CN117273584B - Cargo tracking method and device and computer readable storage medium - Google Patents
Cargo tracking method and device and computer readable storage medium Download PDFInfo
- Publication number
- CN117273584B CN117273584B CN202311564042.3A CN202311564042A CN117273584B CN 117273584 B CN117273584 B CN 117273584B CN 202311564042 A CN202311564042 A CN 202311564042A CN 117273584 B CN117273584 B CN 117273584B
- Authority
- CN
- China
- Prior art keywords
- yuntag
- data
- model
- bluetooth
- power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000003860 storage Methods 0.000 title claims abstract description 9
- 230000008569 process Effects 0.000 claims abstract description 24
- 230000000694 effects Effects 0.000 claims abstract description 15
- 238000001514 detection method Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 36
- 238000003062 neural network model Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 19
- 230000007613 environmental effect Effects 0.000 claims description 13
- 238000012795 verification Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 4
- 238000004806 packaging method and process Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000006978 adaptation Effects 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims description 2
- 230000008054 signal transmission Effects 0.000 claims description 2
- 230000007246 mechanism Effects 0.000 abstract description 3
- 230000008859 change Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000003292 glue Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 230000036632 reaction speed Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0833—Tracking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a cargo tracking method, a cargo tracking device and a computer readable storage medium. By collecting bluetooth signal data in different environments, the BERT model is trained to predict the optimal transmit power. During the cargo transportation process, the YunTag detection signal changes, and the transmitting power is remotely adjusted according to the BERT model prediction result. And meanwhile, analyzing the transportation data to further optimize the model. The self-adaptive power control mechanism can reduce the power consumption of the YunTag while ensuring the Bluetooth tracking effect.
Description
Technical Field
The present invention relates to the field of logistics technologies, and in particular, to a method and apparatus for tracking goods, and a computer readable storage medium.
Background
Apple company published a novel device AirTag in 2021. The working principle of air tag is based on the powerful Find My Network of apple. This network forms a vast search network with hundreds of millions of apple devices, such as iPhone, iPad, and Mac, worldwide. Whenever an AirTag approaches any device in the network, it sends a secure, anonymous signal via bluetooth technology. This signal, when received by a nearby apple device, is uploaded to iboud and the position of the AirTag is recorded therein. The process is completely automatic and anonymous, ensures the privacy of the user and can provide accurate position tracking information.
The AirTag is also equipped with a built-in speaker that can sound when the user looks for it through the Find My application, helping the user locate its position faster. In addition, apple corporation is also equipped with Ultra Wideband (UWB) technology for AirTag, which is applicable to the latest apple devices. UWB technology provides more accurate spatial awareness capabilities than traditional bluetooth, enabling users to more accurately find the specific location of airtags. For example, the user can see the exact direction and distance to the AirTag in the Find My application, which is extremely useful when looking for lost items.
Apple company designed two modes for AirTag: normal mode and lost mode. When the user sets the AirTag to the normal mode, the AirTag will receive the bluetooth signal at a compared frequency, thereby maintaining a relatively low power consumption. When the user sets the AirTag to the lost mode, the AirTag will receive the bluetooth signal at a higher frequency, increasing the probability that the AirTag is found. Apple corporation claims that such a design can ensure that the AirTag does not need to change batteries over a period of one year.
Although this approach may take into account power consumption and tracking efficiency to some extent, it is not able to adaptively balance power consumption and tracking efficiency according to the signal characteristics of the scene. Apple corporation opened the Find My Network alliance in 2021, which can be used to locate items as long as allied products are added. Accordingly, we have devised a new cargo tracking method, apparatus and computer readable storage medium that utilizes artificial intelligence techniques to adapt power consumption and tracking efficiency.
Disclosure of Invention
In view of the problems in the logistics path planning in the prior art, on the one hand, the invention provides a cargo tracking method, which completes cargo tracking by utilizing a yunTag, wherein the yunTag is an electronic device accessed into the Find My Network alliance and used for cargo tracking and positioning, and specifically comprises the following steps:
s1, collecting Bluetooth signal intensity data which can be received by the yunTag under different environments, wherein the Bluetooth signal intensity data comprise different scenes such as indoor, outdoor, crowd-intensive and the like, and the data are used for training a BERT model to predict the optimal yunTag transmitting power. The method comprises the following steps:
s1.1, setting the YunTag in different environmental scenes, including areas with sparse and dense indoor and outdoor areas and crowd, and recording parameters of all Bluetooth signals which can be received by the YunTag, wherein the parameters comprise RSSI, frequency, equipment types and the like.
S1.2, collecting a large amount of YunTag Bluetooth signal data under different environments, making a data cleaning rule, deleting invalid and abnormal data, and ensuring the data quality.
S1.3, dividing the data into a training data set, a verification data set and a test data set according to a certain proportion. The training data set is used to train the BERT model, the validation set is used for parameter adjustment, and the test set is used to evaluate the model performance.
S1.4, establishing a BERT neural network model, wherein an input layer is an environment Bluetooth signal characteristic, and an output layer is a recommended YunTag transmitting power value. And designing a model structure and determining the super parameters.
S1.5, training the BERT model by using training data, and continuously adjusting model parameters by adopting an Adam optimization algorithm to minimize a loss function, wherein a verification set is used for verifying effects. The best parameters are saved.
S1.6, evaluating the performance of the trained BERT model on a test set, recording indexes such as precision, recall rate and the like, and adjusting the model structure and super parameters to improve the effect if necessary.
S2, loading the YunTag on goods to be tracked, wherein when the goods are transported, the YunTag can read Bluetooth signals in the surrounding environment. The method comprises the following steps:
s2.1, selecting representative cargoes of different types to test, wherein the cargoes comprise boxes, packages and the like with different sizes, shapes and materials, and the collection of sufficient data is ensured.
S2.2, selecting proper positions on the surfaces or the interiors of the cargoes, and firmly binding the yunTag on the cargoes by using elastic ropes and the like to ensure that the yunTag cannot fall off in the process of transportation vibration.
S2.3, setting UID of the YunTag, inputting related information of the goods to be tracked, such as the number of the goods, required tracking time length and the like, and establishing a corresponding relation between the goods and the YunTag.
And S2.4, charging the YonTag, so that the electric quantity of the YonTag can be ensured to maintain at least one complete test transportation link, and interruption of data collection due to insufficient half-way electric quantity is avoided.
And S2.5, transporting the goods attached with the YunTag according to a contracted flow, and detecting and recording Bluetooth signals in the transportation process by the YunTag in real time to provide data for model training.
And S3, inputting the read Bluetooth signal data into a pre-trained BERT-based neural network model, wherein the model can predict the optimal transmitting power which is supposed to be used by the YunTag in the environment. The method comprises the following steps:
s3.1, during the cargo transportation process, the YunTag detects Bluetooth signals in the surrounding environment at fixed time intervals, including every 10 seconds, and the Bluetooth signals include RSSI, MAC addresses, frequency and other information. These signals come from the signal transmissions of other bluetooth devices.
S3.2, preprocessing the detected original Bluetooth signal data, including removing repeated redundant information, formatting and converting the redundant information into an input format which can be directly accepted by a model, adding other characteristic information, and removing detection errors.
S3.3, the preprocessed data is input into a previously trained BERT-based neural network model, which can analyze the input data and output a predicted YunTag optimal transmit power value.
S3.4, the input of the model comprises processed Bluetooth signal parameter data, and the output is a specific power value which represents the optimal transmitting power which is set by the YunTag in the environment.
And S3.5, the predicted power value output by the model is used as a reference for adjusting the power of the YunTag, and a control command is sent according to the predicted power value to actually adjust the transmitting power of the YunTag.
And S4, correspondingly adjusting the Bluetooth signal transmitting power of the YenTag according to the prediction result of the BERT model so as to adapt to the environment. The method comprises the following steps:
s4.1, according to the digital power value predicted by the BERT neural network model, the digital power value needs to be converted into a control parameter which can be understood by the YunTag Bluetooth module. This conversion is performed according to the mapping relation specified in the data manual.
S4.2, packaging the obtained control information of the YenTag Bluetooth parameters into an instruction, wherein the instruction comprises information such as a target power value, an adjustment time interval and the like, and checking information to ensure the correctness of the instruction.
S4.3, transmitting the instruction of transmitting power adjustment to the bound YunTag on the goods safely through Bluetooth connection. The transmission requires processing such as encoding, encryption, verification, etc. of the signal.
And S4.4, after the YonTag receives the instruction, analyzing the instruction content, and gradually adjusting the transmitting power parameter of the Bluetooth module within a given time interval until the power value is modified to be the target power value designated by the instruction.
S5, when the goods are continuously transported, the YunTag can continuously read Bluetooth signals in the surrounding environment, and the model prediction and power adjustment process is repeated. The method comprises the following steps:
s5.1, along with the movement of goods in the transportation process, the YunTag can continuously and periodically detect Bluetooth signals in the surrounding environment to acquire the latest Bluetooth equipment data in the transportation environment.
S5.2, preprocessing new Bluetooth signal data repeatedly, including format conversion, denoising, input vector construction and the like, and processing the new Bluetooth signal data into an input format which can be directly accepted by a model.
S5.3, inputting the processed data into the BERT neural network model again to obtain a predicted value of the proposed transmitting power of the YunTag in the brand new environment.
S5.4, comparing the newly predicted power value with the power value actually set by the YonTag currently, and judging whether a new control instruction needs to be sent to adjust the power.
S5.5, if the difference between the predicted power value and the current actual power exceeds a set threshold value, a new power control instruction is constructed, and the instruction is repeatedly sent to adjust the Bluetooth transmitting power of the YunTag.
S6, through coordination adjustment of the transmitting power of the YonTag, the power consumption can be reduced as much as possible and the electric quantity can be saved on the premise that the goods can be effectively tracked. The method comprises the following steps:
and S6.1, after the whole cargo transportation is completed, counting the number of times of the adjustment of the YUNTAG Bluetooth transmitting power in the transportation and the parameter amplitude of each adjustment.
S6.2, analyzing the adjusted parameter amplitude, evaluating the dynamic adjustment range of the YUNTAG transmitting power in the transportation, and observing the characteristics of the adjustment value under different environments.
And S6.3, checking the battery electric quantity of the YunTag, counting the difference value of the electric quantity before and after transportation, and estimating the battery electric quantity consumption condition of the YunTag in the whole transportation process.
And S6.4, collecting electric quantity consumption data of the YonTag in the fixed transmitting power mode, comparing the electric quantity consumption data with the current consumption data, and evaluating the electric quantity percentage saved in the self-adaptive adjustment mode.
And S6.5, finishing the number of times of the adjustment of the YonTag power, the adjustment amplitude and the electric quantity consumption data in the transportation, generating a transportation report, and providing a reference basis for further optimization.
And S7, after the transportation is completed, collecting and analyzing power adjustment data and power consumption conditions of the YonTag in the whole transportation process so as to further optimize the effect of training the BERT model. The method comprises the following steps:
and S7.1, continuously training the BERT neural network model by using the collected new data, and enabling the BERT neural network model to accurately predict the optimal transmitting power of the YunTag in a given environment by adjusting model parameters.
S7.2, manually labeling a data set of a batch of model prediction results, evaluating the adaptation effect of the prediction power values to different environments, analyzing the error types of the models, and finding out the aspect which still needs to be improved.
And S7.3, according to the error analysis of the manual labeling result, the model structure is adjusted, the training strategy is improved, regularization and enhancement measures are added, and training is repeated until the accuracy of model predicted power and recall rate indexes reach expectations.
In addition, the invention also provides a cargo tracking device and a computer readable storage medium, wherein the device is used for executing the cargo tracking method, and the computer storage medium is stored with computer executable instructions for executing the steps in the cargo tracking method.
Drawings
Fig. 1 is a flowchart of a cargo tracking method provided by the present invention.
Detailed Description
As shown in fig. 1, an embodiment of the present invention provides a cargo tracking method, which includes the following steps:
s1, collecting Bluetooth signal intensity data which can be received by the YunTag under different environments;
s2, loading the YunTag on goods to be tracked, wherein when the goods are transported, the YunTag can read Bluetooth signals in the surrounding environment;
s3, inputting the read Bluetooth signal data into a pre-trained BERT-based neural network model, wherein the model can predict the optimal transmitting power which is required to be used by the YunTag in the environment;
s4, according to the prediction result of the BERT model, correspondingly adjusting the Bluetooth signal transmitting power of the YunTag to adapt to the environment;
s5, when goods are continuously transported, the YunTag continuously reads Bluetooth signals in the surrounding environment, and the model prediction and power adjustment process is repeated;
s6, through coordination adjustment of the transmitting power of the YonTag, the power consumption can be reduced as much as possible and the electric quantity can be saved on the premise that the goods can be effectively tracked;
and S7, after the transportation is completed, collecting and analyzing power adjustment data and power consumption conditions of the YonTag in the whole transportation process so as to further optimize the effect of training the BERT model.
The YunTag is an electronic device which is accessed to the Find My Network alliance and used for tracking and positioning cargoes, and is internally provided with a Bluetooth receiving module which is used for receiving Bluetooth signals sent by iPhone, iPad and Mac in the Find My Network.
Specifically, the above steps may be further refined to:
s1, collecting Bluetooth signal intensity data which can be received by the yunTag under different environments, wherein the Bluetooth signal intensity data comprise different scenes such as indoor, outdoor, crowd-intensive and the like, and the data are used for training a BERT model to predict the optimal yunTag transmitting power. The method comprises the following steps:
s1.1, placing the YunTag in different environmental scenes, including areas such as offices, supermarkets, buses, subway carriages, exhibition places, stadiums and the like, and recording parameters of all Bluetooth signals which can be received by the YunTag, wherein the parameters comprise detailed information such as RSSI, frequency, equipment types, signal source positions and the like. The difference of crowd flow and density under different environments needs to be considered, and the influence caused by indoor heterodyne is also needed.
S1.2, collecting a large amount of original data of the YunTag Bluetooth signals under different environments, and customizing a data cleaning flow according to project requirements, such as removing data with signal strength of 0, deleting redundant data with too dense signal sources, eliminating environmental noise influence and the like. And meanwhile, an environment label for recording data is needed for model training. Training sets of sufficient size and variety are collected to train models with good generalization performance.
S1.3 the collected data sets are divided into training data sets, validation data sets and test data sets in proportions of for example 80%, 10%. The data are shuffled according to a certain rule and then divided, so that the data category balance is ensured. The training set is used for training the neural network model, the verification set is used for adjusting the super parameters, and the test set is used for evaluating the performance of the model. Multiple shuffles and split are used to select the combination with the best segmentation result.
S1.4, customizing the BERT type neural network model, and considering factors such as input and output dimensions, network layer number, activation function and the like. The input layer is a Bluetooth signal characteristic sequence subjected to vectorization processing, and the output layer is a model recommended YunTag transmitting power value. Different model structures may be tried for comparative evaluation. And determining the proper number of hidden layers, and the dropout proportion and other super parameters.
S1.5, loading the segmented training set, training a network model by adopting adaptive optimization algorithms such as Adam and the like, and adjusting network weights to minimize a loss function. And recording the change condition of the loss function curves of the training set and the verification set. And saving the model parameters when the loss function is not reduced any more. Training parameters such as learning rate, epoch number and the like can be trained and adjusted for multiple times so as to obtain the optimal effect.
And S1.6, verifying the performance of the trained model by using a test set, and recording various evaluation indexes such as MSE, MAE, R scores and the like. The model structure and key super parameters can be adjusted again to improve if necessary until the performance of the model on the test set reaches the project requirement. The public dataset may be used to further verify the generalization performance of the model.
S2, loading the YunTag on goods to be tracked, wherein when the goods are transported, the YunTag can read Bluetooth signals in the surrounding environment. The method comprises the following steps:
s2.1, selecting cargoes with various shapes, materials and sizes for testing, such as cartons, wooden boxes, plastic boxes, express packages, mask boxes, food bags and the like, and covering different product packaging types. The cargo parameters are recorded each time. The data coverage is ensured to be wide.
S2.2, fastening and mounting the yunTag at a relatively stable position on the surface or in the cargo by using a proper binding belt or glue so as to ensure that the yunTag cannot fall off in the transportation process. The installation position needs to consider to avoid shielding and ensure normal propagation of Bluetooth signals. The test can also be repeated at various positions to select the optimal scheme.
S2.3, compiling a unique ID for each YonTag, and simultaneously marking specific information of the goods to be tracked, such as the name, specification and model of the goods, required tracking time length and the like. And establishing a binding relation database of the goods and the YunTag. Repeated use of the same YunTag requires erasing the data to avoid confusion.
And S2.4, calculating average working current of the YunTag according to historical data, calculating predicted transportation time in combination with battery capacity, and charging the YunTag in advance to ensure sufficient electric quantity. The power consumption condition of different charging time can also be tested by a multiplex. The charge is to a level of charge sufficient to maintain the complete test transportation.
S2.5, according to the simulation test of the standard express delivery transportation environment, the goods attached with the YunTag are subjected to various transportation steps such as loading and unloading, carrying, storage, jolting of transportation means and the like. The YunTag detects and records the bluetooth signals in the process in real time. Multiple repeated tests are required to collect sufficient data.
And S3, inputting the read Bluetooth signal data into a pre-trained BERT-based neural network model, wherein the model can predict the optimal transmitting power which is supposed to be used by the YunTag in the environment. The method comprises the following steps:
and S3.1, in the transportation process, the YunTag continuously detects surrounding Bluetooth equipment signals at a fixed frequency, and records parameters such as signal strength and the like. The detection frequency needs to consider the YunTag data processing capability and real-time requirements. And simultaneously recording information such as detection time stamps, position coordinates and the like. It is necessary to ensure that complete time series data is collected.
S3.2, preprocessing the original detection data, including removing duplicate and abnormal data, summarizing and counting the same signal source information, converting the same signal source information into a model receiving format and the like. It is also necessary to construct input sample weights, highlight important features. The aim of preprocessing is to improve the data quality, highlighting key features.
And S3.3, inputting the processed data into a pre-trained BERT neural network model, and analyzing input characteristics by the model to give a YUNTAG optimal transmission power prediction in the current environment. Intermediate layer characteristics can be visualized and the internal working mechanism of the model can be analyzed.
And S3.4, the model input is encoded Bluetooth signal sequence information, and the output is a corresponding power predicted value. To get a more accurate prediction, the historical sequence may be input for Markov modeling. The model may also predict the trend of signal changes at the same time.
And S3.5, the power prediction result given by the model can be directly used as a reference target value for adjusting the YUNTAG transmitting power. Or after comparing with the current setting value, the adjustment is performed after exceeding the determined threshold value, so that unnecessary adjustment caused by small changes is avoided.
And S4, correspondingly adjusting the Bluetooth signal transmitting power of the YenTag according to the prediction result of the BERT model so as to adapt to the environment. The method comprises the following steps:
s4.1, converting the digital power value predicted by the model into a control parameter format which can be identified by the YunTag Bluetooth module according to a preset mapping rule. The mapping rules require reference to the hardware circuit scheme and data manual of YunTag. After conversion, a control instruction which can be directly executed is formed.
And S4.2, packaging and rechecking the control instruction, and adding information such as check codes and the like for transmission verification. A fixed length instruction format may be provided containing the necessary control fields. Fields such as a joining source address, a destination address, etc. identify both parties of the communication. And adding a CRC check code to ensure the integrity of the instruction.
And S4.3, the packaged control instruction is safely sent to the target YonTag through the encrypted Bluetooth link. The communication needs to be authenticated, and the encryption algorithm can use AES or the like to combat attacks. The addition of a retransmission mechanism ensures that instructions arrive. Instruction transmission requires that the device address corresponds to YunTag.
And S4.4, decoding and checking the information after receiving the instruction by using the YonTag, and starting to execute the transmission power adjustment after confirming that the instruction source is reliable. The value of the relevant control register is modified according to the instruction content, and the value is accurately adjusted to the target power value. To ensure stability, a gradual adjustment may be provided, eventually converging to an ideal value.
S5, when the goods are continuously transported, the YUNTAG continuously detects signals and repeats the model prediction adjustment process. The method comprises the following steps:
s5.1, periodically detecting the latest surrounding Bluetooth signals by using the YunTag along with the change of the transportation environment, and acquiring real-time data of different positions. The continuous monitoring can learn about environmental changes and trigger corresponding adjustments. Detecting the frequency requires balancing the reaction speed and the data load.
And S5.2, preprocessing the new detection data repeatedly, improving the quality and highlighting the characteristics. And adding new data into the sequence history, and maintaining the dependency relationship by combining the context when inputting the model. And parameterization is needed for data preprocessing, so that the method is convenient to quickly adapt to a new environment.
And S5.3, the model analyzes the Bluetooth signal sequence information after new data is input, and updates and predicts the optimal transmitting power value of the YunTag in the brand new environment. Adding new data can help the model determine environmental changes. A portion of the new data verification model output may be sampled.
S5.4, comparing the model predicted new power value with the current actual setting value of the YonTag. If the preset error range is exceeded, an adjustment is necessary. Setting a reasonable error range can avoid frequent adjustment. The tolerance of the signal fluctuations is taken into account.
And S5.5, when the adjustment is needed, repeating the construction control instruction, and sending the construction control instruction to the YunTag to execute new power setting. Repeated verification of instruction content prior to transmission avoids errors. The adjustment amplitude should not be too large, and the conversion needs to be completed gradually.
S6, through coordination adjustment of the YUNTAG transmitting power, power consumption is reduced and electric quantity is saved on the premise of ensuring tracking effect. The method comprises the following steps:
s6.1, counting the times of the adjustment of the YonTag power in the whole transportation process and the parameter change before and after each adjustment. The mode of the adjusted time point and amplitude is analyzed. Multiple test statistical analyses can find the adjustment law.
S6.2, analyzing the influence degree of different environmental factors such as temperature, shielding and transportation means on the adjustment value. the power value and the environmental data are analyzed together to summarize the mapping relation of adjusting the amounts to the environmental sensitivity. This helps the model learn environmental impact factors.
And S6.3, accurately recording the change of the battery power before and after the YUNTAG transportation, and calculating the total power consumption in the transportation process. The influence of natural discharge needs to be considered. Repeated measurement and comparison can improve the calculation accuracy and quantify the influence of power adjustment on energy conservation.
And S6.4, collecting the power consumption in the fixed power mode, and comparing the power consumption with the power consumption in the self-adaptive mode in a similar environment. And calculating the electric quantity proportion which can be saved by the self-adaptive mode average under the condition of ensuring the tracking effect. The result can evaluate the energy saving optimization effect of the design.
And S6.5, generating a transportation report by integrating the statistical data, and displaying information such as adjustment times, environmental factors, power consumption and the like. And (3) comparing the data before and after the optimization, and quantitatively summarizing benefit improvement brought by self-adaptive adjustment. The report may PUSH to the management platform.
And S7, collecting and analyzing the data of the power and the electric quantity of the YonTag in transportation, and further optimizing and training the BERT model to improve the prediction accuracy. The method comprises the following steps:
s7.1, training the BERT model with the collected new data increment, and enabling the model to judge the Bluetooth signal characteristics in each environment more accurately through parameter adjustment. The new key samples can improve the pertinence of the model. The training round number, the regular term and the like are adjusted to find the optimal parameters.
S7.2, manually marking part of sample data, and judging the accuracy of model output power prediction and the environment adaptation effect. And analyzing the error distribution condition of the model under different environments. The manual proposal is given to adjust the gradient direction.
S7.3, analyzing error feedback of the manual annotation, and adopting a new training method such as enhancement learning, transfer learning and other improved models. And (3) adjusting the model structure until the precision and recall index reach the expectations.
The beneficial effects of the invention are as follows: and the optimal power is predicted by training a model, and the YonTag power is controlled by a prediction result, so that the closed-loop control of automatically adjusting the power output in real time according to the environmental signal is realized. Under a complex transportation environment, the method can continuously optimize the model, improve the environmental adaptability, effectively reduce the power consumption on the premise of ensuring the Bluetooth tracking effect, and utilize the self-adaptive control to replace the fixed power output so as to avoid unnecessary energy waste.
Claims (9)
1. A method of tracking cargo, the method comprising the steps of:
s1, collecting Bluetooth signal intensity data which can be received by a YunTag in different environments, wherein the YunTag is electronic equipment which is accessed into the Find My Network alliance and used for tracking and positioning cargoes, and a Bluetooth receiving module is arranged in the electronic equipment and is used for receiving Bluetooth signals sent by iPhone, iPad and Mac in the Find My Network;
s2, loading the YunTag on goods to be tracked, wherein when the goods are transported, the YunTag can read Bluetooth signals in the surrounding environment;
s3, inputting the read Bluetooth signal data into a pre-trained BERT neural network model, wherein the model can predict the optimal transmitting power which is required to be used by the YunTag in the environment, and specifically comprises the following steps:
s3.1, during the cargo transportation process, the YunTag detects Bluetooth signals in the surrounding environment at fixed time intervals, including every 10 seconds, including RSSI, MAC address and frequency information; these signals come from the signal transmissions of other bluetooth devices;
s3.2, preprocessing the detected original Bluetooth signal data, including removing repeated redundant information, formatting and converting the redundant information into an input format which can be directly accepted by a model, and removing detection errors;
s3.3, the preprocessed data is input into a BERT neural network model trained before, and the model can analyze the input data and output a predicted YunTag optimal transmitting power value;
s3.4, the input of the model comprises processed Bluetooth signal parameter data, and the output is a specific power value which represents the optimal transmitting power which is set by the YunTag in the environment;
s3.5, the predicted power value output by the model is used as a reference for adjusting the power of the YunTag, and a control command is sent according to the predicted power value to actually adjust the transmitting power of the YunTag;
s4, according to the prediction result of the BERT neural network model, correspondingly adjusting the Bluetooth signal transmitting power of the YunTag to adapt to the environment;
s5, when goods are continuously transported, the YunTag continuously reads Bluetooth signals in the surrounding environment, and the model prediction and power adjustment process is repeated;
s6, through coordination adjustment of the transmitting power of the YonTag, the power consumption can be reduced as much as possible and the electric quantity can be saved on the premise that the goods can be effectively tracked;
and S7, after the transportation is completed, collecting and analyzing power adjustment data and power consumption conditions of the YonTag in the whole transportation process so as to further optimize the effect of training the BERT neural network model.
2. The method according to claim 1, wherein the step S1 is specifically:
s1.1, placing the YunTag in different environmental scenes, including areas with sparse and dense indoor and outdoor people, and recording parameters of all Bluetooth signals which can be received by the YunTag, including RSSI, frequency and equipment type information;
s1.2, collecting a large amount of YunTag Bluetooth signal data under different environments, making a data cleaning rule, deleting invalid and abnormal data, and ensuring the data quality;
s1.3, dividing data into a training data set, a verification data set and a test data set according to a certain proportion; the training data set is used for training the BERT neural network model, the verification data set is used for parameter adjustment, and the test data set is used for evaluating the performance of the model;
s1.4, establishing a BERT neural network model, wherein an input layer is an environment Bluetooth signal characteristic, and an output layer is a recommended YunTag transmitting power value; designing a model structure and determining super parameters;
s1.5, training the BERT neural network model by using training data, continuously adjusting model parameters by adopting an Adam optimization algorithm to minimize a loss function, and using a verification data set for verification effect.
3. The method according to claim 1, wherein the step S2 is specifically:
s2.1, selecting representative cargoes of different types for testing, wherein the cargoes comprise boxes and packages with different sizes, shapes and materials, and the collection of sufficient data is ensured;
s2.2, selecting proper positions on the surfaces or the interiors of the cargoes, and firmly binding the YunTag on the cargoes by using elastic ropes to ensure that the YunTag cannot fall off in the process of transportation vibration;
s2.3, setting UID of the YunTag, inputting related information of the goods to be tracked, including the number of the goods and the required tracking time length, and establishing a corresponding relation between the goods and the YunTag;
s2.4, charging the YonTag, ensuring that the electric quantity can maintain at least one complete test transportation link, and avoiding interruption of data collection due to insufficient half-way electric quantity;
and S2.5, transporting the goods attached with the YunTag according to a contracted flow, and detecting and recording Bluetooth signals in the transportation process by the YunTag in real time to provide data for model training.
4. The method according to claim 1, wherein the step S4 is specifically:
s4.1, according to the digital power value predicted by the BERT neural network model, converting the digital power value into a control parameter which can be understood by a YunTag Bluetooth module; this conversion is performed according to a mapping relation defined by the data manual;
s4.2, packaging the obtained control information of the YonTag Bluetooth parameters into an instruction, wherein the instruction comprises a target power value, time interval adjustment information and check information to ensure the correctness of the instruction;
s4.3, safely transmitting the instruction for adjusting the transmitting power to the bound yunTag on the goods through Bluetooth connection; the transmission needs to encode, encrypt and check the signal;
and S4.4, after the YonTag receives the instruction, analyzing the instruction content, and gradually adjusting the transmitting power parameter of the Bluetooth module within a given time interval until the power value is modified to be the target power value designated by the instruction.
5. The method according to claim 1, wherein the step S5 is specifically:
s5.1, along with the movement of goods in the transportation process, the YunTag continuously and periodically detects Bluetooth signals in the surrounding environment to acquire the latest Bluetooth equipment data in the transportation environment;
s5.2, preprocessing new Bluetooth signal data repeatedly, including format conversion, denoising and constructing an input vector, and processing the new Bluetooth signal data into an input format which can be directly accepted by a model;
s5.3, inputting the processed data into the BERT neural network model again to obtain a predicted value of the proposed transmitting power of the YunTag in the brand new environment;
s5.4, comparing the newly predicted power value with the power value actually set by the YonTag currently, and judging whether a new control instruction needs to be sent to adjust the power;
s5.5, if the difference between the predicted power value and the current actual power exceeds a set threshold value, a new power control instruction is constructed, and the instruction is repeatedly sent to adjust the Bluetooth transmitting power of the YunTag.
6. The method according to claim 1, wherein the step S6 is specifically:
s6.1, after the whole cargo transportation is completed, summarizing and counting the number of times of the adjustment of the YUNTAG Bluetooth transmitting power in the transportation and the parameter amplitude of each adjustment;
s6.2, analyzing the adjusted parameter amplitude, evaluating the dynamic adjustment range of the YUNTAG transmitting power in the transportation, and observing the characteristics of the adjustment values in different environments;
s6.3, checking the battery electric quantity of the YunTag, counting the difference value of the electric quantity before and after transportation, and estimating the battery electric quantity consumption condition of the YunTag in the whole transportation process;
s6.4, collecting electric quantity consumption data of the YonTag in a fixed transmitting power mode, comparing the electric quantity consumption data with the current consumption data, and evaluating the electric quantity percentage saved in the self-adaptive adjustment mode;
and S6.5, finishing the number of times of the adjustment of the YonTag power, the adjustment amplitude and the electric quantity consumption data in the transportation, generating a transportation report, and providing a reference basis for further optimization.
7. The method according to claim 1, wherein the step S7 is specifically:
s7.1, continuously training the BERT neural network model by using the collected new data, and enabling the BERT neural network model to accurately predict the optimal transmitting power of the YunTag in a given environment by adjusting model parameters;
s7.2, manually labeling a batch of data sets of model prediction results, evaluating the adaptation effect of the prediction power values to different environments, analyzing the error types of the models, and finding out the aspect which still needs to be lifted;
and S7.3, according to the error analysis of the manual labeling result, adjusting the model structure, improving the training strategy, adding regularization and enhancement measures, and repeating training until the accuracy of the model predicted power and the recall rate index reach the expectations.
8. A cargo transportation device, characterized in that the device is arranged to perform the steps of the method according to any one of claims 1-7.
9. A cargo tracking computer readable storage medium having stored thereon computer instructions for performing the steps of any of the methods of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311564042.3A CN117273584B (en) | 2023-11-22 | 2023-11-22 | Cargo tracking method and device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311564042.3A CN117273584B (en) | 2023-11-22 | 2023-11-22 | Cargo tracking method and device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117273584A CN117273584A (en) | 2023-12-22 |
| CN117273584B true CN117273584B (en) | 2024-03-22 |
Family
ID=89220030
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311564042.3A Active CN117273584B (en) | 2023-11-22 | 2023-11-22 | Cargo tracking method and device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117273584B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106332132A (en) * | 2015-07-03 | 2017-01-11 | 中兴通讯股份有限公司 | Communication terminal for performance adjustment of self-adaptable Bluetooth and method |
| CN111027915A (en) * | 2019-12-17 | 2020-04-17 | 哈尔滨工业大学 | Intelligent logistics safety auxiliary system based on Bluetooth technology |
| CN116193409A (en) * | 2022-12-23 | 2023-05-30 | 上海普瑞联网络科技有限公司 | Bluetooth signal receiving method, system, equipment and medium capable of adjusting signal range |
| CN116216075A (en) * | 2023-02-02 | 2023-06-06 | 山东零度供应链有限公司 | Tracking method based on packaging box, system and electronic equipment thereof |
| KR20230150151A (en) * | 2022-04-21 | 2023-10-30 | (주)디에스네트웍스 | IoT-integrated WIFI AP wireless controller considering artificial intelligence-based wireless WiFi security |
| CN116980834A (en) * | 2023-09-25 | 2023-10-31 | 深圳市康凯铭科技有限公司 | Low-power consumption control method and system for Bluetooth beacon |
-
2023
- 2023-11-22 CN CN202311564042.3A patent/CN117273584B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106332132A (en) * | 2015-07-03 | 2017-01-11 | 中兴通讯股份有限公司 | Communication terminal for performance adjustment of self-adaptable Bluetooth and method |
| CN111027915A (en) * | 2019-12-17 | 2020-04-17 | 哈尔滨工业大学 | Intelligent logistics safety auxiliary system based on Bluetooth technology |
| KR20230150151A (en) * | 2022-04-21 | 2023-10-30 | (주)디에스네트웍스 | IoT-integrated WIFI AP wireless controller considering artificial intelligence-based wireless WiFi security |
| CN116193409A (en) * | 2022-12-23 | 2023-05-30 | 上海普瑞联网络科技有限公司 | Bluetooth signal receiving method, system, equipment and medium capable of adjusting signal range |
| CN116216075A (en) * | 2023-02-02 | 2023-06-06 | 山东零度供应链有限公司 | Tracking method based on packaging box, system and electronic equipment thereof |
| CN116980834A (en) * | 2023-09-25 | 2023-10-31 | 深圳市康凯铭科技有限公司 | Low-power consumption control method and system for Bluetooth beacon |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117273584A (en) | 2023-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102027338B (en) | Signal judgment method, signal judgment apparatus, and signal judgment system | |
| Chan et al. | On-road sensor configuration design for traffic flow prediction using fuzzy neural networks and taguchi method | |
| CN102404058A (en) | Electromagnetic wave discrimination device, electromagnetic wave discrimination method, and electromagnetic wave discrimination program | |
| CN107705359B (en) | Channel modeling method and device using three-dimensional visual reconstruction technology | |
| CN107241696A (en) | Multipath effect discriminating conduct and method for estimating distance based on channel condition information | |
| CN111551888A (en) | Electric energy meter metering data fault analysis method based on improved AdaBoost algorithm | |
| CN107295537A (en) | A kind of method and system for wireless sensor network reliability of testing and assessing | |
| CN118070680B (en) | Monitoring and early warning method for space modeling ranging of transformer substation | |
| CN116362359A (en) | User satisfaction prediction method, device, equipment and medium based on AI big data | |
| US11835379B2 (en) | Method and system for shipping container loading and unloading estimation | |
| CN116840764A (en) | Method and system for evaluating comprehensive error state of capacitive voltage transformer | |
| CN117273584B (en) | Cargo tracking method and device and computer readable storage medium | |
| CN105654174A (en) | System and method for prediction | |
| Legatiuk et al. | An abstract approach towards modeling intelligent structural systems | |
| CN116859317A (en) | Method and device for predicting metering error of capacitive voltage transformer | |
| CN103475527B (en) | Network management fault reliability analyzing system and method | |
| CN104568256A (en) | Engine torque detection system and method | |
| Hu | Application of artificial neuron network in analysis of railway delays | |
| CN111628837A (en) | Channel modeling method and device | |
| Waheed et al. | DeepChannel: Robust Multimodal Outdoor Channel Model Prediction in LTE Networks Using Deep Learning | |
| CN111160419B (en) | Deep learning-based electronic transformer data classification prediction method and device | |
| CN114157374A (en) | Method and device for predicting cell wireless signal strength and computer readable storage medium | |
| CN114925789B (en) | Asset equipment intelligent sensing method and device based on RFID and ensemble learning | |
| Wang et al. | Selection Method for Kernel Function in Nonparametric Extrapolation Based on Multicriteria Decision‐Making Technology | |
| Zuo | Automatic Analysis System for Enterprise Financial Information Based on Artificial Intelligence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |