A remote intelligent material storage and transportation monitoring method and system based on intelligent algorithm

Abstract

The present invention discloses a remote intelligent material storage and transportation monitoring method and system based on intelligent algorithms, which relates to the field of intelligent monitoring technology, including: collecting data from material storage and transportation sites, and building a virtual simulation model on a digital twin platform based on the data; using UKF filtering to calculate and update state estimates based on the collected real-time data, dynamically adjusting parameters in the digital twin model, and correcting the virtual simulation model; using the state estimates of the future moment calculated by UKF filtering as the input of a dynamic Bayesian network, evaluating the risk of damage to materials during storage and transportation, triggering an early warning mechanism based on the evaluation results, and realizing safe storage and transportation of materials. The method of the present invention improves the monitoring accuracy and safety during storage and transportation by establishing a virtual model for data update and prediction, calculating risk scores, and dynamically adjusting early warning thresholds, thereby ensuring safe storage and transportation of materials in complex environments.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to a remote intelligent material storage and transportation monitoring method and system based on an intelligent algorithm. Background Art

[0002] With the development of Internet of Things technology, the intelligence and automation of material storage and transportation processes have gradually become the focus of the industry. Traditional material storage and transportation management mainly relies on manual monitoring and experience judgment, and often fails to detect potential risks in time, resulting in damage or loss of materials during storage and transportation. In recent years, the introduction of digital twin technology and intelligent algorithms has provided new solutions for material storage and transportation. Digital twin technology can monitor the status of materials in real time and simulate the response of materials under different storage and transportation conditions by constructing a virtual simulation model corresponding to the actual materials and their environment. However, the existing digital twin models still have certain deficiencies in real-time and accuracy, especially when dealing with complex environmental changes and nonlinear systems, it is often difficult to accurately predict the status of materials. In addition, although algorithms such as dynamic Bayesian networks (DBNs) and Kalman filters (UKFs) are widely used in state estimation and risk assessment, these algorithms are usually applied separately, lacking effective integration and collaborative optimization, resulting in their unsatisfactory performance when dealing with multi-dimensional complex data.

[0003] The existing material storage and transportation monitoring systems mainly face the following deficiencies when dealing with dynamic and changeable storage and transportation environments: First, traditional monitoring methods are difficult to update the material status model in real time, and cannot accurately reflect the external influences on materials during storage and transportation. Secondly, the existing early warning mechanisms mostly use fixed threshold methods, lack dynamic adjustment capabilities, cannot adapt to rapidly changing environmental conditions, and are prone to false alarms or missed reports. Furthermore, although the digital twin model can simulate the state changes of materials, its ability to predict future states is limited, and it is difficult to identify potential risks in advance. These deficiencies significantly affect the safety and reliability of the material storage and transportation process. The present invention proposes a remote intelligent material storage and transportation monitoring method based on an intelligent algorithm. By integrating a UKF filter, a dynamic Bayesian network and digital twin technology, accurate prediction of material status and dynamic assessment of risks are achieved, effectively solving the deficiencies of the existing technology in terms of real-time performance, accuracy and early warning mechanism. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is that the existing material storage and transportation monitoring technology is difficult to achieve efficient risk prediction and management when facing multi-dimensional and complex data.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a remote intelligent material storage and transportation monitoring method based on an intelligent algorithm, comprising: collecting data from material storage and transportation sites, and building a virtual simulation model on a digital twin platform based on the data; updating state estimates using UKF filtering calculations based on the collected real-time data, dynamically adjusting parameters in the digital twin model, and correcting the virtual simulation model; using the state estimates at future moments calculated by UKF filtering as input to a dynamic Bayesian network, evaluating the risk of damage to materials during storage and transportation, triggering an early warning mechanism based on the evaluation results, and realizing safe storage and transportation of materials.

[0007] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, the data of the material storage and transportation site includes attribute data, environmental parameter data, dynamic behavior data, logistics path and status data and historical data of the stored and transported materials; the attribute data of the stored and transported materials include the geometric size, weight and surface state of the material packaging; the dynamic behavior data includes the speed, vibration and tilt angle data of the materials during the storage and transportation process.

[0008] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, wherein: the virtual simulation model includes, according to the collected data, using 3D modeling software to generate a three-dimensional geometric model of the material, and setting corresponding parameters according to the attribute data of the stored and transported materials, modeling the storage and transportation environment, and creating a virtual scene corresponding to the actual warehouse and transportation tools; using the virtual simulation platform to set physical properties and simulation rules for the material model and the environmental model, setting transportation and storage rules, and creating a virtual simulation model; initializing the virtual model according to the collected data, continuously inputting the real-time data obtained from the sensor into the virtual simulation model, adjusting the model parameters through UKF filtering, and realizing online correction of the virtual simulation model.

[0009] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, wherein: correcting the virtual simulation model includes defining the initial state vector and the error covariance matrix , initialize the process noise covariance matrix and the observation noise covariance matrix , according to the state estimation and the error covariance matrix ; Introduce the sigma points generated by the third-order transformation method, discard the sigma points with low contribution through the sparse Sigma point selection strategy, use the sparse Sigma point set to recalculate the state estimation and covariance matrix, and complete the prediction step of UKF correction; After the prediction step is completed, for each sparse Sigma point after screening , the observation value is calculated through the observation model, and the weighted average of the observation value is performed to obtain the predicted observation value:

[0010] ,

[0011] ,

[0012] in, express The corresponding observed value; represents the observation model; represents the predicted observed value; represents the weight coefficient; Represents the filtered sparse Sigma point set; calculate the observation error covariance matrix:

[0013] ,

[0014] in, represents the observation error covariance matrix; Represents the matrix transpose; computes the cross covariance matrix between states and observations , integrating observation data into state estimation and using the observation error covariance and the cross covariance matrix , calculate the Kalman gain And update the state estimate , the error covariance matrix of the updated state Reflecting the uncertainty after observation correction, using the updated state estimate , perform real-time parameter correction on the digital twin model.

[0015] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, the prediction step of the UKF correction includes using the third-order transformation to generate sigma points that are more evenly distributed in the state space. , the number of sigma points is , weights are set to equal weights For each sigma point:

[0016] ,

[0017] Generate sigma points through the third-order transformation method:

[0018] ,

[0019] ,

[0020] in, Indicates the generated sigma points; Indicates the generated negative sigma point; Indicates at time An estimate of the system state; Represents a unit vector; calculate the contribution of each Sigma point, expressed as:

[0021] ,

[0022] in, express Contribution of Indicates time No. sigma points; Indicates the predicted state mean; sets the contribution threshold based on historical data ,when When the corresponding sigma point Keep, discard sigma point; for each sparse sigma point after screening , calculate the next moment through the system's state transfer function Status :

[0023] ,

[0024] in, represents the state transition function; Represents the control input vector; after the state propagation is completed, the sparse Sigma point is used to calculate the predicted state estimate at the next moment :

[0025] ,

[0026] Calculate the error covariance matrix of the predicted state , quantify the uncertainty of the predicted state:

[0027] ,

[0028] in, Represents matrix transpose.

[0029] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, the dynamic Bayesian network includes the predicted state estimation obtained by UKF filtering calculation. , initialize the material state variables of the dynamic Bayesian network and environment variables , using Bayes' theorem, calculate the state of the material and environment variables The joint probability of:

[0030] ,

[0031] in, Indicates the status of the material and environment variables The joint probability of Indicates that under given environmental conditions Next material status Probability of occurrence; Represents the prior probability of environmental variables; uses historical data to initialize the conditional probability distribution of DBN, and directly uses Conduct risk assessment, do not Update; through historical data, establish a physical model of the impact of temperature, humidity and acceleration on materials, and the impact on materials is used Indicates that risk events are defined based on the attribute data of storage and transportation materials If the materials are affected, When the preset threshold is exceeded, it is judged as a risk event Occur; use DBN's conditional probability model to calculate the current state and environment The conditional probability of a specific risk event occurring under :

[0032] ,

[0033] in, Indicates the probability of a risk event occurring under the current material status; Indicates the conditional probability that the material status is affected by the environment; calculates the comprehensive risk score, considering the probability of each risk event and performing a weighted average:

[0034] ,

[0035] in, represents the comprehensive risk score; Represents weight.

[0036] As a preferred solution of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm described in the present invention, the early warning mechanism includes: Calculate current risk score The risk score at the previous time step The difference between The sum of the risk score change rates within time steps is used to identify the overall trend of the risk score:

[0037] ,

[0038] ,

[0039] in, represents the rate of change of risk score; Indicates the trend direction; represents the time step used to analyze the trend; (·) represents the sign function; the trend weighting factor is calculated based on the trend direction and change rate, and the basic risk threshold is set based on historical data , at each time step , dynamically adjust the current threshold according to the trend weighting factor, and prevent drastic fluctuations of the dynamic threshold through time smoothing:

[0040] ,

[0041] ,

[0042] in, represents the trend weighting factor; represents the trend weighting coefficient; Represents a dynamic threshold; when No warning is triggered when Trigger the warning and visualize the material and environmental parameter data that triggers the warning through a virtual simulation model.

[0043] A remote intelligent material storage and transportation monitoring system based on an intelligent algorithm and adopting any of the methods described in the present invention, wherein: a virtual simulation module, based on the data of the material storage and transportation site, uses a virtual simulation platform to build a virtual simulation model; a dynamic correction module, corrects the virtual simulation model through UKF filtering, introduces sigma points generated by the third-order transformation method, discards sigma points with low contribution through a sparse Sigma point selection strategy, calculates state estimates, adjusts model parameters according to real-time data; an early warning model, uses the predicted state estimate obtained by UKF filtering as the input of a dynamic Bayesian network, evaluates the damage risk of materials during storage and transportation, and triggers an early warning mechanism according to the evaluation results.

[0044] A computer device comprises: a memory and a processor; the memory stores a computer program, comprising: the steps of implementing any one of the methods of the present invention when the processor executes the computer program.

[0045] A computer-readable storage medium stores a computer program, comprising: when the computer program is executed by a processor, the steps of implementing any one of the methods of the present invention are implemented.

[0046] Beneficial effects of the present invention: The method of the present invention realizes accurate modeling and dynamic monitoring of the status of materials by collecting data from material storage and transportation sites and constructing a virtual simulation model on a digital twin platform. The UKF filter is used to update the state estimate in real time and dynamically adjust the model parameters to ensure that the virtual simulation model is synchronized with the actual state. The future state calculated by the UKF filter is used as the input of the dynamic Bayesian network to evaluate the damage risk of the materials and trigger the early warning mechanism, thereby realizing the early prediction and processing of future risks. The overall method improves the monitoring accuracy and safety during the storage and transportation process, ensuring the safe storage and transportation of materials in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0048] Figure 1 An overall flow chart of a remote intelligent material storage and transportation monitoring method based on an intelligent algorithm provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0050] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a remote intelligent material storage and transportation monitoring method based on an intelligent algorithm, comprising:

[0051] S1: Collect data from material storage and transportation sites, and build a virtual simulation model on the digital twin platform based on the data.

[0052] Digital Twin Technology refers to the real-time mapping of the data of physical entities (such as products, equipment, processes, systems, etc.) into a virtual digital model through the Internet of Things, sensors, artificial intelligence, simulation and other digital technologies. This virtual model is a "digital twin", which can reflect the current state, historical records, and possible future behaviors of the physical entity.

[0053] Various IoT sensors are installed in materials and their storage and transportation environments to collect data from material storage and transportation sites, such as RFID tags for tracking material locations, temperature and humidity sensors for monitoring environmental conditions, and accelerometers for recording vibrations of materials during transportation.

[0054] The data of material storage and transportation sites include the attribute data of stored and transported materials, environmental parameter data, dynamic behavior data, logistics path and status data and historical data; the attribute data of stored and transported materials include the geometric dimensions, weight and surface status of material packaging; the dynamic behavior data include the speed, vibration and tilt angle data of materials during the storage and transportation process.

[0055] Furthermore, the collected material and environmental data are integrated to form a complete physical property data set, including information such as the weight, size, material, ambient temperature, humidity, vibration frequency, etc. of the materials. 3D modeling software is used to generate a three-dimensional geometric model of the materials, and corresponding parameters are set according to the attribute data of the stored and transported materials. The storage and transportation environment is modeled to create a virtual scene corresponding to the actual warehouse and transportation tools, including the shelf layout, the structure of the transportation tools, etc.

[0056] On virtual simulation platforms (such as Unity, Simulink, etc.), set physical properties and simulation rules for material models and environmental models, such as gravity, friction, elasticity, etc., in order to simulate the behavior of materials in different environments, and set specific transportation and storage rules, such as the stacking method of materials in the warehouse and the fixing method during transportation, to ensure that the simulation model can accurately reflect the actual operation situation.

[0057] The initial physical parameters of the materials (such as size, weight, material, etc.) and environmental parameters (such as temperature, humidity, vibration, etc.) are input into the digital twin model. The basic data of the initialized virtual model will continuously input the real-time data obtained from the sensor into the digital twin platform to synchronize the state of the virtual model with the actual state of the materials. When the sensor data changes significantly, the model parameters are adjusted through UKF filtering, so that the virtual simulation model can reflect the actual state of the materials and environmental conditions in real time, and realize the online correction of the virtual simulation model.

[0058] S2: Based on the collected real-time data, the UKF filter is used to calculate and update the state estimation, dynamically adjust the parameters in the digital twin model, and calibrate the virtual simulation model.

[0059] Furthermore, the virtual simulation model is corrected according to the collected real-time data using UKF filtering. UKF is a recursive filtering algorithm designed for state estimation of nonlinear systems. It propagates the state distribution nonlinearly through Unscented transformation to estimate the system state more accurately. Unlike the standard Kalman filter, UKF does not rely on linear approximation, but uses a set of weighted sampling points called sigma points to capture the true characteristics of the state distribution.

[0060] Although UKF has advantages in dealing with nonlinear problems, there are still problems in the digital twin model correction process directly applied to the present invention. The most important one is the balance between computational complexity and real-time performance. Specifically, UKF needs to generate and propagate multiple sigma points, which will significantly increase the amount of calculation in the high-dimensional state space. In an environment where material storage and transportation monitoring has high requirements for real-time performance, the computational delay caused by the large amount of calculation may cause the correction mechanism to lag, affecting the synchronization and accuracy of the overall system.

[0061] Therefore, we consider improving the UKF algorithm. In the entire UKF process, we first need to initialize the state vector, error covariance matrix, and noise covariance matrix:

[0062] Define the initial state vector , represents the initial state of the system, such as the initial position, velocity or environmental conditions of the material; initialize the error covariance matrix , this matrix reflects the uncertainty of the initial state estimate; the initialization process noise covariance matrix and the observation noise covariance matrix , reflects the uncertainty in the system model, This reflects the uncertainty in the observation process.

[0063] External disturbances (such as vibration and temperature changes) to materials during storage and transportation may have nonlinear effects on the state of materials. These nonlinear effects are difficult to accurately estimate using simple linear approximation methods, so a more accurate state estimation method is needed.

[0064] The sigma point generation method used by the traditional UKF can only capture second-order statistical characteristics, and when faced with highly nonlinear systems, the second-order approximation may not be sufficient to accurately reflect the complexity of the state distribution. Therefore, the introduction of the third-order transformation can generate more evenly distributed sigma points, thereby capturing higher-order statistical characteristics.

[0065] Use the third-order transformation to generate sigma points that are more evenly distributed in the state space. In the third-order transformation, we first need to determine the Cubature rule, which refers to how to choose sigma points to approximate the state distribution: for each dimension , the number of sigma points is , weights are set to equal weights For each sigma point:

[0066] ,

[0067] By using the third-order transformation method, the sigma points can be more evenly distributed on the surface of the state space:

[0068] ,

[0069] ,

[0070] in, Indicates the generated sigma points; Indicates the generated negative sigma point; Indicates at time An estimate of the system state; represents a unit vector.

[0071] It should be noted that the third-order transformation inherently requires higher computational accuracy and more computational steps because it requires a higher-dimensional approximation of the state space. Especially in high-dimensional systems, the number of sigma points grows exponentially with the increase in dimension, which will lead to a significant increase in the computational burden.

[0072] In order to deal with the problem that the number of sigma points increases exponentially with the increase of dimension caused by the third-order transformation, a sparse sigma point selection strategy is designed. The core idea of ​​this strategy is to retain only the sigma points that contribute more to the final estimation according to the local characteristics of the state space when generating sigma points, thereby reducing the computational complexity. This can not only reduce the computational burden, but also maintain the state estimation accuracy of high-dimensional systems.

[0073] The contribution of each sigma point can be defined as its influence on the state estimation, and its contribution is expressed by the weighted deviation of the point in the predicted state (i.e., the distance from the discrete center), which is expressed as:

[0074] ,

[0075] in, express Contribution of Indicates time No. sigma points; Represents the predicted state mean.

[0076] Set contribution thresholds based on historical data ,when When the corresponding sigma point Keep, discard sigma point; for each sparse sigma point after screening , calculate the next moment through the system's state transfer function Status :

[0077] ,

[0078] in, represents the state transition function; Represents the control input vector; each sparse Sigma point is propagated to the next moment through the state transfer function , thereby obtaining the corresponding sparse Sigma point set for subsequent state estimation and error covariance calculation.

[0079] After the state propagation is completed, the sparse Sigma point is used to calculate the predicted state estimate for the next moment :

[0080] ,

[0081] The next state estimate of the system is calculated by weighted averaging of sparse Sigma points. Since only Sigma points with large contributions are used, this process effectively reduces the computational complexity while maintaining a high estimation accuracy.

[0082] Calculate the error covariance matrix of the predicted state , quantify the uncertainty of the predicted state:

[0083] ,

[0084] in, Represents matrix transpose.

[0085] Furthermore, after the prediction step is completed, for each filtered sparse Sigma point , the observation value is calculated through the observation model, and the weighted average of the observation value is performed to obtain the predicted observation value:

[0086] ,

[0087] ,

[0088] in, express The corresponding observed value; represents the observation model; represents the predicted observed value; represents the weight coefficient; Represents the filtered sparse Sigma point set.

[0089] It should be noted that in the remote intelligent material storage and transportation monitoring system of the present invention, the observation model Defined according to the specific application scenario. For example, if the system monitors the status of materials through temperature and humidity sensors and GPS devices, then This may include mapping the actual location, temperature, humidity and other status information of the material to the sensor readings. This mapping process is modeled based on the characteristics of the sensor and the interaction between the material and the environment.

[0090] Compute the observation error covariance matrix:

[0091] ,

[0092] in, represents the observation error covariance matrix; Represents matrix transpose.

[0093] In order to effectively integrate the observation data into the state estimation, the cross covariance matrix between the state and the observation is calculated. The cross covariance matrix represents the covariance relationship between the state variables and the observation variables and is used to update the Kalman gain. The calculation formula is expressed as:

[0094] ,

[0095] Incorporate observation data into state estimation and use the observation error covariance and the cross covariance matrix , calculate the Kalman gain And update the state estimate , Kalman gain It is used to determine the correction amplitude of the observed data to the state estimation, expressed as:

[0096] ,

[0097] The size of the Kalman gain determines the sensitivity of the system to the observation error. Here, the error covariance matrix of the updated state is calculated based on the relationship between the cross covariance and the observation error covariance. Reflecting the uncertainty after observation correction, using the updated state estimate , perform real-time parameter correction on the digital twin model.

[0098] S3: The state estimate at the future moment calculated by UKF filtering is used as the input of the dynamic Bayesian network to evaluate the damage risk of materials during storage and transportation. The early warning mechanism is triggered according to the evaluation results to achieve safe storage and transportation of materials.

[0099] Furthermore, the dynamic Bayesian network (DBN) is a probabilistic graphical model used to model time series data, which can capture the dynamic characteristics of system status changing over time. Through DBN, the status of materials at different times in the storage and transportation process can be modeled, and risk assessment can be performed in combination with historical data.

[0100] The present invention has adopted UKF for state estimation and correction. UKF can handle nonlinear systems and predict states through sigma point propagation. DBN can serve as a supplement to UKF prediction and provide higher accuracy in time series prediction and state evaluation.

[0101] Secondly, DBN relies on a large amount of historical data for structure and parameter learning, and the historical data collection and use in the present invention are already part of the system. Therefore, DBN can directly use these data for modeling and reasoning. At the same time, when performing real-time correction of the virtual model, the predicted state estimate obtained by UKF can be directly initialized as the input of DBN, and the predicted state estimate itself is used as a future prediction data, which eliminates the prediction part in the conventional DBN network, simplifies the overall network model and reduces the computational complexity to a certain extent.

[0102] The specific steps of the dynamic Bayesian network are: based on the predicted state estimation calculated by UKF filtering , initialize the material state variables of the dynamic Bayesian network and environment variables , using Bayes' theorem, calculate the state of the material and environment variables The joint probability of:

[0103] ,

[0104] in, Indicates the status of the material and environment variables The joint probability of Indicates that under given environmental conditions Next material status Probability of occurrence; represents the prior probability of the environmental variable.

[0105] Use historical data to initialize the conditional probability distribution of DBN and directly use Conduct risk assessment, do not Updates.

[0106] Through historical data, a physical model of the impact of temperature, humidity and acceleration on materials is established. Indicates that risk events are defined based on the attribute data of storage and transportation materials If the materials are affected, When the preset threshold is exceeded, it is judged as a risk event occur.

[0107] It should be noted that in most storage environments, the three environmental factors of temperature, humidity and acceleration have the most important impact on materials. Take temperature as an example. The internal temperature of materials is affected by the external temperature. Especially when exposed to high or low temperature environments for a long time, temperature changes may cause changes in the chemical or physical properties of materials, thereby increasing the risk of damage. In the storage and transportation environment, the conduction of temperature can be described by the heat conduction equation. Since the temperature change during storage and transportation is discrete, the heat conduction equation is discretized:

[0108] ,

[0109] in, Indicates the internal temperature of the material; Indicates the external temperature; represents the time step; Represents the thermal conductivity coefficient.

[0110] Use DBN's conditional probability model to calculate the current state and environment The conditional probability of a specific risk event occurring under :

[0111] ,

[0112] in, Indicates the probability of a risk event occurring under the current material status; It represents the conditional probability that the state of materials is affected by the environment.

[0113] Calculate a comprehensive risk score, taking into account the probability of each risk event and taking a weighted average of them:

[0114] ,

[0115] in, represents the comprehensive risk score; Represents weight.

[0116] Furthermore, in the storage and transportation monitoring system, the material status and environmental conditions often change rapidly over time. For example, a sharp fluctuation in external temperature or a change in road conditions during transportation will have a significant impact on the material status. By analyzing the trend of environmental changes, the present invention is designed for complex storage and transportation conditions (such as temperature fluctuations, transportation vibrations, humidity changes, etc.) in the environment, which can reflect the changes in risks in real time and make dynamic responses based on the trends.

[0117] Each time step Calculate current risk score The risk score at the previous time step The difference between The sum of the risk score change rates within time steps is used to identify the overall trend of the risk score:

[0118] ,

[0119] ,

[0120] in, represents the rate of change of risk score; Indicates the trend direction; represents the time step used to analyze the trend; (·) represents a sign function.

[0121] It should be noted that It just provides a trend direction. Specifically, if the trend of the risk score change is positive (that is, the risk score is increasing), sign returns +1, indicating that the risk is increasing; if the trend of the risk score change is negative (that is, the risk score is decreasing), sign returns -1, indicating that the risk is decreasing.

[0122] Calculate trend weighting factors based on trend direction and rate of change, and set basic risk thresholds based on historical data , at each time step , dynamically adjust the current threshold according to the trend weighting factor, and prevent drastic fluctuations of the dynamic threshold through time smoothing:

[0123] ,

[0124] ,

[0125] in, represents the trend weighting factor; represents the trend weighting coefficient; Represents a dynamic threshold; when No warning is triggered when Trigger the warning and visualize the material and environmental parameter data that triggers the warning through a virtual simulation model.

[0126] It should be noted that if the risk score remains unchanged, sign returns to 0. However, in actual conditions, both the material status and environmental conditions are changing, and it is difficult for them to remain unchanged. Therefore, the situation where sign returns to 0 can be ignored. In order to maintain the accuracy of model prediction and prevent false positives, set =0 .

[0127] The present embodiment also provides a remote intelligent material storage and transportation monitoring system based on an intelligent algorithm, including a virtual simulation module, which constructs a virtual simulation model using a virtual simulation platform according to data from material storage and transportation sites; a dynamic correction module, which corrects the virtual simulation model through UKF filtering, introduces sigma points generated by the third-order transformation method, discards sigma points with low contribution through a sparse Sigma point selection strategy, calculates state estimates, adjusts model parameters according to real-time data; an early warning model, uses the predicted state estimate obtained by UKF filtering as the input of a dynamic Bayesian network, evaluates the risk of damage to materials during storage and transportation, and triggers an early warning mechanism based on the evaluation results.

[0128] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0129] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0130] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0131] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0132] Example 2, the following is an embodiment of the present invention, which provides a remote intelligent material storage and transportation monitoring method based on an intelligent algorithm. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0133] The experimental scenario simulates the actual status monitoring and risk assessment process of several different types of materials in a warehouse under different storage and transportation conditions. To ensure the authenticity of the experiment and the accuracy of the data, the experiment first selected five types of materials, each of which has specific requirements for environmental conditions during storage and transportation, such as different environmental parameters such as temperature, humidity and vibration.

[0134] Use 3D modeling software to generate three-dimensional geometric models of materials, and set model parameters such as weight, size and material according to actual measured data. Next, put these models into a virtual simulation environment, which simulates the shelf layout and transportation structure inside the warehouse, and sets corresponding physical properties (such as gravity, friction, elastic coefficient) and simulation rules (such as stacking method, fixing method) according to the properties of materials.

[0135] In the dynamic process of the experiment, the environmental conditions in the warehouse are gradually adjusted, such as increasing the temperature, reducing the humidity, and increasing the vibration frequency at a certain moment. After each environmental change, the sensor transmits the data to the digital twin platform, and the virtual model is corrected in real time through the UKF filtering algorithm to update the model parameters. In this process, the UKF filter generates and propagates sigma points to capture the changes in the state of materials. In each time step, the system inputs the future state estimate calculated by the UKF filter into the dynamic Bayesian network to evaluate the damage risk of materials. By analyzing the changes in the state of materials under different environmental conditions, the system triggers the corresponding early warning mechanism. The experimental data recorded by the simulation experiment are shown in Table 1.

[0136] Table 1 Experimental data table

[0137] ,

[0138] The temperature of the frozen food is simulated by artificially inputting high-temperature external environment data to simulate the effect of refrigeration failure. Under normal circumstances, refrigeration failure will cause such a large temperature change within the predicted time. However, the present invention can also achieve the early warning effect by analyzing the temperature trend and dynamically adjusting the temperature threshold.

[0139] The conditional probability of vibration events represents the movement of materials in the storage environment and is used to monitor the risk of material movement or falling. In this experimental data, no materials were moved, so the conditional probability of vibration events was small.

[0140] It can be seen from the table data that frozen foods and pharmaceutical products are sensitive to changes in temperature and humidity, especially frozen foods, whose temperature forecast rises from the initial -18°C to -10.5°C, resulting in higher conditional probability of risk events and comprehensive risk scores, proving that the present invention can predict future data through existing data and issue early warnings when the risk is high.

[0141] Through accurate risk score calculation, the present invention can dynamically adjust the warning threshold, avoiding the false alarm or missed alarm problems that may be caused by the traditional fixed threshold method, ensuring that warehouse managers can take measures quickly, which not only improves the accuracy of the warning, but also enhances the flexibility and response speed of the system.

[0142] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A remote intelligent material storage and transportation monitoring method based on intelligent algorithm, characterized in that: include: Collect data from material storage and transportation sites, and build a virtual simulation model on the digital twin platform based on the data; Based on the collected real-time data, the UKF filter is used to calculate and update the state estimate, dynamically adjust the parameters in the digital twin model, and calibrate the virtual simulation model; The state estimation at the future moment calculated by UKF filtering is used as the input of the dynamic Bayesian network to evaluate the damage risk of materials during storage and transportation. The early warning mechanism is triggered according to the evaluation results to achieve safe storage and transportation of materials. Initialize the virtual model according to the collected data, continuously input the real-time data obtained from the sensor into the virtual simulation model, adjust the model parameters through UKF filtering, and realize the online correction of the virtual simulation model; Correcting the virtual simulation model includes defining an initial state vector And error covariance matrix P0, initialize the process noise covariance matrix Q0 and observation noise covariance matrix R0, according to the state estimation and the error covariance matrix P k-1 ; The sigma points generated by the third-order transformation method are introduced, and the sigma points with low contribution are discarded through the sparse Sigma point selection strategy. The sparse sigma point set is used to recalculate the state estimation and covariance matrix to complete the prediction step of UKF correction. After the prediction step is completed, for each filtered sparse Sigma point The observation values ​​are calculated through the observation model, and the weighted average of the observation values ​​is performed to obtain the predicted observation values.

2. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 1, characterized in that: The data of the material storage and transportation site includes attribute data, environmental parameter data, dynamic behavior data, logistics path and status data and historical data of the stored and transported materials; The attribute data of storage and transportation materials include the geometric size, weight and surface condition of the material packaging; Dynamic behavior data includes the speed, vibration and tilt angle data of materials during storage and transportation.

3. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 2 is characterized by: The virtual simulation model includes generating a three-dimensional geometric model of materials using 3D modeling software based on the collected data, setting corresponding parameters based on the attribute data of the stored and transported materials, modeling the storage and transportation environment, and creating a virtual scene corresponding to the actual warehouse and transportation tools; Use the virtual simulation platform to set physical properties and simulation rules for material models and environmental models, set transportation and storage rules, and create virtual simulation models.

4. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 3 is characterized by: The weighted average of the observed values ​​is used to obtain the predicted observed values, and the expression is: in, express The corresponding observation value; h represents the observation model; represents the predicted observed value; W i represents the weight coefficient; S represents the sparse Sigma point set after screening; Represents the sparse Sigma points after screening; Compute the observation error covariance matrix: Among them, S k represents the observation error covariance matrix; T represents the matrix transpose; R0 represents the observation noise covariance matrix; Calculate the cross covariance matrix P between states and observations xz , integrate the observation data into the state estimation, and use the observation error covariance S k and the cross covariance matrix P xz , calculate the Kalman gain K k And update the state estimate Update the error covariance matrix P of the state k Reflecting the uncertainty after observation correction, using the updated state estimate Perform real-time parameter correction on the digital twin model.

5. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 4 is characterized by: The prediction step of the UKF correction includes using a third-order transform to generate sigma points that are more evenly distributed in the state space. For each dimension L, the number of sigma points is 2L, and the weights are set to equal weights W. i For each sigma point: Generate sigma points through the third-order transformation method: For the positive sigma point For negative sigma points in, Represents the i-th sigma point generated; Indicates the generated negative sigma point; represents the estimated value of the system state at time k-1; e i represents a unit vector; Calculate the contribution of each Sigma point, expressed as: Among them, D i express Contribution of Represents the i-th sigma point at time k-1; represents the predicted state mean; According to the historical data, the contribution threshold ∈ is set. i ≥∈, the corresponding sigma point Keep, discard D i <∈ sigma point; For each filtered sparse sigma point Calculate the state of the next moment k through the system's state transition function Where, f represents the state transfer function; u k-1 represents the control input vector; After the state propagation is completed, the sparse Sigma point is used to calculate the predicted state estimate for the next moment Calculate the error covariance matrix of the predicted state Quantify the uncertainty of the predicted state: Where T represents the matrix transpose.

6. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 5, characterized in that: The dynamic Bayesian network includes a predicted state estimate obtained by UKF filtering calculation. Initialize the material state variable S of the dynamic Bayesian network t and environment variable E t , using Bayes' theorem, calculate the material state S t and environment variable E t The joint probability of: P(S t ,E t )=P(S t |E t )·P(E t ) Among them, P(S t ,E t ) indicates the material status S t and environment variable E t The joint probability of t |E t ) indicates that under given environmental conditions E t Lower material status S t Probability of occurrence; P(E t ) represents the prior probability of environmental variables; Use historical data to initialize the conditional probability distribution of DBN, and directly use S t Conduct risk assessment without S t Updates; Through historical data, a physical model of the impact of temperature, humidity and acceleration on materials is established. The impact on materials is represented by R, and the risk event R is defined based on the attribute data of the stored and transported materials. i If the impact on the material exceeds the preset threshold, it is judged as a risk event R i occur; Use DBN's conditional probability model to calculate the current state S t and environment t The conditional probability P(R i |S t ,E t ): P(R i |S t ,E t )=P(R i |S t )·P(S t |E t ) Among them, P(R i |S t ) represents the probability of a risk event occurring under the current material status; P(S t |E t ) represents the conditional probability that the state of materials is affected by the environment; Calculate a comprehensive risk score, taking into account the probability of each risk event and taking a weighted average of them: Among them, R t represents the comprehensive risk score; w i Represents weight.

7. The remote intelligent material storage and transportation monitoring method based on intelligent algorithm as claimed in claim 6, characterized in that: The early warning mechanism includes calculating the current risk score R at each time step t. t and the risk score R at the previous time step t-1 The difference between the two time steps is used to obtain the change rate of the risk score. The sum of the risk score change rates in consecutive n time steps is calculated to identify the overall trend of the risk score: ΔR t =R t -R t-1 Among them, ΔR t represents the rate of change of risk score; D represents the trend direction; n represents the time step used to analyze the trend; sign(·) represents the sign function; According to the trend direction and rate of change, the trend weighting factor is calculated, and the basic risk threshold R is set according to historical data. base , at each time step t, the current threshold is dynamically adjusted according to the trend weighting factor, and the drastic fluctuation of the dynamic threshold is prevented by time smoothing: Among them, W t represents the trend weighting factor; k represents the trend weighting coefficient; R dyn,t Represents the dynamic threshold; when R t ≤R dyn,t No warning is triggered when R t >R dyn,t Trigger the warning and visualize the material and environmental parameter data that triggers the warning through a virtual simulation model.

8. A remote intelligent material storage and transportation monitoring system based on an intelligent algorithm using any of the methods of claims 1 to 7, characterized in that: include, The virtual simulation module uses the virtual simulation platform to build a virtual simulation model based on the data of material storage and transportation sites; The dynamic correction module corrects the virtual simulation model through UKF filtering, introduces the sigma points generated by the third-order transformation method, discards the sigma points with low contribution through the sparse Sigma point selection strategy, calculates the state estimation, and adjusts the model parameters according to the real-time data; The early warning model uses the predicted state estimate obtained by UKF filtering as the input of the dynamic Bayesian network to evaluate the damage risk of materials during storage and transportation, and triggers the early warning mechanism based on the evaluation results.

9. A computer device comprising: Memory and processor; The memory stores a computer program, which is characterized in that when the processor executes the computer program, the steps of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm as described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the remote intelligent material storage and transportation monitoring method based on intelligent algorithm as described in any one of claims 1-7 are implemented.

Images