Intelligent adjusting and optimizing system for oyster mushroom growth environment in combination with spectral sensitivity

Abstract

The invention discloses an intelligent oyster mushroom growth environment adjusting and optimizing system combined with spectral sensitivity. The system comprises a multi-dimensional data acquisition layer, a spectrum analysis layer and a spectrum analysis layer, wherein the multi-dimensional data acquisition layer is used for acquiring environmental parameters and multi-period spectrum data in the oyster mushroom growth process; the intelligent preprocessing layer is used for preprocessing the collected spectral data and standardizing environmental parameters; the feature engineering layer is used for extracting key spectral features from the preprocessed spectral data and fusing the key spectral features with the standardized environmental parameters to construct a multi-dimensional feature matrix; the model prediction layer is used for predicting the protein content of the oyster mushrooms by utilizing an integrated machine learning model based on the multi-dimensional feature matrix; and the intelligent regulation and control layer is used for dynamically adjusting the environment parameters and the illumination conditions of oyster mushroom growth based on the prediction result, and the protein content dynamic prediction and the growth environment adaptive adjustment and optimization based on the fusion analysis of the oyster mushroom growth real-time spectral characteristics and the environment parameters are realized. The accuracy, the automation level and the production efficiency of oyster mushroom quality cultivation are fundamentally improved.

Description

Technical Field

[0001] This invention relates to the field of oyster mushroom cultivation technology, and in particular to an intelligent optimization system for the oyster mushroom growth environment that incorporates spectral sensitivity. Background Technology

[0002] Oyster mushrooms, as an important edible fungus, occupy a significant position in the agricultural industry. By integrating spectral sensitivity analysis technology, the physiological response patterns of oyster mushrooms to specific spectral bands at different growth stages can be accurately identified. Combined with an IoT environmental monitoring module to collect multi-dimensional growth parameters such as temperature, humidity, and carbon dioxide concentration in real time, and using artificial intelligence algorithms to construct a dynamic optimization model, intelligent linkage and regulation of spectral parameters and environmental factors can be achieved. This not only significantly improves the biotransformation efficiency and product quality of oyster mushrooms, reduces the risk of pests and diseases, and reduces dependence on chemical agents, but also optimizes energy utilization efficiency through precise spectral regulation, promoting the transformation of oyster mushroom cultivation towards a green and sustainable modern agricultural model. Its role is further reflected in the formation of a complete intelligent regulation link from environmental perception to decision optimization and execution adjustment through a data-driven closed-loop feedback mechanism, ultimately achieving a dynamic balance of the oyster mushroom growth environment and maximizing growth benefits. This provides key technical support for the intelligent and precise upgrading of the edible fungus industry, and at the same time meets the urgent needs of current agricultural technology development for efficient resource utilization, environmentally friendly production, and improved product quality.

[0003] Traditional oyster mushroom cultivation environment control mainly relies on human experience or static control of a single environmental factor, lacking the ability to perceive and regulate the internal quality formation mechanism of oyster mushrooms in real time and quantitatively. This results in unstable protein accumulation efficiency and low precision in quality control. Therefore, a smart optimization system for oyster mushroom growth environment that combines spectral sensitivity is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent optimization system for the growth environment of oyster mushrooms that incorporates spectral sensitivity.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A smart optimization system for the growth environment of oyster mushrooms, incorporating spectral sensitivity, includes: Multidimensional data acquisition layer: Collects environmental parameters and multi-time period spectral data during the growth process of oyster mushrooms; Intelligent preprocessing layer: preprocesses the collected spectral data and standardizes environmental parameters; Feature engineering layer: Extracts key spectral features from preprocessed spectral data and fuses them with standardized environmental parameters to construct a multidimensional feature matrix; Model prediction layer: Based on the multidimensional feature matrix, the protein content of oyster mushrooms is predicted using an integrated machine learning model; Intelligent control layer: Based on the prediction results, dynamically adjust the environmental parameters and light conditions for oyster mushroom growth; Feedback Validation Layer: Evaluates the model's predictive performance and provides visual monitoring and anomaly alerts.

[0006] The above technical solution further includes: Furthermore, the multidimensional data acquisition layer includes an environmental parameter monitoring module and a spectral data acquisition module; The environmental parameter monitoring module includes temperature and humidity sensors and a substrate composition analyzer arranged in the facility greenhouse. It is used to continuously monitor environmental data covering a temperature gradient of 15°C to 25°C and a humidity gradient of 66%RH to 90%RH, and to analyze the substrate composition composed of sawdust, corn cobs and soybean straw in different proportions. The spectral data acquisition module uses a mid-infrared spectrometer and is equipped with a diffuse reflection fiber optic probe to automatically acquire mid-infrared spectral data of oyster mushroom fruiting bodies every 2 hours under non-destructive conditions, while recording the corresponding growth environment parameters.

[0007] Furthermore, the intelligent preprocessing layer includes a spectral preprocessing module and an environmental data standardization module; The spectral preprocessing module integrates a smoothing filter unit, a first-order derivative processing unit, and a multivariate scattering correction unit. The smoothing filter unit uses the Savitzky-Golay algorithm to smooth the original spectral data. The first-order derivative processing unit performs a first-order derivative transformation on the smoothed spectral data to eliminate spectral baseline drift. The multivariate scattering correction unit corrects the spectral scattering effect caused by uneven particle distribution in the sample.

[0008] Furthermore, the environmental data standardization module employs the Z-score standardization method to process the raw environmental data from the temperature and humidity sensors and matrix composition analyzer, eliminating the influence of differences in dimensions and numerical ranges of different environmental parameters.

[0009] Furthermore, the feature engineering layer includes a spectral feature extraction module and an environmental feature fusion module; The spectral feature extraction module includes a principal component analysis unit, which is used to perform dimensionality reduction on the preprocessed spectral data, extract key spectral feature variables, reduce data redundancy, and improve the efficiency and accuracy of subsequent modeling.

[0010] Furthermore, the environmental feature fusion module is used to construct a three-dimensional feature matrix containing information on temperature, humidity and substrate ratio. It also uses the Pearson correlation coefficient to screen out highly correlated environmental features with an absolute correlation value greater than the threshold (0.7) with the protein content of oyster mushroom. The screened environmental features are then fused with the extracted key spectral features to form a multi-dimensional feature matrix for model input.

[0011] Furthermore, the integrated machine learning models are constructed using machine learning-based fusion algorithms, including the PCA-SVR model combining principal component analysis and support vector regression, the PCA-RF model combining principal component analysis and random forest, and the PCA-PLSR model combining principal component analysis and partial least squares regression. The integrated machine learning model grid search method optimizes the hyperparameters of each model. At the same time, the Stacking ensemble learning method is used, with the support vector regression as the meta-model and the random forest and partial least squares regression as the base models to fuse the prediction results, so as to improve the robustness and generalization ability of the final protein content prediction model.

[0012] Furthermore, the intelligent control layer includes a dynamic optimization module, a spectral control module, and an adaptive learning module; The dynamic optimization module adopts a proportional-integral-derivative control algorithm. Based on the protein content prediction value output by the model prediction layer, it performs real-time feedback adjustment of greenhouse environmental parameters to achieve precise control of temperature with an accuracy of ±0.5℃ and humidity with an accuracy of ±2%RH. The spectral control module includes an LED spectral array with an adjustable ratio of red and blue light, which optimizes the protein synthesis physiological pathway of oyster mushroom fruiting bodies by dynamically adjusting the irradiation ratio of 660nm red light and 450nm blue light. The adaptive learning module updates the machine learning model parameters in the model prediction layer with the latest collected data every 72 hours to adapt to the changes in the spectral sensitivity of oyster mushrooms at different growth stages.

[0013] Furthermore, the feedback verification layer includes an online verification system and a visual dashboard; The online verification system uses five-fold cross-validation and leave-one-out cross-validation to continuously evaluate the predictive performance of the model prediction layer. The qualified threshold for model prediction is set as a coefficient of determination of not less than 0.92 and a root mean square error of not more than 0.8 grams per 100 grams. The visualization dashboard is used to display spectral feature heatmaps, historical trends of environmental parameters, and comparison curves between predicted protein content and measured values ​​by the Kjeldahl method in real time. It also integrates automatic anomaly detection and alarm, automatically triggering a warning when prediction deviations or environmental parameters exceed the set range.

[0014] The present invention has the following beneficial effects: In this invention, mid-infrared spectroscopy technology is systematically combined with multiple environmental parameters such as temperature, humidity, and substrate for collection, feature fusion, and modeling. This reveals the intrinsic relationship between substrate composition-temperature and humidity coupling effect and oyster mushroom protein accumulation. Based on this, prediction and regulation are carried out, realizing dynamic prediction of protein content and adaptive optimization of the growth environment based on the fusion analysis of real-time spectral characteristics of oyster mushroom growth and environmental parameters. This fundamentally improves the accuracy, automation level, and production efficiency of oyster mushroom quality cultivation. Attached Figure Description

[0015] Figure 1 This is a system block diagram of an intelligent optimization system for the growth environment of oyster mushrooms that incorporates spectral sensitivity, as proposed in this invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 As shown, this invention is an intelligent optimization system for the growth environment of oyster mushrooms that incorporates spectral sensitivity, comprising: Multidimensional data acquisition layer: Collects environmental parameters and multi-time period spectral data during the growth process of oyster mushrooms; Intelligent preprocessing layer: preprocesses the collected spectral data and standardizes environmental parameters; Feature engineering layer: Extracts key spectral features from preprocessed spectral data and fuses them with standardized environmental parameters to construct a multidimensional feature matrix; Model prediction layer: Based on the multidimensional feature matrix, the protein content of oyster mushrooms is predicted using an integrated machine learning model; Intelligent control layer: Based on the prediction results, dynamically adjust the environmental parameters and light conditions for oyster mushroom growth; Feedback Validation Layer: Evaluates the model's predictive performance and provides visual monitoring and anomaly alerts.

[0018] In one embodiment, the multidimensional data acquisition layer includes an environmental parameter monitoring module and a spectral data acquisition module; The environmental parameter monitoring module includes temperature and humidity sensors and a substrate composition analyzer arranged in the facility greenhouse. It is used to continuously monitor environmental data covering a temperature gradient of 15°C to 25°C and a humidity gradient of 66%RH to 90%RH, and to analyze the substrate composition composed of sawdust, corn cobs and soybean straw in different proportions. The spectral data acquisition module uses a mid-infrared spectrometer and is equipped with a diffuse reflection fiber optic probe to automatically acquire mid-infrared spectral data of oyster mushroom fruiting bodies every 2 hours under non-destructive conditions, while recording the corresponding growth environment parameters.

[0019] In one embodiment, the intelligent preprocessing layer includes a spectral preprocessing module and an environmental data normalization module; The spectral preprocessing module integrates a smoothing filter unit, a first-order derivative processing unit, and a multivariate scattering correction unit. The smoothing filter unit uses the Savitzky-Golay algorithm to smooth the original spectral data. The first-order derivative processing unit performs a first-order derivative transformation on the smoothed spectral data to eliminate spectral baseline drift. The multivariate scattering correction unit corrects the spectral scattering effect caused by uneven particle distribution in the sample.

[0020] It should be noted that the specific analysis process for spectral smoothing of the original spectral data is as follows: The integrated Savitzky-Golay convolutional smoothing algorithm is automatically invoked. Based on preset rules or spectral noise levels, the window width (e.g., 5, 7, or 9 points) and the polynomial fitting order (usually 2nd or 3rd order) are automatically selected. Then, starting from the spectral origin, the center of the sliding window is sequentially aligned with each data point i. For each window position, an nth-order polynomial is fitted using the least squares method with m data points within the window. The independent variable (wavenumber position) corresponding to the window center point i is substituted into the local polynomial obtained by the least squares method, and the calculated function value is used as the smoothed new spectral value for that point. Where, k = (m-1) / 2, coefficient To filter out high-frequency random noise while preserving the effective shape features of the spectrum, fixed weight coefficients (Savitzky-Golay convolution coefficients) are derived in advance based on the window width m and the polynomial order n using the least squares principle. This process traverses all spectral bands and finally outputs smoothed spectral data.

[0021] It should be noted that the specific analysis process for performing the first derivative transformation is as follows: The smoothed spectral data sequence is numerically differentiated using the first-order difference method. A suitable difference interval (such as the spectrometer sampling interval) is selected, and the rate of change of the spectral intensity value at each wavelength point relative to the wavelength is calculated; this is the first-order derivative value, expressed as follows: ,in, and These are the spectral intensity values ​​at g intervals before and after point i, respectively. is the wavenumber interval (in cm⁻¹), and g is the difference interval (usually 1 or 2). This transformation can significantly amplify subtle spectral features and effectively suppress slowly changing baseline interference caused by sample background or measurement conditions. For points near the start and end of the spectrum where complete central difference cannot be performed, boundary processing methods such as forward difference or backward difference can be used for calculation. After processing, the first derivative spectral data is output. The spectral characteristic peaks of this data correspond to the inflection points of the original spectrum, thereby enhancing the spectral resolution related to the target component (such as protein).

[0022] It should be noted that the implementation steps of the multivariate scattering correction unit are as follows: The average spectrum of all samples to be processed (such as samples from the same batch or the same growth stage) is calculated as the ideal reference spectrum. Then, a univariate linear regression analysis is performed on the original spectrum of each sample (the spectrum after smoothing and derivative processing) and the average reference spectrum to obtain the unique correction coefficient (i.e., offset and multiplicative factor) for each sample. The obtained correction coefficient is used to perform an inverse operation on the spectral data of each sample, that is, to subtract the offset and divide by the multiplicative factor, so that the spectra of all samples are aligned on the intensity axis, thereby eliminating the non-chemical differences in the spectrum caused by physical scattering, retaining the spectral information related to the content of chemical components, and outputting the corrected final spectrum for subsequent feature extraction.

[0023] In one embodiment, the environmental data standardization module uses the Z-score standardization method to process the raw environmental data from the temperature and humidity sensor and the matrix composition analyzer, eliminating the influence of differences in the dimensions and numerical ranges of different environmental parameters.

[0024] In one embodiment, the feature engineering layer includes a spectral feature extraction module and an environmental feature fusion module; The spectral feature extraction module includes a principal component analysis unit, which is used to perform dimensionality reduction on the preprocessed spectral data, extract key spectral feature variables, reduce data redundancy, and improve the efficiency and accuracy of subsequent modeling.

[0025] It should be noted that the specific analysis process for dimensionality reduction is as follows: Set the preprocessed spectral data into M samples { Each sample has N-dimensional features. Each feature Each of them has its own eigenvalues; First, all features are decentered, i.e., mean-removed, by calculating the average of each feature. Then, for all samples, each feature is subtracted from its own mean, where the respective means are respectively... ; Calculate the covariance matrix after decentralization The features are shown on the diagonal. and The variance is on the off-diagonal side, and the covariance is on the other side. The calculation formula is: Thus, the covariance matrix C of the M samples under these N-dimensional features is obtained; After obtaining the covariance matrix, according to the characteristic equation Find its eigenvalues ​​and their corresponding eigenvectors, where λ is the eigenvalue and μ is its corresponding eigenvector. Select the top k largest eigenvalues ​​and their corresponding eigenvectors for projection. Projection is the process of dimensionality reduction, which reduces the original features from high dimension to low dimension. After dimensionality reduction, a large amount of redundant information is removed, while at least 85% of the original information is retained.

[0026] In one embodiment, the environmental feature fusion module is used to construct a three-dimensional feature matrix containing information on temperature, humidity and substrate ratio, and to screen out highly correlated environmental features whose absolute correlation with the protein content of oyster mushroom is greater than a threshold (0.7) by calculating the Pearson correlation coefficient. The screened environmental features are then fused with the extracted key spectral features to form a multi-dimensional feature matrix for model input.

[0027] It should be noted that the specific analysis process for fusing the screened environmental features with the extracted key spectral features is as follows: Standardized environmental data preparation: Obtain the Z-score standardized environmental parameter dataset from the environmental data standardization module of the intelligent preprocessing layer. This dataset contains continuous temperature and humidity monitoring values ​​(temperature gradient covering 16-24℃, humidity gradient covering 86-97%RH) and mass or volume ratio data of three matrices, sawdust, corn cob, and soybean straw, obtained by quantitative analysis by a matrix composition analyzer. This ensures that all environmental parameters have eliminated dimensional differences and are on the same numerical scale, providing standardized input for subsequent matrix construction and correlation analysis. Three-dimensional feature matrix construction: Standardized temperature data, humidity data, and substrate ratio data are used as three independent feature vectors. They are aligned and matched with the corresponding oyster mushroom growth samples according to the timestamp. A structured temperature-humidity-substrate three-dimensional feature matrix is ​​constructed with samples as rows and the three types of environmental parameters as columns. Each sample row vector contains specific temperature values, specific humidity values, and specific substrate component ratio information at a specific time or stage during its growth process. Pearson correlation coefficient calculation and high-correlation feature screening: Using the measured protein content of oyster mushroom samples obtained by the Kjeldahl nitrogen determination method as the target variable, the Pearson correlation coefficient between each column of the three-dimensional feature matrix and the target variable was calculated, and expressed as follows: ,in, The correlation coefficient represents the j-th environmental characteristic. This represents the j-th environmental feature value of the i-th sample. It is the mean of this feature across all samples. It is the protein content of the i-th sample. This represents the mean protein content, where n is the total number of samples. A screening threshold θ = 0.7 is set to retain all samples that meet the criteria. Environmental features with >θ are marked as highly relevant environmental features and extracted from the original three-dimensional matrix to form a filtered subset of key environmental features; Fusion of key spectral features and key environmental features: The screened environmental features and spectral features are Z-score standardized so that the mean of each feature dimension is 0 and the standard deviation is 1. Then, the standardized highly correlated environmental feature vector and the key spectral feature vector are concatenated in the sample dimension to form the final multidimensional feature matrix used for model input.

[0028] In one embodiment, the integrated machine learning model is constructed using a machine learning-based fusion algorithm, including the PCA-SVR model combining principal component analysis and support vector regression, the PCA-RF model combining principal component analysis and random forest, and the PCA-PLSR model combining principal component analysis and partial least squares regression. The integrated machine learning model grid search method optimizes the hyperparameters of each model. At the same time, the Stacking ensemble learning method is used, with the support vector regression as the meta-model and the random forest and partial least squares regression as the base models to fuse the prediction results, so as to improve the robustness and generalization ability of the final protein content prediction model.

[0029] In one embodiment, the intelligent control layer includes a dynamic optimization module, a spectral control module, and an adaptive learning module; The dynamic optimization module adopts a proportional-integral-derivative control algorithm. Based on the protein content prediction value output by the model prediction layer, it performs real-time feedback adjustment of greenhouse environmental parameters to achieve precise control of temperature with an accuracy of ±0.5℃ and humidity with an accuracy of ±2%RH. The spectral control module includes an LED spectral array with an adjustable ratio of red and blue light, which optimizes the protein synthesis physiological pathway of oyster mushroom fruiting bodies by dynamically adjusting the irradiation ratio of 660nm red light and 450nm blue light. The adaptive learning module updates the machine learning model parameters in the model prediction layer with the latest collected data every 72 hours to adapt to the changes in the spectral sensitivity of oyster mushrooms at different growth stages.

[0030] In one embodiment, the feedback verification layer includes an online verification system and a visual dashboard; The online verification system uses five-fold cross-validation and leave-one-out cross-validation to continuously evaluate the predictive performance of the model prediction layer. The qualified threshold for model prediction is set as a coefficient of determination of not less than 0.92 and a root mean square error of not more than 0.8 grams per 100 grams. The visualization dashboard is used to display spectral feature heatmaps, historical trends of environmental parameters, and comparison curves between predicted protein content and measured values ​​by the Kjeldahl method in real time. It also integrates automatic anomaly detection and alarm, automatically triggering a warning when prediction deviations or environmental parameters exceed the set range.

[0031] All data obtained in this invention has been authorized by the user.

[0032] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart optimization system for the growth environment of oyster mushrooms, incorporating spectral sensitivity, characterized in that: include: Multidimensional data acquisition layer: Collects environmental parameters and multi-time period spectral data during the growth process of oyster mushrooms; Intelligent preprocessing layer: preprocesses the collected spectral data and standardizes environmental parameters; Feature engineering layer: Extracts key spectral features from preprocessed spectral data and fuses them with standardized environmental parameters to construct a multidimensional feature matrix; Model prediction layer: Based on the multidimensional feature matrix, the protein content of oyster mushrooms is predicted using an integrated machine learning model; Intelligent control layer: Based on the prediction results, dynamically adjust the environmental parameters and light conditions for oyster mushroom growth; Feedback Validation Layer: Evaluates the model's predictive performance and provides visual monitoring and anomaly alerts.

2. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The multidimensional data acquisition layer includes an environmental parameter monitoring module and a spectral data acquisition module; The environmental parameter monitoring module includes temperature and humidity sensors and a matrix composition analyzer arranged in the facility greenhouse, used to monitor environmental data and analyze the matrix composition with different ratios; The spectral data acquisition module uses a mid-infrared spectrometer and is equipped with a diffuse reflection fiber optic probe to automatically acquire mid-infrared spectral data of oyster mushroom fruiting bodies under non-destructive conditions, while recording the corresponding growth environment parameters.

3. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The intelligent preprocessing layer includes a spectral preprocessing module and an environmental data standardization module; The spectral preprocessing module integrates a smoothing filter unit, a first-order derivative processing unit, and a multivariate scattering correction unit. The smoothing filter unit uses the Savitzky-Golay algorithm to smooth the original spectral data. The first-order derivative processing unit performs a first-order derivative transformation on the smoothed spectral data to eliminate spectral baseline drift. The multivariate scattering correction unit corrects the spectral scattering effect caused by uneven particle distribution in the sample.

4. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 3, characterized in that, The environmental data standardization module uses the Z-score standardization method to process the raw environmental data from the temperature and humidity sensor and the matrix composition analyzer, eliminating the influence of differences in the dimensions and numerical ranges of different environmental parameters.

5. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The feature engineering layer includes a spectral feature extraction module and an environmental feature fusion module; The spectral feature extraction module includes a principal component analysis unit, which is used to perform dimensionality reduction processing on the preprocessed spectral data and extract key spectral feature variables.

6. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 5, characterized in that, The environmental feature fusion module is used to construct a three-dimensional feature matrix containing information on temperature, humidity and substrate ratio. It also uses the Pearson correlation coefficient to screen out highly correlated environmental features with an absolute correlation value greater than the threshold (0.7) with the protein content of oyster mushroom. The screened environmental features are then fused with the extracted key spectral features to form a multi-dimensional feature matrix for model input.

7. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The integrated machine learning model is constructed using a machine learning-based fusion algorithm, including the PCA-SVR model combining principal component analysis and support vector regression, the PCA-RF model combining principal component analysis and random forest, and the PCA-PLSR model combining principal component analysis and partial least squares regression. The integrated machine learning model grid search method optimizes the hyperparameters of each model. At the same time, the Stacking ensemble learning method is used, with the support vector regression as the meta-model and the random forest and partial least squares regression as the base models to fuse the prediction results.

8. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The intelligent control layer includes a dynamic optimization module, a spectral control module, and an adaptive learning module. The dynamic optimization module uses a proportional-integral-derivative control algorithm to adjust the greenhouse environmental parameters in real time based on the protein content prediction value output by the model prediction layer. The spectral control module includes an LED spectral array with an adjustable ratio of red and blue light, which optimizes the protein synthesis physiological pathway of oyster mushroom fruiting bodies by dynamically adjusting the irradiation ratio of red and blue light. The adaptive learning module updates the machine learning model parameters in the model prediction layer using the latest collected data.

9. The intelligent optimization system for the growth environment of oyster mushrooms according to claim 1, characterized in that, The feedback verification layer includes an online verification system and a visual dashboard; The online validation system uses five-fold cross-validation and leave-one-out cross-validation to continuously evaluate the prediction performance of the model prediction layer; The visualization dashboard is used to display spectral feature heatmaps, historical trends of environmental parameters, and comparison curves between predicted protein content and measured values ​​by the Kjeldahl method in real time. It also integrates automatic anomaly detection and alarm, automatically triggering a warning when prediction deviations or environmental parameters exceed the set range.

Images