Cross-border ocean e-commerce multi-currency intelligent settlement method and system

Abstract

The invention relates to the technical field of block chain and Internet of Things data processing, and discloses a cross-border ocean e-commerce multi-currency intelligent settlement method and system, which effectively solves the problems of difficult goods authenticity verification and long settlement period in cross-border trade by constructing a time-space alignment and multi-modal verification mechanism based on physical fingerprints, and improves the efficiency of cross-border trade. A time axis is calibrated through container vibration sudden change characteristics, sensor accumulative errors are eliminated, empty container fraud, position counterfeiting and environment simulation attacks are accurately recognized at a physical bottom layer through thermodynamic passive attenuation analysis and kinematic sea wave spectrum verification, and the method is based on a thermal and dynamic two-dimension orthogonal confidence fusion algorithm. The block chain intelligent contract is driven to execute automatic fund delivery, real-time pre-advance payment of cross-border funds is realized on the premise of ensuring financial level risk control security, the trade mutual trust cost is remarkably reduced, and the supply chain fund turnover efficiency is improved.

Description

Technical Field

[0001] This application relates to the fields of blockchain and IoT data processing technology, and in particular to a method and system for multi-currency intelligent settlement in cross-border ocean e-commerce. Background Technology

[0002] With the deepening of global economic integration, cross-border e-commerce has become an important engine for international trade growth, characterized by high-frequency, fragmented orders and extremely high requirements for timely capital turnover. In cross-border ocean trade, due to the inherently long cycle of maritime logistics, goods often take weeks or even months to travel from the port of shipment to the port of destination, involving complex port operations, customs inspections, and multiple transshipment stages. This lengthy logistics delivery cycle places enormous pressure on the capital tied up by small and medium-sized export enterprises at the upstream of the supply chain. Therefore, achieving rapid settlement services with immediate payment upon shipment has become an urgent need for the industry. Traditionally, the settlement of funds in this type of trade mainly relies on letters of credit or bank collection. Financial institutions or settlement platforms verify the authenticity of the transaction background by reviewing trade documents such as bills of lading, commercial invoices, packing lists, and customs declarations, and then execute fund transfers or advance payments accordingly.

[0003] However, under the existing technological system, cross-border trade settlement risk control faces the technical challenge of a severe disconnect between the physical world of goods status and the digital world of document information. Although existing electronic solutions widely employ optical character recognition technology to digitize paper documents and utilize blockchain technology to ensure the immutability of data after it is uploaded to the chain, these methods only solve the security issues in the data transmission and storage stages, but cannot solve the problem of verifying the authenticity of the data source, the so-called oracle problem. Specifically, digital documents are extremely susceptible to being tampered with by image processing tools at the source of generation, or fraudulent phenomena such as separation of documents and goods may occur, where the uploaded document information is perfect, but the actual goods loaded in the container are seriously inconsistent with the declared content or even the container is empty. Existing risk control systems, lacking the ability to perceive and cross-verify physical characteristic data during logistics and transportation, find it difficult to identify such source fraud from a technical perspective. They are forced to adopt a conservative static credit granting model, that is, to rely excessively on static indicators such as the company's registered capital or historical turnover, and cannot conduct risk assessment based on the dynamic physical facts of a single transaction. This technological limitation has resulted in a large number of high-quality merchants with genuine but small-scale transactions failing to pass risk control audits. This not only leads to low settlement efficiency but also prevents the allocation of financial resources from accurately matching the actual trade and logistics process, severely hindering the large-scale expansion of cross-border ocean e-commerce business. Summary of the Invention

[0004] This application proposes a multi-currency intelligent settlement method and system for cross-border ocean e-commerce to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this application adopts the following technical solution: a multi-currency intelligent settlement method for cross-border ocean e-commerce, comprising the following steps:

[0006] Step S1: Acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. Extract physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Construct a time axis mapping model based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features. Use the time axis mapping model to calibrate the relative timestamps of the relative temporal sensing data inside the container to absolute world time and generate a spatiotemporally aligned physical fingerprint matrix.

[0007] Step S2: Extract the temperature change sequence during the passive temperature rise period from the spatiotemporal aligned physical fingerprint matrix generated in Step S1, calculate the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence, obtain the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculate the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generate thermodynamic consistency verification index.

[0008] Step S3: Extract triaxial acceleration data from the spatiotemporal aligned physical fingerprint matrix generated in step S1, perform frequency domain feature transformation on the triaxial acceleration data to obtain the measured environmental spectrum, verify whether there are wave swell characteristic frequencies in the measured environmental spectrum, calculate the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system, and generate kinematic scene verification indicators.

[0009] Step S4: Input the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 into the multimodal confidence assessment model, calculate the comprehensive transaction confidence value, and only when the comprehensive transaction confidence value is greater than the preset security settlement threshold, drive the blockchain smart contract state machine to switch from the locked state to the execution state, and call the cross-border payment gateway to execute the pre-payment transfer operation of the multi-currency fund pool.

[0010] Furthermore, in step S1, the specific operation of extracting physical vibration abrupt change features from the relative temporal sensing data inside the container is as follows:

[0011] A sliding window short-time Fourier transform operation is performed on the triaxial acceleration time-series signal contained in the relative time-series sensing data inside the container to transform the acceleration numerical sequence in the time domain into a power spectral density distribution matrix in the frequency domain. An integral operation is performed on the energy values ​​in the power spectral density distribution matrix that are located in the preset low-frequency gravity wave frequency band to obtain the low-frequency energy component. The numerical ratio between the low-frequency energy component and the sum of the total spectral energy of the entire frequency band is calculated to generate a low-frequency energy proportion factor sequence.

[0012] Monitor the temporal numerical changes of the low-frequency energy proportion factor sequence, identify the numerical mutation time points where the low-frequency energy proportion factor sequence experiences a step decrease in numerical value, determine the numerical mutation time points as the physical phase transition points where the low-frequency surge vibration mode dominated by fluid medium changes to the broadband mechanical shock vibration mode dominated by rigid medium, and mark the physical phase transition points as physical medium switching anchor points.

[0013] Traverse all physical medium switching anchor points to construct a set of physical vibration abrupt change features;

[0014] In the process of constructing the physical vibration abrupt change feature set, a filtering logic based on the hydrodynamic spectrum morphology is executed to identify and remove deep-sea wind and wave vibration data with high amplitude values ​​but whose spectrum structure exhibits the characteristics of a gravity wavelength periodic narrow band. Only port operation vibration data with a spectrum structure exhibiting broadband pulse characteristics are retained. The physical medium switching anchor point after filtering is identified as the physical vibration abrupt change feature, thereby establishing a physical side observation benchmark for time axis calibration.

[0015] Furthermore, in step S1, the specific operation of constructing a time-axis mapping model based on the physical vibration abrupt change characteristics and port operation event characteristics is as follows:

[0016] A random sampling consistency iterative strategy is implemented to extract feature point pairs from the physical vibration mutation feature set and the port operation event feature set to construct candidate linear mapping hypothesis models. In the process of iteratively calculating the candidate linear mapping hypothesis models, the physical frequency stability characteristics of the quartz crystal oscillator are introduced as hard physical constraint boundaries. Numerical filtering operations are performed to directly eliminate non-physical models in the candidate linear mapping hypothesis models whose clock drift rate parameters exceed the nominal physical error limit of the quartz crystal oscillator, and retain the set of interior points that conform to the physical laws of the hardware.

[0017] A spatiotemporal topological consistency energy equation with a robust kernel function is constructed based on the set of interior points. Variational methods are performed to solve for the global minimum of the spatiotemporal topological consistency energy equation. The global minimum is analyzed to obtain the optimal clock drift rate parameter and the optimal initial time deviation parameter. The optimal clock drift rate parameter and the optimal initial time deviation parameter are used to establish a time axis mapping model that can linearly map relative machine time to absolute world time.

[0018] Furthermore, in step S2, the specific operation of extracting the temperature change sequence during the passive temperature rise period from the spatiotemporally aligned physical fingerprint matrix generated in step S1, and calculating the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence is as follows:

[0019] Morphological opening filtering is performed on the container internal temperature numerical sequence contained in the spatiotemporal aligned physical fingerprint matrix to eliminate high-frequency thermal noise interference caused by refrigeration compressor oscillation, thereby obtaining a smooth temperature change sequence;

[0020] Perform a first-order difference operation on the smoothed temperature change sequence to generate a temperature change rate sequence, and further perform a second-order difference operation to generate a temperature acceleration sequence.

[0021] Traverse the smooth temperature change sequence and identify continuous time segments that simultaneously satisfy the condition that the temperature change rate sequence value is positive and the temperature acceleration sequence value is non-positive. If the duration of the continuous time segment exceeds the preset thermal relaxation time threshold, then the continuous time segment is marked as a passive temperature rise period. In this way, the natural cooling process can be locked by using the convex function geometric features of the temperature curve without needing to obtain the power status data of the refrigerator.

[0022] Extract the instantaneous temperature values ​​and corresponding instantaneous temperature change rate values ​​during the passive temperature rise period to construct a phase plane differential regression dataset;

[0023] The random sampling consensus algorithm was used to perform linear regression on the phase plane differential regression dataset to fit a linear regression equation with instantaneous temperature values ​​as independent variables and instantaneous temperature change rate values ​​as dependent variables.

[0024] Extract the slope value of the linear regression equation, calculate the opposite of the reciprocal of the slope value, and determine the calculation result as the measured thermal response parameter characterizing the thermal resistance and capacitance properties of objects inside the container.

[0025] Furthermore, in step S2, the specific operations for obtaining the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration document, calculating the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generating the thermodynamic consistency verification index are as follows:

[0026] The data interface is used to parse digital trade customs declaration documents to obtain customs code information of the declared goods. The customs code information is then used to search in a pre-set multimodal thermophysical property knowledge graph to obtain the lower limit value of the theoretical thermal response parameter that matches the physical properties of the declared goods. The lower limit value of the theoretical thermal response parameter is then confirmed as the boundary benchmark of the theoretical thermal response parameter range.

[0027] Calculate the natural logarithm of the measured thermal response parameter and the natural logarithm of the lower limit of the theoretical thermal response parameter, and calculate the difference between the two. When the measured thermal response parameter is less than the lower limit of the theoretical thermal response parameter, the absolute value of the difference is determined as the numerical deviation degree, which is used to characterize the degree of physical deviation between the thermal inertia of the actual loaded object inside the container and the theoretical thermal inertia of the declared cargo at the order of magnitude level, thereby eliminating the problem of numerical linear comparison distortion caused by the large difference between the thermal capacity of air and the thermal capacity of the actual cargo.

[0028] The numerical deviation is input into a normalized evaluation model based on the hyperbolic tangent function. The numerical deviation is dimensionless using a preset loading rate fluctuation tolerance parameter. The processed result is then subjected to nonlinear penalty calculation using a preset discrimination order parameter. Finally, a thermodynamic consistency verification index between zero and one is generated.

[0029] The closer the value of this indicator is to one, the higher the physical authenticity of the goods; the closer the value of this indicator is to zero, the greater the risk of empty container fraud.

[0030] Furthermore, in step S3, triaxial acceleration data is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1, and frequency domain feature transformation is performed on the triaxial acceleration data to obtain the measured environmental spectrum. The specific operation to verify whether there are wave swell characteristic frequencies in the measured environmental spectrum is as follows:

[0031] Long-window mean filtering is performed on the triaxial acceleration data extracted from the spatiotemporal aligned physical fingerprint matrix to obtain the steady-state gravity vector representing the direction of gravity at the Earth's core. A spatial rotation mapping matrix is ​​constructed based on the steady-state gravity vector. The triaxial acceleration data is then projected onto a virtual geocentric coordinate system using the spatial rotation mapping matrix to separate the acceleration sequence in the absolute vertical direction.

[0032] Welch power spectral density estimation is performed on the absolute vertical acceleration sequence to generate the measured environmental spectrum. Shannon information entropy calculation is performed on the measured environmental spectrum to generate the Shannon information entropy value. The Shannon information entropy value is defined as the fluid disorder index.

[0033] Calculate the energy integral value within the preset wave swell frequency range in the measured environmental spectrum, and calculate the proportion of the energy integral value to the total energy integral value of the entire measured environmental spectrum.

[0034] The numerical comparison logic is executed to determine whether the fluid disorder index is greater than the preset minimum fluid disorder threshold and whether the numerical ratio is greater than the preset physical energy ratio threshold. Only when the fluid disorder index is greater than the preset minimum fluid disorder threshold and the numerical ratio is greater than the preset physical energy ratio threshold, a verification pass signal is generated to confirm the presence of wave surge characteristic frequencies in the measured environmental spectrum; otherwise, an environmental anomaly alarm signal is generated.

[0035] Furthermore, in step S3, the specific operation of calculating the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation state information in the absolute spatiotemporal trajectory data of the external logistics system, and generating kinematic scenario verification indicators, is as follows:

[0036] Monitor the ship's heading change rate in the absolute spatiotemporal trajectory data of the external logistics system. When the ship's heading change rate exceeds the preset maneuver event trigger threshold, initiate the rigid body coupling verification calculation process.

[0037] Perform multiplication operations to calculate the product of the ship's heading rate of change and the ship's speed, and generate a theoretical centrifugal acceleration modulus sequence;

[0038] Obtain the absolute horizontal plane acceleration sequence after processing by the spatial rotation mapping matrix, and calculate the vector modulus of the absolute horizontal plane acceleration sequence to generate the measured resultant horizontal acceleration modulus sequence;

[0039] Within the preset time delay scanning window, a sliding cross-correlation operation is performed on the theoretical centrifugal acceleration modulus sequence and the measured horizontal combined acceleration modulus sequence. The optimal time delay parameter corresponding to the maximum value of the cross-correlation function is searched, and the maximum value of the cross-correlation function is determined as the scale-invariant coupling correlation.

[0040] The scale-invariant coupling correlation is input into the nonlinear activation function model and normalized to obtain the normalized coupling probability value. The validity judgment result of the fluid disorder index is obtained. A logical AND operation is performed on the normalized coupling probability value and the validity judgment result of the fluid disorder index to generate the kinematic scenario verification index. When the fluid disorder index is judged to be invalid, the kinematic scenario verification index is forcibly set to zero.

[0041] Furthermore, in step S4, the specific operation of inputting the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value is as follows:

[0042] Multiplication is performed on the thermodynamic consistency verification index and the kinematic scenario verification index to obtain the index product. The square root operation is then performed on the index product to obtain the geometric mean basis value. The physical existence co-occurrence constraint is constructed by utilizing the sensitive response characteristic of the geometric mean operation to zero values.

[0043] The absolute value of the difference between the thermodynamic consistency verification index and the kinematic scenario verification index is calculated. The absolute value of the difference is defined as the modal inconsistency dispersion. The modal inconsistency dispersion is input into the hyperbolic tangent inverse activation function model. The modal inconsistency dispersion is generated by performing nonlinear amplification operation on the modal inconsistency dispersion using a preset risk sensitivity coefficient. The consistency weight factor is generated based on the difference between the numerical value and the paradox penalty coefficient.

[0044] Determine whether the minimum value of the thermodynamic consistency verification index and the kinematic scenario verification index is lower than the preset minimum physical existence threshold. If the minimum value is lower than the preset minimum physical existence threshold, generate a fuse multiplier with a value of zero. If the minimum value is not lower than the preset minimum physical existence threshold, generate a fuse multiplier with a value of one.

[0045] Perform a multiplication operation on the geometric mean base value, the consistency weight factor, and the circuit breaker multiplier, and determine the product result as the comprehensive transaction confidence value.

[0046] Furthermore, in step S4, only when the comprehensive transaction confidence value is greater than the preset security settlement threshold, the blockchain smart contract state machine is driven to switch from the locked state to the execution state, and the specific operation of calling the cross-border payment gateway to execute the pre-payment transfer operation of the multi-currency fund pool is as follows:

[0047] A numerical quantization mapping operation is performed on the comprehensive transaction confidence value, which ranges from zero to one. By multiplying the comprehensive transaction confidence value by a preset base point magnification factor and performing a floor operation, an integer confidence rank value compatible with the blockchain virtual machine is generated.

[0048] The integer confidence rank is input into the pre-deployed blockchain smart contract finite state machine, which includes locked state, execution state, and penalty state. The execution state transition judgment logic is as follows: when the integer confidence rank is detected to be greater than the preset secure settlement integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the execution state, triggering the oracle interface to call the cross-border payment gateway to lock the corresponding fiat currency amount in the multi-currency fund pool and perform a pre-payment transfer operation to the seller's account.

[0049] When the integer confidence rank value is detected to be less than the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the penalty state, triggering the default record on-chain logic and deducting the reputation score of the corresponding IoT node.

[0050] When the integer confidence rank value is detected to be between the preset secure settlement integer threshold and the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to remain locked and generate a manual review request signal.

[0051] A multi-currency intelligent settlement system for cross-border ocean e-commerce includes a spatiotemporal aligned physical fingerprint generation module, a thermodynamic consistency verification module, a kinematic scene verification module, and a smart contract settlement control module, wherein;

[0052] The spatiotemporal aligned physical fingerprint generation module is used to acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. It extracts physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features, a time axis mapping model is constructed. The relative timestamps of the relative temporal sensing data inside the container are calibrated to absolute world time using the time axis mapping model to generate a spatiotemporal aligned physical fingerprint matrix.

[0053] The thermodynamic consistency verification module is used to extract the temperature change sequence during the passive temperature rise period from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, calculate the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence, obtain the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculate the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generate thermodynamic consistency verification index.

[0054] The kinematic scene verification module is used to extract triaxial acceleration data from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, perform frequency domain feature transformation on the triaxial acceleration data to obtain the measured environment spectrum, verify whether there are wave swell characteristic frequencies in the measured environment spectrum, calculate the coupling correlation between the rigid body motion response characteristics in the measured environment spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system, and generate kinematic scene verification indicators.

[0055] The smart contract settlement control module is used to input the thermodynamic consistency verification index generated by the thermodynamic consistency verification module and the kinematic scenario verification index generated by the kinematic scenario verification module into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value. Only when the comprehensive transaction confidence value is greater than the preset safe settlement threshold, the blockchain smart contract state machine is driven to switch from the locked state to the execution state, and the cross-border payment gateway is called to execute the pre-payment transfer operation of the multi-currency fund pool.

[0056] The beneficial effects of this invention are as follows:

[0057] This invention effectively solves the problems of difficulty in verifying the authenticity of goods and long settlement cycles in cross-border trade by constructing a spatiotemporal alignment and multimodal verification mechanism based on physical fingerprints. It uses the abrupt vibration characteristics of containers to calibrate the time axis, eliminating the cumulative error of sensors. Through thermodynamic passive attenuation analysis and kinematic wave spectrum verification, it accurately identifies empty container fraud, location spoofing, and environmental simulation attacks at the physical level. Based on the orthogonal confidence fusion algorithm of thermal and dynamic dimensions, it drives the blockchain smart contract to execute automated fund delivery. Under the premise of ensuring financial-grade risk control security, it realizes the instant prepayment of cross-border funds, significantly reduces the cost of trade mutual trust, and improves the efficiency of supply chain capital turnover. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort:

[0059] Figure 1 This is a flowchart of the method of the present invention;

[0060] Figure 2 This is a system framework diagram of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1

[0063] like Figure 1 As shown, this invention provides a multi-currency smart settlement method for cross-border e-commerce, comprising the following steps:

[0064] Step S1: Acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. Extract physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Construct a time axis mapping model based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features. Use the time axis mapping model to calibrate the relative timestamps of the relative temporal sensing data inside the container to absolute world time and generate a spatiotemporally aligned physical fingerprint matrix.

[0065] Specifically, in step S1, the specific operation of extracting physical vibration abrupt change features from the relative time-series sensing data inside the container is as follows: perform a sliding window short-time Fourier transform operation on the triaxial acceleration time-series signal contained in the relative time-series sensing data inside the container, and transform the acceleration numerical sequence in the time domain dimension into a power spectral density distribution matrix in the frequency domain dimension.

[0066] In practice, the sampling frequency of the inertial measurement unit is first configured. In this embodiment, the sampling frequency is set to 50 Hz to 100 Hz to ensure that the Nyquist sampling theorem requirement for analyzing low-frequency gravity wave signals is met.

[0067] Define a sliding window for signal interception. The time span of the sliding window is preferably set to five to ten seconds, and the overlap ratio between adjacent windows is set to fifty percent. The physical basis for selecting the range of five to ten seconds is that the wave period of typical ocean swells in physical oceanography is usually distributed between eight and twenty seconds. This window length can ensure that at least half of the wave period is covered.

[0068] The acceleration data within the window is weighted and truncated using the Hamming window function, and then a discrete Fourier transform is performed.

[0069] The purpose of this processing step is to deconstruct the vibration amplitude signal, which originally varies with time, into an energy density signal that varies with frequency, thereby separating the frequency response characteristics of fluid media and rigid media to vibration energy transmission, which are completely different in the frequency domain.

[0070] Subsequently, an integral operation is performed on the energy values ​​in the power spectral density distribution matrix located within the preset low-frequency gravity wave frequency band to obtain the low-frequency energy components. The numerical ratio between the low-frequency energy components and the sum of the total spectral energy of the entire frequency band is calculated to generate a low-frequency energy proportion factor sequence.

[0071] In this embodiment, the frequency range of the preset low-frequency gravity wave band is strictly locked between 0.05 Hz and 0.2 Hz, which represents the frequency domain range where the energy of deep-sea swells is most concentrated. The value is based on the gravity wave dispersion relation theory in fluid dynamics.

[0072] By accumulating the energy spectral density values ​​within the frequency band through discrete integration and dividing them by the total energy integral value of the entire frequency band from 0 Hz to 50 Hz, a dimensionless low-frequency energy proportion factor between the values ​​of zero and one is obtained.

[0073] The physical meaning of this dimensionless low-frequency energy proportion factor is a normalized index characterizing the degree to which the current environment is dominated by ocean waves. When the dimensionless low-frequency energy proportion factor is close to a value of one, it means that the container is being lifted and swayed by ocean waves; when the dimensionless low-frequency energy proportion factor is close to a value of zero, it means that the container is in a static land environment or a mechanical high-frequency vibration environment.

[0074] Next, monitor the temporal numerical changes of the low-frequency energy proportion factor sequence, identify the numerical mutation time point where the low-frequency energy proportion factor sequence experiences a step decrease in numerical value, determine the numerical mutation time point as the physical phase transition point where the low-frequency surge vibration mode dominated by the fluid medium changes to the broadband mechanical shock vibration mode dominated by the rigid medium, and mark the physical phase transition point as the physical medium switching anchor point.

[0075] In practice, the system has a logic gate threshold. A sudden change is determined to have occurred only when the value of the low-frequency energy proportion factor is detected to decrease monotonically from a high value greater than 0.8 to a low value less than 0.2 within a time span of less than 30 seconds.

[0076] The 30-second time limit is based on statistical data from the standard operating cycle of port quay cranes grabbing and placing containers; while the drop range requirement of 0.8 to 0.2 is to effectively eliminate interference from minor changes in sea conditions.

[0077] The purpose of this logic is to accurately capture the moment when the medium in a container changes from the fluid environment of the ship's hold to the rigid ground of the port in the physical world, thereby avoiding the risk of false alarms caused by simply relying on the magnitude of vibration amplitude.

[0078] Furthermore, all physical medium switching anchor points are traversed to construct a set of physical vibration abrupt change features;

[0079] In the process of constructing the physical vibration abrupt change feature set, a filtering logic based on the hydrodynamic spectrum morphology is executed to identify and remove deep-sea wind and wave vibration data with high amplitude values ​​but whose spectrum structure exhibits the characteristics of a gravity wavelength periodic narrow band. Only port operation vibration data with a spectrum structure exhibiting broadband pulse characteristics are retained. The physical medium switching anchor point after filtering is identified as the physical vibration abrupt change feature, thereby establishing a physical side observation benchmark for time axis calibration.

[0080] Specifically, the system calculates the spectral bandwidth value corresponding to each anchor point, such as the negative 3 dB bandwidth. If the bandwidth value is less than 0.5 Hz, it indicates that although the energy is huge, the frequency component is singular, which is consistent with the narrow band characteristics of deep-sea wind and waves. The system marks it as invalid interference and removes it. If the bandwidth value covers a wide frequency band from low frequency to high frequency and exhibits white noise characteristics, it indicates that it is consistent with the wide frequency characteristics of mechanical impact and is confirmed as an effective physical vibration mutation characteristic.

[0081] The physical basis of this filtering logic lies in the essential difference in the spectral morphology between wind and wave impacts and mechanical impacts. Its function is to ensure that the data source input to the subsequent model has a very high degree of physical confidence.

[0082] In step S1, the specific operation of constructing the time axis mapping model based on the physical vibration mutation characteristics and port operation event characteristics is as follows: execute the random sampling consistency iteration strategy, extract feature point pairs from the physical vibration mutation feature set and the port operation event feature set to construct candidate linear mapping hypothesis models, introduce the physical frequency stability characteristics of quartz crystal oscillators as hard physical constraint boundaries during the iterative calculation of candidate linear mapping hypothesis models, and perform numerical filtering operations to directly eliminate non-physical models in the candidate linear mapping hypothesis models whose clock drift rate parameters exceed the nominal physical error limit of quartz crystal oscillators, and retain the set of interior points that conform to the physical laws of hardware.

[0083] In this embodiment, the nominal physical error limit of the quartz crystal oscillator is specifically set to 50 parts per million. This parameter represents the maximum frequency drift of the industrial-grade quartz crystal oscillator under extreme temperatures, and its value is derived from the hardware specification of the Internet of Things device.

[0084] This means that when the system fits the slope parameter of the linear equation, it requires that the slope must fall within the closed interval of 0.99995 to 1.00005.

[0085] The purpose of this constraint mechanism is to transform the physical characteristics of the hardware into hard boundaries for the algorithm, thereby preventing the algorithm from fitting an erroneous model that violates physical common sense, such as time flow doubling or time reversal, when the anchor data is sparse or there is noise interference. This ensures the legality of the time axis mapping results at the physical level.

[0086] Finally, a spatiotemporal topological consistency energy equation containing a robust kernel function is constructed based on the set of interior points. Variational methods are performed to solve for the global minimum of the spatiotemporal topological consistency energy equation. The global minimum is analyzed to obtain the optimal clock drift rate parameter and the optimal initial time deviation parameter. The optimal clock drift rate parameter and the optimal initial time deviation parameter are used to establish a time axis mapping model that can linearly map relative machine time to absolute world time.

[0087] In specific implementation, the robust kernel function preferably adopts the Huber loss function, which increases quadratically for small errors and linearly for large errors. The core parameter corresponding to this function, namely the time uncertainty scale parameter, is set to sixty seconds in this embodiment. The basis for setting this value is that port hoisting operations usually have an operation time window of thirty to sixty seconds. Deviations smaller than this time scale are considered reasonable operation errors, while deviations larger than this scale are considered outlier noise.

[0088] The energy equation is optimized through multiple iterations using the variational method, minimizing the sum of time residuals at all interior points, thereby solving for the globally optimal clock drift rate parameter and initial time deviation parameter.

[0089] The model's role is to use statistical methods to eliminate the influence of nonlinear micro-fluctuations in crystal oscillator temperature changes, and to establish a precise and rigid mapping of the relative machine time in the enclosed environment inside the container onto the absolute world time axis of the external logistics system, thus eliminating the cumulative timing error in long-cycle voyages.

[0090] Further, in step S2, the temperature change sequence during the passive temperature rise period is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1. Based on the temperature change sequence, the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container are calculated. The theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration is obtained. The numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range is calculated, and a thermodynamic consistency verification index is generated.

[0091] In step S2, the temperature change sequence during the passive temperature rise period is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1. The specific operation of calculating the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence is as follows: morphological opening operation filtering is performed on the temperature value sequence inside the container contained in the spatiotemporal aligned physical fingerprint matrix to eliminate high-frequency thermal noise interference generated by the oscillation of the refrigeration compressor, thereby obtaining a smooth temperature change sequence.

[0092] In specific implementation, the structural element required for morphological opening operation is first defined. In this embodiment, the time span of the structural element is preferably set to five minutes. The physical basis for selecting five minutes as the time span of the structural element is that the start-stop vibration cycle of the container refrigeration compressor is usually less than five minutes, while the temperature inertia change cycle of the overall thermal environment inside the container is much greater than five minutes.

[0093] The structural element is used to first perform morphological erosion operation on the temperature numerical sequence inside the container to remove positive impulse noise. Then, the structural element is used to perform morphological dilation operation to restore the main shape of the signal. In this way, while preserving the macroscopic trend of the temperature curve, high-frequency spikes caused by sensor electromagnetic interference or local cold air convection are filtered out, ensuring the stability of subsequent differential calculations.

[0094] Subsequently, a first-order difference operation is performed on the smoothed temperature change sequence to generate a temperature change rate sequence, and a second-order difference operation is further performed to generate a temperature acceleration sequence.

[0095] Next, the smooth temperature change sequence is traversed to identify continuous time segments that simultaneously satisfy the condition that the temperature change rate sequence value is positive and the temperature acceleration sequence value is non-positive. If the duration of the continuous time segment exceeds the preset thermal relaxation time threshold, the continuous time segment is marked as a passive temperature rise period. In this way, the natural cooling process can be locked by using the convex function geometric features of the temperature curve without needing to obtain the power status data of the refrigerator.

[0096] In this embodiment, the preset thermal relaxation time threshold is specifically set to thirty minutes. The preset thermal relaxation time threshold represents the minimum stabilization time required for the thermodynamic system to recover from dynamic disturbance to quasi-static natural convection state.

[0097] The selection of 30 minutes as the preset thermal relaxation time threshold is based on statistical analysis of the heat transfer model of a standard container. Data segments shorter than 30 minutes may be affected by residual cooling from refrigeration unit shutdown and cannot accurately reflect the thermal inertia of the cargo.

[0098] The preset thermal relaxation time threshold acts as a logical gate in the algorithm, forcing the processing of only data segments with sufficient data volume and conforming to the natural physical temperature rise law, thereby eliminating abnormal temperature rise data caused by active heating or refrigerator failure at the physical level.

[0099] Next, the instantaneous temperature values ​​and corresponding instantaneous temperature change rate values ​​during the passive temperature rise period are extracted to construct a phase plane differential regression dataset. The random sampling consensus algorithm is used to perform linear regression on the phase plane differential regression dataset to fit a linear regression equation with instantaneous temperature values ​​as independent variables and instantaneous temperature change rate values ​​as dependent variables.

[0100] In this step, the iterative mechanism of the random sampling consensus algorithm is used to randomly select a subset from the phase plane differential regression dataset to estimate the parameters of the linear model, and calculate the distance from the remaining data points to the linear model. Points with a distance less than the preset tolerance are marked as interior points.

[0101] By maximizing the number of internal points, the multi-currency intelligent settlement system for cross-border ocean e-commerce can effectively resist occasional temperature fluctuations.

[0102] Finally, the slope value of the linear regression equation is extracted, and the opposite of the reciprocal of the slope value is calculated. The result of the calculation of the opposite of the reciprocal of the slope value is determined as the measured thermal response parameter characterizing the thermal resistance and capacitance properties of the objects inside the container.

[0103] The significance of calculating the inverse of the slope of a straight line lies in the fact that, according to the differential form of Newton's law of cooling, the rate of temperature rise is equal to the difference between the current temperature and the ambient temperature divided by the thermal time constant. The instantaneous temperature change rate of a closed system during the passive temperature rise period has a linear relationship with the instantaneous temperature value. The slope of this linear relationship is physically strictly equal to the negative value of the reciprocal of the system's thermal time constant.

[0104] Regardless of changes in the external ambient temperature, the slope of the straight line remains constant. Therefore, by calculating the inverse of the slope of the straight line, the inherent thermal inertia parameter of the system, independent of ambient temperature disturbances, is decoupled.

[0105] Simultaneously, the intercept value of the linear regression equation is extracted and divided by the absolute value of the slope value to blindly calculate the equivalent external environmental temperature during the time period. The temperature is then verified for consistency with the external meteorological trajectory data aligned with step S1. If the deviation exceeds a preset threshold, it is determined to be environmental simulation fraud. The inherent thermal inertia parameter of the system directly reflects the product of the total mass of the cargo inside the container and its specific heat capacity.

[0106] After obtaining the robust measured thermal response parameters, step S2 involves obtaining the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculating the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generating a thermodynamic consistency verification index. The specific operation is as follows: the digital trade customs declaration is parsed through the data interface to obtain the customs code information of the declared commodity, and the customs code information is used to search in the pre-set multimodal thermophysical property knowledge graph to obtain the lower limit value of the theoretical thermal response parameter that matches the physical properties of the declared commodity. The lower limit value of the theoretical thermal response parameter is then confirmed as the boundary benchmark of the theoretical thermal response parameter range.

[0107] For example, when the declared commodity is frozen beef, the lower limit of the theoretical thermal response parameter retrieved in this embodiment is 4.5 hours. The value of 4.5 hours comes from the statistical lower limit of thermal test data of a fully loaded frozen meat container under standard operating conditions. 4.5 hours represents the minimum thermal resistance and capacitance attribute that frozen beef products should have.

[0108] Subsequently, the natural logarithm of the measured thermal response parameter and the natural logarithm of the lower limit of the theoretical thermal response parameter are calculated respectively. The difference between the natural logarithm of the measured thermal response parameter and the natural logarithm of the lower limit of the theoretical thermal response parameter is calculated. When the measured thermal response parameter is less than the lower limit of the theoretical thermal response parameter, the absolute value of the difference is determined as the numerical deviation degree, which is used to characterize the degree of physical deviation between the thermal inertia of the actual loaded object inside the container and the theoretical thermal inertia of the declared cargo at the order of magnitude level. This eliminates the problem of numerical linear comparison distortion caused by the large difference between the thermal capacity of air and the thermal capacity of the actual cargo.

[0109] This step introduces a logarithmic domain transformation because there is an order of magnitude difference in physical values ​​between an empty chamber (mainly air, with a thermal response parameter of about 0.2 hours) and a full-loaded chamber (with a thermal response parameter of about 5 hours). Direct linear subtraction cannot reasonably quantify the difference between the empty and full-loaded chambers, while logarithmic difference can transform the exponential physical difference into a linear metric.

[0110] Finally, the numerical deviation is input into the normalized evaluation model based on the hyperbolic tangent function. The numerical deviation is dimensionless using the preset loading rate fluctuation tolerance parameter, and the processed result is nonlinearly penalized using the preset discrimination order parameter. Finally, a thermodynamic consistency verification index between zero and one is generated.

[0111] In this embodiment, the preset load rate fluctuation tolerance parameter is set to 0.3.

[0112] The preset load factor fluctuation tolerance parameter allows for approximately 30% void or density fluctuation in cargo during loading, which is determined based on average load factor data from the logistics industry.

[0113] The preset discrimination order parameter is set to four. The preset discrimination order parameter is used to adjust the steepness of the edge of the evaluation function. The preset discrimination order parameter of four means that the normalized evaluation model has extremely high sensitivity to numerical deviations that exceed the tolerance range, and can achieve a quasi-step penalty effect.

[0114] The specific calculation logic is as follows: the numerical deviation is divided by the preset loading rate fluctuation tolerance parameter to obtain the relative deviation ratio. The relative deviation ratio is then subjected to a fourth power operation to amplify the significant error. Finally, the hyperbolic tangent function is used to map the calculation result into a normalized score.

[0115] The closer the thermodynamic consistency verification index value is to the value of one, the higher the physical authenticity of the goods. The closer the thermodynamic consistency verification index value is to the value of zero, the higher the risk of empty container fraud.

[0116] This thermodynamic consistency verification index will serve as the trust weight for smart contract-triggered settlement in the subsequent step S4, directly affecting the exchange rate locking strategy and fund transfer ratio of the cross-border multi-currency fund pool, thereby realizing a rigid mapping from physical objective laws to financial settlement credit.

[0117] Further, in step S3, three-axis acceleration data is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1, frequency domain feature transformation is performed on the three-axis acceleration data to obtain the measured environmental spectrum, the presence of wave swell characteristic frequencies in the measured environmental spectrum is verified, the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system is calculated, and kinematic scene verification indicators are generated.

[0118] In step S3, triaxial acceleration data is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1. Frequency domain feature transformation is performed on the triaxial acceleration data to obtain the measured environmental spectrum. The specific operation to verify whether there are characteristic frequencies of ocean waves in the measured environmental spectrum is as follows: a long window mean filtering operation is performed on the triaxial acceleration data extracted from the spatiotemporal aligned physical fingerprint matrix to obtain the steady-state gravity vector representing the direction of gravity at the Earth's center. A spatial rotation mapping matrix is ​​constructed based on the steady-state gravity vector. The triaxial acceleration data is projected onto the virtual geocentric coordinate system using the spatial rotation mapping matrix to separate the absolute vertical acceleration sequence.

[0119] In specific implementation, the time window length of the long window mean filtering operation is first defined. In this embodiment, the time window length of the long window mean filtering operation is preferably set to ten minutes. The physical basis for selecting ten minutes as the time window length is that the roll and pitch motion of a ship on the sea surface is a periodic zero-mean oscillation around the equilibrium position. According to the ergodicity principle, as long as the statistical window is long enough, the time mean of the resultant acceleration vector measured by the inertial sensor will necessarily and uniquely point to the direction of gravity.

[0120] The steady-state gravity vector is obtained by calculating the arithmetic mean of the triaxial acceleration data within the time window. A spatial rotation mapping matrix is ​​constructed based on the Rodrigues rotation formula to project the data in the original coordinate system to a virtual geocentric coordinate system with the steady-state gravity vector as the Z-axis.

[0121] It is worth noting that the system is configured to dynamically and continuously execute the above-mentioned long-window mean filtering operation and spatial rotation mapping matrix construction process. The technical effect of dynamically and continuously executing the above process is that it can adapt to and compensate for long-period changes in the ship's tilt angle caused by the ship's adjustment of ballast water, fuel consumption, or cargo loading center of gravity shift in real time, ensuring that the Z-axis of the virtual geocentric coordinate system is always accurately aligned with the physical direction of gravity, thereby accurately decoupling the absolute vertical acceleration sequence that is only affected by the undulation of the sea waves.

[0122] Subsequently, Welch power spectral density estimation was performed on the absolute vertical acceleration sequence to generate the measured environmental spectrum. Shannon information entropy calculation was then performed on the measured environmental spectrum to generate the Shannon information entropy value, which was defined as the fluid disorder index.

[0123] In this embodiment, the Welch power spectral density estimation operation uses the Hanning window function to reduce spectral leakage. The calculated Shannon information entropy value is used as a fluid disorder index. The physical meaning of the fluid disorder index represents the randomness and disorder of the vibration signal energy distribution in the frequency domain.

[0124] For fluid motion in nature, especially ocean waves formed by long-distance wind action, the energy of ocean waves is dispersed over a wide frequency band and the phase is random, thus having a high Shannon information entropy value; while for mechanically simulated shaking tables driven by motors, the vibration of mechanically simulated shaking tables is usually based on regular sine waves or narrowband noise, and the energy is highly concentrated, thus having a low Shannon information entropy value.

[0125] The fluid disorder index plays a key role in identifying fraudulent high-level waveform simulations in the algorithm, effectively filling the gap in existing technologies that only compare frequency points while ignoring energy distribution patterns.

[0126] Next, the energy integral value within the preset wave swell frequency range in the measured environmental spectrum is calculated, and the proportion of the energy integral value to the total energy integral value of the entire measured environmental spectrum is calculated.

[0127] In this embodiment, the numerical range of the preset wave swell frequency range is strictly set to 0.05 Hz to 0.2 Hz. The basis for selecting this numerical range comes from the gravity wave spectrum theory in physical oceanography. This frequency band is the region where the energy of deep-sea swells generated by the movement of tens of millions of tons of seawater is most concentrated.

[0128] The energy percentage within this frequency band is calculated through integral operations to characterize whether the current environment meets the physical characteristics of deep-sea navigation, distinguishing it from the high-frequency vibration environment of shallow waters or land.

[0129] Finally, the cross-border ocean e-commerce multi-currency intelligent settlement system executes numerical comparison logic to determine whether the fluid disorder index is greater than the preset minimum fluid disorder threshold and whether the numerical ratio is greater than the preset physical energy ratio threshold. Only when the fluid disorder index is greater than the preset minimum fluid disorder threshold and the numerical ratio is greater than the preset physical energy ratio threshold, a verification pass signal is generated to confirm the presence of wave swell characteristic frequencies in the measured environmental spectrum; otherwise, an environmental anomaly alarm signal is generated.

[0130] In this embodiment, the preset minimum fluid disorder threshold is set to 2.0 at the base of the natural logarithm, and the preset physical energy ratio threshold is set to 0.3. The basis for setting the above thresholds is the boundary line obtained by comparing and analyzing a large amount of measured ocean wave data and mechanical shaking table data. Only when the signal simultaneously meets the two physical hard constraints of high entropy (proving that it is a random fluid) and high energy in a specific frequency band (proving that it is a deep-sea swell), the system confirms that the environment is real.

[0131] In step S3, the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system is calculated. The specific operation to generate the kinematic scenario verification index is as follows: monitor the ship heading change rate value in the absolute spatiotemporal trajectory data of the external logistics system. When the ship heading change rate value is detected to exceed the preset maneuver event trigger threshold, the rigid body coupling verification calculation process is started.

[0132] In this embodiment, the preset maneuver event trigger threshold is set to 0.5 degrees per second. The basis for setting this value is that when a ship is cruising in a straight line, it will have a slight yaw due to the influence of waves, but it will not generate a significant centrifugal force. If the coupling degree is forcibly calculated at this time, the mathematical denominator will approach zero, thus causing calculation divergence.

[0133] A preset maneuver event trigger threshold is used as a logic gating mechanism to ensure that the system only activates the verification process when the ship undergoes a significant turning maneuver and generates an inertial force response with a sufficient signal-to-noise ratio, thus guaranteeing the numerical stability of the calculation.

[0134] Subsequently, a multiplication operation is performed to calculate the product of the ship's heading rate of change and the ship's speed, generating a theoretical centrifugal acceleration modulus sequence; the absolute horizontal plane acceleration sequence processed by the spatial rotation mapping matrix is ​​obtained, and the vector modulus of the absolute horizontal plane acceleration sequence is calculated to generate a measured horizontal resultant acceleration modulus sequence.

[0135] Next, within a preset time delay scanning window, the cross-border ocean e-commerce multi-currency intelligent settlement system performs a sliding cross-correlation operation on the theoretical centrifugal acceleration modulus sequence and the measured horizontal combined acceleration modulus sequence, searches for the optimal time delay parameter corresponding to the maximum value of the cross-correlation function, and determines the maximum value of the cross-correlation function as the scale-invariant coupling correlation.

[0136] In this embodiment, considering that the measured horizontal combined acceleration modulus sequence comes from a high-frequency inertial sensor (usually greater than 50 Hz), while the theoretical centrifugal acceleration modulus sequence comes from low-frequency automatic identification system data (usually once every few seconds to tens of seconds), before performing the sliding cross-correlation operation, the measured horizontal combined acceleration modulus sequence is first subjected to low-pass filtering, with the cutoff frequency set to 0.5 Hz, and downsampling is performed to make the time sampling rate of the measured horizontal combined acceleration modulus sequence consistent with the time sampling rate of the theoretical centrifugal acceleration modulus sequence. This signal preprocessing step eliminates the calculation error caused by the inconsistency of sequence density.

[0137] The preset time delay scanning window is set to a range of -300 seconds to +300 seconds. This range is based on the fact that the transmission and processing of data from the satellite automatic identification system typically involves a time delay of minutes.

[0138] The optimal time alignment point is automatically searched through sliding cross-correlation calculation, and the similarity between the waveforms is calculated. It is worth noting that this calculation utilizes the scale invariance principle of rigid body dynamics: although the radius of the container's specific installation position on the ship is unknown, making it impossible to verify the absolute amplitude of the centrifugal force, the fluctuations of the inertial force waveform felt by the container on the time axis must be strictly synchronized with the ship's maneuvering. Therefore, the maximum value of the cross-correlation function can accurately quantify this time-causal locking relationship that does not depend on the installation position.

[0139] Finally, the scale-invariant coupling correlation is input into the nonlinear activation function model and normalized to obtain the normalized coupling probability value. The validity judgment result of the fluid disorder index is obtained. Logical AND operation is performed on the normalized coupling probability value and the validity judgment result of the fluid disorder index to generate the kinematic scenario verification index.

[0140] In this embodiment, the nonlinear activation function model preferably adopts the logistic function, with its gain parameter set to 10 and its bias parameter set to 0.5. The purpose of this parameter setting is to map the linear correlation coefficient to a probability score with steep edges, thereby rewarding highly correlated data and penalizing low-correlation data.

[0141] The logic and computation mechanism constitute a veto logic: if the fluid disorder index shows that the environment is a low-entropy mechanical simulation vibration, regardless of the motion coupling degree calculation result, the system will forcibly determine that the position is fake and set the kinematic scene verification index to zero, thereby triggering the risk lock operation of the smart contract and blocking the illegal disbursement of the multi-currency fund pool.

[0142] Furthermore, in step S4, the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 are input into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value. Only when the comprehensive transaction confidence value is greater than the preset security settlement threshold, the blockchain smart contract state machine is driven to switch from the locked state to the execution state, and the cross-border payment gateway is called to execute the pre-payment transfer operation of the multi-currency fund pool.

[0143] According to claim 1, the method for multi-currency intelligent settlement in cross-border ocean e-commerce is characterized in that, in step S4, the specific operation of inputting the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 into the multimodal confidence evaluation model to calculate the comprehensive transaction confidence value is as follows: performing a multiplication operation on the thermodynamic consistency verification index and the kinematic scenario verification index to obtain the index product, performing a square root operation on the index product to obtain the geometric mean basis value, and using the sensitive response characteristics of the geometric mean operation to zero values ​​to construct physical existence co-occurrence constraints.

[0144] In practice, the geometric mean algorithm is used instead of the arithmetic mean algorithm to construct the basic probability, thereby achieving a veto effect at the algorithm level. By utilizing the mathematical properties of the geometric mean operation, it is mandatory that both the thermodynamic consistency verification index and the kinematic scenario verification index must have high confidence levels.

[0145] If any one of the indicators approaches zero (e.g., the goods are fake or the location is fake), the geometric mean base value will be forced to approach zero no matter how high the other indicator value is. In layman's terms, this processing logic is analogous to the AND gate in electronic circuits, where the output port is only high when all input ports are high.

[0146] In this way, fraudsters effectively combat fraudulent strategies that attempt to use a high-confidence indicator (such as a real GPS location) to cover up another low-confidence indicator (such as a fake cargo loading), thus closely linking the choice of mathematical operations with the purpose of solving specific anti-fraud technology problems.

[0147] Subsequently, the absolute value of the difference between the thermodynamic consistency verification index and the kinematic scenario verification index is calculated. The absolute value of the difference is defined as the modal inconsistency dispersion. The modal inconsistency dispersion is input into the hyperbolic tangent inverse activation function model. The modal inconsistency dispersion is nonlinearly amplified using a preset risk sensitivity coefficient to generate a paradox penalty coefficient. The consistency weight factor is generated based on the difference between the numerical value and the paradox penalty coefficient.

[0148] In this embodiment, the preset risk sensitivity coefficient is specifically set to a value of 2.0. The logic for setting the preset risk sensitivity coefficient is based on regression training of a large amount of historical cross-border trade fraud data. The setting of a value of 2.0 causes the function curve of the hyperbolic tangent inverse activation function model to show a sharp downward trend when the modal inconsistency dispersion exceeds a value of 0.3, thereby imposing a severe penalty on significantly inconsistent data.

[0149] The preset risk sensitivity coefficient plays a role in adjusting the nonlinear gain in the algorithm. Its technical effect is to significantly improve the sensitivity of the cross-border ocean e-commerce multi-currency intelligent settlement system to inconsistent data, thereby greatly reducing the success rate of gray-scale attacks (i.e., attacks that are half true and half false).

[0150] The consistency weight factor characterizes the degree of logical self-consistency between data sources. When there is a logical conflict between thermodynamic features and kinematic features, the sharp drop in the consistency weight factor mathematically simulates the trust collapse caused by the physical logic paradox.

[0151] Next, the cross-border ocean e-commerce multi-currency intelligent settlement system determines whether the minimum value of the thermodynamic consistency verification index and the kinematic scenario verification index is lower than the preset minimum physical existence threshold. When the minimum value is lower than the preset minimum physical existence threshold, a circuit breaker multiplier with a value of zero is generated; when the minimum value is not lower than the preset minimum physical existence threshold, a circuit breaker multiplier with a value of one is generated.

[0152] In this embodiment, the preset minimum physical existence threshold is set to a value of 0.4. The physical meaning of the preset minimum physical existence threshold is the lowest acceptable signal-to-noise ratio for a single physical dimension. The basis for selecting a value of 0.4 is to prevent two indicators with extremely low scores from escaping consistency penalties due to a small difference in values. The circuit breaker multiplier plays the role of bottom-line gating and avalanche breakdown adjustment in the algorithm. Once the physical authenticity of any dimension is lower than the bottom line, it is determined that the current transaction has a fundamental physical defect, and a circuit breaker multiplier with a value of zero is forcibly output to cut off the trust chain.

[0153] Finally, a multiplication operation is performed on the geometric mean base value, the consistency weight factor, and the circuit breaker multiplier, and the product obtained from the multiplication operation is determined as the comprehensive transaction confidence value.

[0154] This multiplication process constructs a rigorous logical closed loop, using the overall transaction confidence value as the final probability measure of physical authenticity. Only when both the material flow and the information flow have high confidence, are logically highly self-consistent, and do not violate the bottom line threshold, can a high score close to one be output.

[0155] In step S4, the blockchain smart contract state machine is driven to switch from the locked state to the execution state only when the comprehensive transaction confidence value is greater than the preset security settlement threshold. The specific operation of calling the cross-border payment gateway to perform the pre-payment transfer operation of the multi-currency fund pool is as follows: the comprehensive transaction confidence value, which is between zero and one, is subjected to a numerical quantization mapping operation. By multiplying the comprehensive transaction confidence value by a preset base point magnification factor and performing a floor operation, an integer confidence rank value compatible with the blockchain virtual machine is generated.

[0156] In this embodiment, the preset base point magnification factor is set to the value of 10,000. The physical meaning of the preset base point magnification factor is to map the floating-point probability space to the financial-grade base point integer space. The basis for selecting the value of 10,000 is that blockchain smart contract virtual machines usually do not support floating-point operations, and financial settlement usually requires accuracy to the ten-thousandth place. The preset base point magnification factor plays a role in data format standardization and computation cost optimization in the algorithm.

[0157] Subsequently, the integer confidence rank value is input into the pre-deployed blockchain smart contract finite state machine, which includes locked, executed, and penalized states. The execution state transition judgment logic is as follows: when the integer confidence rank value is detected to be greater than the preset secure settlement integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the executed state, triggering the oracle interface to call the cross-border payment gateway to lock the corresponding fiat currency amount in the multi-currency fund pool and perform a pre-payment transfer operation to the seller's account.

[0158] In this embodiment, the preset secure settlement integer threshold is set to the value 9,500 (corresponding to a probability of 0.95). The technical consideration for setting the preset secure settlement integer threshold to the value 9,500 is based on historical transaction backtesting data, which requires physical evidence to reach an extremely high level of confidence in order to match the extremely low fault tolerance requirements of financial transactions.

[0159] When this condition is met, the finite state machine of the blockchain smart contract completes the energy level transition. At this time, the oracle retrieves the current real-time international settlement exchange rate, locks the balance of the corresponding buyer's declared currency from the multi-currency fund pool, and completes the accounting on the chain according to atomic operations, triggering the cross-border payment gateway to transfer the advance payment to the seller's local currency account.

[0160] Meanwhile, when the integer confidence rank value is detected to be less than the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the penalty state, triggering the default record on-chain logic and deducting the reputation score of the corresponding IoT node.

[0161] In this embodiment, the preset risk control circuit breaker threshold is set to 8,000 (corresponding to a probability of 0.80). The technical consideration for setting the preset risk control circuit breaker threshold to 8,000 is to promptly execute a circuit breaker operation when the uncertainty of physical verification is high, freezing funds for manual review and preventing potential losses from escalating. This mechanism is similar to the circuit breaker mechanism in financial markets, highlighting the role of the preset risk control circuit breaker threshold as a technical safety valve.

[0162] In addition, when the integer confidence rank value is detected to be between the preset secure settlement integer threshold and the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to remain locked and generate a manual review request signal.

[0163] In this intermediate state, specific physical waveform segments (such as abnormal fluctuation segments of thermal time constant) that cause inconsistencies in the preceding steps S1 to S3 are automatically extracted and pushed to the visualization interface of the risk control center. This logic sets up a grayscale buffer to handle edge cases where the physical data has some noise but is not enough to be judged as fraud, ensuring that the cross-border ocean e-commerce multi-currency intelligent settlement system retains the fault tolerance capability for complex working conditions while pursuing automation.

[0164] Example 2

[0165] like Figure 2 As shown, the present invention also discloses a multi-currency intelligent settlement system for cross-border ocean e-commerce, including a spatiotemporal alignment physical fingerprint generation module, a thermodynamic consistency verification module, a kinematic scene verification module, and a smart contract settlement control module, wherein;

[0166] The spatiotemporal aligned physical fingerprint generation module is used to acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. It extracts physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features, a time axis mapping model is constructed. The relative timestamps of the relative temporal sensing data inside the container are calibrated to absolute world time using the time axis mapping model to generate a spatiotemporal aligned physical fingerprint matrix.

[0167] The thermodynamic consistency verification module is used to extract the temperature change sequence during the passive temperature rise period from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, calculate the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence, obtain the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculate the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generate thermodynamic consistency verification index.

[0168] The kinematic scene verification module is used to extract triaxial acceleration data from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, perform frequency domain feature transformation on the triaxial acceleration data to obtain the measured environment spectrum, verify whether there are wave swell characteristic frequencies in the measured environment spectrum, calculate the coupling correlation between the rigid body motion response characteristics in the measured environment spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system, and generate kinematic scene verification indicators.

[0169] The smart contract settlement control module is used to input the thermodynamic consistency verification index generated by the thermodynamic consistency verification module and the kinematic scenario verification index generated by the kinematic scenario verification module into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value. Only when the comprehensive transaction confidence value is greater than the preset safe settlement threshold, the blockchain smart contract state machine is driven to switch from the locked state to the execution state, and the cross-border payment gateway is called to execute the pre-payment transfer operation of the multi-currency fund pool.

[0170] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-currency intelligent settlement method for cross-border ocean e-commerce, characterized in that, Includes the following steps: Step S1: Acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. Extract physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Construct a time axis mapping model based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features. Use the time axis mapping model to calibrate the relative timestamps of the relative temporal sensing data inside the container to absolute world time and generate a spatiotemporally aligned physical fingerprint matrix. Step S2: Extract the temperature change sequence during the passive temperature rise period from the spatiotemporal aligned physical fingerprint matrix generated in Step S1, calculate the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence, obtain the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculate the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generate thermodynamic consistency verification index. Step S3: Extract triaxial acceleration data from the spatiotemporal aligned physical fingerprint matrix generated in step S1, perform frequency domain feature transformation on the triaxial acceleration data to obtain the measured environmental spectrum, verify whether there are wave swell characteristic frequencies in the measured environmental spectrum, calculate the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system, and generate kinematic scene verification indicators. Step S4: Input the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 into the multimodal confidence assessment model, calculate the comprehensive transaction confidence value, and only when the comprehensive transaction confidence value is greater than the preset security settlement threshold, drive the blockchain smart contract state machine to switch from the locked state to the execution state, and call the cross-border payment gateway to execute the pre-payment transfer operation of the multi-currency fund pool.

2. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 1, characterized in that, In step S1, the specific operation of extracting physical vibration abrupt change features from the relative time-series sensing data inside the container is as follows: A sliding window short-time Fourier transform operation is performed on the triaxial acceleration time-series signal contained in the relative time-series sensing data inside the container to transform the acceleration numerical sequence in the time domain into a power spectral density distribution matrix in the frequency domain. An integral operation is performed on the energy values ​​in the power spectral density distribution matrix that are located in the preset low-frequency gravity wave frequency band to obtain the low-frequency energy component. The numerical ratio between the low-frequency energy component and the sum of the total spectral energy of the entire frequency band is calculated to generate a low-frequency energy proportion factor sequence. Monitor the temporal numerical changes of the low-frequency energy proportion factor sequence, identify the numerical mutation time points where the low-frequency energy proportion factor sequence experiences a step decrease in numerical value, determine the numerical mutation time points as the physical phase transition points where the low-frequency surge vibration mode dominated by fluid medium changes to the broadband mechanical shock vibration mode dominated by rigid medium, and mark the physical phase transition points as physical medium switching anchor points. Traverse all physical medium switching anchor points to construct a set of physical vibration abrupt change features; In the process of constructing the physical vibration abrupt change feature set, a filtering logic based on the hydrodynamic spectrum morphology is executed to identify and remove deep-sea wind and wave vibration data with high amplitude values ​​but whose spectrum structure exhibits the characteristics of a gravity wavelength periodic narrow band. Only port operation vibration data with a spectrum structure exhibiting broadband pulse characteristics are retained. The physical medium switching anchor point after filtering is identified as the physical vibration abrupt change feature, thereby establishing a physical side observation benchmark for time axis calibration.

3. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 2, characterized in that, In step S1, the specific operation of constructing a time-axis mapping model based on the characteristics of sudden physical vibrations and port operation events is as follows: A random sampling consistency iterative strategy is implemented to extract feature point pairs from the physical vibration mutation feature set and the port operation event feature set to construct candidate linear mapping hypothesis models. In the process of iteratively calculating the candidate linear mapping hypothesis models, the physical frequency stability characteristics of the quartz crystal oscillator are introduced as hard physical constraint boundaries. Numerical filtering operations are performed to directly eliminate non-physical models in the candidate linear mapping hypothesis models whose clock drift rate parameters exceed the nominal physical error limit of the quartz crystal oscillator, and retain the set of interior points that conform to the physical laws of the hardware. A spatiotemporal topologically consistent energy equation containing a robust kernel function is constructed based on the set of interior points. Variational methods are performed to solve for the global minimum of the spatiotemporal topologically consistent energy equation. The global minimum is analyzed to obtain the optimal clock drift rate parameter and the optimal initial time deviation parameter. The optimal clock drift rate parameter and the optimal initial time deviation parameter are used to establish a time axis mapping model that can linearly map relative machine time to absolute world time.

4. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 3, characterized in that, In step S2, the specific operation of extracting the temperature change sequence during the passive temperature rise period from the spatiotemporally aligned physical fingerprint matrix generated in step S1, and calculating the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence is as follows: Morphological opening filtering is performed on the container internal temperature numerical sequence contained in the spatiotemporal aligned physical fingerprint matrix to eliminate high-frequency thermal noise interference caused by refrigeration compressor oscillation, thereby obtaining a smooth temperature change sequence; Perform a first-order difference operation on the smoothed temperature change sequence to generate a temperature change rate sequence, and further perform a second-order difference operation to generate a temperature acceleration sequence. Traverse the smooth temperature change sequence and identify continuous time segments that simultaneously satisfy the condition that the temperature change rate sequence value is positive and the temperature acceleration sequence value is non-positive. If the duration of the continuous time segment exceeds the preset thermal relaxation time threshold, then the continuous time segment is marked as a passive temperature rise period. In this way, the natural cooling process can be locked by using the convex function geometric features of the temperature curve without needing to obtain the power status data of the refrigerator. Extract the instantaneous temperature values ​​and corresponding instantaneous temperature change rate values ​​during the passive temperature rise period to construct a phase plane differential regression dataset; The random sampling consensus algorithm was used to perform linear regression on the phase plane differential regression dataset to fit a linear regression equation with instantaneous temperature values ​​as independent variables and instantaneous temperature change rate values ​​as dependent variables. Extract the slope value of the linear regression equation, calculate the opposite of the reciprocal of the slope value, and determine the calculation result as the measured thermal response parameter characterizing the thermal resistance and capacitance properties of objects inside the container.

5. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 4, characterized in that, In step S2, the specific steps for obtaining the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculating the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generating thermodynamic consistency verification indicators are as follows: The data interface is used to parse digital trade customs declaration documents to obtain customs code information of the declared goods. The customs code information is then used to search in a pre-set multimodal thermophysical property knowledge graph to obtain the lower limit value of the theoretical thermal response parameter that matches the physical properties of the declared goods. The lower limit value of the theoretical thermal response parameter is then confirmed as the boundary benchmark of the theoretical thermal response parameter range. Calculate the natural logarithm of the measured thermal response parameter and the natural logarithm of the lower limit of the theoretical thermal response parameter, and calculate the difference between the two. When the measured thermal response parameter is less than the lower limit of the theoretical thermal response parameter, the absolute value of the difference is determined as the numerical deviation degree, which is used to characterize the degree of physical deviation between the thermal inertia of the actual loaded object inside the container and the theoretical thermal inertia of the declared cargo at the order of magnitude level, thereby eliminating the problem of numerical linear comparison distortion caused by the large difference between the thermal capacity of air and the thermal capacity of the actual cargo. The numerical deviation is input into a normalized evaluation model based on the hyperbolic tangent function. The numerical deviation is dimensionless using a preset loading rate fluctuation tolerance parameter. The processed result is then subjected to nonlinear penalty calculation using a preset discrimination order parameter. Finally, a thermodynamic consistency verification index between zero and one is generated. The closer the value of this indicator is to one, the higher the physical authenticity of the goods; the closer the value of this indicator is to zero, the greater the risk of empty container fraud.

6. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 5, characterized in that, In step S3, triaxial acceleration data is extracted from the spatiotemporal aligned physical fingerprint matrix generated in step S1. Frequency domain feature transformation is performed on the triaxial acceleration data to obtain the measured environmental spectrum. The specific operation to verify whether there are wave swell characteristic frequencies in the measured environmental spectrum is as follows: Long-window mean filtering is performed on the triaxial acceleration data extracted from the spatiotemporal aligned physical fingerprint matrix to obtain the steady-state gravity vector representing the direction of gravity at the Earth's core. A spatial rotation mapping matrix is ​​constructed based on the steady-state gravity vector. The triaxial acceleration data is then projected onto the virtual geocentric coordinate system using the spatial rotation mapping matrix to separate the acceleration sequence in the absolute vertical direction. Welch power spectral density estimation is performed on the absolute vertical acceleration sequence to generate the measured environmental spectrum. Shannon information entropy calculation is performed on the measured environmental spectrum to generate the Shannon information entropy value. The Shannon information entropy value is defined as the fluid disorder index. Calculate the energy integral value within the preset wave swell frequency range in the measured environmental spectrum, and calculate the proportion of the energy integral value to the total energy integral value of the entire measured environmental spectrum. The numerical comparison logic is executed to determine whether the fluid disorder index is greater than the preset minimum fluid disorder threshold and whether the numerical ratio is greater than the preset physical energy ratio threshold. Only when the fluid disorder index is greater than the preset minimum fluid disorder threshold and the numerical ratio is greater than the preset physical energy ratio threshold, a verification pass signal is generated to confirm the presence of wave surge characteristic frequencies in the measured environmental spectrum; otherwise, an environmental anomaly alarm signal is generated.

7. A multi-currency intelligent settlement method for cross-border ocean e-commerce according to claim 6, characterized in that, In step S3, the specific operation of calculating the coupling correlation between the rigid body motion response characteristics in the measured environmental spectrum and the ship navigation state information in the absolute spatiotemporal trajectory data of the external logistics system, and generating kinematic scenario verification indicators, is as follows: Monitor the ship's heading change rate in the absolute spatiotemporal trajectory data of the external logistics system. When the ship's heading change rate exceeds the preset maneuver event trigger threshold, initiate the rigid body coupling verification calculation process. Perform multiplication operations to calculate the product of the ship's rate of change of course and the ship's speed, and generate a theoretical sequence of centrifugal acceleration modulus. Obtain the absolute horizontal plane acceleration sequence after processing by the spatial rotation mapping matrix, and calculate the vector modulus of the absolute horizontal plane acceleration sequence to generate the measured resultant horizontal acceleration modulus sequence; Within the preset time delay scanning window, a sliding cross-correlation operation is performed on the theoretical centrifugal acceleration modulus sequence and the measured horizontal combined acceleration modulus sequence. The optimal time delay parameter corresponding to the maximum value of the cross-correlation function is searched, and the maximum value of the cross-correlation function is determined as the scale-invariant coupling correlation. The scale-invariant coupling correlation is input into the nonlinear activation function model and normalized to obtain the normalized coupling probability value. The validity judgment result of the fluid disorder index is obtained. A logical AND operation is performed on the normalized coupling probability value and the validity judgment result of the fluid disorder index to generate the kinematic scenario verification index. When the fluid disorder index is judged to be invalid, the kinematic scenario verification index is forcibly set to zero.

8. The method for multi-currency intelligent settlement in cross-border ocean e-commerce according to claim 7, characterized in that, In step S4, the thermodynamic consistency verification index generated in step S2 and the kinematic scenario verification index generated in step S3 are input into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value. The specific operation is as follows: Multiplication is performed on the thermodynamic consistency verification index and the kinematic scenario verification index to obtain the index product. The square root operation is then performed on the index product to obtain the geometric mean basis value. The physical existence co-occurrence constraint is constructed by utilizing the sensitive response characteristic of the geometric mean operation to zero values. The absolute value of the difference between the thermodynamic consistency verification index and the kinematic scenario verification index is calculated. The absolute value of the difference is defined as the modal inconsistency dispersion. The modal inconsistency dispersion is input into the hyperbolic tangent inverse activation function model. The modal inconsistency dispersion is generated by performing nonlinear amplification operation on the modal inconsistency dispersion using a preset risk sensitivity coefficient. The consistency weight factor is generated based on the difference between the numerical value and the paradox penalty coefficient. Determine whether the minimum value of the thermodynamic consistency verification index and the kinematic scenario verification index is lower than the preset minimum physical existence threshold. If the minimum value is lower than the preset minimum physical existence threshold, generate a fuse multiplier with a value of zero. If the minimum value is not lower than the preset minimum physical existence threshold, generate a fuse multiplier with a value of one. Perform a multiplication operation on the geometric mean base value, the consistency weight factor, and the circuit breaker multiplier, and determine the product result as the comprehensive transaction confidence value.

9. A multi-currency intelligent settlement method for cross-border ocean e-commerce according to claim 8, characterized in that, In step S4, the blockchain smart contract state machine is driven to switch from the locked state to the execution state only when the comprehensive transaction confidence value is greater than the preset security settlement threshold. The specific operation of calling the cross-border payment gateway to execute the pre-payment transfer operation of the multi-currency fund pool is as follows: A numerical quantization mapping operation is performed on the comprehensive transaction confidence value, which ranges from zero to one. By multiplying the comprehensive transaction confidence value by a preset base point magnification factor and performing a floor operation, an integer confidence rank value compatible with the blockchain virtual machine is generated. The integer confidence rank is input into the pre-deployed blockchain smart contract finite state machine, which includes locked state, execution state, and penalty state. The execution state transition judgment logic is as follows: when the integer confidence rank is detected to be greater than the preset secure settlement integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the execution state, triggering the oracle interface to call the cross-border payment gateway to lock the corresponding fiat currency amount in the multi-currency fund pool and perform a pre-payment transfer operation to the seller's account. When the integer confidence rank value is detected to be less than the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to perform an atomic transition from the locked state to the penalty state, triggering the default record on-chain logic and deducting the reputation score of the corresponding IoT node. When the integer confidence rank value is detected to be between the preset secure settlement integer threshold and the preset risk control circuit breaker integer threshold, the blockchain smart contract finite state machine is driven to remain locked and generate a manual review request signal.

10. A multi-currency intelligent settlement system for cross-border e-commerce, applied to the multi-currency intelligent settlement method for cross-border e-commerce as described in any one of claims 1-9, characterized in that, It includes a spatiotemporal alignment physical fingerprint generation module, a thermodynamic consistency verification module, a kinematic scene verification module, and a smart contract settlement control module, among which; The spatiotemporal aligned physical fingerprint generation module is used to acquire relative temporal sensing data inside the container and absolute spatiotemporal trajectory data of the external logistics system in parallel through the IoT communication interface. It extracts physical vibration mutation features from the relative temporal sensing data inside the container and port operation event features from the absolute spatiotemporal trajectory data of the external logistics system. Based on the spatiotemporal topological constraint relationship between the physical vibration mutation features and the port operation event features, a time axis mapping model is constructed. The relative timestamps of the relative temporal sensing data inside the container are calibrated to absolute world time using the time axis mapping model to generate a spatiotemporal aligned physical fingerprint matrix. The thermodynamic consistency verification module is used to extract the temperature change sequence during the passive temperature rise period from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, calculate the measured thermal response parameters characterizing the thermal resistance and capacitance properties of objects inside the container based on the temperature change sequence, obtain the theoretical thermal response parameter range corresponding to the customs code of the declared commodity in the digital trade customs declaration, calculate the numerical deviation between the measured thermal response parameters and the theoretical thermal response parameter range, and generate thermodynamic consistency verification index. The kinematic scene verification module is used to extract triaxial acceleration data from the spatiotemporal aligned physical fingerprint matrix generated by the spatiotemporal aligned physical fingerprint generation module, perform frequency domain feature transformation on the triaxial acceleration data to obtain the measured environment spectrum, verify whether there are wave swell characteristic frequencies in the measured environment spectrum, calculate the coupling correlation between the rigid body motion response characteristics in the measured environment spectrum and the ship navigation status information in the absolute spatiotemporal trajectory data of the external logistics system, and generate kinematic scene verification indicators. The smart contract settlement control module is used to input the thermodynamic consistency verification index generated by the thermodynamic consistency verification module and the kinematic scenario verification index generated by the kinematic scenario verification module into the multimodal confidence assessment model to calculate the comprehensive transaction confidence value. Only when the comprehensive transaction confidence value is greater than the preset safe settlement threshold, the blockchain smart contract state machine is driven to switch from the locked state to the execution state, and the cross-border payment gateway is called to execute the pre-payment transfer operation of the multi-currency fund pool.

Images