the metal layer is generally tested after the coating is completed. In this way, efficient performance testing can be achieved. However, since it is based solely on the test results after the coating is completed, it may lead to relatively low reliability. Summary of the Invention

[0003] In view of this, an object of the present invention is to provide a performance analysis method and system for a metallized thin film flexible capacitor, so as to improve the problem of relatively low reliability of capacitor performance analysis in the prior art.

[0004] To achieve the above object, the present invention adopts the following technical solutions: A performance analysis method for a metallized film flexible capacitor, comprising: Obtaining a target deposition thickness sequence formed during a coating process of a target capacitor, wherein the target deposition thickness sequence includes a plurality of deposition thickness data corresponding to a plurality of deposition stages, each deposition thickness data being used to characterize the thickness of at least one point of the metallized film deposited in the current stage; Determining first deposition thickness data and a comparative deposition thickness sequence from the target deposition thickness sequence, wherein each deposition thickness data in the comparative deposition thickness sequence is located after the first deposition thickness data; performing prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence, wherein each deposition thickness data in the predicted deposition thickness sequence is a prediction result for a deposition stage subsequent to the first deposition thickness data; The predicted deposition thickness sequence and the comparative deposition thickness sequence are compared and analyzed to obtain the metallized film performance data of the target capacitor.

[0005] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the step of performing prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized thin film to obtain a predicted deposition thickness sequence includes: Performing semantic mining on the first deposition thickness data to output A deposition thickness semantic vectors, wherein the deposition thickness semantic vectors are used to reflect potential semantic information of the first deposition thickness data; Determining a timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, and, based on the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, semantically adjusting the corresponding deposition thickness semantic vector in the A deposition thickness semantic vectors, and outputting A deposition thickness adjustment vectors, wherein each of the deposition thickness adjustment vectors is used to predict a deposition thickness data in a required predicted deposition thickness sequence, and the semantic adjustment is used to control a change between each deposition thickness data in the predicted deposition thickness sequence and the first deposition thickness data; Based on the coating deposition parameters, the A deposition thickness adjustment vectors are predicted and processed to output a corresponding predicted deposition thickness sequence, wherein the predicted deposition thickness sequence includes A deposition thickness data.

[0006] In a preferred embodiment of the present invention, in the above-mentioned method for analyzing the performance of the metallized thin film flexible capacitor, the step of performing semantic mining on the first deposition thickness data and outputting A deposition thickness semantic vectors includes: Embedding the first deposition thickness data into a vector space to form a corresponding deposition thickness embedding vector, and performing self-attention processing on the deposition thickness embedding vector to form a corresponding deposition thickness semantic vector; The deposition thickness semantic vector is expanded to form A deposition thickness semantic vectors, wherein every two deposition thickness semantic vectors in the A deposition thickness semantic vectors are the same.

[0007] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the steps of determining the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, and, based on the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, semantically adjusting the corresponding deposition thickness semantic vector in the A deposition thickness semantic vectors, and outputting A deposition thickness adjustment vectors, include: Determining a preconfigured reference adjustment parameter, wherein the reference adjustment parameter is used to reflect the difference between adjacent deposition thickness data in the predicted deposition thickness sequence; Obtaining A training adjustment parameters carried in a target thickness prediction network, and determining A timing adjustment parameters based on the reference adjustment parameter and the A training adjustment parameters, wherein there is a one-to-one correspondence between the A timing adjustment parameters and the A deposition thickness semantic vectors, and the A training adjustment parameters are formed during the training process of the target thickness prediction network, which is a neural network model; Based on each of the A timing adjustment parameters, the corresponding deposition thickness semantic vector is mapped and adjusted to output A deposition thickness adjustment vectors.

[0008] In a preferred embodiment of the present invention, in the above-mentioned method for analyzing the performance of the metallized thin film flexible capacitor, the step of predicting the A deposition thickness adjustment vectors based on the coating deposition parameters and outputting a corresponding predicted deposition thickness sequence includes: Determining A deposition latent semantic vectors, wherein the deposition latent semantic vectors are formed by perturbing the latent semantic vectors in the semantic space of the deposition thickness formed by training, the perturbed vectors obey a normal distribution, and one of the deposition latent semantic vectors is used to predict one deposition thickness data in a required predicted deposition thickness sequence; Performing semantic mining on the coating deposition parameters to form corresponding deposition parameter semantic vectors, wherein the deposition parameter semantic vectors are used to reflect semantic information of the coating deposition parameters, and the coating deposition parameters include at least one of coating distance, gas flow rate, coating temperature, coating bias, deposition atmosphere, sputtering power, and target current; The deposition parameter semantic vector and the A deposition latent semantic vectors are predicted and processed by the target thickness prediction network, and the A deposition thickness adjustment vectors are fused during the prediction process to output a corresponding predicted deposition thickness sequence, wherein the latent semantic vector in the semantic space where the deposition thickness is located is formed during the training process of the target thickness prediction network, which belongs to a neural network model.

[0009] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the step of predicting the deposition parameter semantic vector and the A deposition potential semantic vectors through a target thickness prediction network, and fusing the A deposition thickness adjustment vectors during the prediction process to output a corresponding predicted deposition thickness sequence includes: For each of the A deposited latent semantic vectors, determining a corresponding semantic vector to be predicted based on the deposited latent semantic vector; By using a semantic fusion model in the target thickness prediction network, the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors are fused into the semantic vector to be predicted, and a multi-dimensional fusion vector corresponding to the deposition latent semantic vector is output; The multi-dimensional fusion vector is predicted and processed by the prediction output model in the target thickness prediction network to output a deposition thickness data corresponding to the deposition latent semantic vector, and the A deposition thickness data corresponding to the A deposition latent semantic vectors are combined to form a corresponding predicted deposition thickness sequence.

[0010] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the step of determining the corresponding semantic vector to be predicted based on each of the A deposited latent semantic vectors comprises: Determining the first one of the A deposited latent semantic vectors as the corresponding semantic vector to be predicted; For each deposition latent semantic vector other than the first deposition latent semantic vector among the A deposition latent semantic vectors, splicing is performed based on the multi-dimensional fusion vector corresponding to the deposition latent semantic vector and the previous deposition latent semantic vector of the deposition latent semantic vector to form a corresponding spliced ​​semantic vector, and the spliced ​​semantic vector is downsampled to form a corresponding semantic vector to be predicted.

[0011] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the step of fusing the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition potential semantic vector in the A deposition thickness adjustment vectors into the semantic vector to be predicted through the semantic fusion model in the target thickness prediction network, and outputting a multi-dimensional fusion vector corresponding to the deposition potential semantic vector, includes: Performing significant feature mining on the semantic vector to be predicted through the first mining branch and the second mining branch included in the semantic fusion model in the target thickness prediction network, and outputting the corresponding first significant semantic vector and second significant semantic vector; fusing the deposition parameter semantic vector into the first saliency semantic vector to obtain a corresponding first fused semantic vector; fusing a deposition thickness adjustment vector corresponding to the deposition latent semantic vector among the A deposition thickness adjustment vectors into the second saliency semantic vector to obtain a corresponding second fused semantic vector; The first fused semantic vector and the second fused semantic vector are added to output a multi-dimensional fused vector corresponding to the deposited latent semantic vector.

[0012] In a preferred embodiment of the present invention, in the performance analysis method of the metallized thin film flexible capacitor, the step of fusing the deposition parameter semantic vector into the first saliency semantic vector to obtain the corresponding first fused semantic vector includes: Performing multiple cascaded semantic extractions on the first saliency semantic vector to form corresponding multiple first extracted semantic vectors, wherein, for each semantic extraction, the first extracted semantic vector corresponding to the previous semantic extraction and the deposition parameter semantic vector are spliced, and self-attention processing is performed on the spliced ​​semantic vector, and the semantic vector obtained by the self-attention processing is down-sampled to form a first extracted semantic vector corresponding to the current semantic extraction, and the first extracted semantic vector corresponding to the semantic extraction before the first semantic extraction is the first saliency semantic vector; Perform multiple cascaded semantic diffusions on the last first extracted semantic vector to form corresponding multiple first diffusion semantic vectors, wherein, for each semantic diffusion, the first diffusion semantic vector corresponding to the previous semantic diffusion and a first extracted semantic vector of the corresponding level are spliced, and the spliced ​​semantic vector is self-attention processed, and the semantic vector obtained by the self-attention processing is up-sampled to form the first diffusion semantic vector corresponding to the current semantic diffusion, and the first diffusion semantic vector corresponding to the semantic diffusion before the first semantic diffusion is the last first extracted semantic vector; The last first diffusion semantic vector is determined as the first fusion semantic vector.

[0013] Based on the above, the present invention also provides a performance analysis system for a metallized thin film flexible capacitor, comprising: memory for storing computer programs; The processor connected to the memory is used to execute the computer program stored in the memory to implement the performance analysis method of the metallized film flexible capacitor.

[0014] The present invention provides a performance analysis method and system for a metallized thin film flexible capacitor. First, a target deposition thickness sequence formed during the coating process of the target capacitor is obtained; second, a first deposition thickness data and a comparative deposition thickness sequence are determined from the target deposition thickness sequence; then, a prediction process is performed based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence; finally, the predicted deposition thickness sequence and the comparative deposition thickness sequence are compared and analyzed to obtain the metallized film performance data of the target capacitor. Based on the above content, by predicting the corresponding deposition thickness sequence, it is possible to compare and analyze it as a theoretical value with the comparative deposition thickness sequence as an actual value, thereby achieving a comparison of the data of each deposition stage, that is, ensuring that the dynamic process of multiple deposition stages can be fully analyzed. Therefore, compared with the conventional solution of comparing the data after the deposition is completed, it can be more comprehensive and therefore more reliable, thereby improving the problem of relatively low reliability of capacitor performance analysis in the prior art. In addition, since the predicted corresponding deposition thickness sequence refers to the first deposition thickness data, the initial deposition thickness data of the predicted deposition thickness sequence and the comparison deposition thickness sequence can be made the same, that is, the initial conditions are the same, so that the comparison accuracy of the subsequent deposition thickness data can be higher, and more attention can be paid to the dynamic changes in the subsequent deposition process. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings.

[0016] Figure 1 This is a structural block diagram of a performance analysis system for a metallized film flexible capacitor provided by an embodiment of the present invention.

[0017] Figure 2 A schematic flow chart of a performance analysis method for a metallized thin film flexible capacitor provided in an embodiment of the present invention.

[0018] Figure 3 A schematic diagram of fusing to form a multi-dimensional fusion vector provided by an embodiment of the present invention.

[0019] Figure 4 A schematic diagram of multi-level fusion of semantic vectors provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0022] like Figure 1 As shown, an embodiment of the present invention provides a performance analysis system for a metallized film flexible capacitor, wherein the performance analysis system for a metallized film flexible capacitor may include a memory, a processor, and a performance analysis device for a metallized film flexible capacitor.

[0023] In detail, the memory and the processor are electrically connected directly or indirectly to realize data transmission or interaction. For example, the memory and the processor can be electrically connected through one or more communication buses or signal lines. The performance analysis device of the metallized film flexible capacitor includes at least one software function module stored in the memory in the form of software or firmware. The processor is used to execute the executable computer program stored in the memory, for example, the software function module and computer program included in the performance analysis device of the metallized film flexible capacitor, so as to realize the performance analysis method of the metallized film flexible capacitor provided in the embodiment of the present invention.

[0024] Optionally, the memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0025] Furthermore, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0026] In addition, the performance analysis device for the metallized thin film flexible capacitor may include multiple modules, such as a first module, a second module, a third module, and a fourth module (all of which are software functional modules). The first module is used to obtain a target deposition thickness sequence formed during the coating process of a target capacitor, wherein the target deposition thickness sequence includes multiple deposition thickness data corresponding to multiple deposition stages, each deposition thickness data representing the thickness of at least one point of the metallized thin film deposited during the current stage. The second module is used to determine a first deposition thickness data and a comparative deposition thickness sequence from the target deposition thickness sequence, wherein each deposition thickness data in the comparative deposition thickness sequence is located after the first deposition thickness data. The third module is used to perform predictive processing based on the first deposition thickness data and the coating deposition parameters of the metallized thin film to obtain a predicted deposition thickness sequence, wherein each deposition thickness data in the predicted deposition thickness sequence is a prediction result for the deposition stage after the first deposition thickness data. The fourth module is used to compare and analyze the predicted deposition thickness sequence with the comparative deposition thickness sequence to obtain performance data of the metallized thin film of the target capacitor.

[0027] I understand. Figure 1 The structure shown is for illustration only. The performance analysis system of the metallized film flexible capacitor may also include a comparison Figure 1 More or fewer components than shown, or with Figure 1 The different configurations shown, for example, may further include a communication unit for exchanging information with other devices (such as a device for collecting deposition thickness, etc.).

[0028] Combine Figure 2 The embodiment of the present invention further provides a performance analysis method for a metallized film flexible capacitor that can be applied to the performance analysis system for the metallized film flexible capacitor. The method steps defined in the process related to the performance analysis method for the metallized film flexible capacitor can be implemented by the performance analysis system for the metallized film flexible capacitor (hereinafter referred to as the performance analysis system). Figure 2 The specific process shown is explained in detail.

[0029] Step S110 , obtaining a target deposition thickness sequence formed during the coating process of the target capacitor.

[0030] In an embodiment of the present invention, the performance analysis system can obtain a target deposition thickness sequence formed during the coating process of a target capacitor. The target deposition thickness sequence includes multiple deposition thickness data corresponding to multiple deposition stages (which can be arranged in chronological order). Each deposition thickness data is used to characterize the thickness of at least one location on the metallized film deposited during the current stage (the specific number of locations can be selected based on actual needs; for example, a larger number of locations can be used to ensure accuracy, or a smaller number can be used to reduce computational complexity). Furthermore, the deposition thickness data can be obtained by measuring the reflectivity of the corresponding locations using a spectroscopic reflectometer. Since differences in reflectivity can reflect differences in thickness, the measured reflectivity can be used to characterize the corresponding thickness. Thus, by deploying multiple spectroscopic reflectometers (e.g., forming a corresponding detector array), the thickness of multiple locations can be monitored. In other embodiments, the film deposition thickness can be monitored by monitoring changes in oscillation frequency using a quartz crystal microbalance (QCM), thereby obtaining the corresponding target deposition thickness sequence.

[0031] Step S120 : determining first deposition thickness data and a comparison deposition thickness sequence from the target deposition thickness sequence.

[0032] In an embodiment of the present invention, after obtaining the target deposition thickness sequence, the performance analysis system can determine first deposition thickness data and a comparative deposition thickness sequence from the target deposition thickness sequence. Each deposition thickness data in the comparative deposition thickness sequence is located after the first deposition thickness data. For example, the first deposition thickness data in the target deposition thickness sequence can be determined as the first deposition thickness data, and then the remaining deposition thickness data in the target deposition thickness sequence other than the first deposition thickness data can be sorted in order to form a corresponding comparative deposition thickness sequence.

[0033] Step S130 , performing prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence.

[0034] In an embodiment of the present invention, after determining the first deposition thickness data, the performance analysis system may perform prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence. Each deposition thickness data in the predicted deposition thickness sequence is a prediction result for a deposition stage subsequent to the first deposition thickness data, i.e., it corresponds one-to-one to a deposition thickness data in the comparison deposition thickness sequence.

[0035] Step S140 : performing comparative analysis on the predicted deposition thickness sequence and the comparative deposition thickness sequence to obtain performance data of the metallized film of the target capacitor.

[0036] In an embodiment of the present invention, after obtaining the predicted deposition thickness sequence, the performance analysis system can compare and analyze the predicted deposition thickness sequence and the comparative deposition thickness sequence to obtain the metallized film performance data of the target capacitor. For example, the difference between the corresponding deposition thickness data in the predicted deposition thickness sequence and the comparative deposition thickness sequence is calculated, and then the metallized film performance data is determined based on the obtained difference. For example, the smaller the difference, the smaller the difference between the predicted result and the actual data, indicating that no abnormality occurs in the actual deposition process, making the deposition process stable, and therefore, the corresponding deposition effect is in line with expectations, that is, it has better performance. On the contrary, if the difference is large, it means that in the actual deposition process, there may be abnormalities, which makes the deposition process problematic, and therefore, the corresponding deposition effect may not meet expectations, that is, the performance may be poor. In addition, the difference can refer to the average value of the absolute difference between the values ​​of the deposition thickness of each point in the two sequences. In addition, the obtained metallized film performance data can be used as a pre-screening condition. For example, after screening out capacitors with poor performance based on the metallized film performance data, the performance of the capacitor can be further tested, such as a breakdown test, to avoid waste caused by misjudgment.

[0037] Based on the above, by predicting the corresponding deposition thickness sequence, it is possible to compare and analyze it as a theoretical value with a comparative deposition thickness sequence as an actual value, thereby achieving a comparison of the data of each deposition stage, that is, ensuring that the dynamic process of multiple deposition stages can be comprehensively analyzed. Therefore, compared with the conventional solution of comparing the data after the deposition is completed, it can be more comprehensive and therefore more reliable, thereby improving the relatively low reliability of capacitor performance analysis in the prior art. In addition, because the predicted corresponding deposition thickness sequence refers to the first deposition thickness data, the initial deposition thickness data of the predicted deposition thickness sequence and the comparative deposition thickness sequence can be the same, that is, the initial conditions are the same, so that the comparison accuracy of the subsequent deposition thickness data can be higher, and more attention can be paid to the dynamic changes in the subsequent deposition process.

[0038] In the embodiment of the present invention, with respect to the above-mentioned step S120, the specific implementation method of performing the prediction process based on the first deposition thickness data and the coating deposition parameters of the metallized film is not limited and can be selected according to actual needs.

[0039] For example, in an alternative embodiment, the first deposition thickness data and the coating deposition parameters may be subjected to semantic mining, and then, based on the corresponding semantic mining results, a corresponding predicted deposition thickness sequence may be predicted.

[0040] For another example, in another alternative embodiment, in order to improve the reliability of the prediction process, the above-mentioned step S120 may further include the following steps S121, S122 and S123, and the details of each step are described below.

[0041] Step S121 : performing semantic mining on the first deposition thickness data, and outputting A deposition thickness semantic vectors.

[0042] In an embodiment of the present invention, semantic mining can be performed on the first deposition thickness data to output A deposition thickness semantic vectors. The deposition thickness semantic vectors are used to reflect the potential semantic information of the first deposition thickness data. That is, the corresponding potential semantic information can be mined from the first deposition thickness data and represented in the form of vectors, thereby obtaining the A deposition thickness semantic vectors. A is equal to the number of deposition thickness data in the comparison deposition thickness sequence, or in other words, the number of deposition thickness data in the predicted deposition thickness sequence to be output.

[0043] Step S122, determining the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, and, based on the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, semantically adjusting the corresponding deposition thickness semantic vectors in the A deposition thickness semantic vectors, and outputting A deposition thickness adjustment vectors.

[0044] In an embodiment of the present invention, after obtaining the A deposition thickness semantic vectors, a timing adjustment parameter corresponding to each deposition thickness semantic vector can be determined. Based on the timing adjustment parameter corresponding to each deposition thickness semantic vector, the corresponding deposition thickness semantic vector in the A deposition thickness semantic vectors is semantically adjusted to output A deposition thickness adjustment vectors. Each deposition thickness adjustment vector is used to predict a desired deposition thickness data item in a predicted deposition thickness sequence, and the semantic adjustment is used to control the variation between each deposition thickness data item in the predicted deposition thickness sequence and the first deposition thickness data item. That is, since each deposition thickness data item in the predicted deposition thickness sequence cannot be identical to the first deposition thickness data item, and the deposition thickness gradually increases over time (later in the sequence or later in the deposition phase), the variation between the deposition thickness data item and the first deposition thickness data item increases. Therefore, it is necessary to determine each corresponding timing adjustment parameter so that the deposition thickness adjustment vector obtained after semantically adjusting the deposition thickness semantic vector based on the corresponding timing adjustment parameter better matches the actual deposition thickness data item, thereby improving the accuracy of the semantic representation.

[0045] Step S123 : Based on the coating deposition parameters, the A deposition thickness adjustment vectors are predicted and a corresponding predicted deposition thickness sequence is output.

[0046] In an embodiment of the present invention, after obtaining the A deposition thickness adjustment vectors, the A deposition thickness adjustment vectors can be predicted based on the coating deposition parameters to output a corresponding predicted deposition thickness sequence. The predicted deposition thickness sequence includes A deposition thickness data. In other words, since different deposition progresses directly lead to different deposition thicknesses, it is necessary to perform prediction processing in conjunction with the actual coating deposition parameters to further improve the reliability of the output predicted deposition thickness sequence.

[0047] Optionally, for the above step S121, the specific implementation method of semantic mining the first deposition thickness data is not limited. For example, in an alternative implementation method, considering that deposition thicknesses at different points are generally correlated, in order to ensure that the mined deposition thickness semantic vector has a higher semantic representation capability, the above step S121 may further include the following practicable content: In the first step, the first deposition thickness data can be vector-space embedded to form a corresponding deposition thickness embedding vector, and the deposition thickness embedding vector can be self-attention processed to form a corresponding deposition thickness semantic vector; wherein, the vector space embedding can be implemented by a corresponding word embedding model. For example, the first deposition thickness data can include the deposition thickness of B points (in some embodiments, it can also include the position coordinates of the points). In this way, each deposition thickness (and position coordinates) can be word-embedded separately to obtain a corresponding word embedding vector. Then, the B word embedding vectors are combined to form a corresponding deposition thickness embedding. embedding vectors (for example, the dimension of each word embedding vector can be 1*C, and the dimension of the sediment thickness embedding vector can be B*C). In addition, since there is also a correlation between each point, the sediment thickness embedding vector carrying the semantic information of the sediment thickness of each point can be self-attention processed to thereby mine the correlation between each point. In addition, the result of the self-attention processing can be superimposed on the sediment thickness embedding vector to obtain a sediment thickness semantic vector that can represent both the semantic information of the sediment thickness and the correlation between the points. In this way, the semantic information represented by the sediment thickness semantic vector can be more rich. In the second step, the deposition thickness semantic vector can be expanded to form A deposition thickness semantic vectors, wherein every two deposition thickness semantic vectors in the A deposition thickness semantic vectors are the same. That is to say, based on the deposition thickness semantic vector, A-1 deposition thickness semantic vectors can be copied to obtain A identical deposition thickness semantic vectors.

[0048] Optionally, for the above step S122, the specific implementation method of outputting the A deposition thickness adjustment vectors is not limited. For example, in an alternative implementation method, in order to ensure that the output A deposition thickness adjustment vectors can reliably represent the deposition thickness at each stage or time sequence, the above step S122 may further include the following executable content: In the first step, a pre-configured reference adjustment parameter may be determined, wherein the reference adjustment parameter is used to reflect the difference between adjacent deposition thickness data in the predicted deposition thickness sequence, and the specific value of the reference adjustment parameter may be configured according to actual conditions. For example, the longer the time between two adjacent stages, the greater the corresponding difference, that is, the smaller the value of the reference adjustment parameter (between 0 and 1); In a second step, A training adjustment parameters carried in the target thickness prediction network can be obtained, and A timing adjustment parameters can be determined based on the reference adjustment parameters and the A training adjustment parameters, wherein there is a one-to-one correspondence between the A timing adjustment parameters and the A deposition thickness semantic vectors, and the A training adjustment parameters are formed during the training process of the target thickness prediction network, and the target thickness prediction network belongs to a neural network model; illustratively, the training adjustment parameters can be a parameter distribution matrix having the same size as the deposition thickness semantic vector, so that each parameter in the training adjustment parameters can be multiplied by the reference adjustment parameter to obtain the corresponding timing adjustment parameter; In the third step, based on each of the A timing adjustment parameters, the corresponding deposition thickness semantic vector can be mapped and adjusted to output A deposition thickness adjustment vectors; illustratively, since the size of the timing adjustment parameter is the same as the size of the deposition thickness semantic vector, the timing adjustment parameter and the deposition thickness semantic vector can be bitwise multiplied to obtain the corresponding deposition thickness adjustment vector. For example, the parameters in the first row and first column of the timing adjustment parameter can be multiplied with the parameters in the first row and first column of the deposition thickness semantic vector to obtain the parameters in the first row and first column of the deposition thickness adjustment vector.

[0049] Optionally, for the above-mentioned step S123, the specific implementation method of predicting the A deposition thickness adjustment vectors is not limited. For example, in an alternative implementation, in order to avoid the problem of overfitting during the prediction processing, that is, in order to improve the reliability of the prediction processing, the above-mentioned step S123 may further include the following steps S123a, S123b and S123c, and the details of each step are described below.

[0050] Step S123a: Determine A deposited latent semantic vectors.

[0051] In an embodiment of the present invention, A deposition latent semantic vectors can be determined. These are formed by perturbing the latent semantic vectors within the semantic space of deposition thickness, which is generated during training. The perturbed vectors follow a normal distribution (and their corresponding mean can be 0). These deposition latent semantic vectors are used to predict a deposition thickness data point in a desired predicted deposition thickness sequence. For example, a random vector following a normal distribution can be generated as a perturbation vector. This perturbation vector can then be added to the latent semantic vector within the semantic space of deposition thickness (with consistent dimensions). It should be noted that, in the initial stages of training, A latent semantic vectors can be randomly generated and used as network parameters for the corresponding neural network model. These vectors are continuously updated during training to form the final latent semantic vector. Thus, since the final latent semantic vector is generated during training, semantic information within the semantic space of deposition thickness can be learned. Thus, the A deposition latent semantic vectors can be used to introduce perturbations to mitigate overfitting while also introducing semantic information within the semantic space of deposition thickness, avoiding semantic distortion caused by the introduction of perturbations.

[0052] Step S123b: semantically mine the coating deposition parameters to form corresponding deposition parameter semantic vectors.

[0053] In an embodiment of the present invention, the film deposition parameters may be semantically mined to form corresponding deposition parameter semantic vectors. The deposition parameter semantic vector is used to reflect the semantic information of the coating deposition parameters, and the coating deposition parameters at least include coating distance (target-substrate distance, which affects the angle, speed and density of particles reaching the substrate during sputtering. For example, during sputtering, if the distance between the target and the substrate is too close, the substrate will receive more sputtered particles, which will lead to an excessively thick film layer), gas flow rate (sputtering gas flow rate, the flow rate of the gas (usually argon) used in the sputtering process. This affects the interaction between the gas and the target and the ionization rate during the sputtering process. For example, when argon is used as the sputtering gas, a lower flow rate may lead to a low sputtering rate and an uneven film, while an excessively high flow rate may produce more ion contamination), coating temperature (substrate temperature, which has a significant impact on the crystal structure, composition uniformity and density of the film), coating bias voltage (bias voltage, during the sputtering process, the energy of the deposited particles is adjusted by applying a bias voltage to the substrate), deposition atmosphere (deposition atmosphere). Atmosphere), the gas environment used in the sputtering process, such as a vacuum environment, an argon atmosphere, an oxygen atmosphere, etc. Different atmosphere environments will affect the composition and crystallization state of the thin film), sputtering power (sputtering power, affecting the density and energy of the sputtered particles), target current (target current, the larger the current, the higher the energy of the sputtered particles and the faster the deposition rate), at least one of which may also include other deposition parameters in other embodiments. Exemplarily, each deposition parameter may be word-embedded to obtain a corresponding word-embedded vector, and then each word-embedded vector may be combined to form a corresponding combination vector. Finally, the combination vector may be self-attention processed so that the semantic information of each deposition parameter can be associated mined, and the combination vector and the result of the self-attention processing are added to obtain the corresponding deposition parameter vector. For example, the size of each word embedding vector may be 1*X. If there are Y deposition parameters, the size of the deposition parameter semantic vector may be Y*X. For example, for the deposition atmosphere "argon", the corresponding word embedding vector may be obtained through word embedding processing: [0.53, 0.85, 0.42, 0.43, 0.10, 0.66, 0.98, 0.83, 0.66, 0.77, 0.86, 0.37, 0.81, 0.15, 0.00, 0.74, 0.45, 0.72, 0.59, …, 0.76].

[0054] For example, for the sputtering power "100 W", through word embedding processing, the corresponding word embedding vector can be obtained: [0.23, 0.54, 0.21, 0.11, 0.72, 0.63, 0.39, 0.46, 0.31, 0.26, 0.13, 0.98, 0.04, 0.87, 0.4, 0.18, 0.5, 0.77, 0.98, …, 0.21].

[0055] Based on this, by combining, we can get the corresponding combination vector: {[0.53, 0.85, 0.42, 0.43, 0.10, 0.66, 0.98, 0.83, 0.66, 0.77, 0.86, 0.37, 0.81, 0.15, 0.00, 0.74, 0.45, 0.72, 0.59, ..., 0.76]; [0.23, 0.54, 0.21, 0.11, 0.72, 0.63, 0.39, 0.46, 0.31, 0.26, 0.13, 0.98, 0.04, 0.87, 0.4, 0.18, 0.5, 0.77, 0.98, …, 0.21]}.

[0056] Step S123c: The deposition parameter semantic vector and the A deposition latent semantic vectors are subjected to prediction processing by a target thickness prediction network, and the A deposition thickness adjustment vectors are fused during the prediction processing to output a corresponding predicted deposition thickness sequence.

[0057] In an embodiment of the present invention, after obtaining the deposition parameter semantic vector and the A deposition latent semantic vectors, the deposition parameter semantic vector and the A deposition latent semantic vectors can be predicted by a target thickness prediction network, and the A deposition thickness adjustment vectors can be fused during the prediction process to output a corresponding predicted deposition thickness sequence. The latent semantic vector in the semantic space where the deposition thickness is located formed by training is formed during the training process of the target thickness prediction network, and the target thickness prediction network belongs to a neural network model. That is to say, since the A deposition latent semantic vectors contain disturbance information, the deposition parameter semantic vector and the A deposition thickness adjustment vectors can be guided or fused to remove the disturbance information in the A deposition latent semantic vectors, thereby obtaining a semantic vector with better semantic representation ability, so that the reliability of the predicted deposition thickness sequence obtained based on the semantic vector is relatively high.

[0058] Optionally, for the above-mentioned step S123c, the specific implementation method of outputting the corresponding predicted deposition thickness sequence is not limited. For example, in an alternative implementation method, considering that there is a correlation between adjacent deposition thickness data in the predicted deposition thickness sequence, therefore, in order to ensure the effective implementation of the correlation and make the obtained predicted deposition thickness sequence more reliable, the above-mentioned step S123c may further include step c1, step c2 and step c3, and the details of each step are described below.

[0059] Step c1: for each of the A deposited latent semantic vectors, determine a corresponding semantic vector to be predicted based on the deposited latent semantic vector.

[0060] In an embodiment of the present invention, for each of the A deposited latent semantic vectors, a corresponding semantic vector to be predicted is determined based on the deposited latent semantic vector. For example, a first semantic vector to be predicted may be determined based on the first deposited latent semantic vector, and a second semantic vector to be predicted may be determined based on the second deposited latent semantic vector. In this way, A semantic vectors to be predicted may be determined for each of the A deposited latent semantic vectors.

[0061] Step c2: Through the semantic fusion model in the target thickness prediction network, the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors are fused into the semantic vector to be predicted, and a multi-dimensional fusion vector corresponding to the deposition latent semantic vector is output.

[0062] In an embodiment of the present invention, after obtaining the corresponding semantic vector to be predicted, the semantic fusion model in the target thickness prediction network can be used to fuse the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors into the semantic vector to be predicted, outputting a multi-dimensional fusion vector corresponding to the deposition latent semantic vector. In other words, the semantic fusion model can be a coding network model that, during the coding mining of the deposition latent semantic vector (i.e., the semantic vector to be predicted), achieves fusion of the deposition parameter semantic vector and the deposition thickness adjustment vector, thereby enabling fusion of multi-dimensional semantic information while removing disturbance information. Based on this, A multi-dimensional fusion vectors can be obtained for the A semantic vectors to be predicted.

[0063] Step c3: perform prediction processing on the multi-dimensional fusion vector through the prediction output model in the target thickness prediction network, output a deposition thickness data corresponding to the deposition latent semantic vector, and combine the A deposition thickness data corresponding to the A deposition latent semantic vectors to form a corresponding predicted deposition thickness sequence.

[0064] In an embodiment of the present invention, after obtaining the multi-dimensional fusion vector, the multi-dimensional fusion vector can be subjected to prediction processing by the prediction output model in the target thickness prediction network, outputting a deposition thickness data corresponding to the deposition latent semantic vector, and combining the A deposition thickness data corresponding to the A deposition latent semantic vectors to form a corresponding predicted deposition thickness sequence. That is, the prediction output model can be a decoding network model, for example, and can include a fully connected network layer, so that the multi-dimensional fusion vector can be fully connected to obtain the corresponding deposition thickness data. Alternatively, it can also include a linear mapping function to linearly map the vector obtained by the fully connected processing to obtain the corresponding deposition thickness data. For example, the deposition thickness data can be the deposition thickness of five points. Thus, through the fully connected processing, a vector of size 1*5 can be obtained, and the five parameters in the vector can correspond to the deposition thickness of the five points. In addition, in other embodiments, after the fully connected processing, self-attention processing can also be performed.

[0065] Optionally, for the above step c1, the specific implementation method of determining the semantic vector to be predicted is not limited. For example, in an alternative implementation method, in order to fully consider the influence of the previous deposition thickness on the subsequent deposition thickness so that the semantic representation accuracy of the determined semantic vector to be predicted is higher, the above step c1 may further include the following practicable contents: In the first step, the first deposited latent semantic vector among the A deposited latent semantic vectors can be determined as the corresponding semantic vector to be predicted. That is, for the first deposited latent semantic vector, it can be directly used as the semantic vector to be predicted. In the second step, for each deposition latent semantic vector other than the first deposition latent semantic vector among the A deposition latent semantic vectors, the deposition latent semantic vector and the multi-dimensional fusion vector corresponding to the previous deposition latent semantic vector of the deposition latent semantic vector are spliced ​​to form a corresponding spliced ​​semantic vector, and the spliced ​​semantic vector is downsampled to form a corresponding semantic vector to be predicted, wherein the size of the semantic vector to be predicted can be consistent with the size of the deposition latent semantic vector, and the downsampling process can be achieved by convolution, pooling, etc. For example, the second deposition latent semantic vector and the multi-dimensional fusion vector corresponding to the first deposition latent semantic vector can be spliced ​​to form a corresponding spliced ​​semantic vector, that is, at the second time step, not only the deposition latent semantic vector corresponding to the current time step should be considered, but also the output semantic vector of the first time step.

[0066] Alternatively, for the above step c2, the specific implementation method of outputting the multi-dimensional fusion vector corresponding to the deposited latent semantic vector is not limited. For example, in an alternative implementation method, in order to reliably remove the disturbance information during the fusion process so that the output multi-dimensional fusion vector has a higher semantic representation accuracy, the above step c2 may further include the following implementable content (combined with Figure 3 ): In the first step, the semantic fusion model in the target thickness prediction network includes a first mining branch and a second mining branch, respectively performing significant feature mining on the semantic vector to be predicted, and outputting the corresponding first significant semantic vector and second significant semantic vector; that is, the semantic vector to be predicted can be mined for significant features through the first mining branch to obtain the corresponding first significant semantic vector, and the semantic vector to be predicted can be mined for significant features through the second mining branch to obtain the corresponding second significant semantic vector, wherein the first mining branch and the second mining branch have the same method for mining significant features, but the parameters carried by the mining branches can be different, so that the first mining branch and the second mining branch can respectively mine and capture different significant features in the semantic vector to be predicted, thereby obtaining the first significant semantic vector and the second significant semantic vector that can characterize different significant characteristics. semantic vector; in addition, the specific implementation process of significant feature mining can be: based on the first parameter distribution, the second parameter distribution and the third parameter distribution carried by the first mining branch, the deposition parameter semantic vector is mapped (such as multiplied) respectively to obtain the corresponding first mapping vector to be predicted, the second mapping vector to be predicted and the third mapping vector to be predicted, and then, the dot product result between the transposed vectors of the first mapping vector to be predicted and the second mapping vector to be predicted can be determined, so that the relevant semantics between the parameters inside the semantic vector to be predicted can be represented by the dot product result, and then, based on the dot product result, the third mapping vector to be predicted can be weighted summed to obtain the corresponding first significant semantic vector. In this way, since the relevant semantics are mined, the disturbance information that is not relevant can be ignored, so that the disturbance information can be removed to a certain extent; the mining process of the second significant semantic vector is the same; In the second step, the deposition parameter semantic vector can be fused into the first saliency semantic vector to obtain a corresponding first fused semantic vector. In other words, by further fusing the deposition parameter vector into the first saliency semantic vector, disturbance information therein can be further removed, thereby obtaining a reliable first fused semantic vector. In the third step, a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors can be fused into the second saliency semantic vector to obtain a corresponding second fused semantic vector. That is, by further fusing the deposition thickness adjustment vector into the second saliency semantic vector, the disturbance information therein can be further removed, thereby obtaining a reliable second fused semantic vector. Based on this, it can be seen that since the first saliency semantic vector and the second saliency semantic vector need to fuse with different other semantic vectors (respectively, the deposition parameter semantic vector and the deposition thickness adjustment vector), the semantic information that the first saliency semantic vector and the second saliency semantic vector need to focus on or represent is different. Based on this, it is necessary to mine the first mining branch and the second mining branch separately. In the fourth step, the first fused semantic vector and the second fused semantic vector are added to output a multi-dimensional fused vector corresponding to the deposited latent semantic vector.

[0067] Alternatively, for the second and third steps above, the specific manner of performing the corresponding semantic vector fusion is not limited. For example, in an alternative embodiment, in order to achieve reliable fusion of semantic vectors through multi-level fusion, the fusion of the deposition parameter semantic vector into the first saliency semantic vector is used as an example for explanation (combined with Figure 4 ): In the first step, multiple semantic extractions can be cascaded on the first saliency semantic vector to form corresponding multiple first extracted semantic vectors, wherein, for each semantic extraction, the first extracted semantic vector corresponding to the previous semantic extraction and the deposition parameter semantic vector are spliced, and the spliced ​​semantic vector is self-attention processed, and the semantic vector obtained by the self-attention processing is down-sampled (which can be achieved through convolution and / or pooling) to form the first extracted semantic vector corresponding to the current semantic extraction, and the first extracted semantic vector corresponding to the previous semantic extraction of the first semantic extraction is the first saliency semantic vector; exemplarily, in the process of the first semantic extraction, the first saliency semantic vector and the deposition parameter semantic vector can be spliced, and then self-attention processing and down-sampling processing are performed to obtain the first extracted semantic vector corresponding to the first semantic extraction; in the process of the second semantic extraction, the first extracted semantic vector corresponding to the first semantic extraction and the deposition parameter semantic vector can be spliced, and then self-attention processing and down-sampling processing are performed to obtain the first extracted semantic vector corresponding to the second semantic extraction; In the second step, the last first extracted semantic vector can be subjected to multiple cascaded semantic diffusions to form corresponding multiple first diffusion semantic vectors, wherein, for each semantic diffusion, the first diffusion semantic vector corresponding to the previous semantic diffusion and a first extracted semantic vector of the corresponding level are spliced, and the spliced ​​semantic vector is subjected to self-attention processing, and the semantic vector obtained by the self-attention processing is up-sampled (which can be achieved by transposed convolution and / or interpolation) to form the first diffusion semantic vector corresponding to the current semantic diffusion, and the first diffusion semantic vector corresponding to the previous semantic diffusion of the first semantic diffusion is the last first extracted semantic vector; exemplarily, in the process of the first semantic extraction, the last first extracted semantic vector and the last first extracted semantic vector can be spliced, and then self-attention processing and up-sampling processing are performed to obtain the first diffusion semantic vector corresponding to the first semantic diffusion; in the process of the second semantic extraction, the first diffusion semantic vector corresponding to the first semantic diffusion and the penultimate extracted semantic vector can be spliced, and then self-attention processing and down-sampling processing are performed to obtain the first diffusion semantic vector corresponding to the second semantic diffusion; In the third step, the last first diffusion semantic vector may be determined as the first fusion semantic vector.

[0068] In summary, the present invention provides a performance analysis method and system for a metallized thin film flexible capacitor. First, a target deposition thickness sequence formed during the coating process of the target capacitor is obtained; second, a first deposition thickness data and a comparative deposition thickness sequence are determined from the target deposition thickness sequence; then, a prediction process is performed based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence; finally, the predicted deposition thickness sequence and the comparative deposition thickness sequence are compared and analyzed to obtain the metallized film performance data of the target capacitor. Based on the above content, by predicting the corresponding deposition thickness sequence, it is possible to compare and analyze it as a theoretical value with the comparative deposition thickness sequence as an actual value, thereby achieving a comparison of the data of each deposition stage, that is, ensuring that the dynamic process of multiple deposition stages can be fully analyzed. Therefore, compared with the conventional solution of comparing the data after the deposition is completed, it can be more comprehensive, and therefore, the reliability can be higher, thereby improving the problem of relatively low reliability of capacitor performance analysis in the prior art. In addition, since the predicted corresponding deposition thickness sequence refers to the first deposition thickness data, the initial deposition thickness data of the predicted deposition thickness sequence and the comparison deposition thickness sequence can be made the same, that is, the initial conditions are the same, so that the comparison accuracy of the subsequent deposition thickness data can be higher, and more attention can be paid to the dynamic changes in the subsequent deposition process.

[0069] In the several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of the devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or can be implemented using a combination of dedicated hardware and computer instructions.

[0070] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0071] If the functions are implemented in the form of software modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or device. Without further constraints, an element defined by the phrase "comprises a..." does not preclude the existence of additional identical elements in the process, method, article or apparatus that includes the element.

[0072] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A performance analysis method for a metallized film flexible capacitor, characterized in that: include: Obtaining a target deposition thickness sequence formed during a coating process of a target capacitor, wherein the target deposition thickness sequence includes a plurality of deposition thickness data corresponding to a plurality of deposition stages, each deposition thickness data being used to characterize the thickness of at least one point of the metallized film deposited in the current stage; Determining first deposition thickness data and a comparative deposition thickness sequence from the target deposition thickness sequence, wherein each deposition thickness data in the comparative deposition thickness sequence is located after the first deposition thickness data; performing prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence, wherein each deposition thickness data in the predicted deposition thickness sequence is a prediction result for a deposition stage subsequent to the first deposition thickness data; The predicted deposition thickness sequence and the comparative deposition thickness sequence are compared and analyzed to obtain the metallized film performance data of the target capacitor.

2. The performance analysis method of the metallized film flexible capacitor according to claim 1, characterized in that: The step of performing prediction processing based on the first deposition thickness data and the coating deposition parameters of the metallized film to obtain a predicted deposition thickness sequence includes: Performing semantic mining on the first deposition thickness data to output A deposition thickness semantic vectors, wherein the deposition thickness semantic vectors are used to reflect potential semantic information of the first deposition thickness data; Determining a timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, and, based on the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, semantically adjusting the corresponding deposition thickness semantic vector in the A deposition thickness semantic vectors, and outputting A deposition thickness adjustment vectors, wherein each of the deposition thickness adjustment vectors is used to predict a deposition thickness data in a required predicted deposition thickness sequence, and the semantic adjustment is used to control a change between each deposition thickness data in the predicted deposition thickness sequence and the first deposition thickness data; Based on the coating deposition parameters, the A deposition thickness adjustment vectors are predicted and processed to output a corresponding predicted deposition thickness sequence, wherein the predicted deposition thickness sequence includes A deposition thickness data.

3. The performance analysis method of the metallized film flexible capacitor according to claim 2, characterized in that: The step of performing semantic mining on the first deposition thickness data and outputting A deposition thickness semantic vectors includes: Embedding the first deposition thickness data into a vector space to form a corresponding deposition thickness embedding vector, and performing self-attention processing on the deposition thickness embedding vector to form a corresponding deposition thickness semantic vector; The deposition thickness semantic vector is expanded to form A deposition thickness semantic vectors, wherein every two deposition thickness semantic vectors in the A deposition thickness semantic vectors are the same.

4. The performance analysis method of the metallized film flexible capacitor according to claim 2, characterized in that: The steps of determining a timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, and, based on the timing adjustment parameter corresponding to each of the deposition thickness semantic vectors, semantically adjusting the corresponding deposition thickness semantic vector among the A deposition thickness semantic vectors to output A deposition thickness adjustment vectors, include: Determining a preconfigured reference adjustment parameter, wherein the reference adjustment parameter is used to reflect the difference between adjacent deposition thickness data in the predicted deposition thickness sequence; Obtaining A training adjustment parameters carried in a target thickness prediction network, and determining A timing adjustment parameters based on the reference adjustment parameter and the A training adjustment parameters, wherein there is a one-to-one correspondence between the A timing adjustment parameters and the A deposition thickness semantic vectors, and the A training adjustment parameters are formed during the training process of the target thickness prediction network, which is a neural network model; Based on each of the A timing adjustment parameters, the corresponding deposition thickness semantic vector is mapped and adjusted to output A deposition thickness adjustment vectors.

5. The performance analysis method of the metallized film flexible capacitor according to claim 2, characterized in that: The step of predicting the A deposition thickness adjustment vectors based on the coating deposition parameters and outputting a corresponding predicted deposition thickness sequence includes: Determining A deposition latent semantic vectors, wherein the deposition latent semantic vectors are formed by perturbing the latent semantic vectors in the semantic space of the deposition thickness formed by training, the perturbed vectors obey a normal distribution, and one of the deposition latent semantic vectors is used to predict one deposition thickness data in a required predicted deposition thickness sequence; Performing semantic mining on the coating deposition parameters to form corresponding deposition parameter semantic vectors, wherein the deposition parameter semantic vectors are used to reflect semantic information of the coating deposition parameters, and the coating deposition parameters include at least one of coating distance, gas flow rate, coating temperature, coating bias, deposition atmosphere, sputtering power, and target current; The deposition parameter semantic vector and the A deposition latent semantic vectors are predicted and processed by the target thickness prediction network, and the A deposition thickness adjustment vectors are fused during the prediction process to output a corresponding predicted deposition thickness sequence, wherein the latent semantic vector in the semantic space where the deposition thickness is located is formed during the training process of the target thickness prediction network, which belongs to a neural network model.

6. The performance analysis method of the metallized film flexible capacitor according to claim 5, characterized in that: The step of predicting the deposition parameter semantic vector and the A deposition latent semantic vectors through a target thickness prediction network, fusing the A deposition thickness adjustment vectors during the prediction process, and outputting a corresponding predicted deposition thickness sequence includes: For each of the A deposited latent semantic vectors, determining a corresponding semantic vector to be predicted based on the deposited latent semantic vector; By using a semantic fusion model in the target thickness prediction network, the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors are fused into the semantic vector to be predicted, and a multi-dimensional fusion vector corresponding to the deposition latent semantic vector is output; The multi-dimensional fusion vector is predicted and processed by the prediction output model in the target thickness prediction network to output a deposition thickness data corresponding to the deposition latent semantic vector, and the A deposition thickness data corresponding to the A deposition latent semantic vectors are combined to form a corresponding predicted deposition thickness sequence.

7. The performance analysis method of the metallized film flexible capacitor according to claim 6, characterized in that: The step of determining, for each of the A deposited latent semantic vectors, a corresponding semantic vector to be predicted based on the deposited latent semantic vector, includes: Determining the first one of the A deposited latent semantic vectors as the corresponding semantic vector to be predicted; For each deposition latent semantic vector other than the first deposition latent semantic vector among the A deposition latent semantic vectors, splicing is performed based on the multi-dimensional fusion vector corresponding to the deposition latent semantic vector and the previous deposition latent semantic vector of the deposition latent semantic vector to form a corresponding spliced ​​semantic vector, and the spliced ​​semantic vector is downsampled to form a corresponding semantic vector to be predicted.

8. The performance analysis method of the metallized film flexible capacitor according to claim 6, characterized in that: The step of fusing the deposition parameter semantic vector and a deposition thickness adjustment vector corresponding to the deposition latent semantic vector in the A deposition thickness adjustment vectors into the semantic vector to be predicted through the semantic fusion model in the target thickness prediction network, and outputting a multi-dimensional fusion vector corresponding to the deposition latent semantic vector, includes: Performing significant feature mining on the semantic vector to be predicted through the first mining branch and the second mining branch included in the semantic fusion model in the target thickness prediction network, and outputting the corresponding first significant semantic vector and second significant semantic vector; fusing the deposition parameter semantic vector into the first saliency semantic vector to obtain a corresponding first fused semantic vector; fusing a deposition thickness adjustment vector corresponding to the deposition latent semantic vector among the A deposition thickness adjustment vectors into the second saliency semantic vector to obtain a corresponding second fused semantic vector; The first fused semantic vector and the second fused semantic vector are added to output a multi-dimensional fused vector corresponding to the deposited latent semantic vector.

9. The performance analysis method of the metallized film flexible capacitor according to claim 8, characterized in that: The step of fusing the deposition parameter semantic vector into the first saliency semantic vector to obtain a corresponding first fused semantic vector includes: Performing multiple cascaded semantic extractions on the first saliency semantic vector to form corresponding multiple first extracted semantic vectors, wherein, for each semantic extraction, the first extracted semantic vector corresponding to the previous semantic extraction and the deposition parameter semantic vector are spliced, and self-attention processing is performed on the spliced ​​semantic vector, and the semantic vector obtained by the self-attention processing is down-sampled to form a first extracted semantic vector corresponding to the current semantic extraction, and the first extracted semantic vector corresponding to the semantic extraction before the first semantic extraction is the first saliency semantic vector; Perform multiple cascaded semantic diffusions on the last first extracted semantic vector to form corresponding multiple first diffusion semantic vectors, wherein, for each semantic diffusion, the first diffusion semantic vector corresponding to the previous semantic diffusion and a first extracted semantic vector of the corresponding level are spliced, and the spliced ​​semantic vector is self-attention processed, and the semantic vector obtained by the self-attention processing is up-sampled to form the first diffusion semantic vector corresponding to the current semantic diffusion, and the first diffusion semantic vector corresponding to the semantic diffusion before the first semantic diffusion is the last first extracted semantic vector; The last first diffusion semantic vector is determined as the first fusion semantic vector.

10. A performance analysis system for a metallized film flexible capacitor, characterized in that: include: memory for storing computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the performance analysis method of the metallized thin film flexible capacitor according to any one of claims 1 to 9.

Images