A smart logistics data management method and system based on multi-source data fusion
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUIZHOU MODERN DIGITAL INTELLIGENCE TECH CO LTD
- Filing Date
- 2026-03-05
- Publication Date
- 2026-05-29
Smart Images
Figure CN121788038B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of logistics data processing technology, and in particular to a smart logistics data governance method and system based on multi-source data fusion. Background Technology
[0002] As the digital transformation of the smart logistics industry deepens, logistics parks are placing higher demands on the digital record-keeping and refined management of warehousing operations. Smart logistics data governance, as a key link connecting sensing devices and management decisions, integrates sensor data and business data to characterize the real-time displacement and operational status of stored goods, playing a vital supporting role in improving logistics turnover efficiency and ensuring asset security.
[0003] Existing smart logistics management solutions typically employ a multi-source information fusion framework, acquiring spatial location information of logistics entities through RFID nodes and laser ranging sensors deployed at the work site. To handle random fluctuations in the raw signals, these solutions often use Kalman filtering algorithms to smooth the sensor outputs over time and assign preset static weights based on the nominal accuracy of each sensor to synthesize the motion trajectory of the logistics equipment.
[0004] However, traditional static weighted governance methods have limitations when dealing with complex dynamic environmental data within logistics parks. Considering the large number of metal shelves and mobile handling equipment in logistics operation areas, the resulting multipath effects and electromagnetic shielding can cause non-physical instantaneous jumps in sensor readings. Because existing governance methods only perform equalization mapping based on the grayscale statistics or variance characteristics of sensor data, they lack logical constraints on the data's background and cannot effectively identify these statistically consistent but logically inconsistent abnormal signals. When sensors experience displacement deviations due to interference, the governance system, unable to distinguish between physical noise and actual business operations, will include erroneous displacements in the log data, leading to logical deviations in the rectified logistics trajectory and reducing the reliability of the log data. Summary of the Invention
[0005] To address the aforementioned technical problems arising from the lack of logical constraints on data background in existing governance methods, which makes it difficult to distinguish between instantaneous changes caused by physical environmental interference and actual business operations, thus leading to logical deviations in logistics trajectories and low reliability of trace data, this invention provides solutions in the following aspects.
[0006] In a first aspect, the present invention provides a smart logistics data governance method based on multi-source data fusion, the method comprising the following steps:
[0007] The system acquires raw data from a sensor set at various sampling times and associated business instruction streams at the logistics operation site. It performs a preset normalization process on the raw data to obtain preliminary governance data for each sampling time. The raw data includes the physical coordinates of the sensors. For any sensor in the sensor set, its neighboring sensors are determined based on its physical coordinates. The spatial deviation of the sensor at each sampling time is calculated based on the physical coordinates of the sensor and its neighboring sensors. The temporal change rate of the preliminary governance data at adjacent sampling times is calculated. The temporal change rate and the spatial deviation are fused to determine the physical confidence level of the sensor. Based on a preset instruction response model, the business instruction intensity in the business instruction stream is mapped to a characteristic change amplitude. The business confidence level of the sensor is determined based on the deviation between the characteristic change amplitude and the temporal change rate. Adaptive weights for each sensor are determined based on their physical and business confidence levels. The preliminary governance data is then weighted and fused using these adaptive weights to obtain the target governance data.
[0008] This invention acquires raw data and business instruction flows from logistics operations. Based on preliminary data processing, it introduces a dual-dimensional confidence assessment mechanism encompassing both physical and business dimensions. It utilizes the multiplicative coupling of spatial deviation and temporal rate of change to evaluate the physical observation stability of sensors. Furthermore, it maps the intensity of business instructions to expected characteristic change amplitudes through an instruction response model to assess the consistency between measured data and business logic. Thus, during the data fusion phase, it adaptively adjusts the fusion weights of each sensor based on the joint confidence level. This effectively addresses issues such as sensor value jumps caused by environmental interference and the disconnect between sensor data and actual business actions in smart logistics scenarios. It improves the accuracy and logical consistency of the target data processing, providing a reliable data foundation for accurate digital traceability in logistics warehousing.
[0009] Preferably, the step of calculating the spatial deviation of the sensor at each sampling time based on the physical coordinates of the sensor and neighboring sensors includes: for any sampling time, calculating the geometric center of the physical coordinates of the neighboring sensors to obtain the center coordinates; calculating the Euclidean distance between the physical coordinates of the sensor and the center coordinates; and dividing the Euclidean distance by a preset neighborhood reference distance to obtain the spatial deviation of the sensor at the corresponding sampling time.
[0010] This invention transforms abstract spatial positional relationships into concrete spatial deviation by calculating the center coordinates of neighboring sensors and obtaining the Euclidean distance between the sensors and them. This allows for an intuitive assessment of the sensor's positional offset within the physical array, providing a physically meaningful correction benchmark for subsequent physical confidence calculations.
[0011] Preferably, the physical confidence level satisfies the following relationship:
[0012] ;
[0013] in, It is a sensor At sampling time Physical confidence level; , These are sensors At sampling time , Preliminary governance data; It is the preset sensitivity adjustment coefficient; It is a sensor At the sampling time of its neighboring sensors Spatial deviation; It is the natural logarithm function; It is a natural exponential function; It is the absolute value symbol.
[0014] This invention utilizes the natural exponential function to nonlinearly map the rate of change of time to the degree of spatial deviation, so that the physical confidence can decrease sensitively with the drastic fluctuations of the observation values or the abnormal shift of the spatial location, thereby suppressing abnormal physical observation values and ensuring the stability of data during the fusion process.
[0015] Preferably, the preset instruction response model is implemented through an instruction response function, and the instruction response function satisfies the following relation: ;
[0016] in, It is an instruction response function; Sampling time The characteristic change amplitude; Sampling time The intensity of business instructions; It is the maximum expected rate of change of displacement of the sensor; It is the preset response gain coefficient; It is the logical threshold for the strength of business instructions; It is a natural exponential function.
[0017] This invention transforms the intensity of business instructions into the expected characteristic change amplitude through an instruction response function. Considering the physical characteristics of logistics equipment when responding to instructions, it constructs a mapping process from business instructions to physical actions using logical thresholds and gain coefficients, providing an objective reference for determining whether data conforms to business logic.
[0018] Preferably, determining the service confidence level of the sensor based on the deviation between the feature change amplitude and the time change rate includes: calculating the absolute value of the difference between the time change rate and the feature change amplitude, denoted as the absolute deviation; multiplying the absolute deviation by a preset time normalization factor to obtain a scaling deviation; processing the scaling deviation using a hyperbolic tangent function to obtain a nonlinear mapping value; obtaining the complement of the nonlinear mapping value to obtain the service confidence level of the sensor.
[0019] This invention uses the hyperbolic tangent function to map the absolute deviation between the rate of change of time and the amplitude of characteristic change, and obtains its complement to obtain the service confidence level. This can smoothly evaluate the degree of matching between the measured action and the expected instruction, and reduce misjudgments caused by service response delay or sensor noise.
[0020] Preferably, determining the adaptive weight of each sensor based on the physical confidence and operational confidence of each sensor includes: calculating the product of the physical confidence and operational confidence of each sensor to obtain the joint confidence of each sensor; summing the joint confidence of all sensors to obtain the global confidence sum; and recording the ratio of the joint confidence of any sensor to the global confidence sum as the adaptive weight of the corresponding sensor.
[0021] This invention obtains a joint confidence score by calculating the product of physical confidence score and business confidence score, and then obtains the adaptive weights of each sensor based on this score. This achieves the coordinated linkage between physical layer observation quality and business layer logical compliance, ensuring that the final target governance data is biased towards sensing sources that are both stable and in line with business expectations.
[0022] Preferably, determining the neighboring sensors for any sensor in the sensor set based on its physical coordinates includes: for any sensor in the sensor set, obtaining the physical coordinates of the sensor and the remaining sensors in the sensor set; calculating the straight-line distance between the physical coordinates of the sensor and the physical coordinates of each remaining sensor; sorting the remaining sensors in ascending order based on the straight-line distance; and extracting the remaining sensors ranked in the top preset order as neighboring sensors.
[0023] Preferably, the step of performing a preset normalization process on the original data to obtain preliminary governance data at each sampling time includes: acquiring the historical observation set of each sensor within a preset time window; calculating the difference between the original data and the minimum value in the historical observation set, dividing it by the difference between the maximum value and the minimum value in the historical observation set, to obtain the preliminary governance data at each sampling time.
[0024] Preferably, the step of using the adaptive weights of each sensor to perform weighted fusion of the preliminary governance data to obtain the target governance data includes: for any sampling time, multiplying the preliminary governance data of each sensor with its corresponding adaptive weight to obtain the weighted governance component of each sensor; and accumulating the weighted governance components of all sensors in the sensor set to obtain the target governance data at the corresponding sampling time.
[0025] In a second aspect, the present invention provides a smart logistics data governance system based on multi-source data fusion. The smart logistics data governance system based on multi-source data fusion includes a memory and a processor. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the smart logistics data governance method based on multi-source data fusion of the first aspect of the present invention is implemented.
[0026] By adopting the above technical solution, a computer program is generated from the intelligent logistics data governance method based on multi-source data fusion of the first aspect of the present invention, and stored in a memory so that it can be loaded and executed by a processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.
[0027] The beneficial effects of this invention are as follows: This invention utilizes spatial deviation to calculate the consistency of sensors in their physical layout and combines this with an instruction response model to obtain the expected change patterns in the business dimension. This allows for accurate assessment of the reliability of raw data from each sensor under complex logistics environment interference, and the adaptive weights can be adjusted accordingly. On the one hand, this invention suppresses random jumps in sensors caused by ambient light, occlusion, or electromagnetic interference; on the other hand, it verifies the business rationality of data generation, effectively reducing the probability of false data being entered into the smart logistics system. By weighted fusion of the preliminary governance data, the target governance data output by this invention not only possesses physical smoothness but also business traceability. This provides high-reliability decision support for accurate scheduling of logistics equipment and digital supervision of cargo assets, reducing reliance on manual intervention in the data governance process and improving the data quality and automated governance level of the smart logistics data center. Attached Figure Description
[0028] Figure 1 A flowchart illustrating a smart logistics data governance method based on multi-source data fusion, provided as an embodiment of the present invention;
[0029] Figure 2 This is a schematic diagram of the evolution of adaptive weight allocation for multi-source sensors provided in an embodiment of the present invention;
[0030] Figure 3 This is a schematic diagram comparing the sensor data management effects provided in an embodiment of the present invention;
[0031] Figure 4 This is a structural block diagram of a smart logistics data governance system based on multi-source data fusion, provided for an embodiment of the present invention. Detailed Implementation
[0032] The first aspect of this invention provides a smart logistics data governance method based on multi-source data fusion, such as... Figure 1 As shown, the method includes steps S100-S400:
[0033] Step S100: Obtain the raw data of the sensor set at each sampling time in the logistics operation site and the associated business instruction stream, perform preset normalization processing on the raw data, and obtain the preliminary management data at each sampling time.
[0034] It should be noted that in the automated operation process of smart logistics parks, considering the need for real-time synchronization of physical spatial displacement trajectories and logical spatial business status, this invention requires obtaining evaluation benchmarks for logistics status through multiple spatially distributed sensing sources. Specifically, the cargo identification is used to establish a unique mapping relationship between the physical object and its digital file, representing the identity of the entity being transferred; precise spatial coordinates are used to frequently measure the absolute position of the entity in the automated warehouse, representing the real-time movement status and physical trajectory of the cargo; and the associated operation instruction flow reflects the management system's expected behavior settings for the equipment, serving as a reference benchmark for representing the compliance of the current operational business. Through the collaborative acquisition of these multi-source heterogeneous data, the original inputs for physical perception and logical verification can be provided for subsequent determination of the authenticity of the logistics status.
[0035] Specifically, this invention first establishes a unified global sampling clock and determines the sampling period. It then deploys multiple sensing devices, including RFID readers, laser rangefinders, and positioning modules, at preset locations at the work site, forming a sensor set. At each sampling moment, the invention acquires the identification of the cargo's attached tag and its logical location information via the RFID reader; it simultaneously acquires multiple sets of three-dimensional precise coordinate data (hereinafter collectively referred to as physical coordinates) of the work entity at each sampling moment using the laser rangefinder and positioning module; and it also acquires the work instruction stream issued by the WMS (Warehouse Management System) at each sampling moment in real time through a data interface.
[0036] It should be noted that due to the fundamental differences in the working principles, measurement ranges, and data protocols of the various sensing devices in the sensor array, the original acquired data differs significantly in numerical magnitude and physical units. Direct logical fusion would result in dimensional conflicts, thereby masking subtle fluctuations in the data that reflect the true characteristics of the operation. Therefore, this invention uses mathematical mapping to initially manage these discrete observations, mapping them to a unified dimensionless scalar space to provide a standardized data benchmark for subsequent confidence assessments across various dimensions.
[0037] Specifically, this invention utilizes linear mapping operators to perform preliminary processing on data from a sensor set. It should be noted that, for each dimensional component of the physical coordinates, such as the x-coordinate, y-coordinate, and height component, this invention uses the aforementioned linear mapping operators to perform calculations to obtain preliminary processed data corresponding to each dimensional component.
[0038] Specifically, this invention pre-builds a storage buffer in the database to store the raw data of each sensor at each sampling time in real time; this invention sets a preset time window based on the average operation cycle of logistics operations. By retrieving the sampling time prior to the sampling time from the storage buffer A series of continuous sampling points are used to construct a historical observation set for each sensor. It should be noted that a preset time window... The value needs to cover the time span for the logistics equipment to complete a single full operation, to ensure that the historical observation set can cover the complete fluctuation range of the equipment in both static and dynamic states. In this embodiment, The preferred sampling period is 300 to 600.
[0039] For any sensor, the preliminary governance data mapped at the sampling time satisfies the following relationship:
[0040] ;
[0041] in, It is a sensor At any moment Preliminary governance data; It is a sensor At any moment The original data; It is a preset time window Internal sensor A collection of historical observations; It is a collection of historical observations The minimum value in; It is a collection of historical observations The maximum value in; It is a preset first tiny value used to prevent the denominator from being zero, and its preferred range is... to .
[0042] In this relation, This is a bias cancellation unit used to cancel sampling time. Raw data relative to the sensor In historical observation set The baseline offset ensures that the calculation results only reflect the net change of the current moment relative to the historical fluctuation baseline. This is a range normalization unit used to calculate the sensor's range. In historical observation set The reciprocal of the internal fluctuation range proportionally compresses observations with different physical units into a unit interval. This is because the numerator and denominator differ at the sampling time... With completely consistent physical dimensions, the division achieves dimensionless transformation of the original observation data, ensuring mathematical comparability of data from different sources in the sensor set during subsequent fusion, thus improving data consistency.
[0043] Thus, preliminary governance data for each sensor in the sensor set at each sampling time has been obtained.
[0044] Step S200: For any sensor in the sensor set, determine its neighboring sensors based on its physical coordinates, and calculate the spatial deviation of the sensor at each sampling time based on the physical coordinates of the sensor and the neighboring sensors; calculate the time change rate of the preliminary treatment data at adjacent sampling times, and fuse the time change rate and the spatial deviation to determine the physical confidence level of the sensor.
[0045] It should be noted that in the automated warehousing scenarios of smart logistics parks, considering the multipath effects and electromagnetic pulse interference generated by moving metal shelves, forklifts, and communication equipment, the data sequences used to sense the operational status are often accompanied by non-physical instantaneous jumps. Given that the displacement of logistics entities follows the principle of motion continuity at the physical level, the numerical evolution of the entities should be within a reasonable fluctuation range during the sampling period. This invention captures the gradient changes of preliminary governance data on the time axis and couples them with their topological consistency in the spatial neighborhood to identify and suppress abnormal disturbance signals caused by environmental disturbances.
[0046] First, the spatial deviation between each sensor and its neighboring sensors is obtained. It should be noted that in complex warehouse environments, the observation results of a single sensor are easily affected by local physical obstructions, leading to deviations. Considering the geographical distribution characteristics of the sensors in the sensor set, for any given sensor, the surrounding sensors can be used as reference nodes to form a neighboring sensor set. Since the overall observation center of multiple sensors within a neighborhood typically exhibits higher consistency, by comparing the geometric distribution differences between the target sensor and its neighboring sensors, the degree of outlier status of the sensor in the spatial dimension can be obtained, which can be used to determine whether there are spatial observation anomalies.
[0047] Specifically, for any sensor in the sensor set, neighboring sensors are determined based on its physical coordinates at the current sampling time. For example, a predetermined number of sensors in the sensor set that are spatially closest to the target sensor can be retrieved as neighboring sensors. The physical coordinates of the neighboring sensors are obtained, and their geometric centers are calculated and used as the center coordinates. The Euclidean distance between the physical coordinates of the sensor and the center coordinates is calculated, and the Euclidean distance is divided by a predetermined neighboring reference distance to achieve dimensionless processing. The resulting dimensionless ratio is determined as the spatial deviation of the sensor at the current sampling time. The predetermined neighboring reference distance can be set according to the average spacing of the sensing devices deployed. In an optional embodiment, the number of selected neighboring sensors can be set according to the deployment density of the sensing devices in the logistics park to ensure that the selected neighboring sensors can cover the physical monitoring area where the target sensor is located. The preferred number of selected sensors is 4 to 8.
[0048] In one feasible implementation, the process of determining the neighborhood sensors includes: for any sensor, obtaining the physical coordinates of the sensor and other remaining sensors in the sensor set; calculating the straight-line distance between the physical coordinates of the sensor and the physical coordinates of each remaining sensor; sorting the remaining sensors in ascending order according to the straight-line distance; extracting the remaining sensors that rank in the top preset position, preferably 4 to 8, or extracting the remaining sensors whose straight-line distance is less than a preset search radius, as the neighborhood sensors of the sensor.
[0049] Then, the physical confidence level of each sensor in the sensor set at each sampling time is obtained. It should be noted that non-physical disturbances usually manifest as drastic numerical jumps in the time dimension and as significant outliers in the spatial dimension. Considering that the exponential mapping structure is used to multiplicatively couple the temporal gradient and spatial deviation, the system can generate a decaying response to data whose spatiotemporal characteristics deviate from physical laws, which can be used to obtain a confidence metric that reflects the true physical state of the data.
[0050] Based on the above logic, any sensor at the sampling time The physical confidence level satisfies the following relationship:
[0051] ;
[0052] in, It is a sensor At sampling time Physical confidence level; , These are sensors At sampling time , Preliminary governance data; It is the preset sensitivity adjustment coefficient; It is a sensor At the sampling time of its neighboring sensors Spatial deviation; It is the natural logarithm function; It is the natural exponential function; It is the absolute value symbol.
[0053] In this relation, Used to evaluate the rate of change of values between adjacent sampling times. When the sensor When an electromagnetic pulse interference causes a jump, this term increases. Because it lies in the negative term of the exponential function, it can affect the physical confidence level. It exhibits a monotonically decreasing trend and is used to suppress transient disturbances. Used to measure the local consistency of feature distribution. When the sensor Spatial deviation caused by physical obstruction As the value increases, the logarithmic value of this term also increases, and it amplifies the impact of the time gradient on the physical confidence level as a weighting factor. The suppression intensity is used to reduce the data weight of spatial outliers.
[0054] It should be noted that the sensitivity adjustment coefficient The value needs to be set according to the intensity of the actual working conditions: for low-speed handling scenarios with low operating frequency and stable equipment operation, the value can be appropriately increased. Setting it to 4.5 to 5 enhances sensitivity to minor data fluctuations and strengthens noise removal; for high-dynamic scenarios with intensive tasks and frequent equipment acceleration or turning, the value can be appropriately reduced. For example, setting it to 2.5 to 3 increases tolerance to normal physical maneuvers. In this invention, The preferred value is 3.5.
[0055] Thus, the physical confidence level of each sensor in the sensor set at each sampling time was obtained.
[0056] Step S300: Based on the preset instruction response model, the intensity of the service instruction in the service instruction stream is mapped to the characteristic change amplitude; based on the deviation between the characteristic change amplitude and the time change rate, the service confidence of the sensor is determined.
[0057] It should be noted that the automated operation of the smart logistics park is driven by business instructions issued by the WMS, and the evolution of physical sensor values should be highly correlated with the action logic of the business instructions. Considering that simple physical smoothing cannot identify abnormal data that conforms to continuous motion patterns but deviates from business logic, such as unauthorized manual handling, goods slipping, or false displacement reports from the system, this invention introduces business instruction flow as an external prior constraint to determine the degree of matching between preliminary governance data and management logic.
[0058] First, the expected characteristic change amplitude of the service command at each sampling time is obtained. It should be noted that service commands typically exist as discrete task parameters, while the sensor feedback represents continuous physical changes. Considering the construction of a command response function, the abstract command intensity can be transformed into expected numerical fluctuations within the logical space, providing a logical reference for verifying the authenticity of the physical data.
[0059] Specifically, a command response model is pre-established. For each sampling moment, the business command intensity value is extracted from the synchronously acquired WMS data; this invention uses the command response function in the model to map the business command intensity value to the expected characteristic change amplitude. Considering that the physical response characteristics of logistics equipment typically have nonlinear start-up and braking processes, and that different business commands correspond to different expected accelerations, the command response function in this invention satisfies the following relationship:
[0060] ;
[0061] in, It is an instruction response function; Sampling time The characteristic change amplitude; Sampling time The intensity of the business instruction is preset by the WMS based on task priority and distance to the storage location, and the value range is [value missing]. ; It is the maximum expected rate of change of displacement of the sensor, which is obtained by dividing the maximum physical displacement velocity of the logistics equipment carrying the sensor by the sensor sampling frequency, and then performing dimensionless processing through the linear mapping operator described in step S100; wherein, the maximum physical displacement velocity is determined by the physical performance parameters of the logistics equipment carrying the sensor. It is a preset response gain coefficient used to adjust the device's sensitivity to command strength; This is a logical threshold for the strength of business commands, used to characterize the critical trigger strength for device startup. It should be noted that at the physical level... Corresponding to the minimum command amplitude required to overcome the static resistance of logistics equipment; when At this time, the equipment is expected to be in a static or slightly moving state, and the corresponding expected characteristic change amplitude approaches zero; It is a natural exponential function.
[0062] In this relation, Used to measure the effective deviation of the current service command strength relative to the device's startup threshold. When the service command strength... Below the logical threshold When this value is negative, the value inside the parentheses is rapidly increased through exponential operation, thereby significantly reducing the output amplitude and filtering out non-operational minor fluctuations generated by the system. Used to depict the dynamic response process of operating equipment switching from a static state to a high-speed operating state. With increasing command intensity... The increase and crossing the threshold Then, this value is used in the response gain coefficient. It exhibits nonlinearity under regulation. The trend is such that when the instruction strength is in the high range, the output approaches 1, thus causing the expected characteristic change amplitude to move towards the maximum expected displacement rate. Approaching is used to characterize the expected motion trajectory of the equipment under full-load operating conditions; conversely, in the range of low command intensity, the output remains at an extremely low level, used to characterize the equipment in a stationary or slightly moving state.
[0063] It should be noted that the response gain coefficient With logical threshold The value of needs to be set according to the dynamic characteristics of the operating equipment: for heavy stacker cranes with large self-weight and significant inertia, the response gain coefficient can be appropriately reduced, such as setting it to 0.05 to 0.1, and the logic threshold can be increased, such as setting it to 20 to 30, to reflect its slow start-up process; for light automated guided vehicles with sensitive movements and extremely fast acceleration and deceleration, the response gain coefficient can be appropriately increased, such as setting it to 0.2 to 0.4, and the logic threshold can be decreased, such as setting it to 5 to 10. In this invention, The preferred value is 0.15. The preferred value is 15.
[0064] Then, the operational confidence level of each sensor in the sensor set at each sampling time is obtained. It should be noted that when the measured physical displacement rate deviates significantly from the expected change amplitude of the operational command, it indicates a logical risk of unreliability in the data sequence. Considering the saturation mapping characteristics of the hyperbolic tangent function, the logical deviation between physical displacement and operational expectations can be transformed into a normalized confidence level index, which can be used to filter valid observation data with operational compliance from a management perspective.
[0065] Based on the above logic, any sensor at the sampling time The business confidence level satisfies the following relationship:
[0066] ;
[0067] in, It is a sensor At sampling time Business confidence level; , These are sensors At sampling time , Preliminary governance data; Sampling time The characteristic change amplitude; It is a preset time normalization factor used to make the deviation of the rate of change dimensionless; It is the hyperbolic tangent function; It is the absolute value symbol.
[0068] In this relation, This is used to measure the absolute difference between the measured rate of change of preliminary governance data and the expected amplitude of business instructions. The logical deviation unit increases when the WMS does not issue a work instruction but the sensor reports a significant displacement, or when the instruction requires high-speed operation but the sensor reports static data. The logical deviation is mapped to a unit interval using the hyperbolic tangent operator. Because the formula uses a subtractive structure, as the logical deviation increases, the output of the hyperbolic tangent operator approaches 1, resulting in a monotonically decreasing business confidence level, used to assess the unreliability of the perceived data at the business level.
[0069] It should be noted that the time normalization factor The value of the time normalization factor needs to be set according to the sampling frequency of the sensing system: for high-frequency sampling precision positioning scenarios, since the displacement gradient between adjacent time points is small, the time normalization factor can be appropriately increased, such as set to 1.8 to 2.5, to amplify the subtle mismatches at the logic level; for low-frequency sampling intermittent monitoring scenarios, the time normalization factor can be appropriately decreased, such as set to 0.8 to 1.2, to improve the system's tolerance to logic misalignments caused by communication delays. In this invention, the factor is preferably 1.5.
[0070] At this point, the service confidence level of each sensor in the sensor set at each sampling time has been obtained.
[0071] Step S400: Determine the adaptive weight of each sensor based on the physical confidence and business confidence of each sensor, and use the adaptive weight of each sensor to perform weighted fusion on the preliminary governance data to obtain the target governance data.
[0072] It should be noted that after calculating the physical and operational confidence levels, the discrete observations from each sensor at the sampling time need to be synthesized. Considering the varying intensity of physical interference experienced by different sensors, such as electromagnetic shielding and differences in their compatibility with operational instructions, using traditional static weighting methods would lead to logical distortion of the processed trajectory. This invention constructs an adaptive weight allocation mechanism based on dual-dimensional confidence levels to adjust the contribution of each sensor's data to the final processing result in real time, thereby generating a highly reliable logistics tracking trajectory.
[0073] First, the adaptive weights of each sensor at each sampling time are calculated. It should be noted that physical confidence reflects the authenticity of data at the perception level, while operational confidence reflects the compliance of instructions at the logical level. Considering that only when the confidence levels of both dimensions are at a high level is the sensor data valid for tracking logistics trajectories, this invention achieves deep suppression of anomalous observation sources by multiplicatively fusing physical and operational confidence, which can be used to construct an importance assessment benchmark for each sensor.
[0074] Specifically, for any sensor in the sensor set, at the sampling time... The adaptive weights satisfy the following relationship:
[0075] ;
[0076] in, It is a sensor At sampling time Adaptive weights; , It is a sensor , At sampling time Physical confidence level; , It is a sensor , At sampling time Business confidence level; It is a collection of sensors; It is a preset second minute value used to prevent the denominator from being zero, and its preferred range is... to .
[0077] In this relation, Used to measure sensors Comprehensive trustworthiness across both physical and operational dimensions. When sensors... When the displacement jumps due to metal interference, or when the displacement feedback deviates from the WMS operation instructions, the corresponding physical confidence or business confidence value decreases, causing the product term to show a non-linear shrinkage trend, which is used to achieve automatic isolation of untrusted data sources at the weight allocation level. This is a normalized reference unit used to calculate the sum of the overall confidence strengths of all sensing sources within the sensor set. By comparing the individual confidence strength with the overall set strength, it ensures that the sum of the weights of all sensors is always equal to 1, thus maintaining a dynamic balance in weight allocation under different interference environments.
[0078] Then, the target governance data at each sampling time is acquired. It should be noted that the target governance data should be the result of collaborative representation from multiple high-reliability sensing sources. Considering that using adaptive weights to linearly map the preliminary governance data from each sensor can preserve the true physical characteristics consistent with business logic to the greatest extent possible, while eliminating environmental noise interference, this approach is recommended.
[0079] Specifically, for any given sampling time, the target governance data satisfies the following relationship:
[0080] ;
[0081] in, Sampling time Target governance data; It is a sensor At sampling time Adaptive weights; It is a sensor At sampling time Preliminary governance data; It is a collection of sensors.
[0082] In this relation, Used for real-time adjustment of sensors The contribution ratio to trajectory synthesis. The adaptive weighting of a sensing source when its physical or operational reliability decreases. If the value approaches 0, the observations from that source are effectively ignored during the summation process; conversely, sensor data operating under normal logic will have a dominant weight to ensure the accuracy of target governance data. It is always anchored within a range of logically constrained physical real values.
[0083] It should be noted that, to prevent system crashes caused by a zero denominator, when the sum of the two-dimensional confidence products of all sensors in the sensor set is lower than a preset threshold, such as... At that time, the present invention will target governance data Locked to sampling time The reason for setting up the boundary locking mechanism is that when all sensing sources are untrusted, the system determines that it is currently in a state of severe physical interference or logical anomaly. By maintaining the historical trajectory state, it is used to avoid non-physical jumps being included in the trace data.
[0084] like Figure 2 As shown in the figure, this is a schematic diagram of the adaptive weight allocation evolution of multi-source sensors. The horizontal axis represents sampling time, and the vertical axis represents adaptive weights. The solid line, dashed line, and dotted line represent the adaptive fusion weights of sensor 1, sensor 2, and sensor 3, respectively. Within the sampling time interval of 180 to 260 seconds, due to sensor 2's feedback deviating from the expected business instructions or the presence of observational anomalies, the dashed line rapidly decreases to a lower level, achieving automatic isolation of untrusted data sources. Simultaneously, the solid line and dotted line rise synchronously and occupy the dominant weight, maintaining a dynamic balance in the allocation. This demonstrates that the present invention can detect performance degradation, shift the focus of governance to high-confidence sensors, and ensure the stability of the fusion results.
[0085] like Figure 3 As shown in the figure, this is a comparative diagram of the sensor data management effects. The horizontal axis represents sampling time, and the vertical axis represents the managed data. The dashed line represents the initial managed data from sensor 2, and the solid line represents the target managed data from sensor 2. Within the sampling time range of 180 to 260 seconds, the dashed line exhibits sharp shifts and oscillations, reflecting that the sensor is subject to physical sensing interference. Meanwhile, the solid line maintains stable fluctuations and does not shift with the dashed line. This indicates that the present invention successfully corrects the jumps in the observation stream by assessing confidence in real time, ensuring that the target managed data is anchored within the range of the physical true value, and eliminating false alarms caused by sensor anomalies.
[0086] It should be noted that the target governance data obtained by this invention is used to address the problem of the disconnect between physical perception interference and business logic. By using this data as the sole trusted input source for the digital twin platform, the entity coordinates in the virtual space can follow the real displacement constrained by business logic in real time, thereby eliminating false business warnings caused by sensor jumps and providing highly reliable digital traceability evidence for asset supervision.
[0087] Specifically, the target governance data at each sampling time is transmitted via a data bus. The data is pushed to the asset management backend of the logistics park in real time; the asset management backend calls a preset electronic fence spatial algorithm to determine the target governance data. Whether it enters or leaves a specific storage location logical area; if the determination result is that it has entered the target storage location area, the corresponding goods' entry status update instruction in the database will be automatically triggered; if the determination result is that the trajectory deviates from the preset business path, a logical anomaly alarm will be automatically sent to the management terminal.
[0088] The second aspect of this embodiment provides a smart logistics data governance system based on multi-source data fusion, such as... Figure 4 As shown, the intelligent logistics data governance system based on multi-source data fusion includes a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the intelligent logistics data governance method based on multi-source data fusion of the first aspect of the present invention is implemented.
[0089] The intelligent logistics data governance system based on multi-source data fusion also includes other components well known to those skilled in the art, such as communication buses and communication interfaces. Their settings and functions are known in the art and will not be described in detail here.
[0090] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (DRAM), high-bandwidth memory, hybrid memory cube, etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.
[0091] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.
Claims
1. A smart logistics data governance method based on multi-source data fusion, characterized in that, include: The system acquires raw data from sensors at various sampling times in the logistics operation site, along with associated business instruction streams. It then performs a preset normalization process on the raw data to obtain preliminary processing data for each sampling time. The raw data includes the physical coordinates of the sensors. For any sensor in the sensor set, its neighboring sensors are determined based on its physical coordinates, and the spatial deviation of the sensor at each sampling time is calculated based on the physical coordinates of the sensor and its neighboring sensors. The temporal change rate of the preliminary treatment data at adjacent sampling times is calculated, and the temporal change rate is fused with the spatial deviation to determine the physical confidence level of the sensor. , ;in, , These are sensors At sampling time , Preliminary governance data; It is the preset sensitivity adjustment coefficient; It is a sensor At the sampling time of its neighboring sensors Spatial deviation; It is the natural logarithm function; It is the natural exponential function; It is the absolute value symbol; Based on a preset instruction response model, the intensity of the service instruction in the service instruction stream is mapped to a characteristic change amplitude. Based on the deviation between the characteristic change amplitude and the time change rate, the service confidence of the sensor is determined, including: calculating the absolute value of the difference between the time change rate and the characteristic change amplitude, denoted as the absolute deviation; multiplying the absolute deviation by a preset time normalization factor to obtain a scaling deviation; processing the scaling deviation using a hyperbolic tangent function to obtain a nonlinear mapping value; obtaining the complement of the nonlinear mapping value to obtain the service confidence of the sensor. The intensity of the business instruction is preset by the WMS based on task priority and distance to the storage location, and the value range is: ; The adaptive weights of each sensor are determined based on the physical and operational confidence levels of each sensor. The preliminary governance data is then weighted and fused using these adaptive weights to obtain the target governance data.
2. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The calculation of the spatial deviation of the sensor at each sampling time based on the physical coordinates of the sensor and neighboring sensors includes: For any sampling time, calculate the geometric center of the physical coordinates of the neighborhood sensor to obtain the center coordinates; Calculate the Euclidean distance between the physical coordinates of the sensor and the center coordinates; Dividing the Euclidean distance by a preset neighborhood reference distance yields the spatial deviation of the sensor at the corresponding sampling time.
3. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The preset command response model is implemented through a command response function, which satisfies the following relation: ; in, It is an instruction response function; Sampling time The characteristic change amplitude; Sampling time The intensity of business instructions; It is the maximum expected rate of change of displacement of the sensor; It is the preset response gain coefficient; It is the logical threshold for the strength of business instructions; It is a natural exponential function.
4. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The process of determining the adaptive weights of each sensor based on its physical and operational confidence levels includes: Calculate the product of the physical confidence level and the operational confidence level of each sensor to obtain the joint confidence level of each sensor; The joint confidence scores of all sensors are summed to obtain the global confidence score. The ratio of the joint confidence level of any sensor to the sum of the global confidence levels is denoted as the adaptive weight of the corresponding sensor.
5. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The step of determining the neighboring sensors of any sensor in the sensor set based on its physical coordinates includes: For any sensor in the sensor set, obtain the physical coordinates of the sensor and the remaining sensors in the sensor set; Calculate the straight-line distance between the physical coordinates of the sensor and the physical coordinates of each of the remaining sensors; The remaining sensors are sorted in ascending order according to the straight-line distance; Extract the remaining sensors that rank in the top preset order as neighborhood sensors.
6. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The step of performing a preset normalization process on the original data to obtain preliminary processing data for each sampling time includes: Obtain the historical observation set of each sensor within a preset time window; The difference between the original data and the minimum value in the historical observation set is calculated, and then divided by the difference between the maximum value and the minimum value in the historical observation set to obtain the preliminary treatment data for each sampling time.
7. The intelligent logistics data governance method based on multi-source data fusion according to claim 1, characterized in that, The step of weighting and fusing the preliminary governance data using adaptive weights from each sensor to obtain the target governance data includes: For any sampling time, the preliminary treatment data of each sensor is multiplied by its corresponding adaptive weight to obtain the weighted treatment component of each sensor. The weighted governance components of all sensors in the sensor set are summed to obtain the target governance data at the corresponding sampling time.
8. A smart logistics data governance system based on multi-source data fusion, characterized in that, The intelligent logistics data governance system based on multi-source data fusion includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the intelligent logistics data governance method based on multi-source data fusion according to any one of claims 1-7 is implemented.