Warehousing digital twin system integrated with UWB-RFID technology

Abstract

The invention discloses a storage digital twinning system integrated with a UWB-RFID technology. The storage digital twinning system comprises an entity storage, an edge server and a storage digital twinning system. The entity storage comprises a fixed UWB base station, a UWB tag, a fixed RFID base station, a passive tag and a movable RFID base station. And the edge server collects all base station equipment data of the entity warehouse, processes and calculates cargo information according to a built-in algorithm, and performs data interaction with an upper-layer warehouse digital twin system at the same time, so that the operation condition of the entity warehouse is remotely monitored. According to the system, through the ultra-wideband technology, the radio frequency identification technology and the digital twinning technology, the storage position of stored goods can be determined, entity storage data can be visualized, and the problems of low positioning precision and low informatization degree in the existing storage management technology are solved.

Description

Technical Field

[0001] The present invention relates to a warehouse digital management system, and in particular to a warehouse digital twin system integrating UWB-RFID technology. Background Art

[0002] In modern logistics and supply chain management, warehouse management is a key link. With the rapid development of e-commerce and the increasing global logistics demand, the traditional warehouse management model can no longer meet the needs of efficient, accurate and real-time management.

[0003] RFID technology can quickly identify and track items by reading tag information. However, the positioning accuracy of RFID is limited, usually only within a meter range, which is difficult to meet the needs of high-precision positioning. In addition, RFID signals are easily interfered by metals, liquids and other materials, resulting in unstable performance in complex environments and affecting recognition accuracy. UWB technology can provide centimeter-level high-precision positioning, which is suitable for real-time tracking of movable targets. However, UWB deployment costs are high, especially in large storage environments, where a large number of base stations may be required, increasing deployment and maintenance costs. The key advantage of digital twins lies in their data integration and analysis capabilities, but digital twins rely on a large amount of real-time data, including location, temperature and humidity, inventory status, etc., which usually come from a variety of different sensors and devices. The data collection, cleaning, transmission and fusion process is very complicated, and any slight deviation or data loss may affect the accuracy of the digital twin model. Summary of the invention

[0004] Purpose of the invention: In view of the above-mentioned existing technologies, a warehouse digital twin system integrating UWB-RFID technology is proposed to solve the problems of low positioning accuracy and low degree of informatization in existing warehouse management technologies.

[0005] Technical solution: A warehouse digital twin system integrating UWB-RFID technology, including physical warehouse, edge server, and warehouse digital twin system;

[0006] The physical warehouse includes 4 fixed UWB base stations, multiple fixed RFID base stations, multiple shelves, a mobile RFID base station with a UWB tag attached, multiple passive tags and multiple target goods, and a spatial positioning model of the physical warehouse is established;

[0007] The edge server is used to collect data from base station equipment in the physical warehouse in real time, transmit the processed data to the positioning calculation center, and infer the coordinate information of the tag to be tested in combination with the warehouse static data stored in the edge database, and integrate the cargo information and tag information data and send them to the warehouse digital twin;

[0008] The warehouse digital twin system includes a twin data system and a virtual system; the twin data system is used to complete the interactive mapping of physical warehouse data and virtual warehouse models; the virtual system is used to display the three-dimensional model and operation data of the physical warehouse in a visual way.

[0009] Furthermore, in the physical warehouse, fixed UWB base stations are installed at the four corners of the warehouse ceiling and can communicate with UWB tags; fixed RFID base stations are installed on the side of each shelf, and an antenna is installed at the end point of each shelf; the mobile RFID base station is a handheld scanner or a vehicle-mounted RFID device, and a UWB tag is attached to the surface; the passive RFID tag attached to the goods is defined as the tag to be tested.

[0010] Furthermore, edge data processing, edge database, positioning computing center, and edge application layer are deployed in the edge server;

[0011] The edge data processing communicates wirelessly with the base station equipment in the physical warehouse; an antenna information table is created in the edge database to bind the mapping relationship between the antenna ID and the antenna coordinates;

[0012] A rack information table is created in the edge database to bind the mapping relationship between the rack coordinates and the corresponding layer columns; a cargo information table is created in the edge database to bind the mapping relationship between cargo attributes and the attributes of the tag to be tested, wherein the attributes of the tag to be tested include EPC code, RSSI, and coordinates to be tested, wherein the RSSI is the signal strength read by the RFID base station, and the EPC code is used as the unique identification indicator of the tag to be tested;

[0013] The positioning calculation center is deployed with a positioning algorithm, and the positioning calculation center receives new data output after edge data processing, and performs passive positioning or active positioning through the positioning algorithm;

[0014] The edge application layer is used to cache the output data of the positioning and computing center. The output data is encapsulated into an API interface of the Http protocol. The API interface returns JSON format data of cargo information and equipment operation data; among which, the cargo information includes cargo ID, shelf ID, layer column, warehousing date, and inventory date. The equipment operation data includes antenna operating frequency, base station output power, and base station operating temperature.

[0015] Furthermore, the passive positioning includes the following steps:

[0016] S100: constructing an RSSI signal path loss model through a GA-BP neural network to calculate the predicted value of the distance between the antenna of the fixed RFID base station and the tag to be tested;

[0017] S110: Determine the spatial coordinates of the tag to be tested by a four-side positioning method;

[0018] S120: Determine the area where the target goods with the tags to be tested are located by using a grid coordinate similarity matching criterion;

[0019] Wherein, the S100 includes the following steps:

[0020] S101: In the storage room, the distance between the antenna of the fixed RFID base station and the RFID tag is adjusted multiple times, and the actual value of the distance between the two and the RSSI value read by the base station are recorded each time;

[0021] S102: construct a three-layer neural network, including an input layer, a hidden layer, and an output layer, use the RSSI value as the input layer, and the distance value as the output layer, and specify that the number of neurons in the input layer, the hidden layer, and the output layer are m, t, and n respectively;

[0022] S103: During the forward propagation of the neural network, the output calculation formula is Y j =W ij X i +B j , after the Sigmoid activation function, the output is The error produced is Where X i is the input value, Y j is the output value, W ij is the weight, B j is the bias, O j is the predicted value, T j is the true value, E j is the error between the predicted value and the true value;

[0023] S104: Initialize N populations, the number of genes of each individual is S = S w +S b , where the total number of weights S w =m*t+t*n, total offset S b =t+n, real number encoding for each individual gene;

[0024] S105: Constructing fitness function The roulette method is used to select individuals. The probability of an individual being selected is Taking each individual as a sector of the roulette wheel, the cumulative probability is

[0025] S106: After the selection operation, select N1 first-generation individuals, implement the crossover operation through gene weighted averaging, and set the crossover probability to P c , suppose there are two individuals A p and A qA genetic crossover occurs at the k-th position, and the two second-generation individuals obtained are and Their genetic characteristics are respectively and where α ∈ (0, 1); set the mutation probability to P m , and let the individual be A mutation occurs at the i-th position to generate a third-generation individual Its genetic characteristic is where r ∈ (-1, 1), is the maximum value of the genetic characteristic, is the minimum value of the genetic characteristic;

[0026] S107: After the first population evolution ends, sort the fitness of the three generations of individuals by size, select M individuals with larger fitness and use them as the initial population for the next evolution, repeat operations S105 and S106 until the number of evolutions reaches the preset value, and then execute S108;

[0027] S108: After the number of evolutions reaches the preset value, obtain the individual with the largest fitness from the three generations of individuals, decode this individual, obtain the optimal weights and biases and assign them to the BP neural network, substitute the RSSI value recorded in S101 into the BP neural network for training. If the error obtained from the training is greater than the preset value, then perform backpropagation through the gradient descent method, continuously update the weights and biases, and the weight update amount The bias update amount where l is the learning rate and 0 < l < 1, until the error does not exceed the preset value to end the training;

[0028] S109: Use the trained GA-BP neural network as the RSSI signal path loss model, substitute the RSSI value read by the RFID base station into the RSSI signal path loss model, obtain the distance prediction value and lay the foundation for the subsequent positioning of the coordinates of the待测标签.

[0029] Furthermore, in S110,联立4个空间距离公式 Calculate the coordinates of the待测标签, where (x i , y i , z i ) represents the antenna coordinates, (x, y, z) represents the coordinates of the待测标签, d i represents the distance prediction value converted from the RSSI signal value received by the antenna for the i-th待测标签, i = 1, 2, 3, 4.

[0030] Furthermore, in S120, based on the coordinates of the tag to be tested inferred by the RSSI signal path loss model and the four-side positioning method, the distance from the coordinates of the tag to be tested to all the shelf coordinates is calculated, the shelf coordinates with the smallest distance are selected, and the corresponding layer column is found in the shelf information table, that is, the shelf layer and column of the tag to be tested is determined, and according to the EPC code of the tag to be tested, the corresponding cargo ID is found in the cargo information table, and the placement position of the target cargo bound to the tag to be tested on the shelf in the warehouse is further determined.

[0031] Furthermore, the active positioning includes the following steps:

[0032] S200: Determine the distance between the UWB base station and the attached UWB tag through a bilateral two-way ranging algorithm;

[0033] S210: determining the coordinates of the mobile RFID base station to which the UWB tag is attached and the coordinates of the tag to be measured in sequence by a four-side positioning method;

[0034] S220: Determine the area where the target goods with the tags to be tested are located by using a grid coordinate similarity matching criterion;

[0035] Wherein, the S200 includes the following steps:

[0036] S201: The UWB base station sends a signal to the UWB tag and records the timestamp τ1. When the UWB tag receives the signal, it also records the timestamp τ2. The signal flight time is T f ;

[0037] S202: UWB tag delays for a fixed period of time T d1 , send a signal to the UWB base station, record the timestamp τ3 at this moment, the base station receives the signal and records the timestamp τ4, during which the signal flight time is T f ;

[0038] S203: The UWB base station delays for a fixed period of time T d2 , and then send another signal to the UWB tag, recording the timestamp τ5 at this moment. The UWB tag receives the signal and records the timestamp τ6. The signal flight time during this period is T f ;

[0039] S204: In the time period from τ1 to τ4, τ4-τ1=2T f +T d1 , from τ3 to τ6, τ6-τ3=2T f +T d2 ,get

[0040] S205: According to the formula d = Tf c is used to calculate the distance d from the UWB base station to the mobile RFID base station with the UWB tag attached, where c represents the flight speed of the UWB signal, i.e., the speed of light.

[0041] Furthermore, the S210 includes the following steps:

[0042] S211: The mobile RFID base station with the UWB tag attached reads the information of all the tags to be tested at a certain location, records the EPC and the corresponding RSSI value of each tag, and calculates the distance between the mobile RFID base station at the current location and the i-th tag to be tested according to the RSSI signal path loss model, i=1,2,..,S, S is the total number of tags to be tested;

[0043] S212: Establishing a coordinate memory of the mobile RFID base station, the coordinate memory records the coordinates of the first base station, and the base station is adjusted to the next position;

[0044] S213: If the coordinates recorded in the coordinate memory reach 4 times, execute S214, otherwise continue to execute S212;

[0045] S214: Get the latest 4 base station coordinate records in the coordinate memory and combine 4 spatial distance formulas Calculate the coordinates of the jth tag to be tested, where (X i ,Y i ,Z i ) represents the coordinates of the mobile RFID base station at the i-th position, (X, Y, Z) represents the coordinates of the tag to be tested, D ij It represents the distance prediction value converted from the RSSI signal value received by the base station at position i for the jth tag to be tested, i=1,2,3,4.

[0046] Furthermore, the twin data system includes twin data processing, virtual-reality interactive mapping, and twin data storage;

[0047] The twin data processing requests cargo information data and storage equipment working data from the edge application layer of the edge server through the Http protocol; the virtual-reality interactive mapping matches the virtual variables with the data variables parsed by the twin data processing, and updates the warehouse digital twin in real time; the twin data storage stores the static data of the warehousing scene and the dynamic data uploaded by the edge server in the database, the static data includes shelf location information and fixed base station location information, and the dynamic data is the cargo location information that the edge server collects, processes, and calculates and outputs in real time from the storage equipment data.

[0048] Further, the virtual system includes a twin model and data visualization;

[0049] The twin model includes a geometric model of the virtual warehouse; the twin model also includes a simulation model, the simulation model includes simulating the motion trajectory of a mobile RFID base station with a UWB tag attached, inputting the coordinates of the UWB tag in several time periods from historical data and drawing the motion trajectory for each second;

[0050] The data visualization displays the virtual warehousing environment in 3D form, and summarizes the quantity of stored goods and the working status of equipment in the form of charts; the data visualization also provides interactive tools, allowing users to freely explore and operate the virtual warehousing environment; the data visualization is also configured with an alarm engine, which includes trigger conditions and triggers. The trigger conditions include setting a temperature range for normal operation of the base station, and the trigger detects the acquired temperature according to the trigger conditions. When a temperature abnormality is detected, the three-dimensional model is highlighted.

[0051] Beneficial effect: A warehouse digital twin system integrating UWB-RFID technology of the present invention, under normal circumstances, uses a fixed RFID base station to conduct real-time inventory of goods in the warehouse, and UWB-RFID technology can be used in unmanned vehicle systems to realize the record of goods in and out of the warehouse. In order to reduce the computing load of the upper-level warehouse digital twin system, a large amount of base station equipment data is uploaded to the edge server, and the location of the goods is output through edge computing and a new API interface is developed for the warehouse digital twin system. The warehouse digital twin system parses the API data and completes the virtual-reality mapping to realize the visualization of all data in the warehouse. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a structural schematic diagram of the warehouse digital twin system of the present invention;

[0053] Figure 2 This is a flow chart of constructing an RSSI signal path loss model based on a GA-BP neural network in the present invention;

[0054] Figure 3 It is a schematic diagram of the bilateral two-way ranging principle in the present invention;

[0055] Figure 4 This is a flow chart of determining the coordinates of the tag to be measured using the four-side positioning method in the present invention. DETAILED DESCRIPTION

[0056] The present invention will be further explained below in conjunction with the accompanying drawings.

[0057] like Figure 1 As shown, a warehouse digital twin system integrating UWB-RFID technology includes a physical warehouse, an edge server, and a warehouse digital twin system.

[0058] 1. Deployment of physical warehousing.

[0059] The physical warehouse includes 4 fixed UWB base stations, multiple fixed RFID base stations, multiple shelves, multiple mobile RFID base stations with UWB tags attached, multiple passive tags and multiple target goods.

[0060] 1) Fixed UWB base stations are installed at the four corners of the warehouse ceiling and can communicate with UWB tags.

[0061] 2) Fixed RFID base stations can support multiple antenna interfaces and be installed on the side of each shelf, without occupying the space for placing goods. Fixed RFID base stations can read data from passive tags in real time. The shelf contains a grid area with layer a and column b, and an antenna is installed at the end of each layer of the shelf.

[0062] 3) Mobile RFID base station The base station has a built-in single antenna and can be a handheld scanner or a vehicle-mounted RFID device. The movement trajectory of the mobile RFID base station is marked by the attached UWB tag, which is used to actively locate the target goods with attached unattached tags.

[0063] 4) The passive tag is an RFID tag, and the memory data of the passive tag includes data indicators such as EPC code, RSSI, number of reads, frequency, etc. The passive tag is attached to the target goods and is used to locate and track the target goods in real time. In this system, it serves as a tag to be measured at the position to be measured, and receives radio energy from the RFID base station to activate and transmit data.

[0064] Establish a spatial positioning model for the above-mentioned physical warehouse. Specifically, the spatial positioning model abstracts the physical warehouse into a three-dimensional space, takes a corner of the ground as the origin, establishes a spatial rectangular coordinate system, stipulates that the ground is the XOY plane, the adjacent walls perpendicular to the ground are the XOZ plane and the YOZ plane respectively, and the shelf is perpendicular to the XOY plane. The positions of all equipment and goods in the warehouse can be expressed by spatial coordinates.

[0065] 2. Deployment of edge servers.

[0066] Edge data processing, edge database, positioning computing center, and edge application layer are deployed in the edge server. The edge server is used to collect base station equipment data in the physical warehouse in real time, and transmit the data to the internal positioning computing center after processing. The positioning computing center deploys the cargo positioning algorithm, combines the warehouse static data stored in the edge database to infer the coordinate information of the tag to be tested, combines the edge database to bind the target cargo, and finally constructs JSON data and outputs it to the edge application layer and uploads it to the warehouse digital twin system.

[0067] 1) Edge data processing can communicate wirelessly with base station equipment in physical storage through routers. The base station equipment acts as a server and the edge data processing acts as a client. The edge data processing connects to the IP and port of the base station equipment. The edge data processing sends an API request to obtain the original data read by the base station and filters, parses, and integrates the original data. Among them, the API is provided by the equipment manufacturer and encapsulates the read and write operations, frequency settings, and output power functions of the base station.

[0068] 2) The edge database includes an antenna information table, which binds the mapping relationship between antenna ID and antenna coordinates. The antenna information table includes antenna attributes such as antenna ID, base station, and antenna coordinates. The antenna ID is the unique identification indicator of the antenna.

[0069] The edge database includes a shelf information table. The shelf is defined as each grid area of ​​the shelf. The shelf attributes include the layer column, shelf ID, and shelf coordinates. The shelf information table binds the mapping relationship between the shelf coordinates and the layer column. Each layer column corresponds to only one shelf coordinate. The shelf coordinate is defined as the coordinate of the geometric center point of the grid area. The layer column represents the layer and column of the grid on the shelf.

[0070] The edge database includes a cargo information table, which binds the mapping relationship between cargo attributes and the attributes of the tag to be tested. The cargo information table includes cargo attributes such as cargo ID, specifications, and weight. The cargo ID is the unique identification indicator of the cargo attribute, and each cargo corresponds to only one tag to be tested. The attributes of the tag to be tested include EPC, RSSI, and coordinates to be tested. RSSI is the signal strength read by the RFID base station, and EPC is the unique identification indicator of the tag to be tested.

[0071] 3) The edge application layer receives the output data from the positioning and computing center, and the output data is encapsulated into an HTTP protocol API interface, which returns cargo information and equipment operation data. The cargo information includes cargo ID, shelf ID, layer column, storage date, inventory date, etc. The equipment operation data includes antenna operating frequency, base station output power, base station operating temperature, etc.

[0072] 4) The positioning computing center receives the new data output by edge data processing. The positioning functions of the positioning computing center include passive positioning and active positioning. Passive positioning is defined as a fixed RFID base station taking real-time inventory of goods on shelves in the warehouse, and active positioning is defined as a mobile RFID base station tracking goods and updating the goods’ in and out status.

[0073] 4.1) Specifically, passive positioning includes the following steps:

[0074] S100: constructing an RSSI signal path loss model through a GA (genetic algorithm)-BP (back propagation) neural network to calculate a predicted value of the distance between the antenna of the fixed RFID base station and the tag to be tested.

[0075] S110: Determine the spatial coordinates of the tag to be tested by a four-side positioning method.

[0076] S120: Determine the area where the target goods with the tags to be tested are located by using a grid coordinate similarity matching criterion.

[0077] like Figure 2 As shown, S100 includes the following specific steps:

[0078] S101: Perform multiple experimental tests on RSSI and distance.

[0079] Specifically, the positions of the antenna and RFID tag of the fixed RFID base station are adjusted, and multiple sets of real distance values ​​and RSSI values ​​read by the base station are recorded.

[0080] S102: Construct a neural network and initialize the number of neurons.

[0081] Specifically, a three-layer neural network is constructed, including an input layer, a hidden layer, and an output layer. The RSSI value is used as the input layer, and the distance value is used as the output layer. The number of neurons in the input layer, the hidden layer, and the output layer is specified to be m, t, and n, respectively.

[0082] S103: Derive the excitation output and error formula of the neural network forward propagation.

[0083] Specifically, during the forward propagation of the neural network, the calculation formula for its output is Y j =W ij X i +B j , after Sigmoid activation function The output is The error produced is Where X i is the input value, Y j is the output value, W ij is the weight, B j is the bias, O j is the predicted value, T j is the true value, E j is the error between the predicted value and the true value.

[0084] S104: Initialize the population of the genetic algorithm and encode the individuals with real numbers.

[0085] Specifically, N populations are initialized, the genetic characteristics of individuals are composed of weights and biases, and the number of genes of each individual is S = Sw +S b , where the total number of weights S w =m*t+t*n, total offset S b =t+n, and each individual gene is encoded with a real number.

[0086] S105: Construct a fitness function and select individuals using a roulette wheel method.

[0087] Specifically, construct the fitness function It can be seen that the smaller the error, the greater the individual fitness. The roulette method is used to select individuals. The probability of an individual being selected is Consider each individual as a sector of the roulette wheel, and mark the two ends of the sector arc with the cumulative probability. The cumulative probability formula is: It seems that for each individual in a probability interval, the individual with a larger sector area is more likely to be selected.

[0088] S106: Perform crossover mutation operation on the selected individuals.

[0089] Specifically, after the selection operation, N1 first-generation individuals are selected, and the gene weighted average of the specific position of the individual is performed to realize the crossover operation. Set the crossover probability to P c , assuming that two individuals A p and A q When a gene crossover occurs at the kth position, the two second-generation individuals obtained are and Their genetic characteristics are and Where α∈(0,1); set the mutation probability to P m , assuming that the individual is A mutation occurs at the i-th position, generating the third generation of individuals Its genetic characteristics are where r∈9-1,1), is the maximum value of the gene feature, is the minimum value of the gene feature.

[0090] S107: Select individuals with greater fitness as the initial population for the next evolution.

[0091] Specifically, after the first population evolution is completed, the fitness of the three generations of individuals is sorted by size, and M individuals with larger fitness are selected as the initial population for the next evolution. Operations S105 and S106 are repeated until the number of evolutions reaches a preset value, and then S108 is executed.

[0092] S108: Decode the optimal individual and substitute it into the neural network for continuous training.

[0093] Specifically, when the number of evolutions reaches the preset value, the individual with the maximum fitness is obtained from the three generations of individuals, decoded, and the optimal weights and biases are obtained and assigned to the BP neural network. The RSSI value recorded in S101 is substituted into the BP neural network for training. If the error obtained from the training is greater than the preset value, backpropagation is performed through the gradient descent method to continuously update the weights and biases Weight update amount Bias update amount where l is the learning rate and 0 < l < 1, and the training ends until the error does not exceed the preset value.

[0094] S109: Substitute the read RSSI value into the RSSI signal path loss model to obtain the distance prediction value.

[0095] Specifically, use the trained GA-BP neural network as the RSSI signal path loss model, substitute the RSSI value read by the RFID base station into the RSSI signal path loss model to obtain the distance prediction value, laying a foundation for the subsequent positioning of the coordinates of the tag to be measured.

[0096] The four-side positioning method of S110 needs to联立4个空间距离公式 calculate the coordinates of the tag to be measured, where (x i , y i , z i ) represents the antenna coordinates, (x, y, z) represents the coordinates of the tag to be measured, and d i represents the distance prediction value converted from the RSSI signal value received by the antenna for the i-th tag to be measured, i = 1, 2, 3, 4.

[0097] Based on the RSSI signal path loss model and the four-side positioning method to infer the coordinates of the tag to be measured, the grid coordinate similarity matching criterion of S120 calculates the distances from the coordinates of the tag to be measured to all grid coordinates, selects the grid coordinate with the minimum distance, searches for the corresponding layer column in the grid information table of the edge database, and determines that the tag to be measured is in the grid on the severalth layer and severalth column of the shelf. According to the EPC of the tag to be measured, the corresponding goods ID is found in the goods information table, further determining the placement position of the target goods bound by the tag to be measured on the shelf in the warehouse.

[0098] 4.2) Active positioning includes the following steps:

[0099] S200: Determine the distance between the UWB base station and the attached UWB tag through the two-way two-way ranging algorithm.

[0100] S210: Determine the coordinates of the mobile RFID base station with the attached UWB tag and the coordinates of the tag to be measured through the four-side positioning method in sequence.

[0101] S220: Determine the area where the target goods with the labels to be tested are located by using a grid coordinate similarity matching criterion.

[0102] like Figure 3 As shown, the bilateral two-way ranging algorithm of S200 includes the following steps:

[0103] S201: The fixed UWB base station sends a signal to the UWB tag and records the sending timestamp τ1. After the UWB tag receives the signal, it also records the receiving timestamp τ2. The signal flight time from sending to receiving is T f .

[0104] S202: UWB tag delays for a period of time T d1 After that, a signal is sent to the fixed UWB base station, and the sending timestamp τ3 is recorded. The base station receives the signal and records the receiving timestamp τ4. The signal flight time is T f .

[0105] S203: The fixed UWB base station delays for a period of time T d2 , and then send a signal to the UWB tag, record the sending timestamp τ5, the UWB tag receives the signal and records the receiving timestamp τ6, during which the signal flight time is T f .

[0106] S204: In the time period from τ1 to τ4, τ4-τ1=2T f +T d1 , from τ3 to τ6, τ6-τ3=2T f +T d2 , it can be deduced from the two formulas

[0107] S205: According to the formula d = T f c is used to calculate the distance d from the UWB base station to the mobile RFID base station with the UWB tag attached, where c represents the flight speed of the UWB signal, i.e., the speed of light.

[0108] like Figure 4 As shown, the four-side positioning method execution process of S210 includes the following steps:

[0109] S211: When the mobile RFID base station with UWB tags attached is at a certain location, it reads the information of all the tags to be tested and records the EPC and corresponding RSSI value of each tag, regardless of whether the antenna rules used by the fixed RFID base station and the mobile RFID base station are the same; the RSSI signal path loss model construction method of S100 is also applicable to mobile RFID base stations. The RSSI measurement value is substituted into the RSSI signal path loss model to calculate the distance between the mobile RFID base station at the current location and the i-th tag to be tested, where i = 1, 2, .., S, and S is the total number of tags to be tested.

[0110] S212: Establishing a coordinate memory of the mobile RFID base station, the coordinate memory records the coordinates of the base station for the first time, and the base station is adjusted to the next position.

[0111] S213: If the coordinates recorded in the coordinate memory reach 4 times, execute S214, otherwise continue to execute S212.

[0112] S214: Get the latest 4 base station coordinate records in the container and combine 4 spatial distance formulas Calculate the coordinates of the jth tag to be tested, where (X i ,Y i ,Z i ) represents the coordinates of the mobile RFID base station at the i-th position, (X, Y, Z) represents the coordinates of the tag to be tested, D ij It represents the distance prediction value converted from the RSSI signal value received by the base station at position i for the jth tag to be tested, i=1,2,3,4.

[0113] In one implementation, the grid coordinate similarity matching criterion implementation process of S220 may refer to the grid coordinate similarity matching criterion of S120.

[0114] 3. Deployment of warehouse digital twin system.

[0115] Twin data processing receives cargo information and equipment information uploaded by the edge application layer, and the virtual-reality interactive mapping completes the interactive mapping of physical warehouse data and virtual warehouse model. The virtual system displays the three-dimensional model and operation data of the physical warehouse in a visual way, and can also perform simulation tests based on historical data.

[0116] In one embodiment, twin data processing is deployed in the warehouse digital twin system, and the twin data processing requests cargo information data and warehouse equipment working data from the edge application layer of the edge server through the Http protocol.

[0117] In one embodiment, virtual-reality interactive mapping is deployed in the warehouse twin data system, and the virtual-reality interactive mapping matches the virtual variables with the data variables processed and parsed by the twin data, and updates the warehouse digital twin system in real time.

[0118] In one embodiment, twin data storage is deployed in the warehouse twin data system. The twin data storage stores the static data of the warehouse scene and the dynamic data uploaded by the edge server in the database. The static data mainly includes shelf location information and fixed base station location information. The dynamic data is the cargo location information collected, processed, calculated and output by the edge server in real time from the warehouse equipment data.

[0119] In one embodiment, a twin model is deployed in a warehouse twin data system. The twin model includes establishing a geometric model of a virtual warehouse, collecting physical data such as geometric data of warehouse facilities (such as floor plans, building structures), inventory data, equipment locations, etc., and using three-dimensional modeling software such as SketchUp to create a geometric model of the virtual warehouse.

[0120] Furthermore, the twin model also includes a simulation model, which includes simulating the movement trajectory of a mobile RFID base station with an attached UWB tag. Since the timestamp recorded by the UWB device can reach the nanosecond level, the UWB tag data is extracted from the historical records, the timestamp is converted to the second level, and a script is written to update the coordinates of the UWB tag frame by frame according to the second-level timestamp. The simulation model is run to observe the movement trajectory of the mobile RFID base station.

[0121] In one embodiment, data visualization is deployed in the warehouse twin data system to realize 3D display of the virtual warehouse environment, summarize the quantity of goods and the working status of equipment, and provide interactive tools and alarm functions.

[0122] Specifically, in one implementation, the 3D model exported from SketchUp is imported into a Unity project, and the quantity of stored goods and the working status of equipment are summarized in the form of charts in combination with js technology.

[0123] Furthermore, data visualization configures interactive tools, allowing users to freely explore and operate the virtual warehousing environment. Virtual targets use a zoom-type display panel, and users can display detailed properties of the target by hovering the mouse over the virtual target.

[0124] Furthermore, data visualization also configures an alarm engine. The alarm engine sets trigger conditions and triggers. The trigger conditions need to be set beyond the normal operating temperature range of the base station equipment. The trigger detects the acquired temperature. When the temperature reaches the trigger conditions, the target three-dimensional model is highlighted.

[0125] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A warehouse digital twin system integrating UWB-RFID technology, characterized in that: Includes physical warehousing, edge servers, and warehousing digital twin systems; The physical warehouse includes 4 fixed UWB base stations, multiple fixed RFID base stations, multiple shelves, a mobile RFID base station with a UWB tag attached, multiple passive tags and multiple target goods, and a spatial positioning model of the physical warehouse is established; The edge server is used to collect data from base station equipment in the physical warehouse in real time, transmit the processed data to the positioning calculation center, and infer the coordinate information of the tag to be tested in combination with the warehouse static data stored in the edge database, and integrate the cargo information and tag information data and send them to the warehouse digital twin; The warehouse digital twin system includes a twin data system and a virtual system; the twin data system is used to complete the interactive mapping of physical warehouse data and virtual warehouse models; the virtual system is used to display the three-dimensional model and operation data of the physical warehouse in a visual way.

2. The warehouse digital twin system integrating UWB-RFID technology according to claim 1 is characterized in that: In the physical warehouse, fixed UWB base stations are installed at the four corners of the warehouse ceiling and can communicate with UWB tags; fixed RFID base stations are installed on the side of each shelf, and an antenna is installed at the end of each shelf layer; mobile RFID base stations are handheld scanners or vehicle-mounted RFID devices, and a UWB tag is attached to the surface; The passive RFID tag attached to the goods is defined as the tag to be tested.

3. The warehouse digital twin system integrating UWB-RFID technology according to claim 1 is characterized in that: The edge server deploys edge data processing, edge database, positioning computing center, and edge application layer; The edge data processing communicates wirelessly with the base station equipment in the physical warehouse; an antenna information table is created in the edge database to bind the mapping relationship between the antenna ID and the antenna coordinates; Create a grid information table in the edge database to bind the mapping relationship between the grid coordinates and the corresponding layer columns; A cargo information table is created in the edge database to bind the mapping relationship between cargo attributes and the attributes of the tag to be tested. The attributes of the tag to be tested include EPC code, RSSI, and coordinates to be tested. The RSSI is the signal strength read by the RFID base station, and the EPC code is used as the unique identification indicator of the tag to be tested. The positioning calculation center is deployed with a positioning algorithm, and the positioning calculation center receives new data output after edge data processing, and performs passive positioning or active positioning through the positioning algorithm; The edge application layer is used to cache the output data of the positioning and computing center. The output data is encapsulated into an API interface of the Http protocol. The API interface returns JSON format data of cargo information and equipment operation data; among which, the cargo information includes cargo ID, shelf ID, layer column, warehousing date, and inventory date. The equipment operation data includes antenna operating frequency, base station output power, and base station operating temperature.

4. The warehouse digital twin system integrating UWB-RFID technology according to claim 3 is characterized in that: The passive positioning comprises the following steps: S100: constructing an RSSI signal path loss model through a GA-BP neural network to calculate the predicted value of the distance between the antenna of the fixed RFID base station and the tag to be tested; S110: Determine the spatial coordinates of the tag to be tested by a four-side positioning method; S120: Determine the area where the target goods with the tags to be tested are located by using a grid coordinate similarity matching criterion; Wherein, the S100 includes the following steps: S101: In the storage room, the distance between the antenna of the fixed RFID base station and the RFID tag is adjusted multiple times, and the actual value of the distance between the two and the RSSI value read by the base station are recorded each time; S102: construct a three-layer neural network, including an input layer, a hidden layer, and an output layer, use the RSSI value as the input layer, and the distance value as the output layer, and specify that the number of neurons in the input layer, the hidden layer, and the output layer are m, t, and n respectively; S103: During the forward propagation of the neural network, the output calculation formula is Y j =W ij X i +B j , after the Sigmoid activation function, the output is The error produced is Where X i is the input value, Y j is the output value, W ij is the weight, B j is the bias, O j is the predicted value, T j is the true value, E j is the error between the predicted value and the true value; S104: Initialize N populations, the number of genes of each individual is S = S w +S b , where the total number of weights S w =m*t+t*n, total offset S b =t+n, real number encoding for each individual gene; S105: Constructing fitness function The roulette method is used to select individuals. The probability of an individual being selected is Taking each individual as a sector of the roulette wheel, the cumulative probability is S106: After the selection operation, select N1 first-generation individuals, implement the crossover operation through gene weighted averaging, and set the crossover probability to P c , suppose there are two individuals A p and A q Gene crossover occurs at the kth position, and the two second-generation individuals obtained are and Their genetic characteristics are and Where α∈(0,1); set the mutation probability to P m , let the individual be A mutation occurs at the i-th position, generating the third generation of individuals Its genetic characteristics are where r∈(-1,1), is the maximum value of the gene feature, is the minimum value of the gene feature; S107: After the first population evolution is completed, the fitness of the three generations of individuals is sorted by size, and M individuals with larger fitness are selected as the initial population for the next evolution. Operations S105 and S106 are repeated until the number of evolutions reaches a preset value, and then S108 is executed; S108: After the number of evolutions reaches the preset value, obtain the individual with the maximum fitness from the three generations of individuals, decode this individual, obtain the optimal weights and biases and assign them to the BP neural network, substitute the RSSI values recorded in S101 into the BP neural network for training. If the error obtained from the training is greater than the preset value, perform backpropagation through the gradient descent method, continuously update the weights and biases, the weight update amount the bias update amount where l is the learning rate and 0 < l < 1, and end the training until the error does not exceed the preset value; S109: Using the trained GA-BP neural network as an RSSI signal path loss model, substituting the RSSI value read by the RFID base station into the RSSI signal path loss model, obtaining a distance prediction value and paving the way for subsequent coordinate positioning of the tag to be tested.

5. The warehouse digital twin system integrating UWB-RFID technology according to claim 4 is characterized in that: In S110, four spatial distance formulas are combined Calculate the coordinates of the tag to be tested, where (x i ,y i ,z i ) represents the antenna coordinates, (x, y, z) represents the coordinates of the tag to be tested, d i It represents the distance prediction value converted from the RSSI signal value received by the antenna for the i-th tag to be tested, i=1,2,3,4.

6. The warehouse digital twin system integrating UWB-RFID technology according to claim 4 is characterized in that: In the S120, based on the coordinates of the tag to be tested inferred by the RSSI signal path loss model and the four-side positioning method, the distance from the coordinates of the tag to be tested to all the shelf coordinates is calculated, the shelf coordinates with the smallest distance are selected, and the corresponding layer column is found in the shelf information table, that is, the shelf of the layer and column of the tag to be tested is determined, and according to the EPC code of the tag to be tested, the corresponding cargo ID is found in the cargo information table, and the placement position of the target cargo bound to the tag to be tested on the shelf in the warehouse is further determined.

7. The warehouse digital twin system integrating UWB-RFID technology according to claim 3 is characterized in that: Active positioning The following steps are involved: S200: Determine the distance between the UWB base station and the attached UWB tag through a bilateral two-way ranging algorithm; S210: determining the coordinates of the mobile RFID base station to which the UWB tag is attached and the coordinates of the tag to be measured in sequence by a four-side positioning method; S220: Determine the area where the target goods with the tags to be tested are located by using a grid coordinate similarity matching criterion; Wherein, the S200 includes the following steps: S201: The UWB base station sends a signal to the UWB tag and records the timestamp τ1. When the UWB tag receives the signal, it also records the timestamp τ2. The signal flight time is T f ; S202: UWB tag delays for a fixed period of time T d1 , send a signal to the UWB base station, record the timestamp τ3 at this moment, the base station receives the signal and records the timestamp τ4, during which the signal flight time is T f ; S203: The UWB base station delays for a fixed period of time T d2 , and then send another signal to the UWB tag, recording the timestamp τ5 at this moment. The UWB tag receives the signal and records the timestamp τ6. The signal flight time during this period is T f ; S204: In the time period from τ1 to τ4, τ4-τ1=2T f +T d1 , from τ3 to τ6, τ6-τ3=2T f +T d2 ,get S205: According to the formula d = T f c is used to calculate the distance d from the UWB base station to the mobile RFID base station with the UWB tag attached, where c represents the flight speed of the UWB signal, i.e., the speed of light.

8. The warehouse digital twin system integrating UWB-RFID technology according to claim 7 is characterized in that: The S210 includes the following steps: S211: The mobile RFID base station with the UWB tag attached reads the information of all the tags to be tested at a certain location, records the EPC and the corresponding RSSI value of each tag, and calculates the distance between the mobile RFID base station at the current location and the i-th tag to be tested according to the RSSI signal path loss model, i=1,2,..,S, S is the total number of tags to be tested; S212: Establishing a coordinate memory of the mobile RFID base station, the coordinate memory records the coordinates of the first base station, and the base station is adjusted to the next position; S213: If the coordinates recorded in the coordinate memory reach 4 times, execute S214, otherwise continue to execute S212; S214: Get the latest 4 base station coordinate records in the coordinate memory and combine 4 spatial distance formulas Calculate the coordinates of the jth tag to be tested, where (X i ,Y i ,Z i ) represents the coordinates of the mobile RFID base station at the i-th position, (X, Y, Z) represents the coordinates of the tag to be tested, D ij It represents the distance prediction value converted from the RSSI signal value received by the base station at position i for the jth tag to be tested, i=1,2,3,4.

9. The warehouse digital twin system integrating UWB-RFID technology according to any one of claims 1 to 3, characterized in that: The twin data system includes twin data processing, virtual-reality interactive mapping, and twin data storage; The twin data processing requests cargo information data and storage equipment working data from the edge application layer of the edge server through the Http protocol; the virtual-reality interactive mapping matches the virtual variables with the data variables parsed by the twin data processing, and updates the warehouse digital twin in real time; the twin data storage stores the static data of the warehousing scene and the dynamic data uploaded by the edge server in the database, the static data includes shelf location information and fixed base station location information, and the dynamic data is the cargo location information that the edge server collects, processes, and calculates and outputs in real time from the storage equipment data.

10. The warehouse digital twin system integrating UWB-RFID technology according to claim 9 is characterized in that: The virtual system includes a twin model and data visualization; The twin model includes a geometric model of the virtual warehouse; the twin model also includes a simulation model, the simulation model includes simulating the motion trajectory of a mobile RFID base station with a UWB tag attached, inputting the coordinates of the UWB tag in several time periods from historical data and drawing the motion trajectory for each second; The data visualization displays the virtual warehousing environment in 3D form, and summarizes the quantity of stored goods and the working status of equipment in the form of charts; the data visualization also provides interactive tools, allowing users to freely explore and operate the virtual warehousing environment; the data visualization is also configured with an alarm engine, which includes trigger conditions and triggers. The trigger conditions include setting a temperature range for normal operation of the base station, and the trigger detects the acquired temperature according to the trigger conditions. When a temperature abnormality is detected, the three-dimensional model is highlighted.

Images