Crude oil electric dehydration electrode film preparation method and electric field response film preparation method

Abstract

The invention provides a preparation method of a crude oil electric dehydration electrode film and a preparation method of an electric field response film, and the preparation method comprises the following steps: mixing a functional material with a solvent to form an initial dispersion liquid; mechanically stirring the initial dispersion liquid in an oil bath at normal temperature to obtain a uniform solution; uniformly mixing the solution; coating a copper sheet with the solution, and putting the copper sheet into a drying oven for curing treatment; measuring the dehydration rate; the functional material is selected from at least one of matrix resin, neutral resin, PEI polyetherimide, PMI polymethacrylimide foam and aldehyde ketone resin, and the concentration of the functional material is 0.1-0.2 g / ml; the solvent is N, N-dimethyl formamide (DMF). The method further comprises the steps of obtaining a crude oil sample moisture content historical data sequence, processing the sequence by adopting an autoregressive integral moving average model, extracting moisture content change trend feature vectors including a mean value, a variance and a trend slope, and obtaining a feature vector set; according to the optimized dispersion liquid proportioning scheme, a gradient optimization algorithm is adopted to calculate a mechanical stirring speed and duration sequence, a stirring parameter control sequence containing a plurality of time nodes is generated, stirring speed and duration are contained every fixed time interval, and a stirring control sequence is obtained; according to the uniform mixed solution, a temperature control curing method is adopted for treatment, a curing time and temperature curve is generated according to a pre-established material thermal characteristic curve, the temperature is increased to a preset temperature and kept for a fixed time length, and an initial cured film structure is obtained.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for preparing a crude oil electro-dehydration electrode thin film and a method for preparing an electric field responsive thin film. Background Technology

[0002] With the promotion of tertiary oil recovery technology in major oilfields across China, while increasing crude oil production, the composition of produced fluids has become more complex, and the dehydration of produced crude oil has become increasingly difficult. As oilfield development enters the high water-cut stage, the physicochemical properties of produced fluids undergo significant changes, specifically manifested in increased stability of crude oil emulsions and improved interfacial film strength, posing a severe challenge to traditional electrostatic dehydration processes. Industry statistics show that after onshore oilfields enter the later stages of development, the incidence of excessive water content in exported crude oil increases by 3-5 times compared to the early stages, resulting in a 15%-20% increase in energy consumption of the gathering and transportation system. Due to the unique conditions of crude oil production, the industry currently widely uses bare electrode electrostatic dehydrators without electrode films. This method has the following two problems: First, when processing crude oil emulsions with a water content exceeding 30%, the working voltage between electrodes needs to be increased to 1.5-2 times that of conventional operating conditions. At this time, local discharge is very likely to occur on the surface of the bare electrode, leading to severely uneven electric field distribution and even breakdown. Secondly, bare electrode electrostatic dehydrators without electrode films often experience a vicious cycle of "electric field collapse," which severely affects dehydration efficiency.

[0003] The dehydration electrode is made of stainless steel and is processed from a 2mm stainless steel plate. The edges are rounded with a radius of 2.5mm and burrs are removed to prevent tip discharge during the dehydration process. The dimensions are 40mm×40mm×2mm and the electrode spacing is 2cm. The dehydration electrode is connected with a polytetrafluoroethylene plate, which has the characteristics of high mechanical strength and good insulation performance.

[0004] The high-voltage power supply device used in the test can generate DC and pulse voltages. The DC output voltage is 0-15kV, the pulse output voltage is 500-5000Hz, and the duty cycle is 10%-80%.

[0005] Electric field distribution in emulsion

[0006] Assumptions: The thickness of the insulating layer is d1, and the dielectric constant is ε1; the width of the emulsion between the insulating electrodes is d2, and the dielectric constant is ε2 (ε1 > ε2).

[0007] According to the principle of potential continuity (D=εE), the electric field strengths in the insulating oil and the emulsion are respectively:

[0008]

[0009] The total voltage satisfies:

[0010]

[0011] Therefore, the electric displacement D is:

[0012]

[0013] The electric field strength E2 in the emulsion is:

[0014]

[0015] Therefore, preparing an insulating varnish film with excellent adhesion and coating it on the surface of the dehydration electrode is of great significance for improving the crude oil electro-dehydration process, and can also improve equipment stability and dehydration efficiency. Summary of the Invention

[0016] To overcome the challenge of increasing the inter-electrode operating voltage to 1.5-2 times that under normal operating conditions when processing crude oil emulsions with a water content exceeding 30%, which easily leads to partial discharge on the bare electrode surface and instability in the electric field strength distribution, resulting in a vicious cycle of "field collapse," this invention provides an insulating varnish film with excellent adhesion, which is then coated onto the surface of the dehydration electrode to improve the electric field strength experienced by the crude oil between the dehydration electrodes and the electro-dehydration efficiency of the crude oil.

[0017] In the process of crude oil sample processing, the accurate analysis of water content change trends and the preparation of electric field responsive films face a core technical problem: how to accurately control the electric field strength, dispersion ratio, stirring parameters and curing process under dynamically changing water content conditions in order to achieve the preparation of efficient dehydration films, while ensuring the matching of the electric field response characteristics of the films with the actual water content load.

[0018] This problem stems from the complexity and nonlinear fluctuations of historical water content data for crude oil samples. This makes it difficult for the mean, variance, and trend slope extracted from feature vectors to accurately represent dynamic changes, which in turn affects the accurate classification of electric field strength requirements by the support vector classification algorithm, resulting in unstable threshold division of low, medium, and high water content ranges.

[0019] Furthermore, the preset parameter range of the electric field-material mapping table may not be fully adaptable to the actual moisture content distribution, which limits the optimization of the dispersion ratio scheme, and makes it difficult to ensure the uniformity of the material distribution density by controlling the stirring speed and duration.

[0020] While real-time monitoring by online density sensors can detect deviations, iterative adjustments based on the gradient descent method may fail to quickly achieve a homogeneous mixture state due to slow convergence or getting stuck in local optima.

[0021] During the curing process, the preset temperature and time settings of the material's thermal characteristic curve may not match the actual electric field response requirements. The mean square error of the randomly sampled electric field response data is too high, reflecting the unstable performance of the initially cured film structure.

[0022] Ultimately, in the simulated dehydration environment test, the membrane dehydration efficiency may not meet expectations due to the dynamic changes in water content load, resulting in the prepared membrane being unable to operate stably and efficiently in actual crude oil dehydration scenarios.

[0023] The core of this problem lies in the insufficient dynamic coupling and real-time adaptability of parameters in multiple stages, involving multiple sub-problems such as feature extraction, classification threshold, material ratio, stirring control, curing process and dehydration efficiency, which together restrict the accuracy and applicability of thin film preparation.

[0025] This invention provides a method for preparing a thin film electrode for crude oil electro-dehydration, the method comprising the following steps:

[0026] S1. The functional material is mixed with a solvent to form an initial dispersion;

[0027] S2. The initial dispersion is mechanically stirred in an oil bath at room temperature to obtain a homogeneous solution;

[0028] S3. Mix the solution thoroughly;

[0029] S4. Apply the solution to the copper sheet and place it in an oven for curing.

[0030] S5. Measure the dehydration rate;

[0031] S6. The functional material is selected from at least one of the following: matrix resin, neutral resin, PEI polyetherimide, PMI polymethacrylimide foam, and aldehyde-ketone resin, and the concentration is 0.1-0.2 g / ml;

[0032] S7. The solvent is N,N-dimethylformamide (DMF).

[0033] Furthermore, the functional materials in step S1 are PMI (polymethacrylimide) and aldehyde-ketone resin.

[0034] Furthermore, the curing temperature in step S4 is 70-100℃, and the time is 70-100 minutes; then the temperature is increased to 110-140℃ at a rate of 10-20℃ / min, and held at that temperature for 70-100 minutes before cooling with the chamber.

[0035] Furthermore, in step S2, the mechanical stirring speed at room temperature is 200-600 rpm, and the processing time is 2-4 hours.

[0036] Furthermore, in step S2, the oil bath mechanical stirring speed is 200-600 rpm, the temperature is 100-150℃, and the processing time is 2-4 hours.

[0037] Furthermore, the copper sheet in step S4 must undergo acid washing, ultrasonic cleaning, and drying before use to obtain a clean substrate.

[0038] This invention also provides a method for preparing an electric field-responsive thin film based on the water content characteristics of crude oil, mainly comprising:

[0039] Historical data sequence of water content of crude oil samples is obtained, and the sequence is processed by an autoregressive integral moving average model to extract the feature vector of water content change trend, which includes mean, variance and trend slope, to obtain a set of feature vectors.

[0040] Based on the set of feature vectors, the electric field strength requirements are classified using a support vector classification algorithm to generate electric field strength grading threshold sets corresponding to low, medium, and high moisture content ranges, defined as low range, medium range, and high range, thus obtaining the electric field strength threshold set.

[0041] The threshold range of each interval is obtained from the set of electric field intensity thresholds. Based on the pre-established electric field-material mapping table, the functional material type and proportion parameter corresponding to the interval are queried. If the proportion parameter exceeds the preset range, it is adjusted to the nearest value within the range to obtain the optimized dispersion ratio scheme.

[0042] Based on the optimized dispersion ratio scheme, a gradient optimization algorithm is used to calculate the mechanical stirring speed and duration sequence, generating a stirring parameter control sequence containing multiple time nodes. The stirring speed and duration are included at fixed time intervals to obtain the stirring control sequence.

[0043] The stirring speed value at the current time node is obtained from the stirring control sequence, and the corresponding stirring operation is applied to the optimized dispersion. The material distribution density is monitored in real time by an online density sensor. If the density deviation exceeds a preset threshold, the speed value is iteratively adjusted based on the gradient descent method until the deviation converges to within the preset threshold to obtain a uniform mixture.

[0044] Based on the homogeneous mixture, a temperature-controlled curing method is used to process it. A curing time and temperature curve is generated according to the pre-established material thermal characteristic curve. The temperature is raised to a preset temperature and held for a fixed time to obtain the initial cured film structure.

[0045] Multiple regions are randomly sampled from the initial cured film structure, and the electric field response data at each point is measured. The relative mean square error between the response data and the electric field intensity threshold set is calculated. If the error exceeds the preset threshold, the curing time and temperature curve are adjusted based on the gradient optimization algorithm to reduce the temperature and extend the curing time to obtain the final electric field response film.

[0046] Based on the final electric field response film, a simulated dehydration environment test method is used. The water content load generated based on the distribution of the feature vector set is input, and the dehydration efficiency index is measured. It is defined as the ratio of the dehydrated amount to the initial water content. If the efficiency index reaches the preset standard, the film preparation is confirmed to be complete.

[0047] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0048] This invention discloses a method for preparing a thin film electrode for crude oil electro-dehydration. Compared with conventional bare electrode crude oil electro-dehydrators, in the process of electro-dehydration of crude oil with high water content, the bare electrode can no longer dehydrate and the electric field is unbalanced, while the thin film electrode can still dehydrate normally with a dehydration rate of over 99.5%.

[0049] Compared to conventional bare electrode crude oil electrostatic dehydrators, the membrane electrode dehydrator increases the electric field strength experienced by water droplets in the crude oil by 5 times and the maximum electric field density by 3 times, significantly improving the electrostatic dehydration efficiency.

[0050] This invention also discloses a method for preparing an electric field-responsive thin film based on the water content characteristics of crude oil, addressing the business scenario problem of how to dynamically optimize the dispersion ratio, stirring control, and curing process according to the water content to improve dehydration efficiency during crude oil dehydration. The method extracts water content trend features using an autoregressive integral moving average model, generates electric field strength grading thresholds using a support vector classification algorithm, maps these thresholds to the functional material ratio, and dynamically adjusts the stirring speed and duration using a gradient optimization algorithm to ensure the homogeneity of the mixture. Then, through temperature-controlled curing and error optimization, an electric field-responsive thin film is generated, and its efficiency is verified in a simulated dehydration environment. This invention significantly improves the adaptability of thin film preparation and dehydration efficiency through an integrated method of feature extraction, classification mapping, dynamic optimization, and error correction. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the experimental apparatus for preparing a crude oil electro-dehydration electrode thin film according to Example 1 of the present invention.

[0052] Figure 2 This is a schematic diagram of the experimental apparatus for preparing a crude oil electro-dehydration electrode thin film according to Example 2 of the present invention.

[0053] Figure 3 This is a schematic diagram of the experimental apparatus for preparing a crude oil electro-dehydration electrode thin film according to Example 3 of the present invention.

[0054] Figure 4 This is a flowchart of an electric field responsive thin film preparation method based on the water content characteristics of crude oil according to the present invention.

[0055] Figure 5This is a schematic diagram of an electric field responsive thin film preparation method based on the water content characteristics of crude oil according to the present invention.

[0056] Figure 6 This is another schematic diagram of an electric field responsive thin film preparation method based on the water content characteristics of crude oil according to the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Example 1

[0059] like Figure 1 The present invention provides a method for preparing a crude oil electro-dehydration electrode thin film, the preparation method comprising the following steps:

[0060] First, preheat the oven to 80℃. Then, evenly apply the prepared solution onto the copper sheet and place it stably into the oven. Keep the copper sheet horizontal during the curing process. Before use, the copper sheet should be acid-washed, ultrasonically cleaned, and dried to obtain a clean substrate.

[0061] Based on this, keep the copper sheet in the oven for 80 minutes, then raise the oven temperature to 120°C at a rate of 10-20°C / min, continue to keep it for 80 minutes, then turn off the oven and let it cool to room temperature before taking it out.

[0062] Example 2

[0063] Electrode films were prepared using a method for preparing electrode films for crude oil electro-dehydration.

[0064] The steps are as follows:

[0065] Using an electronic balance, measure polymethacrylimide (PMI) granules and aldehyde-ketone resin in a 1:1 mass ratio; then pour the aldehyde-ketone resin and PMI granules into two separate three-necked flasks; then add the resin to the two flasks again.

[0066] The DMF (N,N-dimethylformamide) was calculated using the following formula based on the PMI solubility of 0.11 g / mL.

[0067] VDMF = 9MPMI

[0068] The three-necked flask containing aldehyde-ketone resin and DMF was mechanically stirred at room temperature at 200 rpm for 2 hours; the three-necked flask containing PMI particles was mechanically stirred at 200 rpm in an oil bath at 120°C for 2 hours to fully dissolve the PMI particles and bring them to a liquid state.

[0069] Pour the solution from the three-necked flask containing the aldehyde-ketone resin and DMF into the three-necked flask containing the PMI granules and DMF mixture, and continue mechanical stirring for 2 hours at 200 rpm to ensure complete dissolution. Then pour the mixture into a sealed container for storage.

[0070] like Figure 2 As shown

[0071] First, preheat the oven to 80℃. Then, evenly apply the prepared solution onto the copper sheet and place it stably into the oven. Keep the copper sheet horizontal during the curing process. Before use, the copper sheet should be acid-washed, ultrasonically cleaned, and dried to obtain a clean substrate.

[0072] Based on this, keep the copper sheet in the oven for 80 minutes, then raise the oven temperature to 120°C at a rate of 10-20°C / min, continue to keep it for 80 minutes, then turn off the oven and let it cool to room temperature before taking it out.

[0073] Example 3

[0074] Electrode films were prepared using a method for preparing electrode films for crude oil electro-dehydration.

[0075] The steps are as follows:

[0076] Using an electronic balance, measure polymethacrylimide (PMI) particles and aldehyde-ketone resin at a mass ratio of 1:2. Then, pour the aldehyde-ketone resin and polymethacrylimide (PMI) particles into two separate three-necked flasks. Calculate the DMF (N,N-dimethylformamide) in each flask using the following formula, based on the solubility of PMI (0.11 g / mL).

[0077] V DMF =9M PMI

[0078] The three-necked flask containing aldehyde-ketone resin and DMF was mechanically stirred at 200 rpm at room temperature for 2 hours. The three-necked flask containing PMI granules was mechanically stirred at 200 rpm in an oil bath at 120°C for 2 hours to ensure complete dissolution and liquid state.

[0079] Pour the solution from the three-necked flask containing the aldehyde-ketone resin and DMF into the three-necked flask containing the PMI granules and DMF mixture, and continue mechanical stirring for 2 hours at a stirring speed of 200 rpm. Allow it to dissolve completely, then pour into a sealed container for storage.

[0080] like Figure 3 As shown

[0081] First, preheat the oven to 80℃. Then, evenly apply the prepared solution onto the copper sheet and place it stably into the oven. Keep the copper sheet horizontal during the curing process. Before use, the copper sheet should be acid-washed, ultrasonically cleaned, and dried to obtain a clean substrate.

[0082] Based on this, keep the copper sheet in the oven for 80 minutes, then raise the oven temperature to 120°C at a rate of 10-20°C / min, continue to keep it for 80 minutes, then turn off the oven and let it cool to room temperature before taking it out.

[0083] The embodiments described above are merely preferred embodiments of the present invention, and not all feasible embodiments of the present invention. Any obvious modifications made by those skilled in the art without departing from the principles and spirit of the present invention should be considered to be included within the scope of protection of the claims of the present invention.

[0084] Example 4

[0085] The material obtained in Example 1 was coated onto the electrode with a thickness of 0.2 mm.

[0086] The electrode spacing is 2cm, and the electrode thickness is 2mm;

[0087] The voltage across the electrode is 0.1 kV / mm, and the voltage type is pulse voltage.

[0088] The processing temperature was 55℃, and the time was 80 minutes.

[0089] The final dehydration rate was 91.62%.

[0090] Example 5

[0091] The material obtained in Example 1 was coated onto the electrode at a thickness of 0.6 mm;

[0092] The electrode spacing is 2cm, and the electrode thickness is 2mm;

[0093] The voltage across the electrode is 0.5 kV / mm, and the voltage type is pulse voltage.

[0094] The processing temperature was 55℃, and the time was 80 minutes.

[0095] The final dehydration rate was 98.33%.

[0096] Example 5

[0097] The material obtained in Example 1 was coated onto the electrode with a thickness of 1.0 mm.

[0098] The electrode spacing is 2cm, and the electrode thickness is 2mm;

[0099] The voltage across the electrode is 0.5 kV / mm, and the voltage type is pulse voltage.

[0100] The processing temperature was 55℃, and the time was 80 minutes.

[0101] The final dehydration rate was 99.53%.

[0102] like Figure 4-6 As shown in the following embodiments, a method for preparing an electric field-responsive thin film based on the water content characteristics of crude oil is disclosed, which may specifically include:

[0103] Step S101: Obtain historical data sequence of water content of crude oil samples, process the sequence using an autoregressive integral moving average model, extract feature vectors of water content change trend, including mean, variance and trend slope, to obtain a set of feature vectors.

[0104] Historical data sequences of crude oil sample water content are obtained from the storage system. These historical data sequences include multiple timestamps and corresponding water content values, resulting in the raw data sequence. The raw data sequence is preprocessed to remove outliers and missing values, resulting in a cleaned data sequence. An autoregressive integral moving average model is used to fit the cleaned data sequence, generating a smoothed data sequence. The trend is extracted from the smoothed data sequence, and the mean and variance within a continuous time window are calculated, resulting in a mean sequence and a variance sequence. Based on the mean and variance sequences, a linear regression method is used to calculate the slope of the trend, obtaining a slope value. If the slope value exceeds a preset threshold, a sliding window analysis is used to determine the significance of the trend change, obtaining a significance index. Based on the mean sequence, variance sequence, and slope value, a feature vector set containing the mean, variance, and slope is constructed, resulting in the final feature vector set.

[0105] For example, a historical data sequence of water content of crude oil samples is obtained from a storage system. This historical data sequence includes multiple timestamps and corresponding water content values, resulting in a raw data sequence. The raw data sequence is preprocessed to remove outliers and missing values, resulting in a cleaned data sequence. An autoregressive integral moving average model is used to fit the cleaned data sequence, generating a smoothed data sequence. The trend is extracted from the smoothed data sequence, and the mean and variance within a continuous time window are calculated, resulting in a mean sequence and a variance sequence. Based on the mean and variance sequences, a linear regression method is used to calculate the slope of the trend, obtaining a slope value. If the slope value exceeds a preset threshold, the significance of the trend change is determined through sliding window analysis, obtaining a significance index. Based on the mean sequence, variance sequence, and slope value, a feature vector set containing the mean, variance, and slope is constructed, resulting in the final feature vector set.

[0106] Step S102: Based on the feature vector set, the electric field intensity requirement is classified using the support vector classification algorithm to generate electric field intensity graded threshold sets corresponding to low, medium and high water content ranges, defined as low range, medium range and high range, thus obtaining the electric field intensity threshold set.

[0107] Mean, variance, and slope data are extracted from the feature vector set to construct a classification input dataset. The classification input dataset is then trained using a support vector classification algorithm to generate a classification model for water content intervals. This classification model is used to divide the water content intervals, determining the boundaries of low, medium, and high water content intervals, resulting in interval division results. If the boundary values ​​of the interval division results are inconsistent with preset water content thresholds, the parameters of the classification model are adjusted using cross-validation to obtain an optimized classification model. Based on the optimized classification model, the electric field intensity values ​​corresponding to the low, medium, and high water content intervals are calculated, resulting in an electric field intensity threshold set. Low, medium, and high interval thresholds are extracted from this electric field intensity threshold set to construct a hierarchical threshold mapping table, resulting in the hierarchical threshold mapping table. Using this hierarchical threshold mapping table, a correspondence between water content intervals and electric field intensity values ​​is generated, resulting in an interval intensity mapping set.

[0108] For example, when extracting mean, variance, and slope data from a set of feature vectors, we can envision a crude oil sample water content monitoring system. The system stores feature vector sets for multiple time windows, each containing the mean, variance, and trend slope for a given time period. Suppose a certain time window has a mean of 35%, a variance of 2.5, and a slope of 0.03, reflecting a slow upward trend in water content. These data are extracted through database queries to form a categorical input dataset. During data preprocessing, outliers, such as invalid data with a mean exceeding 100% or a negative slope, are removed to ensure the reliability of the dataset.

[0109] In one possible implementation, when training the input dataset using a support vector classification algorithm, a radial basis function kernel can be selected to map the data to a high-dimensional space to distinguish between moisture content ranges. During training, the input dataset contains 1000 samples, each consisting of a mean, variance, and slope, labeled as low, medium, or high moisture content categories. The classification model divides the samples into moisture content ranges by optimizing the hyperplane, for example, 0-30% for the low range, 30-60% for the medium range, and 60-100% for the high range. After training, the model outputs classification boundary values, such as 30.5% for the low-medium range and 59.8% for the medium-high range.

[0110] Specifically, if the boundary values ​​of the classification model are inconsistent with the preset threshold (e.g., 28% for the low-to-medium range), parameters can be adjusted through cross-validation. Using 5-fold cross-validation, the classification accuracy of the model on different subsets is evaluated, and the regularization parameter C and kernel function parameter γ are adjusted to optimize model performance. The optimized model then re-divides the intervals to obtain boundary values ​​closer to the preset threshold, such as adjusting the low-to-medium range to 28.2%.

[0111] In one embodiment, when calculating the electric field strength value, an optimized classification model is used, combined with the physical relationship between water content and electric field strength, to determine the electric field strength threshold for each interval. It is assumed that the low interval corresponds to an electric field strength of 100 V / m, the middle interval to 150 V / m, and the high interval to 200 V / m. These thresholds are obtained by mapping experimental data, constructing a graded threshold mapping table. The mapping table records the correspondence between water content intervals and electric field strength, such as 0-28.2% corresponding to 100 V / m, and 28.2-59.8% corresponding to 150 V / m.

[0112] For example, when generating the final interval intensity mapping set, the mapping table can be stored as structured data for subsequent real-time monitoring. In crude oil dehydration equipment, the system automatically adjusts the electric field strength according to the mapping set. When a water content of 40% is detected, the electric field strength of the intermediate interval is set to 150V / m. This mapping method facilitates dynamic equipment control and improves dehydration efficiency. The mapping set can also be used for trend prediction to analyze the long-term impact of water content changes on the electric field strength.

[0113] In one possible implementation, if the scheme needs to be expanded, a time-weighted factor can be introduced to combine the time characteristics of historical data sequences, thereby enhancing the model's sensitivity to recent data.

[0114] For example, the mean and variance of recent data are given higher weights, and the slope calculation considers a shorter time window, thus reflecting the trend of moisture content changes more accurately. This approach improves the adaptability of the classification model, making it particularly suitable for scenarios with large fluctuations in moisture content.

[0115] Understandably, the construction of the grading threshold mapping table needs to be updated regularly to adapt to changes in crude oil properties.

[0116] For example, in the later stages of oilfield extraction, water cut generally increases, requiring recalibration of the mapping table to ensure the accuracy of the electric field strength threshold. This dynamic adjustment mechanism ensures the system's applicability under different operating conditions and improves the stability of the crude oil processing process.

[0117] Step S103: Obtain the threshold range of each interval from the set of electric field strength thresholds. Based on the pre-established electric field-material mapping table, query the functional material type and proportion parameter corresponding to the interval. If the proportion parameter exceeds the preset range, adjust it to the nearest value within the range to obtain the optimized dispersion ratio scheme.

[0118] The threshold ranges for each interval are obtained from the set of electric field strength thresholds to construct an interval threshold dataset. Based on this dataset, a pre-established electric field-material mapping table is queried to match the corresponding functional material type and proportion parameters, resulting in an initial material ratio set. If the proportion parameter values ​​in the initial material ratio set exceed a preset threshold range, they are adjusted to the nearest value within the range using linear interpolation, resulting in an adjusted material ratio set. Using the adjusted material ratio set and a dispersion composition database, the content of each component in the optimized ratio scheme is calculated, yielding dispersion ratio parameters. Based on these parameters, a ratio scheme generation record is generated and stored in the ratio scheme database, resulting in a ratio scheme storage result. The consistency between the optimized ratio scheme and the interval threshold range is verified based on the storage result. If inconsistent, the ratio parameters are adjusted using an iterative optimization algorithm to obtain the final optimized ratio scheme. Functional material types and proportion parameters are extracted from the final optimized ratio scheme to generate ratio scheme execution instructions, resulting in a ratio scheme execution instruction set.

[0119] For example, threshold ranges for each interval are obtained from the electric field strength threshold set to construct an interval threshold dataset. Based on the interval threshold dataset, a pre-established electric field-material mapping table is queried to match the corresponding functional material type and proportion parameters, resulting in an initial material ratio set. If the proportion parameter value in the initial material ratio set exceeds a preset range threshold, it is adjusted to the nearest value within the range using a linear interpolation method, resulting in an adjusted material ratio set. Using the adjusted material ratio set and a dispersion composition database, the content of each component in the optimized ratio scheme is calculated, resulting in dispersion ratio parameters. Based on the dispersion ratio parameters, a ratio scheme generation record is generated and stored in the ratio scheme database, resulting in a ratio scheme storage result. Based on the ratio scheme storage result, the consistency between the optimized ratio scheme and the interval threshold range is verified. If they are inconsistent, the ratio parameters are adjusted using an iterative optimization algorithm to obtain the final optimized ratio scheme. Functional material types and proportion parameters are extracted from the final optimized ratio scheme to generate ratio scheme execution instructions, resulting in a ratio scheme execution instruction set.

[0120] Step S104: Based on the optimized dispersion ratio scheme, a gradient optimization algorithm is used to calculate the mechanical stirring speed and duration sequence, generating a stirring parameter control sequence containing multiple time nodes. The stirring speed and duration are included at fixed time intervals to obtain the stirring control sequence.

[0121] Component ratio parameters are obtained from the optimized dispersion formulation scheme to construct a formulation parameter dataset. Based on this dataset, the mechanical stirring speed and duration at each time point are calculated using a gradient descent algorithm to generate an initial stirring parameter sequence. If the stirring speed in the initial stirring parameter sequence exceeds a preset range, it is adjusted to the nearest value within the range using linear interpolation, resulting in an adjusted stirring parameter sequence. Based on the adjusted stirring parameter sequence and the time point sequence, a stirring parameter control sequence containing fixed time intervals is generated, resulting in a stirring parameter control sequence. The stirring speed and duration at each time point are extracted from the stirring parameter control sequence, and their consistency with the formulation parameter dataset is verified. If inconsistent, the stirring parameters are iteratively optimized using Newton's method to obtain an optimized stirring parameter sequence. Based on the optimized stirring parameter sequence, a stirring control command sequence is generated and stored in a stirring control database, resulting in a stirring control command storage result. The final stirring control sequence is extracted from the stirring control command storage result to generate equipment execution commands, resulting in a set of equipment execution commands.

[0122] For example, component ratio parameters are obtained from the optimized dispersion formulation scheme to construct a formulation parameter dataset. Based on this dataset, the mechanical stirring speed and duration at each time point are calculated using a gradient descent algorithm to generate an initial stirring parameter sequence. If the stirring speed in the initial stirring parameter sequence exceeds a preset range, it is adjusted to the nearest value within the range using linear interpolation, resulting in an adjusted stirring parameter sequence. Based on the adjusted stirring parameter sequence and the time point sequence, a stirring parameter control sequence containing fixed time intervals is generated, resulting in a stirring parameter control sequence. The stirring speed and duration at each time point are extracted from the stirring parameter control sequence, and their consistency with the formulation parameter dataset is verified. If inconsistent, the stirring parameters are iteratively optimized using Newton's method to obtain an optimized stirring parameter sequence. Based on the optimized stirring parameter sequence, a stirring control command sequence is generated and stored in a stirring control database, resulting in a stirring control command storage result. The final stirring control sequence is extracted from the stirring control command storage result to generate equipment execution commands, resulting in a set of equipment execution commands.

[0123] Step S105: Obtain the stirring speed value of the current time node from the stirring control sequence, apply the corresponding stirring operation to the optimized dispersion, monitor the material distribution density in real time through an online density sensor, and if the density deviation exceeds a preset threshold, iteratively adjust the speed value based on the gradient descent method until the deviation converges to within the preset threshold to obtain a uniform mixture.

[0124] The stirring speed value at the current time point is obtained from the stirring control sequence, and material distribution density data is collected in real time using an online density sensor to obtain a material distribution density dataset. If the density deviation in the material distribution density dataset exceeds a preset threshold, the loss function L is defined as the root mean square error between the density value and the target uniform density, the learning rate is initialized to α = 0.01, and the iteration update speed V is set to V. new =V old -αdL / dv, where dL / dv is the derivative of loss with respect to velocity, is repeated until L is below a threshold to obtain the adjusted stirring speed value. Based on the adjusted stirring speed value, the stirring speed value at the current time node in the stirring control sequence is updated to obtain the updated stirring control sequence. The stirring speed value at the next time node is extracted from the updated stirring control sequence. After applying this stirring speed value, material distribution density data is collected again by an online density sensor to obtain a new material distribution density dataset. If the density deviation of the new material distribution density dataset still exceeds the preset threshold, the minimize function in Python's SciPy library is used to iteratively optimize the stirring speed value using the 'Newton-CG' method. The input objective function is the sum of squares of the density deviation, and the initial guess is the current velocity, to obtain the optimized stirring speed value. Based on the optimized stirring speed value, a stirring equipment control command sequence is generated, and a set of equipment execution commands is generated to obtain the equipment execution command set.

[0125] For example, the stirring speed value at the current time point is obtained from the stirring control sequence, and material distribution density data is collected in real time using an online density sensor to obtain a material distribution density dataset. If the density deviation in the material distribution density dataset exceeds a preset threshold, the loss function L is defined as the root mean square error between the density value and the target uniform density, the learning rate is initialized to α = 0.01, and the iteration update speed V is set to V. new =V old -αdL / dv, where dL / dv is the derivative of loss with respect to velocity, is repeated until L is below a threshold to obtain the adjusted stirring speed value. Based on the adjusted stirring speed value, the stirring speed value at the current time node in the stirring control sequence is updated to obtain the updated stirring control sequence. The stirring speed value at the next time node is extracted from the updated stirring control sequence. After applying this stirring speed value, material distribution density data is collected again by an online density sensor to obtain a new material distribution density dataset. If the density deviation of the new material distribution density dataset still exceeds the preset threshold, the minimize function in Python's SciPy library is used to iteratively optimize the stirring speed value using the 'Newton-CG' method. The input objective function is the sum of squares of the density deviation, and the initial guess is the current velocity, to obtain the optimized stirring speed value. Based on the optimized stirring speed value, a stirring equipment control command sequence is generated, and a set of equipment execution commands is generated to obtain the equipment execution command set.

[0126] Step S106: Based on the homogeneous mixture, a temperature-controlled curing method is used to process it. A curing time and temperature curve is generated according to the pre-established material thermal characteristic curve. The temperature is raised to a preset temperature and held for a fixed time to obtain the initial cured film structure.

[0127] Temperature data during the curing process is collected in real time using an online temperature sensor to obtain a temperature dataset. If the deviation between the temperature value in the dataset and the preset temperature exceeds a preset threshold, the heating power is iteratively adjusted using the gradient descent method. The loss function L is defined as the mean square error between the temperature value and the preset temperature, where the temperature value refers to the actual temperature collected, and the preset temperature refers to the target temperature. The initial learning rate α is 0.01. Linear regression from the NumPy library is used to fit the relationship between the temperature dataset and the current power, and the derivative of the loss function with respect to power, dL / dP, is calculated, where dL / dP equals twice the mean temperature deviation multiplied by the fitting slope. The power P_new is iteratively updated to P_old minus α multiplied by dL / dP, repeated until the loss function value is below the threshold, yielding the adjusted heating power. Based on the adjusted heating power, the temperature control sequence is updated, generating a new curing time and temperature curve, resulting in the updated temperature control sequence. The temperature value at the current time point is extracted from the updated temperature control sequence, and the corresponding temperature is applied through the heating device. The curing state data of the thin film structure is collected in real time to obtain the curing state dataset. If the structural uniformity in the cured state dataset is lower than the preset standard, the `minimize` function from Python's SciPy library is used to optimize the curing time using the BFGS method. The input objective function is the sum of squared variances of the structure image pixels calculated using the OpenCV library, with the current time as the initial guess, resulting in the optimized curing time. Based on the optimized curing time, the curing time-temperature curve is adjusted to generate a new sequence of heating device control commands, resulting in a set of device execution commands. The heating device is controlled using this set of commands, applying the adjusted temperature and time to obtain the final cured thin film structure.

[0128] For example, initial material parameters are obtained from a homogeneous mixture to obtain an initial material dataset. Temperature data during the curing process is collected in real time using an online temperature sensor to obtain a temperature dataset. If the temperature value in the temperature dataset deviates from the preset temperature by more than a preset threshold, the heating power is iteratively adjusted using the gradient descent method. The loss function is defined as the mean square error between the temperature value and the preset temperature. The learning rate is initialized, and the heating power is iteratively updated to obtain the adjusted heating power. Based on the adjusted heating power, the temperature control sequence is updated to generate a new curing time and temperature curve, resulting in an updated temperature control sequence. The temperature value at the current time node is extracted from the updated temperature control sequence, and the corresponding temperature is applied through the heating device. The curing state data of the thin film structure is collected in real time to obtain a curing state dataset. If the structural uniformity in the curing state dataset is lower than a preset standard, the minimum function in the SciPy library is used to optimize the curing time using the BFGS method. The input objective function is the sum of squares of the structural uniformity deviation, and the initial guess is the current time, resulting in an optimized curing time. Based on the optimized curing time, the curing time and temperature curve are adjusted to generate a new heating device control command sequence, resulting in a set of device execution commands. The heating equipment is controlled by a set of instructions executed by the device, and the adjusted temperature and time are applied to obtain the final cured film structure.

[0129] Step S107: Randomly sample multiple regions from the initial cured film structure, measure the electric field response data at each point, calculate the relative mean square error between the response data and the electric field intensity threshold set, and if the error exceeds the preset threshold, adjust the curing time and temperature curve based on the gradient optimization algorithm, reduce the temperature and extend the duration to obtain the final electric field response film.

[0130] Multiple regions are randomly selected from the initial cured film structure, and electric field response data are collected at each point using an electric field sensor to obtain an electric field response dataset. Based on this dataset, the mean square error (MSE) between the response data at each point and a preset electric field intensity threshold set is calculated to obtain an error dataset. If the MSE in the error dataset exceeds the preset threshold, the gradient descent method is used to iteratively optimize the curing time and temperature curves. The loss function L is calculated as the sum of squared MSEs, the learning rate α is initialized, and the curing time and temperature parameters are updated to obtain an optimized parameter set. Based on the optimized parameter set, a new curing time and temperature curve is generated to obtain an updated control sequence. Using the updated control sequence, the temperature and curing time applied by the heating device are adjusted, and the electric field response data of the adjusted film is collected to obtain a new electric field response dataset. The MSE is calculated from the new electric field response dataset. If the error still exceeds the preset threshold, the minimum function of the SciPy library is used with the BFGS method to further optimize the curing time. The objective function is the variance of the electric field response data, and the final curing time is obtained. Based on the final curing time, the heating device control command is updated, and the adjusted temperature and time are applied to obtain an optimized electric field response film.

[0131] For example, multiple regions are randomly selected from the initial cured film structure, and electric field response data are collected at each point using an electric field sensor to obtain an electric field response dataset. Based on the electric field response dataset, the mean square error (MSE) between the response data at each point and a preset electric field intensity threshold set is calculated to obtain an error dataset. If the MSE in the error dataset exceeds the preset threshold, the gradient descent method is used to iteratively optimize the curing time and temperature curves. The loss function L is calculated as the sum of squared MSEs, the learning rate α is initialized, and the curing time and temperature parameters are updated to obtain an optimized parameter set. Based on the optimized parameter set, a new curing time and temperature curve is generated to obtain an updated control sequence. Using the updated control sequence, the temperature applied by the heating device and the curing time are adjusted, and the electric field response data of the adjusted film is collected to obtain a new electric field response dataset. The MSE is calculated from the new electric field response dataset. If the error still exceeds the preset threshold, the minimum function of the SciPy library is used with the BFGS method to further optimize the curing time. The objective function is the variance of the electric field response data, and the final curing time is obtained. Based on the final curing time, the heating equipment control command is updated, and the adjusted temperature and duration are applied to obtain an optimized electric field response film.

[0132] Step S108: Based on the final electric field response film, a simulated dehydration environment test method is used. The water content load generated based on the distribution of the feature vector set is input, and the dehydration efficiency index is measured. It is defined as the ratio of the dehydrated amount to the initial water content. If the efficiency index reaches the preset standard, the film preparation is confirmed to be complete.

[0133] Multiple points are randomly selected from the thin film, and electric field response data are collected at each point using an electric field sensor to obtain the first electric field response dataset. Based on the first electric field response dataset, the average electric field intensity and variance of each point are calculated to generate the first eigenvector set. Principal component analysis is performed on the first eigenvector set using the PCA function of the sklearn library to extract the first principal eigenvectors related to water content, resulting in the first principal eigenvector set. The Euclidean norm of each vector is calculated using the first principal eigenvector set to construct the first water content loading sequence. The dehydration environment is simulated using ANSYS software, and the first water content loading sequence is input to obtain the first dehydration test dataset. From the first dehydration test dataset, the first dehydration efficiency index is calculated. The first dehydration efficiency index is defined as the dimensionless ratio of the dehydrated amount D divided by the initial water content I, i.e., D / I, where D is the mass of water reduced in the simulation and I is the initial water mass. The first dehydration efficiency index is judged to be greater than a preset threshold of 0.8 to obtain the first efficiency judgment result. If the first efficiency determination result is not greater than the preset threshold of 0.8, the `minimize` function of the `scipy.optimize` library is used to optimize the first moisture content loading sequence based on gradient descent, updating the first principal feature vector set to obtain the second moisture content loading sequence. Using the second moisture content loading sequence, a simulated dehydration environment test is re-executed in ANSYS software, collecting the second dehydration test dataset, calculating the second dehydration efficiency index, and determining whether the second dehydration efficiency index is greater than the preset threshold of 0.8 to obtain the second efficiency determination result. Based on the second moisture content loading sequence where the second efficiency determination result is greater than the preset threshold of 0.8, the film preparation parameters are adjusted, and the final film preparation is confirmed to be complete.

[0134] For example, multiple points are randomly selected from the final electric field response film, and electric field response data are collected at each point using an electric field sensor to obtain an electric field response dataset. Based on the electric field response dataset, a set of feature vectors is generated, and principal component analysis is used to extract the principal feature vectors related to the water content load, resulting in a set of principal feature vectors. A water content load sequence is constructed using the set of principal feature vectors, and a simulated dehydration environment test method is used as the input to obtain a dehydration test dataset. From the dehydration test dataset, a dehydration efficiency index is calculated, defined as the ratio of dehydrated water to initial water content. The efficiency index is then judged to meet a preset threshold, yielding an efficiency judgment result. If the efficiency judgment result does not meet the preset threshold, the gradient descent method is used to optimize the water content load sequence, updating the set of principal feature vectors to obtain an optimized load sequence. Using the optimized load sequence, the simulated dehydration environment test is re-executed, a new dehydration test dataset is collected, and the dehydration efficiency index is calculated to obtain an updated efficiency index. If the updated efficiency index meets the preset threshold, the film preparation parameters are adjusted according to the optimized load sequence, confirming the completion of the final film preparation.

[0135] The above are only some preferred embodiments of the present invention, but the present invention is not limited thereto, and many improvements and modifications can be made. Any improvements and modifications made based on the basic principles of the present invention should be considered to fall within the protection scope of the present invention.

Claims

1. A method for preparing a thin film electrode for crude oil electro-dehydration, characterized in that, The preparation method includes the following steps: S1. The functional material is mixed with a solvent to form an initial dispersion; S2. The initial dispersion is mechanically stirred in an oil bath at room temperature to obtain a homogeneous solution; S3. Mix the solution thoroughly; S4. Apply the solution to the copper sheet and place it in an oven for curing. S5. Measure the dehydration rate; S6. The functional material is selected from at least one of the following: matrix resin, neutral resin, PEI polyetherimide, PMI polymethacrylimide foam, and aldehyde-ketone resin, and the concentration is 0.1-0.2 g / ml; S7. The solvent is N,N-dimethylformamide (DMF).

2. A method for preparing an electric field-responsive thin film, characterized in that, The method includes: Historical data sequence of water content of crude oil samples is obtained, and the sequence is processed by an autoregressive integral moving average model to extract the feature vector of water content change trend, which includes mean, variance and trend slope, to obtain a set of feature vectors. Based on the set of feature vectors, the electric field strength requirements are classified using a support vector classification algorithm to generate electric field strength grading threshold sets corresponding to low, medium, and high moisture content ranges, defined as low range, medium range, and high range, thus obtaining the electric field strength threshold set. The threshold range of each interval is obtained from the set of electric field intensity thresholds. Based on the pre-established electric field-material mapping table, the functional material type and proportion parameter corresponding to the interval are queried. If the proportion parameter exceeds the preset range, it is adjusted to the nearest value within the range to obtain the optimized dispersion ratio scheme. Based on the optimized dispersion ratio scheme, a gradient optimization algorithm is used to calculate the mechanical stirring speed and duration sequence, generating a stirring parameter control sequence containing multiple time nodes. The stirring speed and duration are included at fixed time intervals to obtain the stirring control sequence. The stirring speed value at the current time node is obtained from the stirring control sequence, and the corresponding stirring operation is applied to the optimized dispersion. The material distribution density is monitored in real time by an online density sensor. If the density deviation exceeds a preset threshold, the speed value is iteratively adjusted based on the gradient descent method until the deviation converges to within the preset threshold to obtain a uniform mixture. Based on the homogeneous mixture, a temperature-controlled curing method is used to process it. A curing time and temperature curve is generated according to the pre-established material thermal characteristic curve. The temperature is raised to a preset temperature and held for a fixed time to obtain the initial cured film structure. Multiple regions are randomly sampled from the initial cured film structure, and the electric field response data at each point is measured. The relative mean square error between the response data and the electric field intensity threshold set is calculated. If the error exceeds the preset threshold, the curing time and temperature curve are adjusted based on the gradient optimization algorithm to reduce the temperature and extend the curing time to obtain the final electric field response film. Based on the final electric field response film, a simulated dehydration environment test method is used. The water content load generated based on the distribution of the feature vector set is input, and the dehydration efficiency index is measured. It is defined as the ratio of the dehydrated amount to the initial water content. If the efficiency index reaches the preset standard, the film preparation is confirmed to be complete.

3. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The historical data sequence of water content in crude oil samples is obtained, and the sequence is processed using an autoregressive integral moving average model to extract a feature vector of water content change trend, which includes the mean, variance, and trend slope, resulting in a feature vector set including: The historical data sequence of water content of crude oil samples is obtained from the storage system. The historical data sequence includes multiple timestamps and corresponding water content values ​​to obtain the original data sequence. The original data sequence is preprocessed to remove outliers and missing values, resulting in a cleaned data sequence. An autoregressive integral moving average model is used to fit the cleaned data sequence to generate a smooth data sequence; Extract the trend of change from the smoothed data sequence, calculate the mean and variance within a continuous time window, and obtain the mean sequence and variance sequence; Based on the mean sequence and variance sequence, the slope of the trend is calculated using a linear regression method to obtain the slope value; If the slope value exceeds a preset threshold, the significance of the trend change is determined by sliding window analysis to obtain a significance index. Based on the mean sequence, variance sequence, and slope value, a feature vector set containing the mean, variance, and slope is constructed to obtain the final feature vector set.

4. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The electric field intensity requirements are classified using a support vector classification algorithm based on the feature vector set, generating electric field intensity grading threshold sets corresponding to low, medium, and high water content ranges, defined as low range, medium range, and high range, resulting in an electric field intensity threshold set including: The mean data, variance data, and slope data are extracted from the feature vector set to construct the classification input dataset, thus obtaining the classification input dataset. The classification input dataset is trained using a support vector classification algorithm to generate a classification model for the water content range, thus obtaining the classification model; The classification model is used to divide the moisture content range, determine the boundaries of low, medium and high moisture content ranges, and obtain the range division results; If the boundary value of the interval division result is inconsistent with the preset moisture content threshold, the parameters of the classification model are adjusted by cross-validation to obtain an optimized classification model. Based on the optimized classification model, the electric field intensity values ​​corresponding to the low, medium and high moisture content ranges are calculated to obtain the electric field intensity threshold set. Extract low-range thresholds, medium-range thresholds, and high-range thresholds from the electric field intensity threshold set, and construct a hierarchical threshold mapping table to obtain the hierarchical threshold mapping table; The hierarchical threshold mapping table is used to generate the correspondence between water content intervals and electric field strength values, thus obtaining the interval strength mapping set.

5. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves obtaining threshold ranges for each interval from the set of electric field strength thresholds, querying the functional material type and proportion parameters corresponding to the intervals based on a pre-established electric field-material mapping table, and adjusting the proportion parameters to the nearest value within the preset range if the proportion parameters exceed the preset range, thereby obtaining an optimized dispersion formulation scheme. This includes: The threshold ranges of each interval are obtained from the set of electric field intensity thresholds, and the interval threshold dataset is constructed to obtain the interval threshold dataset; Based on the interval threshold dataset, query the pre-established electric field-material mapping table, match the corresponding functional material type and proportion parameters, and obtain the initial material ratio set; If the proportion parameter values ​​in the initial material ratio set exceed the preset range threshold, they are adjusted to the nearest value within the range using a linear interpolation method to obtain the adjusted material ratio set. Using the adjusted material ratio set and the dispersion composition database, the content of each component in the optimized ratio scheme is calculated to obtain the dispersion ratio parameters; Based on the dispersion ratio parameters, a ratio scheme generation record is generated and stored in the ratio scheme database to obtain the ratio scheme storage result; Based on the storage results of the ratio scheme, verify the consistency between the optimized ratio scheme and the interval threshold range. If they are inconsistent, adjust the ratio parameters through an iterative optimization algorithm to obtain the final optimized ratio scheme. The functional material types and proportion parameters are extracted from the final optimized proportioning scheme to generate the proportioning scheme execution instructions, thus obtaining the proportioning scheme execution instruction set.

6. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves using a gradient optimization algorithm to calculate the mechanical stirring speed and duration sequence based on the optimized dispersion ratio scheme, generating a stirring parameter control sequence containing multiple time nodes. Each fixed time interval includes both stirring speed and duration, resulting in a stirring control sequence, including: The component ratio parameters are obtained from the optimized dispersion ratio scheme, and a ratio parameter dataset is constructed to obtain the ratio parameter dataset; Based on the aforementioned proportioning parameter dataset, the mechanical stirring speed and duration at each time point are calculated using the gradient descent algorithm to generate an initial stirring parameter sequence. If the stirring speed in the initial stirring parameter sequence exceeds the preset range, it is adjusted to the nearest value within the range using a linear interpolation method to obtain the adjusted stirring parameter sequence. Based on the adjusted stirring parameter sequence and the time node sequence, a stirring parameter control sequence containing fixed time intervals is generated, thus obtaining the stirring parameter control sequence. The stirring speed and duration at each time point are extracted from the stirring parameter control sequence to verify the consistency with the ratio parameter dataset. If they are inconsistent, the stirring parameters are iteratively optimized using Newton's method to obtain an optimized stirring parameter sequence. Based on the optimized stirring parameter sequence, a stirring control command sequence is generated and stored in the stirring control database to obtain the stirring control command storage result. Extract the final stirring control sequence from the storage results of the stirring control instructions, generate the equipment execution instructions, and obtain the set of equipment execution instructions.

7. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves obtaining the stirring speed value at the current time point from the stirring control sequence, applying a corresponding stirring operation to the optimized dispersion, monitoring the material distribution density in real time using an online density sensor, and iteratively adjusting the speed value based on the gradient descent method until the deviation converges to within the preset threshold to obtain a homogeneous mixture. The stirring speed value at the current time node is obtained from the stirring control sequence, and the material distribution density data is collected in real time using an online density sensor to obtain the material distribution density dataset; If the density deviation in the material distribution density dataset exceeds the preset threshold, then the loss function L is defined as the mean square error between the density value and the target uniform density, the learning rate is initialized to α = 0.01, and the iteration update speed is v_new = v_old - αdL / dv, where dL / dv is the derivative of the loss with respect to the speed. This process is repeated until L is lower than the threshold to obtain the adjusted stirring speed value. Based on the adjusted stirring speed value, update the stirring speed value at the current time node in the stirring control sequence to obtain the updated stirring control sequence; Extract the stirring speed value of the next time node from the updated stirring control sequence, apply the stirring speed value and then collect the material distribution density data again through the online density sensor to obtain a new material distribution density dataset; If the density deviation of the new material distribution density dataset still exceeds the preset threshold, the minimum function in Python's SciPy library is used to iteratively optimize the stirring speed value using the 'Newton-CG' method. The input objective function is the sum of squares of the density deviation, the initial guess is the current speed, and the optimized stirring speed value is obtained. Based on the optimized stirring speed value, a sequence of control instructions for the stirring equipment is generated, a set of equipment execution instructions is generated, and the set of equipment execution instructions is obtained.

8. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves treating the homogeneous mixture using a temperature-controlled curing method, generating a curing time-temperature curve based on a pre-established material thermal characteristic curve, raising the temperature to a preset temperature and holding it for a fixed duration to obtain an initial cured film structure, including: Temperature data during the curing process is collected in real time using online temperature sensors to obtain a temperature dataset; If the temperature value in the temperature dataset deviates from the preset temperature by more than a preset threshold, the heating power is iteratively adjusted using the gradient descent method. The loss function L is defined as the mean square error between the temperature value and the preset temperature, where the temperature value refers to the actual temperature collected and the preset temperature refers to the target temperature. The learning rate α was initialized to 0.

01. The linear regression of the NumPy library was used to fit the relationship between the temperature dataset and the current power. The derivative of the loss function with respect to power, dL / dP, was calculated, where dL / dP is equal to twice the mean temperature deviation multiplied by the fitting slope. The power P_new is iteratively updated to P_old minus α multiplied by dL / dP. This process is repeated until the loss function value is below the threshold, thus obtaining the adjusted heating power. Based on the adjusted heating power, the temperature control sequence is updated, a new curing time and temperature curve is generated, and the updated temperature control sequence is obtained. Extract the temperature value of the current time node from the updated temperature control sequence, apply the corresponding temperature through the heating device, and collect the curing state data of the thin film structure in real time to obtain the curing state dataset. If the structural uniformity in the solidified state dataset is lower than the preset standard, the minimum function in Python's SciPy library is used to optimize the solidification time using the BFGS method. The input objective function is the sum of squares of the pixel variances of the structural image calculated using the OpenCV library, and the initial guess is the current time. The optimized solidification time is then obtained. Based on the optimized curing time, the curing time and temperature curves are adjusted to generate a new sequence of control instructions for the heating equipment, thus obtaining a set of equipment execution instructions. The heating equipment is controlled by a set of instructions executed by the device, and the adjusted temperature and time are applied to obtain the final cured film structure.

9. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves randomly sampling multiple regions from the initially cured film structure, measuring the electric field response data at each point, calculating the relative mean square error between the response data and the electric field intensity threshold set, and if the error exceeds a preset threshold, adjusting the curing time and temperature curves based on a gradient optimization algorithm to reduce the temperature and extend the curing time to obtain the final electric field response film. This includes: Multiple regions were randomly selected from the initial cured film structure, and electric field response data of each point were collected using an electric field sensor to obtain an electric field response dataset. Based on the electric field response dataset, the mean square error between the response data at each point and the preset electric field intensity threshold set is calculated to obtain the error dataset; If the mean square error in the error dataset exceeds the preset threshold, the gradient descent method is used to iteratively optimize the curing time and temperature curves. The loss function L is calculated as the sum of squared mean square errors. The learning rate α is initialized, and the curing time and temperature parameters are updated to obtain the set of optimized parameters. Based on the optimized parameter set, a new curing time and temperature curve is generated, resulting in an updated control sequence; By updating the control sequence, adjusting the temperature and curing time applied by the heating equipment, and collecting the electric field response data of the adjusted film, a new electric field response dataset is obtained. The mean square error is calculated from the new electric field response dataset. If the error still exceeds the preset threshold, the minimum function of the SciPy library is used to further optimize the curing time using the BFGS method. The objective function is the variance of the electric field response data, and the final curing time is obtained. Based on the final curing time, the heating equipment control command is updated, and the adjusted temperature and duration are applied to obtain an optimized electric field response film.

10. The method for preparing an electric field-responsive thin film according to claim 2, characterized in that, The process involves testing the final electric field response film using a simulated dehydration environment method. The input is a water content load generated based on the distribution of the feature vector set. The dehydration efficiency index is measured, defined as the ratio of dehydrated water to initial water content. If the efficiency index reaches a preset standard, the film preparation is confirmed complete. This includes: Multiple regions were randomly selected from the thin film, and electric field response data at each point were collected using an electric field sensor to obtain the first electric field response dataset. Based on the first electric field response dataset, calculate the average electric field intensity and variance at each point to generate the first feature vector set; The PCA function of the sklearn library is used to perform principal component analysis on the first feature vector set to extract the first principal feature vector related to the moisture content, thus obtaining the first principal feature vector set. The first water content loading sequence is constructed by calculating the Euclidean norm of each vector from the first set of principal eigenvectors. The first dehydration test dataset was obtained by simulating the dehydration environment using ANSYS software and inputting the first moisture content load sequence. From the first dehydration test dataset, the first dehydration efficiency index is calculated. The first dehydration efficiency index is defined as the dimensionless ratio of the dehydration amount D divided by the initial water content I, i.e., D / I, where D is the mass of water reduced in the simulation and I is the initial water mass. It is determined whether the first dehydration efficiency index is greater than the preset threshold of 0.8 to obtain the first efficiency judgment result. If the first efficiency determination result is not greater than the preset threshold of 0.8, the minimum function of the scipy.optimize library is used to optimize the first water content load sequence based on gradient descent, update the first main feature vector set, and obtain the second water content load sequence. Using the second moisture content load sequence, the simulated dehydration environment test was re-executed in ANSYS software, the second dehydration test dataset was collected, the second dehydration efficiency index was calculated, and it was determined whether the second dehydration efficiency index was greater than the preset threshold of 0.8, thus obtaining the second efficiency judgment result. Based on the second water content load sequence where the second efficiency determination result is greater than the preset threshold of 0.8, the thin film preparation parameters are adjusted to determine that the final thin film preparation is complete.

Images