Edible mushroom factory digital twinning construction method and device based on Internet of Things

Abstract

The invention discloses an edible mushroom factory digital twinning construction method and device based on the Internet of Things. The method comprises the steps that edible mushroom growth environment data, strain gene expression data, equipment operation data, real-time meteorological data and historical production data are collected; according to the edible mushroom growth environment data and the strain gene expression data, a multi-scale digital twinborn model is constructed, and parameters of the multi-scale digital twinborn model are dynamically calibrated; according to the real-time meteorological data and the historical production data, updating the multi-scale digital twinborn model, and utilizing the updated multi-scale digital twinborn model to compensate environmental fluctuation; and integrating equipment operation data, edible mushroom growth environment data and fault early warning information, establishing a fault knowledge graph, identifying a fault propagation path, and positioning a fault point and a fault occurrence reason. According to the invention, the production efficiency is improved, the resource use is optimized, and the production stability is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and more specifically, to a method and apparatus for constructing a digital twin of an edible mushroom factory based on the Internet of Things. Background Technology

[0002] A digital twin system for edible mushroom factories is an intelligent production management system that integrates technologies such as the Internet of Things (IoT), big data, 3D modeling, and simulation algorithms. It aims to achieve real-time monitoring, precise simulation, and predictive optimization of the physical production process by constructing a "digital mirror" of the edible mushroom factory in virtual space. While IoT-based digital twin systems for edible mushroom factories have shown potential in technological integration and production optimization, they still face many challenges in practical applications due to limitations imposed by the characteristics of agricultural production and the maturity of the technology.

[0003] The shortcomings of existing technologies are as follows: First, existing biological models have two major drawbacks in describing the complex metabolic processes of edible fungi: on the one hand, they are difficult to accurately characterize the dynamic relationship between gene expression, environmental factors and growth status; on the other hand, they cannot quickly adapt to external interference such as weather fluctuations, resulting in a disconnect between model predictions and actual production. Second, the efficiency of fault diagnosis is low. When equipment suddenly fails, the virtual system cannot quickly integrate multi-source information such as equipment operation logs, maintenance records, and environmental data, making it difficult to locate the root cause of the fault within seconds, which affects the continuity of production. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a method for constructing a digital twin of an edible fungus factory based on the Internet of Things, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for constructing a digital twin of an edible mushroom factory based on the Internet of Things, comprising:

[0006] Collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data, and historical production data;

[0007] Based on the edible fungi growth environment data and strain gene expression data, a multi-scale digital twin model is constructed, and the parameters of the multi-scale digital twin model are dynamically calibrated; wherein, the multi-scale digital twin model outputs the edible fungi growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information;

[0008] The multi-scale digital twin model is updated based on the real-time meteorological data and historical production data, and the updated multi-scale digital twin model is used to compensate for environmental fluctuations.

[0009] By integrating the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model, a fault knowledge graph is established. The graph neural network is then used to analyze the fault knowledge graph to identify the fault propagation path, locate the fault point, and determine the cause of the fault.

[0010] In a preferred embodiment, the step of constructing a multi-scale digital twin model based on the edible fungus growth environment data and strain gene expression data includes:

[0011] use In principle, outliers in the edible fungi growth environment data are removed, and missing values ​​are filled in by interpolation. The edible fungi growth environment data is then standardized, and an environmental data vector is constructed based on the standardized edible fungi growth environment data.

[0012] Outliers were removed from the gene expression data of the bacterial strain using interquartile range, missing values ​​were filled by interpolation, and the data was normalized. A gene expression data matrix was constructed based on the normalized gene expression data of the bacterial strain.

[0013] The environmental data vector and gene expression data matrix are integrated into a multi-scale framework to construct a multi-scale input data space, which covers different time and spatial scales.

[0014] The multi-scale input data space is input into a preset model structure to construct a multi-scale digital twin model.

[0015] In a preferred embodiment, the multi-scale digital twin model is constructed based on a Bayesian network, and the parameters of the multi-scale digital twin model are dynamically calibrated, including:

[0016] Using the real-time collected equipment operation data and edible fungus growth environment data, the parameters in the Bayesian network are dynamically adjusted, and the conditional probabilities in the Bayesian network are gradually updated.

[0017] The model parameters are calibrated using the device's operating data.

[0018] Verify the accuracy and predictive ability of the model, output the verification results, and adjust the Bayesian network structure or optimize the data preprocessing process based on the verification results;

[0019] The model is continuously optimized by collecting real-time data on the equipment operation, the growth environment of the edible fungi, and the gene expression data of the fungi strains.

[0020] Based on the calibrated multi-scale digital twin model, the system outputs monitoring results of edible fungi growth status, instructions for adjusting culture conditions, and early warning information for equipment malfunctions. The monitoring results of edible fungi growth status reflect the real-time growth status of edible fungi, and the instructions for adjusting culture conditions are used to regulate the environmental parameters of edible fungi culture. The instructions for adjusting culture conditions are output based on changes in environmental data and gene expression data.

[0021] In a preferred embodiment, updating the multi-scale digital twin model based on the real-time meteorological data and historical production data includes:

[0022] The real-time meteorological data is obtained based on the edible fungus growth environment data;

[0023] The historical production data is obtained based on the equipment operation data;

[0024] An environmental fluctuation model is established based on the real-time meteorological data. The environmental fluctuation model is used to capture the nonlinear relationship between meteorological changes and production processes.

[0025] The environmental fluctuation model is initially trained and calibrated using the historical production data and real-time meteorological data, and meteorological-production correlation model data is output.

[0026] The real-time meteorological data, historical production data, and meteorological-production correlation pattern data are input into an online learning algorithm. The parameters of the multi-scale digital twin model are updated through the online learning algorithm. In each round of learning, the weights and parameters of the multi-scale digital twin model are adjusted according to the newly received data so that the weights and parameters adapt to the nonlinear impact of meteorological changes on production.

[0027] Based on the comparison results between the output of the multi-scale digital twin model and the actual production, real-time error feedback adjustments are made.

[0028] In a preferred embodiment, compensating for environmental fluctuations using the updated multi-scale digital twin model includes:

[0029] The updated multi-scale digital twin model is used to predict production changes in the next period, and production parameters are adjusted based on meteorological forecasts.

[0030] Real-time monitoring of environmental changes and production deviations, and adjustments made through compensation strategies of the online learning model; periodic comparison of the deviation between model predictions and actual production results to evaluate the compensation capability of the multi-scale digital twin model for environmental fluctuations;

[0031] If the model prediction error exceeds a preset threshold, the parameters or model structure of the online learning algorithm will be optimized.

[0032] In a preferred embodiment, establishing the fault knowledge graph includes:

[0033] By integrating the equipment operation data, edible fungus growth environment data, and equipment fault early warning information output by the multi-scale digital twin model, the nodes and edges of the fault knowledge graph are defined, where nodes include equipment, fault type, environmental factors, and equipment components, and edges represent the relationships between nodes;

[0034] Integrate historical failure cases and historical equipment operation data, label failure types and failure propagation paths, and construct a failure knowledge graph;

[0035] Design a graph neural network and train it using the collected historical fault cases and historical equipment operation data. During the training process, the relationship between nodes and potential fault propagation paths are learned through message passing and aggregation operations between nodes.

[0036] In a preferred embodiment, analyzing the fault knowledge graph using a graph neural network to identify fault propagation paths includes:

[0037] The trained graph neural network is used to analyze the fault knowledge graph and identify the propagation path of each fault from one device to other devices.

[0038] Based on the propagation path, identify the device nodes related to the fault and trace the source of the fault.

[0039] Identify the interrelationships between various devices and determine which devices are affected or cause the fault to propagate when it occurs.

[0040] In a preferred embodiment, locating the fault point and the cause of the fault includes:

[0041] The system continuously monitors the equipment operation data and the edible fungus growth environment data, inputs the equipment operation data and the edible fungus growth environment data into a graph neural network, predicts whether a fault has occurred, and generates a fault prediction result.

[0042] Based on the fault prediction results, the fault point is located and the root cause of the fault is identified.

[0043] By combining the fault propagation path, the specific equipment and its location where the fault occurred can be identified, and the upstream root cause of the fault can be traced.

[0044] In a preferred embodiment, the edible fungi growth environment data includes temperature and humidity data, light data, gas composition data, and culture medium data;

[0045] The bacterial strain gene expression data includes transcriptome data and gene expression level data;

[0046] The equipment operation data includes basic operating parameters and fault-related data;

[0047] The real-time meteorological data includes temperature, humidity, wind speed, and precipitation.

[0048] In another aspect, the present invention provides a digital twin construction device for an edible mushroom factory based on the Internet of Things, employing the above-described method, comprising:

[0049] The data acquisition module is used to collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data, and historical production data.

[0050] A multi-scale digital twin model construction module is used to construct a multi-scale digital twin model based on the edible fungus growth environment data and fungal gene expression data, and to dynamically calibrate the parameters of the multi-scale digital twin model; wherein, the multi-scale digital twin model outputs the edible fungus growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information;

[0051] An adaptive dynamic model module is used to update the multi-scale digital twin model based on the real-time meteorological data and historical production data, and to use the updated multi-scale digital twin model to compensate for environmental fluctuations.

[0052] The fault diagnosis module integrates the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model to establish a fault knowledge graph. The graph neural network is then used to analyze the fault knowledge graph to identify the fault propagation path, locate the fault point, and determine the cause of the fault.

[0053] The technical effects and advantages of this invention are as follows:

[0054] This invention provides a method for constructing a digital twin for edible mushroom factories based on the Internet of Things (IoT). By building a multi-scale digital twin model and combining it with a dynamic optimization mechanism, it achieves precise control and efficient management of the entire edible mushroom production process. This multi-scale digital twin model is constructed based on edible mushroom growth environment data and strain gene expression data. Parameters are dynamically calibrated using a Bayesian network, and continuously updated using real-time equipment operation data and environmental data. It outputs growth status monitoring results and cultivation condition adjustment instructions, stabilizing production conditions within the optimal range and avoiding resource waste caused by human error. Simultaneously, an online learning algorithm is introduced, combining real-time meteorological data and historical production data to iteratively optimize the parameters of the multi-scale digital twin model, responding in real-time to nonlinear fluctuations in the external environment and significantly enhancing stability under complex meteorological conditions. Furthermore, by integrating equipment operation data, environmental data, and fault warning information output by the model, a fault knowledge graph is constructed. Graph neural networks are used to analyze node relationships, accurately identifying fault propagation paths and locating fault points. Through deep linkage between fault diagnosis and the digital twin model, it achieves second-level fault location and efficient response, and supports intelligent decision-making throughout the entire process through multi-source data fusion. Attached Figure Description

[0055] Figure 1 A flowchart illustrating a method for constructing a digital twin of an edible mushroom factory based on the Internet of Things, according to an embodiment of the present invention;

[0056] Figure 2 A schematic diagram of a digital twin construction device for an edible fungus factory based on the Internet of Things provided in another embodiment of the present invention.

[0057] The reference numerals in the attached figures are as follows:

[0058] 300: Digital Twin Construction Device for Edible Mushroom Factories;

[0059] 310: Data Acquisition Module;

[0060] 320: Multi-scale digital twin model construction module;

[0061] 330: Adaptive Dynamic Model Module;

[0062] 340: Fault diagnosis module. Detailed Implementation

[0063] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0064] Example 1, Figure 1 A schematic diagram of the overall process of constructing a digital twin of an edible fungus factory based on the Internet of Things is given in this invention.

[0065] A method for constructing a digital twin of an edible mushroom factory based on the Internet of Things includes the following steps:

[0066] Step S1: Collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data, and historical production data.

[0067] In one specific implementation, sensor networks and Internet of Things (IoT) technologies are used to monitor key data such as temperature, humidity, carbon dioxide concentration, and light intensity in the edible fungus growth environment in real time, ensuring that environmental parameters remain stable within the range most suitable for edible fungus growth, thereby improving production efficiency and quality.

[0068] Specifically, data on the growth environment of edible fungi include temperature and humidity data, light data, gas composition data, and culture medium data. To meet environmental monitoring needs, temperature and humidity sensors, light sensors, and gas composition sensors are installed, and culture medium data is monitored regularly. Temperature and humidity sensors facilitate the control of the thermal and humid environment for fungi growth. Light and gas composition sensors monitor light and gas composition data, as some edible fungi are sensitive to light; for example, Ganoderma lucidum requires specific intensity and duration of diffused light during fruiting body differentiation. Excessive carbon dioxide concentration inhibits fungi growth, while changes in volatile organic compound (VOC) concentration reflect the health status of the fungi and the presence of contamination. Regular monitoring of culture medium data includes pH value, water content, and nutrient composition (such as carbon and nitrogen source content), as these parameters affect nutrient absorption and growth rate.

[0069] Microbial gene expression data includes transcriptome data and gene expression level data; transcriptome data and gene expression level data are obtained in the early stage of edible fungi cultivation using a gene sequencer and a real-time PCR instrument.

[0070] It should be noted that transcriptome data is obtained through RNA sequencing technology to acquire gene transcription information of edible fungi at different growth stages. This analysis identifies which genes are upregulated or downregulated during key processes such as mycelial growth and fruiting body differentiation, revealing the molecular mechanisms of edible fungi growth and development. Gene expression data is obtained using quantitative real-time PCR technology to quantitatively analyze the expression levels of specific functional genes (such as those related to polysaccharide synthesis and stress resistance) to determine the growth activity and metabolic state of the fungal strain.

[0071] Equipment operation data includes basic operating parameters and fault-related data. Basic operating parameters include operating time, speed, and power, which are read directly through the built-in controller. Fault-related data includes vibration frequency, noise intensity, and historical fault information. Vibration frequency and noise intensity are obtained through vibration and noise sensors, while historical fault information is obtained through equipment logs and databases.

[0072] In one specific implementation, real-time meteorological data, including temperature, humidity, wind speed, and precipitation, is used to analyze the growth environment of edible fungi. Historical production data is also obtained based on equipment operation data.

[0073] Step S2: Based on the edible fungus growth environment data and strain gene expression data, construct a multi-scale digital twin model and dynamically calibrate the parameters of the multi-scale digital twin model; wherein, the multi-scale digital twin model outputs the edible fungus growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information.

[0074] Data on the growth environment of edible fungi include temperature and humidity data, light data, gas composition data, and culture medium data. The principle is to remove outliers from the data, fill in missing values ​​using interpolation, standardize the edible fungi growth environment data, and construct an environmental data vector based on the standardized edible fungi growth environment data.

[0075] The bacterial gene expression data includes transcriptome data and gene expression level data. Outliers were removed using interquartile range, missing values ​​were filled using interpolation, and the bacterial gene expression data were normalized. A gene expression data matrix was constructed based on the normalized bacterial gene expression data.

[0076] Then, the environmental data vectors and gene expression data matrices are integrated into a multi-scale framework to construct a multi-scale input data space. This multi-scale framework covers different temporal and spatial scales, where the temporal scales include stages such as mycelial growth and fruiting body differentiation, and the spatial scales include the interaction between the growth environment of a single mycelium and the entire culture environment. Integrating the environmental data vectors and gene expression data matrices to construct a multi-scale input data space ensures that the model can perform effective predictions and analyses at multiple scales.

[0077] The multi-scale input data space is input into a preset model structure to construct a multi-scale digital twin model.

[0078] In one embodiment, the preset model structure is based on a Bayesian network, and the parameters of the multi-scale digital twin model are dynamically calibrated through the Bayesian network.

[0079] Specifically, Bayesian networks are used to model the causal relationships between different factors, such as the relationship between environmental factors (temperature, humidity, light, etc.) and bacterial gene expression (transcriptome data, gene expression level data). Bayesian networks can infer the potential state and dynamic changes of a system under conditions of uncertainty. The specific process is as follows:

[0080] Determine the dependencies between variables, such as the influence of environmental factors on gene expression; construct the structure of the Bayesian network model, select appropriate nodes and directed edges (causal relationships) between them; and set the probability distribution of each node (usually using Gaussian or multinomial distribution).

[0081] Using real-time collected data on the operation of the equipment and the growth environment of the edible fungi, the parameters in the Bayesian network are dynamically adjusted, and the conditional probabilities in the Bayesian network are updated step by step to ensure the accuracy and real-time performance of the model.

[0082] By using equipment operation data (such as vibration, noise, and historical fault information), model parameters are calibrated to optimize the accuracy of model predictions. For example, when the temperature and humidity are too high or the gas concentration is abnormal, the model parameters are automatically adjusted to optimize the environment for edible fungi growth.

[0083] Use partial experimental data (such as data not used for model training) to verify the model's accuracy and predictive ability, and output the verification results, including: whether it can accurately predict the growth status of edible fungi, the trend of gene expression changes, and the regulation of the culture environment; use cross-validation technology to evaluate the model's performance on different datasets to ensure its generalization ability.

[0084] Based on the validation results, adjust the Bayesian network structure or optimize the data preprocessing process, such as using more accurate interpolation methods or standardization techniques. Different algorithms, such as deep learning and support vector machines, can also be tried to improve prediction accuracy.

[0085] By collecting real-time data on equipment operation, edible fungi growth environment, and strain gene expression, the model is continuously optimized to ensure rapid adaptation to changing environments and edible fungi growth requirements in practical applications.

[0086] Based on the calibrated multi-scale digital twin model, the system outputs monitoring results of edible fungi growth status, cultivation condition adjustment instructions, and equipment fault early warning information. The monitoring results reflect the real-time growth status of the edible fungi. The cultivation condition adjustment instructions are used to regulate environmental parameters, and are output based on changes in environmental and gene expression data to optimize the growth process. Based on equipment operation data, potential equipment faults are identified and early warnings are issued to avoid negative impacts on the edible fungi growth environment.

[0087] In this embodiment, a multi-scale digital twin model is constructed based on edible fungi growth environment data and strain gene expression data. By dynamically calibrating parameters using a Bayesian network, real-time monitoring of edible fungi growth status and precise control of cultivation conditions are achieved. Compared to traditional empirical control, this model combines multi-scale data at both the environmental and genetic levels, avoiding human error, stabilizing production conditions within the optimal range, improving the consistency of edible fungi quality, and reducing resource waste.

[0088] Step S3: Update the multi-scale digital twin model based on the real-time meteorological data and historical production data, and use the updated multi-scale digital twin model to compensate for environmental fluctuations.

[0089] In one specific implementation, real-time meteorological data, including temperature, humidity, wind speed, and precipitation, is used to analyze the growth environment of edible fungi; historical production data is also obtained based on equipment operation data. Then, preprocessing operations such as cleaning, noise reduction, and missing value imputation are performed on both the real-time meteorological data and the historical production data.

[0090] An environmental fluctuation model is established based on real-time meteorological data to represent the impact of meteorological changes on the production system. This environmental fluctuation model can capture the nonlinear relationship between meteorological changes and the production process.

[0091] Historical production data and real-time meteorological data are used to initially train and calibrate the environmental fluctuation model, and output meteorological-production correlation model data to improve its prediction accuracy.

[0092] Furthermore, the process of updating the multi-scale digital twin model using an online learning algorithm is as follows:

[0093] The real-time meteorological data, historical production data, and meteorological-production correlation pattern data are input into an online learning algorithm. This algorithm updates the parameters of the multi-scale digital twin model. In each learning round, the weights and parameters of the multi-scale digital twin model are adjusted based on newly received data, ensuring that the weights and parameters adapt to the nonlinear impact of meteorological changes on production. Specifically, the weights and parameters are adjusted slightly based on real-time meteorological data and historical production data, enabling the model to respond in real-time to fluctuations in the external environment and ensuring that the model can respond to environmental changes and production status in real time.

[0094] Based on the comparison between the output of the multi-scale digital twin model and actual production, real-time error feedback adjustments are made, and the model can be continuously optimized through an adaptive mechanism even if the environment changes significantly.

[0095] In one specific implementation, the updated multi-scale digital twin model is used to compensate for environmental fluctuations, specifically including:

[0096] The updated multi-scale digital twin model is used to predict production changes in the next period and to adjust production parameters, such as the impact of temperature and humidity on equipment, based on meteorological forecasts.

[0097] Real-time monitoring of environmental changes and production deviations allows for adjustments based on compensation strategies learned from online learning models. For example, when the external temperature changes suddenly, production process parameters and energy consumption can be adjusted.

[0098] Regularly evaluate the model's performance, including comparing the deviation between the model's predictions and actual production results, and assessing its ability to compensate for environmental fluctuations.

[0099] If the model's prediction error is large and exceeds a preset threshold, the parameters or model structure of the online learning algorithm will be optimized, such as adjusting the learning rate and model complexity.

[0100] Through continuous processing and feedback of real-time data streams, the model will be continuously optimized to improve prediction accuracy and adaptability. As production processes change and meteorological data accumulates, the model should be continuously iterated and updated to ensure it can always adapt to new production and environmental conditions.

[0101] In this embodiment, the parameters of a multi-scale digital twin model are iteratively optimized by combining real-time meteorological data, historical production data, and the "meteorological-production correlation pattern" output by the environmental fluctuation model. This process can respond in real time to nonlinear fluctuations in the external environment, and by dynamically adjusting model weights and parameters, it ensures that the production process remains stable under complex meteorological conditions, reducing production deviations caused by environmental fluctuations.

[0102] Step S4: Integrate the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model to establish a fault knowledge graph. Analyze the fault knowledge graph using a graph neural network to identify fault propagation paths, locate fault points, and determine the causes of fault occurrence.

[0103] In one specific implementation, the process of establishing a fault knowledge graph is as follows:

[0104] By integrating the equipment operation data, edible fungus growth environment data, and equipment fault early warning information output by the multi-scale digital twin model, the nodes and edges of the fault knowledge graph are defined. The nodes include equipment, fault type, environmental factors, and equipment components; the edges represent the relationships between nodes, such as the relationship between equipment and fault, and the relationship between equipment and environmental factors.

[0105] By integrating historical failure cases and historical equipment operation data, and labeling failure types and failure propagation paths, a failure knowledge graph is constructed. Nodes in the failure knowledge graph represent various elements of the system, and edges represent causal or influence relationships between these elements.

[0106] A graph neural network is further designed, where each node represents a device, fault type, or environmental factor. The graph neural network can capture the spatial relationships and structural information between nodes. The features of each node include information such as the device's operating status, environmental data, and fault history. Using the collected historical fault cases and historical device operating data, the graph neural network is trained. During the training process, through message passing and aggregation operations between nodes, the relationships between nodes and potential fault propagation paths are learned. An appropriate loss function is set so that the graph neural network model can accurately identify fault points and their propagation paths.

[0107] In one specific implementation, the process of using a graph neural network to analyze the fault knowledge graph and identify the fault propagation path is as follows:

[0108] The trained graph neural network is used to analyze the fault knowledge graph and identify the propagation path of each fault from one device to other devices. Based on the propagation path, reasoning is used to identify the propagation path after the fault occurs, find the device nodes related to the fault, and trace the source of the fault. Based on the graph neural network analysis results, the correlation between various devices is identified, and it is determined which devices were affected or caused the fault to propagate when it occurred.

[0109] The process of locating the fault point and its cause, achieving second-level fault point location, is as follows:

[0110] The system continuously monitors the equipment operation data and the edible fungus growth environment data, inputs the equipment operation data and the edible fungus growth environment data into a graph neural network, predicts whether a fault has occurred, and generates a fault prediction result.

[0111] Based on the fault prediction results, the fault point is located, and the root cause of the fault is identified. Furthermore, by combining the fault propagation path, the specific equipment and its location where the fault occurred are identified, and the upstream root cause that triggered the fault is traced.

[0112] Ultimately, based on the location of the fault and its cause, immediate response measures are provided, such as automatically adjusting equipment parameters and activating backup equipment, to achieve fault handling and recovery.

[0113] By continuously optimizing the graph neural network model—through increasing datasets, improving algorithms, or adjusting model architecture—the accuracy and real-time performance of fault diagnosis are enhanced. Validation is performed using different datasets to ensure the model can efficiently and accurately locate faults under various conditions, validating key performance indicators such as accuracy, recall, and fault location time. The fault diagnosis module is deployed in a real-world environment, integrated with industrial equipment and environmental monitoring systems, to achieve real-time monitoring and fault diagnosis. Based on feedback from actual use, the model is further optimized to improve system stability and accuracy.

[0114] In this embodiment, equipment operation data, environmental data, and fault warning information output by a multi-scale digital twin model are integrated to construct a fault knowledge graph containing equipment, fault types, and environmental factors. Graph neural networks are used to analyze node relationships, enabling fault propagation path identification and second-level location tracking. The location results, combined with cultivation condition adjustment instructions output by the model, can quickly trigger response measures, such as adjusting gas composition sensor parameters in the affected area, shortening fault handling time and significantly reducing negative impacts on the edible fungi growth environment.

[0115] In summary, the IoT-based digital twin construction method for edible mushroom factories provided by this invention achieves precise control and efficient management of the entire edible mushroom production process by constructing a multi-scale digital twin model and combining it with a dynamic optimization mechanism. This multi-scale digital twin model is constructed based on edible mushroom growth environment data and strain gene expression data. Parameters are dynamically calibrated using a Bayesian network, and continuously updated using real-time equipment operation data and environmental data. It outputs growth status monitoring results and cultivation condition adjustment instructions, stabilizing production conditions within the optimal range and avoiding resource waste caused by human error. Simultaneously, an online learning algorithm is introduced, combining real-time meteorological data and historical production data to iteratively optimize the parameters of the multi-scale digital twin model, responding in real-time to nonlinear fluctuations in the external environment and significantly enhancing stability under complex meteorological conditions. Furthermore, by integrating equipment operation data, environmental data, and fault warning information output by the model, a fault knowledge graph is constructed. Graph neural networks are used to analyze node relationships, accurately identifying fault propagation paths and locating fault points. Through deep integration of fault diagnosis and the digital twin model, second-level fault location and efficient response are achieved, and intelligent decision-making throughout the entire process is supported by multi-source data fusion. In addition, by simultaneously acquiring multi-dimensional data on environment, gene expression, and equipment operation, comprehensive input is provided for the multi-scale digital twin model. Combined with the growth status monitoring results, fault warning information, and adaptively updated environmental fluctuation compensation strategies output by the model, the factory's operating status can be monitored in real time through the digital twin visualization interface, enabling the formulation of scientific scheduling strategies.

[0116] The following are device embodiments corresponding to the above method embodiments, such as... Figure 2 As shown, Figure 2This diagram illustrates the structure of a digital twin construction device for an edible mushroom factory based on the Internet of Things (IoT) according to an embodiment of the present invention. This device embodiment can be implemented in conjunction with the above-described method embodiment. The relevant technical details mentioned in the above method embodiment remain valid in this device embodiment, and will not be repeated here to avoid repetition.

[0117] A digital twin construction device 300 for an edible mushroom factory based on the Internet of Things includes:

[0118] The data acquisition module 310 is used to collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data and historical production data;

[0119] The multi-scale digital twin model construction module 320 is used to construct a multi-scale digital twin model based on the edible fungus growth environment data and fungal gene expression data, and to dynamically calibrate the parameters of the multi-scale digital twin model; wherein, the multi-scale digital twin model outputs the edible fungus growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information;

[0120] The adaptive dynamic model module 330 is used to update the multi-scale digital twin model based on the real-time meteorological data and historical production data, and to use the updated multi-scale digital twin model to compensate for environmental fluctuations.

[0121] The fault diagnosis module 340 is used to integrate the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model, establish a fault knowledge graph, and use a graph neural network to analyze the fault knowledge graph to identify the fault propagation path, locate the fault point, and determine the cause of the fault.

[0122] This device embodiment can be implemented in conjunction with the implementation methods described above. The relevant technical details mentioned in the implementation methods of the above embodiments remain valid in the implementation methods of this method embodiment, and will not be repeated here to avoid repetition.

[0123] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0124] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0127] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a digital twin of an edible mushroom factory based on the Internet of Things, characterized in that, include: Collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data, and historical production data; Based on the edible fungi growth environment data and strain gene expression data, a multi-scale digital twin model is constructed, and the parameters of the multi-scale digital twin model are dynamically calibrated; wherein, the multi-scale digital twin model outputs the edible fungi growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information; The multi-scale digital twin model is updated based on the real-time meteorological data and historical production data, and the updated multi-scale digital twin model is used to compensate for environmental fluctuations. By integrating the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model, a fault knowledge graph is established. The graph neural network is then used to analyze the fault knowledge graph to identify the fault propagation path, locate the fault point, and determine the cause of the fault.

2. The method according to claim 1, characterized in that, The construction of a multi-scale digital twin model based on the edible fungi growth environment data and strain gene expression data includes: use In principle, outliers in the edible fungi growth environment data are removed, and missing values ​​are filled in by interpolation. The edible fungi growth environment data is then standardized, and an environmental data vector is constructed based on the standardized edible fungi growth environment data. Outliers were removed from the gene expression data of the bacterial strain using interquartile range, missing values ​​were filled by interpolation, and the data was normalized. A gene expression data matrix was constructed based on the normalized gene expression data of the bacterial strain. The environmental data vector and gene expression data matrix are integrated into a multi-scale framework to construct a multi-scale input data space, which covers different time and spatial scales. The multi-scale input data space is input into a preset model structure to construct a multi-scale digital twin model.

3. The method according to claim 1, characterized in that, The multi-scale digital twin model is constructed based on a Bayesian network. Dynamic calibration of the parameters of the multi-scale digital twin model includes: Using the real-time collected equipment operation data and edible fungus growth environment data, the parameters in the Bayesian network are dynamically adjusted, and the conditional probabilities in the Bayesian network are gradually updated. The model parameters are calibrated using the device's operating data. Verify the accuracy and predictive ability of the model, output the verification results, and adjust the Bayesian network structure or optimize the data preprocessing process based on the verification results; The model is continuously optimized by collecting real-time data on the equipment operation, the growth environment of the edible fungi, and the gene expression data of the fungi strains. Based on the calibrated multi-scale digital twin model, the system outputs monitoring results of edible fungi growth status, instructions for adjusting culture conditions, and early warning information for equipment malfunctions. The monitoring results of edible fungi growth status reflect the real-time growth status of edible fungi, and the instructions for adjusting culture conditions are used to regulate the environmental parameters of edible fungi culture. The instructions for adjusting culture conditions are output based on changes in environmental data and gene expression data.

4. The method according to claim 1, characterized in that, The step of updating the multi-scale digital twin model based on the real-time meteorological data and historical production data includes: The real-time meteorological data is obtained based on the edible fungus growth environment data; The historical production data is obtained based on the equipment operation data; An environmental fluctuation model is established based on the real-time meteorological data. The environmental fluctuation model is used to capture the nonlinear relationship between meteorological changes and production processes. The environmental fluctuation model is initially trained and calibrated using the historical production data and real-time meteorological data, and meteorological-production correlation model data is output. The real-time meteorological data, historical production data, and meteorological-production correlation pattern data are input into an online learning algorithm. The parameters of the multi-scale digital twin model are updated through the online learning algorithm. In each round of learning, the weights and parameters of the multi-scale digital twin model are adjusted according to the newly received data so that the weights and parameters adapt to the nonlinear impact of meteorological changes on production. Based on the comparison results between the output of the multi-scale digital twin model and the actual production, real-time error feedback adjustments are made.

5. The method according to claim 4, characterized in that, Compensation for environmental fluctuations using the updated multi-scale digital twin model includes: The updated multi-scale digital twin model is used to predict production changes in the next period, and production parameters are adjusted based on meteorological forecasts. Real-time monitoring of environmental changes and production deviations, and adjustments made through compensation strategies based on online learning models; The deviation between the model's predictions and actual production results is compared periodically to assess the multi-scale digital twin model's ability to compensate for environmental fluctuations. If the model prediction error exceeds a preset threshold, the parameters or model structure of the online learning algorithm will be optimized.

6. The method according to claim 1, characterized in that, The establishment of the fault knowledge graph includes: By integrating the equipment operation data, edible fungus growth environment data, and equipment fault early warning information output by the multi-scale digital twin model, the nodes and edges of the fault knowledge graph are defined, where nodes include equipment, fault type, environmental factors, and equipment components, and edges represent the relationships between nodes; Integrate historical failure cases and historical equipment operation data, label failure types and failure propagation paths, and construct a failure knowledge graph; Design a graph neural network and train it using the collected historical fault cases and historical equipment operation data. During the training process, the relationship between nodes and potential fault propagation paths are learned through message passing and aggregation operations between nodes.

7. The method according to claim 6, characterized in that, The fault knowledge graph is analyzed using a graph neural network to identify fault propagation paths, including: The trained graph neural network is used to analyze the fault knowledge graph and identify the propagation path of each fault from one device to other devices. Based on the propagation path, identify the device nodes related to the fault and trace the source of the fault. Identify the interrelationships between various devices and determine which devices are affected or cause the fault to propagate when it occurs.

8. The method according to claim 7, characterized in that, The location of the fault point and the cause of the fault include: The system continuously monitors the equipment operation data and the edible fungus growth environment data, inputs the equipment operation data and the edible fungus growth environment data into a graph neural network, predicts whether a fault has occurred, and generates a fault prediction result. Based on the fault prediction results, the fault point is located and the root cause of the fault is identified. By combining the fault propagation path, the specific equipment and its location where the fault occurred can be identified, and the upstream root cause of the fault can be traced.

9. The method according to claim 1, characterized in that, The data on the growth environment of edible fungi include temperature and humidity data, light data, gas composition data, and culture medium data; The bacterial strain gene expression data includes transcriptome data and gene expression level data; The equipment operation data includes basic operating parameters and fault-related data; The real-time meteorological data includes temperature, humidity, wind speed, and precipitation.

10. A device for constructing a digital twin of an edible mushroom factory based on the Internet of Things, characterized in that, include: The data acquisition module is used to collect data on the growth environment of edible fungi, gene expression data of fungi strains, equipment operation data, real-time meteorological data, and historical production data. A multi-scale digital twin model construction module is used to construct a multi-scale digital twin model based on the edible fungus growth environment data and fungal gene expression data, and to dynamically calibrate the parameters of the multi-scale digital twin model; wherein, the multi-scale digital twin model outputs the edible fungus growth status monitoring results, and identifies potential equipment faults based on equipment operation data and outputs fault warning information; An adaptive dynamic model module is used to update the multi-scale digital twin model based on the real-time meteorological data and historical production data, and to use the updated multi-scale digital twin model to compensate for environmental fluctuations. The fault diagnosis module integrates the equipment operation data, the edible fungus growth environment data, and the fault warning information output by the multi-scale digital twin model to establish a fault knowledge graph. The graph neural network is then used to analyze the fault knowledge graph to identify the fault propagation path, locate the fault point, and determine the cause of the fault.

Images