An intelligent environmental monitoring method, system, device and readable storage medium

Abstract

The present invention discloses an intelligent environment monitoring method, system, device and readable storage medium. The method includes: obtaining the current environment data of the totipotent laboratory, performing intelligent data processing on it to obtain accurate current environment data, and performing intelligent analysis on the accurate current environment data to obtain the current analysis result; obtaining the historical environment data of the totipotent laboratory, performing optimization processing on it, and analyzing the optimized historical environment data and the accurate current environment data using the totipotent laboratory environment prediction model to obtain the predictive analysis result; evaluating the test environment status of the totipotent laboratory based on the current analysis result and the predictive analysis result to achieve the environment monitoring of the totipotent laboratory. The present invention solves the problems that the current environment monitoring technology cannot capture the subtle changes of the environment in real time, lacks the prediction ability for future environmental condition changes, and the accuracy of the generated environmental management decisions is relatively low.

Description

Technical Field

[0001] The present invention belongs to the technical field of environmental monitoring, and particularly relates to an intelligent environmental monitoring method, system, device and readable storage medium. Background Art

[0002] In the current all-round laboratory environment, accurate environmental monitoring and management are crucial. Environmental parameters have a significant impact on the accuracy and reliability of experimental results. Therefore, laboratories need an efficient and precise system to monitor these environmental parameters to ensure the best experimental conditions. With the progress of technology, especially in the fields of Internet of Things (IoT) technology and machine learning, the methods of laboratory environmental monitoring are also constantly evolving and improving. Traditional monitoring methods rely on simple sensor networks and manual adjustments. These methods often fail to capture the subtle changes in the environment in real time and lack the ability to predict future changes in environmental conditions. In addition, problems such as data inconsistency, missing values, and noise often interfere with the accuracy of data analysis, resulting in inaccurate environmental management decisions.

[0003] Chinese Patent with Publication No. CN116451865A discloses an intelligent environmental monitoring method based on deep learning, including obtaining historical monitoring data of pollution points; constructing a prediction and early warning model; inputting the historical monitoring data into the prediction and early warning model for training to obtain an environmental prediction result; and formulating a treatment plan based on the environmental prediction result. When extracting data during the environmental monitoring process, this invention does not use special data processing methods, cannot fully express the characteristic information of the data, and only uses one deep learning network model to analyze the historical monitoring data, making it difficult to comprehensively capture the complex relationships contained in the data, resulting in limited analysis results.

[0004] Chinese Patent with Announcement No. CN116485202B discloses a real-time monitoring method and system for industrial pollution based on the Internet of Things, including: obtaining environmental pollution information within a preset area, determining pollution sources and pollution monitoring indicators according to the environmental pollution information; generating a sensor layout plan within the preset area according to the pollution monitoring indicators and pollution source information, obtaining pollution monitoring information collected by different sensors, extracting features from the multi-source pollution monitoring information, constructing a multi-modal feature fusion model for the pollution monitoring information, obtaining the correlation features of different pollution monitoring information, using the correlation features to monitor the industrial pollution status within the preset area, and analyzing the pollution development trend to generate pollution prediction information for pollution early warning. This invention optimizes the layout positions of sensors within the monitoring range of each pollution source in turn through a genetic algorithm and performs conventional preprocessing such as data cleaning on the pollution monitoring information collected by the sensors, but still cannot completely retain the effective data information. Summary of the Invention

[0005] The present invention provides an intelligent environmental monitoring method, system, device and readable storage medium, aiming to solve the problems that the current environmental monitoring technology cannot capture the subtle changes in the environment in real time, lacks the ability to predict the changes in future environmental conditions, and the accuracy of the generated environmental management decisions is relatively low.

[0006] To solve the above technical problems, the present invention provides an intelligent environmental monitoring method, including the following steps:

[0007] Obtain the current environmental data of the totipotency laboratory, perform intelligent data processing on the current environmental data to obtain accurate current environmental data, perform intelligent analysis on the accurate current environmental data using the SVC algorithm to obtain the current analysis result, and the current analysis result is an environmental quantification level parameter, which is used to represent the test environment state of the current totipotency laboratory.

[0008] Obtain the historical environmental data of the totipotency laboratory, perform optimization processing on the historical environmental data, and perform predictive analysis on the optimized historical environmental data and the accurate current environmental data using the totipotency laboratory environment prediction model constructed based on the LSTM network to obtain the predictive analysis result, and the predictive analysis result is the change trend of future environmental parameters.

[0009] Evaluate the test environment state of the totipotency laboratory based on the current analysis result and the predictive analysis result to achieve environmental monitoring of the totipotency laboratory.

[0010] Preferably, the intelligent data processing of the current environmental data is specifically: preprocess the current environmental data, extract the current environmental data features from the preprocessed current environmental data using feature engineering to obtain a multi-source data feature set, and perform data feature fusion processing on the multi-source data feature set using a multi-source data fusion algorithm.

[0011] Preferably, the preprocessing includes noise removal, outlier correction, data normalization and data completion fusion algorithm, and the specific operation of using the data completion fusion algorithm to complete the data of the current environmental data is:

[0012] For each missing data point, find its nearest neighbor data point in the dataset of the current environmental data by calculating the Euclidean distance, select the nearest neighbor data points, and assign weights to the nearest neighbor data points , and the weight calculation formula is specifically:

[0013] ;

[0014] In the formula, is the local gradient rate of the th nearest neighbor data point relative to the missing data point; is a regulating factor used to balance the effects of distance and local gradient rate; is the distance between the th nearest neighbor data point and the missing data point; .

[0015] The imputation value of the missing data point is calculated using a weighted average formula to obtain the imputed data point. The specific formula is:

[0016] ;

[0017] In the formula, is the imputation value of the th missing data point, is the th nearest neighbor data point of the missing data point.

[0018] Data consistency adjustment and enhancement processing is performed on the imputation value of the missing data point. The specific formula is:

[0019] ;

[0020] In the formula, is the non - linear adjustment exponent used to control the intensity and direction of adjustment to adapt to different data distribution characteristics; , are the variance and mean of the overall current environmental data; is the th data point after data consistency adjustment and enhancement processing; is the mean of the nearest neighbor data points; is the variance of the nearest neighbor data points.

[0021] Preferably, the data feature fusion processing using the multi - data fusion algorithm based on the multi - data feature set is specifically as follows:

[0022] The multi - data feature set is standardized to obtain the standardized multi - data feature set, and the standardized multi - data feature sets of samples are used to form a multi - dimensional data feature set .

[0023] Each feature in the multi - dimensional data feature set is non - linearly extended by combining the polynomial kernel and the radial basis kernel. The specific calculation formula is:

[0024] ;

[0025] In the formula, is the feature data of the -th sample on the -th extended feature dimension; is the -th sample of the multi-dimensional data feature set; is the -th sample of the multi-dimensional data feature set; is the kernel function; is the contribution coefficient in the polynomial kernel function, used to adjust the contribution of the -th feature dimension in the extended feature; is the order of the polynomial; is the feature dimension; is the parameter in the radial basis kernel function; is the sample and the square of the Euclidean distance between them; finally, it is composed of to form the feature data on the extended feature dimension .

[0026] Map the feature data on the extended feature dimension through the multi-kernel combination function to obtain the feature data after hybrid kernel mapping. The mapping function is specifically:

[0027] ;

[0028] In the formula, is the feature data after hybrid kernel mapping, denoted by ; is the -th kernel function; is the weight coefficient of the -th kernel function, obtained through Lagrangian optimization; is the number of kernel functions.

[0029] Perform wavelet transform on the mapped feature data, and perform collaborative feature interaction fusion according to the wavelet coefficients. By analyzing the interaction between different features, construct an interaction matrix, and thus generate the final fusion feature set. The calculation formula of the interaction matrix is:

[0030] ;

[0031] In the formula, is the interaction matrix element between the features and after hybrid kernel mapping, representing the interaction strength between features; is the wavelet coefficient of the feature at the -th layer; is a feature the wavelet coefficients of the layer; Adjust the intensity of the interaction between features through periodic changes; is the number of wavelet coefficient layers.

[0032] Finally, calculate the covariance matrix of the feature data for the multi-dimensional data feature set after normalization processing, and calculate the eigenvalues and corresponding eigenvectors of the covariance feature matrix. According to the magnitude of the eigenvalues, select the first eigenvectors corresponding to the largest eigenvalues. The selected eigenvectors are the main features. Based on the selected main feature matrix , obtain the final fused feature representation:

[0033] ;

[0034] wherein, is the final fused feature, that is, accurate environmental parameter data; is through constituted interaction matrix, is the matrix constituted by the selected main features.

[0035] Preferably, the optimization processing of the historical environmental data is specifically: perform data processing on the historical environmental data including filling missing values, removing noise, correcting outliers, and data normalization. After data processing, use the dynamic periodic capture algorithm to reconstruct and optimize the historical environmental data, and use feature engineering to construct a historical data feature set representing the historical environmental data.

[0036] Preferably, the specific operation of using the dynamic periodic capture algorithm to reconstruct and optimize the historical environmental data after data processing is:

[0037] Use Fourier transform to decompose the historical current environmental data at different time scales into components of different frequencies.

[0038] By calculating the autocorrelation of each frequency component and combining its amplitude to determine the periodicity score of the frequency component, specifically:

[0039] ;

[0040] In the formula, is the periodicity score of the th frequency component, is the th frequency component at a delay of autocorrelation function, and the delay is the time interval used in the autocorrelation calculation, representing the a delay; is the number of delays; is the amplitude of the -th frequency component; and

[0041] is the total number of decomposed frequency components. The weights are dynamically assigned using the Softmax function according to the periodicity score of each frequency component.

[0042] The time series is reconstructed using the weighted frequency components. The decomposed and weight-adjusted frequency components are recombined to form the optimized historical and current environmental data. The specific formula is:

[0043] ;

[0044] In the formula, is the optimized historical and current environmental data; is the weight of the -th frequency component; is the frequency of the -th frequency component; is the time; is the phase of the

[0045]

[0046] Preferably, based on the current analysis result and the predictive analysis result, the test environment state of the totipotency laboratory is evaluated to realize the environmental monitoring of the totipotency laboratory, specifically:For the current analysis result, according to the statistical analysis method, the environmental quantification level of the current totipotency laboratory is obtained; for the predictive analysis result, according to the time series analysis technique, the change trend of the future environmental parameters is analyzed using the predictive analysis result to identify the rising and falling fluctuation ranges of the future trend.

[0047] Using the decision tree analysis method, operation suggestions are provided according to the obtained environmental quantification level of the current totipotency laboratory and the change trend of the future environmental parameters to realize the environmental monitoring of the totipotency laboratory.

[0048] On the other hand, the present invention provides an intelligent environmental monitoring system. The system uses an intelligent environmental monitoring method as described in any embodiment of the present invention for environmental monitoring, including a current environmental analysis module, a historical environmental analysis module, and an environmental monitoring and evaluation module.

[0049] The current environmental analysis module is used to obtain the current environmental data of the totipotency laboratory, perform intelligent data processing on the current environmental data to obtain accurate current environmental data, and perform intelligent analysis on the accurate current environmental data using the SVC algorithm to obtain the current analysis result, and the current analysis result is the environmental quantification level parameter;

[0050] A historical environment analysis module, configured to obtain historical environment data of a totipotency laboratory, optimize the historical environment data, perform predictive analysis on the optimized historical environment data and accurate current environment data by using a totipotency laboratory environment prediction model constructed based on an LSTM network, obtain a predictive analysis result, and the predictive analysis result is the change trend of future environmental parameters.

[0051] An environment monitoring and evaluation module, configured to evaluate the test environment status of the totipotency laboratory based on the current analysis result and the predictive analysis result, and implement environment monitoring of the totipotency laboratory.

[0052] On the other hand, the present invention further provides an electronic device, where the electronic device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, it implements an intelligent environment monitoring method as described in any embodiment of the present invention.

[0053] On the other hand, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements an intelligent environment monitoring method as described in any embodiment of the present invention.

[0054] Compared with the prior art, the present invention has the following technical effects:

[0055] 1. By introducing a data completion and fusion algorithm, the present invention can effectively handle the problem of missing values and data inconsistency, ensure the integrity and unity of the data set, and provide accurate data for subsequent data analysis and model training; by calculating the covariance matrix, eigenvalues, eigenvectors, and selecting the main features for data transformation, the present invention realizes the reduction of data dimensions and the effective fusion of information, reduces the complexity of subsequent processing, and enhances the representation ability of the most important information in the data set; the whole process provides a systematic method to process and analyze current environment data, adjusts and optimizes according to different environment monitoring requirements, and improves the adaptability of the processing flow.

[0056] 2. The present invention intelligently analyzes the fused data feature set through SVM technology, can accurately classify the environmental status of the current totipotency laboratory, and represents it with environmental quantification level parameters, improving the accuracy of environmental status monitoring; analyzes historical environmental data using the LSTM network, combines with the dynamic periodic capture algorithm, can accurately capture the non-linear trends and periodic changes in the current environmental data, and the enhanced prediction ability enables laboratory managers to make adjustments in advance to adapt to the expected environmental changes and optimize the operation efficiency of the laboratory; decomposes the historical and current environmental data through Fourier transform, and reconstructs the time series using the dynamic periodic capture algorithm, can deeply understand and reveal the multi-level periodic characteristics in the environmental data, and the high-level understanding helps to predict and respond to complex environmental changes to ensure that the laboratory conditions are always in the optimal state. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is the overall flowchart of an intelligent environmental monitoring method according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present application and with reference to the accompanying drawings.

[0059] Embodiment 1

[0060] This embodiment provides an intelligent environmental monitoring method. Referring to Figure 1 as shown, it includes the following steps:

[0061] Obtain the current environmental data of the totipotency laboratory. The obtained current environmental data is the current environmental data, and the specific acquisition time range of the current data depends on the monitoring frequency and requirements. Further, use sensors to obtain the current environmental data. The sensors can include, for example, temperature sensors, humidity sensors, barometric pressure sensors, etc.; perform intelligent data processing on the current environmental data to obtain accurate current environmental data, perform intelligent analysis on the accurate current environmental data using the SVC algorithm to obtain the current analysis result, and the current analysis result represents the test environmental status of the current totipotency laboratory with environmental quantification level parameters. Specifically, it can be the label parameter data obtained according to different defined numerical ranges for various data such as temperature, humidity, and barometric pressure corresponding to the output sensor detection data.

[0062] Obtain the historical environmental data of the totipotency laboratory, that is, obtain all the data collected before the current environmental data collected currently; perform optimization processing on the historical environmental data, and perform predictive analysis on the optimized historical environmental data and the accurate current environmental data using a totipotency laboratory environmental prediction model constructed based on the LSTM (Long Short-Term Memory) network to obtain a predictive analysis result. The predictive analysis result is the change trend of future environmental parameters, which can specifically be the predicted value of future environmental parameters and the change trend of environmental parameters, and is represented in the form of a prediction report.

[0063] Evaluate the test environment status of the totipotency laboratory based on the current analysis result and the predictive analysis result to achieve environmental monitoring of the totipotency laboratory.

[0064] As a preferred implementation manner of this embodiment, the intelligent data processing of the current environmental data is specifically as follows: perform preprocessing on the current environmental data, and extract the current environmental data features from the preprocessed current environmental data using feature engineering such as statistical analysis methods, time series feature extraction techniques, frequency domain feature extraction techniques, etc. The current environmental data features are such as statistical features (mean, standard deviation, etc.) of the current environmental data, time series features, etc.; obtain a multi-dimensional data feature set, and perform data feature fusion processing on the multi-dimensional data feature set using a multi-dimensional data fusion algorithm.

[0065] As a preferred implementation manner of this embodiment, the preprocessing includes noise removal, outlier correction, data normalization, and data completion fusion algorithm. The data completion fusion algorithm is used to avoid the inconsistency of data and the interference of missing values on the accuracy of subsequent analysis. The specific operation of using the data completion fusion algorithm to perform data completion on the current environmental data is as follows:

[0066] For each missing data point, find its nearest neighbor data point in the dataset of the current environmental data by calculating the Euclidean distance, and select the nearest neighbor data points. In this embodiment, is a parameter preset according to the empirical method, and assign weights to the nearest neighbor data points , and the weight The specific calculation formula is:

[0067] ;

[0068] In the formula, is the local gradient rate of the th nearest neighbor data point relative to the missing data point, that is, the local gradient adjustment factor; is the adjustment factor used to balance the influence of distance and local gradient rate; is the The distance between the k nearest neighbor data points and the missing data point; is a parameter that adjusts the influence of the distance and is used to control how the distance affects the weight assignment; .

[0069] Furthermore, the local gradient rate of the k-th nearest neighbor data point relative to the missing data point is specifically calculated as follows:

[0070] ;

[0071] In the formula, is an adjustment coefficient, is the local density of the k-th nearest neighbor data point, is a small positive number set to avoid division by zero.

[0072] The imputation value of the missing data point is calculated using the weighted average formula to obtain the imputed data point. The specific formula is:

[0073] ;

[0074] In the formula, is the imputation value of the k-th missing data point, is the k-th nearest neighbor data point of the missing data point.

[0075] Data consistency adjustment and enhancement processing is performed on the imputation value of the missing data point. The specific formula is:

[0076] ;

[0077] In the formula, is a non-linear adjustment exponent used to control the intensity and direction of the adjustment to adapt to different data distribution characteristics; , are the variance and mean of the overall current environmental data; is the k-th data point after data consistency adjustment and enhancement processing; is the mean of the nearest neighbor data points, which is used to reflect the central tendency of the local data; is the variance of the nearest neighbor data points, which is used to measure the degree of dispersion of the local data point values.

[0078] As a preferred implementation manner of this embodiment, the data feature fusion processing based on the multi-source data feature set using the multi-source data fusion algorithm is specifically as follows:

[0079] Normalize the multivariate data feature set to obtain the normalized multivariate data feature set, and form a multi-dimensional data feature set from the normalized multivariate data feature sets of samples .

[0080] To expand the original feature space and enable the feature data to show richer relationships in the non-linear space, a combination of polynomial kernel and radial basis kernel is used to perform non-linear expansion on each feature in the multi-dimensional data feature set. The specific calculation formula is:

[0081] ;

[0082] In the formula, is the feature data of the th sample in the multi-dimensional data feature set on the th expanded feature dimension; is the th sample in the multi-dimensional data feature set; is the th sample in the multi-dimensional data feature set; is the kernel function; is the contribution coefficient in the polynomial kernel function, used to adjust the contribution of the th feature dimension in the expanded feature; is the order of the polynomial, which determines the polynomial complexity during feature expansion; is the feature dimension; is the parameter in the radial basis kernel function, which controls the sensitivity of the distance between samples in the calculation of the kernel function; is the sample and The square of the Euclidean distance between them represents the similarity of these two samples in the original feature space; finally, it is composed of to form the feature data on the expanded feature dimension .

[0083] On the basis of non-linear feature expansion, a hybrid kernel mapping and optimization process is introduced. By combining multiple kernel functions, the correlation between features is further optimized, making the features more closely related in the new high-dimensional space. The specific operation is to map the feature data on the expanded feature dimension through a multi-kernel combination function to obtain the feature data after hybrid kernel mapping; the kernel weight coefficients in the mapping process are obtained through Lagrangian optimization. In the Lagrangian optimization process, maximizing the correlation between features and avoiding overfitting are considered. The mapping function is specifically:

[0084] ;

[0085] In the formula, is the feature data after mixed kernel mapping, denoted by . is the th kernel function, which can be selected by professionals according to the actual application scenario and will not be elaborated here; is the weight coefficient of the th kernel function, obtained through Lagrangian optimization; is the number of kernel functions.

[0086] After mixed kernel mapping and optimization, the obtained high-dimensional feature space still contains a large amount of redundant information. Multiscale feature selection is introduced. By performing multiscale analysis on the features, the features with the maximum amount of information at each scale are extracted. The specific operation is to perform wavelet transform on the mapped features to decompose the features into different scales to capture the feature information at different scales. The formula for wavelet transform is:

[0087] ;

[0088] where is the wavelet coefficient of the th layer, representing the feature information of the feature data after mixed kernel mapping at the th scale, obtained through wavelet transform, reflecting the feature performance of the data at different scales; is the wavelet function of the th layer, used to decompose the feature signal; is the value of the wavelet function at the th layer and position for calculating the signal components at different scales; is the number of decomposition positions in wavelet transform; is the sample size.

[0089] According to the wavelet coefficients, collaborative feature interaction fusion is performed. By analyzing the interaction between different features, an interaction matrix is constructed to generate the final fusion feature set. The calculation formula for the interaction matrix is:

[0090] ;

[0091] In the formula, is the element of the interaction matrix between the features and after mixed kernel mapping, representing the interaction strength between the features, and the interaction matrix is formed by ; is the feature th The wavelet coefficients of the layer, representing the features after the mixed kernel mapping The performance at different scales; Is the feature The The wavelet coefficients of the layer, representing the features after the mixed kernel mapping The performance at different scales; Adjust the intensity of the interaction between features through periodic changes; Is the number of layers of wavelet coefficients.

[0092] Finally, calculate the covariance matrix of the feature data for the multi-dimensional data feature set after standardization processing, and calculate the eigenvalues and corresponding eigenvectors of the covariance feature matrix. According to the magnitudes of the eigenvalues, select the Eigenvectors corresponding to the top Largest eigenvalues. The selected eigenvectors are the principal features. Based on the selected principal feature matrix

[0093] ;

[0094] Wherein, Is the final fused feature, that is, the accurate environmental parameter data; Is through The constructed interaction matrix, Is the matrix composed of the selected principal features.

[0095] As a preferred implementation manner of this embodiment, the accurate current environmental data is intelligently analyzed using the SVC algorithm, and the specific current analysis result is:

[0096] First, perform standardization processing on the accurate current environmental data, that is, the fused data feature set, to obtain the standardized data feature set.

[0097] Use the existing SVC algorithm for classification to obtain the analysis result based on the fused data feature set. The analysis result represents the current omnipotent laboratory environmental state using environmental quantization level parameters. In machine learning, SVC (Support Vector Classification) is an application form of the support vector machine (SVM, Support Vector Machine) for classification tasks. The SVC algorithm realizes the accurate division of different categories by finding the optimal separation surface between data points.

[0098] The working principle of the SVC algorithm is as follows: First, using the selected kernel function and configured hyperparameters, the training dataset is used to train the SVC model. This typically involves optimizing an objective function that attempts to find a hyperplane that maximizes the margin between different classes, resulting in a trained SVC model. The standardized data feature set is input into the trained SVC model, and the model will output a class label for each input sample, and these labels represent the environmental state classes predicted for each sample.

[0099] As a preferred implementation manner of this embodiment, the optimization process for historical environmental data is specifically as follows: Perform data processing on historical environmental data including filling missing values, removing noise, correcting outliers, and data normalization. After data processing, use the dynamic periodic capture algorithm to reconstruct and optimize the historical environmental data, and use feature engineering to construct a historical data feature set representing the historical environmental data.

[0100] As a preferred implementation manner of this embodiment, in order to more accurately capture the non-linear trends and periodic changes in historical environmental data, the specific operation of using the dynamic periodic capture algorithm to reconstruct and optimize the historical environmental data after data processing is as follows:

[0101] Perform Fourier transform on the historical current environmental data Decompose it at different time scales into components of different frequencies to reveal the multi-level periodic characteristics in the data. The specific calculation formula is:

[0102] ;

[0103] In the formula, is the amplitude of the th frequency component; is the frequency of the th frequency component; is time; is the phase of the th frequency component; is the total number of frequency components of the decomposition.

[0104] Evaluate the importance and periodic intensity of each frequency component in the historical current environmental data sequence to identify the components that contribute the most to the changes in the historical current environmental data. By calculating the autocorrelation of each frequency component and combining its amplitude to determine the periodicity score of the frequency component, a frequency component with high autocorrelation shows strong periodicity. Specifically:

[0105] ;

[0106] In the formula, is the The periodicity score of each frequency component, is the autocorrelation function of the th frequency component at a delay of where the delay is the time interval used in the autocorrelation calculation, representing the time difference considered in the autocorrelation calculation for analyzing the self-similarity of the signal after this time difference, and represents the th delay; is the number of delay times; is the amplitude of the th frequency component; , and

[0107] is the total number of decomposed frequency components. The weights are dynamically assigned using the Softmax function according to the periodicity scores of each frequency component to emphasize the periodic components that contribute the most to the changes in the time series. The specific calculation formula is:

[0108] ;

[0109] where in the formula, is the weight of the th frequency component, ensuring that the sum of all weights is 1.

[0110] The weighted frequency components are used to reconstruct the time series to better capture the periodic characteristics and trends of the data. The decomposed and weight-adjusted frequency components are recombined to form the optimized historical current environmental data. The specific formula is:

[0111] ;

[0112] where in the formula, is the optimized historical current environmental data; is the weight of the th frequency component; is the frequency of the th frequency component; is time; is the phase of the th frequency component.

[0113] As a preferred implementation manner of this embodiment, the predictive analysis of the optimized historical environmental data and accurate current environmental data by using the all-round laboratory environment prediction model constructed based on the LSTM network is specifically as follows: The feature set is divided into a training set and a test set. The LSTM network is used to design a model to obtain the all-round laboratory environment prediction model. The model is trained by using the training set and tested by using the test set to obtain the final all-round laboratory environment prediction model. Then, the accurate current environmental data is used as the input, and the trained model is used to predict the future environmental parameters to obtain the predictive analysis result.

[0114] As a preferred implementation manner of this embodiment, based on the current analysis result and the predictive analysis result, the test environment state of the all-round laboratory is evaluated to realize the environmental monitoring of the all-round laboratory, specifically as follows:

[0115] For the current analysis result, according to the statistical analysis method, the analysis result is deeply analyzed to obtain the environmental quantification level of the current all-round laboratory. The quantification level annotation can be: The original current analysis result will be converted into a specific quantification level. For example, the environmental state is labeled as "suitable", "warning", or "unsuitable". For the predictive analysis result, according to the time series analysis technique, the predictive analysis result is used to analyze the change trend of the future environmental parameters and identify the rising and falling fluctuation ranges of the future trend. For example, the predictive analysis result will be detailedly analyzed into specific future trend changes, such as the expected increase or decrease of the temperature within the next week. Specifically, the statistical analysis methods that can be adopted include descriptive statistical analysis, box plot analysis, skewness and kurtosis analysis; the time series analysis techniques that can be adopted include moving average and exponential smoothing, autoregressive moving average, and Fourier transform. Compared with the classification output directly obtained from the SVC algorithm, the above statistical analysis methods can provide more levels and more detailed information. The SVC algorithm provides the direct classification result based on the dividing line learned by the model, mainly manifested as discrete class labels. While the statistical analysis and time series analysis provide deeper data insights, including:

[0116] Deeper data understanding: For example, through statistical analysis methods such as descriptive statistics and box plots, the distribution of the data and potential outliers can be understood in detail, which may affect the setting of the environmental parameter quantification level.

[0117] More accurate trend prediction: The time series analysis can not only predict the future trend, but also evaluate the stability and possible fluctuation range of the trend, providing more reliable support for future environmental management.

[0118] Furthermore, the comprehensive analysis results can be applied to the laboratory environmental monitoring system using real-time data monitoring technology to update the environmental status display in real-time: The current environmental status and predicted trends are displayed on the monitoring dashboard in real-time using Internet of Things (IoT) technology and real-time data visualization tools (such as Grafana).

[0119] Furthermore, using the decision tree analysis method, operation suggestions are provided based on the obtained environmental quantification level parameters of the current totipotency laboratory and the change trends of future environmental parameters to achieve environmental monitoring of the totipotency laboratory. Specifically, the environmental quantification level parameters of the current totipotency laboratory and the change trends of future environmental parameters are used as inputs, and specific operation suggestions for the current environmental parameters and predicted future trends are output through the decision tree model. For example, if it is predicted that the environmental temperature will rise, the decision tree output will suggest lowering the air conditioner temperature setting.

[0120] As a preferred implementation manner of this embodiment, the accuracy of the environmental monitoring strategy and prediction model can be evaluated regularly according to the performance evaluation criteria formulated by the expert experience method. By comparing the prediction results with the actual observed values, a performance report is obtained; according to the machine learning feedback adjustment mechanism, the SVC and the totipotency laboratory environmental prediction model are adjusted and retrained according to the performance monitoring results to improve the prediction accuracy and classification efficiency.

[0121] As a preferred implementation manner of this embodiment, the data acquisition frequency, input features of the prediction model, etc. can be adjusted according to the model optimization results and laboratory requirements to optimize the environmental monitoring strategy.

[0122] As a preferred implementation manner of this embodiment, specific measures such as air quality improvement and temperature and humidity control can be implemented according to the provided operation suggestions in accordance with the environmental management protocol formulated by the experience method to ensure that the laboratory environment reaches the optimal state. At the same time, according to the experimental planning guiding principles formulated by the experience method, using the environmental trend prediction information, the experimental plan and schedule are adjusted to avoid performing sensitive experiments under unsuitable environmental conditions.

[0123] Embodiment 2

[0124] Correspondingly, this embodiment provides an intelligent environmental monitoring system. The system uses the intelligent environmental monitoring method described in any embodiment of the present invention for environmental monitoring, including a current environment analysis module, a historical environment analysis module, and an environmental monitoring evaluation module.

[0125] The current environment analysis module is used to obtain the current environmental data of the totipotency laboratory, perform intelligent data processing on the current environmental data to obtain accurate current environmental data, and perform intelligent analysis on the accurate current environmental data using the SVC algorithm to obtain the current analysis result, and the current analysis result is the environmental quantification level parameter;

[0126] A historical environment analysis module, which is used to obtain the historical environment data of the totipotency laboratory, optimize the historical environment data, and perform predictive analysis on the optimized historical environment data and accurate current environment data by using a totipotency laboratory environment prediction model constructed based on the LSTM network, so as to obtain a predictive analysis result, and the predictive analysis result is the change trend of future environmental parameters.

[0127] An environmental monitoring and evaluation module, which is used to evaluate the test environment status of the totipotency laboratory based on the current analysis result and the predictive analysis result, so as to realize the environmental monitoring of the totipotency laboratory.

[0128] Embodiment III

[0129] This embodiment provides an electronic device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the intelligent environment monitoring method described in any embodiment of the present invention is implemented.

[0130] Embodiment IV

[0131] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the intelligent environment monitoring method described in any embodiment of the present invention is implemented.

[0132] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent the situation where A exists alone, A and B exist simultaneously, and B exists alone. Where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after. "At least one of the following" and its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0133] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be implemented by a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0134] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0135] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, 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. This 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 application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (hereinafter referred to as ROM), random access memories (hereinafter referred to as RAM), magnetic disks, or optical discs.

[0136] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An intelligent environment monitoring method, characterized in that: The following steps are involved: Acquire the current environmental data of the universal laboratory, perform intelligent data processing on the current environmental data to obtain accurate current environmental data, use the SVC algorithm to perform intelligent analysis on the accurate current environmental data to obtain the current analysis result, which is an environmental quantitative level parameter used to represent the current test environment status of the universal laboratory; The intelligent data processing specifically includes: preprocessing the current environment data including data completion and fusion algorithm, feature engineering extraction, and data feature fusion processing based on multivariate data fusion algorithm; Obtain the historical environmental data of the universal laboratory, optimize the historical environmental data, and use the universal laboratory environmental prediction model built based on the LSTM network to perform predictive analysis on the optimized historical environmental data and the accurate current environmental data to obtain the predictive analysis results, which are the changing trends of future environmental parameters; The optimization processing of the historical environmental data includes: performing data processing on the historical environmental data, and reconstructing and optimizing the historical environmental data using a dynamic periodic capture algorithm after the data processing; the reconstruction optimization is specifically: Fourier transform is used to decompose historical and current environmental data at different time scales into components of different frequencies; The periodicity score of each frequency component is determined by calculating the autocorrelation of the frequency component and combining it with its amplitude; Dynamically assign weights using the Softmax function based on the periodicity score of each frequency component; Reconstruct the time series using weighted frequency components, and recombine the decomposed and weighted frequency components to form optimized historical current environmental data; Based on the current analysis results and predictive analysis results, the test environment status of the universal laboratory is evaluated to achieve environmental monitoring of the universal laboratory.

2. The intelligent environment monitoring method according to claim 1, characterized in that: The specific method of intelligent data processing of current environmental data is as follows: preprocessing the current environmental data, extracting the features of the current environmental data from the preprocessed current environmental data by feature engineering, obtaining a multivariate data feature set, and performing data feature fusion processing by using a multivariate data fusion algorithm based on the multivariate data feature set.

3. The intelligent environment monitoring method according to claim 2, characterized in that: Preprocessing includes noise removal, outlier correction, data normalization and data completion fusion algorithm. The data completion fusion algorithm is used to complete the current environment data as follows: For each missing data point, find its nearest neighbor data point in the current environment data set by calculating the Euclidean distance, and select the nearest neighbor data points and the nearest neighbor Assign weights to data points , weight The specific calculation formula is: ； In the formula, It is The local gradient rate of the nearest neighbor data point relative to the missing data point; is a regulating factor to balance the effects of distance and local gradient rate; It is The distance between the nearest neighbor data point and the missing data point; is the parameter that adjusts the effect of distance; ; Use the weighted average formula to calculate the interpolation value of the missing data point to obtain the interpolated data point. The specific formula is: ； In the formula, It is The interpolated values ​​for missing data points, is the missing data point nearest neighbor data points; The interpolation values ​​of missing data points are adjusted and enhanced for data consistency. The specific formula is: ； In the formula, It is a nonlinear adjustment index, which is used to control the intensity and direction of adjustment to adapt to different data distribution characteristics; , is the variance and mean of the overall current environment data; It is the first data points; is the nearest neighbor The mean of the data points; is the nearest neighbor The variance of the data points.

4. The intelligent environment monitoring method according to claim 2, characterized in that: The data feature fusion processing based on the multivariate data feature set using the multivariate data fusion algorithm is as follows: The multivariate data feature set is standardized to obtain the standardized multivariate data feature set, and then The multivariate data feature set after the standardized processing of samples constitutes the multidimensional data feature set ; The polynomial kernel and radial basis kernel are combined to perform nonlinear expansion on each feature in the multidimensional data feature set. The specific calculation formula is: ； In the formula, It is the multidimensional data feature set The sample in Feature data on the extended feature dimension; It is the multidimensional data feature set samples; It is the multidimensional data feature set samples; is the kernel function; is the contribution coefficient in the polynomial kernel function, which is used to adjust the The contribution of each feature dimension in the extended feature; is the degree of the polynomial; is the characteristic dimension; is the parameter in the radial basis kernel function; It is a sample and The square of the Euclidean distance between Constitute feature data on the extended feature dimension ; The feature data on the extended feature dimension is combined through the multi-core function Mapping is performed to obtain the feature data after hybrid kernel mapping. The specific mapping function is: ； In the formula, is the feature data after mixed kernel mapping, express; It is Kernel function; It is The weight coefficients of the kernel functions are obtained through Lagrangian optimization; is the number of kernel functions; The mapped feature data is subjected to wavelet transformation, and collaborative feature interaction fusion is performed according to the wavelet coefficients. By analyzing the interaction between different features, an interaction matrix is ​​constructed to generate the final fusion feature set. The calculation formula of the interaction matrix is: ； In the formula, is the feature after mixed kernel mapping and The interaction matrix elements between represent the interaction strength between features; It is a feature No. The wavelet coefficients of the layer; It is a feature No. The wavelet coefficients of the layer; The strength of the interaction between features is adjusted through periodic changes; is the number of wavelet coefficient layers; Finally, the covariance matrix of the feature data is calculated for the standardized multidimensional data feature set, and the eigenvalues ​​and corresponding eigenvectors of the covariance feature matrix are calculated. According to the size of the eigenvalue, the previous The eigenvector corresponding to the largest eigenvalue is the selected eigenvector as the main feature. Based on the selected main feature matrix , and get the final fusion feature representation: ； in, It is the final fusion feature, i.e., accurate environmental parameter data; is through The interaction matrix is ​​composed of is the matrix consisting of the selected main features.

5. The intelligent environment monitoring method according to claim 1, characterized in that: The specific optimization processing of historical environmental data includes: performing data processing on historical environmental data including filling missing values, removing noise, correcting outliers, and normalizing data; after data processing, reconstructing and optimizing the historical environmental data using a dynamic periodic capture algorithm; and constructing a historical data feature set representing the historical environmental data using feature engineering for the optimized historical environmental data.

6. The intelligent environment monitoring method according to claim 5, characterized in that: The periodicity score of each frequency component is determined by calculating the autocorrelation of the frequency component and combining it with its amplitude, specifically: ； In the formula, It is The periodicity score of the frequency components, It is The frequency components are delayed The autocorrelation function of the time delay is the time interval used in the autocorrelation calculation, indicating the delay; is the number of delays; It is The amplitude of a frequency component; , is the total number of decomposed frequency components; The time series is reconstructed using weighted frequency components, and the decomposed and weighted frequency components are recombined to form optimized historical current environmental data. The specific formula is: ； In the formula, It is the optimized historical current environment data; It is The weight of the frequency components; It is The frequency of the frequency component; It’s time; It is The phase of a frequency component.

7. The intelligent environment monitoring method according to claim 1, characterized in that: Based on the current analysis results and predictive analysis results, the test environment status of the universal laboratory is evaluated to achieve environmental monitoring of the universal laboratory as follows: According to the current analysis results, the current environmental quantitative level of the universal laboratory is obtained according to the statistical analysis method; according to the predictive analysis results, the change trend of future environmental parameters is obtained by using the predictive analysis results according to the time series analysis technology, and the fluctuation range of future trends is identified; Using the decision tree analysis method, operational suggestions are provided based on the current environmental quantitative level of the universal laboratory and the changing trends of future environmental parameters, thereby realizing environmental monitoring of the universal laboratory.

8. An intelligent environment monitoring system, characterized in that: The system adopts the intelligent environmental monitoring method according to any one of claims 1 to 7 to perform environmental monitoring, including a current environmental analysis module, a historical environmental analysis module and an environmental monitoring and evaluation module; The current environment analysis module is used to obtain the current environment data of the universal laboratory, perform intelligent data processing on the current environment data to obtain accurate current environment data, and use the SVC algorithm to perform intelligent analysis on the accurate current environment data to obtain the current analysis result, which is the environmental quantitative level parameter; The historical environment analysis module is used to obtain the historical environment data of the universal laboratory, optimize the historical environment data, and use the universal laboratory environment prediction model built based on the LSTM network to perform predictive analysis on the optimized historical environment data and the accurate current environment data to obtain the predictive analysis results. The predictive analysis results are the changing trends of future environmental parameters. The environmental monitoring and evaluation module is used to evaluate the test environment status of the universal laboratory based on the current analysis results and predictive analysis results, so as to realize the environmental monitoring of the universal laboratory.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the intelligent environment monitoring method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the intelligent environment monitoring method according to any one of claims 1 to 7 is implemented.

Images