Positioning method and system suitable for different network signals in cross-border transportation scene

Abstract

The invention discloses a positioning method and system suitable for different network signals in a cross-border transportation scene, and belongs to the technical field of cross-border logistics transportation cargo positioning. The method comprises the following steps: receiving a signal during network switching, carrying out weight distribution on the signal during network switching, carrying out weighted calculation on the network signal with the weight through a weighted summation algorithm to serve as the input of a BP neural network, outputting the longitude and latitude of the network signal, and carrying out deviation calculation on the longitude and latitude of the network signal and the longitude and latitude of a pre-stored fixed landmark, and positioning adapting to different network modes in a cross-border transportation scene is obtained. According to the method, the signal characteristics of 2G, 3G, 4G and 5G networks are fused, a machine learning model is combined, and the positioning weight is dynamically adjusted according to the signal quality in the network switching process, so that the positioning process is not interrupted when the equipment is switched among different networks, reliable positioning is realized by integrating the advantages of multiple networks, powerful support is provided for informatization management of cross-border transportation, and the method is suitable for popularization and application. And the positioning accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of positioning of goods in cross-border logistics transportation, and particularly to a positioning method and system adaptable to different network signals in a cross-border transportation scenario. Background Art

[0002] Cross-border e-commerce business has put forward higher requirements for the real-time position accuracy of goods during transportation. In cross-border transportation, transportation equipment frequently shuttles through different network coverage areas, and network switching is frequent. How to effectively integrate positioning information under different network modes and give full play to the advantages of each network has become a problem to be solved urgently. During cross-border transportation, it is difficult to rely on a single network mode to achieve accurate and continuous positioning of goods, which seriously affects the informatization management and efficient operation of cross-border transportation.

[0003] However, with the rapid development of machine learning technology, its application in the field of positioning has gradually deepened. In particular, neural network models have shown strong advantages in signal feature fusion and position estimation. However, most of the existing machine learning models are optimized for a single network mode and fail to make full use of signal features under multiple network modes, resulting in limited application effects in cross-border transportation scenarios. Therefore, how to design a method that can adapt to different network modes, dynamically adjust positioning weights, and ensure positioning accuracy and continuity has become an urgent problem in the current technical field. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a positioning method and system adaptable to different network signals in a cross-border transportation scenario.

[0005] To achieve the above technology, specifically as follows:

[0006] A positioning method adaptable to different network signals in a cross-border transportation scenario includes the following steps:

[0007] S1. Receive signals during network switching through a signal receiving module; the signals during network switching are switched from a signal with high quality to a signal with low quality, including: switching from 5G signal to 4G signal, switching from 4G signal to 3G signal, and switching from 3G signal to 2G signal;

[0008] When the signal receiving module enters a 2G signal area, obtain the strength vector (Received Signal Strength Indicator, RSSI) of the network signal to reflect distance attenuation;

[0009] When the signal receiving module enters a 3G signal area, obtain the time difference of arrival vector (Time Difference of Arrival, TDOA) of the network signal for hyperbolic positioning;

[0010] When the signal receiving module enters the 4G signal area, obtain the intensity vector of the network signal and the time of arrival (TOA) of the network signal for calculating the distance.

[0011] When the signal receiving module enters the 5G signal area, obtain the multipath signal feature vector for assisting in environment recognition and the angle of arrival vector (AOA) of the network signal for providing direction information.

[0012] S2. Assign weights to the signals during network switching to obtain network signals with weights.

[0013] The weight assignment is as follows: the network weight before the network switching signal is set to ≥60%, and the network weight after the network switching signal is set to ≤40%.

[0014] During network switching, the signal weight of the network is linearly positively correlated with the network signal quality.

[0015] When the network signal quality increases, the network weight after the network switching signal increases to be higher than the network weight before the network switching signal and satisfies the weight ≥60%.

[0016] The expression of the weight assignment is as follows:

[0017] ω curr (t) = ω curr (0) + γ·(1 - e -λt )

[0018] ω target (t) = 1 - ω curr (t)

[0019]

[0020] In the formula, ω curr (t) represents the proportion of the positioning weight of the network after the traditional network switching signal at time t; ω curr (0) represents the proportion of the positioning weight of the network after the traditional network switching signal at the initial moment (t = 0); γ represents the adjustment rate for controlling the speed of weight change; ω target (t) is the proportion of the positioning weight of the network before the traditional network switching signal at time t; λ represents the attenuation coefficient for dynamically adjusting the speed of weight change; λ base represents the basic attenuation coefficient; Q target represents the signal quality score of the target network; Q current represents the signal quality score of the network after the traditional network switching signal; Q max and Q minThe maximum and minimum values of the signal quality score are used for normalization calculation.

[0021] S3. Perform weighted calculation on the network signals with weights through the weighted summation algorithm as the input of the BP neural network;

[0022] The input of the BP neural network is two or more signals with weights among 5G signal, 4G signal, 3G signal, and 2G signal; when input, the weights of the input signals add up to 100%;

[0023] The expression of the weighted summation algorithm is as follows:

[0024]

[0025] In the formula, X represents the fused result; ω i represents the weight of the i-th network; x i represents the i-th network.

[0026] S4. Perform calculation on the fused result through the BP neural network with a hidden layer where the number of neurons decreases layer by layer for 3 layers, and output the longitude and latitude of the network signal;

[0027] The BP neural network includes an input layer, a hidden layer, and an output layer; among them, the input of the input layer is the weighted calculation result of S3, and the output layer outputs the longitude and latitude of the network signal; data passes through the input layer, the hidden layer, and the output layer in sequence;

[0028] All hidden layers are fully connected layers;

[0029] The number of neurons decreases as follows: the first layer uses 128 neurons and adopts the ReLU activation function; the second hidden layer contains 64 neurons and adopts the LeakyReLU activation function; the third hidden layer contains 32 neurons and adopts the Sigmoid activation function;

[0030] The number of neurons in the first layer needs to be greater than the dimension of the input features. Using 128 neurons can fully capture low-order non-linear features and avoid information loss; the ReLU activation function can avoid the problem of gradient disappearance;

[0031] The number of neurons in the second layer reduces the number of neurons, enabling the network to fuse high-order interaction information of multi-network signals; the LeakyReLU activation function improves the robustness of feature fusion;

[0032] The number of neurons in the third layer is further reduced, enabling the network to learn the most critical abstract features, that is, the "information distillation" operation; the Sigmoid activation function compresses the output to (0,1);

[0033] The expressions of the 3 hidden layers are respectively:

[0034] h1 = ReLU(W1·X + b1)

[0035] h2 = LeakyReLU(W2·h1 + b2)

[0036] h3 = Sigmoid(W3·h2 + b3)

[0037] Wherein, h1, h2, and h3 respectively represent the first hidden layer, the second hidden layer, and the third hidden layer; W1, W2, and W3 respectively represent the weight matrices of the first hidden layer, the second hidden layer, and the third hidden layer; b1, b2, and b3 respectively represent the bias vectors of the first hidden layer, the second hidden layer, and the third hidden layer.

[0038] S5. By calculating the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmarks, if it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual location. If it is greater than the preset threshold, re-execute S1 - S3 until it is less than the preset threshold, and then obtain the location adapted to different network modes in the cross-border transportation scenario;

[0039] The pre-stored fixed landmarks include: base station coordinates, customs inspection points, traffic lights, bridges, tunnel entrances;

[0040] The formula for deviation calculation is:

[0041]

[0042] Wherein, A 融合 and B 融合 represent the longitude and latitude of the network signal output by the BP neural network; A' 地标 and B' 地标 represent the longitude and latitude of the pre-stored fixed landmarks.

[0043] A positioning device adapted to different network modes in a cross-border transportation scenario includes: a signal receiving module, a weight adjustment module, a data input module, a data processing module, and a positioning evaluation module;

[0044] The signal receiving module is used to execute: S1. Receive the signal during network switching through the signal receiving module;

[0045] The weight adjustment module is used to execute: S2. Allocate the weight of the signal during network switching according to a preset ratio to obtain a weighted network signal;

[0046] The data input module is used to execute: S3. Perform weighted calculation on the weighted network signal through the weighted summation algorithm as the input of the BP neural network;

[0047] The data processing module is used to perform: S4. Use a BP neural network to perform calculations on the fused results through a hidden layer with a decreasing number of neurons in three layers, and output the longitude and latitude of the network signal.

[0048] The positioning evaluation module is used to perform: S5. Calculate the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmarks. If it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual positioning. If it is greater than the preset threshold, re-execute S1 - S3 until it is less than the preset threshold.

[0049] The positioning device in the cross-border transportation scenario of the present invention receives periodic signals according to the transportation route and the communication network coverage area.

[0050] The periodic signal reception is as follows: In the area where the transportation route is stable and the communication network coverage is continuous, extend the signal sampling period. At the key nodes of the route or in the signal attenuation area, shorten the signal sampling period. By adaptively adjusting the signal sampling interval, reduce the energy consumption of redundant signal search, and at the same time ensure the positioning continuity and the integrity of the transportation trajectory to meet the endurance requirements of long-distance cross-border transportation. The specific sampling period formula is:

[0051]

[0052] In the formula, S stable represents the length of the stable section; S total represents the total path length; T base represents the basic sampling period; T represents the actual signal sampling period.

[0053] The beneficial effects of the present invention:

[0054] The present invention integrates the signal characteristics of 2G, 3G, 4G, and 5G networks, combines with a machine learning model, and gives full play to the advantages of each network. Compared with the traditional single-network positioning method, the positioning accuracy is significantly improved.

[0055] During the network switching process of the present invention, the positioning weight is dynamically adjusted according to the signal quality, so that the positioning process is not interrupted when the device switches between different networks, ensuring the stability and continuity of positioning, and ensuring the continuous and reliable acquisition of the position of goods in cross-border transportation.

[0056] The present invention can effectively cope with the challenges of network signal differences in different countries and regions. In the areas where the network signals are unstable or have large differences, reliable positioning is achieved by integrating the advantages of multiple networks, providing strong support for the informatization management of cross-border transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is the step flow chart of the present invention;

[0058] Figure 2 This is the network structure diagram of the present invention. Detailed implementation manners

[0059] The present invention will be further described in detail below in conjunction with specific embodiments.

[0060] As Figure 1 shown, a positioning method for adapting to different network signals in a cross-border transportation scenario includes the following steps:

[0061] S1. Receive the signal during network switching through a signal receiving module; the signal during network switching is switched from a high-quality signal to a low-quality signal, including: switching from a 5G signal to a 4G signal, switching from a 4G signal to a 3G signal, and switching from a 3G signal to a 2G signal;

[0062] When the signal receiving module enters a 2G signal area, obtain the intensity vector (Received Signal Strength Indicator, RSSI) of the network signal to reflect distance attenuation;

[0063] When the signal receiving module enters a 3G signal area, obtain the time difference of arrival vector (Time Difference of Arrival, TDOA) of the network signal for hyperbolic positioning;

[0064] When the signal receiving module enters a 4G signal area, obtain the intensity vector of the network signal and the time of arrival (Time of Arrival, TOA) of the network signal to calculate the distance;

[0065] When the signal receiving module enters a 5G signal area, obtain the multipath signal feature vector for assisting in environment recognition and the angle of arrival vector (Angle of Arrival, AOA) of the network signal to provide direction information.

[0066] S2. Assign weights to the signal during network switching to obtain a network signal with weights;

[0067] The weight assignment is as follows: the network weight before the network switching signal is set to ≥60%, and the network weight after the network switching signal is set to ≤40%;

[0068] During network switching, the signal weight of the network is linearly positively correlated with the network signal quality;

[0069] When the network signal quality increases, the network weight after the network switching signal increases to be higher than the network weight before the network switching signal and satisfies the weight ≥60%;

[0070] The expression of the weight assignment is as follows:

[0071] ω curr (t) = ω curr (0) + γ·(1 - e -λt )

[0072] ω target (t) = 1 - ω curr (t)

[0073]

[0074] Wherein, ω curr (t) represents the proportion of the positioning weight of the network after the traditional network handover signal at time t; ω curr (0) represents the proportion of the positioning weight of the network after the traditional network handover signal at the initial time (t = 0); γ represents the adjustment rate, which is used to control the speed of weight change; ω target (t) is the proportion of the positioning weight of the network before the traditional network handover signal at time t; λ represents the attenuation coefficient, which is used to dynamically adjust the change speed of the weight; λ base represents the basic attenuation coefficient; Q target represents the signal quality score of the target network; Q current represents the signal quality score of the network after the traditional network handover signal; Q max and Q min are the maximum and minimum values of the signal quality score, which are used for normalization calculation.

[0075] S3. Perform weighted calculation on the network signals with weights through the weighted summation algorithm as the input of the BP neural network;

[0076] The input of the BP neural network is: two or more signals with weights among 5G signal, 4G signal, 3G signal and 2G signal; when input, the weights of the input signals add up to 100%;

[0077] The expression of the weighted summation algorithm is as follows:

[0078]

[0079] Wherein, X represents the fused result; ω i represents the weight of the i-th network; x i represents the i-th network.

[0080] S4. Perform calculation on the fused result through the BP neural network with a hidden layer whose number of neurons decreases layer by layer, and output the longitude and latitude of the network signal;

[0081] Such as Figure 2As shown, the BP neural network includes: an input layer, a hidden layer, and an output layer; among them, the input of the input layer is the weighted calculation result of S3, and the output layer outputs the longitude and latitude of the network signal; the data passes through the input layer, the hidden layer, and the output layer in sequence;

[0082] All hidden layers are fully connected layers;

[0083] The number of neurons decreases as follows: the first layer uses 128 neurons and adopts the ReLU activation function; the second hidden layer contains 64 neurons and adopts the LeakyReLU activation function; the third hidden layer contains 32 neurons and adopts the Sigmoid activation function;

[0084] The number of neurons in the first layer needs to be greater than the dimension of the input features. Using 128 neurons can fully capture low-order non-linear features and avoid information loss; the ReLU activation function can avoid the problem of gradient disappearance;

[0085] The number of neurons in the second layer reduces the number of neurons, enabling the network to fuse high-order interaction information of multiple network signals; the LeakyReLU activation function enhances the robustness of feature fusion;

[0086] The number of neurons in the third layer is further reduced, enabling the network to learn the most critical abstract features, namely the "information distillation" operation; the Sigmoid activation function compresses the output to (0, 1);

[0087] The expressions of the 3 hidden layers are respectively:

[0088] h1 = ReLU(W1·X + b1)

[0089] h2 = LeakyReLU(W2·h1 + b2)

[0090] h3 = Sigmoid(W3·h2 + b3)

[0091] In the formula, h1, h2, and h3 respectively represent the first hidden layer, the second hidden layer, and the third hidden layer; W1, W2, and W3 respectively represent the weight matrices of the first hidden layer, the second hidden layer, and the third hidden layer; b1, b2, and b3 respectively represent the bias vectors of the first hidden layer, the second hidden layer, and the third hidden layer.

[0092] S5. By calculating the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmarks, if it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual location. If it is greater than the preset threshold, S1 - S3 are re-executed until it is less than the preset threshold, and then the location adapted to different network modes in the cross-border transportation scenario is obtained;

[0093] Pre-stored fixed landmarks include: base station coordinates, customs checkpoints, traffic lights, bridges, tunnel entrances;

[0094] The formula for deviation calculation is:

[0095]

[0096] In the formula, A 融合 and B 融合 represent the longitude and latitude of the network signal output by the BP neural network; A' 地标 and B' 地标 represent the longitude and latitude of the pre-stored fixed landmarks.

[0097] A positioning device adapted to different network modes in a cross-border transportation scenario includes: a signal reception module, a weight adjustment module, a data input module, a data processing module, and a positioning evaluation module;

[0098] The signal reception module is used to perform: S1, receive the signal during network switching through the signal reception module;

[0099] The weight adjustment module is used to perform: S2, allocate weights to the signal during network switching according to a preset ratio to obtain a weighted network signal;

[0100] The data input module is used to perform: S3, perform weighted calculation on the weighted network signal through a weighted summation algorithm as the input of the BP neural network;

[0101] The data processing module is used to perform: S4, perform a hidden layer calculation with a decreasing number of neurons in 3 layers on the fused result through the BP neural network, and output the longitude and latitude of the network signal;

[0102] The positioning evaluation module is used to perform: S5, calculate the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmarks. If it is less than the preset threshold, the longitude and latitude of the network signal output by the BP neural network are considered as the actual positioning. If it is greater than the preset threshold, re-execute S1 - S3 until it is less than the preset threshold.

[0103] The positioning device adapted to different network modes in the cross-border transportation scenario of the present invention performs periodic signal reception according to the transportation path and the communication network coverage area;

[0104] The periodic signal reception is: extend the signal sampling period in the area where the transportation path is stable and the communication network coverage is continuous, and shorten the signal sampling period at the key path nodes or signal attenuation areas; by adaptively adjusting the signal sampling interval, reduce the energy consumption of redundant signal search, and at the same time ensure the positioning continuity and the integrity of the transportation trajectory to meet the endurance requirements of long-distance cross-border transportation. The specific sampling period formula is:

[0105]

[0106] In the formula, S stable represents the length of the stable section; S total represents the total path length; T base represents the basic sampling period; T represents the actual signal sampling period.

[0107] The device includes a detachable battery pack for power supply. The battery pack is configured with a high-capacity lithium-ion battery pack to provide continuous power support for the device throughout the full cycle of cross-border transportation. The device can optimize positioning and save power consumption according to the characteristics of the transportation route.

[0108] In summary, the present invention provides a precise positioning method and system for cross-border transported goods under different network modes. By integrating the signal characteristics of 2G, 3G, 4G, and 5G networks and combining the powerful processing capabilities of machine learning models, the advantages of each network are fully utilized, significantly improving the accuracy of positioning.

[0109] The above specific embodiments of the present invention are only used for exemplary illustration or explanation of the principle of the present invention, and do not constitute a limitation to the present invention. Therefore, any modifications, equivalent replacements, improvements, etc. made without departing from the spirit and scope of the present invention shall be included within the protection scope of the present invention. In addition, the appended claims of the present invention are intended to cover all variations and modification examples that fall within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A positioning method for adapting to different network signals in a cross-border transportation scenario, characterized in that, It includes the following steps: S1. Receive the signal during network switching through the signal receiving module; the signal during network switching is switched from a signal with high quality to a signal with low quality; S2. Assign weights to the signal during network switching to obtain a network signal with weights; The weight assignment is as follows: the network weight before the network switching signal is set to ≥60%, and the network weight after the network switching signal is set to ≤40%; During network switching, the signal weight of the network is linearly positively correlated with the network signal quality; When the network signal quality increases, the network weight after the network switching signal increases to be higher than the network weight before the network switching signal and satisfies the weight ≥60%; S3. Perform weighted calculation on the network signal with weights through the weighted summation algorithm as the input of the BP neural network; The input of the BP neural network is: two or more signals with weights among 5G signal, 4G signal, 3G signal and 2G signal; when inputting, the weights of the input signals add up to 100%; S4. Perform calculation on the fused result through the BP neural network with a hidden layer whose number of neurons decreases layer by layer in three layers, and output the longitude and latitude of the network signal; S5. Calculate the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmark. If it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual positioning. If it is greater than the preset threshold, re-execute S1 - S3 until it is less than the preset threshold, and then obtain the positioning adapted to different network modes in the cross-border transportation scenario.

2. The positioning method adapted to different network signals in a cross-border transportation scenario according to claim 1, wherein The expression of the weight assignment for assigning weights to the signal during network switching according to a preset ratio to obtain a network signal with weights is as follows: ω curr (t) = ω curr (0) + γ·(1 - e -λt ) ω target (t) = 1 - ω curr (t) where ω curr (t) represents the proportion of the positioning weight of the network after the traditional network handover signal at time t; ω curr (0) represents the proportion of the positioning weight of the network after the traditional network handover signal at the initial moment (t = 0); γ represents the adjustment rate; ω target (t) represents the proportion of the positioning weight of the network before the traditional network handover signal at time t; λ represents the attenuation coefficient; λ base represents the basic attenuation coefficient; Q target represents the signal quality score of the target network; Q current represents the signal quality score of the network after the traditional network handover signal; Q max and Q min represent the maximum and minimum values of the signal quality score.

3. The positioning method adapted to different network signals in a cross-border transportation scenario according to claim 1, characterized in that Perform weighted calculation on the network signal with weights through the weighted summation algorithm as the input of the BP neural network; The expression of the weighted summation algorithm is as follows: where X represents the fused result; ω i represents the weight of the i-th network; x i represents the i-th network.

4. A positioning method for adapting to different network signals in a cross-border transportation scenario according to claim 1, characterized in that, The BP neural network that performs calculation on the fused result through the BP neural network with a hidden layer whose number of neurons decreases layer by layer in three layers and outputs the longitude and latitude of the network signal includes: an input layer, a hidden layer and an output layer; data passes through the input layer, the hidden layer and the output layer in sequence; The number of neurons decreases as follows: 128 neurons are used in the first layer, and the ReLU activation function is adopted; the second hidden layer contains 64 neurons, and the Leaky ReLU activation function is adopted; the third hidden layer contains 32 neurons, and the Sigmoid activation function is adopted.

5. A positioning method for adapting to different network signals in a cross-border transportation scenario according to claim 1, characterized in that, The formula for calculating the deviation in calculating the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmark. If it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual positioning. If it is greater than the preset threshold, re-execute S1 - S3 until it is less than the preset threshold, and then obtain the positioning adapted to different network modes in the cross-border transportation scenario is: where A 融合 and B 融合 represent the longitude and latitude of the output network signal of the BP neural network; A' 地标 and B' 地标 represent the longitude and latitude of the pre-stored fixed landmark.

6. A positioning device adapted to different network modes in a cross-border transportation scenario, comprising: A signal receiving module, a weight adjustment module, a data input module, a data processing module and a positioning evaluation module; The signal receiving module is used to execute: S1. Receive the signal during network switching through the signal receiving module; The weight adjustment module is used to execute: S2. Distribute the weights of the signals during network switching according to a preset ratio to obtain network signals with weights. The data input module is used to execute: S3. Perform weighted calculation on the network signals with weights through the weighted summation algorithm as the input of the BP neural network. The data processing module is used to execute: S4. Perform resolution on the fused result through a hidden layer with a decreasing number of neurons in three layers by the BP neural network, and output the longitude and latitude of the network signal. The positioning evaluation module is used to execute: S5. Calculate the deviation between the longitude and latitude of the network signal output by the BP neural network and the longitude and latitude of the pre-stored fixed landmark. If it is less than the preset threshold, it is considered that the longitude and latitude of the network signal output by the BP neural network are the actual positioning. If it is greater than the preset threshold, re-execute S1-S3 until it is less than the preset threshold, and then obtain the positioning adapted to different network modes in the cross-border transportation scenario.

7. The positioning device adapted to different network modes in the cross-border transportation scenario according to claim 6, wherein the device receives signals periodically according to the transportation path and the communication network coverage area, and the sampling period formula is: where S stable represents the length of the stable section; S total represents the total path length; T base represents the basic sampling period; T represents the actual signal sampling period.

Images