Full logistics object visualization method and system based on digital twinning

Abstract

The invention discloses a full logistics object visualization method and system based on digital twinning, and the method comprises the steps: collecting key state parameters and other environment parameters of a logistics object in real time through multiple types of sensors disposed on the logistics object; preprocessing the key state parameters to form a time sequence data set; extracting key features influencing the state of the logistics object from the time sequence data set by using a feature selection algorithm; constructing a digital twinborn model of the logistics object through a generative adversarial network technology by using the extracted key features; and on the basis of the environmental parameters, constructing an environmental model based on machine learning, and integrating the environmental model and the digital twinborn model to form a comprehensive virtual environment and visually display the virtual environment. According to the embodiment of the invention, comprehensive monitoring and dynamic display of the logistics object can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital twins, and in particular to a method and system for visualizing all logistics objects based on digital twins. Background Art

[0002] With the rapid development of the logistics industry, modern logistics management is facing higher and higher requirements, including improving transportation efficiency, reducing costs, and ensuring cargo safety. In order to achieve these goals, the logistics industry is gradually adopting digital and intelligent technologies to improve the transparency and flexibility of operations. In this context, digital twin technology, as an emerging solution, has begun to attract widespread attention. By creating a virtual model of a physical object, digital twins can reflect its status and performance in real time, providing new ideas and tools for logistics management. However, existing digital twin applications still have some limitations, which affect the reliability and accuracy of digital twin models. Summary of the invention

[0003] The purpose of the present invention is to provide a full logistics object visualization method and system based on digital twins to address the deficiencies in the prior art and enable comprehensive monitoring and dynamic display of logistics objects.

[0004] An embodiment of the present application provides a full logistics object visualization method based on digital twins, the method comprising: Through the multi-type sensors installed on the logistics objects, the key status parameters and other environmental parameters of the logistics objects are collected in real time; Preprocessing the key state parameters to remove noise and outliers to form a time series data set; Using a feature selection algorithm, extract key features that affect the state of logistics objects from the time series data set; By using the extracted key features and generative adversarial network technology, a digital twin model of the logistics object is constructed, which can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; Based on the environmental parameters, a machine learning-based environmental model is constructed to predict the performance of logistics objects under different environmental conditions in real time. The environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it.

[0005] Optionally, the method of using the extracted key features to construct a digital twin model of the logistics object by generating adversarial network technology includes: Standardize or normalize the extracted key features to ensure that the feature values ​​at different scales are comparable; The generator receives random noise and extracted features as input, and through multiple fully connected layers or convolutional layers, the generator uses a deep learning structure to generate fake simulated logistics object status data; The discriminator receives the real logistics object status data and the fake data generated by the generator as input. The discriminator outputs a probability value through multiple fully connected layers or convolutional layers to indicate whether the input data is real data. The generator and the discriminator are trained until the logistics object status data generated by the generator and the real logistics object status data tend to be consistent in feature distribution, and a trained generative adversarial network is obtained as the digital twin model of the logistics object; Visually compare the generated logistics object status data with the actual logistics object status data, observe the distribution of the two in the feature space, and ensure that the results generated by the digital twin model are logically reasonable.

[0006] Optionally, constructing an environment model based on machine learning based on the environment parameters includes: Selecting target variables according to application scenarios, wherein the target variables are used to evaluate the performance of logistics objects under different environmental conditions; Determine the machine learning model to be trained based on the type of target variable; Preprocess the environmental parameter data, and divide the preprocessed environmental parameter data set into a training set and a test set; The determined machine learning model is trained using the training set, and the model parameters are adjusted so that it can adapt to the relationship between the environmental input and the response variable, thereby obtaining the trained machine learning model as the environmental model.

[0007] Optionally, the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize the environment, including: The prediction results of the environmental model are embedded in the digital twin model to update the status of the logistics object in real time. When the environmental conditions change, the environmental model predicts the impact on the status of the logistics object and feeds the impact information back to the digital twin model. In addition, the digital twin model can simulate the status according to the output parameters of the environmental model. The state simulation of the embedded digital twin model and the prediction results of the environmental adaptability model are visualized and displayed as a comprehensive virtual environment, allowing users to monitor the status of logistics objects and their environment in real time.

[0008] Another embodiment of the present application provides a full logistics object visualization system based on digital twins, the system comprising: The acquisition module is used to collect key status parameters and other environmental parameters of the logistics object in real time through multiple types of sensors installed on the logistics object; A processing module, used for preprocessing the key state parameters to remove noise and outliers to form a time series data set; An extraction module, used to extract key features that affect the state of logistics objects from the time series data set using a feature selection algorithm; The first building module is used to use the extracted key features to build a digital twin model of the logistics object through generative adversarial network technology, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; The second construction module is used to construct an environmental model based on machine learning based on the environmental parameters to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

[0009] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when running.

[0010] Yet another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the methods described above.

[0011] Compared with the prior art, the present invention provides a full logistics object visualization method based on digital twins, which collects key state parameters and other environmental parameters of logistics objects in real time through multi-type sensors set on logistics objects; preprocesses the key state parameters to form a time series data set; uses a feature selection algorithm to extract key features that affect the state of logistics objects from the time series data set; uses the extracted key features to build a digital twin model of the logistics object through generative adversarial network technology; based on environmental parameters, builds an environmental model based on machine learning, integrates the environmental model with the digital twin model, forms a comprehensive virtual environment and visualizes it, thereby enabling comprehensive monitoring and dynamic display of logistics objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A hardware structure block diagram of a computer terminal for a full logistics object visualization method based on digital twins provided in an embodiment of the present invention; Figure 2 A schematic diagram of a flow chart of a method for visualizing a full logistics object based on digital twins provided in an embodiment of the present invention; Figure 3 A structural schematic diagram of a full logistics object visualization system based on digital twins provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0014] The embodiment of the present invention first provides a full logistics object visualization method based on digital twins, which can be applied to electronic devices such as computer terminals, specifically ordinary computers, etc.

[0015] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for a full logistics object visualization method based on digital twins provided in an embodiment of the present invention. Figure 1 As shown, the computer device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0016] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any one of the digital twin-based full logistics object visualization methods.

[0017] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0018] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any full logistics object visualization method based on digital twins.

[0019] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0020] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0021] See also Figure 2 , an embodiment of the present invention provides a full logistics object visualization method based on digital twins, which may include the following steps: S201, collecting key state parameters and other environmental parameters of the logistics object in real time through multiple types of sensors installed on the logistics object; By setting up multiple types of sensors on logistics objects, key state parameters of logistics objects (such as temperature, humidity, position, vibration, etc.) and surrounding environmental parameters (such as climate conditions, traffic conditions, etc.) can be collected in real time. These sensors can be temperature sensors, humidity sensors, GPS positioning systems, acceleration sensors, etc., which can timely capture various dynamic changes of logistics objects during transportation. The real-time collection of these data provides raw materials for subsequent analysis and model construction, ensuring that the digital twin model can reflect the true state of the logistics object.

[0022] Real-time collection of key status parameters and other environmental parameters of logistics objects is of great significance in logistics management. First, it can help managers obtain the real-time status of logistics objects in a timely manner to ensure the safety and quality of goods, especially in the transportation of goods that are sensitive to temperature and humidity, and can effectively prevent losses. Secondly, by fully understanding the environmental parameters, the logistics management system can make more accurate decisions, such as adjusting transportation routes or arranging more suitable transportation methods in bad weather conditions. In addition, real-time data provides a basis for subsequent data analysis and model training, which helps to improve the accuracy and reliability of the digital twin model, thereby optimizing the overall logistics operations.

[0023] In terms of specific implementation methods, first of all, it is necessary to strategically deploy a variety of sensors on each logistics object. These sensors include temperature sensors, humidity sensors, location sensors (such as GPS), acceleration sensors, and vibration sensors. When selecting these sensors, the characteristics of the logistics object and the type of goods stored or transported should be considered. For example, perishable food may require more sensitive temperature and humidity monitoring. The installation location of the sensor should also be carefully designed to achieve the best monitoring effect and ensure that every key area is covered. Secondly, through wireless communication technology (such as Bluetooth, Wi-Fi or cellular network), these sensors transmit the real-time collected data to the central data processing system. The transmission protocol should choose a low-power and reliable protocol to maintain data consistency in long-distance transportation. During the data collection process, the sensor should automatically collect data regularly, such as every few seconds or minutes. The specific collection frequency can be set according to business needs.

[0024] S202, preprocessing the key state parameters to remove noise and outliers to form a time series data set; The process of preprocessing key state parameters aims to improve the quality and availability of data in order to more accurately describe the dynamic changes of logistics objects. The preprocessing steps mainly include noise removal and outlier detection. Noise is usually a random fluctuation caused by factors such as sensor errors, environmental interference or communication failures; outliers refer to data points that deviate significantly from the normal range, which may be caused by failures, incorrect operations or other unexpected events. When preprocessing data, filtering techniques (such as moving average filtering or median filtering) are first applied to smooth time series data and reduce the impact of noise. Then, by setting reasonable thresholds and standards, statistical methods (such as Z-score or IQR method) are used to identify and eliminate outliers. Ultimately, this process will form a clear and accurate time series data set, providing a reliable foundation for subsequent feature extraction and model building.

[0025] Effective preprocessing of key state parameters has important practical significance. First, it can significantly improve the accuracy and consistency of data, ensuring that the basis for subsequent analysis and modeling is reliable. This kind of data processing can not only avoid misunderstandings and wrong decisions caused by noise and outliers, but also improve the training effect of the model, thereby enhancing the accuracy and predictive ability of the digital twin model. Secondly, by cleaning the data, managers can more clearly identify the real state trend of logistics objects, which is convenient for making quick responses and effective decisions. For example, in the process of cargo transportation, once an abnormal state is found, timely measures can effectively reduce losses and risks, thereby improving the efficiency and safety of logistics operations. In addition, good data preprocessing lays a solid foundation for the subsequent work of big data analysis and machine learning, and ensures the stability and reliability of the entire logistics intelligent management system.

[0026] In terms of specific implementation, we first need to set up a data preprocessing module, which will receive the original state parameter data collected from multiple types of sensors. The preprocessing program can specify a time window for data collection according to the characteristics of the data stream, so as to ensure the continuity of the data and the integrity of the time series. After the data is acquired, the module will first perform preliminary noise filtering, using moving average filtering or weighted average methods to reduce the impact of instantaneous fluctuations by calculating the average value within a certain time range. Then, the system will detect abnormalities in the data by setting thresholds for standard deviations or extreme values, identify those abnormal values ​​that deviate significantly from the average level, and usually use statistical methods such as Z-score to quantify the "normality" of each data point and eliminate those abnormal data points that exceed the threshold. In order to ensure the integrity and accuracy of the final data set, the system can generate a data quality report that lists the data profile and changes before and after processing. At the same time, this preprocessing process should be real-time, which means that the system needs to dynamically update the monitoring time range to reflect the current status in a timely manner. After processing, the system will store the cleaned data in the database and organize it into a time series data set in chronological order to facilitate subsequent feature extraction and analysis. In addition, all operations of the data preprocessing module should be traceable so that retrospective analysis and adjustments can be performed when problems arise, which can better support intelligent decision-making and optimization of the entire logistics management system.

[0027] S203, using a feature selection algorithm to extract key features that affect the state of the logistics object from the time series data set; Using feature selection algorithms to extract key features that affect the state of logistics objects from time series data sets is an important step in the data analysis and modeling process. The purpose of feature selection is to screen out features that have a significant impact on the target variable (for example, the state change of logistics objects, transportation efficiency, etc.) from the original data, thereby improving the performance and interpretability of the model. The process usually includes several stages: first, the relationship between each feature (such as temperature, humidity, location, etc.) and the target variable is evaluated using correlation analysis (such as Pearson correlation coefficient or mutual information) to identify features with potential influence; second, various feature selection methods (such as filtering, wrapping, or embedding) are used for in-depth analysis to reduce the data dimension. These methods can include stepwise regression, Lasso regression, random forest feature importance scoring, etc., to ensure that the extracted features retain key information while removing redundant and noise features. Finally, the selected key features will form a set of high-quality input data to support subsequent modeling, so that the digital twin model can more accurately reflect the dynamic changes of logistics objects.

[0028] Feature selection plays a vital role in data analysis and modeling, and its importance is reflected in many aspects. First, by extracting key features, we can significantly improve the prediction accuracy and generalization ability of the model, avoid overfitting caused by redundant or irrelevant features, and thus improve the performance of the digital twin model. Secondly, feature selection can reduce computational complexity and training time, allowing the model to run more efficiently in large-scale data sets. In addition, a clear feature selection process helps to enhance the interpretability of the model, allowing managers to understand which factors play a dominant role in the state changes of logistics objects, thereby better guiding actual operations and decision-making. For example, identifying the impact of key environmental parameters (such as temperature and humidity) on the quality of cargo transportation can help logistics companies optimize transportation routes and storage conditions, thereby improving overall service quality and customer satisfaction. In short, feature selection not only improves the efficiency and accuracy of analysis, but also provides a solid data foundation for subsequent model applications and business decisions.

[0029] In the process of implementing the feature selection algorithm, it is first necessary to conduct a preliminary analysis of the time series data set to understand the basic structure and distribution characteristics of the data. Next, use correlation analysis methods (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) to evaluate the correlation between each feature and the target variable, and preliminarily screen out high-correlation features based on the analysis results. For example, a correlation threshold can be set to filter out features with low correlation with the target variable. Then, a more complex feature selection algorithm is adopted to further optimize the feature set, such as using a tree-based model (such as random forest) to evaluate the importance of features, and sorting features according to their importance scores, so as to select the features that contribute the most to the model. In addition, cross-validation methods can be used in combination with wrapping methods, and recursive feature elimination (RFE) and other techniques can be used to verify the impact of selected features on model performance.

[0030] In the entire feature selection process, ensuring the standardization and normalization of the data is also a very important step, so that features of different scales can remain comparable and avoid analytical bias caused by different feature dimensions. In the final stage of feature selection, the extracted features need to be visualized to intuitively understand the relationship and distribution characteristics between the features. To this end, dimensionality reduction techniques such as principal component analysis (PCA) can be used to project high-dimensional features into low-dimensional space to help analysts identify potential feature associations and patterns. Finally, after this series of steps, the selected key features will be integrated into the data preprocessing process as input for the subsequent construction of the digital twin model, providing strong data support for the intelligent management of the entire logistics object.

[0031] S204, using the extracted key features and generative adversarial network technology, a digital twin model of the logistics object is constructed, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; Building a digital twin model of logistics objects through generative adversarial networks (GAN) is an innovative approach of this application, which uses deep learning technology to combine the extracted key features with noise input to generate high-quality logistics object status data. Specifically, the generator uses multiple levels of neural network structures to take random noise and key features as input to generate simulated logistics status. This process ensures that the digital twin model can adaptively reflect the real-time logistics status and then perform dynamic simulation under various environmental conditions. This digital twin model can not only analyze the performance of logistics objects in real time, but also provide reference for decision support and promote the intelligentization of logistics management.

[0032] The digital twin model constructed by this method can provide high-fidelity state simulation and help logistics companies accurately predict the performance of items under different environmental conditions. This real-time feedback capability enhances the flexibility and adaptability of the logistics system, which can effectively reduce risks during transportation and improve customer satisfaction. In addition, the introduction of the digital twin model provides strong data support for subsequent optimization of management processes and improvement of resource allocation efficiency, which helps to achieve the goal of smart logistics.

[0033] Specifically, the extracted key features can be standardized or normalized to ensure that the feature values ​​at different scales are comparable; Standardization and normalization are key steps in data preprocessing, which aim to eliminate the scale differences between features and make all features within the same range, thereby improving the training efficiency and accuracy of the model. In this way, it can ensure that the generator gives equal attention to each feature during training, avoiding some features dominating the model's learning process due to excessive values.

[0034] This processing can significantly promote the training effect of the generator in the generative adversarial network, improve the quality of the generated data, and ultimately make the generated logistics object status data more in line with the actual situation, thereby enhancing the practical application value of the digital twin model.

[0035] In actual operation, for each key feature, first calculate its mean and standard deviation (for standardization), or minimum and maximum values ​​(for normalization). Then, apply these statistics to transform the feature data to ensure that the processed feature values ​​are in the same numerical range (for example, between 0 and 1), which will provide a consistent data basis for the input of the subsequent generator, making its learning process more efficient and stable.

[0036] The generator receives random noise and extracted features as input, and through multiple fully connected layers or convolutional layers, the generator uses a deep learning structure to generate fake simulated logistics object status data; In this step, the generator is designed to accept random noise and extracted feature data, and gradually transform them into logical and reasonable simulated logistics object status data after being processed by multiple layers of fully connected layers or convolutional layers. This process simulates the generation mechanism of real data, so that the generated data can reflect the actual status of the system.

[0037] Through this complex network structure, the generator can better capture the relationship between features, so that the generated state data is not only authentic but also diverse, able to cope with different changes in environmental conditions, providing a solid foundation for the construction of digital twin models.

[0038] In actual network design, the generator will adapt to the complexity of the model by setting different numbers of layers and neurons. Its structural selection can be adjusted based on existing experience, using activation functions such as ReLU or Leaky ReLU to ensure that the data always maintains effective learning ability during the transmission of each layer. During the training process, the weight of the generator is gradually adjusted by calculating the loss and back-propagating the generated data to improve the authenticity of the generated data.

[0039] The discriminator receives the real logistics object status data and the fake data generated by the generator as input. The discriminator outputs a probability value through multiple fully connected layers or convolutional layers to indicate whether the input data is real data. The main task of the discriminator is to distinguish whether the input data comes from the real state of the logistics object or the fake data generated by the generator. By using multiple layers of fully connected layers or convolutional layers, the discriminator can extract features from the input data and represent the authenticity of the input data through an output value.

[0040] The effectiveness of the discriminator directly affects the training effect of the generator. Through a continuous feedback mechanism, the generator can continuously optimize its output, making the generated data closer to the real data, and ultimately making the digital twin model more predictive.

[0041] When building a discriminator, you need to set the number of layers and neurons so that it can effectively capture the distribution of features, and use an appropriate activation function (such as the Sigmoid function) to limit the output value to between (0, 1), representing the probability of true or false. During the training process, real data and generated data are compared, and the weight of the discriminator is updated by calculating the loss function of the two to improve its discrimination ability.

[0042] The generator and the discriminator are trained until the logistics object status data generated by the generator and the real logistics object status data tend to be consistent in feature distribution, and a trained generative adversarial network is obtained as the digital twin model of the logistics object; In this stage, the generator and the discriminator are trained together in a confrontational manner. The generator strives to improve the quality of the generated data, while the discriminator continuously enhances its discrimination ability until the two reach a balance point, that is, the generated data tends to be consistent with the real data in terms of feature distribution.

[0043] This adversarial training mechanism can effectively improve the quality of data generated by the generator and ensure that the output data of the digital twin model is in line with the actual application scenario, so that the state simulation of logistics objects can show good accuracy and reliability under various conditions.

[0044] During the training process, the discriminator is first trained once to improve its discrimination ability, and then the generator is trained once to make the generated data more realistic. This process is repeated, and the distance between the generated data and the real data in various indicators is monitored until the loss function converges and the data output by the generator can pass the verification of the discriminator, thus forming an efficient digital twin model.

[0045] Visually compare the generated logistics object status data with the actual logistics object status data, observe the distribution of the two in the feature space, and ensure that the results generated by the digital twin model are logically reasonable.

[0046] The last step is to compare the generated logistics object status data with the real data through visualization, observe the distribution of the two in the feature space, and ensure that the results of the digital twin model are logically reasonable and realistic.

[0047] Visual comparison can not only help analysts intuitively judge the quality of generated data and the rationality of the model, but also help discover potential model adjustment needs, further optimize the performance of the digital twin model, and ensure its effectiveness in practical applications.

[0048] When implementing visual comparison, you can use dimensionality reduction techniques (such as principal component analysis PCA) to project high-dimensional data into low-dimensional space, and then draw a scatter plot to observe the distribution difference between the generated data and the real data. In addition, you can also use histograms or box plots to display the statistical characteristics of the data. Through these visualization methods, analysts can more clearly identify the advantages and disadvantages of the model, so as to make targeted optimization adjustments.

[0049] S205, based on the environmental parameters, construct an environmental model based on machine learning to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

[0050] In this digital twin-based full logistics object visualization method, by combining real-time collected environmental parameters and key state parameters, a machine learning-based environmental model is constructed to predict the performance of logistics objects under different environmental conditions in real time, greatly enhancing the intelligence level of the logistics system. The construction of the environmental model depends on the analysis of environmental parameters, so that logistics objects can adapt to environmental changes and achieve dynamic and real-time status monitoring and prediction. By integrating the environmental model with the digital twin model to form a comprehensive virtual environment, the system can provide more accurate data support under complex and changing conditions, which helps to improve decision-making efficiency.

[0051] This integrated approach not only improves the accuracy of logistics object state simulation, but also enhances the real-time and effectiveness of decision analysis. In actual operation, this integrated virtual environment can enable managers to understand the dynamic changes of logistics objects and the impact of environmental factors on their states in a timely manner through visual display, thereby optimizing logistics management strategies, reducing operational risks, and improving the efficiency and responsiveness of the entire logistics chain.

[0052] Specifically, based on the environmental parameters, an environmental model based on machine learning is constructed, and a target variable can be selected according to the application scenario, and the target variable is used to evaluate the performance of the logistics object under different environmental conditions; This is the first step in building an environmental model based on machine learning. The target variable can be some key performance indicator of the logistics object, such as transportation time, temperature response, or loss rate. This choice is crucial because it directly affects the training and prediction effect of the subsequent model.

[0053] By clarifying the target variable, we can ensure that the subsequent data processing and model selection are centered around improving the predictive power of the variable. This targeted selection helps to focus subsequent analysis, making the environmental model more targeted and practical in actual applications, thereby effectively supporting logistics management decisions.

[0054] In actual operation, you can first conduct a preliminary screening of each factor that may affect the performance of the logistics object, and combine professional knowledge and historical data to determine the most influential target variables. Then you need to set clear evaluation criteria to help benchmark when evaluating model performance in the future. Finally, the development team should use the insights of domain experts to discuss and confirm the selection of target variables to ensure that they are closely linked to the actual needs of the logistics business.

[0055] Determine the machine learning model to be trained based on the type of target variable; This step involves choosing an appropriate machine learning model based on the nature of the target variable. The target variable may be continuous (such as shipping time) or discrete (such as whether cargo loss occurs), and based on these characteristics, the analysis team needs to choose different types of machine learning algorithms. For example, for predicting continuous variables, linear regression or support vector regression (SVR) may be appropriate choices, while when the target variable is a classification problem, classification algorithms such as decision trees or random forests will be more effective. Wise model selection can greatly improve the accuracy and efficiency of subsequent models.

[0056] Clearly selecting the right machine learning model helps improve the prediction accuracy of the environmental model, thereby enhancing the real-time response capability of the digital twin model. This not only improves the scientific nature of business decisions, but also makes the model more robust in practical applications, thereby reducing the risks caused by improper model selection and ensuring the maximization of the economic benefits of logistics management.

[0057] During the implementation process, the team first needs to classify the target variable and understand its classification or regression characteristics. Then, based on existing domain knowledge and experience, list several commonly used machine learning models suitable for the target variable. After selecting the model, the team can conduct preliminary tests by setting up small-scale experiments to compare and evaluate the performance of different models on this specific problem. Finally, the best model is selected as the basis for subsequent training, ensuring flexibility throughout the project cycle so that adjustments can be made when necessary.

[0058] Preprocess the environmental parameter data, and divide the preprocessed environmental parameter data set into a training set and a test set; In this step, the collected environmental parameter data needs to be cleaned and processed to eliminate outliers, missing values ​​and noise. After processing, the data will be divided into training sets and test sets to facilitate subsequent model training and verification.

[0059] Through effective data preprocessing and partitioning, the effectiveness of the training process and the reliability of the model performance can be ensured. The training set is used for model learning, and the test set is used to evaluate the generalization ability of the model on unseen data, which is crucial for building a high-performance environmental model.

[0060] In the specific implementation, the original environmental parameter data set is first reviewed through statistical analysis and visualization to identify and process missing or outliers. Subsequently, standardization or normalization is performed to ensure that all features are in the same numerical range. Finally, the data set is randomly allocated according to a certain ratio (such as 80% for training and 20% for testing) to ensure the representativeness of the training set and test set so that the subsequent model can be effectively learned and evaluated.

[0061] The determined machine learning model is trained using the training set, and the model parameters are adjusted so that it can adapt to the relationship between the environmental input and the response variable, thereby obtaining the trained machine learning model as the environmental model.

[0062] In this stage, the training set prepared in the previous step is used to input into the selected machine learning model for training, and the parameters of the model are adjusted so that the model can effectively capture the relationship between environmental parameters and target variables. Through the effective training of the machine learning model, it can be ensured that the obtained environmental model has strong predictive and generalization capabilities, so that it can provide accurate environmental impact prediction for the digital twin model in practical applications, thereby realizing more intelligent logistics management.

[0063] During this process, the development team will use the training set to train the model, and use algorithm optimization techniques such as gradient descent and cross-validation to gradually adjust the model parameters to improve prediction accuracy. The training effect should be evaluated regularly, the learning progress of the model should be monitored through the loss function, and hyperparameter optimization should be performed when necessary to ensure the stability and reliability of the model under various conditions. After the training is completed, a test set should be retained for subsequent verification to ensure that the obtained environmental model can effectively cope with the changing environment in actual operations.

[0064] Specifically, the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it. The prediction results of the environmental model can be embedded in the digital twin model to update the status of the logistics object in real time. In this step, the prediction results generated by the environmental model are integrated into the digital twin model to ensure that the status of the logistics object can be updated in real time when the environmental conditions change. The environmental model generates predictions about the performance of the logistics object based on real-time environmental data, so that the digital twin model can reflect a more accurate real-time status. The effective implementation of this process depends on good data structure and interface design to ensure the smooth and timely flow of information.

[0065] By embedding the prediction results of the environmental model into the digital twin model, a more accurate dynamic state simulation can be achieved. This integration not only improves the responsiveness of logistics objects under different environmental conditions, but also enhances the intelligence level of the decision support system, allowing managers to make adjustments and optimize decisions more quickly, thereby improving the efficiency and flexibility of overall logistics management.

[0066] In the specific implementation, it is first necessary to ensure that the digital twin model has interfaces and algorithms that can receive and process the prediction results from the environmental model. This may include implementing dynamic data transmission procedures to quickly obtain updated information when environmental conditions change. By setting scheduled tasks or triggers and monitoring changes in environmental data in real time, the system can quickly call the prediction method of the environmental model and feed the results back to the digital twin model. On this basis, the digital twin model will use the updated prediction information to adjust its internal state representation and realize the dynamic update of the logistics object state.

[0067] Among them, when environmental conditions change, the environmental model predicts the impact on the state of the logistics object and feeds back the impact information to the digital twin model; This step involves the environmental model timely analyzing and predicting the possible impact of changes on the state of logistics objects when environmental conditions change. This prediction is obtained by modeling the relationship between environmental parameters and the state of logistics objects. The prediction results need to be passed to the digital twin model in the form of information feedback to ensure that it can quickly adjust the corresponding state representation and behavior strategy.

[0068] When environmental conditions change, timely and accurate prediction results can help the digital twin model better reflect the dynamic state of the actual logistics object and improve the system's response speed to environmental changes. This not only enhances the intelligence level of the logistics system, but also significantly reduces the risks caused by failure to respond to environmental changes in a timely manner, ensuring the safety and reliability of logistics operations.

[0069] In terms of implementation, the environmental model needs to have the ability to monitor changes in environmental parameters in real time, such as temperature, humidity, and transportation speed. When changes in environmental conditions are detected, the environmental model will automatically run the prediction algorithm to calculate the potential impact of these changes on the status of logistics objects. The prediction results will be formatted as feedback information through the data interface and transmitted to the digital twin model in real time. In order to ensure accurate information transmission, the system will regularly verify the performance of the interface and perform necessary optimization to reduce delays and improve the timeliness of feedback.

[0070] Moreover, the digital twin model can simulate the state according to the output parameters of the environmental model; In this step, the digital twin model performs state simulation using the output parameters provided by the environmental model. These output parameters include the analyzed environmental influencing factors and relevant state indicators of the logistics objects. The digital twin model adjusts its internal model based on this information to reflect the actual transportation and operation status. The state simulation process needs to ensure the effective flow and logical consistency of data to improve the authenticity and reliability of the final simulation results.

[0071] This state simulation based on the output parameters of the environmental model enables the digital twin model to adapt to changing external conditions. This capability not only improves the accuracy of the state of logistics objects, but also promotes intelligent decision-making in the system, allowing logistics companies to monitor and optimize processes in real time, ultimately improving overall operational efficiency and resource allocation efficiency.

[0072] In the specific implementation, the digital twin model must first have the ability to parse the output parameters of the environmental model to ensure that it can understand and process the data type it provides. This can be achieved by building a data adaptation layer so that data in different formats can be seamlessly connected. Subsequently, the digital twin model will apply the prediction parameters, run the state simulation algorithm, and calculate the latest state of the logistics object. This process can use visualization tools to display the simulation results, which is convenient for users to monitor and analyze in real time, and finally form a highly interactive monitoring interface.

[0073] The state simulation of the embedded digital twin model and the prediction results of the environmental adaptability model are visualized and displayed as a comprehensive virtual environment, allowing users to monitor the status of logistics objects and their environment in real time.

[0074] This step aims to combine the embedded digital twin model with the prediction results of the environmental adaptability model to form a comprehensive virtual environment for visualization. The comprehensive display not only includes the status of logistics objects, but also covers relevant information about environmental factors, so that users can monitor the operation of the entire logistics system at a glance. This visualization needs to include real-time data updates to ensure that users can obtain the most accurate information flow.

[0075] By centrally displaying the results of these models, users can fully understand the dynamic performance of logistics objects and changes in their environment. This intuitive visual interface enhances decision support capabilities, allowing users to evaluate and make decisions more quickly and accurately, thereby optimizing transportation and operation processes, improving work efficiency, and enhancing customer experience.

[0076] During the specific implementation process, the development team will create a visualization platform to integrate the data of the digital twin model and the environmental adaptability model. Use data visualization technology, such as dashboards, dynamic maps, and charts, to display real-time logistics status and environmental indicators. At the same time, the system must have user interaction functions, allowing users to customize the indicators and data ranges to view, and provide a multi-dimensional analysis perspective. In order to keep the data updated in real time, the platform should regularly obtain the latest information from various data sources to ensure that users can monitor the logistics objects and their environmental status in real time and improve the overall operational transparency.

[0077] It can be seen that by setting up multiple types of sensors on logistics objects, the key state parameters and other environmental parameters of logistics objects are collected in real time; the key state parameters are preprocessed to form a time series data set; the feature selection algorithm is used to extract the key features that affect the state of logistics objects from the time series data set; the extracted key features are used to build a digital twin model of the logistics object through the generative adversarial network technology; based on the environmental parameters, a machine learning-based environmental model is built, and the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it, so as to achieve comprehensive monitoring and dynamic display of logistics objects.

[0078] Another embodiment of the present invention provides a full logistics object visualization system based on digital twins, see Figure 3 , the system may include: The acquisition module 301 is used to acquire key state parameters and other environmental parameters of the logistics object in real time through multiple types of sensors arranged on the logistics object; The processing module 302 is used to pre-process the key state parameters to remove noise and abnormal values ​​to form a time series data set; An extraction module 303 is used to extract key features that affect the state of logistics objects from the time series data set using a feature selection algorithm; The first construction module 304 is used to construct a digital twin model of the logistics object by using the extracted key features and generating adversarial network technology, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; The second construction module 305 is used to construct an environmental model based on machine learning based on the environmental parameters to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

[0079] It can be seen that by setting up multiple types of sensors on logistics objects, the key state parameters and other environmental parameters of logistics objects are collected in real time; the key state parameters are preprocessed to form a time series data set; the feature selection algorithm is used to extract the key features that affect the state of logistics objects from the time series data set; the extracted key features are used to build a digital twin model of the logistics object through the generative adversarial network technology; based on the environmental parameters, a machine learning-based environmental model is built, and the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it, so as to achieve comprehensive monitoring and dynamic display of logistics objects.

[0080] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0081] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps: S201, collecting key state parameters and other environmental parameters of the logistics object in real time through multiple types of sensors installed on the logistics object; S202, preprocessing the key state parameters to remove noise and outliers to form a time series data set; S203, using a feature selection algorithm to extract key features that affect the state of the logistics object from the time series data set; S204, using the extracted key features and generative adversarial network technology, a digital twin model of the logistics object is constructed, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; S205, based on the environmental parameters, construct an environmental model based on machine learning to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

[0082] It can be seen that by setting up multiple types of sensors on logistics objects, the key state parameters and other environmental parameters of logistics objects are collected in real time; the key state parameters are preprocessed to form a time series data set; the feature selection algorithm is used to extract the key features that affect the state of logistics objects from the time series data set; the extracted key features are used to build a digital twin model of the logistics object through the generative adversarial network technology; based on the environmental parameters, a machine learning-based environmental model is built, and the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it, so as to achieve comprehensive monitoring and dynamic display of logistics objects.

[0083] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0084] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0085] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program: S201, collecting key state parameters and other environmental parameters of the logistics object in real time through multiple types of sensors installed on the logistics object; S202, preprocessing the key state parameters to remove noise and outliers to form a time series data set; S203, using a feature selection algorithm to extract key features that affect the state of the logistics object from the time series data set; S204, using the extracted key features and generative adversarial network technology, a digital twin model of the logistics object is constructed, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; S205, based on the environmental parameters, construct an environmental model based on machine learning to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

[0086] It can be seen that by setting up multiple types of sensors on logistics objects, the key state parameters and other environmental parameters of logistics objects are collected in real time; the key state parameters are preprocessed to form a time series data set; the feature selection algorithm is used to extract the key features that affect the state of logistics objects from the time series data set; the extracted key features are used to build a digital twin model of the logistics object through the generative adversarial network technology; based on the environmental parameters, a machine learning-based environmental model is built, and the environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it, so as to achieve comprehensive monitoring and dynamic display of logistics objects.

[0087] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made according to the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. A digital twin-based full logistics object visualization method, characterized in that: The method comprises: Through the multi-type sensors installed on the logistics objects, the key status parameters and other environmental parameters of the logistics objects are collected in real time; Preprocessing the key state parameters to remove noise and outliers to form a time series data set; Using a feature selection algorithm, extract key features that affect the state of logistics objects from the time series data set; By using the extracted key features and generative adversarial network technology, a digital twin model of the logistics object is constructed, which can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; Based on the environmental parameters, a machine learning-based environmental model is constructed to predict the performance of logistics objects under different environmental conditions in real time. The environmental model is integrated with the digital twin model to form a comprehensive virtual environment and visualize it.

2. The method according to claim 1, characterized in that The method uses the extracted key features and generative adversarial network technology to construct a digital twin model of the logistics object, including: Standardize or normalize the extracted key features to ensure that the feature values ​​at different scales are comparable; The generator receives random noise and extracted features as input, and through multiple fully connected layers or convolutional layers, the generator uses a deep learning structure to generate fake simulated logistics object status data; The discriminator receives the real logistics object status data and the fake data generated by the generator as input. The discriminator outputs a probability value through multiple fully connected layers or convolutional layers to indicate whether the input data is real data. The generator and the discriminator are trained until the logistics object status data generated by the generator and the real logistics object status data tend to be consistent in feature distribution, and a trained generative adversarial network is obtained as the digital twin model of the logistics object; Visually compare the generated logistics object status data with the actual logistics object status data, observe the distribution of the two in the feature space, and ensure that the results generated by the digital twin model are logically reasonable.

3. The method according to claim 2, characterized in that The step of constructing a machine learning-based environment model based on the environment parameters includes: Selecting target variables according to application scenarios, wherein the target variables are used to evaluate the performance of logistics objects under different environmental conditions; Determine the machine learning model to be trained based on the type of target variable; Preprocess the environmental parameter data, and divide the preprocessed environmental parameter data set into a training set and a test set; The determined machine learning model is trained using the training set, and the model parameters are adjusted so that it can adapt to the relationship between the environmental input and the response variable, thereby obtaining the trained machine learning model as the environmental model.

4. The method according to claim 3, characterized in that The step of integrating the environment model with the digital twin model to form a comprehensive virtual environment and visually displaying the environment model includes: The prediction results of the environmental model are embedded in the digital twin model to update the status of the logistics object in real time. When the environmental conditions change, the environmental model predicts the impact on the status of the logistics object and feeds the impact information back to the digital twin model. In addition, the digital twin model can simulate the status according to the output parameters of the environmental model. The state simulation of the embedded digital twin model and the prediction results of the environmental adaptability model are visualized and displayed as a comprehensive virtual environment, allowing users to monitor the status of logistics objects and their environment in real time.

5. A digital twin-based full logistics object visualization system, characterized in that: The system comprises: The acquisition module is used to collect key status parameters and other environmental parameters of the logistics object in real time through multiple types of sensors installed on the logistics object; A processing module, used for preprocessing the key state parameters to remove noise and outliers to form a time series data set; An extraction module, used to extract key features that affect the state of logistics objects from the time series data set using a feature selection algorithm; The first building module is used to use the extracted key features to build a digital twin model of the logistics object through generative adversarial network technology, so that it can simulate the state under various environmental conditions and reflect the dynamic changes of the logistics object in real time; The second construction module is used to construct an environmental model based on machine learning based on the environmental parameters to predict the performance of logistics objects under different environmental conditions in real time, integrate the environmental model with the digital twin model to form a comprehensive virtual environment and visualize it.

6. The system according to claim 5, characterized in that The first building module is specifically used for: Standardize or normalize the extracted key features to ensure that the feature values ​​at different scales are comparable; The generator receives random noise and extracted features as input, and through multiple fully connected layers or convolutional layers, the generator uses a deep learning structure to generate fake simulated logistics object status data; The discriminator receives the real logistics object status data and the fake data generated by the generator as input. The discriminator outputs a probability value through multiple fully connected layers or convolutional layers to indicate whether the input data is real data. The generator and the discriminator are trained until the logistics object status data generated by the generator and the real logistics object status data tend to be consistent in feature distribution, and a trained generative adversarial network is obtained as the digital twin model of the logistics object; Visually compare the generated logistics object status data with the actual logistics object status data, observe the distribution of the two in the feature space, and ensure that the results generated by the digital twin model are logically reasonable.

7. The system according to claim 6, characterized in that The second building block is specifically used for: Selecting target variables according to application scenarios, wherein the target variables are used to evaluate the performance of logistics objects under different environmental conditions; Determine the machine learning model to be trained based on the type of target variable; Preprocess the environmental parameter data, and divide the preprocessed environmental parameter data set into a training set and a test set; The determined machine learning model is trained using the training set, and the model parameters are adjusted so that it can adapt to the relationship between the environmental input and the response variable, thereby obtaining the trained machine learning model as the environmental model.

8. The system according to claim 7, characterized in that The second building block is specifically used for: The prediction results of the environmental model are embedded in the digital twin model to update the status of the logistics object in real time. When the environmental conditions change, the environmental model predicts the impact on the status of the logistics object and feeds the impact information back to the digital twin model. In addition, the digital twin model can simulate the status according to the output parameters of the environmental model. The state simulation of the embedded digital twin model and the prediction results of the environmental adaptability model are visualized and displayed as a comprehensive virtual environment, allowing users to monitor the status of logistics objects and their environment in real time.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 4 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 4.

Images