Dangerous chemical cargo transportation data sharing method and system

Abstract

The invention discloses a hazardous chemical substance cargo transportation data sharing method and system, and relates to the technical field of data management, and the method comprises the steps: obtaining hazardous chemical substance cargo transportation data, extracting cargo transportation feature data, and generating cargo transportation sharing data; performing anomaly detection on the cargo transportation shared data by using an anomaly detection algorithm to obtain safe cargo transportation shared data; the safe cargo transportation shared data are stored in a data sharing platform in a distributed mode, the stored cargo transportation shared data are dynamically deployed through a data scheduling algorithm, and data sharing access authority is distributed. According to the method, the abnormal condition in the data can be found in time through the abnormal detection algorithm, the accuracy and the safety of the transportation data are ensured, the storage of the transportation data is more flexible and efficient, and different related parties are ensured to share necessary data under a compliant framework by reasonably distributing the data sharing access authority.

Description

Technical Field

[0001] The present invention relates to the field of data management technology, and in particular to a method and system for sharing hazardous chemicals cargo transportation data. Background Art

[0002] With the rapid development of the chemical industry, the demand for hazardous chemical transportation has increased year by year, and the scale of hazardous chemical transportation has continued to expand. Currently, the hazardous chemical transportation industry exhibits several characteristics: widely distributed transportation companies, long transportation routes, and frequent cross-regional transportation, which makes transportation management and supervision extremely complex. This is especially true during transportation, as the variety of goods involved is diverse, transportation conditions are stringent, and transportation routes often pass through densely populated areas or areas with rapid economic development. However, the current hazardous chemical transportation industry faces problems such as weak rescue capabilities, inadequate supervision of transportation vehicles, and a lack of effective supervision and inspection methods, which have led to numerous transportation accidents. Furthermore, due to the severe phenomenon of data silos and the low level of information sharing and collaboration between different transportation companies, emergency resources cannot be quickly and accurately dispatched in the event of an emergency.

[0003] Existing technologies lack effective sharing mechanisms, leading to severe information silos between different transportation companies and regulatory authorities, hindering real-time data exchange and collaborative management. This prevents stakeholders from quickly obtaining critical information in the event of an emergency, resulting in slow response times. Furthermore, the lack of effective real-time monitoring and detection methods for abnormalities during transportation makes it difficult to promptly identify potential data security risks, creating certain security risks.

[0004] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention proposes a method and system for sharing data on the transportation of hazardous chemicals, which solves the problem mentioned in the above background technology that the existing method lacks an effective sharing mechanism, resulting in serious information islands between different transportation companies and regulatory departments, and the inability to achieve real-time data exchange and collaborative management. This makes it impossible for relevant parties to quickly obtain key information in the event of an emergency, and the response speed is slow. Secondly, there is a lack of effective real-time monitoring and detection methods for abnormal situations during transportation, making it difficult to promptly discover potential data security risks, resulting in certain security risks.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] According to one aspect of the present invention, a method for sharing hazardous chemicals transportation data is provided, comprising:

[0008] Obtain hazardous chemical cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data;

[0009] Use anomaly detection algorithms to detect anomalies in cargo transportation shared data and obtain safe cargo transportation shared data;

[0010] Distribute and store secure cargo transportation shared data on a data sharing platform, dynamically allocate the stored cargo transportation shared data through a data scheduling algorithm, and assign data sharing access rights;

[0011] Among them, the anomaly detection algorithm is used to detect anomalies in the cargo transportation shared data, and the safe cargo transportation shared data obtained includes:

[0012] Obtain shared data samples of cargo transportation, classify data of different abnormal types, divide the data of different abnormal types into training sets and test sets, and establish an anomaly detection model;

[0013] Optimizing the anomaly detection model through learning algorithms and inputting the test set into the optimized anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data; including:

[0014] Initialize the population of anomaly detection models, calculate the fitness of each anomaly detection model, sort them, and optimize the population based on the Cauchy mutation strategy and reverse learning strategy;

[0015] According to the fitness value, it is judged whether the preset conditions are met, and then the global search phase is entered; otherwise, the local search phase is carried out;

[0016] If the maximum number of iterations is reached, the optimal anomaly detection model is output, and the test set is input into the optimal anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data.

[0017] Furthermore, the hazardous chemicals transportation data is obtained, the cargo transportation characteristic data is extracted, and the cargo transportation shared data is generated, including:

[0018] Obtain raw data on hazardous chemical cargo transportation, clean and process the raw data, and remove noise and outliers;

[0019] By constructing data sequences and training samples, selecting embedding dimensions, and extracting cargo transportation feature data;

[0020] The least squares support vector machine is used to analyze the extreme points and trends in the freight transportation characteristic data, calculate the core indicators of the transportation characteristics, and generate the transportation characteristic analysis results; and based on the analysis results, generate freight transportation shared data.

[0021] Furthermore, the formulas for optimizing the population according to the Cauchy mutation strategy and the reverse learning strategy are:

[0022] K new =K better (t)+Cauchy(m0)·K better (t)

[0023] K i (t+1)=ub+δ·(lb-K i (t));

[0024] Where K new represents the parameter value of the anomaly detection model after optimization by the Cauchy mutation strategy; K better (t) represents the parameter value of the anomaly detection model with the highest fitness at the current iteration t;

[0025] Cauchy(m0) represents the Cauchy distribution function; K i (t+1) represents the parameter value of the anomaly detection model after the i-th anomaly detection model is optimized by the reverse learning strategy at iteration t+1; K i (t) represents the parameter value of the i-th anomaly detection model at iteration t; ub represents the maximum allowable value of the anomaly detection model parameter; lb represents the minimum allowable value of the anomaly detection model parameter; δ represents a random number.

[0026] Furthermore, the fitness value is used to determine whether the preset conditions are met, and then the global search phase is entered. Otherwise, the local search phase includes:

[0027] Initialize the population and conversion probability of the anomaly detection model, calculate the fitness value of the population of the initial anomaly detection model, and record the anomaly detection model with the largest fitness as the initial optimal solution;

[0028] In each iteration, the fitness value of each anomaly detection model is calculated, the current optimal solution is found, and compared with the initial optimal solution;

[0029] A new anomaly detection model is selected through a tournament strategy, random numbers are generated, and the random numbers are compared with the transition probabilities. Based on the comparison results, the global search phase or the local search phase is selectively entered.

[0030] In the global search phase, a large-scale search is performed by generating Levy flights to change the solution space. If it is the local search phase, small-scale adjustments are made within the solution space.

[0031] Furthermore, a new anomaly detection model is selected through a tournament strategy, a random number is generated, and the random number is compared with the transition probability. The global search phase or the local search phase is selectively entered according to the comparison result, including:

[0032] Use the tournament strategy to select a new anomaly detection model from the current population. By comparing the fitness of several anomaly detection models, the anomaly detection model with the best fitness is selected as the candidate solution.

[0033] Generate a random number and compare it with the transition probability. If the random number is greater than the transition probability, enter the global search phase and search a large range of solution space;

[0034] If the random number is less than or equal to the conversion probability, the local search phase is entered, and the current solution is adjusted, local optimization is performed, and a small range of solution space is searched.

[0035] Furthermore, the secure cargo transportation shared data is distributed and stored on the data sharing platform. The stored cargo transportation shared data is dynamically allocated through the data scheduling algorithm, and the data sharing access rights are allocated, including:

[0036] According to the set storage capacity and number of storage nodes, the stored cargo transportation shared data is randomly initialized, and the corresponding storage location is allocated according to the cargo transportation shared data type;

[0037] The stored cargo transportation shared data is optimized using a scheduling optimization algorithm, and the optimized cargo transportation shared data is updated using differential mutation operations. The distribution of cargo transportation shared data in the data sharing platform is dynamically adjusted based on the access frequency of the cargo transportation shared data and the load of the storage nodes.

[0038] Cross-process the stored cargo transportation shared data with its access policy, and calculate the access rights of each cargo transportation shared data item based on the importance and priority of the cargo transportation shared data;

[0039] Based on the access frequency and demand of cargo transportation shared data, matching cargo transportation shared data is selected for optimized storage; for frequently accessed cargo transportation shared data, its storage location and access path are optimized through cross-operation;

[0040] According to the access requirements and permission configuration of cargo transportation shared data, the access rights of cargo transportation shared data are dynamically allocated, and the storage path and permission allocation of cargo transportation shared data are optimized through the taboo search method.

[0041] Furthermore, the stored cargo transportation shared data is optimized using a scheduling optimization algorithm, including:

[0042] Use the data segmenter to segment the stored cargo transportation shared data, mark the attributes of each cargo transportation shared data item, and remove irrelevant and inactive data;

[0043] Perform association rule mining on the cut cargo transportation shared data and define the elements that have association relationships between the same cargo transportation shared data items;

[0044] For each shared data item in cargo transportation, calculate its associated weight in the frequent itemset and optimize its weight according to the occurrence probability of the frequent itemset;

[0045] Traverse the link matrix of the entire frequent item set and calculate the weight value of each cargo transportation shared data item;

[0046] Output the final optimized scheduling plan and determine the optimal stored cargo transportation sharing data.

[0047] Furthermore, the taboo search method is used to optimize the storage path and authority allocation of cargo transportation shared data, including:

[0048] Set the maximum number of iterations of the tabu search algorithm and initialize the tabu table to empty;

[0049] Generate new candidate storage paths through local search and calculate the fitness value of each candidate storage path;

[0050] Determine whether each candidate storage path is in a taboo state, and select the candidate storage path with the best fitness value that is not in the taboo table;

[0051] Compare the optimal candidate storage path that is not in the taboo table with the historical optimal storage path. If the fitness of the optimal candidate storage path that is not in the taboo table is the highest, then update the historical optimal storage path and the taboo table. Otherwise, select a non-tabu path and update the taboo table.

[0052] Ignore the taboo status of the optimal candidate storage path, treat it as a new historical candidate storage path, add it to the current path and taboo table, and update the status of the taboo table;

[0053] If the optimal candidate storage path does not surpass the historical optimal storage path, the candidate storage path with the highest fitness outside the taboo table is selected, the current storage path is updated and added to the taboo table, and the shared data storage path and permission allocation for cargo transportation are further optimized;

[0054] Check whether the maximum number of iterations has been reached. If so, output the historical optimal candidate storage path; otherwise, continue the iterative optimization process.

[0055] Output the historical optimal cargo transportation shared data storage path and permission allocation plan after taboo search optimization as the final cargo transportation shared data storage plan.

[0056] Furthermore, the formula for calculating the weight value of each cargo transportation shared data item is:

[0057]

[0058] M(d a ) represents the weight value of the a-th cargo transportation shared data item; represents the initial weight value of the a-th cargo transportation shared data item; out(d b ) represents the number of data items associated with the b-th cargo transportation shared data item; In(d a ) represents the set of data items associated with the a-th cargo transportation shared data item; M(d b ) represents the weight value of the b-th cargo transportation shared data item; M represents the weight value of the cargo transportation shared data item; W represents the average association entropy weight value of the cargo transportation shared data item.

[0059] According to another aspect of the present invention, a hazardous chemicals cargo transportation data sharing system is provided, the system comprising:

[0060] The data acquisition module is used to obtain hazardous chemical cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data;

[0061] Anomaly detection module, used to detect anomalies in cargo transportation shared data using anomaly detection algorithms to obtain safe cargo transportation shared data;

[0062] The data sharing module is used to distribute and store secure cargo transportation shared data on the data sharing platform, dynamically allocate the stored cargo transportation shared data through the data scheduling algorithm, and allocate data sharing access rights.

[0063] The beneficial effects of the present invention are:

[0064] 1. By extracting cargo transportation characteristic data and generating shared data, the present invention can achieve comprehensive monitoring and information sharing of the entire transportation process, effectively reducing risks in transportation. The anomaly detection algorithm can promptly detect anomalies in the data and ensure the accuracy and security of transportation data. The application of distributed storage and data scheduling algorithms makes the storage of transportation data more flexible and efficient, ensuring that resources can be quickly deployed in the event of an emergency and improving emergency response capabilities. In addition, by reasonably allocating data sharing access rights, it ensures that different stakeholders share necessary data within a compliant framework, thereby enhancing cross-enterprise and cross-regional collaboration capabilities.

[0065] 2. The present invention optimizes the method for sharing hazardous chemical cargo transportation data by utilizing an anomaly detection algorithm, which can significantly improve the security and accuracy of data. The anomaly detection algorithm ensures the validity and reliability of shared data by eliminating abnormal data that does not meet expectations. In addition, the model is optimized by combining the Cauchy mutation strategy and the reverse learning strategy, making anomaly detection more accurate and able to quickly respond to abnormal situations during transportation.

[0066] 3. The present invention optimizes the hazardous chemical cargo transportation data, which not only improves the flexibility and efficiency of data storage, but also enhances the security and reliability of data access. By dynamically adjusting the storage location and optimizing the storage path, it can achieve more efficient resource allocation based on the access frequency and demand of cargo transportation data. At the same time, the taboo search method plays an important role in authority allocation and storage path optimization, ensuring the optimal path for data access, avoiding redundancy and unnecessary storage burden, thereby improving the data processing efficiency during the transportation process, enhancing the cross-departmental collaborative operation capability, and effectively ensuring the safety and efficiency of hazardous chemical transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in 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 paying any creative work.

[0068] Figure 1 This is a flow chart of a method for sharing hazardous chemicals transportation data according to an embodiment of the present invention;

[0069] Figure 2 The present invention is a block diagram of a hazardous chemicals transportation data sharing system.

[0070] In the picture:

[0071] 1. Data acquisition module; 2. Anomaly detection module; 3. Data sharing module. DETAILED DESCRIPTION

[0072] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0073] In the description of the present invention, unless otherwise specified, "plurality" means two or more. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0074] According to an embodiment of the present invention, a method and system for sharing hazardous chemicals transportation data are provided.

[0075] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, the method for sharing hazardous chemicals transportation data according to an embodiment of the present invention includes:

[0076] S1. Obtain hazardous chemical cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data;

[0077] Specifically, hazardous chemicals transportation data includes:

[0078] 1) Cargo information:

[0079] Types of cargo: such as flammable, explosive, toxic chemicals, etc.

[0080] Quantity and volume of goods: weight, volume and packaging method of the goods.

[0081] Hazardous characteristics: such as chemical properties, reactivity, toxicity, environmental hazards, etc.

[0082] 2) Transportation route information:

[0083] Origin and destination: The origin and destination of the shipment.

[0084] Route: cities, counties, roads, etc. passed through during transportation.

[0085] Mode of transportation: road, rail, sea, air, etc.

[0086] 3) Transport vehicle and driver information:

[0087] Type and number of transport vehicle: e.g. truck, tanker, etc.

[0088] Driver's qualification certificate and health status: whether he is qualified to transport dangerous goods, driver's work experience, etc.

[0089] 4) Environmental data during transportation:

[0090] Climatic conditions: such as temperature, wind speed, precipitation, etc., especially weather factors that may affect safety during transportation.

[0091] 5) Real-time location data: real-time location information obtained through GPS.

[0092] Transportation time: the start and end time of the transportation process, as well as the time spent on the way.

[0093] 6) Safety measures and emergency plan data:

[0094] Safety equipment: whether equipped with fire extinguishers, leak prevention equipment, etc.

[0095] Emergency contact person and plan: Emergency response contact person and plan in case of an accident.

[0096] Specifically, freight transportation shared data includes:

[0097] 1) Freight transportation characteristic data:

[0098] It includes characteristic information such as transportation routes, cargo types, and transportation time, helping different departments or companies to have a consistent understanding and monitoring of the transportation process.

[0099] 2) Real-time transportation monitoring data:

[0100] Location data: Real-time location updates during transport (via technologies such as GPS).

[0101] Speed ​​and time data: shipping speed, estimated time of arrival, etc.

[0102] Transportation status: information such as whether the transportation is proceeding as planned, whether there are any delays, and whether the goods are safe.

[0103] 3) Safety data during transportation:

[0104] Safety inspection records: results of safety inspections during transportation, such as whether dangerous goods have been properly handled and whether the vehicle meets safety standards.

[0105] Risk assessment data: Risk assessment results generated based on factors such as transportation routes and cargo characteristics.

[0106] 4) Emergency Response Information:

[0107] Emergency contacts and plan information: When an accident occurs, how to quickly coordinate and handle it, early warning information, emergency response steps, etc.

[0108] 5) Historical data and analysis reports:

[0109] Transport history: historical data for each transport, including transport time, any accidents or special circumstances during transport, etc.

[0110] Performance evaluation and feedback: A summary report after the transportation process, including the efficiency of the transportation, whether it was completed as planned, and whether there is room for optimization.

[0111] Specifically, obtaining hazardous chemicals transportation data includes:

[0112] 1) Obtaining data from transportation vehicles and logistics systems:

[0113] GPS positioning system: can obtain the location information of transport vehicles in real time.

[0114] On-board sensors: monitor vehicle speed, temperature, humidity and other environmental data to ensure that transportation conditions meet safety requirements.

[0115] Transportation Management Software (TMS): Obtain relevant data such as transportation routes, time, and cargo through the transportation management system.

[0116] 2) Collect data through cargo delivery platforms:

[0117] Logistics platform interface: Logistics companies collect and share all information related to cargo transportation through open platform interfaces.

[0118] EDI system: a data exchange platform between enterprises, such as obtaining order, shipment, arrival and other information through the Electronic Data Interchange (EDI) system.

[0119] 3) Real-time monitoring and data collection:

[0120] Sensors and IoT devices: For example, sensors can monitor environmental factors related to hazardous chemicals, such as temperature and humidity.

[0121] Mobile applications and devices: For example, applications used by drivers and managers to record real-time transportation data.

[0122] 4) Sharing data with emergency response systems:

[0123] Emergency response platform: During transportation, if an emergency occurs, relevant data (such as the location of the accident, type of cargo, hazard level, etc.) can be quickly shared with the emergency response system.

[0124] Safety monitoring system: By integrating monitoring cameras, sensors and other equipment, the entire transportation process can be tracked.

[0125] S2. Use an anomaly detection algorithm to detect anomalies in the cargo transportation shared data to obtain safe cargo transportation shared data;

[0126] Specific, secure cargo transportation sharing data includes:

[0127] 1) Normal cargo transportation data:

[0128] Cargo information: correctly labeled hazardous chemicals name, category, quantity, packaging method, hazard level, etc.

[0129] Transportation route information: transportation routes, starting points, destinations, transfer stations, etc. that are reasonably planned and comply with regulatory requirements.

[0130] Transit time data: Shipments, estimated arrivals, and real-time transit time data that are in line with the transportation plan, with no unusual delays or advances.

[0131] 2) Vehicle and personnel data that meets regulatory requirements:

[0132] Legally registered transport vehicle information, including license plate number, vehicle type, load capacity, etc.

[0133] Compliant driver information, such as qualifications for transporting hazardous chemicals and no record of violations.

[0134] The vehicle is in normal condition, with no abnormalities in fuel level, brakes, tire pressure, temperature control system, etc.

[0135] 3) Stable environment and monitoring data:

[0136] The GPS track is normal, the vehicle has not deviated from the designated route, and there is no abnormal parking or speeding behavior.

[0137] Sensor data such as temperature and humidity are normal and do not exceed the safety threshold required for cargo transportation.

[0138] There were no abnormalities in the on-board surveillance video, and no dangerous behavior was detected (such as driver fatigue driving, cargo leakage, etc.).

[0139] 4) Emergency management data that meets safety standards:

[0140] The safety plan is normal. In the event of an emergency (such as bad weather or an accident), the system will trigger a reasonable emergency plan.

[0141] The emergency contact person is available, the information of the dispatch center and emergency response personnel is complete, and communication is smooth.

[0142] Consistency of historical data: newly acquired data is consistent with historical transportation data, with no sudden or unexplained abnormal changes.

[0143] Specifically, the abnormal data of cargo transportation shared data include:

[0144] 1) Abnormal object information:

[0145] Incomplete cargo data: such as missing cargo name, hazard level, quantity, packaging information, etc.

[0146] Misclassification of goods: Mislabeling of hazardous chemicals can lead to errors in storage, transportation, or emergency response.

[0147] Exceeding the transportation limit: The quantity of hazardous chemicals transported in a single shipment exceeds the regulations or the vehicle's carrying capacity.

[0148] 2) Abnormal transportation routes and times:

[0149] Vehicle deviates from the route: The vehicle does not follow the prescribed route, which may involve illegal operation or emergency.

[0150] Abnormal parking: long parking periods or frequent route changes may involve illegal transactions, leaks, accidents, and other issues.

[0151] Abnormal transportation time: significantly advanced or delayed, which may be affected by traffic control, accidents, vehicle failure, etc.

[0152] 3) Abnormalities of vehicle and driver:

[0153] Driver violations: fatigue driving, speeding, violation of traffic regulations, etc.

[0154] Abnormal vehicle status: brake, engine, and sensor failures affect transportation safety.

[0155] No transportation qualifications: The driver or vehicle does not meet the requirements of hazardous chemicals transportation regulations.

[0156] 4) Abnormal sensor and environmental data:

[0157] GPS signal loss or abnormal jumps: This may be due to system failure, signal shielding, or device tampering.

[0158] Exceeding temperature and humidity standards: Hazardous chemicals require specific temperature and humidity for transportation, and exceeding the threshold may trigger chemical reactions or accidents.

[0159] Dangerous events are not identified: for example, a sensor detects a gas leak but does not trigger the alarm system.

[0160] 5) Emergency management exceptions:

[0161] Invalid emergency contact: The contact information is incorrect and the relevant person cannot be contacted in an emergency.

[0162] Abnormal dispatch response: After an emergency occurs, the system fails to correctly dispatch the nearest rescue resources.

[0163] Historical data anomalies: Transportation records and historical data are inconsistent, such as a vehicle appearing in different locations at the same time.

[0164] S3. Distribute and store secure cargo transportation shared data on a data sharing platform, dynamically allocate the stored cargo transportation shared data through a data scheduling algorithm, and assign data sharing access rights;

[0165] Among them, the anomaly detection algorithm is used to detect anomalies in the cargo transportation shared data, and the safe cargo transportation shared data obtained includes:

[0166] Obtain shared data samples of cargo transportation, classify data of different abnormal types, divide the data of different abnormal types into training sets and test sets, and establish an anomaly detection model;

[0167] Optimizing the anomaly detection model through learning algorithms and inputting the test set into the optimized anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data; including:

[0168] Initialize the population of anomaly detection models, calculate the fitness of each anomaly detection model, sort them, and optimize the population based on the Cauchy mutation strategy and reverse learning strategy;

[0169] According to the fitness value, it is judged whether the preset conditions are met, and then the global search phase is entered; otherwise, the local search phase is carried out;

[0170] If the maximum number of iterations is reached, the optimal anomaly detection model is output, and the test set is input into the optimal anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data.

[0171] Specifically, the anomaly detection algorithm is an extreme learning machine autoencoder, a deep learning model that combines the extreme learning machine (ELM) and autoencoder architecture. Its basic concept is to leverage the rapid training capabilities of the extreme learning machine to construct an autoencoder, thereby reducing training time while maintaining the effectiveness of the autoencoder. It has demonstrated excellent performance in detecting anomalies in shared freight transportation data. By using the anomaly detection algorithm to detect anomalies in shared freight transportation data, the ELM-AE can efficiently and accurately identify anomalous data in the transportation process and eliminate records that do not conform to actual conditions, thereby ensuring the accuracy and security of system decisions.

[0172] Specifically, the learning algorithm is the hunter-prey optimization algorithm. The hunter-prey optimization algorithm (HPO) is a swarm intelligence optimization algorithm based on predation behavior in nature. It simulates the strategies adopted by predators (hunters) in the process of tracking and capturing prey to find the optimal solution in the search space.

[0173] It should be explained that historical transportation data provided by freight companies are collected and sorted by data categories, such as normal data (such as reasonable freight tracks, accurate order information), abnormal data (such as duplicate orders, invalid GPS tracks, false transportation records, etc.), and the data are divided into training sets (for training anomaly detection models) and test sets (for evaluating the anomaly detection capabilities of the models); anomaly detection algorithms suitable for cargo transportation data are selected, such as machine learning models, neural networks, statistical analysis, extreme learning machine autoencoders and other methods, and the model is initialized, and the learning algorithm is used to optimize the anomaly detection model to enable it to more accurately identify abnormal data.

[0174] Population initialization: Create multiple initial anomaly detection models, each trained with different parameters; Fitness calculation: Evaluate the detection accuracy of each model, rank them, and select the model with the best performance.

[0175] Cauchy mutation strategy optimization: For low-fitness models, small-scale parameter adjustments are introduced to improve anomaly detection capabilities.

[0176] Backward learning strategy optimization: Generate new model structures to improve the ability to recognize unseen abnormal patterns.

[0177] If the detection effect of the current model does not reach the preset threshold, it enters the local search phase, optimizes the model in a small range, and improves the recognition ability of specific anomaly types; if the detection ability of the model meets the set requirements, it enters the global search phase to ensure the applicability of the model in different data scenarios; after multiple rounds of optimization, the optimal anomaly detection model is selected and the test data is input into the model; identify and eliminate abnormal data, and finally generate safe cargo transportation shared data for use by logistics companies and regulatory agencies.

[0178] In this optional embodiment, obtaining hazardous chemical cargo transportation data, extracting cargo transportation characteristic data, and generating cargo transportation shared data includes:

[0179] Obtain raw data on hazardous chemical cargo transportation, clean and process the raw data, and remove noise and outliers;

[0180] By constructing data sequences and training samples, selecting embedding dimensions, and extracting cargo transportation feature data;

[0181] The least squares support vector machine is used to analyze the extreme points and trends in the freight transportation characteristic data, calculate the core indicators of the transportation characteristics, and generate the transportation characteristic analysis results; and based on the analysis results, generate freight transportation shared data.

[0182] It is important to clarify that hazardous chemical cargo transportation data is collected from various logistics companies, transport vehicles, GPS systems, and monitoring platforms. The raw data typically includes transportation timestamps, origin and destination, transportation routes, vehicle numbers, driver information, hazardous chemical type, temperature and humidity, transportation conditions (such as wind speed and air pressure), safety incident records, and loading and unloading information. The collected raw data is cleaned to remove noise, such as invalid data, duplicate records, and entries that do not meet format requirements. Outliers, such as those caused by sensor failures or network issues, are processed, and missing data is properly filled or removed to ensure data quality and accuracy. Based on the raw data, time series data is constructed, including but not limited to various dynamic information during transportation (such as transportation speed, cargo status, ambient temperature and humidity, transportation distance, and time). Training samples are extracted by labeling and classifying the known transportation data, defining normal and abnormal transportation conditions, and forming training and test sets for subsequent model training and validation. Data analysis and statistical methods are used to select an appropriate embedding dimension. The embedding dimension indicates how to extract time series features from data by selecting an appropriate time window. This process helps uncover underlying patterns in the data, capture dynamic changes during transportation, and provide useful information for subsequent analysis. Based on the constructed time series and training samples, transportation feature data is extracted. This feature data includes important indicators of the transportation process, such as temperature and humidity fluctuations, transportation speed changes, driving behavior (such as sudden braking and speeding), and cargo status (such as whether there is severe vibration). This feature data helps identify key factors in transportation and ensure transportation safety. The extracted cargo transportation feature data is analyzed using a least squares support vector machine (LS-SVM) model, specifically identifying extreme values ​​and trend changes. Extreme values ​​often represent abnormal changes, such as temperature and humidity anomalies or sudden changes in transportation routes. Trend changes reveal long-term patterns of change during transportation, such as temperature increases and vehicle speed changes. Based on the analysis results of LS-SVM, the core indicators of the transportation process are calculated, such as: average transportation speed, stability of the transportation environment (such as the range of temperature and humidity fluctuations), safety of driving behavior (such as the number of sudden brakes, the number of speeding, etc.), potential risk levels in the transportation of hazardous chemicals (such as unstable factors in transportation), etc. By analyzing the characteristic data and core indicators, the analysis results of the transportation characteristics are generated. These analysis results not only reflect the safety and risks in the transportation process, but also reveal the optimization space in the transportation process. For example, it may be found that the temperature and humidity fluctuations of a certain transportation route are large, and it is recommended to choose a more stable route in future transportation, or it may be found that a certain driver has frequent sudden brakes and needs further training. Based on the analysis results of transportation characteristics, cargo transportation sharing data is generated, including but not limited to:

[0183] Safety score: Comprehensively considers various safety indicators during the transportation process to give an overall transportation safety score.

[0184] Risk Alert: If there are any abnormalities or potential risks during transportation, corresponding alert information will be generated for real-time monitoring and intervention by relevant departments.

[0185] Optimization suggestions: Based on the analysis results, optimization suggestions for the transportation process are put forward, such as changing the transportation route, adjusting driving behavior, improving cargo storage conditions, etc.

[0186] The analyzed and processed cargo transportation shared data is uploaded to the data sharing platform for access by different users.

[0187] For example:

[0188] Logistics companies can optimize transportation scheduling and management based on shared data.

[0189] Regulators can monitor safety during transportation in real time and take intervention measures if anomalies occur.

[0190] Insurance companies can adjust cargo transportation insurance strategies based on data analysis results.

[0191] In this optional embodiment, the formulas for optimizing the population according to the Cauchy mutation strategy and the reverse learning strategy are:

[0192] K new =K better (t)+Cauchy(m0)·K better (t)

[0193] K i (t+1)=ub+δ·(lb-K i (t));

[0194] Where K new represents the parameter value of the anomaly detection model after optimization by the Cauchy mutation strategy; K better (t) represents the parameter value of the anomaly detection model with the highest fitness at the current iteration t; Cauchy (m0) represents the Cauchy distribution function; K i (t+1) represents the parameter value of the anomaly detection model after the i-th anomaly detection model is optimized by the reverse learning strategy at iteration t+1; K i (t) represents the parameter value of the i-th anomaly detection model at iteration t; ub represents the maximum allowable value of the anomaly detection model parameter; lb represents the minimum allowable value of the anomaly detection model parameter; δ represents a random number.

[0195] In this optional embodiment, whether the preset conditions are met is determined based on the fitness value, and then the global search phase is entered; otherwise, the local search phase includes:

[0196] Initialize the population and conversion probability of the anomaly detection model, calculate the fitness value of the population of the initial anomaly detection model, and record the anomaly detection model with the largest fitness as the initial optimal solution;

[0197] In each iteration, the fitness value of each anomaly detection model is calculated, the current optimal solution is found, and compared with the initial optimal solution;

[0198] A new anomaly detection model is selected through a tournament strategy, random numbers are generated, and the random numbers are compared with the transition probabilities. Based on the comparison results, the global search phase or the local search phase is selectively entered.

[0199] In the global search phase, a large-scale search is performed by generating Levy flights to change the solution space. If it is the local search phase, small-scale adjustments are made within the solution space.

[0200] Specifically, this invention performs global and local searches by improving the discrete flower pollination algorithm. The flower pollination algorithm is a swarm intelligence optimization algorithm inspired by the natural pollination process of flowers. It simulates the pollination of flowers through wind and insects to solve optimization problems. This invention discretizes the traditional flower pollination algorithm and introduces an improved algorithm based on a tournament selection strategy. This reduces the number of iterations, shortens the optimization time, and quickly and accurately obtains the optimal solution.

[0201] In order to facilitate understanding of the above technical solution of the present invention, the present invention will be described in detail below in which the fitness value is used to determine whether the preset conditions are met and then enter the global search stage; otherwise, the local search stage is performed.

[0202] Step 1: Initialization phase:

[0203] 1) Set parameters:

[0204] Population size = 10.

[0205] Maximum number of iterations = 100.

[0206] Conversion probability = 0.5.

[0207] 2) Initialize the population: Randomly generate 10 anomaly detection models, denoted as Y1, Y2, ..., Y 10 .

[0208] 3) Calculate the fitness value: Calculate the fitness value f(Y x ), record the model with the largest fitness as the initial optimal solution G * .

[0209] Example:

[0210] f(Y1)=0.85.

[0211] f(Y2)=0.92.

[0212] f(Y3)=0.78.

[0213] ….

[0214] G * =Y2 (f(Y2)=0.92 is the maximum fitness value).

[0215] Step 2: Iterative process:

[0216] 1) Calculate the fitness value and update the optimal solution:

[0217] Calculate the fitness value: Calculate the fitness value f(Y x ).

[0218] Example:

[0219] f(Y1)=0.88.

[0220] f(Y2)=0.91.

[0221] f(Y3)=0.82.

[0222] ….

[0223] Update the optimal solution: update the current optimal solution g * and the initial optimal solution G * , compare, if g * Better, then update G * .

[0224] Example:

[0225] Current optimal solution g * =Y2(f(Y2)=0.91).

[0226] Initial optimal solution G * =Y2(f(Y2)=0.92).

[0227] Since f(g * ) is less than f(G * ), do not update G * .

[0228] Step 3. Select a new model:

[0229] 1) Tournament strategy: Select two models from the current population, compare their fitness values, and select the model with the larger fitness value as the new model.

[0230] Example: Select Y1 and Y3, compare f(Y1) = 0.88 and f(Y3) = 0.82, and select Y1 as the new model.

[0231] 2) Judgment and search stage:

[0232] Generate random number: Generate a random number rand(0, 1).

[0233] Example: rand(0, 1) = 0.6.

[0234] Compare and choose:

[0235] If rand(0, 1) is greater than the transition probability, the global search phase is entered.

[0236] If rand(0, 1) is less than or equal to the transition probability, the local search phase is entered.

[0237] Example: Since 0.6 is less than 0.5, the global search phase is entered.

[0238] Step 4: Global search phase:

[0239] 1) Lévy flight: The step length is generated by the Lévy flight formula, which is:

[0240]

[0241] Among them, u and v obey the normal distribution, α=1, β=1.5.

[0242] Example:

[0243] u=0.5,v=0.3, substitute into Levy flight formula and we get L≈1.44.

[0244] 2) Update the model: Update the model parameters according to the Levy step size and expand the search range.

[0245] Example: Update the parameters of Y1 to generate a new model Y1 * .

[0246] 3) Local search stage:

[0247] Small-scale adjustment: Fine-tune the model parameters within the solution space and fine-tune the model.

[0248] Example:

[0249] Fine-tune the parameters of Y1 to generate a new model Y1 ** .

[0250] 4) Termination conditions:

[0251] Check the number of iterations. If the maximum number of iterations reaches 100, the process ends and the optimal solution G is output. * .

[0252] Example:

[0253] The number of iterations is 100, and the output is G * =Y2(f(Y2)=0.92).

[0254] 5) Model application:

[0255] Test set input: Input the test set data into the optimal anomaly detection model G * , detect and remove abnormal data.

[0256] Generate secure data: retain normal data and generate secure cargo transportation sharing data.

[0257] In this optional embodiment, selecting a new anomaly detection model through a tournament strategy, generating a random number, comparing the random number with the transition probability, and selectively entering the global search phase or the local search phase based on the comparison result include:

[0258] Use the tournament strategy to select a new anomaly detection model from the current population. By comparing the fitness of several anomaly detection models, the anomaly detection model with the best fitness is selected as the candidate solution.

[0259] Generate a random number and compare it with the transition probability. If the random number is greater than the transition probability, enter the global search phase and search a large range of solution space;

[0260] If the random number is less than or equal to the conversion probability, it enters the local search phase, adjusts the current solution, performs local optimization, and searches a small range of solution space.

[0261] It should be noted that in the initial phase, multiple candidate anomaly detection models are generated, each with a different parameter combination and training set data. These models form the initial population, and the goal is to find a model that can best identify anomalies. The tournament strategy selects the best model from the current population. In each iteration, several models are randomly selected for the tournament. Assume that three models are selected for the tournament each time. Each candidate model has a fitness value, typically evaluated based on its precision, recall, or F1 score on the test dataset. Models with higher fitness values ​​are considered superior. By comparing the fitness values ​​of these three models, the model with the best fitness value is selected as the "candidate solution." This candidate solution will serve as the key point in the next optimization process. During the optimization process, a random number is generated, uniformly distributed in the range [0, 1]. This random number determines whether to conduct a global search or a local search. A transition probability is set, which determines the probability of entering a global search during a given optimization process. The generated random number is compared with the transition probability. If the random number is greater than the transition probability, the global search phase begins. If the random number is less than or equal to the transition probability, the local search phase begins. Global search aims to conduct a broad search across the entire solution space, exploring a wider range of parameter combinations and model structures. This search method can discover potentially excellent solutions and avoid being trapped in local optima. Global search is typically performed by random jumps or by generating new candidate models. During the search process, the model's parameter space undergoes significant changes, which may involve major parameter adjustments or new hyperparameter configurations. After the global search, a new, more optimal solution may be found. This new solution is compared with the current candidate solution, and if the fitness value is higher, the candidate solution is updated as the new optimal solution. Local search fine-tunes the current solution, aiming to improve model performance through small adjustments. This optimization method allows for more refined model adjustments and improves model performance within the current solution space. Local search typically involves adjusting certain parameters of the current model (such as the regularization parameter and learning rate) to conduct a detailed search within the solution space. The primary goal of local search is to improve the current solution, rather than extensively exploring new solution spaces. During the local search phase, the model is optimized through small changes. For example, the kernel function of the model can be fine-tuned, its weighting of the training data can be optimized, or other hyperparameters can be adjusted. The core of local search is to find the best solution near the current solution, rather than completely redefining the solution space. After the local search phase, if the new solution has a higher fitness, it will be selected as a new candidate for subsequent operations. Tournament selection, random number comparison, global search, and local search are repeated until a preset stopping condition is met. These stopping conditions typically include reaching a preset maximum number of iterations or the fitness value of the current solution reaching the desired target. In each round of optimization, the best model is selected and further optimized.Through repeated selection, comparison, and search, the model's performance is gradually improved. When the maximum number of iterations is reached or the preset performance criteria are met, the optimal anomaly detection model is output. This optimal model, the result of multiple optimizations using global and local searches, efficiently identifies anomalies in the data. The optimal anomaly detection model can be used for anomaly detection in actual cargo transportation data. This model examines the input data, identifies anomalous records, and filters them out, ensuring data quality and security.

[0262] In this optional embodiment, the secure cargo transportation shared data is distributedly stored on a data sharing platform, the stored cargo transportation shared data is dynamically allocated through a data scheduling algorithm, and data sharing access rights are allocated, including:

[0263] According to the set storage capacity and number of storage nodes, the stored cargo transportation shared data is randomly initialized, and the corresponding storage location is allocated according to the cargo transportation shared data type;

[0264] The stored cargo transportation shared data is optimized using a scheduling optimization algorithm, and the optimized cargo transportation shared data is updated using differential mutation operations. The distribution of cargo transportation shared data in the data sharing platform is dynamically adjusted based on the access frequency of the cargo transportation shared data and the load of the storage nodes.

[0265] Cross-process the stored cargo transportation shared data with its access policy, and calculate the access rights of each cargo transportation shared data item based on the importance and priority of the cargo transportation shared data;

[0266] Based on the access frequency and demand of cargo transportation shared data, matching cargo transportation shared data is selected for optimized storage; for frequently accessed cargo transportation shared data, its storage location and access path are optimized through cross-operation;

[0267] According to the access requirements and permission configuration of cargo transportation shared data, the access rights of cargo transportation shared data are dynamically allocated, and the storage path and permission allocation of cargo transportation shared data are optimized through the taboo search method.

[0268] Specifically, the data scheduling algorithm is an improved differential evolution algorithm. The differential evolution algorithm (DE) is a population-based random optimization algorithm that is inspired by the natural selection process, especially the mutation and selection mechanism in species evolution. Differential evolution algorithms are widely used in global optimization problems. The disadvantage is that they are prone to falling into local optimal solutions, have a slow convergence speed, and may not be able to find the global optimum quickly for complex multimodal problems. Therefore, the present invention is based on the traditional differential evolution algorithm and combines it with taboo search for improvement. The purpose is to overcome the shortcomings of the traditional differential evolution algorithm, especially to prevent the algorithm from falling into local optimal solutions during the search process and to speed up the convergence speed. The taboo search method records historical solutions to avoid repeated visits to similar solutions during the search process, thereby enhancing the global search capability.

[0269] Specifically, the scheduling optimization algorithm is an improved TextRank algorithm. The traditional TextRank algorithm is a graph-based sorting algorithm. Its basic idea is to represent text as a graph structure, using nodes to represent words or sentences, edges to represent their similarity or co-occurrence relationship, and to measure their importance by iteratively calculating the weights of the nodes. In the present invention, the traditional TextRank algorithm cannot distinguish the specific degree of association of data items. It only calculates weights based on connection relationships, which may lead to poor optimization effects, and cannot fully utilize the distribution characteristics of data items. It cannot accurately measure the storage optimization priority of data items. Therefore, by combining the association entropy, the association entropy (AE) is a measurement method for measuring the degree of association between data items. It can effectively supplement the deficiencies of the TextRank algorithm in data weight evaluation. By calculating the association entropy of data items, the optimization ability of TextRank is enhanced, making storage scheduling more in line with data access rules.

[0270] It should be noted that the stored shared freight transportation data is randomly initialized based on the specified storage capacity and number of storage nodes. Assume that the platform provides multiple storage nodes with a certain storage capacity capable of storing different types of freight transportation data. This shared freight transportation data includes key information such as transportation time, cargo type, transportation route, and transportation status. This data is categorized and stored according to different categories. Each type of freight transportation data is assigned to an appropriate storage location based on pre-defined rules. For example, transportation route data and vehicle information may be stored on one node, while cargo status data may be stored on another. The system categorizes and stores data based on data type (such as vehicle data, route data, order data, etc.). Each type of data is assigned to a pre-defined storage location to ensure efficient storage and access. Storage nodes may have different processing capabilities and storage capacities, so the system selects the most appropriate storage location based on the characteristics of the storage node and the data type. A scheduling optimization algorithm is used to optimize the stored shared freight transportation data. The optimization goals include improving storage efficiency, reducing data access latency, and improving overall system load balancing. The scheduling optimization algorithm dynamically adjusts storage locations based on the following factors:

[0271] Data access frequency: Frequently accessed data will be preferentially stored in nodes with faster storage speeds.

[0272] Storage node load: The node load will also affect the distribution of data to avoid overloading a node.

[0273] The scheduling optimization algorithm will regularly optimize based on the status of storage nodes and data access requirements to ensure efficient access to data.

[0274] During the data optimization process, the optimized shared cargo transportation data is updated through differential mutation operations. Differential mutation operations help the system explore new data storage solutions and adapt to data changes. Differential mutation increases the system's adaptability to data changes, ensuring flexible and reliable data storage. Based on the access frequency of shared cargo transportation data and the load of storage nodes, the system dynamically adjusts the distribution of shared cargo transportation data within the distributed storage platform. For example, if the access frequency of a certain type of data increases, the system will migrate it to a storage node with faster access speeds. If the load on a storage node is too high, the system will migrate some data to other nodes with lighter loads. Based on the optimized data storage location, the system updates the data storage path to ensure that data can be read and written along the optimal path. The stored shared cargo transportation data is cross-processed with its access policy. The system calculates access permissions for each shared cargo transportation data item based on the data's importance, priority, and access requirements. For example, some data (such as sensitive transportation information) may require high permissions to access, while other data (such as general transportation route information) can be open to more users. Access rights for each data item are calculated based on its importance and priority, ensuring that only authorized users or systems can access sensitive or important data. Storage of shared cargo transportation data is optimized based on access frequency and demand. For example, the system optimizes storage of frequently accessed cargo transportation data on nodes with faster read speeds and higher bandwidth to reduce access latency. For certain important transportation data, the system may select multiple backups to ensure high data availability and reliability. The system optimizes storage locations and data access paths through cross-mining operations. For example, the system may adjust data storage node selection based on access frequency and move frequently queried data to more frequently accessed areas to reduce access latency. Data access rights are dynamically assigned based on access requirements and permission configuration for shared cargo transportation data. For example, some users may only have access to non-sensitive data, while other users (such as managers or high-level users) have access to all data. Permission configuration is not only based on user roles but also closely correlated with real-time data demand changes. For example, in an emergency, certain data may be temporarily open to more users for emergency response. Tabu search methods are used to further optimize data storage paths and permission assignment. Tabu search is a local search algorithm that avoids being trapped in local optimal solutions and seeks the global optimal solution by restricting certain infeasible paths. At this stage, the tabu search algorithm explores multiple permission allocation schemes to find the most appropriate one, ensuring data security and efficient access.

[0275] In this optional embodiment, optimizing the stored cargo transportation sharing data using a scheduling optimization algorithm includes:

[0276] Use the data segmenter to segment the stored cargo transportation shared data, mark the attributes of each cargo transportation shared data item, and remove irrelevant and inactive data;

[0277] Perform association rule mining on the cut cargo transportation shared data and define the elements that have association relationships between the same cargo transportation shared data items;

[0278] For each shared data item in cargo transportation, calculate its associated weight in the frequent itemset and optimize its weight according to the occurrence probability of the frequent itemset;

[0279] Traverse the link matrix of the entire frequent item set and calculate the weight value of each cargo transportation shared data item;

[0280] Output the final optimized scheduling plan and determine the optimal stored cargo transportation sharing data.

[0281] It's important to note that shared freight transport data items often contain multiple attributes and information, such as transport time, route, vehicle type, and cargo type. To better analyze and optimize the data, the stored shared freight transport data is first segmented using a data segmenter. The data segmenter processes each data item, breaking down complex transport data into smaller, meaningful segments. These segments can be attributes, values, or other related information, such as:

[0282] Transit time: date, hour, minute, etc.

[0283] Type of cargo: chemicals, electronic products, etc.

[0284] Transportation route: starting point, end point, transfer station, etc.

[0285] During the segmentation process, the data tokenizer labels the different attributes of each shared freight transport data item for subsequent analysis. For example, "transport time" is labeled as time-related data, and "transport route" is labeled as geographic information data. During the segmentation process, inactive data irrelevant to the actual transport data (such as meaningless punctuation, spaces, and malformed data) is identified and removed to improve data accuracy and analysis efficiency. After data segmentation and labeling, the system uses association rule mining techniques to analyze the inherent connections between different data items. For example, there may be certain regular relationships between transport time and cargo type, or between transport route and vehicle type. The goal of association rule mining is to discover which data items frequently appear together and, in turn, to uncover correlations between different data items. For example, if a certain type of cargo is often transported along a specific transport route and within a specific time period, such correlations can be automatically identified using rule mining tools. Based on the mining results, the system defines elements that correlate between the same shared freight transport data items and creates a correlation matrix between the data. For example, if "chemicals" are commonly transported on "hazardous goods transport vehicles," the correlation between them is defined as high. The system analyzes frequent itemsets in the data to identify itemsets that appear frequently across multiple shared data items in freight transportation. Frequent itemsets typically represent events or data item combinations that frequently occur together during transportation. For example, certain types of freight consistently occur with specific transportation routes or modes of transport. For each shared data item in the frequent itemset, the system calculates its association weight within the frequent itemset. The association weight reflects the strength of the relationship between a data item and other items; a higher weight indicates a closer association. Based on the probability of occurrence of the frequent itemset, the system optimizes the weight of each shared data item in freight transportation. Items that appear more frequently receive higher weights and are preferentially stored in storage nodes with higher access rates to improve access efficiency. For example, if a particular transportation route appears frequently in a frequent itemset, its weight will be increased, making it more prioritized in subsequent scheduling. After frequent itemset analysis, the system constructs a frequent itemset link matrix, which describes the strength of associations between different data items. Each element in the matrix represents the degree of association between two data items; the stronger the association, the higher the value in the matrix. The entire link matrix is ​​traversed to calculate the weight of each shared data item for cargo transportation. Higher weights indicate a greater importance within the system, and the system prioritizes these high-weighted data items during storage and scheduling. Furthermore, weight calculations take into account factors such as storage node load and access frequency to ensure that the optimized storage solution is both efficient and balanced. After traversing and calculating the weight of each data item, the system uses a scheduling optimization algorithm to output the final optimized solution.The optimized scheduling plan determines the optimal storage location for each shared cargo transport data item, minimizing data access latency and storage node load. The system selects the most appropriate storage node based on the weight of the data item and the correlation of frequent itemsets. For example, frequently accessed and highly weighted cargo transport data is assigned to a node with better storage performance. Based on the optimized scheduling plan, the system also optimizes storage paths to ensure prompt response for frequently accessed data. By rationally configuring storage paths, the system not only reduces access times but also effectively avoids network bandwidth congestion.

[0286] In this optional embodiment, optimizing the storage path and authority allocation of cargo transportation shared data by using a taboo search method includes:

[0287] Set the maximum number of iterations of the tabu search algorithm and initialize the tabu table to empty;

[0288] Generate new candidate storage paths through local search and calculate the fitness value of each candidate storage path;

[0289] Determine whether each candidate storage path is in a taboo state, and select the candidate storage path with the best fitness value that is not in the taboo table;

[0290] Compare the optimal candidate storage path that is not in the taboo table with the historical optimal storage path. If the fitness of the optimal candidate storage path that is not in the taboo table is the highest, then update the historical optimal storage path and the taboo table. Otherwise, select a non-tabu path and update the taboo table.

[0291] Ignore the taboo status of the optimal candidate storage path, treat it as a new historical candidate storage path, add it to the current path and taboo table, and update the status of the taboo table;

[0292] If the optimal candidate storage path does not surpass the historical optimal storage path, the candidate storage path with the highest fitness outside the taboo table is selected, the current storage path is updated and added to the taboo table, and the shared data storage path and permission allocation for cargo transportation are further optimized;

[0293] Check whether the maximum number of iterations has been reached. If so, output the historical optimal candidate storage path; otherwise, continue the iterative optimization process.

[0294] Output the historical optimal cargo transportation shared data storage path and permission allocation plan after taboo search optimization as the final cargo transportation shared data storage plan.

[0295] It should be explained that in order to ensure the convergence and computational efficiency of the algorithm, a maximum number of iterations is set. For example, if the algorithm still does not find a better solution within the set number of iterations, the search is terminated and the current optimal storage path is returned. The taboo table is used to record the storage paths that have been searched recently to prevent the search process from falling into a local optimum and being unable to escape. Initially, the taboo table is empty, which means that all storage paths can be considered. Through the local search method, multiple new candidate storage paths are generated from the current storage path. These paths can be obtained by modifying the data access policy, changing the server storage node, adjusting the permission hierarchy, etc.

[0296] For example, if the original storage path accesses data through nodes A→B→C, local search may generate new paths, such as A→C→B or A→D→C, to try to find a better solution. Each candidate storage path is calculated with a fitness value, which measures the quality of the path. The fitness value is usually calculated based on the following factors:

[0297] Access time: Whether the path can reduce data access latency.

[0298] Data load balancing: Whether the storage pressure is reasonably distributed among storage nodes.

[0299] Security: Whether the path meets the permission control policy to prevent unauthorized access.

[0300] Storage cost: Different storage paths may involve different storage resource consumption.

[0301] Traverse all candidate storage paths and determine whether they are in the taboo table:

[0302] If a candidate path is in the taboo table, it is excluded first unless it can provide a historically optimal breakthrough solution.

[0303] If the candidate path is not in the taboo table, it can be used as an optimization candidate.

[0304] Select the candidate storage path with the best fitness: Among all non-tabu paths, the path with the highest fitness is selected as the current optimal candidate storage path. If the fitness of the current optimal candidate storage path is higher than the previously optimal storage path, the previously optimal storage path is updated. This path is added to the taboo table to prevent the algorithm from repeatedly searching the same path in a short period of time. If the fitness of the current optimal candidate path does not exceed the previously optimal path, the path with the highest fitness outside the tabu table is selected and the current storage path is updated. If the current optimal path is in the tabu table but its fitness exceeds the previously optimal path, its taboo status is ignored and it is set as the new historical optimal path. The tabu table is dynamically adjusted; each time a new path is added, the oldest path may be removed to ensure search flexibility and diversity. If the maximum number of iterations has not been reached, the optimization process is repeated to generate new candidate paths and select the optimal solution. The iteration terminates and the historical optimal storage path is returned as the final optimization result. After the tabu search optimization, the system ultimately selects the storage path with the shortest access time, balanced storage load, and high data security as the final solution. Based on the final storage path, user access permissions are readjusted to ensure that: critical data is accessible only to authorized users, frequently accessed data items have faster access paths, and low-priority data storage does not affect high-priority tasks. Applying the optimized storage path and permission configuration to the data sharing platform improves overall system access efficiency and security.

[0305] In this optional embodiment, the formula for calculating the weight value of each cargo transportation shared data item is:

[0306]

[0307] M(d a ) represents the weight value of the a-th cargo transportation shared data item; represents the initial weight value of the a-th cargo transportation shared data item; out(d b ) represents the number of data items associated with the b-th cargo transportation shared data item; In(d a ) represents the set of data items associated with the a-th cargo transportation shared data item; M(d b ) represents the weight value of the b-th cargo transportation shared data item; M represents the weight value of the cargo transportation shared data item; W represents the average association entropy weight value of the cargo transportation shared data item.

[0308] According to another embodiment of the present invention, Figure 2 As shown, a hazardous chemicals cargo transportation data sharing system is also provided, which includes:

[0309] Data acquisition module 1, used to obtain hazardous chemicals cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data;

[0310] Anomaly detection module 2, used to perform anomaly detection on the cargo transportation shared data using an anomaly detection algorithm to obtain safe cargo transportation shared data;

[0311] The data sharing module 3 is used to store the secure cargo transportation shared data in a distributed manner on the data sharing platform, dynamically allocate the stored cargo transportation shared data through the data scheduling algorithm, and allocate data sharing access rights.

[0312] The data acquisition module 1 is connected through the anomaly detection module 2 and the data sharing module 3.

[0313] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for sharing hazardous chemicals cargo transportation data, characterized in that: include: Obtain hazardous chemical cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data; Use anomaly detection algorithms to detect anomalies in cargo transportation shared data and obtain safe cargo transportation shared data; Distribute and store secure cargo transportation shared data on a data sharing platform, dynamically allocate the stored cargo transportation shared data through a data scheduling algorithm, and assign data sharing access rights; The method of performing anomaly detection on the cargo transportation shared data using an anomaly detection algorithm to obtain safe cargo transportation shared data includes: Obtain shared data samples of cargo transportation, classify data of different abnormal types, divide the data of different abnormal types into training sets and test sets, and establish an anomaly detection model; Optimizing the anomaly detection model through learning algorithms and inputting the test set into the optimized anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data; including: Initialize the population of anomaly detection models, calculate the fitness of each anomaly detection model, sort them, and optimize the population based on the Cauchy mutation strategy and reverse learning strategy; According to the fitness value, it is judged whether the preset conditions are met, and then the global search phase is entered; otherwise, the local search phase is carried out; If the maximum number of iterations is reached, the optimal anomaly detection model is output, and the test set is input into the optimal anomaly detection model to eliminate abnormal data and obtain safe cargo transportation sharing data.

2. A method for sharing hazardous chemicals transportation data according to claim 1, characterized in that: The steps of obtaining hazardous chemical cargo transportation data, extracting cargo transportation characteristic data, and generating cargo transportation shared data include: Obtain raw data on hazardous chemical cargo transportation, clean and process the raw data, and remove noise and outliers; By constructing data sequences and training samples, selecting embedding dimensions, and extracting cargo transportation feature data; The least squares support vector machine is used to analyze the extreme points and trends in the freight transportation characteristic data, calculate the core indicators of the transportation characteristics, and generate the transportation characteristic analysis results; and based on the analysis results, generate freight transportation shared data.

3. A method for sharing hazardous chemicals transportation data according to claim 1, characterized in that: The formulas for optimizing the population according to the Cauchy mutation strategy and the reverse learning strategy are: K new =K better (t)+Cauchy(m0)·K better (t) K i (t+1)=ub+δ·(lb-K i (t)); Where K new represents the parameter value of the anomaly detection model after optimization by the Cauchy mutation strategy; K better (t) represents the parameter value of the anomaly detection model with the highest fitness at the current iteration t; Cauchy(m0) represents the Cauchy distribution function; K i (t+1) represents the parameter value of the anomaly detection model after the i-th anomaly detection model is optimized by the reverse learning strategy at iteration t+1; K i (t) represents the parameter value of the i-th anomaly detection model at iteration t; ub represents the maximum allowable value of the anomaly detection model parameter; lb represents the minimum allowable value of the anomaly detection model parameter; δ represents a random number.

4. A method for sharing hazardous chemicals transportation data according to claim 1, characterized in that: If the fitness value is used to determine whether the preset conditions are met, the global search phase is entered. Otherwise, the local search phase includes: Initialize the population and conversion probability of the anomaly detection model, calculate the fitness value of the population of the initial anomaly detection model, and record the anomaly detection model with the largest fitness as the initial optimal solution; In each iteration, the fitness value of each anomaly detection model is calculated, the current optimal solution is found, and compared with the initial optimal solution; A new anomaly detection model is selected through a tournament strategy, random numbers are generated, and the random numbers are compared with the transition probabilities. Based on the comparison results, the global search phase or the local search phase is selectively entered. In the global search phase, a large-scale search is performed by generating Levy flights to change the solution space. If it is the local search phase, small-scale adjustments are made within the solution space.

5. A method for sharing hazardous chemicals transportation data according to claim 4, characterized in that: The method of selecting a new anomaly detection model through a tournament strategy, generating a random number, comparing the random number with the transition probability, and selectively entering the global search phase or the local search phase according to the comparison result includes: Use the tournament strategy to select a new anomaly detection model from the current population. By comparing the fitness of several anomaly detection models, the anomaly detection model with the best fitness is selected as the candidate solution. Generate a random number and compare it with the transition probability. If the random number is greater than the transition probability, enter the global search phase and search a large range of solution space; If the random number is less than or equal to the conversion probability, it enters the local search phase, adjusts the current solution, performs local optimization, and searches a small range of solution space.

6. A method for sharing hazardous chemicals transportation data according to claim 1, characterized in that: The distributed storage of secure cargo transportation shared data on the data sharing platform, the dynamic allocation of the stored cargo transportation shared data through the data scheduling algorithm, and the allocation of data sharing access rights include: According to the set storage capacity and number of storage nodes, the stored cargo transportation shared data is randomly initialized, and the corresponding storage location is allocated according to the cargo transportation shared data type; The stored cargo transportation shared data is optimized using a scheduling optimization algorithm, and the optimized cargo transportation shared data is updated using differential mutation operations. The distribution of cargo transportation shared data in the data sharing platform is dynamically adjusted based on the access frequency of the cargo transportation shared data and the load of the storage nodes. Cross-process the stored cargo transportation shared data with its access policy, and calculate the access rights of each cargo transportation shared data item based on the importance and priority of the cargo transportation shared data; Based on the access frequency and demand of cargo transportation shared data, matching cargo transportation shared data is selected for optimized storage; for frequently accessed cargo transportation shared data, its storage location and access path are optimized through cross-operation; According to the access requirements and permission configuration of cargo transportation shared data, the access rights of cargo transportation shared data are dynamically allocated, and the storage path and permission allocation of cargo transportation shared data are optimized through the taboo search method.

7. A method for sharing hazardous chemicals transportation data according to claim 6, characterized in that: The optimization of the stored cargo transportation shared data using the scheduling optimization algorithm includes: Use the data segmenter to segment the stored cargo transportation shared data, mark the attributes of each cargo transportation shared data item, and remove irrelevant and inactive data; Perform association rule mining on the cut cargo transportation shared data and define the elements that have association relationships between the same cargo transportation shared data items; For each shared data item in cargo transportation, calculate its associated weight in the frequent itemset and optimize its weight according to the occurrence probability of the frequent itemset; Traverse the link matrix of the entire frequent item set and calculate the weight value of each cargo transportation shared data item; Output the final optimized scheduling plan and determine the optimal stored cargo transportation sharing data.

8. A method for sharing hazardous chemicals transportation data according to claim 6, characterized in that: The method of optimizing the storage path and authority allocation of cargo transportation shared data by using the taboo search method includes: Set the maximum number of iterations of the tabu search algorithm and initialize the tabu table to empty; Generate new candidate storage paths through local search and calculate the fitness value of each candidate storage path; Determine whether each candidate storage path is in a taboo state, and select the candidate storage path with the best fitness value that is not in the taboo table; Compare the optimal candidate storage path that is not in the taboo table with the historical optimal storage path. If the fitness of the optimal candidate storage path that is not in the taboo table is the highest, then update the historical optimal storage path and the taboo table. Otherwise, select a non-tabu path and update the taboo table. Ignore the taboo status of the optimal candidate storage path, treat it as a new historical candidate storage path, add it to the current path and taboo table, and update the status of the taboo table; If the optimal candidate storage path does not surpass the historical optimal storage path, the candidate storage path with the highest fitness outside the taboo table is selected, the current storage path is updated and added to the taboo table, and the shared data storage path and permission allocation for cargo transportation are further optimized; Check whether the maximum number of iterations has been reached. If so, output the historical optimal candidate storage path; otherwise, continue the iterative optimization process. Output the historical optimal cargo transportation shared data storage path and permission allocation plan after taboo search optimization as the final cargo transportation shared data storage plan.

9. A method for sharing hazardous chemicals transportation data according to claim 8, characterized in that: The formula for calculating the weight value of each cargo transportation shared data item is: M(d a ) represents the weight value of the a-th cargo transportation shared data item; represents the initial weight value of the a-th cargo transportation shared data item; out(d b ) represents the number of data items associated with the b-th cargo transportation shared data item; In(d a ) represents the set of data items associated with the a-th cargo transportation shared data item; M(d b ) represents the weight value of the b-th cargo transportation shared data item; M represents the weight value of the cargo transportation shared data item; W represents the average association entropy weight value of the cargo transportation shared data item.

10. A hazardous chemicals cargo transportation data sharing system, used to implement the hazardous chemicals cargo transportation data sharing method according to any one of claims 1 to 9, characterized in that: The system includes: The data acquisition module is used to obtain hazardous chemical cargo transportation data, extract cargo transportation feature data, and generate cargo transportation shared data; Anomaly detection module, used to detect anomalies in cargo transportation shared data using anomaly detection algorithms to obtain safe cargo transportation shared data; The data sharing module is used to distribute and store secure cargo transportation shared data on the data sharing platform, dynamically allocate the stored cargo transportation shared data through the data scheduling algorithm, and allocate data sharing access rights.

Images