Ultra-high performance liquid chromatography data preprocessing method and system based on bidirectional long short-term memory network

Abstract

The invention provides an ultra-high performance liquid chromatography data preprocessing method and system based on a bidirectional long short-term memory network. The method comprises the following steps: acquiring target ultra-high performance liquid chromatography data to be processed; inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain a target preprocessing adjustment parameter output by the liquid chromatography data preprocessing parameter prediction model; wherein the liquid chromatogram data pre-processing parameter model is obtained by training sample ultra-high performance liquid chromatogram data marked with a pre-processing adjustment parameter label; and based on the target preprocessing adjustment parameter, performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data. According to the method, the noise reduction effect of the ultra-high performance liquid chromatography data is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular, to an ultra-high performance liquid chromatography data preprocessing method and system based on a bidirectional long short-term memory network. Background Art

[0002] Ultra-High Performance Liquid Chromatography (UHPLC), as an advanced separation and analysis technology, has shown extensive application value in multiple scientific research and industrial fields. Compared with traditional High Performance Liquid Chromatography (HPLC), UHPLC, with its finer packing particles and higher operating pressure, has achieved a significant improvement in separation efficiency and analysis speed. It can quickly and accurately analyze the multiple components in complex samples, perfectly meeting the requirements of high-throughput and high-sensitivity analysis pursued by modern laboratories.

[0003] In the application process of UHPLC, data processing technology plays a core role in ensuring the reliability and accuracy of analysis results. With the continuous progress of UHPLC technology, the amount of data generated has increased sharply, and at the same time, the complexity and diversity of the data have also increased. An effective data preprocessing process can remove background noise, correct baseline drift, accurately identify peaks, and extract key feature information from them, providing a solid foundation for subsequent accurate quantitative and qualitative analysis.

[0004] However, the existing data preprocessing methods have shown obvious limitations when dealing with UHPLC data. When processing data containing high-intensity noise or complex interference signals, the existing methods may be difficult to fully extract the target information, resulting in the loss or misjudgment of data information. In addition, since chromatographic data often contains complex non-linear characteristics, the existing linear models cannot effectively capture these characteristics when processing such data. And currently, many preprocessing steps still require manual participation, which not only increases the operation complexity but also may introduce errors due to human factors, thus affecting the consistency and repeatability of the results, leading to poor denoising effects of ultra-high performance liquid chromatography data. Therefore, there is an urgent need for an ultra-high performance liquid chromatography data preprocessing method and system based on a bidirectional long short-term memory network to solve the above problems. Summary of the Invention

[0005] Aiming at the problems existing in the prior art, the present invention provides an ultra-high performance liquid chromatography data preprocessing method and system based on a bidirectional long short-term memory network.

[0006] The present invention provides an ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network, including: Obtaining target ultra-high performance liquid chromatography data to be processed; Input the target ultra - performance liquid chromatography (UPLC) data into the liquid chromatography data pre - processing parameter prediction model to obtain the target pre - processing adjustment parameters output by the liquid chromatography data pre - processing parameter prediction model; wherein, the liquid chromatography data pre - processing parameter model is trained based on the sample UPLC data marked with pre - processing adjustment parameter tags. Based on the target pre - processing adjustment parameters, perform noise reduction pre - processing on the target UPLC data to obtain the noise reduction result data corresponding to the target UPLC data.

[0007] According to a UPLC data pre - processing method based on a bidirectional long short - term memory network provided by the present invention, the target pre - processing adjustment parameters include the first target pre - processing adjustment parameter, the second target pre - processing adjustment parameter, and the third target pre - processing adjustment parameter; wherein, the first target pre - processing adjustment parameter includes the pre - processing adjustment parameters corresponding to the wavelet threshold denoising process, the second target pre - processing adjustment parameter includes the pre - processing adjustment parameters corresponding to the baseline drift correction process, and the third target pre - processing adjustment parameter includes the pre - processing adjustment parameters corresponding to the smoothing filtering process. The performing noise reduction pre - processing on the target UPLC data based on the target pre - processing adjustment parameters to obtain the noise reduction result data corresponding to the target UPLC data includes: Based on the first target pre - processing adjustment parameter, perform wavelet threshold denoising on the target UPLC data to obtain the target UPLC data after wavelet threshold denoising. Based on the second target pre - processing adjustment parameter, perform baseline drift correction on the target UPLC data after wavelet threshold denoising to obtain the target UPLC data after baseline drift correction. Based on the third target pre - processing adjustment parameter, perform smoothing filtering on the target UPLC data after baseline drift correction to obtain the noise reduction result data.

[0008] According to a UPLC data pre - processing method based on a bidirectional long short - term memory network provided by the present invention, the liquid chromatography data pre - processing parameter prediction model is trained through the following steps: Obtain sample UPLC data. Based on the noise reduction pre - processing processes of multiple different noise reduction pre - processing types, perform noise reduction pre - processing on the sample UPLC data in sequence, and after determining that the sample UPLC data after noise reduction pre - processing meets the preset noise reduction requirements, obtain the pre - processing adjustment parameters corresponding to each noise reduction pre - processing process. Label each of the preprocessing adjustment parameter tags corresponding to the noise reduction preprocessing process for the sample ultra-high performance liquid chromatography data to obtain a training sample set, where the preprocessing adjustment parameter tags are constructed according to the preprocessing adjustment parameters; Train a bidirectional long short-term memory network according to the training sample set to obtain the liquid chromatography data preprocessing parameter prediction model.

[0009] According to a method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network provided by the present invention, the labeling each of the preprocessing adjustment parameter tags corresponding to the noise reduction preprocessing process for the sample ultra-high performance liquid chromatography data to obtain a training sample set includes: When the noise reduction preprocessing process is a wavelet threshold denoising process, construct a first adjustment parameter tag according to the wavelet basis type and threshold type used in the wavelet threshold denoising process; When the noise reduction preprocessing process is a baseline drift correction process, construct a second adjustment parameter tag according to the polynomial fitting order used in the baseline drift correction process; When the noise reduction preprocessing process is a smoothing filtering process, construct a third adjustment parameter tag according to the window size and polynomial order used in the smoothing filtering process; Construct the preprocessing adjustment parameter tags corresponding to the sample ultra-high performance liquid chromatography data in each of the noise reduction preprocessing processes according to the first adjustment parameter tag, the second adjustment parameter tag, and the third adjustment parameter tag; Based on the sample ultra-high performance liquid chromatography data labeled with the preprocessing adjustment parameter tags, construct the training sample set.

[0010] According to a method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network provided by the present invention, the training a bidirectional long short-term memory network according to the training sample set to obtain the liquid chromatography data preprocessing parameter prediction model includes: Train the bidirectional long short-term memory network according to the training sample set and a preset loss function until the loss function value in the training process converges to obtain the liquid chromatography data preprocessing parameter prediction model, where the preset loss function is constructed based on the mean square error and the peak-to-peak signal-to-noise ratio.

[0011] The present invention also provides a system for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network, including: A liquid chromatography data acquisition module for acquiring target ultra-high performance liquid chromatography data to be processed; A preprocessing parameter prediction module for inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain target preprocessing adjustment parameters output by the liquid chromatography data preprocessing parameter prediction model; wherein, the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels; A preprocessing module for performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data based on the target preprocessing adjustment parameters to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

[0012] According to a preprocessing system for ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network provided by the present invention, the target preprocessing adjustment parameters include a first target preprocessing adjustment parameter, a second target preprocessing adjustment parameter, and a third target preprocessing adjustment parameter; wherein, the first target preprocessing adjustment parameter includes preprocessing adjustment parameters corresponding to the wavelet threshold denoising process, the second target preprocessing adjustment parameter includes preprocessing adjustment parameters corresponding to the baseline drift correction process, and the third target preprocessing adjustment parameter includes preprocessing adjustment parameters corresponding to the smoothing filtering process; The preprocessing module includes a wavelet transform denoising module, a baseline drift correction module, and a smoothing filtering module, wherein: The wavelet transform denoising module is used to perform wavelet threshold denoising processing on the target ultra-high performance liquid chromatography data based on the first target preprocessing adjustment parameters to obtain the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing; The baseline drift correction module is used to perform baseline drift correction processing on the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing based on the second target preprocessing adjustment parameters to obtain the target ultra-high performance liquid chromatography data after baseline drift correction processing; The smoothing filtering module is used to perform smoothing filtering processing on the target ultra-high performance liquid chromatography data after baseline drift correction processing based on the third target preprocessing adjustment parameters to obtain the noise reduction result data.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the preprocessing method for ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network as described in any one of the above.

[0014] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the preprocessing method for ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network as described in any one of the above.

[0015] The present invention also provides a computer program product, including a computer program which, when executed by a processor, implements the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network as described in any one of the above.

[0016] For the ultra-high performance liquid chromatography data preprocessing method and system based on a bidirectional long short-term memory network provided by the present invention, a liquid chromatography data preprocessing parameter model trained by sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter tags is used to predict the preprocessing adjustment parameters of the target ultra-high performance liquid chromatography data to be processed. Then, based on the target preprocessing adjustment parameters predicted by the liquid chromatography data preprocessing parameter model, noise reduction preprocessing is performed on the target ultra-high performance liquid chromatography data, thereby improving the noise reduction effect of the ultra-high performance liquid chromatography data. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic flowchart of the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network provided by the present invention; Figure 2 It is a schematic diagram of the effect of wavelet transform denoising based on multi-threshold and multi-wavelet basis selection provided by the present invention; Figure 3 It is a schematic diagram of the effect of baseline drift correction based on polynomial fitting provided by the present invention; Figure 4 It is a schematic diagram of the effect of smoothing filtering based on a Savitzky-Golay filter provided by the present invention; Figure 5 It is a schematic diagram of the structure of the bidirectional long short-term memory network provided by the present invention; Figure 6 It is a schematic diagram of the structure of the ultra-high performance liquid chromatography data preprocessing system based on a bidirectional long short-term memory network provided by the present invention; Figure 7 It is a schematic diagram of the overall structure of the ultra-high performance liquid chromatography data preprocessing system based on a bidirectional long short-term memory network provided by the present invention; Figure 8 It is a schematic diagram of the structure of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] Figure 1 The following is a schematic flowchart of the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network provided by the present invention. As Figure 1 shown, the present invention provides an ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network, including: Step 101: Obtain the target ultra-high performance liquid chromatography data to be processed.

[0021] In the present invention, according to the characteristics of the sample to be tested, appropriate chromatographic conditions and parameters are first selected, such as flow rate, column temperature, detection wavelength, etc. Then, the sample to be tested is subjected to corresponding treatments, such as dissolution, dilution, and filtration, etc., so as to ensure that a clear chromatogram can be generated after the sample to be tested is injected into the UHPLC system. Further, the processed sample to be tested is injected into the UHPLC system, and through the corresponding data acquisition software, the chromatogram and related data are recorded to obtain the ultra-high performance liquid chromatography data of the sample to be tested, that is, the target ultra-high performance liquid chromatography data to be processed. In the present invention, in order to better analyze the target ultra-high performance liquid chromatography data, a subsequent noise reduction preprocessing process needs to be performed on the target ultra-high performance liquid chromatography data.

[0022] Step 102: Input the target ultra-high performance liquid chromatography data into the liquid chromatography data preprocessing parameter prediction model to obtain the target preprocessing adjustment parameters output by the liquid chromatography data preprocessing parameter prediction model; wherein, the liquid chromatography data preprocessing parameter model is trained based on the sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels.

[0023] In the present invention, through the pre-trained liquid chromatography data preprocessing parameter prediction model, the optimal preprocessing adjustment parameters for specific ultra-high performance liquid chromatography data are automatically determined, and these adjustment parameters will be used to optimize the original data (i.e., the target ultra-high performance liquid chromatography data) to improve the accuracy and reliability of subsequent analysis.

[0024] Specifically, in the present invention, in the model training stage, a large number of sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels need to be collected. These sample ultra-high performance liquid chromatography data should cover a variety of experimental conditions, sample types, and instrument settings to ensure the generalization ability of the liquid chromatography data preprocessing parameter prediction model.

[0025] In the present invention, for each sample ultra-high performance liquid chromatography (UHPLC) data, it is necessary to obtain the optimal pretreatment adjustment parameters, which include baseline correction parameters, smoothing parameters, noise reduction parameters, peak identification parameters, and so on. Further, the labeled sample UHPLC data is input into a deep learning algorithm for model training. During the training process, the deep learning algorithm will learn the mapping relationship between the data features and the pretreatment adjustment parameters, thereby constructing a prediction model for the pretreatment parameters of liquid chromatography data.

[0026] When new UHPLC data (i.e., target UHPLC data) needs to be processed, first, these target UHPLC data are input into the trained prediction model for the pretreatment parameters of liquid chromatography data. The prediction model for the pretreatment parameters of liquid chromatography data predicts the optimal pretreatment adjustment parameters according to the features of the input data, and these parameters will be used to guide the subsequent pretreatment operations.

[0027] Step 103: Based on the target pretreatment adjustment parameters, perform noise reduction pretreatment on the target UHPLC data to obtain the noise reduction result data corresponding to the target UHPLC data.

[0028] In the present invention, according to the target pretreatment adjustment parameters, the noise reduction algorithms corresponding to the target UHPLC data are determined, including moving average filtering, median filtering, Kalman filtering, and wavelet transform, etc. These algorithms are applicable to different types of noise and data.

[0029] Further, the selected noise reduction algorithm is applied to the target UHPLC data. During the algorithm application process, the target UHPLC data will be processed according to the preset parameters to remove noise and retain useful information. After the noise reduction process, the noise reduction result data can also be evaluated, which involves comparing the differences in the data before and after noise reduction, and evaluating whether the noise reduction effect meets the preset quality standards. At the same time, during the noise reduction pretreatment process, it is necessary to ensure that the integrity of the data is not affected, including the continuity and consistency of the data, etc., to avoid introducing new errors or biases.

[0030] The method for pretreating ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network provided by the present invention predicts the pretreatment adjustment parameters of the target ultra-high performance liquid chromatography data to be processed through a prediction model for the pretreatment parameters of liquid chromatography data trained by sample ultra-high performance liquid chromatography data labeled with pretreatment adjustment parameter tags, and then performs noise reduction pretreatment on the target ultra-high performance liquid chromatography data according to the target pretreatment adjustment parameters predicted by the prediction model for the pretreatment parameters of liquid chromatography data, thereby improving the noise reduction effect of ultra-high performance liquid chromatography data.

[0031] Based on the above embodiments, the target preprocessing adjustment parameters include a first target preprocessing adjustment parameter, a second target preprocessing adjustment parameter, and a third target preprocessing adjustment parameter; wherein, the first target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the wavelet threshold denoising process, the second target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the baseline drift correction process, and the third target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the smoothing filtering process; Based on the target preprocessing adjustment parameters, performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data to obtain the denoising result data corresponding to the target ultra-high performance liquid chromatography data, including: Based on the first target preprocessing adjustment parameter, performing wavelet threshold denoising processing on the target ultra-high performance liquid chromatography data to obtain the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing; Based on the second target preprocessing adjustment parameter, performing baseline drift correction processing on the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing to obtain the target ultra-high performance liquid chromatography data after baseline drift correction processing; Based on the third target preprocessing adjustment parameter, performing smoothing filtering processing on the target ultra-high performance liquid chromatography data after baseline drift correction processing to obtain the denoising result data.

[0032] In the present invention, the first target preprocessing adjustment parameters include the wavelet basis type, the threshold processing selection strategy (such as hard threshold, soft threshold or adaptive threshold), and the threshold size, etc. These parameters together determine the effect of wavelet threshold denoising, so as to select different wavelet bases and thresholds for different inputs to improve the denoising effect. Based on these parameters, the wavelet threshold denoising algorithm is applied to the target ultra-high performance liquid chromatography data. This algorithm decomposes the data into different scales through wavelet transform, and then applies threshold processing to the coefficients at each scale to remove the noise components. Finally, the data is reconstructed through inverse wavelet transform to obtain the denoised result, that is, the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing. Figure 2 It is a schematic diagram of the effect of wavelet transform denoising based on multi-threshold and multi-wavelet basis selection provided by the present invention. The wavelet threshold denoising effect in the present invention can be referred to Figure 2 as shown.

[0033] The second target preprocessing adjustment parameters include the type of correction algorithm (such as polynomial fitting, moving average filtering, etc.) and the range of fitting orders, etc. These parameters determine the accuracy and smoothness of baseline drift correction. Different polynomial fitting orders are selected for different inputs to reduce the mean square error between the data after baseline drift correction and the real data. In the present invention, based on the data after wavelet threshold denoising processing, a baseline drift correction algorithm is applied. By fitting the baseline trend of the data and subtracting this trend from the original data, the influence of baseline drift is eliminated. Figure 3 The figure shows the effect diagram of baseline drift correction based on polynomial fitting provided by the present invention. The effect of baseline drift correction in the present invention can be referred to Figure 3 as shown.

[0034] The third target preprocessing adjustment parameters include the type of filter (such as Gaussian filter, mean filter, etc.), the window size of the filter, and the range of polynomial orders, etc. These parameters determine the balance between the degree of smoothing filtering and the retention of details. Different windows and polynomial orders are selected for different inputs to improve the noise suppression effect without distortion. Based on the data after baseline drift correction processing, a smoothing filtering algorithm is applied to smooth the high-frequency fluctuations in the data, further reducing noise while trying to retain the useful information in the data. Figure 4 The figure shows the effect diagram of smoothing filtering based on Savitzky-Golay filter provided by the present invention. The effect of smoothing filtering in the present invention can be referred to Figure 4 as shown.

[0035] Specifically, in the present invention, first, through the liquid chromatography data preprocessing parameter prediction model, the parameters required in the noise reduction preprocessing process of the target ultra-high performance liquid chromatography data (i.e., the target preprocessing adjustment parameters) are predicted. Then, based on the first target preprocessing adjustment parameter in the predicted target preprocessing adjustment parameters, wavelet threshold denoising processing is performed on the target ultra-high performance liquid chromatography data to obtain the preliminarily denoised data, that is, the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing. Further, based on the second target preprocessing adjustment parameter in the target preprocessing adjustment parameters, baseline drift correction processing is performed on the preliminarily denoised data to obtain the data after baseline drift correction. Finally, based on the third target preprocessing adjustment parameter in the target preprocessing adjustment parameters, smoothing filtering processing is performed on the data after baseline drift correction to obtain the final noise reduction result data.

[0036] Based on the above embodiments, the liquid chromatography data preprocessing parameter prediction model is trained through the following steps: Obtain sample ultra-high performance liquid chromatography data; A noise reduction preprocessing process based on multiple different noise reduction preprocessing types is used to perform noise reduction preprocessing on the sample ultra-high performance liquid chromatography data in sequence. After determining that the sample ultra-high performance liquid chromatography data after noise reduction preprocessing meets the preset noise reduction requirements, the preprocessing adjustment parameters corresponding to each noise reduction preprocessing process are obtained; Each of the preprocessing adjustment parameter tags corresponding to each noise reduction preprocessing process is marked on the sample ultra-high performance liquid chromatography data to obtain a training sample set, where the preprocessing adjustment parameter tag is constructed based on the preprocessing adjustment parameter; Based on the training sample set, a bidirectional long short-term memory network is trained to obtain the liquid chromatography data preprocessing parameter prediction model.

[0037] In the present invention, sample ultra-high performance liquid chromatography data is first collected from experiments or databases. Ultra-high performance liquid chromatography data is a chromatographic technique used to separate, identify, and quantify compounds in complex mixtures. The collected sample ultra-high performance liquid chromatography data includes a series of signal intensities that change over time, and these signal intensities reflect the concentrations of different compounds in the sample. Further, feature extraction is performed on these sample ultra-high performance liquid chromatography data, for example, statistical features (mean, variance) and frequency domain features (FFT), so as to convert the original signal into a feature matrix. When constructing the training sample set subsequently, corresponding tags are marked on these features.

[0038] In data analysis, noise reduction is an important step that can remove noise and interference in the original data and improve data quality. In the present invention, multiple different noise reduction methods (such as moving average filtering, wavelet transform, Kalman filtering, etc.) are tried to process the sample ultra-high performance liquid chromatography data, and each method has its specific parameter settings, and these parameters will affect the noise reduction effect.

[0039] For each effective noise reduction method, specific parameter settings (such as the window size of the filter, the number of levels of wavelet transform, etc.) are obtained to obtain the preprocessing adjustment parameters. Further, the preprocessing adjustment parameters used in each noise reduction preprocessing process are associated with the corresponding sample ultra-high performance liquid chromatography data, and these data samples are labeled, and these labels contain information on the preprocessing adjustment parameters used to generate the noise-reduced data.

[0040] Through the above steps, the present invention can construct a training sample set containing sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter tags, and this training sample set will be used to train a deep learning model.

[0041] The Bidirectional Long Short-Term Memory (Bi-LSTM) network is a special type of Recurrent Neural Network (RNN) that can capture long-term dependencies in data and is suitable for processing time series data. In the present invention, a training sample set is used to train the Bi-LSTM network, enabling it to learn to predict the optimal preprocessing adjustment parameters based on the characteristics of the sample ultra-high performance liquid chromatography data. Consequently, the parameters to be adjusted during the denoising process of one-dimensional signal data such as spectrograms are specifically adjusted, significantly enhancing the adaptability and flexibility of the model and strengthening the generalization ability and accuracy of the ultra-high performance liquid chromatography data preprocessing process in a diverse data environment. Figure 5 FIG. shows the structural schematic diagram of the bidirectional long short-term memory network provided by the present invention. For the specific structure of the Bi-LSTM network, reference can be made to Figure 5 as shown.

[0042] After training, the obtained Bi-LSTM network can be used as a prediction model for liquid chromatography data preprocessing parameters to automatically predict the optimal denoising preprocessing parameters for new ultra-high performance liquid chromatography data, thereby simplifying the data processing flow and improving the data processing efficiency and accuracy.

[0043] Based on the above embodiments, the step of labeling the corresponding preprocessing adjustment parameter tags in each of the noise reduction preprocessing processes for the sample ultra-high performance liquid chromatography data to obtain a training sample set includes: When the noise reduction preprocessing process is a wavelet threshold denoising process, a first adjustment parameter tag is constructed according to the wavelet basis type and threshold type used in the wavelet threshold denoising process; When the noise reduction preprocessing process is a baseline drift correction process, a second adjustment parameter tag is constructed according to the polynomial fitting order used in the baseline drift correction process; When the noise reduction preprocessing process is a smoothing filtering process, a third adjustment parameter tag is constructed according to the window size and polynomial order used in the smoothing filtering process; According to the first adjustment parameter tag, the second adjustment parameter tag, and the third adjustment parameter tag, the preprocessing adjustment parameter tags corresponding to each of the noise reduction preprocessing processes for the sample ultra-high performance liquid chromatography data are constructed; Based on the sample ultra-high performance liquid chromatography data labeled with the preprocessing adjustment parameter tags, the training sample set is constructed.

[0044] In the present invention, when wavelet threshold denoising is selected as the noise reduction preprocessing method, two key factors need to be considered: the type of wavelet basis and the type of threshold. Among them, the types of wavelet bases include Haar wavelet, Daubechies wavelets (db1 to db10), Symlets wavelets, Coiflets wavelets, Biorthogonal wavelets, Meyer wavelet, Gaussian wavelet, and Shannon wavelet. Each wavelet basis has its unique time-frequency characteristics and is suitable for different types of signal features.

[0045] Threshold processing determines how to handle the noise part in the wavelet coefficients. In the present invention, the types of thresholds include hard threshold, soft threshold, adaptive threshold, and deterministic threshold (such as VisuShrink). Different threshold processing methods have a significant impact on the denoising effect.

[0046] During the wavelet threshold processing, according to the difference in the numerical values of the wavelet coefficients of the signal and noise in different frequency bands, a suitable threshold is selected to retain or remove them to achieve noise filtering, and then a pure signal is obtained through wavelet reconstruction. The specific steps are as follows: (1) Wavelet decomposition: Perform wavelet transform to obtain decomposition coefficients; (2) Threshold processing: Perform threshold processing on the coefficients at each scale; (3) Wavelet coefficient reconstruction: Obtain the denoised signal through inverse wavelet transform.

[0047] Furthermore, based on the above-mentioned various types of wavelet bases and threshold types, a first adjustment parameter label can be constructed, which uniquely identifies the specific parameter combination used in the wavelet threshold denoising process of the sample ultra-high performance liquid chromatography data.

[0048] Baseline drift is a common problem in chromatographic data. When the present invention uses polynomial fitting for correction, the key lies in selecting an appropriate polynomial fitting order. In the present invention, polynomials of order 3 to 6 are used for fitting because too low an order may not accurately describe the complex changes of the baseline, while too high an order may lead to overfitting and introduce unnecessary fluctuations.

[0049] In the present invention, the specific steps of the baseline drift correction process based on polynomial fitting are as follows: (1) Select an appropriate polynomial fitting order according to the characteristics of the signal; (2) Perform polynomial fitting. (3) Calculate the fitted baseline, subtract the fitted baseline from the original signal to obtain the corrected signal.

[0050] Then, according to the baseline drift correction effect of the sample ultra-high performance liquid chromatography data (it should be noted that the sample ultra-high performance liquid chromatography data here has undergone wavelet denoising processing) under different polynomial fitting orders, a corresponding label, that is, a second adjustment parameter label, is constructed for this sample ultra-high performance liquid chromatography data.

[0051] Smoothing filtering is very effective in removing high-frequency noise. In the present invention, a Savitzky-Golay filter is adopted, and two main parameters are considered: window size and polynomial order. Specifically, the window size determines the number of data points participating in the smoothing calculation, ranging from 2 to 15; the choice of polynomial order affects the smoothing effect and signal fidelity. The polynomial order used in the Savitzky-Golay filter in the present invention is 1 or 2.

[0052] In the present invention, the data is fitted by polynomial least squares method, and then the window size and order are adjusted to be applicable to various data. Specifically, the Savitzky-Golay filter is adopted in the present invention. First, a suitable window size is selected, and the values in the window are fitted by polynomial least squares. Then, the obtained data is used as the value at the middle position of the filtered window, and the signal is filtered by continuously moving the window.

[0053] Furthermore, according to the smoothing filtering effect of the sample ultra-high performance liquid chromatography data (it should be noted that the sample ultra-high performance liquid chromatography data here has been subjected to wavelet denoising processing and baseline drift correction processing) under different window sizes and polynomial orders, a corresponding label, that is, the third adjustment parameter label, is constructed for this sample ultra-high performance liquid chromatography data.

[0054] Finally, the above three adjustment parameter labels (the first adjustment parameter label, the second adjustment parameter label, and the third adjustment parameter label) are combined to generate a complete preprocessing adjustment parameter label for each sample ultra-high performance liquid chromatography data in each noise reduction preprocessing process. These labels record the specific parameters of each preprocessing method, and thus a training sample set can be constructed based on these labels.

[0055] In the present invention, an unprocessed sample ultra-high performance liquid chromatography data, based on the combined conditions of different preprocessing adjustment parameters, after wavelet threshold denoising processing, baseline drift correction processing, and smoothing filtering processing, will generate 17 (wavelet basis) × 4 (threshold) × 4 (polynomial fitting order) × 14 (window size) × 2 (polynomial order) = 7616 denoised preprocessing data, and then the preprocessing adjustment parameters corresponding to the sample ultra-high performance liquid chromatography data with better denoising effect are selected as labels for training.

[0056] Based on the above embodiments, the training of the bidirectional long short-term memory network according to the training sample set to obtain the liquid chromatography data preprocessing parameter prediction model includes: Based on the training sample set and a preset loss function, the bidirectional long short-term memory network is trained until the value of the loss function in the training process converges, obtaining the prediction model for liquid chromatography data preprocessing parameters, where the preset loss function is constructed based on the mean square error and the peak-to-peak signal-to-noise ratio.

[0057] In the present invention, the Bi-LSTM network can consider both the forward and reverse information of the sequence, thereby improving the model's ability to understand context information. The training sample set contains a large amount of liquid chromatography data and their corresponding preprocessing parameters. These data are used to train the Bi-LSTM network so that it can learn the mapping relationship from the original liquid chromatography data to the preprocessing parameters.

[0058] The loss function is used to evaluate the difference between the predicted value and the actual value of the model. The goal of the training process is to minimize this loss function. In the present invention, a preset loss function is constructed through the mean square error (MSE) and the peak-to-peak (Peak-to-Peak) signal-to-noise ratio. Among them, the mean square error is a commonly used method to measure the difference between the predicted value and the actual value, and is used to calculate the average of the squares of the differences between the predicted value and the actual value. The smaller the MSE, the more accurate the prediction of the model. The peak-to-peak signal-to-noise ratio reflects the ratio between the signal peak and the noise peak. A high signal-to-noise ratio means that the signal is clearer and the noise interference is smaller.

[0059] Specifically, the calculation formula for the mean square error is: ; where N represents the total number of data points, is the i-th value of the original signal (i.e., the signal without denoising), is the i-th value of the denoised signal.

[0060] The calculation formula for the peak-to-peak signal-to-noise ratio is: ; where is the square of the signal peak, is the square of the root mean square value of the noise.

[0061] In the present invention, the preset loss function is constructed based on the weighted sum of the mean square error and the peak-to-peak signal-to-noise ratio. The weights can adjust the importance of the two according to actual needs. For example, if more attention is paid to the accuracy of the prediction, a higher weight can be assigned to the mean square error; if more attention is paid to the clarity of the signal, a higher weight can be assigned to the peak-to-peak signal-to-noise ratio.

[0062] During the training process, the Bi-LSTM network continuously adjusts its internal parameters through the backpropagation algorithm to minimize the preset loss function value. When the change in the loss function value is very small within a certain number of iterations or reaches the preset threshold, it is considered that the training process has converged, and the model obtained at this time is the required prediction model for the pretreatment parameters of liquid chromatography data.

[0063] The ultra-high performance liquid chromatography data pretreatment system based on the bidirectional long short-term memory network provided by the present invention will be described below. The ultra-high performance liquid chromatography data pretreatment system based on the bidirectional long short-term memory network described below can be correspondingly referred to the ultra-high performance liquid chromatography data pretreatment method described above.

[0064] Figure 6 It is a schematic structural diagram of the ultra-high performance liquid chromatography data pretreatment system based on the bidirectional long short-term memory network provided by the present invention. As Figure 6 shown, the present invention provides an ultra-high performance liquid chromatography data pretreatment system based on the bidirectional long short-term memory network, including a liquid chromatography data acquisition module 601, a pretreatment parameter prediction module 602, and a pretreatment module 603. Among them, the liquid chromatography data acquisition module 601 is used to obtain the target ultra-high performance liquid chromatography data to be processed; the pretreatment parameter prediction module 602 is used to input the target ultra-high performance liquid chromatography data into the liquid chromatography data pretreatment parameter prediction model to obtain the target pretreatment adjustment parameter output by the liquid chromatography data pretreatment parameter prediction model; wherein, the liquid chromatography data pretreatment parameter model is trained based on the sample ultra-high performance liquid chromatography data marked with the pretreatment adjustment parameter label; the pretreatment module 603 is used to perform noise reduction pretreatment on the target ultra-high performance liquid chromatography data based on the target pretreatment adjustment parameter to obtain the noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

[0065] The ultra-high performance liquid chromatography data pretreatment system provided by the present invention predicts the pretreatment adjustment parameter of the target ultra-high performance liquid chromatography data to be processed through the liquid chromatography data pretreatment parameter model trained by the sample ultra-high performance liquid chromatography data marked with the pretreatment adjustment parameter label, and then performs noise reduction pretreatment on the target ultra-high performance liquid chromatography data according to the target pretreatment adjustment parameter predicted by the liquid chromatography data pretreatment parameter model, thereby improving the noise reduction effect of the ultra-high performance liquid chromatography data.

[0066] Based on the above embodiments, the target preprocessing adjustment parameters include a first target preprocessing adjustment parameter, a second target preprocessing adjustment parameter, and a third target preprocessing adjustment parameter; wherein, the first target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the wavelet threshold denoising process, the second target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the baseline drift correction process, and the third target preprocessing adjustment parameter includes the preprocessing adjustment parameter corresponding to the smoothing filtering process; The preprocessing module includes a wavelet transform denoising module, a baseline drift correction module, and a smoothing filtering module. Among them, the wavelet transform denoising module is used to perform wavelet threshold denoising processing on the target ultra-high performance liquid chromatography data based on the first target preprocessing adjustment parameter to obtain the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing; the baseline drift correction module is used to perform baseline drift correction processing on the target ultra-high performance liquid chromatography data after wavelet threshold denoising processing based on the second target preprocessing adjustment parameter to obtain the target ultra-high performance liquid chromatography data after baseline drift correction processing; the smoothing filtering module is used to perform smoothing filtering processing on the target ultra-high performance liquid chromatography data after baseline drift correction processing based on the third target preprocessing adjustment parameter to obtain the noise reduction result data.

[0067] Figure 7 For the overall structural schematic diagram of the ultra-high performance liquid chromatography data preprocessing system based on the bidirectional long short-term memory network provided by the present invention, reference can be made to Figure 7 As shown, in the present invention, in the training stage, the wavelet transform denoising module based on the selection of multiple thresholds and multiple wavelet bases includes 17 wavelet bases such as Haar wavelet, Daubechies wavelets (db1 to db10), Symlets wavelets, Coiflets wavelets, Biorthogonal wavelets, Meyer wavelets, Gaussian wavelets, and Shannon wavelets; and the threshold processing includes four threshold processing methods such as hard threshold, soft threshold, adaptive threshold, and deterministic threshold (VisuShrink). A sample ultra-high performance liquid chromatography data can generate 17×4 = 68 wavelet denoised data. In the application stage, the wavelet transform denoising module will perform denoising using the wavelet basis and threshold processing in the optimal adjustment parameters (i.e., the target preprocessing adjustment parameters) output by the trained BI-LSTM network (i.e., the liquid chromatography data preprocessing parameter prediction model).

[0068] Further, in the training stage, after the above-mentioned wavelet denoising of a sample ultra-high performance liquid chromatography data, the baseline drift correction module based on polynomial fitting performs 4 times of baseline drift correction on the generated 68 denoised data at different orders, generating 68×4 = 272 baseline drift corrected data. Among them, different polynomial fitting orders are adopted for each baseline offset correction process. In the present invention, the range of the polynomial fitting order is from 3 to 6. In the application stage, the baseline drift correction module will perform baseline drift correction using the polynomial fitting order in the target preprocessing adjustment parameters.

[0069] Further, in the training stage, after the above-mentioned wavelet denoising and baseline drift correction of a sample ultra-high performance liquid chromatography data, the smoothing filter module based on Savitzky-Golay filter performs smoothing processing on the generated 272 data with different window sizes and different polynomial orders, generating 272×14×2 = 7616 smoothed data. In the present invention, the range of the window size is from 2 to 15, and the range of the polynomial order is from 1 to 2. In the application stage, the smoothing filter module will perform smoothing filtering using the window size and polynomial order in the target preprocessing adjustment parameters.

[0070] In the present invention, in the training stage of the bidirectional long short-term memory network, the original data (i.e., the sample ultra-high performance liquid chromatography data) and the data processing part parameters used in the best denoising result corresponding to the sample ultra-high performance liquid chromatography data are used as labels for training. After the bidirectional long short-term memory network is trained, a liquid chromatography data preprocessing parameter prediction model is obtained. At this time, the new original signal is input to the liquid chromatography data preprocessing parameter prediction model, and the liquid chromatography data preprocessing parameter prediction model will output the preprocessing adjustment parameters required for the data processing part.

[0071] The system provided by the embodiment of the present invention is used to execute the above-mentioned method embodiments. For the specific process and detailed content, please refer to the above embodiments and will not be elaborated here.

[0072] Figure 8 The structural schematic diagram of the electronic device provided by the present invention is as Figure 8As shown in the figure, the electronic device may include: a processor 801, a communications interface 802, a memory 803, and a communication bus 804. Among them, the processor 801, the communications interface 802, and the memory 803 complete communication with each other through the communication bus 804. The processor 801 may call the logical instructions in the memory 803 to execute an ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network. The method includes: obtaining target ultra-high performance liquid chromatography data to be processed; inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain a target preprocessing adjustment parameter output by the liquid chromatography data preprocessing parameter prediction model; where the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter tags; based on the target preprocessing adjustment parameter, performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

[0073] In addition, when the logical instructions in the above-mentioned memory 803 can be implemented in the form of software function units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0074] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network provided by the above-mentioned various methods. The method includes: obtaining target ultra-high performance liquid chromatography data to be processed; inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain a target preprocessing adjustment parameter output by the liquid chromatography data preprocessing parameter prediction model; wherein, the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels; and based on the target preprocessing adjustment parameter, performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

[0075] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is configured to execute the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network provided by the above-mentioned various embodiments. The method includes: obtaining target ultra-high performance liquid chromatography data to be processed; inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain a target preprocessing adjustment parameter output by the liquid chromatography data preprocessing parameter prediction model; wherein, the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels; and based on the target preprocessing adjustment parameter, performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

[0076] The device embodiments described above are merely illustrative. The units described as separation components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0077] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network, characterized in that: include: Acquiring target ultra-high performance liquid chromatography data to be processed; Inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain target preprocessing adjustment parameters output by the liquid chromatography data preprocessing parameter prediction model; wherein the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels; Based on the target preprocessing adjustment parameters, the target ultra-high performance liquid chromatography data is subjected to noise reduction preprocessing to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

2. The method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network according to claim 1, characterized in that: The target preprocessing adjustment parameters include a first target preprocessing adjustment parameter, a second target preprocessing adjustment parameter and a third target preprocessing adjustment parameter; wherein the first target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a wavelet threshold denoising process, the second target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a baseline drift correction process, and the third target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a smoothing filtering process; The method of performing noise reduction preprocessing on the target ultra-high performance liquid chromatography data based on the target preprocessing adjustment parameter to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data includes: Based on the first target preprocessing adjustment parameter, the target ultra-high performance liquid chromatography data is subjected to wavelet threshold denoising to obtain the target ultra-high performance liquid chromatography data after wavelet threshold denoising; Based on the second target preprocessing adjustment parameter, performing baseline drift correction processing on the target ultra-high performance liquid chromatography data after the wavelet threshold denoising processing to obtain the target ultra-high performance liquid chromatography data after the baseline drift correction processing; Based on the third target preprocessing adjustment parameter, the target ultra-high performance liquid chromatography data after the baseline drift correction process is smoothed and filtered to obtain the noise reduction result data.

3. The method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network according to claim 2, characterized in that: The liquid chromatography data preprocessing parameter prediction model is trained by the following steps: Obtain sample ultra-high performance liquid chromatography data; Based on a plurality of noise reduction preprocessing processes of different noise reduction preprocessing types, the sample ultra-high performance liquid chromatography data are subjected to noise reduction preprocessing in sequence, and after determining that the obtained sample ultra-high performance liquid chromatography data after noise reduction preprocessing meets the preset noise reduction requirements, the preprocessing adjustment parameters corresponding to each of the noise reduction preprocessing processes are obtained; Marking the sample ultra-high performance liquid chromatography data with the corresponding preprocessing adjustment parameter labels in each noise reduction preprocessing process to obtain a training sample set, wherein the preprocessing adjustment parameter labels are constructed based on the preprocessing adjustment parameters; According to the training sample set, a bidirectional long short-term memory network is trained to obtain the liquid chromatography data preprocessing parameter prediction model.

4. The method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network according to claim 3, characterized in that: The step of marking the sample ultra-high performance liquid chromatography data with the corresponding preprocessing adjustment parameter labels in each noise reduction preprocessing process to obtain a training sample set includes: When the denoising preprocessing process is a wavelet threshold denoising process, constructing a first adjustment parameter tag according to the wavelet basis type and threshold type used in the wavelet threshold denoising process; When the noise reduction preprocessing process is a baseline drift correction process, constructing a second adjustment parameter label according to the polynomial fitting order used in the baseline drift correction process; When the noise reduction preprocessing process is a smoothing filter processing process, constructing a third adjustment parameter tag according to the window size and polynomial order used in the smoothing filter processing process; According to the first adjustment parameter tag, the second adjustment parameter tag and the third adjustment parameter tag, constructing the preprocessing adjustment parameter tag corresponding to the sample ultra-high performance liquid chromatography data in each of the noise reduction preprocessing processes; The training sample set is constructed based on the sample ultra-high performance liquid chromatography data marked with the preprocessing adjustment parameter label.

5. The method for preprocessing ultra-high performance liquid chromatography data based on a bidirectional long short-term memory network according to claim 3, characterized in that: The method of training a bidirectional long short-term memory network according to the training sample set to obtain the liquid chromatography data preprocessing parameter prediction model comprises: According to the training sample set and the preset loss function, the bidirectional long short-term memory network is trained until the loss function value in the training process converges, thereby obtaining the liquid chromatography data preprocessing parameter prediction model, wherein the preset loss function is constructed based on the mean square error and the peak-to-peak signal-to-noise ratio.

6. An ultra-high performance liquid chromatography data preprocessing system based on a bidirectional long short-term memory network, characterized in that: include: A liquid chromatography data acquisition module, used for acquiring target ultra-high performance liquid chromatography data to be processed; A preprocessing parameter prediction module, used for inputting the target ultra-high performance liquid chromatography data into a liquid chromatography data preprocessing parameter prediction model to obtain target preprocessing adjustment parameters output by the liquid chromatography data preprocessing parameter prediction model; wherein the liquid chromatography data preprocessing parameter model is trained based on sample ultra-high performance liquid chromatography data marked with preprocessing adjustment parameter labels; A preprocessing module is used to perform noise reduction preprocessing on the target ultra-high performance liquid chromatography data based on the target preprocessing adjustment parameters to obtain noise reduction result data corresponding to the target ultra-high performance liquid chromatography data.

7. The ultra-high performance liquid chromatography data preprocessing system based on a bidirectional long short-term memory network according to claim 6, characterized in that: The target preprocessing adjustment parameters include a first target preprocessing adjustment parameter, a second target preprocessing adjustment parameter and a third target preprocessing adjustment parameter; wherein the first target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a wavelet threshold denoising process, the second target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a baseline drift correction process, and the third target preprocessing adjustment parameter includes a preprocessing adjustment parameter corresponding to a smoothing filtering process; The preprocessing module includes a wavelet transform denoising module, a baseline drift correction module and a smoothing filter module, wherein: The wavelet transform denoising module is used to perform wavelet threshold denoising on the target ultra-high performance liquid chromatography data based on the first target preprocessing adjustment parameter to obtain the target ultra-high performance liquid chromatography data after the wavelet threshold denoising; The baseline drift correction module is used to perform baseline drift correction processing on the target ultra-high performance liquid chromatography data after the wavelet threshold denoising processing based on the second target preprocessing adjustment parameter to obtain the target ultra-high performance liquid chromatography data after the baseline drift correction processing; The smoothing filter module is used to perform smoothing filter processing on the target ultra-high performance liquid chromatography data after the baseline drift correction processing based on the third target preprocessing adjustment parameter to obtain the noise reduction result data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network as described in any one of claims 1 to 5 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network as described in any one of claims 1 to 5 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the ultra-high performance liquid chromatography data preprocessing method based on a bidirectional long short-term memory network as described in any one of claims 1 to 5 is implemented.

Images