Liquefied oil product flow management method based on Internet of Things technology

Abstract

The invention discloses a liquefied oil product flow management method based on the Internet of Things technology, and the method comprises the following steps: installing a high-precision flowmeter at a key node of a liquefied oil product pipeline, monitoring the flow data in real time, and collecting the historical data; local processing and analysis are carried out on real-time and historical data by using a multi-source domain antagonistic transfer learning algorithm, and the processed data are transmitted to an Internet of Things platform through a self-adaptive multi-hop wireless sensor network. And storing and processing the data through space-time self-correlation analysis, and identifying a space-time mode and abnormal change. And real-time monitoring and prediction are performed in combination with a mixed event triggering prediction control algorithm. A multivariable time sequence decomposition and reconstruction algorithm is utilized to deeply analyze flow data, and a multi-kernel learning algorithm is combined to optimize a flow prediction model, extract key features, identify abnormities and optimize transportation and storage strategies of liquefied oil products. According to the invention, real-time monitoring, automatic management, data fusion and analysis and abnormal early warning of flow data are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of liquefied oil product management, and in particular to a liquefied oil product flow management method based on Internet of Things technology. Background Art

[0002] With the rapid development of the industrial and energy sectors, the importance of liquefied oil products in transportation and storage has become increasingly prominent. However, the traditional methods of liquefied oil product flow management have many problems and need to be improved through advanced technologies. In the existing technology, liquefied oil product flow management usually relies on manual monitoring and conventional flow metering equipment. These methods have the defects of untimely data acquisition, large metering errors, frequent manual intervention, and inability to monitor and manage in real time.

[0003] Traditional liquefied oil flow management usually relies on mechanical or simple electronic flow meters. The data collection frequency of these devices is low, and most of them rely on manual recording and manual uploading, which leads to a slow data acquisition process and inability to reflect flow changes in a timely manner. Untimely data acquisition will lead to insufficient real-time monitoring capabilities of liquefied oil flow, and will not be able to detect and handle abnormal situations in a timely manner, thus affecting the safe and efficient management of liquefied oil. Traditional flow meters are greatly affected by factors such as ambient temperature and pressure changes, and are prone to metering errors. In addition, the measurement and recording methods of manual intervention are prone to human errors and data omissions. Large metering errors directly affect the accurate measurement of liquefied oil products, leading to problems such as loss or overage during transportation and storage, which in turn affects economic benefits and safety.

[0004] The existing liquefied oil flow management system is highly dependent on manual operation, including data recording, uploading, analysis and exception handling. Manual operation is not only inefficient, but also prone to human errors. Too much manual intervention leads to low management efficiency, making it difficult to achieve automated and intelligent management, increasing operating costs and management difficulties. The existing system lacks real-time data transmission and processing capabilities, making it difficult to achieve real-time monitoring and management of flow data. Once an abnormal situation occurs, it cannot be warned and handled in time. The inability to monitor and manage in real time will lead to the passivity of liquefied oil flow management, and it will be impossible to respond and handle emergencies in a timely manner, increasing safety hazards and management risks.

[0005] Traditional methods rely on wired transmission, which is complex and costly, and difficult to maintain. Once a line failure occurs, data transmission will be interrupted, seriously affecting the continuity and reliability of flow monitoring. Even if some systems use wireless transmission technology, the stability and real-time performance of data transmission are still difficult to guarantee due to the limitations of transmission protocols and equipment.

[0006] Traditional methods lack intelligent analysis and processing capabilities, and are unable to conduct in-depth analysis and trend prediction of flow data, resulting in a lack of scientific basis for liquefied oil transportation and storage strategies, and serious waste of resources. Systems lacking intelligent analysis cannot provide effective decision support, limiting the improvement of management efficiency. The functions of various parts of traditional flow management systems are scattered and lack system integration. The lack of collaborative working capabilities between different functional modules leads to serious information islands and low overall system efficiency. The low system integration makes it difficult for managers to obtain comprehensive and accurate flow management information, affecting the timeliness and accuracy of decision-making.

[0007] Therefore, how to provide a liquefied oil flow management method based on Internet of Things technology is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0008] One purpose of the present invention is to propose a liquefied oil product flow management method based on the Internet of Things technology. The present invention installs high-precision flow meters at key nodes of liquefied oil product pipelines, uses multi-source domain adversarial transfer learning algorithms to perform local real-time processing and analysis of real-time and historical data, and transmits the processed data to the Internet of Things platform through an adaptive multi-hop wireless sensor network. The flow data is stored and processed through spatiotemporal autocorrelation analysis, spatiotemporal patterns and abnormal changes are identified, and real-time monitoring and predictive control are performed in combination with a hybrid event-triggered predictive control algorithm. Finally, a multivariate time series decomposition and reconstruction algorithm is used to deeply analyze the flow data, and a multi-core learning algorithm is used to optimize the flow prediction model, extract key features, identify abnormal changes, and optimize the transportation and storage strategy of liquefied oil products.

[0009] A liquefied oil product flow management method based on Internet of Things technology according to an embodiment of the present invention includes the following steps: 1. A liquefied oil product flow management method based on Internet of Things technology, characterized in that it includes the following steps: S1. Install high-precision flow meters at key nodes of liquefied oil pipelines to monitor the flow data of liquefied oil in real time and collect historical flow data; S2, using multi-source domain adversarial transfer learning algorithm to perform local real-time processing and preliminary analysis of real-time traffic data and historical traffic data; S3, the processed data is transmitted to the IoT platform through an adaptive multi-hop wireless sensor network. The network can dynamically adjust the transmission path according to the real-time environment and data flow, and use Search algorithms predict the best transmission routes; S4. Use spatiotemporal autocorrelation analysis to store and process the traffic data transmitted to the IoT platform, and identify spatiotemporal patterns and abnormal changes in the traffic data by combining historical traffic data through spatiotemporal autocorrelation analysis; S5. Use a hybrid event-triggered predictive control algorithm to monitor and predict traffic data in real time, combine the event trigger mechanism to collect and process data when a specific event occurs, and use the predictive control model in combination with historical data to predict and optimize traffic data in real time; S6. Use multivariate time series decomposition and reconstruction algorithms to conduct in-depth analysis of real-time and historical traffic data, extract key features of traffic data, identify abnormal changes, and optimize traffic prediction models; S7. Based on the analysis results and trend prediction of historical flow data, sparse representation and dictionary learning optimization algorithm are used to learn and analyze historical flow data, extract key features and patterns, optimize the transportation and storage strategies of liquefied oil products, and achieve the global optimal solution.

[0010] Optionally, the S2 specifically includes: S21. Collect data from multiple source domains, where the source domain data includes real-time traffic data and historical traffic data; S22, preprocessing the collected source domain data, including data cleaning, data normalization and data enhancement; S23. Use the adaptive weighting mechanism to dynamically adjust the weight of each source domain: ; in, For the The weight of the source domain, To dynamically adjust parameters, For the Source domain data Target domain data The alignment between is the total number of source domains; S24. Build adversarial transfer learning models, including generators and the discriminator , generator Used to convert source domain data Convert to target domain data The representation of the discriminator Used to distinguish source domain data from target domain data; S25. Introduce sparse regularization and multiple gradient penalty terms in the adversarial training process to prevent model overfitting: ; in, is the total loss, To combat losses, For the Layer gradient penalty coefficient, is the sparse regularization coefficient, Sparsity constraints for the generator output; S26. Align the source domain features with the target domain features through a multi-layer adaptive alignment layer. The parameters of the multi-layer adaptive alignment layer are dynamically adjusted according to the real-time data and historical data of the target domain: ; in, For the adaptive alignment layer in time Parameters, is the learning rate, For the Alignment loss function for source domain data; S27. Use hybrid discriminator to combine global features and local features To determine: ; in, To determine the results, is the activation function, and is the weight matrix, is the bias term; S28. Output the processed real-time traffic data and historical traffic data.

[0011] Optionally, the S3 specifically includes: S31, transmitting the processed real-time traffic data and historical traffic data to the Internet of Things platform through an adaptive multi-hop wireless sensor network; S32. Implement a dynamic transmission path optimization algorithm to dynamically adjust the transmission path according to the real-time environment and data traffic: ; in, is the optimal transmission path, For path The transmission energy consumption on is the energy consumption weight, For path The signal strength on For path The transmission delay on is the delay weight, For path The packet loss rate on and are the number of nodes and the number of transmission segments on the path, respectively; S33. Utilization Search algorithms predict the best transmission routes, The search algorithm evaluates the function Calculate the optimal path: ; ; in, For Node The evaluation function value of From the start point to the node The actual cost of For Node The estimated cost to reach the destination, For Node The transmission energy consumption, is the transmission energy consumption weight, For Node The link quality, is the time weight, For Node timestamp, is the cost weight, For Node The transmission cost, and are the number of nodes on the path respectively; S34. During data transmission, monitor the transmission status of the network and detect the packet loss rate and transmission delay: ; in, For transmission efficiency, For path The number of packets successfully transmitted. For path The total number of packets on For path The transmission delay time on For path Total transfer time on ; S35: If it is detected that the transmission efficiency is lower than the preset threshold, dynamically adjust the transmission path and recalculate the optimal path , and reconfigure network parameters.

[0012] Optionally, the S4 specifically includes: S41, storing and processing the traffic data transmitted to the IoT platform; S42. Processing flow data using spatiotemporal autocorrelation analysis: ; in, For the adjusted index, is the sample size, For the The observed value of a sample, is the mean of the observed values, For the The samples and The spatial weights between samples, is the sum of spatial weights, ; S43, according to the adjusted index , combined with historical traffic data, identify spatiotemporal patterns and abnormal changes in traffic data: ; in, is the pattern matrix, For the The spatiotemporal characteristic values ​​of samples, is the corresponding eigenvector, represents eigenvalue decomposition; S44, according to the pattern matrix Calculate the results and identify abnormal data points; S45. Output the results of spatiotemporal autocorrelation analysis and abnormal data identification results.

[0013] Optionally, the S5 specifically includes: S51, using the results of spatiotemporal autocorrelation analysis and abnormal data identification results, using a hybrid event-triggered predictive control algorithm to perform real-time monitoring and predictive control of traffic data; S52. Combine the event trigger mechanism to collect and process data when a specific event occurs: ; in, To trigger the event set, For event set The events, Trigger thresholds for events; S53. Use the predictive control model to predict and optimize the flow data in real time: ; in, For time The control input, For the predicted state, is the state weight matrix, is the control input weight matrix, is the terminal state weight matrix, is the length of the prediction time domain; S54, dynamically updating parameters of the predictive control model and adjusting the model according to real-time data; S55: Output real-time predicted flow data and optimized control input.

[0014] Optionally, the S6 specifically includes: S61, using the output real-time predicted traffic data and optimized control input, using a multivariable time series decomposition and reconstruction algorithm to perform in-depth analysis on the real-time traffic data and historical traffic data; S62, decomposing the flow data into multiple time series components, extracting features from each time series component, and identifying key features of the flow data; S63, inputting the extracted features into a nonlinear prediction model to optimize the flow prediction model; S64. Combining the identified spatiotemporal patterns and abnormal data points, the multi-core learning algorithm is used to optimize the prediction model: ; in, is the output prediction value, For the Kernel function, used to measure the feature vector and The similarity between For the The weight of the kernel function, is the number of kernel functions; S65. Output the results of the in-depth analysis, including the identified key features, the optimized traffic prediction model, abnormal changes, and control suggestions.

[0015] Optionally, the S7 specifically includes: S71. Optimize the transportation and storage strategies of liquefied petroleum products based on the analysis results and trend prediction of historical flow data; S72, using sparse representation and dictionary learning optimization algorithm to extract key features and patterns by learning and analyzing historical traffic data; S73. Optimize the dictionary matrix and the sparse representation coefficient matrix by minimizing the reconstruction error and the sparsity constraint: ; in, The reconstruction error norm, is the mixed norm of the sparsity constraint, is the regularization parameter, is the historical traffic data matrix, is the dictionary matrix, is the sparse representation coefficient matrix; S74. Using the optimized dictionary matrix and sparse representation coefficient matrix, identify key features and patterns of traffic data: ; in, is the identified feature matrix, For data points The probability density function of To adjust the parameters, is the mixed norm of the feature matrix; S75. Based on the identified key features and patterns and combined with the in-depth analysis results, optimize the transportation and storage management strategy of liquefied petroleum products.

[0016] The beneficial effects of the present invention are: The present invention provides a liquefied oil product flow management method based on the Internet of Things technology, which solves the problems of untimely data acquisition, large measurement errors, frequent manual intervention, and inability to monitor and manage in real time in the prior art, and realizes efficient, accurate and intelligent management of the liquefied oil product flow. By installing high-precision flow meters at key nodes of the liquefied oil product pipeline, the present invention can monitor flow data in real time and collect detailed historical flow data. This high-precision flow meter effectively overcomes the influence of environmental factors, ensures the stability and accuracy of the data, and provides a reliable basis for subsequent data analysis.

[0017] The present invention uses a multi-source domain adversarial transfer learning algorithm to perform local real-time processing and analysis of real-time and historical data, reducing delays and errors in data transmission and improving the timeliness and accuracy of data processing. The processed data is transmitted to the Internet of Things platform through an adaptive multi-hop wireless sensor network. The adaptive network dynamically adjusts the transmission path according to the real-time environment and data flow, and uses The search algorithm predicts the best transmission route to ensure the stability and real-time performance of data transmission. The flow data transmitted to the IoT platform is stored and processed through spatiotemporal autocorrelation analysis to identify spatiotemporal patterns and abnormal changes in the flow data. This analysis method can reveal the inherent laws of flow changes and provide a scientific basis for further optimizing management strategies. Combined with the hybrid event-triggered predictive control algorithm for real-time monitoring and predictive control, the control strategy is adjusted dynamically to improve the flexibility and response speed of flow management.

[0018] The present invention uses a multivariate time series decomposition and reconstruction algorithm to deeply analyze the flow data, extract key features, identify abnormal changes, and optimize the flow prediction model in combination with a multi-core learning algorithm to further improve the transportation and storage efficiency of liquefied oil products. Based on the analysis results and trend prediction of historical data, a sparse representation and dictionary learning optimization algorithm is used to extract key features and patterns through learning the sparse representation of flow data, optimize the transportation and storage strategy of liquefied oil products, and achieve a global optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is an overall flow chart of a liquefied oil product flow management method based on Internet of Things technology proposed by the present invention. DETAILED DESCRIPTION

[0020] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0021] refer to Figure 1 , a liquefied oil product flow management method based on Internet of Things technology, comprising the following steps: 1. A liquefied oil product flow management method based on Internet of Things technology, characterized in that it includes the following steps: S1. Install high-precision flow meters at key nodes of liquefied oil pipelines to monitor the flow data of liquefied oil in real time and collect historical flow data; S2, using multi-source domain adversarial transfer learning algorithm to perform local real-time processing and preliminary analysis of real-time traffic data and historical traffic data; S3, the processed data is transmitted to the IoT platform through an adaptive multi-hop wireless sensor network. The network can dynamically adjust the transmission path according to the real-time environment and data flow, and use Search algorithms predict the best transmission routes; S4. Use spatiotemporal autocorrelation analysis to store and process the traffic data transmitted to the IoT platform, and identify spatiotemporal patterns and abnormal changes in the traffic data by combining historical traffic data through spatiotemporal autocorrelation analysis; S5. Use a hybrid event-triggered predictive control algorithm to monitor and predict traffic data in real time, combine the event trigger mechanism to collect and process data when a specific event occurs, and use the predictive control model in combination with historical data to predict and optimize traffic data in real time; S6. Use multivariate time series decomposition and reconstruction algorithms to conduct in-depth analysis of real-time and historical traffic data, extract key features of traffic data, identify abnormal changes, and optimize traffic prediction models; S7. Based on the analysis results and trend prediction of historical flow data, sparse representation and dictionary learning optimization algorithm are used to learn and analyze historical flow data, extract key features and patterns, optimize the transportation and storage strategies of liquefied oil products, and achieve the global optimal solution.

[0022] In this implementation, S2 specifically includes: S21. Collect data from multiple source domains, where the source domain data includes real-time traffic data and historical traffic data; S22, preprocessing the collected source domain data, including data cleaning, data normalization and data enhancement; S23. Use the adaptive weighting mechanism to dynamically adjust the weight of each source domain: ; in, For the The weight of the source domain, To dynamically adjust parameters, For the Source domain data Target domain data The alignment between is the total number of source domains; S24. Build adversarial transfer learning models, including generators and the discriminator , generator Used to convert source domain data Convert to target domain data The representation of the discriminator Used to distinguish source domain data from target domain data; S25. Introduce sparse regularization and multiple gradient penalty terms in the adversarial training process to prevent model overfitting: ; in, is the total loss, To combat losses, For the Layer gradient penalty coefficient, is the sparse regularization coefficient, Sparsity constraints for the generator output; S26. Align the source domain features with the target domain features through a multi-layer adaptive alignment layer. The parameters of the multi-layer adaptive alignment layer are dynamically adjusted according to the real-time data and historical data of the target domain: ; in, For the adaptive alignment layer in time Parameters, is the learning rate, For the Alignment loss function for source domain data; S27. Use hybrid discriminator to combine global features and local features To determine: ; in, To determine the results, is the activation function, and is the weight matrix, is the bias term; S28. Output the processed real-time traffic data and historical traffic data.

[0023] In this implementation, S3 specifically includes: S31, transmitting the processed real-time traffic data and historical traffic data to the Internet of Things platform through an adaptive multi-hop wireless sensor network; S32. Implement a dynamic transmission path optimization algorithm to dynamically adjust the transmission path according to the real-time environment and data traffic: ; in, is the optimal transmission path, For path The transmission energy consumption on is the energy consumption weight, For path The signal strength on For path The transmission delay on is the delay weight, For path The packet loss rate on and are the number of nodes and the number of transmission segments on the path, respectively; S33. Utilization Search algorithms predict the best transmission routes, The search algorithm evaluates the function Calculate the optimal path: ; ; in, For Node The evaluation function value of From the start point to the node The actual cost of For Node The estimated cost to reach the destination, For Node The transmission energy consumption, is the transmission energy consumption weight, For Node The link quality, is the time weight, For Node timestamp, is the cost weight, For Node The transmission cost, and are the number of nodes on the path respectively; S34. During data transmission, monitor the transmission status of the network and detect the packet loss rate and transmission delay: ; in, For transmission efficiency, For path The number of packets successfully transmitted. For path The total number of packets on For path The transmission delay time on For path Total transfer time on ; S35: If it is detected that the transmission efficiency is lower than the preset threshold, dynamically adjust the transmission path and recalculate the optimal path , and reconfigure network parameters.

[0024] In this implementation manner, the S4 specifically includes: S41, storing and processing the traffic data transmitted to the IoT platform; S42. Processing flow data using spatiotemporal autocorrelation analysis: ; in, For the adjusted index, is the sample size, For the The observed value of a sample, is the mean of the observed values, For the The samples and The spatial weights between samples, is the sum of spatial weights, ; S43, according to the adjusted index , combined with historical traffic data, identify spatiotemporal patterns and abnormal changes in traffic data: ; in, is the pattern matrix, For the The spatiotemporal characteristic values ​​of samples, is the corresponding eigenvector, represents eigenvalue decomposition; S44, according to the pattern matrix Calculate the results and identify abnormal data points; S45. Output the results of spatiotemporal autocorrelation analysis and abnormal data identification results.

[0025] In this implementation manner, S5 specifically includes: S51, using the results of spatiotemporal autocorrelation analysis and abnormal data identification results, using a hybrid event-triggered predictive control algorithm to perform real-time monitoring and predictive control of traffic data; S52. Combine the event trigger mechanism to collect and process data when a specific event occurs: ; in, To trigger the event set, For event set The events, Trigger thresholds for events; S53. Use the predictive control model to predict and optimize the flow data in real time: ; in, For time The control input, For the predicted state, is the state weight matrix, is the control input weight matrix, is the terminal state weight matrix, is the length of the prediction time domain; S54, dynamically updating parameters of the predictive control model and adjusting the model according to real-time data; S55: Output real-time predicted flow data and optimized control input.

[0026] In this implementation manner, S6 specifically includes: S61, using the output real-time predicted traffic data and optimized control input, using a multivariable time series decomposition and reconstruction algorithm to perform in-depth analysis on the real-time traffic data and historical traffic data; S62, decomposing the flow data into multiple time series components, extracting features from each time series component, and identifying key features of the flow data; S63, inputting the extracted features into a nonlinear prediction model to optimize the flow prediction model; S64. Combining the identified spatiotemporal patterns and abnormal data points, the multi-core learning algorithm is used to optimize the prediction model: ; in, is the output prediction value, For the Kernel function, used to measure the feature vector and The similarity between For the The weight of the kernel function, is the number of kernel functions; S65. Output the results of the in-depth analysis, including the identified key features, the optimized traffic prediction model, abnormal changes, and control suggestions.

[0027] In this implementation manner, the S7 specifically includes: S71. Optimize the transportation and storage strategies of liquefied petroleum products based on the analysis results and trend prediction of historical flow data; S72, using sparse representation and dictionary learning optimization algorithm to extract key features and patterns by learning and analyzing historical traffic data; S73. Optimize the dictionary matrix and the sparse representation coefficient matrix by minimizing the reconstruction error and the sparsity constraint: ; in, The reconstruction error norm, is the mixed norm of the sparsity constraint, is the regularization parameter, is the historical traffic data matrix, is the dictionary matrix, is the sparse representation coefficient matrix; S74. Using the optimized dictionary matrix and sparse representation coefficient matrix, identify key features and patterns of traffic data: ; in, is the identified feature matrix, For data points The probability density function of To adjust the parameters, is the mixed norm of the feature matrix; S75. Based on the identified key features and patterns and combined with the in-depth analysis results, optimize the transportation and storage management strategy of liquefied petroleum products.

[0028] Embodiment 1: In order to verify the feasibility of the present invention in implementation, the present invention is applied to the transportation and storage of liquefied oil products in City A, where the flow needs to be accurately managed and monitored in real time. Traditional flow management methods cannot meet existing needs, and there are problems such as untimely data acquisition, large measurement errors, frequent manual intervention, and inability to monitor and manage in real time. Therefore, a liquefied oil flow management method based on Internet of Things technology is introduced. The method aims to achieve efficient, accurate and intelligent management of liquefied oil flow through advanced sensor technology, data processing algorithms and wireless transmission networks.

[0029] During the transportation and storage of liquefied oil products in City A, high-precision flow meters are first installed at key nodes of the pipeline. These flow meters can monitor the flow data of liquefied oil products in real time and collect historical flow data simultaneously. The data of the flow meters are processed and analyzed locally in real time through a multi-source domain adversarial transfer learning algorithm to ensure the accuracy and timeliness of the data. The processed data is transmitted to the IoT platform through an adaptive multi-hop wireless sensor network. The adaptive network dynamically adjusts the transmission path according to the real-time environment and data flow, and uses The search algorithm predicts the best transmission route to ensure the stability and real-time performance of data transmission. The data transmitted to the IoT platform is stored and processed through spatiotemporal autocorrelation analysis to identify spatiotemporal patterns and abnormal changes in flow data. These spatiotemporal patterns and abnormal changes provide a scientific basis for further optimizing the transportation and storage strategies of liquefied oil products.

[0030] Combined with the hybrid event-triggered predictive control algorithm, the system monitors and predicts the flow data in real time. The predictive control model improves the flexibility and response speed of flow management by dynamically adjusting the control strategy. Finally, the multivariable time series decomposition and reconstruction algorithm is used to deeply analyze the flow data, extract key features, identify abnormal changes, and optimize the flow prediction model in combination with the multi-core learning algorithm. Based on the analysis results and trend prediction of historical data, the sparse representation and dictionary learning optimization algorithm are used to optimize the transportation and storage strategy of liquefied oil products and achieve the global optimal solution.

[0031] From May 2023 to October 2023, City A implemented a liquefied petroleum product flow management method based on the Internet of Things technology. In terms of data acquisition and real-time monitoring, high-precision flow meters installed at key nodes of the liquefied petroleum product pipeline achieved a data acquisition frequency of once per second, and the real-time monitoring data enabled operators to grasp the flow changes in a timely manner. During this period, more than 1 million data were monitored in total, and the timeliness and accuracy of data acquisition were significantly improved. In terms of measurement error and accuracy, the measurement error of the flow meter was reduced from the original ±1.5% to ±0.5% through data processing using a multi-source domain adversarial transfer learning algorithm. During a large-scale transportation process in June 2023, the total flow detected by the system was 500,000 cubic meters, which was only 2,500 cubic meters different from the actual measurement value, with a relative error of 0.5%, greatly improving the accuracy of measurement.

[0032] In terms of data transmission stability and real-time performance, adaptive multi-hop wireless sensor networks effectively solve the problem of data transmission stability. The search algorithm dynamically adjusts the transmission path, reducing the data transmission delay from an average of 10 seconds to 1 second, and the packet loss rate to 0.1%. In a test in July, the average data delay from the flow meter to the IoT platform was 0.9 seconds, and the stability and real-time performance of the transmission were guaranteed. In terms of spatiotemporal data analysis and anomaly detection, the system identified multiple spatiotemporal patterns in the traffic data and detected multiple abnormal traffic events through spatiotemporal autocorrelation analysis. In a data analysis in August, the system successfully identified 5 abnormal traffic events and issued warnings in a timely manner. Among these abnormal events, 3 were caused by equipment failures, and timely warnings avoided potential safety hazards.

[0033] In terms of optimized flow management strategies, the flow prediction model was optimized using multivariable time series decomposition and reconstruction algorithms and multi-core learning algorithms. During the entire project implementation period, these optimization strategies increased the transportation and storage efficiency of liquefied oil products by 20%. For example, in a large-scale transportation mission in September, the optimized model and strategy shortened the transportation time from the original 48 hours to 40 hours, saving 8 hours of transportation time and improving transportation efficiency.

[0034] In terms of economic benefits and safety, the sparse representation and dictionary learning optimization algorithm was used to further optimize the transportation and storage strategies of liquefied oil products and achieve the global optimal solution. Through this system, the loss of liquefied oil products during transportation was reduced and costs were saved. Within six months of the project implementation, the cumulative cost savings were approximately RMB 5 million, and the safety and reliability of the system were significantly improved.

[0035] Table 1 Data acquisition, real-time monitoring and measurement accuracy

[0036] Table 2 Stability and real-time performance of data transmission, spatiotemporal data analysis, and anomaly detection

[0037] Table 3 Traffic management strategy optimization, economic benefits and security

[0038] From the above data, it can be seen that from May 2023 to October 2023, City A implemented a liquefied oil product flow management method based on the Internet of Things technology and achieved remarkable results. In terms of data acquisition and real-time monitoring, the high-precision flow meters installed at key nodes of the liquefied oil product pipeline achieved a data collection frequency of once per second, monitored more than 1 million data, and significantly improved the timeliness and accuracy of data acquisition. The measurement error was reduced from the original ±1.5% to ±0.5%. During a large-scale transportation process, the total flow detected was 500,000 cubic meters, which was only 2,500 cubic meters different from the actual measurement value, with a relative error of 0.5%. In terms of the stability and real-time performance of data transmission, the adaptive multi-hop wireless sensor network effectively solved the problem of data transmission stability. By dynamically adjusting the transmission path, the delay time of data transmission was reduced from the original average of 10 seconds to 1 second, and the packet loss rate was reduced to 0.1%. In a test in July, the average delay time of data transmitted from the flow meter to the Internet of Things platform was 0.9 seconds, ensuring the stability and real-time performance of data transmission.

[0039] In terms of spatiotemporal data analysis and anomaly detection, the system identified multiple spatiotemporal patterns in traffic data and detected multiple abnormal traffic events through spatiotemporal autocorrelation analysis. In a data analysis in August, the system successfully identified 5 abnormal traffic events and issued warnings in a timely manner, avoiding 3 potential safety hazards caused by equipment failures. In terms of optimizing traffic management strategies, the multivariate time series decomposition and reconstruction algorithm and the multi-core learning algorithm were used to optimize the traffic prediction model and improve the transportation and storage efficiency of liquefied oil products. In a large-scale transportation mission in September, the optimized model and strategy were used to shorten the transportation time from the original 48 hours to 40 hours, saving 8 hours of transportation time and improving transportation efficiency by 20%.

[0040] In terms of economic benefits and safety, the sparse representation and dictionary learning optimization algorithm was used to optimize the transportation and storage strategy of liquefied oil products, achieve the global optimal solution, reduce losses during transportation, and save costs. Within six months of the project implementation, the cumulative cost savings were approximately RMB 5 million, and the safety and reliability of the system were significantly improved.

[0041] By applying the liquefied oil flow management method based on the Internet of Things technology of the present invention in the transportation and storage process of liquefied oil products in City A, many problems in traditional flow management methods are solved, real-time data acquisition, high-precision measurement, automated management, high-stability transmission and intelligent analysis are achieved, and the efficiency and reliability of liquefied oil flow management are greatly improved.

[0042] The present invention realizes real-time monitoring, automated management, data fusion and analysis, abnormal warning and system integration of flow data, effectively solving many defects in the prior art. Through these technical means and innovations, the present invention significantly improves the efficiency and reliability of liquefied oil flow management, ensures the safety and economy of liquefied oil during transportation and storage, and has broad application prospects and practical significance.

[0043] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The following steps are involved: S1. Install high-precision flow meters at key nodes of liquefied oil pipelines to monitor the flow data of liquefied oil in real time and collect historical flow data; S2, using multi-source domain adversarial transfer learning algorithm to perform local real-time processing and preliminary analysis of real-time traffic data and historical traffic data; S3, the processed data is transmitted to the IoT platform through an adaptive multi-hop wireless sensor network. The network can dynamically adjust the transmission path according to the real-time environment and data flow, and use Search algorithms predict the best transmission routes; S4. Use spatiotemporal autocorrelation analysis to store and process the traffic data transmitted to the IoT platform, and identify spatiotemporal patterns and abnormal changes in the traffic data by combining historical traffic data through spatiotemporal autocorrelation analysis; S5. Use a hybrid event-triggered predictive control algorithm to monitor and predict traffic data in real time, combine the event trigger mechanism to collect and process data when a specific event occurs, and use the predictive control model in combination with historical data to predict and optimize traffic data in real time; S6. Use multivariate time series decomposition and reconstruction algorithms to conduct in-depth analysis of real-time and historical traffic data, extract key features of traffic data, identify abnormal changes, and optimize traffic prediction models; S7. Based on the analysis results and trend prediction of historical flow data, sparse representation and dictionary learning optimization algorithm are used to learn and analyze historical flow data, extract key features and patterns, optimize the transportation and storage strategies of liquefied oil products, and achieve the global optimal solution.

2. According to claim 1, a liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The S2 specifically includes: S21. Collect data from multiple source domains, where the source domain data includes real-time traffic data and historical traffic data; S22, preprocessing the collected source domain data, including data cleaning, data normalization and data enhancement; S23. Use the adaptive weighting mechanism to dynamically adjust the weight of each source domain: ； in, For the The weight of the source domain, To dynamically adjust parameters, For the Source domain data Target domain data The alignment between is the total number of source domains; S24. Build adversarial transfer learning models, including generators and the discriminator , generator Used to convert source domain data Convert to target domain data The representation of the discriminator Used to distinguish source domain data from target domain data; S25. Introduce sparse regularization and multiple gradient penalty terms in the adversarial training process to prevent model overfitting: ； in, is the total loss, To combat losses, For the Layer gradient penalty coefficient, is the sparse regularization coefficient, Sparsity constraints for the generator output; S26. Align the source domain features with the target domain features through a multi-layer adaptive alignment layer. The parameters of the multi-layer adaptive alignment layer are dynamically adjusted according to the real-time data and historical data of the target domain: ； in, For the adaptive alignment layer in time Parameters, is the learning rate, For the Alignment loss function for source domain data; S27. Use hybrid discriminator to combine global features and local features To determine: ； in, To determine the results, is the activation function, and is the weight matrix, is the bias term; S28. Output the processed real-time traffic data and historical traffic data.

3. According to the method for managing the flow of liquefied oil products based on the Internet of Things technology as described in claim 1, It is characterized in that The S3 specifically includes: S31, transmitting the processed real-time traffic data and historical traffic data to the Internet of Things platform through an adaptive multi-hop wireless sensor network; S32. Implement a dynamic transmission path optimization algorithm to dynamically adjust the transmission path according to the real-time environment and data traffic: ； in, is the optimal transmission path, For path The transmission energy consumption on is the energy consumption weight, For path The signal strength on For path The transmission delay on is the delay weight, For path The packet loss rate on and are the number of nodes and the number of transmission segments on the path, respectively; S33. Utilization Search algorithms predict the best transmission routes, The search algorithm evaluates the function Calculate the optimal path: ； ； in, For Node The evaluation function value of From the start point to the node The actual cost, For Node The estimated cost to reach the destination, For Node The transmission energy consumption, is the transmission energy consumption weight, For Node The link quality, is the time weight, For Node timestamp, is the cost weight, For Node The transmission cost, and are the number of nodes on the path respectively; S34. During data transmission, monitor the transmission status of the network and detect the packet loss rate and transmission delay: ； in, For transmission efficiency, For path The number of packets successfully transmitted. For path The total number of packets on For path The transmission delay time on For path Total transfer time on ; S35: If it is detected that the transmission efficiency is lower than the preset threshold, dynamically adjust the transmission path and recalculate the optimal path , and reconfigure network parameters.

4. According to claim 1, a liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The S4 specifically includes: S41, storing and processing the traffic data transmitted to the IoT platform; S42. Processing flow data using spatiotemporal autocorrelation analysis: ； in, For the adjusted index, is the sample size, For the The observed value of a sample, is the mean of the observed values, For the The samples and The spatial weight between samples, is the sum of spatial weights, ; S43, according to the adjusted index , combined with historical traffic data, identify spatiotemporal patterns and abnormal changes in traffic data: ； in, is the pattern matrix, For the The spatiotemporal characteristic values ​​of samples, is the corresponding eigenvector, represents eigenvalue decomposition; S44, according to the pattern matrix Calculate the results and identify abnormal data points; S45. Output the results of spatiotemporal autocorrelation analysis and abnormal data identification results.

5. According to claim 1, a liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The S5 specifically includes: S51, using the results of spatiotemporal autocorrelation analysis and abnormal data identification results, using a hybrid event-triggered predictive control algorithm to perform real-time monitoring and predictive control of traffic data; S52. Combine the event trigger mechanism to collect and process data when a specific event occurs: ； in, To trigger the event set, For event set The events, Trigger thresholds for events; S53. Use the predictive control model to predict and optimize the flow data in real time: ； in, For time The control input, For the predicted state, is the state weight matrix, is the control input weight matrix, is the terminal state weight matrix, is the length of the prediction time domain; S54, dynamically updating the parameters of the predictive control model and adjusting the model according to the real-time data; S55: Output real-time predicted flow data and optimized control input.

6. According to claim 1, a liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The S6 specifically includes: S61, using the output real-time predicted traffic data and optimized control input, using a multivariable time series decomposition and reconstruction algorithm to perform in-depth analysis on the real-time traffic data and historical traffic data; S62, decomposing the flow data into multiple time series components, extracting features from each time series component, and identifying key features of the flow data; S63, inputting the extracted features into a nonlinear prediction model to optimize the flow prediction model; S64. Combining the identified spatiotemporal patterns and abnormal data points, the multi-core learning algorithm is used to optimize the prediction model: ； in, is the output prediction value, For the Kernel function, used to measure the feature vector and The similarity between For the The weight of the kernel function, is the number of kernel functions; S65. Output the results of the in-depth analysis, including the identified key features, the optimized traffic prediction model, abnormal changes, and control suggestions.

7. According to claim 1, a liquefied oil product flow management method based on Internet of Things technology, It is characterized in that The S7 specifically includes: S71. Optimize the transportation and storage strategies of liquefied petroleum products based on the analysis results and trend prediction of historical flow data; S72, using sparse representation and dictionary learning optimization algorithm to extract key features and patterns by learning and analyzing historical traffic data; S73. Optimize the dictionary matrix and the sparse representation coefficient matrix by minimizing the reconstruction error and the sparsity constraint: ； in, The reconstruction error norm, is the mixed norm of the sparsity constraint, is the regularization parameter, is the historical traffic data matrix, is the dictionary matrix, is the sparse representation coefficient matrix; S74. Using the optimized dictionary matrix and sparse representation coefficient matrix, identify key features and patterns of traffic data: ； in, is the identified feature matrix, For data points The probability density function of To adjust the parameters, is the mixed norm of the feature matrix; S75. Based on the identified key features and patterns and combined with the in-depth analysis results, optimize the transportation and storage management strategy of liquefied petroleum products.

Images