Ultrasonic subcritical water extraction device and method

Abstract

The invention provides an ultrasonic subcritical water extraction device and method, relates to the field of intelligent extraction, and obtains time sequence data of a flow velocity value of deionized water, time sequence data of an extraction temperature value and time sequence data of an extraction pressure value. A data analysis and processing algorithm based on deep learning is adopted to carry out time sequence coding and parameter collaborative reasoning representation on the flow rate value, the extraction temperature value and the extraction pressure value of the deionized water; and according to extraction parameter optimization deionized water flow velocity reasoning representation, principal component time sequence aggregation representation is carried out to adaptively control ultrasonic power at a next time point. Therefore, the ultrasonic power can be adjusted in real time according to the time sequence change conditions of the flow velocity, the temperature and the pressure parameters monitored in real time, and the optimal extraction effect can be ensured to be achieved at each time point, so that the intelligent control of the extraction process is realized, and the manual intervention is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent extraction, and more particularly, in an embodiment of the present application, to an ultrasonic subcritical water extraction device and method. Background Art

[0002] Subcritical water extraction technology combines the environmental friendliness of water with the good solubility of organic solvents, and can achieve efficient extraction at lower temperatures, reducing damage to heat-sensitive substances. Ultrasonic extraction technology can effectively destroy plant cell walls through mechanical effects, thermal effects, and cavitation effects, accelerating the release of active ingredients, thereby improving extraction efficiency and extraction rate. The synergistic effect of the two can not only speed up the extraction speed, but also protect heat-sensitive ingredients, thereby increasing the yield and quality of the extract.

[0003] Chinese patent CN110448934B proposes a method for extracting polyphenols from emblica and an ultrasonically coupled dynamic subcritical water extraction device, which extracts polyphenols from emblica by combining ultrasound with dynamic subcritical water. Specifically, the treated raw materials are first placed in a preparation container, and then preheated deionized water is injected. After the pressure and temperature meet the standards, a high-pressure pump is started to circulate the water, and the extract is collected after ultrasonic treatment for a certain period of time. Finally, high-purity emblica polyphenols are obtained by reduced pressure concentration and vacuum freeze drying.

[0004] The above patent uses the traditional method to control the power of ultrasound when using ultrasound for extraction. However, in the traditional method, the ultrasonic power is pre-set based on empirical values ​​or experimental results, and it remains fixed once the extraction begins. It cannot be dynamically adjusted according to the real-time changes in environmental conditions (temperature and pressure) during the extraction process. This may cause the ultrasonic power to be too high at certain stages, causing unnecessary heat accumulation and energy waste, or extend the extraction time due to too low power, affecting the stability of the extraction. In addition, the fixed power and time control method ignores the temporal fluctuations of environmental factors such as temperature and pressure in the early and late stages of extraction, resulting in the inability to accurately match the optimal extraction conditions for each stage, making it difficult to achieve the ideal extraction effect.

[0005] Therefore, an optimized ultrasonic subcritical water extraction protocol is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an ultrasonic subcritical water extraction device and method, which obtains the time series data of the flow rate value of deionized water, and obtains the time series data of the extraction temperature value and the time series data of the extraction pressure value, and adopts a data analysis and processing algorithm based on deep learning to perform time series encoding and parameter collaborative reasoning representation of the flow rate value of the deionized water, the extraction temperature value and the extraction pressure value, so as to optimize the deionized water flow rate reasoning representation according to the extraction parameters and perform principal component time series aggregation representation to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of the flow rate, temperature and pressure parameters monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

[0007] According to one aspect of the present application, there is provided an ultrasonic subcritical water extraction method, comprising:

[0008] Placing the emblica raw material in a preparation container, and injecting preheated deionized water into the material basket of the preparation container; in response to the pressure and temperature in the preparation container reaching the set value, starting the high-pressure pump to allow the deionized water to circulate in the material basket; using ultrasound to extract the emblica in the material basket to obtain an extract; performing reduced pressure concentration and vacuum freeze-drying on the extract to obtain emblica polyphenols; wherein, using ultrasound to extract the material basket in the material basket comprises:

[0009] Obtaining time series data of flow rate values ​​of deionized water;

[0010] Obtaining time series data of extraction temperature value and time series data of extraction pressure value;

[0011] Performing parameter collaborative reasoning based on time series characteristics on the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value to obtain a sequence of local time series reasoning representations of the deionized water flow rate for extraction parameter optimization;

[0012] Performing characteristic principal component time series decay aggregation on the sequence of local time series reasoning representation of the extraction parameter optimization deionized water flow rate to obtain a characteristic aggregation representation of the principal component time series of the deionized water flow rate;

[0013] Based on the aggregation characterization of the main component characteristics of the deionized water flow rate time series, a control result is obtained, and the control result is used to indicate whether the ultrasonic power at the next time point should increase, decrease or remain unchanged.

[0014] According to another aspect of the present application, an ultrasonic subcritical water extraction device is provided, wherein the ultrasonic subcritical water extraction device is used to perform the ultrasonic subcritical water extraction method described above.

[0015] Compared with the prior art, the ultrasonic subcritical water extraction device and method provided by the present application obtains the time series data of the flow rate value of deionized water, and obtains the time series data of the extraction temperature value and the time series data of the extraction pressure value, and adopts a data analysis and processing algorithm based on deep learning to perform time series encoding and parameter collaborative reasoning representation of the flow rate value of deionized water, the extraction temperature value and the extraction pressure value, thereby optimizing the deionized water flow rate reasoning representation according to the extraction parameters to perform principal component time series aggregation representation to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of the flow rate, temperature and pressure parameters monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 Flow chart of the ultrasonic subcritical water extraction method according to an embodiment of the present application.

[0018] Figure 2 This is a flow chart of using ultrasound to extract the material basket in the material basket in the ultrasonic subcritical water extraction method according to an embodiment of the present application.

[0019] Figure 3 It is a data flow diagram for extracting the material basket in the material basket using ultrasound in the ultrasonic subcritical water extraction method according to an embodiment of the present application.

[0020] Figure 4 A flow chart of performing characteristic principal component time series attenuation aggregation on the sequence of local time series reasoning representation of deionized water flow rate optimization of the extraction parameters in the ultrasonic subcritical water extraction method according to an embodiment of the present application to obtain a characteristic aggregation representation of the principal component time series of the deionized water flow rate.

[0021] Figure 5A flowchart is provided for calculating the time series propagation attenuation factor of each local time series kernel principal component representation vector of the historical extraction parameter-optimized deionized water flow rate in the sequence of local time series kernel principal component representation vectors of the historical extraction parameter-optimized deionized water flow rate in the ultrasonic subcritical water extraction method according to an embodiment of the present application, and time-modulating the obtained sequence of time series propagation attenuation factors of the extraction parameter-optimized deionized water flow rate to obtain a sequence of time series span modulation propagation attenuation factors of the extraction parameter-optimized deionized water flow rate. DETAILED DESCRIPTION

[0022] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0023] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0024] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0025] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0026] Subcritical water extraction technology is a process that utilizes the good solubility and environmental friendliness of water in a subcritical state to perform efficient extraction, and is suitable for extracting heat-sensitive components. Subcritical water extraction combines the environmental friendliness of water with the high efficiency of organic solvents to effectively extract heat-sensitive components at low temperatures, while ultrasonic extraction accelerates the release of effective ingredients by destroying plant cell walls. Chinese patent CN110448934B proposes a method for extracting polyphenols from emblica fruit that combines these two technologies, but the traditional ultrasonic power control method is fixed and cannot be dynamically adjusted according to changes in temperature and pressure during the extraction process, which may lead to energy waste and unstable extraction effects.

[0027] In response to the above technical problems, the present application proposes an ultrasonic subcritical water extraction method. Figure 1Flow chart of ultrasonic subcritical water extraction method according to an embodiment of the present application. Figure 1 As shown, the ultrasonic subcritical water extraction method according to the embodiment of the present application includes: S110, placing the emblica raw material in a preparation container, and injecting preheated deionized water into the material basket of the preparation container; S120, in response to the pressure and temperature in the preparation container reaching the set value, starting the high-pressure pump to allow the deionized water to circulate in the material basket; S130, using ultrasound to extract the emblica in the material basket to obtain an extract; S140, performing reduced pressure concentration and vacuum freeze-drying on the extract to obtain emblica polyphenols.

[0028] In the above-mentioned ultrasonic subcritical water extraction method, the step S110 is to place the emblica raw material in a preparation container, and to inject preheated deionized water into the material basket of the preparation container. It should be understood that as a plant rich in nutrients and bioactive substances, the extraction of its effective ingredients depends on the temperature and soaking time of water. Preheated deionized water can quickly penetrate the emblica raw material, promote the rupture of cell walls, thereby releasing more effective ingredients and improving the extraction efficiency. In the specific operation process, first, the emblica raw material is evenly placed in the preparation container to ensure that the raw material is evenly distributed, so as to avoid the problem of local over-extraction or uneven extraction. Then, deionized water preheated to a suitable temperature is slowly injected into the material basket, and the water temperature is generally between 60°C and 80°C to ensure that the solubility during the extraction process is maximized and promote the dissolution of the effective ingredients. Among them, in this process, it is necessary to control the water level to ensure that the water completely covers the emblica raw material for full soaking. In addition, regular stirring or circulating water flow can further improve the extraction effect, increase the contact area between the raw material and water, and thus accelerate the release and extraction of the ingredients. Through this scientific operation method, the extraction rate of Phyllanthus emblica can be effectively improved, ensuring the quality of the final product and the integrity of the bioactive ingredients.

[0029] In the above-mentioned ultrasonic subcritical water extraction method, in step S120, in response to the pressure and temperature in the preparation container reaching the set value, the high-pressure pump is started to allow the deionized water to circulate in the material basket. It should be understood that when the pressure and temperature in the container reach the preset value, it indicates that the environmental conditions are suitable for extraction. At this time, starting the high-pressure pump can enhance the fluidity of the water, thereby accelerating the contact between the material and the water, so that the deionized water can more fully penetrate and soak the emblica raw material. The circulating deionized water can effectively take away the extracted components, avoid the formation of obstacles on the surface of the material by solutions with too high concentrations, and ensure the continuous release of effective ingredients. The high-pressure pump can also increase the flow rate and turbulence, thereby promoting uniform immersion of the material and improving the extraction efficiency. At the same time, this process can also control the temperature to prevent the degradation of components caused by local overheating and ensure the quality of the extract.

[0030] In the above-mentioned ultrasonic subcritical water extraction method, the step S130 uses ultrasound to extract the emblica in the material basket to obtain an extract. It should be understood that ultrasound produces a cavitation effect in the liquid through high-frequency sound waves, causing tiny bubbles to be rapidly generated and collapsed in the liquid. The high energy generated by this process can destroy the cell walls and release the effective ingredients in the cells, such as polyphenols, sugars and other bioactive substances. Compared with traditional immersion or heating methods, ultrasonic extraction has a shorter extraction time and a higher extraction rate, can effectively retain heat-sensitive components, reduce the degradation or loss of components, and the extracted liquid will contain rich bioactive components. Ultrasonic extraction improves the extraction efficiency and ensures the quality and stability of the product.

[0031] Accordingly, in the process of using ultrasound to extract the material basket in the material basket, the technical concept of the present application is to obtain the time series data of the flow rate value of deionized water, and obtain the time series data of the extraction temperature value and the time series data of the extraction pressure value, and use a data analysis and processing algorithm based on deep learning to perform time series encoding and parameter collaborative reasoning representation of the flow rate value of deionized water, the extraction temperature value and the extraction pressure value, so as to optimize the deionized water flow rate reasoning representation according to the extraction parameters and perform principal component time series aggregation representation to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of parameters such as flow rate, temperature and pressure monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

[0032] Figure 2 This is a flow chart of using ultrasound to extract the material basket in the material basket in the ultrasonic subcritical water extraction method according to an embodiment of the present application. Figure 3 This is a data flow diagram of using ultrasound to extract the material basket in the material basket according to the ultrasonic subcritical water extraction method of the embodiment of the present application. Figure 2 and Figure 3As shown, in the step S130, ultrasonic waves are used to extract the material basket in the material basket, including: S131, obtaining time series data of the flow rate value of deionized water; S132, obtaining time series data of the extraction temperature value and time series data of the extraction pressure value; S133, performing parameter collaborative reasoning based on time series characteristics on the time series data of the flow rate value of deionized water, the time series data of the extraction temperature value and the time series data of the extraction pressure value to obtain a sequence of local time series reasoning representations of the extraction parameters optimizing the deionized water flow rate; S134, performing feature principal component time series attenuation aggregation on the sequence of local time series reasoning representations of the extraction parameters optimizing the deionized water flow rate to obtain a deionized water flow rate time series principal component feature aggregation representation; S135, obtaining a control result based on the deionized water flow rate time series principal component feature aggregation representation, and the control result is used to indicate whether the ultrasonic power at the next time point should increase, decrease or remain unchanged.

[0033] Specifically, in step S131 and step S132, the time series data of the flow rate value of deionized water is obtained, and the time series data of the extraction temperature value and the time series data of the extraction pressure value are obtained. It should be understood that in the technical solution of the present application, an appropriate flow rate can increase the propagation efficiency of ultrasound in the liquid, thereby improving the extraction effect. Similarly, an increase in temperature can increase the fluidity of the liquid, making it easier to release the effective ingredients. In addition, pressure changes will affect the boiling point and solubility of the liquid, thereby directly affecting the extraction efficiency. When the flow rate, temperature and pressure reach an optimized state, the effect of ultrasound will be further enhanced to form a synergistic effect, thereby significantly improving the extraction rate and extraction quality. By monitoring the time series data of these parameters, the extraction process can be more accurately grasped, and the operating conditions can be optimized to improve the quality and stability of the final product.

[0034] In an embodiment of the present application, the step S133 performs parameter collaborative reasoning based on time series characteristics on the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value to obtain a sequence of local time series reasoning representations of the deionized water flow rate for extraction parameter optimization, including: S1331, passing the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value through a time series encoder based on a forward LSTM model to obtain a sequence of local time series correlation feature vectors of the deionized water flow rate, a sequence of local time series correlation feature vectors of the extraction temperature, and a sequence of local time series correlation feature vectors of the extraction pressure. S1332, passing each group of corresponding deionized water flow rate local time series association feature vectors, extraction temperature local time series association feature vectors and extraction pressure local time series association feature vectors in the sequence of deionized water flow rate local time series association feature vectors, the sequence of extraction temperature local time series association feature vectors and the sequence of extraction pressure local time series association feature vectors through an extraction process parameter collaborative reasoning characterizer based on a quasi-Bayesian probability network to obtain a sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization vectors as the extraction parameter optimized deionized water flow rate local time series reasoning characterization sequence.

[0035] Specifically, in step S1331, the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value are passed through a time series encoder based on a forward LSTM model to obtain a sequence of local time series associated feature vectors of the flow rate of the deionized water, a sequence of local time series associated feature vectors of the extraction temperature, and a sequence of local time series associated feature vectors of the extraction pressure. It should be understood that, considering that the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value have mutual correlation influences in the local time segment, the forward LSTM model is a neural network model suitable for processing time series data, which can effectively capture the time dependency in the data and overcome the gradient vanishing problem that may occur in traditional neural networks in long sequence learning. That is, the forward LSTM model can effectively capture the dynamic relationship between such time series data through its internal state and gating mechanism. Specifically, the forward LSTM identifies the time series dependency and mutual influence between variables in different time intervals by processing the input time series data, and extracts potential feature vectors. In this way, it can not only reflect the state of a single variable at a specific time point, but also comprehensively consider the changing trends and interactions of these variables in the time dimension. Based on this, in the technical solution of the present application, the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value are captured and extracted through a time series encoder based on a forward LSTM model to capture and extract the time series correlation relationship in different time intervals, and obtain a sequence of local time series correlation feature vectors of the deionized water flow rate, a sequence of local time series correlation feature vectors of the extraction temperature, and a sequence of local time series correlation feature vectors of the extraction pressure.

[0036] Specifically, in the step S1332, each group of corresponding deionized water flow rate local time series association feature vectors, extraction temperature local time series association feature vectors and extraction pressure local time series association feature vectors in the sequence of deionized water flow rate local time series association feature vectors, the sequence of extraction temperature local time series association feature vectors and the sequence of extraction pressure local time series association feature vectors are passed through an extraction process parameter collaborative reasoning characterizer based on a Bayesian probability network to obtain a sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization vectors as the sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization. It should be understood that, considering that parameters such as flow rate, temperature and pressure may be affected by a variety of uncertain factors during the extraction process, in order to more accurately capture the mutual influence between deionized water flow rate, extraction temperature and extraction pressure, in the technical solution of the present application, each group of corresponding deionized water flow rate local time series correlation feature vectors, extraction temperature local time series correlation feature vectors and extraction pressure local time series correlation feature vectors in the sequence of deionized water flow rate local time series correlation feature vectors, the sequence of extraction temperature local time series correlation feature vectors and the sequence of extraction pressure local time series correlation feature vectors are subjected to the extraction process parameter collaborative reasoning characterizer based on the quasi-Bayesian probability network to obtain a sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization vectors. That is, the quasi-Bayesian probability network (Bayesian network) is a graphical model that can explicitly represent the causal relationship and uncertainty between variables, and the network can represent the joint probability distribution between multiple random variables, which is crucial to understanding the interaction between different extraction parameters, thereby obtaining the optimal local time series reasoning characterization.

[0037] Specifically, in step S134, the sequence of local temporal reasoning representations of the extraction parameter optimization deionized water flow rate is subjected to characteristic principal component temporal decay aggregation to obtain a characteristic aggregation representation of the deionized water flow rate temporal principal component. It should be understood that, considering that the importance of the local temporal reasoning representation vector of the extraction parameter optimization deionized water flow rate at different time points varies over the entire time range, that is, the impact of early data on the current state will gradually decrease with the passage of time, in order to more accurately identify and express the key features in the extraction parameter optimization parameters, and better understand the pattern of data changes over time, so as to achieve precise control of ultrasonic power, it is necessary to consider the impact of time factors on information weight. Therefore, in the technical solution of the present application, the sequence of local temporal reasoning representation vectors of the extraction parameter optimization deionized water flow rate is subjected to characteristic principal component temporal decay aggregation to obtain a characteristic aggregation representation vector of the deionized water flow rate temporal principal component as a characteristic aggregation representation of the deionized water flow rate temporal principal component.

[0038] Figure 4A flowchart of performing characteristic principal component time series decay aggregation on the sequence of local time series reasoning characterization of the deionized water flow rate optimization of the extraction parameters in the ultrasonic subcritical water extraction method according to the embodiment of the present application to obtain the characteristic aggregation characterization of the principal component time series of the deionized water flow rate. Figure 4 As shown, in an embodiment of the present application, the step S134, performs feature principal component time series attenuation aggregation on the sequence of local time series reasoning representation of the extraction parameter optimization deionized water flow rate to obtain a feature aggregation representation of the time series principal component of the deionized water flow rate, including: S1341, extracting a sequence of the current extraction parameter optimization deionized water flow rate local time series reasoning representation vector and a sequence of the historical extraction parameter optimization deionized water flow rate local time series reasoning representation vector from the sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation vector, and performing KPCA processing on them to obtain a sequence of the current extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector and a sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector; S1342, calculating the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector. The temporal propagation attenuation factor of each historical extraction parameter-optimized local temporal kernel principal component representation vector of the deionized water flow rate in the sequence of sequential kernel principal component representation vectors is obtained, and the obtained sequence of temporal propagation attenuation factors of the extraction parameter-optimized deionized water flow rate is time-modulated to obtain a sequence of temporal span-modulated propagation attenuation factors of the extraction parameter-optimized deionized water flow rate; S1343, based on the sequence of temporal span-modulated propagation attenuation factors of the extraction parameter-optimized deionized water flow rate, the sequence of historical extraction parameter-optimized local temporal kernel principal component representation vectors of the deionized water flow rate and the current extraction parameter-optimized local temporal kernel principal component representation vector of the deionized water flow rate are subjected to temporal gradient attenuation dynamic propagation aggregation to obtain a deionized water flow rate temporal principal component feature aggregation representation vector as the deionized water flow rate temporal principal component feature aggregation representation.

[0039] Specifically, in the step S1341, the sequence of the local time series reasoning characterization vector of the current extraction parameter optimization deionized water flow rate and the sequence of the local time series reasoning characterization vector of the historical extraction parameter optimization deionized water flow rate are extracted from the sequence of the local time series reasoning characterization vector of the extraction parameter optimization deionized water flow rate, and KPCA processing is performed on them to obtain the sequence of the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate and the local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate. It should be understood that KPCA is a nonlinear feature extraction method that can find the nonlinear structure of data in a high-dimensional space. By mapping the data to a higher-dimensional space, KPCA can discover the nonlinear relationship in the original data, which can better capture the time series interaction relationship between dynamic parameters such as flow rate, temperature and pressure. In particular, when KPCA processing is performed on it, a kernel function based on a Gaussian process is introduced to capture and mine the linear change characteristics between every two time points. That is, although the change of deionized water flow rate in the entire time series is nonlinear, if the entire time series is divided into sufficiently small local time periods, then the change of flow rate in these small time periods can be approximated as linear. In this way, when KPCA searches for nonlinear features in high-dimensional space, it can better capture and characterize the characteristics of nonlinear data through local linear approximation.

[0040] In an embodiment of the present application, the step S1341 extracts the current extraction parameter optimization deionized water flow rate local time series reasoning representation vector and the historical extraction parameter optimization deionized water flow rate local time series reasoning representation vector from the sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation vector, and performs KPCA processing on them to obtain the current extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector and the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, including: extracting the current extraction parameter optimization deionized water flow rate local time series reasoning representation vector from the sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation vector, and performing KPCA processing on the current extraction parameter optimization deionized water flow rate local time series reasoning representation vector The representation vector is processed by KPCA to obtain the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate; other extraction parameter optimization deionized water flow rate local time series reasoning representation vectors in the sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation vectors are defined as historical extraction parameter optimization deionized water flow rate local time series reasoning representation vectors to obtain a sequence of historical extraction parameter optimization deionized water flow rate local time series reasoning representation vectors, and each historical extraction parameter optimization deionized water flow rate local time series reasoning representation vector in the sequence of historical extraction parameter optimization deionized water flow rate local time series reasoning representation vectors is processed by KPCA to obtain a sequence of historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vectors.

[0041] Specifically, in step S1342, the time series propagation attenuation factor of each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector is calculated, and the obtained sequence of extraction parameter optimization deionized water flow rate time series propagation attenuation factor is time modulated to obtain a sequence of extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factor. It should be understood that in order to measure and analyze the time series change trend and consistency of the set deionized water flow rate, the absolute factor of the time series propagation attenuation entropy between each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector and the current extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector is calculated to measure the degree of change over time. Afterwards, in order to better characterize and understand the long-term time series change trend, each time series propagation attenuation factor is modulated by introducing a time span to obtain a sequence of extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factor.

[0042] Figure 5The present invention is a flowchart for calculating the time series propagation attenuation factor of each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the ultrasonic subcritical water extraction method according to the embodiment of the present application, and performing time modulation on the obtained sequence of extraction parameter optimization deionized water flow rate time series propagation attenuation factor to obtain the sequence of extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factor. Figure 5 As shown, in an embodiment of the present application, the step S1342 calculates the time series propagation attenuation factor of each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, and performs time modulation on the obtained sequence of extraction parameter optimization deionized water flow rate time series propagation attenuation factors to obtain a sequence of extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factors, including: S1342-1, calculates each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector relative to the current extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector. The time propagation attenuation factor of the local time series kernel principal component representation vector of the water flow rate is obtained to obtain the sequence of the extraction parameter-optimized deionized water flow rate time series propagation attenuation factors; S1342-2, based on the time span between each historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector in the sequence of historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector, each extraction parameter-optimized deionized water flow rate time series propagation attenuation factor in the sequence of extraction parameter-optimized deionized water flow rate is modulated in the time dimension to obtain the sequence of the extraction parameter-optimized deionized water flow rate time span modulated propagation attenuation factors.

[0043] In an embodiment of the present application, the step S1342-1 calculates the timing propagation attenuation factor of each historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector in the sequence of the historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector relative to the current extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector to obtain the sequence of timing propagation attenuation factors of the extraction parameter-optimized deionized water flow rate, including: dividing the eigenvalues ​​of the corresponding positions of the historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector and the current extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector to obtain the timing propagation attenuation factor of the extraction parameter-optimized deionized water flow rate timing kernel principal component. Quantity; calculate the base-two logarithmic function value of the absolute value of each eigenvalue of the extraction parameter optimization deionized water flow rate timing kernel principal component characteristic attenuation vector to obtain the extraction parameter optimization deionized water flow rate timing kernel principal component characteristic attenuation logarithmic vector; calculate the position-by-position point product of the historical extraction parameter optimization deionized water flow rate local timing kernel principal component representation vector and the extraction parameter optimization deionized water flow rate timing kernel principal component characteristic attenuation logarithmic vector, and perform point-by-position multiplication on the obtained point product vector to obtain the extraction parameter optimization deionized water flow rate timing propagation attenuation value; calculate the exponential function of the extraction parameter optimization deionized water flow rate timing propagation attenuation value with the natural constant e as the base to obtain the extraction parameter optimization deionized water flow rate timing propagation attenuation factor.

[0044] In an embodiment of the present application, the step S1342-2, based on the time span between each historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector in the sequence of historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector and the current extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector, performs time dimension modulation on each extraction parameter-optimized deionized water flow rate timing propagation attenuation factor in the sequence of extraction parameter-optimized deionized water flow rate timing propagation attenuation factors to obtain the sequence of extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation factors, including: modulating the current extraction parameter-optimized deionized water flow rate local timing kernel principal component The timestamp of the representation vector is subtracted from the timestamp of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, and the resultant is rounded down to obtain the extraction parameter optimization deionized water flow rate time span value; an exponential function with the natural constant e as the base and the extraction parameter optimization deionized water flow rate time span value as the exponent is calculated to obtain the extraction parameter optimization deionized water flow rate time span modulation value; the extraction parameter optimization deionized water flow rate time series propagation attenuation factor corresponding to the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector is divided by the extraction parameter optimization deionized water flow rate time span modulation value to obtain the extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factor.

[0045] Specifically, in step S1343, based on the sequence of the extraction parameter-optimized deionized water flow rate time series span modulation propagation attenuation factors, the sequence of the historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vectors and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector are subjected to time series gradient attenuation dynamic propagation aggregation to obtain the deionized water flow rate time series principal component feature aggregation representation vector. It should be understood that the sequence of the extraction parameter-optimized deionized water flow rate time series span modulation propagation attenuation factors is input into the gated function screening module to identify the most influential propagation path and assign corresponding weights accordingly, thereby obtaining the sequence of the extraction parameter-optimized deionized water flow rate time series span modulation propagation attenuation weights. Finally, based on the sequence of weights, the sequence of local time series kernel principal component representation vectors of historical extraction parameter optimization of deionized water flow rate is weighted and calculated to integrate the long-term time series change trend in the historical data to obtain the significant propagation aggregation representation vector of the kernel principal component of historical extraction parameter optimization of deionized water flow rate, and the weighted historical aggregation representation vector and the current extraction parameter optimization local time series reasoning representation vector of deionized water flow rate are summed by position to make full use of the long-term trend in the historical data and the immediate changes in the current data, to obtain the deionized water flow rate time series principal component feature aggregation representation vector.

[0046] In an embodiment of the present application, the step S1343, based on the sequence of the extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation factors, performs time gradient attenuation dynamic propagation aggregation on the sequence of the historical extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vectors and the current extraction parameter-optimized deionized water flow rate local timing kernel principal component representation vector to obtain a deionized water flow rate timing principal component feature aggregation representation vector, including: inputting the sequence of the extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation factors into a gating function screening module to obtain a sequence of the extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation weights, including: comparing each extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation factor in the sequence of the extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation factors with a predetermined threshold value to obtain a sequence of the extraction parameter-optimized deionized water flow rate timing span modulation propagation attenuation weights; wherein, in response to the extraction parameter-optimized deionized water If the velocity timing span modulation propagation attenuation factor is greater than the predetermined threshold, the extraction parameter optimized deionized water flow velocity timing span modulation propagation attenuation factor greater than the predetermined threshold is input into the sigmoid function; in response to the extraction parameter optimized deionized water flow velocity timing span modulation propagation attenuation factor being less than or equal to the predetermined threshold, the extraction parameter optimized deionized water flow velocity timing span modulation propagation attenuation factor less than or equal to the predetermined threshold is set to zero; based on the sequence of the extraction parameter optimized deionized water flow velocity timing span modulation propagation attenuation weights, the weighted sum of the sequence of the historical extraction parameter optimized deionized water flow velocity local timing span kernel principal component representation vectors is calculated to obtain the historical extraction parameter optimized deionized water flow velocity core principal component significant propagation aggregation representation vector; the positional sum of the historical extraction parameter optimized deionized water flow velocity core principal component significant propagation aggregation representation vector and the current extraction parameter optimized deionized water flow velocity local timing span kernel principal component representation vector is calculated to obtain the deionized water flow velocity timing principal component feature aggregation representation vector.

[0047] Specifically, the sequence of the local time series reasoning characterization vectors of the extraction parameter optimization deionized water flow rate is subjected to characteristic principal component time series decay aggregation, which can be expressed by the formula:

[0048] X={x1,x2,...,x i ,...,x j ,...,x p}

[0049] v i =KPCA(x i )

[0050]

[0051] V={v1,v2,...,v i ,...,v p}

[0052]

[0053]

[0054] Wherein, X is the sequence of local temporal reasoning representation vectors for optimizing the deionized water flow rate by the extraction parameters, x1, x2, ..., x i , ..., x j , ..., x p is the first, second, ..., i-th, ..., i-th, ..., p-th local temporal reasoning representation vector for optimizing the deionized water flow rate by extraction parameters in the sequence of local temporal reasoning representation vectors for optimizing the deionized water flow rate by extraction parameters, KPCA(x i ) is x i Perform kernel principal component analysis, v i For x i The corresponding extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, σ is the standard deviation of the local time series reasoning representation vector of the i-th extraction parameter optimization deionized water flow rate, exp(·) represents the exponential function with the natural constant e as the base, l is the attenuation rate modulation value, γ is the linear modulation parameter, k(x i , x j ) is the kernel function, which means x i and x j The similarity between them, H is x i The corresponding extraction parameter optimization deionized water flow rate centering matrix, I n is the identity matrix, n is the scaling parameter, 1 n is a column vector of all 1s, 1 n T Serves 1 n The transposed vector of x i The corresponding extraction parameter optimization deionized water flow rate kernel matrix composed of multiple similarities, K′ is x i The corresponding extraction parameters are optimized for the deionized water flow rate and the central kernel matrix, λ i is the eigenvalue of the i-th extraction parameter optimization deionized water flow rate kernel principal component vector in the sequence of extraction parameter optimization deionized water flow rate kernel principal component vector, α i is i The extraction parameters corresponding to the optimization of the deionized water flow rate kernel principal component vector, λ j is the jth eigenvalue after removing the maximum value each time, To return the j value corresponding to the maximum value, k s is the sth largest eigenvalue, {·;·;…;·} is the cascade processing, αz is the extraction parameter optimization deionized water flow rate cascade vector composed of the selected extraction parameter optimization deionized water flow rate kernel principal component vector, k i is x i The corresponding extraction parameter optimization deionized water flow rate kernel matrix row vector, k i T is the k i The transposed vector of , V is the sequence of vectors representing the local time series kernel principal component of the deionized water flow rate for the extraction parameter optimization, v1, v2, ..., v i , ..., v p are respectively the 1st, 2nd, ..., ith, ..., pth extraction parameter optimization deionized water flow rate local time series kernel principal component representation vectors in the sequence of the extraction parameter optimization deionized water flow rate local time series kernel principal component representation vectors, v i The eigenvalues ​​at each position in , v p The eigenvalues ​​of each position in, L is v i The number of eigenvalues ​​in, log represents the logarithmic function value with base 2, d i v i and v p The extraction parameters between the optimized deionized water flow rate and the time series propagation attenuation factor, t p and t i They represent the timestamps of the p-th and i-th extraction parameter optimization deionized water flow rate local time series kernel principal component representation vectors, For floor rounding operation, e (i→p) Yes i and v p The extraction parameters between are optimized, the deionized water flow rate, the time span modulation, the propagation attenuation factor, mask(·) is the masking process, sigmoid(·) is the sigmoid function, θ is the predetermined threshold, w i Yes i The corresponding extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation weight, p is the number of vectors in the sequence of the extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, v m It is the principal component feature aggregation characterization vector of the deionized water flow rate time series.

[0055] In the above-mentioned ultrasonic subcritical water extraction method, the step S140 is to perform reduced pressure concentration and vacuum freeze drying on the extract to obtain the polyphenols of Phyllanthus emblica. It should be understood that polyphenols are easily degraded under the influence of heat and oxygen, so a gentle treatment method is required to avoid the loss of components. In the technical solution of the present application, reduced pressure concentration reduces the pressure of the extract so that it evaporates at a lower boiling point, thereby effectively removing water. This method can not only reduce the damage of heat to polyphenols, but also shorten the processing time and improve efficiency. Vacuum freeze drying is to quickly freeze the liquid at a low temperature, and then remove the water through a sublimation process in a vacuum environment. In this process, the solid polyphenols will not be affected by high temperature and can maintain their structure and active ingredients. In this way, by performing reduced pressure concentration and vacuum freeze drying on the extract, the polyphenol components in Phyllanthus emblica can be effectively separated and concentrated, while retaining its biological activity and nutrients to the maximum extent.

[0056] In an embodiment of the present application, the step S140, based on the principal component characteristic aggregation representation of the deionized water flow rate time series, obtains a control result, including: passing the principal component characteristic aggregation representation vector of the deionized water flow rate time series through an ultrasonic power controller based on a classifier to obtain the control result. That is, the deionized water flow rate time series principal component characteristic aggregation representation obtained by performing time series principal component aggregation by optimizing the sequence of local time series reasoning representation vectors of the deionized water flow rate using the extraction parameters is classified and processed, so as to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of parameters such as flow rate, temperature and pressure monitored in real time, ensuring that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

[0057] Preferably, the deionized water flow rate time series principal component feature aggregation characterization vector is passed through a classifier-based ultrasonic power controller to obtain a control result, including: determining a deionized water flow rate time series principal component feature aggregation association response matrix and a deionized water flow rate time series principal component feature aggregation distance response matrix based on the square root of the correlation value of the eigenvalue of the deionized water flow rate time series principal component feature aggregation characterization vector and the L2 distance value, that is:

[0058]

[0059] Among them, v i represents the i-th eigenvalue of the principal component feature aggregation characterization vector of the deionized water flow rate time series, v jrepresents the jth eigenvalue of the principal component feature aggregation characterization vector of the deionized water flow rate time series, M1(i,j) represents the (i,j)th eigenvalue in the association response matrix of the principal component feature aggregation of the deionized water flow rate time series, M2(i,j) represents the (i,j)th eigenvalue in the distance response matrix of the principal component feature aggregation of the deionized water flow rate time series, v i ,v j ∈V∈R 1×L ;

[0060] The deionized water flow rate time series principal component feature aggregation characterization vector is matrix-multiplied with the deionized water flow rate time series principal component feature aggregation association response matrix to obtain the deionized water flow rate time series principal component feature aggregation association response vector; the deionized water flow rate time series principal component feature aggregation distance response matrix is ​​matrix-multiplied with the transposed vector of the deionized water flow rate time series principal component feature aggregation characterization vector to obtain the deionized water flow rate time series principal component feature aggregation distance response vector; the deionized water flow rate time series principal component feature aggregation association response matrix and the deionized water flow rate time series principal component feature aggregation distance response matrix are point-multiplied. After multiplication, matrix multiplication is performed with the transposed vector of the deionized water flow rate time series principal component feature aggregation characterization vector to obtain the deionized water flow rate time series principal component feature aggregation bias vector; the deionized water flow rate time series principal component feature aggregation association response vector, the deionized water flow rate time series principal component feature aggregation distance response vector and the deionized water flow rate time series principal component feature aggregation bias vector are dotted to obtain an optimized deionized water flow rate time series principal component feature aggregation characterization vector; and the optimized deionized water flow rate time series principal component feature aggregation characterization vector is passed through a classifier-based ultrasonic power controller to obtain a control result.

[0061] Here, the optimized deionized water flow rate time series principal component feature aggregation characterization vector V' is expressed as:

[0062]

[0063] In V1∈R 1×L , and V2∈R L×1 In the case of:

[0064]

[0065] Among them, V represents the aggregation characterization vector of the main component features of the deionized water flow rate time series, M1 represents the aggregation association response matrix of the main component features of the deionized water flow rate time series, M2 represents the aggregation distance response matrix of the main component features of the deionized water flow rate time series, (·) Trepresents the transposed vector of the vector, ⊙ represents the point multiplication by position, R represents a real number, L represents the length of the vector, V1 represents the first principal component feature vector of the deionized water flow rate time series, V2 represents the principal component feature aggregation bias vector of the deionized water flow rate time series, represents addition by position, and V' represents the optimized principal component feature aggregation characterization vector of the deionized water flow rate time series.

[0066] That is, when the sequence of local time-series correlation feature vectors of deionized water flow rate, the sequence of local time-series correlation feature vectors of extraction temperature and the sequence of local time-series correlation feature vectors of extraction pressure respectively represent the time-series correlation characteristics of the flow rate value, extraction temperature value and extraction pressure value of deionized water, when performing Bayesian-like probability parameter collaborative reasoning in the local time domain and local time domain node feature aggregation in the global time domain, the uneven distribution of collaborative reasoning corresponding to the inconsistent distribution of parameter features in the local time domain will lead to the lack of local sparse aggregation in the global time domain, that is, it will lead to the simple repetition of the time-series aggregation distribution in the local time domain, which will affect the logical dependency of the set time-series features and reduce the accuracy of the control results.

[0067] Therefore, by using the deionized water flow rate time series principal component feature aggregation characterization vector as its statistical reference-free distribution response framework based on the self-correlation matrix and the self-distance matrix, the simple repetition of feature distribution is avoided based on the retrieval-enhanced reverse response construction of the deionized water flow rate time series principal component feature aggregation characterization vector, and by ensuring the intrinsic mapping logic of the deionized water flow rate time series principal component feature aggregation characterization vector based on the retrieval-response context to avoid its surface combination mapping, so as to achieve the logical dependency mapping of the deionized water flow rate time series principal component feature aggregation characterization vector to the classification target domain while maintaining the intuitive response correlation, thereby improving the accuracy of the control result obtained by the deionized water flow rate time series principal component feature aggregation characterization vector through the classifier-based ultrasonic power controller. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of parameters such as flow rate, temperature and pressure monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

[0068] In summary, the ultrasonic subcritical water extraction method based on the embodiment of the present application is explained, which obtains the time series data of the flow rate value of deionized water, and obtains the time series data of the extraction temperature value and the time series data of the extraction pressure value, and adopts a data analysis and processing algorithm based on deep learning to perform time series encoding and parameter collaborative reasoning representation of the flow rate value of the deionized water, the extraction temperature value and the extraction pressure value, so as to optimize the deionized water flow rate reasoning representation according to the extraction parameters to perform principal component time series aggregation representation to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of the flow rate, temperature and pressure parameters monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

[0069] Specifically, the technical solution of the present application also provides an ultrasonic subcritical water extraction device, which is used to perform the ultrasonic subcritical water extraction method described above.

[0070] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned ultrasonic subcritical water extraction device have been described in detail above. Figures 1 to 5 The ultrasonic subcritical water extraction method has been described in detail, and therefore, its repeated description will be omitted.

[0071] In summary, an ultrasonic subcritical water extraction device based on an embodiment of the present application is explained, which obtains the time series data of the flow rate value of deionized water, and obtains the time series data of the extraction temperature value and the time series data of the extraction pressure value, and adopts a data analysis and processing algorithm based on deep learning to perform time series encoding and parameter collaborative reasoning representation of the flow rate value of the deionized water, the extraction temperature value and the extraction pressure value, so as to optimize the deionized water flow rate reasoning representation according to the extraction parameters and perform principal component time series aggregation representation to adaptively control the ultrasonic power at the next time point. In this way, the ultrasonic power can be adjusted in real time according to the time series changes of the flow rate, temperature and pressure parameters monitored in real time to ensure that the best extraction effect can be achieved at each time point, thereby realizing intelligent control of the extraction process and reducing manual intervention.

Claims

1. An ultrasonic subcritical water extraction method comprising: Placing the raw material of Phyllanthus emblica in a preparation container, and injecting preheated deionized water into the material basket of the preparation container; in response to the pressure and temperature in the preparation container reaching the set value, starting the high-pressure pump to make the deionized water circulate in the material basket; using ultrasound to extract the Phyllanthus emblica in the material basket to obtain an extract; performing reduced pressure concentration and vacuum freeze-drying on the extract to obtain Phyllanthus emblica polyphenols; characterized in that the material basket in the material basket is extracted using ultrasound, comprising: Obtaining time series data of flow rate values ​​of deionized water; Obtaining time series data of extraction temperature value and time series data of extraction pressure value; Performing parameter collaborative reasoning based on time series characteristics on the time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value to obtain a sequence of local time series reasoning representations of the deionized water flow rate for extraction parameter optimization; Performing characteristic principal component time series decay aggregation on the sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation to obtain the deionized water flow rate time series principal component characteristic aggregation representation; Based on the aggregation characterization of the main component characteristics of the deionized water flow rate time series, a control result is obtained, and the control result is used to indicate whether the ultrasonic power at the next time point should increase, decrease or remain unchanged.

2. The ultrasonic subcritical water extraction method according to claim 1, characterized in that: The time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value are subjected to parameter collaborative reasoning based on time series characteristics to obtain a sequence of local time series reasoning representations of the deionized water flow rate for extraction parameter optimization, including: The time series data of the flow rate value of the deionized water, the time series data of the extraction temperature value, and the time series data of the extraction pressure value are passed through a time series encoder based on a forward LSTM model to obtain a sequence of local time series associated feature vectors of the deionized water flow rate, a sequence of local time series associated feature vectors of the extraction temperature, and a sequence of local time series associated feature vectors of the extraction pressure; Each group of corresponding deionized water flow rate local time series association feature vectors, extraction temperature local time series association feature vectors and extraction pressure local time series association feature vectors in the sequence of deionized water flow rate local time series association feature vectors, the sequence of extraction temperature local time series association feature vectors and the sequence of extraction pressure local time series association feature vectors is passed through an extraction process parameter collaborative reasoning characterizer based on a Bayesian probability network to obtain a sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization vectors as the sequence of extraction parameter optimized deionized water flow rate local time series reasoning characterization.

3. The ultrasonic subcritical water extraction method according to claim 2, characterized in that: The sequence of the extraction parameter optimization deionized water flow rate local time series reasoning representation is subjected to characteristic principal component time series decay aggregation to obtain the deionized water flow rate time series principal component characteristic aggregation representation, including: Extracting a sequence of the local time series reasoning representation vector of the current extraction parameter optimizing the deionized water flow rate and a sequence of the local time series reasoning representation vector of the historical extraction parameter optimizing the deionized water flow rate from the sequence of the local time series reasoning representation vector of the extraction parameter optimizing the deionized water flow rate, and performing KPCA processing on the sequence to obtain a sequence of the local time series kernel principal component representation vector of the current extraction parameter optimizing the deionized water flow rate and a sequence of the local time series kernel principal component representation vector of the historical extraction parameter optimizing the deionized water flow rate; Calculating the time series propagation attenuation factor of each local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate in the sequence of local time series kernel principal component representation vectors of the historical extraction parameter optimization deionized water flow rate, and performing time modulation on the obtained sequence of time series propagation attenuation factors of the extraction parameter optimization deionized water flow rate to obtain a sequence of time series span modulation propagation attenuation factors of the extraction parameter optimization deionized water flow rate; Based on the sequence of extraction parameter-optimized deionized water flow rate time series span modulation propagation attenuation factors, the sequence of historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vectors and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector are subjected to time series gradient attenuation dynamic propagation aggregation to obtain a deionized water flow rate time series principal component feature aggregation representation vector as the deionized water flow rate time series principal component feature aggregation representation.

4. The ultrasonic subcritical water extraction method according to claim 3, characterized in that: Extracting the sequence of the local time series reasoning representation vector of the current extraction parameter optimization deionized water flow rate and the local time series reasoning representation vector of the historical extraction parameter optimization deionized water flow rate from the sequence of the local time series reasoning representation vector of the extraction parameter optimization deionized water flow rate, and performing KPCA processing on them to obtain the sequence of the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate and the local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate, including: Extracting a local time series reasoning representation vector of the current extraction parameter optimization deionized water flow rate from the sequence of local time series reasoning representation vectors of the extraction parameter optimization deionized water flow rate, and performing KPCA processing on the local time series reasoning representation vector of the current extraction parameter optimization deionized water flow rate to obtain the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate; The other local temporal reasoning representation vectors for optimizing deionized water flow rate of extraction parameters in the sequence of local temporal reasoning representation vectors for optimizing deionized water flow rate of extraction parameters are defined as historical local temporal reasoning representation vectors for optimizing deionized water flow rate of extraction parameters to obtain a sequence of historical local temporal reasoning representation vectors for optimizing deionized water flow rate of extraction parameters, and KPCA processing is performed on each historical local temporal reasoning representation vector for optimizing deionized water flow rate in the sequence of historical local temporal reasoning representation vectors for optimizing deionized water flow rate to obtain a sequence of historical local temporal kernel principal component representation vectors for optimizing deionized water flow rate of extraction parameters.

5. The ultrasonic subcritical water extraction method according to claim 4, characterized in that: Calculating the time series propagation attenuation factor of each historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter optimization deionized water flow rate local time series kernel principal component representation vector, and performing time modulation on the obtained sequence of extraction parameter optimization deionized water flow rate time series propagation attenuation factor to obtain a sequence of extraction parameter optimization deionized water flow rate time series span modulation propagation attenuation factor, including: Calculate the time series propagation attenuation factor of each local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate in the sequence of local time series kernel principal component representation vectors of the historical extraction parameter optimization deionized water flow rate relative to the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate to obtain the sequence of time series propagation attenuation factors of the extraction parameter optimization deionized water flow rate; Based on the time span between each historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector in the sequence of historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vectors and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector, each extraction parameter-optimized deionized water flow rate time series propagation attenuation factor in the sequence of extraction parameter-optimized deionized water flow rate time series propagation attenuation factor is modulated in the time dimension to obtain the sequence of extraction parameter-optimized deionized water flow rate time series span modulated propagation attenuation factors.

6. The ultrasonic subcritical water extraction method according to claim 5, characterized in that: Calculating the time series propagation attenuation factor of each local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate in the sequence of local time series kernel principal component representation vectors of the historical extraction parameter optimization deionized water flow rate relative to the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate to obtain the sequence of the time series propagation attenuation factors of the extraction parameter optimization deionized water flow rate, including: Dividing the characteristic value of the corresponding position of the local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate by the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate to obtain the extraction parameter optimization deionized water flow rate time series kernel principal component characteristic attenuation vector; Calculate the base-two logarithmic function value of the absolute value of each eigenvalue of the extraction parameter optimization deionized water flow rate time series kernel principal component characteristic attenuation vector to obtain the extraction parameter optimization deionized water flow rate time series kernel principal component characteristic attenuation logarithmic vector; Calculate the positional point product of the local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate and the characteristic attenuation logarithm vector of the time series kernel principal component of the extraction parameter optimization deionized water flow rate, and perform point-by-point multiplication on the obtained point product vector to obtain the extraction parameter optimization deionized water flow rate time series propagation attenuation value; An exponential function of the extraction parameter optimization deionized water flow rate time series propagation attenuation value with the natural constant e as the base is calculated to obtain the extraction parameter optimization deionized water flow rate time series propagation attenuation factor.

7. The ultrasonic subcritical water extraction method according to claim 6, characterized in that: Based on the time span between each historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector in the sequence of the historical extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector, each extraction parameter-optimized deionized water flow rate time series propagation attenuation factor in the sequence of the extraction parameter-optimized deionized water flow rate time series propagation attenuation factor is modulated in the time dimension to obtain the sequence of the extraction parameter-optimized deionized water flow rate time series span modulation propagation attenuation factor, including: The timestamp of the local time series kernel principal component representation vector of the current extraction parameter optimization deionized water flow rate is subtracted from the timestamp of the local time series kernel principal component representation vector of the historical extraction parameter optimization deionized water flow rate, and then rounded down to obtain the time span value of the extraction parameter optimization deionized water flow rate; an exponential function with the natural constant e as the base and the time span value of the extraction parameter optimization deionized water flow rate as the exponent is calculated to obtain the time span modulation value of the extraction parameter optimization deionized water flow rate; The extraction parameter optimized deionized water flow rate time series propagation attenuation factor corresponding to the historical extraction parameter optimized deionized water flow rate local time series kernel principal component representation vector is divided by the extraction parameter optimized deionized water flow rate time span modulation value to obtain the extraction parameter optimized deionized water flow rate time series span modulation propagation attenuation factor.

8. The ultrasonic subcritical water extraction method according to claim 7, characterized in that: Based on the sequence of the extraction parameter optimized deionized water flow rate time series span modulation propagation attenuation factors, the sequence of the historical extraction parameter optimized deionized water flow rate local time series kernel principal component representation vectors and the current extraction parameter optimized deionized water flow rate local time series kernel principal component representation vector are subjected to time series gradient attenuation dynamic propagation aggregation to obtain a deionized water flow rate time series principal component feature aggregation representation vector, including: Inputting the sequence of the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factors into the gated function screening module to obtain the sequence of the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation weights, including: comparing each extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factor in the sequence of the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factors with a predetermined threshold value to obtain the sequence of the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation weights; wherein, in response to the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factor being greater than the predetermined threshold value, inputting the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factor greater than the predetermined threshold value into the sigmoid function; in response to the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factor being less than or equal to the predetermined threshold value, setting the extraction parameter optimized deionized water flow rate timing span modulation propagation attenuation factor less than or equal to the predetermined threshold value to zero; Based on the sequence of the time series span modulation propagation attenuation weights of the extraction parameter optimized deionized water flow rate, the weighted sum of the sequence of the local time series kernel principal component representation vectors of the historical extraction parameter optimized deionized water flow rate is calculated to obtain the significant propagation aggregation representation vector of the kernel principal component of the historical extraction parameter optimized deionized water flow rate; The position-wise sum of the historical extraction parameter-optimized deionized water flow rate kernel principal component significant propagation aggregation representation vector and the current extraction parameter-optimized deionized water flow rate local time series kernel principal component representation vector is calculated to obtain the deionized water flow rate time series principal component feature aggregation representation vector.

9. The ultrasonic subcritical water extraction method according to claim 8, characterized in that: Based on the aggregation characterization of the main component features of the deionized water flow rate time series, a control result is obtained, including: passing the aggregation characterization vector of the main component features of the deionized water flow rate time series through an ultrasonic power controller based on a classifier to obtain the control result.

10. An ultrasonic subcritical water extraction device, characterized in that: The ultrasonic subcritical water extraction device is used to perform the ultrasonic subcritical water extraction method of claims 1-9.

Images