CN-121836552-B - Method and system for tracking whole process of medicine cold chain logistics transportation

Abstract

The invention provides a full-range tracking method and system for medical cold-chain logistics transportation, which relate to the technical field of logistics transportation monitoring, and are characterized in that an original sensing tracking data stream output by a sensor cluster in a transportation carrier in a continuous acquisition time window is firstly obtained, then a transportation carrier running track line is generated according to geographic positioning parameters, physical parameters at the same time as each geographic position coordinate point are combined to generate a multi-dimensional environment state label set, a logistics transportation node topological connection network comprising a logistics transportation node unit and a directional transportation path edge is constructed, the running track line, the multi-dimensional environment state label set and the topological connection network are input into a pre-constructed space-time state characteristic fusion tracking model, a full-range state tracking characteristic tensor is generated, abnormal state mode matching is carried out on the full-range state tracking characteristic tensor, abnormal state type labels and space-time distribution characteristics are obtained, and an abnormal early warning instruction set is generated. The intelligent monitoring system can comprehensively and accurately track the whole transportation process of the medicine cold chain, intelligently identify abnormality and early warn in time.

Inventors

• YUAN KAIJUN
• ZENG XIAOLAN
• WU HAIYAN

Assignees

• 成都易速物流有限公司

Dates

Publication Date

20260508

Application Date

20260312

Claims (10)

1. 1. A method for tracking the whole process of cold chain logistics transportation of medicines, which is characterized by comprising the following steps: Acquiring an original sensing tracking data stream output by a sensor cluster deployed in a transportation carrier in a medicine cold chain logistics transportation process in a continuous acquisition time window, wherein the original sensing tracking data stream comprises a temperature sensing parameter subsequence, a humidity sensing parameter subsequence, a vibration sensing parameter subsequence and a geographic positioning parameter subsequence with acquisition time stamp marks; Generating a transportation carrier running track line according to the moving sequence of the geographic position coordinate points in the geographic positioning parameter subsequence, and generating a multidimensional environment state marking set associated with each geographic position coordinate point in the transportation carrier running track line according to the temperature sensing parameter subsequence, the humidity sensing parameter subsequence and the physical parameters with the same time stamp as each geographic position coordinate point in the vibration sensing parameter subsequence; constructing a logistics transfer node topological connection network through which the transportation carrier operation track line passes, wherein the logistics transfer node topological connection network comprises a plurality of logistics transfer node units and directional transfer path edges connected with the logistics transfer node units, and the logistics transfer node units are obtained by matching geographic position coordinate points in the transportation carrier operation track line with a preset logistics transfer node database; Inputting the transportation carrier running track line, the multidimensional environment state label set and the logistics transportation node topological connection network into a pre-constructed space-time state characteristic fusion tracking model for joint analysis treatment, and generating a medicine cold chain logistics transportation whole-course state tracking characteristic tensor; And carrying out abnormal state pattern matching processing on the medicine cold chain logistics transportation whole-process state tracking feature tensor to obtain an abnormal state type identifier and abnormal state space-time distribution feature corresponding to the medicine cold chain logistics transportation whole-process state tracking feature tensor, and generating a medicine cold chain logistics transportation abnormal early warning instruction set according to the abnormal state type identifier and the abnormal state space-time distribution feature.
2. 2. The method of claim 1, wherein generating a transport carrier trajectory according to the order of movement of the geographic location coordinate points in the geographic location parameter sub-sequence comprises: Extracting a plurality of geographic position coordinate points arranged according to the sequence of acquisition time from the geographic positioning parameter subsequence, and performing redundant point elimination processing on the geographic position coordinate points to obtain an effective geographic position coordinate point sequence for constructing the transport carrier running track line, wherein the redundant point elimination processing is realized by calculating displacement distance parameters between adjacent geographic position coordinate points and filtering the geographic position coordinate points of which the displacement distance parameters are smaller than a preset static judging threshold value; Calculating time interval parameters and displacement direction angle parameters between adjacent effective geographic position coordinate points according to the acquisition time stamp corresponding to each effective geographic position coordinate point in the effective geographic position coordinate point sequence; respectively carrying out normalization processing on the time interval parameter and the displacement direction angle parameter, and generating a motion state change description vector between adjacent effective geographic position coordinate points according to the normalized time interval parameter and the displacement direction angle parameter, wherein the motion state change description vector comprises a standardized motion duration component and a standardized motion direction change component; Performing association combination processing on the effective geographic position coordinate point sequence and the motion state change description vector between the adjacent effective geographic position coordinate points to generate an initial track fragment set with time sequence consistency, wherein the initial track fragment set consists of the continuous effective geographic position coordinate points in the effective geographic position coordinate point sequence and the motion state change description vector corresponding to the continuous effective geographic position coordinate points; performing track segment splicing processing on the initial track segment set, and connecting the adjacent track segments meeting preset splicing conditions according to the geographic position coordinate point matching degree of the adjacent track segment endpoints in the initial track segment set and the continuity parameters calculated based on the motion state change description vector to generate the transport carrier running track line; carrying out logistics node proximity marking processing on each track point in the transportation carrier running track line, calculating the spatial distance parameter between each track point and all logistics transportation nodes in the preset logistics transportation node database, marking the track point with the spatial distance parameter smaller than a preset proximity distance threshold as a logistics node proximity point, and simultaneously recording a logistics transportation node identifier closest to the track point as a node association identifier; And determining a logistics transportation node sequence passed by the transportation carrier running track line according to the distribution density and the distribution sequence of the logistics node adjacent points on the transportation carrier running track line, wherein the logistics transportation node sequence is composed of the logistics transportation node identifiers which are sequentially arranged according to the running direction of the transportation carrier running track line.
3. 3. The method of claim 1, wherein generating the set of multi-dimensional environmental status markers associated with each of the geographic location coordinate points in the transport carrier travel trajectory from the temperature sensing parameter sub-sequence, the humidity sensing parameter sub-sequence, and the physical parameters in the vibration sensing parameter sub-sequence having the same time stamp as each of the geographic location coordinate points comprises: Extracting temperature instantaneous sampling values which are the same as the acquisition time stamp of each geographic position coordinate point from the temperature sensing parameter subsequence, arranging the temperature instantaneous sampling values according to the appearance sequence of the geographic position coordinate points on the transport carrier running track line, and generating a temperature parameter time sequence association list corresponding to the transport carrier running track line; Extracting humidity instantaneous sampling values which are the same as the acquisition time stamp of each geographic position coordinate point from the humidity sensing parameter subsequence, arranging the humidity instantaneous sampling values according to the appearance sequence of the geographic position coordinate points on the transport carrier running track line, and generating a humidity parameter time sequence association list corresponding to the transport carrier running track line; Extracting vibration instantaneous sampling values which are the same as the acquisition time stamp of each geographic position coordinate point from the vibration sensing parameter subsequence, arranging the vibration instantaneous sampling values according to the appearance sequence of the geographic position coordinate points on the transportation carrier running track line, and generating a vibration parameter time sequence association list corresponding to the transportation carrier running track line; Performing temperature state interval mapping processing on each temperature instantaneous sampling value in the temperature parameter time sequence association list, and converting each temperature instantaneous sampling value into a corresponding temperature state mark according to a preset temperature state division rule, wherein the temperature state mark is used for representing the deviation degree category of the internal environment temperature of the transportation carrier relative to the medicine storage requirement temperature range; Performing humidity state interval mapping processing on each humidity instantaneous sampling value in the humidity parameter time sequence association list, and converting each humidity instantaneous sampling value into a corresponding humidity state mark according to a preset humidity state division rule, wherein the humidity state mark is used for representing the deviation degree type of the internal environment humidity of the transport carrier relative to the medicine storage requirement humidity range; performing vibration intensity level mapping processing on each vibration instantaneous sampling value in the vibration parameter time sequence association list, and converting each vibration instantaneous sampling value into a corresponding vibration intensity level mark according to a preset vibration intensity level dividing rule, wherein the vibration intensity level mark is used for representing the level of mechanical impact intensity born by the transportation carrier in the transportation process; And carrying out combined packaging treatment on the temperature state mark, the humidity state mark and the vibration intensity level mark according to the association relation with the geographic position coordinate points to generate a multidimensional environment state mark set corresponding to each geographic position coordinate point in the transport carrier running track line.
4. 4. The method of claim 1, wherein the constructing the physical distribution node topology connection network along which the transport carrier running track line is routed comprises: extracting all track points marked as logistics node adjacent points from the transportation carrier running track line, and clustering the logistics node adjacent points belonging to the same logistics transportation node identifier according to the node association identifier corresponding to each logistics node adjacent point to obtain a node track point set corresponding to each logistics transportation node identifier; calculating the average value of the geographical position coordinates of all the logistics node adjacent points in each node track point set, and generating the actual passing position coordinates of the logistics transportation node units represented by each logistics transportation node identifier in the transportation carrier running track line; According to the time sequence of each logistics transfer node unit passing through the transportation carrier running track line in sequence, arranging the logistics transfer node units according to the time sequence, and generating a logistics transfer node sequence through which the transportation carrier is actually passed; for two adjacent logistics transfer node units in the logistics transfer node sequence, calculating a space linear distance between the two logistics transfer node units as a path distance parameter according to the actual passing position coordinate corresponding to the former logistics transfer node unit and the actual passing position coordinate corresponding to the latter logistics transfer node unit; determining a directional transfer path edge which is connected with a previous logistics transfer node unit and points to a next logistics transfer node unit according to the sequence of adjacent logistics transfer node units in the logistics transfer node sequence, and adding the path distance parameter on the directional transfer path edge as edge attribute information; And taking all the logistics transfer node units as network nodes, taking all the directed transfer path edges as directed edges connected with the network nodes, and constructing and forming the logistics transfer node topological connection network by combining the edge attribute information.
5. 5. The method of claim 1, wherein the step of inputting the carrier running trace, the multi-dimensional environmental state label set, and the logistics transportation node topology connection network into a pre-constructed space-time state feature fusion tracking model for joint analysis processing to generate a pharmaceutical cold chain logistics transportation whole-process state tracking feature tensor comprises the steps of: Inputting the transport carrier running track line to a track coding module of the space-time state feature fusion tracking model, and performing spatial position embedding representation processing on a geographic position coordinate point sequence in the transport carrier running track line to generate a spatial track embedding feature vector sequence corresponding to the transport carrier running track line; The multi-dimensional environment state mark set is input to a state coding module of the space-time state feature fusion tracking model, the temperature state mark, the humidity state mark and the vibration intensity level mark corresponding to each geographic position coordinate point in the multi-dimensional environment state mark set are subjected to joint embedding representation processing, environment state embedded feature vectors corresponding to each geographic position coordinate point are generated, all the environment state embedded feature vectors are arranged according to the occurrence sequence of the geographic position coordinate points, and an environment state embedded feature vector sequence is generated; inputting the logistics transfer node topology connection network to a topology coding module of the space-time state feature fusion tracking model, performing graph structure embedding representation processing on logistics transfer node units and directed transfer path edges in the logistics transfer node topology connection network, and generating a network topology embedding feature representation corresponding to the logistics transfer node topology connection network, wherein the network topology embedding feature representation comprises node embedding feature vectors of each logistics transfer node unit and edge embedding feature vectors of each directed transfer path edge; performing feature alignment and splicing processing on the space track embedded feature vector sequence and the environment state embedded feature vector sequence on a time axis to generate a primary combined feature tensor fusing space-time information, wherein each time step of the primary combined feature tensor simultaneously comprises space position information and environment state information of a corresponding time point; the primary combined feature tensor and the network topology embedded feature representation are input into a cross attention fusion module of the space-time state feature fusion tracking model, the cross attention fusion module takes the primary combined feature tensor as a query basis, takes the network topology embedded feature representation as a key value basis, calculates the association weights of the features of each time step in the primary combined feature tensor and each node and edge in the logistics transit node topology connection network through a multi-head cross attention mechanism, and carries out weighted aggregation on the network topology embedded feature representation according to the association weights to generate an enhanced feature tensor with network topology context information; And inputting the enhanced feature tensor into a time sequence convolution coding layer of the space-time state feature fusion tracking model, and checking the enhanced feature tensor through a plurality of time sequence convolution cores with different expansion rates to perform multi-scale time receptive field feature extraction processing on the enhanced feature tensor to generate the state tracking feature tensor in the whole process of medicine cold chain logistics transportation.
6. 6. The method of claim 5, wherein performing feature alignment and stitching on the spatial trajectory embedded feature vector sequence and the environmental state embedded feature vector sequence on a time axis to generate a primary joint feature tensor that fuses spatio-temporal information comprises: Acquiring an original acquisition time stamp corresponding to each space track embedded feature vector in the space track embedded feature vector sequence, and acquiring an original acquisition time stamp corresponding to each environment state embedded feature vector in the environment state embedded feature vector sequence; Performing alignment verification processing on a time axis on the space track embedded feature vector sequence and the environment state embedded feature vector sequence according to the original acquisition time stamp, and confirming that the space track embedded feature vector sequence and the environment state embedded feature vector sequence have a one-to-one correspondence in a time dimension; Splicing and combining the spatial track embedded feature vector and the environment state embedded feature vector at each time point after alignment in a feature channel dimension to generate a space-time combined feature vector corresponding to each time point; Stacking and arranging the space-time combined feature vectors corresponding to all time points according to a time sequence to form a two-dimensional feature matrix as the primary combined feature tensor, wherein the row dimension of the two-dimensional feature matrix corresponds to a time step length, and the column dimension corresponds to the total number of the spliced feature channels; and carrying out feature dimension normalization processing on the primary combined feature tensor, adjusting the numerical range of each feature channel in the primary combined feature tensor to a preset model input interval, and generating a normalized primary combined feature tensor with unified scale representation.
7. 7. The method of claim 5, wherein the cross-attention fusion module that inputs the primary combined feature tensor and the network topology embedded feature representation to the spatio-temporal state feature fusion tracking model generates an enhanced feature tensor with network topology context information, comprising: Linearly mapping the primary combined characteristic tensor into a query matrix required by a multi-head cross attention mechanism, wherein the number of rows of the query matrix corresponds to the number of time steps of the primary combined characteristic tensor, and the number of columns of the query matrix corresponds to a preset query characteristic dimension; Integrating all node embedded feature vectors and edge embedded feature vectors contained in the network topology embedded feature representation, and linearly mapping the node embedded feature vectors and the edge embedded feature vectors into a key matrix and a value matrix required by a multi-head cross attention mechanism, wherein the number of rows of the key matrix and the value matrix corresponds to the total number of the node embedded feature vectors and the edge embedded feature vectors, and the number of columns of the key matrix and the value matrix respectively corresponds to preset key feature dimensions and value feature dimensions; Performing dot product operation on the query matrix and the key matrix to obtain an attention score matrix, wherein each element in the attention score matrix represents the original score of the association strength of a time step feature in the primary combined feature tensor and a node or an edge in the network topology embedded feature representation; Performing scaling processing and softmax normalization processing on the attention score matrix to generate an attention weight matrix, wherein the sum of elements in each row in the attention weight matrix is one and is used for representing attention degree distribution of the characteristics of each time step to all nodes and edges; Carrying out weighted summation operation on the attention weight matrix and the value matrix to obtain a weighted network topology context feature matrix, wherein the number of lines of the weighted network topology context feature matrix is the same as the number of time steps of the primary combined feature tensor, and the number of columns of the weighted network topology context feature matrix is the same as the value feature dimension; and splicing and fusing the primary combined characteristic tensor and the weighted network topology context characteristic matrix in the characteristic channel dimension to generate the enhanced characteristic tensor with the network topology context information.
8. 8. The method according to claim 1, wherein the performing abnormal state pattern matching processing on the medical cold chain logistics transportation whole-process state tracking feature tensor to obtain an abnormal state type identifier and an abnormal state space-time distribution feature corresponding to the medical cold chain logistics transportation whole-process state tracking feature tensor comprises: Inputting the medicine cold chain logistics transportation whole-process state tracking feature tensor into a pre-trained abnormal state mode identification classifier, performing global feature extraction and mode classification processing on the medicine cold chain logistics transportation whole-process state tracking feature tensor by the abnormal state mode identification classifier, and outputting an abnormal state type identifier of the medicine cold chain logistics transportation whole-process state tracking feature tensor; Meanwhile, inputting the medicine cold chain logistics transportation whole-process state tracking feature tensor into a pre-trained abnormal space-time positioning network, and performing time-step feature analysis processing on the medicine cold chain logistics transportation whole-process state tracking feature tensor by the abnormal space-time positioning network to generate an abnormal confidence score sequence corresponding to each time step; screening out a continuous time step interval of which the abnormal confidence score exceeds a preset abnormal judgment threshold according to the abnormal confidence score of each time step in the abnormal confidence score sequence, and determining a corresponding time range of the continuous time step interval on the transportation carrier running track line as an abnormal state occurrence time period; Extracting a track segment corresponding to the abnormal state occurrence time period from the transportation carrier running track line as an abnormal track segment according to the abnormal state occurrence time period, and extracting an environment state mark subset corresponding to the abnormal state occurrence time period from the multi-dimensional environment state mark set as an abnormal environment state mark subset; Calculating a geographical area range covered by the abnormal state in the abnormal state occurrence time period according to the geographical position coordinate point sequence in the abnormal track segment, and generating an abnormal change trend description of the environment state in the abnormal state occurrence time period according to the distribution condition of each state mark in the abnormal environment state mark subset; and combining and packaging the abnormal state occurrence time period, the abnormal track segment, the abnormal environment state mark subset, the geographical area range covered by the abnormal state and the abnormal change trend description to generate the abnormal state space-time distribution characteristic.
9. 9. The method of claim 1, wherein generating the abnormal early warning instruction set for the transportation of the pharmaceutical cold chain logistics according to the abnormal state type identifier and the abnormal state space-time distribution feature comprises: analyzing the abnormal state type identifier, and searching an early warning level code corresponding to the abnormal state type identifier from a preset mapping relation library, wherein the early warning level code is used for indicating the severity level of the abnormal state; Extracting a starting time point and an ending time point of the abnormal state occurrence time period from the abnormal state space-time distribution characteristics, and extracting a starting geographic position coordinate point and an ending geographic position coordinate point in an abnormal track segment from the abnormal state space-time distribution characteristics as early warning trigger position coordinates; Extracting an abnormal environment state mark subset from the abnormal state space-time distribution characteristics, and generating an environment state abnormal evolution description vector according to a change sequence of a temperature state mark, a humidity state mark and a vibration intensity level mark in the abnormal environment state mark subset in the abnormal state occurrence time period; Combining the early warning level code, the starting time point, the ending time point, the early warning trigger position coordinate and the environment state abnormal evolution description vector to generate a basic early warning instruction unit aiming at the current detected abnormal state; judging whether the abnormal state type identifier belongs to a preset chain reaction abnormal type set, if the abnormal state type identifier belongs to the chain reaction abnormal type set, predicting a downstream logistics transfer node unit which is possibly affected by the abnormal state according to the logistics transfer node topological connection network, and generating a predictive early warning instruction unit aiming at the downstream logistics transfer node unit; And summarizing and sequencing all the generated basic early warning instruction units and the generated predictive early warning instruction units according to the time sequence and the spatial association relation to form the abnormal early warning instruction set for the medical cold chain logistics transportation.
10. 10. A full-range tracking system for cold chain logistics transportation of medicine, comprising: A processor; A machine-readable storage medium storing machine-executable instructions for the processor; Wherein the processor is configured to perform the pharmaceutical cold chain logistics transportation whole-process tracking method of any one of claims 1 to 9 via execution of the machine executable instructions.

Description

Method and system for tracking whole process of medicine cold chain logistics transportation Technical Field The invention relates to the technical field of logistics transportation monitoring, in particular to a full-range tracking method and system for medicine cold-chain logistics transportation. Background The medical product is extremely sensitive to environmental conditions such as temperature, humidity and the like, and abnormal fluctuation of any environmental parameter can cause deterioration and failure of the medical product, thereby affecting the health and safety of patients. Therefore, the whole process accurate tracking and effective monitoring of the medicine cold chain logistics transportation process are key requirements for industry development. At present, the existing medicine cold-chain logistics transportation tracking method has a plurality of limitations. Some methods rely on single sensor data to monitor, for example, only concern temperature parameters, but ignore other factors that may affect the quality of medical products, such as humidity and vibration, and cannot fully reflect the environmental state in the transportation process. In addition, although some methods collect various environmental parameter data, the data are simply recorded and displayed, the data are not subjected to deep analysis and fusion processing, and the potential risk and abnormal mode of the data hidden after the data are difficult to mine. In the aspect of transportation track tracking, the prior art can only provide general position information of a transportation carrier, can not accurately construct a running track line of the transportation carrier, and can not clearly show all logistics transportation nodes passing through in the transportation process and connection relations among the logistics transportation nodes. When a problem occurs, the specific position and link where the problem occurs are difficult to quickly locate, and effective countermeasures are not easy to take in time. In addition, the existing tracking method lacks an intelligent recognition and early warning mechanism for abnormal states. When the conditions of abnormal environmental parameters, deviated transportation track and the like occur in the transportation process, early warning cannot be sent out in time, so that problems cannot be treated in time, and the quality risk of medical products is increased. Disclosure of Invention In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a method for tracking the whole-process transportation of a pharmaceutical cold-chain logistics, which includes: Acquiring an original sensing tracking data stream output by a sensor cluster deployed in a transportation carrier in a medicine cold chain logistics transportation process in a continuous acquisition time window, wherein the original sensing tracking data stream comprises a temperature sensing parameter subsequence, a humidity sensing parameter subsequence, a vibration sensing parameter subsequence and a geographic positioning parameter subsequence with acquisition time stamp marks; Generating a transportation carrier running track line according to the moving sequence of the geographic position coordinate points in the geographic positioning parameter subsequence, and generating a multidimensional environment state marking set associated with each geographic position coordinate point in the transportation carrier running track line according to the temperature sensing parameter subsequence, the humidity sensing parameter subsequence and the physical parameters with the same time stamp as each geographic position coordinate point in the vibration sensing parameter subsequence; constructing a logistics transfer node topological connection network through which the transportation carrier operation track line passes, wherein the logistics transfer node topological connection network comprises a plurality of logistics transfer node units and directional transfer path edges connected with the logistics transfer node units, and the logistics transfer node units are obtained by matching geographic position coordinate points in the transportation carrier operation track line with a preset logistics transfer node database; Inputting the transportation carrier running track line, the multidimensional environment state label set and the logistics transportation node topological connection network into a pre-constructed space-time state characteristic fusion tracking model for joint analysis treatment, and generating a medicine cold chain logistics transportation whole-course state tracking characteristic tensor; And carrying out abnormal state pattern matching processing on the medicine cold chain logistics transportation whole-process state tracking feature tensor to obtain an abnormal state type identifier and abnormal state space-time distri