Multifunctional goods tracking method based on RFID

Abstract

The invention relates to the technical field of radio frequency identification and intelligent logistics management, in particular to a multifunctional cargo tracking method based on RFID, which comprises the following steps: step 1, cargo binding and data synchronous acquisition: acquiring cargo state parameters through an RFID tag integrated with multiple sensors, and when a reader-writer transmits a query instruction, sending a query instruction to the reader-writer; synchronously acquiring a temperature value, a humidity value and a three-axis vibration acceleration, and packaging the temperature value, the humidity value and the three-axis vibration acceleration with a cargo identity code and a timestamp into a combined data unit; 2, dynamic environment interference suppression; step 3, mobile scene routing optimization: acquiring the moving speed of the carrier in real time, dynamically calculating an optimal query interval by combining with an environment interference coefficient, and transmitting a query instruction according to an interval period; and 4, multi-dimensional state fusion verification is carried out. Through innovative measures of multi-sensor integration, dynamic environment compensation, mobile scene optimization and the like, the accuracy, stability and intelligent level of cargo tracking are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wireless radio frequency identification and intelligent logistics management, and in particular to a multifunctional cargo tracking method based on RFID. BACKGROUND

[0002] In the logistics management of cross-border high-value cargos (such as biological and pharmaceutical preparations, precision electronic components, and cold-chain food), the cargo tracking technology based on RFID faces three technical bottlenecks: The existing RFID tags only support ID information storage, and key state parameters such as cargo temperature, humidity, and vibration need to be collected through independent sensors. Since each sensor has an independent clock source and data transmission channel, it leads to: Time desynchronization: there is a millisecond to second level deviation between the temperature collection time and the RFID reading time, which cannot determine the cause-and-effect relationship between the vibration peak value and the temperature mutation (for example, the cause-and-effect relationship determination of the temperature sudden rise after violent vibration in vaccine transportation is invalid); Spatial dislocation: the vibration sensor installation position is separated from the RFID tag, and when the cargo is displaced in the container, the vibration data is decoupled from the actual position of the cargo (for example, local collision of precision instruments in the inclined state cannot be accurately positioned).

[0003] In addition, in the typical cross-border multimodal transport environment: Metal interference: the metal shelves in the airport cargo station cause the radio frequency signal attenuation to exceed 20 dB, and the miss reading rate is as high as 35%; Liquid influence: the liquid cargo in the sea shipping container causes the dielectric constant to fluctuate, and the tag resonance frequency offset causes the read-write distance to be shortened by 50%; Dynamic miss reading: when the port AGV forklift moves at a speed of 2 m / s, the read success rate of the traditional fixed interval query scheme is less than 65%.

[0004] Therefore, there is an urgent need for a multifunctional cargo tracking method based on RFID to solve the above problems. SUMMARY

[0005] Based on the above purpose, the present application provides a multifunctional cargo tracking method based on RFID, comprising: Step 1: cargo binding and data synchronization collection: Through the RFID tag integrated with multiple sensors, the cargo state parameters are collected, and when the reader transmits a query instruction, the temperature value, humidity value, and three-axis vibration acceleration are synchronously obtained, and are packaged into a combined data unit together with the cargo identity code and time stamp; Step 2: dynamic environment interference suppression: The reader measures the signal strength value, and matches the environment interference feature library based on the current position coordinates: Multipath attenuation compensation in metal-intensive area: trigger repeated reading when signal strength value drops beyond dynamic baseline; Power self-adaptive adjustment for liquid goods: gradually increase transmission power until signal is stable; Step 3: Mobile scene route optimization: Real-time acquisition of vehicle moving speed, combined with dynamic calculation of environmental interference coefficient to calculate the optimal query interval, and transmit query instructions according to interval period; Step 4: Multi-dimensional state fusion verification: Perform the following on the combined data unit of continuous multiple readings: Extract temperature change trend, humidity fluctuation characteristics and acceleration modulus peak value; When temperature anomaly is accompanied by excessive vibration, generate physical anomaly identification, and when only signal strength is abnormal, generate environmental interference identification.

[0006] Preferably, the generation of the combined data unit in step 1 includes: The tag synchronously triggers the analog-to-digital conversion of the temperature sensor, humidity sensor and three-axis accelerometer within the same clock cycle of receiving the reader instruction; After the sensor data is sampled by 12-bit ADC, it is stored in the cache area, and the tag processor encapsulates it according to the preset frame format: The frame header contains the protocol version number and the check code; The EPC area writes 96-bit goods identity code, and the code structure conforms to the ISO / IEC18000-63 standard; The sensor data area is arranged in the order of temperature (2 bytes), humidity (2 bytes), and X / Y / Z axis acceleration (2 bytes each); The timestamp is taken from the millisecond timer generated by the internal crystal oscillator of the tag; After encapsulation, it is immediately backscattered to the reader, with a total delay of no more than 5 radio frequency periods.

[0007] Preferably, the dynamic baseline of step 2 is determined by the following method: Establish a sliding time window, and the window length is inversely adjusted according to the vehicle moving speed: the faster the speed, the shorter the window; Statistically count the signal strength values of valid readings within the window, and eliminate abnormal low values caused by shielding; Calculate the weighted moving average of the remaining data, and the weight is distributed according to the negative exponential law according to the reading time; When the current signal strength value is lower than the weighted moving average by a set proportion threshold, it is determined as effective drop.

[0008] Preferably, the construction of the environmental interference feature library includes: Reference tag arrays are deployed in typical areas of the warehouse, including: metal shelf area, liquid stacking area, open loading and unloading area; The signal strength attenuation characteristics at each position coordinate are continuously collected through the fixed reader network, and the attenuation characteristics include: The attenuation slope of signal strength with frequency change; The signal fluctuation variance caused by multipath interference; A mapping relationship table between position coordinates and interference characteristics is established, and the interference characteristics are automatically calibrated by the reference tag array every 24 hours.

[0009] Preferably, the calculation of the optimal query interval in step 3 satisfies: ; Where: is the moving speed of the carrier, which is obtained in real time by the inertial measurement unit carried by the reader; is the environmental interference coefficient, which is obtained by matching the current position in the environmental interference characteristic library; is the tag response characteristic factor, which is determined by analyzing the statistical distribution of tag response delay in historical reading records; The specific form of the function is dynamically generated by a machine learning model, which takes the maximum reading success rate as the target, and the combined features of speed, interference coefficient, and characteristic factor as input, and outputs the interval value.

[0010] Preferably, the determination of temperature abnormalities in step 4 includes: Establish a temperature change rate threshold : Take the first-order difference of the historical temperature data of the same type of goods under safe transportation conditions; Take the 99% quantile of the absolute value of the difference as ; When the temperature change rate of three consecutive readings exceeds ; and the acceleration module value exceeds the vibration threshold at the same time, the physical abnormality flag is triggered.

[0011] Preferably, the generation of the environmental interference flag also includes: When the signal strength value is abnormal, check the reading state of other tags at the same position: If the signal attenuation of tags exceeding the proportion threshold within the radius setting distance occurs at the same time, it is determined as regional environmental interference; If only the current tag is abnormal, judge whether it is a tag hardware failure combined with the historical interference record of the goods; The determination distance threshold of regional environmental interference is calculated by the path loss model:

[0012] wherein P is the transmit power, P is the receive sensitivity, P is the 1 meter reference distance loss, P is the environmental attenuation factor.

[0013] Preferably, it further comprises a data compression strategy, specifically including the following: When the goods are stationary, the combined data units of continuous reading are executed: The temperature data adopts a sliding window linear fitting, and the points with fitting residual less than the set sensitivity are deleted; The vibration data adopts a peak detection algorithm, and the points with acceleration modulus exceeding the local window average value are retained; The determination condition of the stationary state is: The sliding variance of the acceleration modulus is continuously lower than the activity threshold; The temperature change rate is continuously lower than the stability threshold.

[0014] Preferably, the synchronization control of the internal crystal oscillator of the tag comprises: When the reader / writer is queried for the first time, the tag records the time when the instruction arrives ; Each subsequent reading is based on The time stamp is generated by dividing the frequency of the crystal oscillator; After receiving a set number of reading instructions, the tag time stamp is compared with the reader / writer system time, and when the deviation exceeds the tolerance, the crystal oscillator frequency is adjusted.

[0015] Preferably, the training process of the machine learning model comprises: Collecting a training data set under typical scenarios: Input features: moving speed , interference coefficient , label response delay distribution matrix; Label data: status identification of whether the actual reading is successful or not; A random forest algorithm is used to construct a regression model: The number of decision trees is adaptively determined according to the feature dimension; The node splitting criterion is to maximize the information gain rate; The model is updated online with new data every set time period.

[0016] The beneficial effects of the present application are: 1、The present application synchronously triggers temperature, humidity and vibration data collection in the same clock cycle when the tag receives the query instruction by combining the synchronous collection of data units, and encapsulates the data after sampling by 12-bit ADC. This synchronization mechanism eliminates the time deviation of temperature and vibration data, ensures that the sequence of vibration changes and temperature changes can be accurately determined, and avoids the failure of causal judgment of vibration and temperature in vaccine transportation and other scenarios.

[0017] 2、The present application realizes the unification of state data and goods location by integrating RFID tags with multiple sensors and combining dynamic monitoring of physical state and location. Through multi-dimensional state fusion verification, temperature, humidity and vibration parameters are combined with spatial information of goods for comprehensive analysis, which can accurately judge the actual state of goods and avoid data misjudgment caused by spatial misplacement.

[0018] 3、The present application uses a dynamic environmental interference suppression method to start multi-path attenuation compensation in metal-intensive areas, and triggers repeated reading when the signal strength value drops below the dynamic reference. This intelligent dynamic compensation method effectively reduces the missed reading phenomenon caused by metal interference, significantly improves the stability and reliability of data reading.

[0019] 4、The present application uses power self-adaptive adjustment to gradually increase the transmission power of the reader until the signal is stable when liquid goods are detected. This adaptive strategy effectively overcomes the influence of liquid goods on RFID signals and improves the RFID reading success rate in liquid goods transportation.

[0020] 5、The present application uses mobile scenario routing optimization to obtain the moving speed of the carrier in real time, and dynamically calculates the optimal query interval combined with the environmental interference coefficient. This scheme allows the query interval of the RFID tag to be adjusted in real time according to the moving speed of the carrier and the interference environment, effectively improving the reading success rate in fast-moving environments, especially in dynamic environments such as ports, significantly reducing the missed reading phenomenon. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0022] Fig. 1 The step flow chart of the method of the present application; Fig. 2 The step flow chart of the method of the present application for constructing the environmental interference feature library; Fig. 3 The step flow chart of the data compression strategy of the method of the present application. DETAILED DESCRIPTION

[0023] The application will be described in detail below with reference to the drawings and specific embodiments. It should be noted here that in order to make the embodiments more detailed, the following embodiments are the best, preferred embodiments, and other alternative ways can also be implemented by those skilled in the art for some known technologies; and the drawings are only used to more specifically describe the embodiments, and are not intended to specifically limit the application.

[0024] Please refer to Figs. 1-3 The embodiment of the application provides a multifunctional goods tracking method based on RFID. In step 1, the state of goods is monitored in real time by an RFID tag integrated with multiple sensors. When the RFID reader transmits a query instruction, the integrated sensors simultaneously collect temperature values, humidity values, and three-axis vibration acceleration data. These sensor data are packaged into a combined data unit together with the identity code of the goods and a time stamp, and are transmitted to the RFID reader. This synchronous data collection mechanism solves the problem of time asynchronization, ensures that temperature and vibration data are collected at the same time, thereby avoiding the time difference problem between data, especially in the scene of analyzing the correlation between temperature change and vibration (such as the causal relationship between violent vibration and temperature sudden rise in vaccine transportation), which can provide accurate data support.

[0025] In step 2, a scheme of dynamic environmental interference suppression is proposed for possible interference in a complex environment (such as the influence of metal objects and liquid substances on RFID signals). First, the RFID reader measures the signal strength value in real time, and matches the pre-established environmental interference feature library according to the current position coordinates: When the signal strength value drops more than the dynamic reference, the system will automatically start the compensation mechanism to perform repeated reading, ensuring accurate transmission of information, and significantly reducing the phenomenon of missed reading caused by metal interference.

[0026] For liquid goods, the resonance frequency of the RFID tag may be affected by the liquid, resulting in unstable signal transmission. The application gradually increases the transmission power of the RFID reader until the signal is stable, thereby overcoming the influence of liquid on the signal and improving the reading success rate.

[0027] Step 3 realizes query optimization in a mobile scenario. By obtaining the moving speed of the carrier in real time, the system dynamically calculates the optimal query interval in combination with the environmental interference coefficient. According to the speed of the carrier and the interference condition of the environment, the query period is adjusted to ensure stable reading of the RFID tag in a high-speed mobile environment (such as a port forklift). This optimization strategy solves the problem of dynamic missed reading, and in the scene of fast movement, it can greatly improve the reading success rate and reduce data loss.

[0028] Step 4: Multi-dimensional state fusion verification is performed on the combined data units of consecutive multiple readings. The specific implementation is as follows: Extract the temperature change trend, humidity fluctuation characteristics, and peak value of the three-axis acceleration.

[0029] When an abnormal temperature change is detected and accompanied by excessive vibration, a physical abnormality identifier is generated, indicating that the goods may be damaged or have other abnormal conditions.

[0030] When only signal strength anomalies occur, the system generates an environmental interference identifier, indicating that the current reading data may be affected by external environments (such as metal or liquid interference).

[0031] Based on RFID technology, the present application effectively improves the accuracy, stability, and intelligent level of goods tracking through multi-sensor integration, dynamic environment compensation, mobile scene optimization, and other innovative measures, solving various technical challenges faced in cross-border logistics, and having significant application prospects and practical value.

[0032] In one possible implementation, within the same clock cycle of receiving the RFID reader instruction, the RFID tag triggers the analog-to-digital conversion (ADC) of the temperature sensor, humidity sensor, and three-axis accelerometer for data collection. The sensors inside the tag will simultaneously start sampling environmental parameters such as temperature, humidity, and vibration, and convert the analog signals into digital signals. The synchronous collection mechanism ensures that all sensor data is obtained at the same time, effectively avoiding data inconsistency due to time differences, especially in scenarios where multiple physical quantity changes need to be monitored, ensuring data integrity and accuracy.

[0033] The collected sensor data is sampled by a 12-bit ADC (analog-to-digital converter), and the data is stored in the cache area after conversion for processing. Then, the processor of the RFID tag encapsulates the data according to the preset frame format. The frame format includes: Frame header: contains the protocol version number and check code to ensure the reliability of data transmission and prevent information loss or misinterpretation due to data packet errors.

[0034] EPC area: writes a 96-bit goods identity code, which conforms to the ISO / IEC18000-63 standard, which is a standard protocol in the RFID field, ensuring the international universality and interoperability of the tag, and compatibility with readers from different manufacturers.

[0035] Sensor data area: arranged in order of temperature (2 bytes), humidity (2 bytes), and X / Y / Z-axis acceleration (2 bytes each), ensuring standardized data structure for subsequent analysis and application.

[0036] Timestamp: The timestamp comes from a millisecond-level timer generated by a crystal oscillator inside the tag, accurate to the millisecond level, ensuring the timeliness of the data, and can trace the status of the goods at each time.

[0037] After data encapsulation, the tag immediately sends the encapsulated data back to the RFID reader through backscattering. In order to improve the data transmission efficiency and response speed, the total delay of the entire data encapsulation and transmission does not exceed 5 radio frequency cycles, ensuring the high responsiveness of the system in real-time tracking of the status of goods, especially suitable for fast-moving logistics environment.

[0038] The present application optimizes the sensor data acquisition, processing and encapsulation process of RFID tag, enhances the synchronization, transmission efficiency and response speed of data, and provides more accurate, real-time and reliable technical support for goods tracking.

[0039] In a possible implementation, in an RFID system, in order to effectively monitor the change of signal strength when the carrier (such as a transport vehicle, a storage device, etc.) moves, first, the length of the sliding time window needs to be dynamically adjusted according to the moving speed of the carrier. Specifically, when the moving speed of the carrier is fast, the length of the time window will be shortened accordingly, so as to capture the signal change in a shorter time; on the contrary, when the speed of the carrier is slow, the time window will be extended. This adjustment mechanism ensures that the data acquisition frequency is optimized according to the speed characteristics of different carriers, thereby improving the detection accuracy of signal strength.

[0040] In each sliding time window, the system will perform statistics on all valid read signal strength values. Signal strength is usually affected by environmental factors (such as obstructions, reflections, etc.), so it is necessary to eliminate those abnormally low values caused by obstructions. By identifying and eliminating these abnormal data, the reliability of the signal strength value can be improved, ensuring that the system accurately judges the status and position of the goods.

[0041] After eliminating abnormal values, the system will calculate the weighted moving average of the remaining data, and assign different weights to each signal strength value according to the proximity of the reading time. The weight distribution follows the negative exponential law, that is, the closer the time point, the greater the weight, and the farther the time point, the smaller the weight. Such weight distribution can ensure that the system pays more attention to the signal strength in the current time period, avoiding the interference of historical data affecting the accuracy of real-time judgment.

[0042] Through the calculation of the weighted moving average, the system obtains a dynamic reference value. When the current signal strength value is lower than the weighted average value and the drop amplitude reaches the set proportion threshold, the system will determine that the signal has effectively decreased. This determination mechanism can help the system to identify the status change of the goods in time, such as the goods being blocked or lost, so as to trigger relevant alarm or tracking measures.

[0043] In this way, the application can more accurately and in real-time track the status of goods, reduce errors and improve response speed, thereby playing an important role in various goods transportation and warehouse management applications.

[0044] In one possible implementation, in order to effectively collect the interference characteristics in the warehouse environment, it is first necessary to deploy a reference tag array in typical areas of the warehouse. These areas are classified as: Metal shelf area: Due to the strong reflection or absorption effect of metal surfaces, it can have a significant impact on RFID signals.

[0045] Liquid stacking area: Liquid materials can change the propagation characteristics of signals, especially in areas with more liquid stacking, the propagation of RFID signals can be severely attenuated.

[0046] Open loading and unloading area: In an open environment, weather factors and external signal interference can affect the stability of RFID signals.

[0047] The reference tag array arranged in these areas continuously collects signals through a fixed reader network. The position and number of these tag arrays are carefully arranged to ensure coverage of typical environmental interference scenarios.

[0048] In these areas, the reader continuously monitors the RFID signal strength attenuation characteristics at each location, mainly including the following two key parameters: Signal strength attenuation slope with frequency: The attenuation characteristics of RFID signals vary with frequency. By measuring the signal strength attenuation at different frequencies, the influence of different frequencies on signal propagation can be better understood, providing data support for subsequent signal analysis.

[0049] Signal fluctuation variance caused by multipath interference: Multipath effect refers to the signal reaching the receiver through multiple paths, which can cause signal interference and fluctuation. By measuring the signal fluctuation variance at different locations, the strength of the multipath effect in the environment can be analyzed, and the impact on signal stability can be evaluated.

[0050] By recording the signal strength attenuation characteristics collected in each typical area in detail, the system will establish a mapping table of location coordinates and interference characteristics. This table records the signal attenuation characteristics of each specific location (such as the metal shelf area, liquid stacking area, etc.) in detail and associates it with the actual coordinates. Through this mapping relationship, the system can more accurately identify and predict the attenuation pattern of RFID signals in a specific environment.

[0051] To ensure the real-time accuracy and adaptability of the environmental interference feature library, the reference tag array will be automatically calibrated every 24 hours. This process updates the decay features by continuously collecting new rounds of signal data and adjusts the interference feature library according to the actual changes in the environment, ensuring that the system can always adapt to the dynamic changes of the warehouse environment.

[0052] By establishing an accurate interference feature library in the warehouse, the accuracy and reliability of the RFID-based cargo tracking system can be significantly improved, especially in the face of complex environments and variable interference, while still maintaining high efficiency and stable tracking capabilities.

[0053] In one possible implementation, in this embodiment, the moving speed of the carrier is a key input variable. In order to accurately obtain the real-time speed of the carrier, the inertial measurement unit (IMU) carried by the reader is used to measure the speed of the carrier in real time. IMU usually includes accelerometer and gyroscope, through real-time acquisition and processing of these sensor data, the instantaneous speed of the carrier can be accurately obtained. This speed data will directly affect the calculation of the query interval, because the faster the carrier speed, the reading frequency usually needs to be increased accordingly to ensure timely and effective capture of cargo information.

[0054] Various interference factors in the environment will significantly affect the strength of the RFID signal and the success rate of reading. In this method, the environmental interference coefficient is obtained by matching the current position of the carrier in the environmental interference feature library. This feature library records the environmental interference features (such as metal reflection, liquid interference, etc.) of different positions. By obtaining the current position of the carrier in real time, the system can find the corresponding environmental interference coefficient in the feature library and use it for the calculation of the query interval. This interference coefficient can help the system dynamically adjust the query frequency according to the different changes of the environment, thereby optimizing the reading effect of the RFID signal.

[0055] The response characteristics of the tag will affect the reading efficiency of the RFID system. In this method, the response characteristic factor of the tag is determined by analyzing the statistical distribution of the tag response delay in the historical reading records. The tag response delay refers to the time difference between the query request sent by the reader and the tag response. After statistical analysis of these delay data, a characteristic factor can be obtained, which reflects the response speed of the tag. This characteristic factor will be used together with other factors to calculate the optimal query interval, thereby improving the response efficiency of the RFID system.

[0056] The specific calculation of the optimal query interval is dynamically generated by a machine learning model. The inputs to this model include the moving speed of the vehicle, environmental interference coefficients, and tag response characteristic factors. The combined characteristics of these factors will serve as inputs to the model. Through learning a large amount of historical data, the machine learning model can generate an optimal query interval function that maximizes the reading success rate. The goal of this model is to dynamically adjust the query interval based on the current environment and system state, thereby improving the reading efficiency and accuracy of RFID signals.

[0057] Through dynamic calculation and optimization of various factors, the performance of the RFID multifunctional cargo tracking system is effectively improved, ensuring efficient operation in different environments and states.

[0058] In one possible implementation, the system can collect real-time speed information of the vehicle through the inertial measurement unit (IMU) mounted on the RFID reader. The IMU integrates an accelerometer and a gyroscope, which can accurately measure the motion state of the vehicle and calculate its instantaneous speed. This information is crucial for calculating the query interval, as in high-speed moving situations, the RFID reader needs to query more frequently to ensure accurate reading of target tags, while in low-speed situations, the query interval can be appropriately increased to save energy.

[0059] The propagation of RFID signals can be affected by environmental factors, especially in complex environments. To quantify these effects, the system obtains environmental interference coefficients through an environmental interference feature library. This feature library records the interference conditions of different environmental areas (such as metal shelves, liquid storage areas, etc.). When the vehicle moves, the corresponding interference coefficient is found and matched in the feature library according to the current location. This interference coefficient can reflect the effects of attenuation, reflection, and multipath effects that the RFID signal may encounter at this location, thereby helping the system dynamically adjust the query strategy.

[0060] The system determines the response characteristics of each tag by analyzing historical reading records and calculating the response delay of the tag. The statistical distribution of the tag response delay can reveal the reaction speed of the tag under different conditions, which is crucial for adjusting the query interval. For example, near a tag with a slower response, the system may need to increase the query frequency to ensure that the tag's response can be read.

[0061] Based on the speed of the carrier, environmental interference coefficient, and tag response characteristics, the system generates an optimal query interval function using a machine learning model. This model is trained with historical data to learn how to dynamically adjust the query frequency in different environments. The goal of this model is to maximize the reading success rate, i.e., to ensure that the RFID system can successfully read as much tag information as possible while maintaining a low query frequency. The query interval output by the machine learning model can be adjusted based on real-time environmental data, thereby optimizing the performance of the RFID tracking system.

[0062] By comprehensively considering factors such as carrier speed, environmental interference, and tag response characteristics, and combining dynamic adjustment of the machine learning model, the RFID multifunctional cargo tracking system can maintain high efficiency and accuracy in various complex environments, significantly improving the performance and reliability of the system.

[0063] In one possible implementation, when the signal strength value received by the RFID reader is abnormal, the system first identifies possible interference problems by detecting the signal strength of the tag. Signal strength abnormalities can manifest as a signal strength received far below the normal value, possibly due to environmental interference or tag hardware problems.

[0064] When a certain tag signal anomaly is detected, the system further checks the reading status of other tags in the same location. Specifically, the system evaluates the signal quality of other tags in the same area. If the signals of these tags also simultaneously decay, and the decay rate exceeds a set threshold, it can be determined that the area is experiencing environmental interference. This step uses the signal conditions of a group of tags to help determine whether the interference is caused by environmental factors or a single tag hardware problem.

[0065] If only the signal of the current tag is abnormal, while the signals of other tags in the same area do not change significantly, it may be a hardware failure of the tag itself. At this time, the system will combine the historical interference records of the tag to further determine whether the problem is caused by a hardware failure. If the tag has a history of multiple signal abnormalities or interference situations, it is more likely to be determined as a hardware failure.

[0066] When the system determines that the area is experiencing environmental interference, the determination distance threshold of the area environmental interference is calculated by a path loss model. The path loss model takes into account multiple factors, such as transmission power, reception sensitivity, loss of reference distance, and environmental attenuation factor, to calculate the loss of the signal during transmission. The specific calculation formula is as follows: ; where is the transmission power, is the reception sensitivity, is the loss of 1 meter reference distance, The environmental attenuation factor is used to calculate the loss of the signal during transmission.

[0067] The model calculates the loss of the signal during transmission by considering the environmental attenuation factor, thereby determining a decision distance threshold. If the signal of the tag attenuates within the distance range and exceeds the set proportion threshold, it is determined that there is environmental interference in the area.

[0068] Through multi-dimensional judgment and the application of the path loss model, the interference recognition and processing capacity of the RFID cargo tracking system is effectively improved, and the stability and reliability of signal transmission can be ensured in various environments.

[0069] In one possible implementation, first, the system needs to determine whether the cargo is in a stationary state. The determination of the stationary state is based on the following two conditions: By calculating the sliding variance of the acceleration module value in a continuous period of time, if the variance is continuously lower than the preset activity threshold, it means that the acceleration change of the cargo is very small, indicating that the cargo is in a stationary state.

[0070] The temperature change rate refers to the change amplitude of the temperature per unit time. When the temperature change rate continuously falls below the set stability threshold, it indicates that the temperature change of the environment or the cargo is relatively stable, which can also be used to assist in determining that the cargo is in a stationary state.

[0071] The two determination conditions work together to accurately determine whether the cargo is in a stationary state, thereby avoiding collecting too much unnecessary data when the cargo is stationary.

[0072] When the cargo is stationary, the system will perform data compression strategies on the continuously read combined data units, specifically including: Temperature data sliding window linear fitting: the system performs sliding window linear fitting processing on the temperature data. Within the window, the system uses linear fitting method to estimate the data trend, and the fitting residual is the difference between the data points and the fitting curve.

[0073] Delete points with residual less than the set sensitivity: if the residual of a data point is less than the set sensitivity threshold, it means that the deviation of the data point from the fitting curve is small, and it does not have significant information value. Therefore, the system will delete these unnecessary points to reduce redundant data and achieve data compression.

[0074] Under the stationary state, the vibration data is usually relatively stable, but occasionally there will be sudden acceleration changes. The system uses a peak detection algorithm to identify and retain points whose acceleration module value exceeds the local window average. In this way, only abnormal vibration data is retained, and most of the normally fluctuating data is removed, further reducing the data volume.

[0075] By combining temperature data linear fitting, vibration data peak detection and accurate determination of the stationary state, data can be effectively compressed, system performance can be improved, and the accuracy and efficiency of data transmission can be guaranteed.

[0076] In one possible implementation, in an RFID system, when a reader first queries a tag, the tag records the time when the received instruction arrives. This record of the time is very important for subsequent time synchronization, as it provides a reference point for subsequent time stamp generation. After receiving the first query instruction, the tag matches the current system time (or the time of the internal clock of the tag) with the received instruction and records the time when the instruction arrives. In this way, the tag's time stamp can be accumulated from this reference time.

[0077] In each subsequent reading process, the tag takes the time stamp of the first query as a reference and generates a time stamp according to the internal crystal oscillator frequency division. The crystal oscillator frequency refers to the frequency of the internal clock signal of the tag, which determines the accuracy and speed of the tag generating time stamps. The tag's crystal oscillator generates time stamps through frequency division mechanism, ensuring that each read data can be associated with a unique time stamp, facilitating accurate recording and querying of goods tracking data.

[0078] After the tag receives a set number of read instructions, the tag compares its generated time stamp with the reader system time. This comparison operation is to ensure that the internal clock of the tag and the system time of the reader can be synchronized. Because the RFID system needs to ensure the time coordination between the tag and the reader, in order to accurately record the position and state of each goods.

[0079] If the deviation between the tag's time stamp and the reader system time exceeds the preset tolerance, the tag will start the crystal oscillator frequency fine-tuning mechanism. The purpose of crystal oscillator frequency fine-tuning is to adjust the frequency of the tag's internal crystal oscillator, so that its clock is more accurately synchronized with the reader system time. During the fine-tuning process, the tag will adjust the frequency of the crystal oscillator according to the size of the deviation, to ensure that future time stamps are more accurate and avoid the gradual accumulation of time deviation affecting tracking accuracy.

[0080] Through accurate time stamp generation, time comparison and crystal oscillator frequency fine-tuning, the time synchronization between the RFID tag and the reader is ensured, the reliability, accuracy and stability of the system are improved, and it has important application value.

[0081] In one possible implementation, first, the system needs to collect training data sets in typical scenarios, which will be used to train the machine learning model. The collection of training data includes the following two parts: Input features: where the moving speed is a characteristic reflecting how fast or slow the goods or tags move in the physical environment. By tracking the moving speed of the goods, the system can identify the moving patterns of the goods at different locations, thus better predicting the success probability of tag reading.

[0082] The interference coefficient is used to represent the degree of signal interference in the environment, which may come from walls, metal objects, other RFID devices, etc. By quantifying the interference coefficient, the system can consider the impact of environmental factors on the effectiveness of RFID tag reading.

[0083] The tag response delay distribution matrix describes the distribution of tag response time, mainly used to measure the response delay of tags to the requests of the reader. Different types of tags and different environments will have different performances of response delay, which will affect the success rate of reading.

[0084] Tag data: The actual reading success or failure status identifier is the target variable of the model, used to mark whether the tag is successfully read by the reader under given conditions. This identifier is obtained by recording the results of each reading attempt, and is the key data for model training.

[0085] After collecting the data set, a random forest algorithm is used to build a regression model to predict the probability of tag reading success. Random forest is an ensemble learning method that improves the accuracy and robustness of prediction by building multiple decision trees. The specific implementation includes: According to the dimension of input features, the system will adaptively determine the number of decision trees in the random forest. A more complex feature space may require more decision trees to effectively model, while a simple feature space may only require a small number of decision trees.

[0086] When building each decision tree, the information gain ratio is chosen as the standard for node splitting. The information gain ratio measures the degree of reduction in uncertainty of the data set under a particular feature, and maximizing the information gain ratio for splitting nodes helps improve the prediction accuracy of the model.

[0087] Over time and with changes in the environment, the conditions of the RFID system may change, so the accuracy of the model may be affected. In order to ensure the real-time and effectiveness of the model, the system will update the model online according to the newly added training data: The system will regularly (such as every hour or every day) update the model with newly added tag reading data. This allows the model to be adjusted in time to adapt to new environmental conditions or changing parameters, so that the RFID system can maintain high accuracy in the long-term running process.

[0088] By combining typical scenario data, random forest regression model, and online updating mechanism, the accuracy, adaptability, and stability of the RFID system are significantly improved, effectively improving the performance and reliability of cargo tracking.

[0089] The following is described in detail through examples: The present application relates to a tracking technology for intelligent logistics systems, aiming to improve the accuracy and real-time performance of cargo tracking. Traditional logistics tracking systems usually rely on barcode scanning or GPS positioning, but these systems have problems of delay and low accuracy in complex environments. The present application uses RFID technology combined with the random forest algorithm in machine learning to achieve more efficient and accurate real-time cargo tracking.

[0090] In large logistics warehouses, the storage and warehouse management of goods often rely on manual operation or traditional barcode technology, resulting in high labor costs and high error rates. Through the present application, RFID technology and random forest algorithm are used for automatic cargo tracking and management.

[0091] Specifically, each cargo is equipped with an RFID tag when entering the warehouse, and the tag stores the unique identification information of the cargo. Multiple entrances and exits of the warehouse are equipped with RFID readers, which periodically scan the cargo tags and collect the following data: Tag ID (unique identifier of the cargo); location data (reader location, located through warehouse map); environmental data (temperature, humidity) To ensure the accuracy of the data, the following sensor parameters are set: Temperature sensor accuracy: ±0.5°C; humidity sensor accuracy: ±3%; RFID reader reading speed: up to 50 tags per second.

[0092] The raw data obtained from the RFID system will be processed through data cleaning and standardization: All cargo tag IDs are converted to a uniform format (e.g., string length fixed at 16 bits).

[0093] The environmental data is processed through data normalization to ensure that the temperature and humidity sensor data in different warehouses can be compared uniformly.

[0094] The collected data is modeled using the random forest algorithm. The specific steps are as follows: Feature selection: select the following features as input: ID of RFID tag (associated with location through historical records); environmental temperature and humidity (affecting cargo storage conditions); RFID tag reading timestamp (for time synchronization); data set division: divide the data set into 80% training set and 20% test set.

[0095] Random forest algorithm parameters: Number of trees: 100 trees; Maximum depth: 10; Maximum number of features per tree: sqrt(total number of features); Minimum number of samples per split: 10.

[0096] The model performance is evaluated using accuracy, recall, and F1 score to ensure that the model can effectively predict the storage location of goods in different environments.

[0097] Once the model is trained, the system can start processing data in real time. When goods enter the warehouse, the RFID tag is scanned, and the system immediately predicts the storage location of the goods, the estimated delivery time, etc. through the trained model.

[0098] Each time the goods are scanned, the RFID reader will update the location and environmental data in real time and pass it into the system.

[0099] The system outputs the latest location of the goods, temperature and humidity status, and determines whether the goods meet the storage conditions based on the random forest model.

[0100] In the same warehouse environment, a comparative experiment was conducted to compare the intelligent logistics tracking system of the invention with the traditional system based on barcode scanning.

[0101] Traditional method: relying on barcode scanning and manual input, with an error range of ±10 meters and a goods scanning time of 5 seconds per item.

[0102] Invention method: based on RFID and random forest algorithm, with a reading accuracy of ±2 meters and a goods scanning time of 2 seconds per item.

[0103] The experimental results are as follows:

[0104] Through comparison, it can be seen that the invention has significant advantages in goods reading accuracy, scanning speed and system efficiency, and is particularly suitable for large-scale warehouse management, improving the intelligent level of logistics management.

[0105] By combining RFID technology with random forest algorithm, high-precision tracking of goods location is achieved. The system can collect and process data in real time, ensuring the timeliness of the goods tracking information. Compared with traditional manual scanning, the invention provides a completely automated goods tracking system, significantly improving work efficiency and reducing human error rate.

[0106] This embodiment demonstrates how to apply RFID technology and machine learning algorithms for cargo tracking in a smart warehouse. By comparing with traditional methods, it is proved that the present application can greatly improve the accuracy and efficiency of cargo tracking, and through detailed parameter setting, model training and evaluation, the stability and reliability of the system are ensured.

[0107] The present application encompasses any substitutions, modifications, equivalent methods and solutions made to the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details to those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0108] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can also be made, which should be considered as the protection scope of the present application.

Claims

1. A multifunctional cargo tracking method based on RFID, characterized in that, include: Step 1: Cargo binding and data synchronization collection: The RFID tag, which integrates multiple sensors, collects cargo status parameters. When the reader sends a query command, it simultaneously acquires temperature, humidity and triaxial vibration acceleration, and packages them together with cargo identification code and timestamp into a combined data unit. Step 2: Suppression of dynamic environmental interference: The reader measures the signal strength value and matches it against an environmental interference feature library based on the current location coordinates. Multipath attenuation compensation is activated in densely metal areas: repeated readings are triggered when the signal strength value drops below the dynamic reference. For liquid cargo, the starting power is adaptively adjusted: the transmission power is gradually increased until the signal stabilizes; Step 3: Routing optimization in mobile scenarios: Real-time vehicle speed is acquired, and the optimal query interval is dynamically calculated based on the environmental interference coefficient. Query commands are then sent at intervals. Step 4: Multi-dimensional state fusion verification: Perform the following on a combined data unit that has been read multiple times consecutively: Extract temperature change trends, humidity fluctuation characteristics, and peak acceleration modulus values; When abnormal temperature is accompanied by excessive vibration, a physical anomaly indicator is generated; when only signal strength is abnormal, an environmental interference indicator is generated.

2. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The generation of combined data units in step 1 includes: Within the same clock cycle of receiving the reader's command, the tag synchronously triggers the analog-to-digital conversion of the temperature sensor, humidity sensor, and triaxial accelerometer. Sensor data is sampled by a 12-bit ADC and stored in a buffer, then packaged by the tag processor according to a preset frame format. The frame header contains the protocol version number and checksum; The EPC area is written with a 96-bit cargo identification code, the coding structure of which conforms to the ISO / IEC 18000-63 standard; The sensor data area is arranged in the following order: temperature (2 bytes), humidity (2 bytes), and X / Y / Z axis acceleration (2 bytes each); The timestamp is taken from a millisecond-level timer generated by frequency division of the crystal oscillator inside the tag; After packaging, the signal is immediately backscattered to the reader, with a total delay of no more than 5 RF cycles.

3. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The dynamic baseline in step 2 is determined in the following way: Establish a sliding time window, with the window length adjusted inversely based on the vehicle's movement speed: the faster the speed, the shorter the window. Within the window, count the valid signal strength values ​​and remove abnormally low values ​​caused by obstruction; Calculate the weighted moving average of the remaining data, with the weights distributed according to the negative exponential law based on the time of data access. When the current signal strength value is lower than the weighted moving average by a set percentage threshold, it is considered a valid decrease.

4. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The construction of the environmental interference feature library includes: Deploy reference tag arrays in typical warehouse areas, including: metal shelving area, liquid storage area, and open loading and unloading area; Signal strength attenuation characteristics at various coordinates are continuously collected using a fixed reader network. These attenuation characteristics include: The attenuation slope of signal strength as a function of frequency; Signal fluctuation variance caused by multipath interference; A mapping table between location coordinates and interference features is established, and the interference features are automatically calibrated every 24 hours using a reference tag array.

5. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The calculation of the optimal query interval in step 3 satisfies: ； in: The vehicle's moving speed is obtained in real time through the inertial measurement unit mounted on the reader; This is the environmental interference coefficient, obtained by matching the current location in the environmental interference feature database. The tag response characteristic factor is determined by analyzing the statistical distribution of tag response latency in historical read records; function The specific form is dynamically generated by a machine learning model. This model aims to maximize the success rate of reading, and its input is a combination of speed, interference coefficient, and characteristic factors. The output is the interval value.

6. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The determination of temperature anomalies in step 4 includes: Establish a threshold for the rate of temperature change : Calculate the first difference of historical temperature data for similar goods under safe transportation conditions; Take the 99th percentile of the absolute value of the difference as ; When the rate of temperature change exceeds 3 consecutive readings ; Furthermore, when the acceleration modulus exceeds the vibration threshold, a physical anomaly indicator is triggered.

7. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, The generation of environmental disturbance labels also includes: When the signal strength value is abnormal, check the reading status of other tags at the same location: If tags within a set radius distance that exceed a proportional threshold all experience signal attenuation, it is determined to be regional environmental interference. If only the current tag is abnormal, determine whether it is a tag hardware failure by combining the historical interference records of the goods; The distance threshold for determining regional environmental interference is calculated using the path loss model: ； in For transmission power, For receiving sensitivity, For a reference distance of 1 meter, the loss is calculated. It is an environmental degradation factor.

8. The multifunctional cargo tracking method based on RFID according to claim 1, characterized in that, It also includes data compression strategies, specifically the following: When the goods are stationary, the following is executed on the continuously read combined data units: Temperature data were fitted using a sliding window linear fit, and points with fitting residuals less than the set sensitivity were removed. The vibration data uses a peak detection algorithm to retain points where the acceleration modulus exceeds the local window average. The condition for determining a static state is: The sliding variance of the acceleration modulus is continuously below the activity threshold; The rate of temperature change remains below the stability threshold.

9. A multifunctional cargo tracking method based on RFID according to claim 2, characterized in that, Synchronization control of the internal crystal oscillator of the tag includes: When the reader makes its first query, the tag records the time of arrival. ; Subsequent reads are as follows Based on the crystal oscillator frequency, timestamps are generated by dividing the crystal oscillator frequency. After receiving a set number of read commands, the tag timestamp is compared with the reader's system time. When the deviation exceeds the tolerance, the crystal oscillator frequency is fine-tuned.

10. A multifunctional cargo tracking method based on RFID according to claim 5, characterized in that, The training process for a machine learning model includes: Collect training datasets in typical scenarios: Input feature: Movement speed Interference coefficient Tag response delay distribution moment; Tag data: A status indicator indicating whether the read was successful or not; A regression model is constructed using the random forest algorithm: The number of decision trees is adaptively determined based on the feature dimensions; The criterion for node splitting is maximizing the information gain ratio; The model is updated online with new data at set time intervals.

Images