Liquid nitrogen preservation effect evaluation method and system based on big data

Abstract

The invention discloses a liquid nitrogen preservation effect evaluation method and system based on big data. The method comprises the steps of data acquisition, data preprocessing, establishment of a liquid nitrogen preservation effect evaluation model and liquid nitrogen preservation effect evaluation. The invention belongs to the field of effect evaluation, and particularly relates to a liquid nitrogen preservation effect evaluation method and system based on big data, according to the scheme, a dynamic activation function is designed by introducing learnable parameters, and the prediction stability is improved; a regularization item is constructed based on the module weight so as to design a loss function, and the prediction accuracy of liquid nitrogen preservation effect evaluation is improved; through gradual updating module by module, the capability of capturing subtle changes of liquid nitrogen preservation effect evaluation data is enhanced; and the liquid nitrogen preservation effect evaluation efficiency is improved based on secondary dynamic optimization correction parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of effect evaluation, and specifically to a liquid nitrogen preservation effect evaluation method and system based on big data. Background Art

[0002] The liquid nitrogen preservation effect evaluation method is a method used to monitor and evaluate the quality and stability of samples or biological materials preserved under ultra-low temperature liquid nitrogen conditions. Liquid nitrogen preservation is widely used in the long-term preservation of biological samples, cells, tissues and other materials to ensure that their activity, quality and characteristics do not change significantly over time. However, the general liquid nitrogen preservation effect evaluation method has insufficient processing capabilities for the distribution of liquid nitrogen data under different preservation conditions, and has a serious impact on the noise of liquid nitrogen preservation data, which leads to poor accuracy of the preservation effect evaluation results; the general liquid nitrogen preservation effect evaluation method has poor nonlinear response capabilities to the liquid nitrogen preservation process, and cannot achieve rapid and real-time liquid nitrogen preservation effect evaluation. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a liquid nitrogen preservation effect evaluation method and system based on big data. In view of the problem that the general liquid nitrogen preservation effect evaluation method has insufficient processing capacity for the distribution of liquid nitrogen data under different preservation conditions, and has a serious impact on the noise of liquid nitrogen preservation data, which leads to poor accuracy of the preservation effect evaluation results, this scheme introduces learnable parameters to design a dynamic activation function to improve the stability of the prediction; constructs a regularization term based on the module weight and then designs a loss function to improve the prediction accuracy of the liquid nitrogen preservation effect evaluation; in view of the problem that the general liquid nitrogen preservation effect evaluation method has poor nonlinear response ability to the liquid nitrogen preservation process and cannot realize fast and real-time liquid nitrogen preservation effect evaluation, this scheme enhances the ability to capture subtle changes in liquid nitrogen preservation effect evaluation data by gradually updating each module; and improves the efficiency of liquid nitrogen preservation effect evaluation based on secondary dynamic optimization correction parameters.

[0004] The technical solution adopted by the present invention is as follows: The liquid nitrogen preservation effect evaluation method based on big data provided by the present invention comprises the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: establishing a liquid nitrogen preservation effect evaluation model;

[0008] Step S4: Evaluation of liquid nitrogen preservation effect.

[0009] Furthermore, in step S1, the data collection is to collect historical liquid nitrogen preservation monitoring data; the historical liquid nitrogen preservation monitoring data includes environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters, equipment performance parameters and preservation effect evaluation level; the preservation effect evaluation level is used as a data label.

[0010] Furthermore, in step S2, the data preprocessing is to clean, convert and standardize the collected data; the data cleaning is to process missing values ​​and duplicate values; the data conversion is to convert the data into a vector form; and the standardization is to standardize the data based on the maximum and minimum normalization method.

[0011] Further, in step S3, the establishment of the liquid nitrogen preservation effect evaluation model specifically includes the following steps:

[0012] Step S31: Model architecture design; the liquid nitrogen preservation effect evaluation model adopts a recurrent probabilistic neural network as the basic architecture, and introduces a dynamic activation function and a parallel gradient update mechanism; the input layer inputs a feature vector, which contains relevant parameters in the liquid nitrogen preservation process; the hidden layer uses K PS modules, each module contains a weight vector, which is used to calculate and transmit information layer by layer; each PS module calculates and outputs an intermediate value, which is expressed as: ;in, is the output of the lth sample in the kth PS module; α k is the weight of the kth PS module; k1 is the total number of weights within the module, and i is the weight index within the module; is the i-th weight vector in the k-th PS module; is the input feature vector of the lth sample;

[0013] Step S32: Dynamic activation function design; dynamically adjust the shape of the activation function through adjustable parameters to adapt to sample data under different liquid nitrogen storage conditions; the dynamic activation function is expressed as: ; where g(·) is a dynamic activation function; a, b, and c are learnable parameters used to dynamically adjust the shape of the activation function; x is the activation function input;

[0014] Step S33: loss function definition; the loss function is used to evaluate the difference between the output of the model and the actual preservation effect, expressed as: ; Where E is the loss function; L is the total number of samples, and l is the sample index; is the true label; is the predicted label; is the regularization coefficient;

[0015] Step S34: parameter update: using optimized parallel gradient parameter update, specifically including:

[0016] Step S341: Calculate partial derivatives; expressed as: ; ;

[0017] ; ;in, , , and They are the loss functions for , , and The partial derivative of , and They are a, b and c during the mth training respectively; is the derivative of the activation function; yes Output during the mth training;

[0018] Step S342: Adaptive update; expressed as: ; ; ; ;in, , , and are a, b, c and ; is the mth training ; is the learning rate;

[0019] Step S343: gradually update the PS module weights; after the adaptive update is completed, the weight vector of each PS module is gradually updated in sequence, and for the weight vector of the t-th PS module, the gradient of the loss function with respect to the weight vector is calculated, which is expressed as: ; Update the weight vector of the t-th PS module, and update the weight vector based on the gradient descent method. The gradient of the j-th weight vector is expressed as: ; After completing the update of all weight vectors, the new weight parameter W is obtained m+1 ; Where i1 is the index of the first t-1 PS modules; and are the outputs of the i1th PS module for the lth sample during the m+1th and mth training respectively; i2 is the weight vector index;

[0020] Step S35: model determination; the pre-processed historical liquid nitrogen preservation monitoring data is randomly divided into a training set and a test set; when the loss of the liquid nitrogen preservation effect evaluation model for the training set converges, the model training is completed; a prediction threshold is set in advance, and when the prediction accuracy of the trained liquid nitrogen preservation effect evaluation model for the test set is higher than the prediction threshold, the liquid nitrogen preservation effect evaluation model is established; otherwise, the initial parameters are corrected and the data set training model is re-divided;

[0021] Step S36: initial parameter correction; specifically including:

[0022] Step S361: Establish parameter correction space based on initial module weights and initial activation function learnable parameters, regularization coefficients and learning rates; introduce dual dynamic mapping, expressed as: ; Initialize the individual particle position, expressed as: ; Among them, y u+1 and u are the dynamic values ​​of the u+1th and uth individuals respectively; μ is the dynamic factor; x u+1,d is the position of the u+1th individual in the dth dimension; L d and U d are the lower and upper limits of the dth dimension respectively;

[0023] Step S362: Dynamic correction; update the position based on the gray wolf search algorithm; use the prediction accuracy of the liquid nitrogen preservation effect evaluation model based on the individual position training for the test set as the individual fitness value; select the individual with the largest individual fitness value as the optimal individual; dynamically correct the optimal individual, expressed as: ; ;in, is the dynamic position of the u-th individual in the d-th dimension; is the optimal position of the individual at the Tth iteration; and are the upper and lower bounds of the dth dimension in the population, respectively; is the position of the optimal individual after dynamic correction;

[0024] Step S363: iterative determination; if there is an individual fitness value higher than the prediction threshold, the correction iteration ends and the established liquid nitrogen preservation effect evaluation model is obtained; if the maximum number of iterations is reached, return to step S361; otherwise, repeat the position update and dynamic correction.

[0025] Furthermore, in step S4, the liquid nitrogen preservation effect evaluation is based on the established liquid nitrogen preservation effect evaluation model, and environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters and equipment performance parameters are collected in real time, and are input into the model after preprocessing, and the preservation effect evaluation level predicted by the model is used as the liquid nitrogen preservation effect evaluation result.

[0026] The liquid nitrogen preservation effect evaluation system based on big data provided by the present invention comprises a data acquisition module, a data preprocessing module, a liquid nitrogen preservation effect evaluation model establishment module and a liquid nitrogen preservation effect evaluation module;

[0027] The data acquisition module collects historical liquid nitrogen storage monitoring data;

[0028] The data preprocessing module performs data cleaning, data conversion and standardization on the collected data;

[0029] The liquid nitrogen preservation effect evaluation model establishment module introduces learnable parameters to design a dynamic activation function; constructs a regularization term based on the module weight and then designs a loss function; and establishes a liquid nitrogen preservation effect evaluation model based on module-by-module gradual updating and secondary dynamic optimization correction parameters;

[0030] The liquid nitrogen preservation effect evaluation module implements liquid nitrogen preservation effect evaluation on real-time collected data based on the established liquid nitrogen preservation effect evaluation model.

[0031] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0032] (1) In view of the problem that the general liquid nitrogen preservation effect evaluation method has insufficient processing capabilities for the distribution of liquid nitrogen data under different preservation conditions, which has a serious impact on the noise of liquid nitrogen preservation data and leads to poor accuracy of preservation effect evaluation results, this scheme introduces learnable parameters to design a dynamic activation function to improve the stability of prediction; constructs a regularization term based on the module weight and then designs a loss function to improve the prediction accuracy of liquid nitrogen preservation effect evaluation.

[0033] (2) In view of the problem that the general liquid nitrogen preservation effect evaluation method has poor ability to respond to the nonlinearity of the liquid nitrogen preservation process and cannot achieve rapid and real-time liquid nitrogen preservation effect evaluation, this solution gradually updates each module to enhance the ability to capture subtle changes in the liquid nitrogen preservation effect evaluation data; based on the secondary dynamic optimization correction parameters, the efficiency of liquid nitrogen preservation effect evaluation is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A schematic diagram of the process of the liquid nitrogen preservation effect evaluation method based on big data provided by the present invention;

[0035] Figure 2 A schematic diagram of a liquid nitrogen preservation effect evaluation system based on big data provided by the present invention;

[0036] Figure 3 It is a schematic diagram of the process of step S3.

[0037] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0040] Example 1, see Figure 1 The present invention provides a method for evaluating the liquid nitrogen preservation effect based on big data, and the method comprises the following steps:

[0041] Step S1: data collection, collecting historical liquid nitrogen storage monitoring data;

[0042] Step S2: Data preprocessing, data cleaning, data conversion and standardization of the collected data;

[0043] Step S3: Establish a liquid nitrogen preservation effect evaluation model, introduce learnable parameters to design a dynamic activation function; construct a regularization term based on the module weight and then design a loss function; establish a liquid nitrogen preservation effect evaluation model based on module-by-module gradual updating and secondary dynamic optimization correction parameters;

[0044] Step S4: Liquid nitrogen preservation effect evaluation: liquid nitrogen preservation effect evaluation is performed on the real-time collected data based on the established liquid nitrogen preservation effect evaluation model.

[0045] Example 2, see Figure 1This embodiment is based on the above embodiment. In step S1, the historical liquid nitrogen preservation monitoring data includes environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters, equipment performance parameters and preservation effect evaluation level; the preservation effect evaluation level is used as a data label; the environmental parameters include liquid nitrogen temperature, temperature change rate, humidity, preservation container pressure and container type; the preservation operation parameters include pre-freezing pretreatment time, preservation time, access times and retrieval duration; the sample characteristic parameters include sample type, sample density, sample volume, sample concentration and cryoprotectant type; the preservation medium parameters include preservation solution type, preservation solution pH value and preservation solution ion strength; the equipment performance parameters include liquid nitrogen tank model, liquid nitrogen tank sealing and liquid nitrogen addition frequency; the preservation effect evaluation level includes normal preservation and abnormal preservation.

[0046] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, data cleaning is to process missing values ​​and duplicate values; data conversion is to convert data into vector form; and standardization is to standardize data based on the maximum and minimum normalization method.

[0047] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S3, establishing a liquid nitrogen preservation effect evaluation model specifically includes the following steps:

[0048] Step S31: Model architecture design; the liquid nitrogen preservation effect evaluation model adopts a recurrent probabilistic neural network as the basic architecture, and introduces a dynamic activation function and a parallel gradient update mechanism; the input layer inputs a feature vector, which contains relevant parameters in the liquid nitrogen preservation process; the hidden layer uses K PS modules, each module contains a weight vector, which is used to calculate and transmit information layer by layer; each PS module calculates and outputs an intermediate value, which is expressed as: ;in, is the output of the lth sample in the kth PS module; α k is the weight of the kth PS module; k1 is the total number of weights within the module, and i is the weight index within the module; is the i-th weight vector in the k-th PS module; is the input feature vector of the lth sample;

[0049] Step S32: Dynamic activation function design; dynamically adjust the shape of the activation function through adjustable parameters to adapt to sample data under different liquid nitrogen storage conditions; the dynamic activation function is expressed as: ; where g(·) is a dynamic activation function; a, b, and c are learnable parameters used to dynamically adjust the shape of the activation function; x is the activation function input;

[0050] Step S33: loss function definition; the loss function is used to evaluate the difference between the output of the model and the actual preservation effect, expressed as: ; Where E is the loss function; L is the total number of samples, and l is the sample index; is the true label; is the predicted label; is the regularization coefficient;

[0051] Step S34: parameter update;

[0052] Step S35: model determination; the pre-processed historical liquid nitrogen preservation monitoring data is randomly divided into a training set and a test set; when the loss of the liquid nitrogen preservation effect evaluation model for the training set converges, the model training is completed; a prediction threshold is set in advance, and when the prediction accuracy of the trained liquid nitrogen preservation effect evaluation model for the test set is higher than the prediction threshold, the liquid nitrogen preservation effect evaluation model is established; otherwise, the initial parameters are corrected and the data set training model is re-divided;

[0053] Step S36: initial parameter correction.

[0054] By performing the above operations, the general liquid nitrogen preservation effect evaluation method has insufficient processing capabilities for the distribution of liquid nitrogen data under different preservation conditions, which has a serious impact on the noise of liquid nitrogen preservation data, and thus leads to poor accuracy of preservation effect evaluation results. This scheme introduces learnable parameters to design a dynamic activation function to improve the prediction stability; constructs a regularization term based on the module weight and then designs a loss function to improve the prediction accuracy of liquid nitrogen preservation effect evaluation.

[0055] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment, and in steps S34 and S36, specifically includes the following steps:

[0056] Step S34: parameter update: using optimized parallel gradient parameter update, specifically including:

[0057] Step S341: Calculate partial derivatives; expressed as: ; ;

[0058] ; ;in, , , and They are the loss functions for , , and The partial derivative of , and They are a, b and c during the mth training respectively; is the derivative of the activation function; yes Output during the mth training;

[0059] Step S342: Adaptive update; expressed as: ; ; ; ;in, , , and are a, b, c and ; is the mth training ; is the learning rate;

[0060] Step S343: gradually update the PS module weights; after the adaptive update is completed, the weight vector of each PS module is gradually updated in sequence, and for the weight vector of the t-th PS module, the gradient of the loss function with respect to the weight vector is calculated, which is expressed as: ; Update the weight vector of the t-th PS module, and update the weight vector based on the gradient descent method. The gradient of the j-th weight vector is expressed as: ; After completing the update of all weight vectors, the new weight parameter W is obtained m+1 ; Where i1 is the index of the first t-1 PS modules; and are the outputs of the i1th PS module for the lth sample during the m+1th and mth training respectively; i2 is the weight vector index;

[0061] Step S36: initial parameter correction; specifically including:

[0062] Step S361: Establish parameter correction space based on initial module weights and initial activation function learnable parameters, regularization coefficients and learning rates; introduce dual dynamic mapping, expressed as: ; Initialize the individual particle position, expressed as: ; Among them, y u+1 and u are the dynamic values ​​of the u+1th and uth individuals respectively; μ is the dynamic factor; x u+1,d is the position of the u+1th individual in the dth dimension; L d and U d are the lower and upper limits of the dth dimension respectively;

[0063] Step S362: Dynamic correction; update the position based on the gray wolf search algorithm; use the prediction accuracy of the liquid nitrogen preservation effect evaluation model based on the individual position training for the test set as the individual fitness value; select the individual with the largest individual fitness value as the optimal individual; dynamically correct the optimal individual, expressed as: ; ;in, is the dynamic position of the u-th individual in the d-th dimension; is the optimal position of the individual at the Tth iteration; and are the upper and lower bounds of the dth dimension in the population, respectively; is the position of the optimal individual after dynamic correction;

[0064] Step S363: iterative determination; if there is an individual fitness value higher than the prediction threshold, the correction iteration ends and the established liquid nitrogen preservation effect evaluation model is obtained; if the maximum number of iterations is reached, return to step S361; otherwise, repeat the position update and dynamic correction.

[0065] By executing the above operations, the general liquid nitrogen preservation effect evaluation method has poor nonlinear response ability to the liquid nitrogen preservation process and cannot achieve fast and real-time liquid nitrogen preservation effect evaluation. This solution gradually updates each module to enhance the ability to capture subtle changes in the liquid nitrogen preservation effect evaluation data; based on the secondary dynamic optimization correction parameters, the efficiency of liquid nitrogen preservation effect evaluation is improved.

[0066] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S4, the liquid nitrogen preservation effect evaluation is based on the established liquid nitrogen preservation effect evaluation model. Environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters and equipment performance parameters are collected in real time, and are input into the model after preprocessing. The preservation effect evaluation level predicted by the model is used as the liquid nitrogen preservation effect evaluation result. When the liquid nitrogen preservation effect is evaluated as preservation abnormality, early warning processing is performed.

[0067] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, and the liquid nitrogen preservation effect evaluation system based on big data provided by the present invention includes a data acquisition module, a data preprocessing module, a liquid nitrogen preservation effect evaluation model establishment module and a liquid nitrogen preservation effect evaluation module;

[0068] The data acquisition module collects historical liquid nitrogen storage monitoring data;

[0069] The data preprocessing module performs data cleaning, data conversion and standardization on the collected data;

[0070] The liquid nitrogen preservation effect evaluation model establishment module introduces learnable parameters to design a dynamic activation function; constructs a regularization term based on the module weight and then designs a loss function; and establishes a liquid nitrogen preservation effect evaluation model based on module-by-module gradual updating and secondary dynamic optimization correction parameters;

[0071] The liquid nitrogen preservation effect evaluation module implements liquid nitrogen preservation effect evaluation on real-time collected data based on the established liquid nitrogen preservation effect evaluation model.

[0072] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0073] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0074] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A method for evaluating the effect of liquid nitrogen preservation based on big data, characterized in that: The method comprises the following steps: Step S1: data collection, collecting historical liquid nitrogen storage monitoring data; Step S2: data preprocessing; Step S3: establishing a liquid nitrogen preservation effect evaluation model; Step S4: liquid nitrogen preservation effect evaluation, based on the established liquid nitrogen preservation effect evaluation model, the liquid nitrogen preservation effect is evaluated on the real-time collected data; Step S3 includes step S32: dynamic activation function design; dynamically adjusting the shape of the activation function through adjustable parameters to adapt to sample data under different liquid nitrogen storage conditions; the dynamic activation function is expressed as: ; where g(·) is a dynamic activation function; a, b, and c are learnable parameters used to dynamically adjust the shape of the activation function; and x is the activation function input.

2. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 1, characterized in that: In step S3, the establishment of a liquid nitrogen preservation effect evaluation model specifically includes the following steps: Step S31: Model architecture design; the liquid nitrogen preservation effect evaluation model adopts a recurrent probabilistic neural network as the basic architecture, and introduces a dynamic activation function and a parallel gradient update mechanism; the input layer inputs a feature vector, which contains relevant parameters in the liquid nitrogen preservation process; the hidden layer uses K PS modules, each module contains a weight vector, which is used to calculate and transmit information layer by layer; each PS module calculates and outputs an intermediate value, which is expressed as: ;in, is the output of the lth sample in the kth PS module; α k is the weight of the kth PS module; k1 is the total number of weights within the module, and i is the weight index within the module; is the i-th weight vector in the k-th PS module; is the input feature vector of the lth sample; Step S32: Dynamic activation function design; dynamically adjust the shape of the activation function through adjustable parameters to adapt to sample data under different liquid nitrogen storage conditions; the dynamic activation function is expressed as: ; where g(·) is a dynamic activation function; a, b, and c are learnable parameters used to dynamically adjust the shape of the activation function; x is the activation function input; Step S33: loss function definition; the loss function is used to evaluate the difference between the output of the model and the actual preservation effect, expressed as: ; Where E is the loss function; L is the total number of samples, and l is the sample index; is the true label; is the predicted label; is the regularization coefficient; Step S34: parameter update; Step S35: model determination; the pre-processed historical liquid nitrogen preservation monitoring data is randomly divided into a training set and a test set; when the loss of the liquid nitrogen preservation effect evaluation model for the training set converges, the model training is completed; a prediction threshold is set in advance, and when the prediction accuracy of the trained liquid nitrogen preservation effect evaluation model for the test set is higher than the prediction threshold, the liquid nitrogen preservation effect evaluation model is established; otherwise, the initial parameters are corrected and the data set training model is re-divided; Step S36: initial parameter correction.

3. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 2, characterized in that: In step S34, the parameter update is performed using optimized parallel gradient parameter update, specifically including: Step S341: Calculate partial derivatives; expressed as: ; ; ; ;in, , , and They are the loss functions for , , and The partial derivative of , and They are a, b and c during the mth training respectively; is the derivative of the activation function; yes Output during the mth training; Step S342: Adaptive update; expressed as: ; ; ; ;in, , , and are a, b, c and ; is the mth training ; is the learning rate; Step S343: gradually update the PS module weights; after the adaptive update is completed, the weight vector of each PS module is gradually updated in sequence, and for the weight vector of the t-th PS module, the gradient of the loss function with respect to the weight vector is calculated, which is expressed as: ; Update the weight vector of the t-th PS module, and update the weight vector based on the gradient descent method. The gradient of the j-th weight vector is expressed as: ; After completing the update of all weight vectors, the new weight parameter W is obtained m+1 ; Where i1 is the index of the first t-1 PS modules; and are the outputs of the i1th PS module for the lth sample during the m+1th and mth training respectively; i2 is the weight vector index.

4. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 3, characterized in that: In step S36, the initial parameter correction specifically includes: Step S361: Establish parameter correction space based on initial module weights and initial activation function learnable parameters, regularization coefficients and learning rates; introduce dual dynamic mapping, expressed as: ; Initialize the individual particle position, expressed as: ; Among them, y u+1 and u are the dynamic values ​​of the u+1th and uth individuals respectively; μ is the dynamic factor; x u+1,d is the position of the u+1th individual in the dth dimension; L d and U d are the lower and upper limits of the dth dimension respectively; Step S362: Dynamic correction; update the position based on the gray wolf search algorithm; use the prediction accuracy of the liquid nitrogen preservation effect evaluation model based on the individual position training for the test set as the individual fitness value; select the individual with the largest individual fitness value as the optimal individual; dynamically correct the optimal individual, expressed as: ; ;in, is the dynamic position of the u-th individual in the d-th dimension; is the optimal position of the individual at the Tth iteration; and are the upper and lower bounds of the dth dimension in the population, respectively; is the position of the optimal individual after dynamic correction; Step S363: iterative determination; if there is an individual fitness value higher than the prediction threshold, the correction iteration ends and the established liquid nitrogen preservation effect evaluation model is obtained; if the maximum number of iterations is reached, return to step S361; otherwise, repeat the position update and dynamic correction.

5. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 4, characterized in that: In step S1, the historical liquid nitrogen preservation monitoring data includes environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters, equipment performance parameters and preservation effect evaluation level; the preservation effect evaluation level is used as a data label.

6. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 5, characterized in that: In step S4, the liquid nitrogen preservation effect evaluation is based on the established liquid nitrogen preservation effect evaluation model, and environmental parameters, preservation operation parameters, sample characteristic parameters, preservation medium parameters and equipment performance parameters are collected in real time, and input into the model after preprocessing, and the preservation effect evaluation level predicted by the model is used as the liquid nitrogen preservation effect evaluation result.

7. The method for evaluating liquid nitrogen preservation effect based on big data according to claim 6, characterized in that: In step S2, the data preprocessing is to clean, convert and standardize the collected data; the data cleaning is to process missing values ​​and duplicate values; the data conversion is to convert the data into vector form; the standardization is to standardize the data based on the maximum and minimum normalization method.

8. A liquid nitrogen preservation effect evaluation system based on big data, used to implement the liquid nitrogen preservation effect evaluation method based on big data as described in any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a data preprocessing module, a liquid nitrogen preservation effect evaluation model establishment module and a liquid nitrogen preservation effect evaluation module; The data acquisition module collects historical liquid nitrogen storage monitoring data; The data preprocessing module performs data cleaning, data conversion and standardization on the collected data; The liquid nitrogen preservation effect evaluation model establishment module introduces learnable parameters to design a dynamic activation function; constructs a regularization term based on the module weight and then designs a loss function; and establishes a liquid nitrogen preservation effect evaluation model based on module-by-module gradual updating and secondary dynamic optimization correction parameters; The liquid nitrogen preservation effect evaluation module implements liquid nitrogen preservation effect evaluation on real-time collected data based on the established liquid nitrogen preservation effect evaluation model.

Images