Simple industrial vehicle indoor positioning device and method

Abstract

The invention discloses a simple industrial vehicle indoor positioning device and method, relates to the technical field of indoor positioning, and solves the problem of huge consumption of manpower, material resources and time cost caused by indoor space transformation when an industrial vehicle is positioned indoors. The indoor positioning device comprises a vehicle-mounted positioning end and a monitoring display end; the vehicle-mounted positioning end comprises a switching execution module, an acquisition processing module and a communication transmission module; the monitoring display end comprises an information storage module, a track fusion module and a monitoring display module; performing data fusion on the initial position of the vehicle and the driving component analysis table through a track fusion module by using a route fusion model to generate a vehicle positioning track; the route fusion model is subjected to lightweight adjustment and optimization through the bidirectional wolf pack algorithm and the multi-target whale optimization algorithm, the training time of the route fusion model is shortened, the precision of the vehicle positioning track in the positioning map is improved, and the management difficulty of the industrial vehicle is reduced.

Description

Technical Field

[0001] The present invention relates to the field of indoor positioning technology, and in particular to a simple indoor positioning device and method for industrial vehicles. Background Art

[0002] Industrial vehicles have a huge market demand in China. As special industrial vehicles, industrial vehicles have reliable quality and flexible scene adaptability. They can be seen in airports, logistics areas, ports, docks and other places, and play a vital role in the national economic system. With the growth of industrial vehicle sales and the rapid development of science and technology, industrial vehicles are also moving rapidly towards intelligence, electrification, networking, and safety. Especially in terms of safety, people are paying more and more attention to it.

[0003] At present, most managers have the need for remote monitoring of vehicles, hoping to know the operation status, location and other information of the vehicles in a timely manner, so as to adjust the operation status of the vehicles through digital management technology to improve the operation efficiency of the vehicles. The operation scenes of industrial vehicles are divided into two categories: open outdoor spaces and closed indoor spaces. In open outdoor spaces, satellite positioning works well, and managers can accurately know the location of the vehicles, which can greatly improve the operation efficiency of the vehicles. In closed indoor spaces, satellite signals cannot penetrate walls constructed of steel bars, cement and iron sheets, resulting in managers being unable to obtain the location information of the vehicles, making supervision ineffective and improving the operation efficiency of the vehicles impossible.

[0004] Patent No. CN2020108312967 discloses a vehicle scheduling method and system based on an industrial vehicle chain, including: a weight sensor sends the weight of the goods and the forking time detected during forking to an IOT device; the IOT device collects the starting area of ​​the goods through an indoor positioning device; the IOT device determines whether the current forking task is the first forking in the starting area of ​​the goods, and if so, sets the forking time to the starting time; the weight sensor sends the unloading time detected after unloading to the IOT device; after receiving the unloading time, the IOT device collects the terminal area of ​​the goods through an indoor positioning device, and adds 1 to the number of handling times; based on the number of handling times, starting time and unloading time, the IOT device calculates the busyness of the industrial vehicle, and sends synchronization data to other industrial vehicles on the industrial vehicle chain according to the busyness. The industrial vehicle of the above invention dispatches other vehicles according to the busyness calculated in real time to achieve rapid handling of goods.

[0005] From the above patents, it can be seen that the traditional indoor positioning methods are divided into two types: Option 1, using radio positioning methods such as Bluetooth, WiFi, and UWB; Option 2, using inertial navigation. For Option 1, the surrounding environment needs to be modified. Not only does it need to install radio signal generators at intervals in the indoor space, but it also needs to install radio receivers on the vehicle. Considering the wide operating range of industrial vehicles, Option 1 requires the modification of every indoor space that may be involved, which consumes huge manpower and material costs. At the same time, when there are multiple devices in the scene, the radio solution is prone to signal confusion and interference, which has a serious impact on the accuracy of positioning. In addition, Option 1 also requires a separate map construction for the indoor space environment, which cannot be integrated with the mainstream map, and the positioning effect is not intuitive and accurate. For Option 2, the solution using inertial navigation requires high-computing hardware equipment to perform calculations, and continuously solves and infers information such as azimuth, speed, acceleration, etc. to obtain positioning information accurate to the centimeter level. It is usually used as a lane-level navigation solution. Considering the working environment of industrial vehicles, the introduction of inertial navigation solutions will inevitably cause a sharp increase in the cost of the entire vehicle, and the precise location information it brings is too detailed for managers, so inertial navigation solutions are not suitable for industrial vehicles. Therefore, how to achieve effective positioning indoors to meet usage needs, while at the same time making the price cost compatible with industrial vehicles, has become an urgent problem to be solved in the industry. Summary of the invention

[0006] The purpose of the present invention is to provide a simple industrial vehicle indoor positioning device and method, which can use the route fusion model through the trajectory fusion module to fuse the vehicle initial position and the driving component analysis table to generate the vehicle positioning trajectory; the route fusion model is lightweight tuned through the two-way wolf pack algorithm and the multi-objective whale optimization algorithm to reduce the training time of the route fusion model, improve the accuracy of the vehicle positioning trajectory in the positioning map, and reduce the management difficulty of industrial vehicles.

[0007] The present invention utilizes the following technical solutions: A simple industrial vehicle indoor positioning device includes a vehicle-mounted positioning terminal and a monitoring display terminal; wherein: The vehicle-mounted positioning terminal is used to initialize the positioning information of the industrial vehicle by switching the execution module according to the initialization instruction to obtain the initial position of the vehicle; it is also used to obtain the driving data of the industrial vehicle through the acquisition and processing module, and perform data cleaning and data solving on the driving data, and then construct a driving component analysis table according to the time series; it is also used to transmit the initial position of the vehicle and the driving component analysis table to the monitoring display terminal through the communication transmission module according to the time series to display the vehicle's forward route; The monitoring and display end is used to generate a vehicle positioning trajectory by fusing the received vehicle initial position and driving component analysis table using a trajectory generation algorithm or a route fusion model through a trajectory fusion module; it is also used to display the vehicle positioning trajectory on a positioning map through the monitoring and display module; it is also used to store the vehicle positioning trajectory through an information storage module.

[0008] Preferably, the vehicle-mounted positioning terminal includes a switching execution module, a collection processing module and a communication transmission module; The switching execution module is used to execute the initialization command sent by the monitoring display terminal to initialize the positioning information of the industrial vehicle after the industrial vehicle is started indoors, and obtain the initial position of the vehicle; the positioning information includes the vehicle identity, real-time position and movement status; The acquisition and processing module is used to collect the driving data of industrial vehicles at regular intervals, clean and solve the driving data, and build a driving component analysis table based on the time series; the driving data includes vehicle speed, travel direction and acquisition time; The communication transmission module is used to transmit the vehicle initial position and the driving component analysis table to the monitoring display terminal in time sequence; The monitoring display terminal includes an information storage module, a trajectory fusion module and a monitoring display module; The trajectory fusion module is used to fuse the received vehicle initial position and driving component analysis table to generate a vehicle positioning trajectory; Monitoring and display module, used to display the vehicle positioning trajectory and movement status on the positioning map; The information storage module is used to store the positioning trajectory, driving data, initial position and driving component analysis table of the industrial vehicle.

[0009] Preferably, the acquisition and processing module first periodically obtains the vehicle speed and travel direction respectively measured by the speed sensor and steering sensor of the industrial vehicle, and records the acquisition time of the speed sensor and the steering sensor; then uses the data deduplication algorithm to deduplicate the driving data according to the acquisition time, and uses the filtering algorithm to denoise the deduplicated driving data; then uses the association fitting algorithm to associate the denoised driving data according to the time series in combination with the spatial interpolation algorithm to obtain a dynamic driving sequence table; finally, uses the component resolution algorithm to decompose the vehicle speed and travel direction of the dynamic driving sequence table according to the initial position of the vehicle and the direction position, and constructs a driving component analysis table.

[0010] Preferably, the trajectory fusion module uses the data extraction layer of the route fusion model to extract the azimuth angle component data and the speed component data in the driving component analysis table, and integrates the azimuth angle component data and the speed component data according to the time series to construct a simultaneous sequence vector; uses the feature extraction layer of the route fusion model to extract vector features and obstacle features from the simultaneous sequence vector and the indoor distribution map to construct a vector trajectory feature matrix; uses the learning and training layer of the route fusion model to call the bidirectional wolf pack algorithm, and performs several iterative training on the vector trajectory feature matrix in combination with the initial position of the vehicle to obtain the vector weight feature matrix; uses the update recognition layer of the route fusion model to call the multi-objective whale optimization algorithm, and optimizes and updates the vector weight feature matrix in combination with the obstacle features to identify and output several segments of simulated trajectories; uses the trajectory reconciliation layer of the route fusion model to call the fitting curve algorithm to perform data fusion on several segments of simulated trajectories to generate a vehicle positioning trajectory.

[0011] Preferably, the feature extraction layer includes 4 3X3 convolutional layers, 3 3X3 maximum pooling layers and 2 normalization layers; the learning and training layer includes 4 combined convolution blocks consisting of 1X3 convolutional layers and 3X1 convolutional layers, 1 7X7 upsampling layer, 3 5X5 downsampling layers, 1 3X3 hole convolutional layer, 1 3X3 upsampling layer, 1 3X3 downsampling layer, 4 shuffled blocks of different dimensions, 6 residual blocks of different dimensions, RELU activation function and Tanh activation function; the shuffled block includes 1 1X1 upsampling group layer, 1 1X1 downsampling group layer, 2 3X3 depth separation convolutions, 1 3X3 convolutional layer, 3 normalization layers, 1 3X3 average pooling layer and RELU activation function; the update recognition layer includes 3 average pooling layers, 2 fully connected layers, 1 random dropout layer and cross entropy-KL divergence loss function.

[0012] Preferably, the learning training layer calls the bidirectional wolf pack algorithm to train the vector trajectory feature matrix, and the steps of obtaining the vector weight feature matrix are: A: Initialize the position, number, maximum number of iterations, wolf detection ratio factor, update ratio factor, maximum number of wandering times, maximum number of running times, distance determination factor, step size factor and preset accuracy value of the artificial wolves in the wolf pack; B: The learning and training layer uses 4 combined convolution blocks consisting of 1X3 convolution layers and 3X1 convolution layers, 1 7X7 upsampling layer and 3 5X5 downsampling layers to expand the vector trajectory feature matrix to form a feature polynomial; C: Select the best artificial wolf as the alpha wolf, and divide the artificial wolves other than the alpha wolf into scout wolves and fierce wolves. The scout wolves perform wandering behavior within the characteristic polynomial until the concentration of prey odor detected by a scout wolf is greater than the concentration of prey odor perceived by the alpha wolf or the maximum wandering times is reached, and the first optimized characteristic formula is obtained; D: The fierce wolf rushes into the learning and training layer. If the concentration of prey odor perceived by the fierce wolf is greater than the concentration of prey odor perceived by the alpha wolf, the fierce wolf and the alpha wolf simultaneously summon the pack of wolves to initiate a siege behavior, that is, the fierce wolf combines the network layer architectures in the learning and training layer, and iterates the first optimized feature formula reached by the alpha wolf several times to obtain the second optimized feature formula; if the concentration of prey odor perceived by the fierce wolf is less than the concentration of prey odor perceived by the alpha wolf, the fierce wolf continues to rush in both directions in the learning and training layer until the distance between the fierce wolf and the alpha wolf is less than or equal to the judgment distance or the maximum number of rushes is reached; E: Update the positions of all artificial wolves participating in the siege behavior. Update the position of the alpha wolf according to the "winner takes all" rule, and then update the group according to the "survival of the fittest" wolf pack update mechanism. F: Judge the accuracy value or the number of iterations. If the accuracy value and / or the number of iterations reach the preset value, output the position of the leader, that is, the vector weight feature matrix; if the accuracy value or the number of iterations does not reach the preset value, repeat steps C to F.

[0013] Preferably, the communication transmission module uses the NetBEUI / TCP network protocol to transmit the driving component analysis table and the vehicle initial position to the supervision platform end.

[0014] Preferably, the monitoring and display terminal is a mobile phone APP terminal.

[0015] A simple industrial vehicle indoor positioning method, applied to an industrial vehicle indoor positioning device, comprises the following steps: S1: Install the vehicle-mounted positioning terminal of the indoor positioning device on an industrial vehicle; S2: The industrial vehicle starts, and the speed sensor, steering sensor and vehicle positioning terminal on the vehicle start working; S3: The monitoring display terminal is connected to the switching execution module of the vehicle positioning terminal, and sends an initialization command to initialize the positioning information of the industrial vehicle to obtain the initial position of the vehicle; S4: The acquisition and processing module of the vehicle positioning terminal regularly collects and cleans the driving data of the industrial vehicle and obtains the speed and angle of the industrial vehicle; S5: The acquisition and processing module calculates the angle and speed and obtains the relative position of the vehicle. Quantity and Component, build driving component analysis table; S6: The communication transmission module transmits the driving component analysis table to the trajectory fusion module of the monitoring display terminal; S7: The trajectory fusion module receives the initial position of the vehicle and driving component analysis table and , Indicates the acquisition time, and uses the route fusion model to perform data fusion and obtain the coordinates of the simulated trajectory ( , ), Indicates the current moment; S8: The route fusion model calls the fitting curve algorithm to fuse the data of several simulated trajectories to generate a vehicle positioning trajectory, which is displayed through the monitoring display module and stored through the information storage module; S9: When the industrial vehicle drives outdoors, the signal switching command is executed by the monitoring display terminal to be converted to the satellite positioning terminal to update the positioning information, and the positioning information is displayed on the monitoring display module of the monitoring display terminal, thereby continuously updating the vehicle position.

[0016] The present invention utilizes the route fusion model through the trajectory fusion module to fuse the vehicle initial position and the driving component analysis table to generate the vehicle positioning trajectory; the route fusion model is lightweight tuned through the two-way wolf pack algorithm and the multi-objective whale optimization algorithm, which reduces the training time of the route fusion model, improves the accuracy of the vehicle positioning trajectory in the positioning map, and reduces the management difficulty of industrial vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0018] Figure 1 It is a schematic diagram of an indoor positioning device for industrial vehicles; Figure 2 This is the principle block diagram of the indoor positioning method for industrial vehicles; Figure 3 This is the flow chart of the bidirectional wolf pack algorithm. DETAILED DESCRIPTION

[0019] The present invention is described in detail below with reference to the accompanying drawings and embodiments: like Figure 1 As shown, the simple industrial vehicle indoor positioning device of the present invention includes a vehicle-mounted positioning terminal and a monitoring display terminal; wherein, The vehicle-mounted positioning terminal is used to initialize the positioning information of the industrial vehicle by switching the execution module according to the initialization instruction to obtain the initial position of the vehicle; it is also used to obtain the driving data of the industrial vehicle through the acquisition and processing module, and perform data cleaning and data solving on the driving data, and then construct a driving component analysis table according to the time series; it is also used to transmit the initial position of the vehicle and the driving component analysis table to the monitoring display terminal through the communication transmission module according to the time series to display the vehicle's forward route; The monitoring and display end is used to generate a vehicle positioning trajectory by fusing the received vehicle initial position and the driving component analysis table using a trajectory generation algorithm or a route fusion model through a trajectory fusion module; it is also used to display the vehicle positioning trajectory on a positioning map through the monitoring and display module; it is also used to store the vehicle positioning trajectory through an information storage module; In the present invention, the vehicle-mounted positioning terminal includes a switching execution module, a collection processing module and a communication transmission module; The switching execution module is used to execute the initialization command sent by the monitoring display terminal to initialize the positioning information of the industrial vehicle after the industrial vehicle is started indoors, and obtain the initial position of the vehicle; the positioning information includes the vehicle identity, real-time position and movement status; The acquisition and processing module is used to collect the driving data of industrial vehicles at regular intervals, clean and solve the driving data, and build a driving component analysis table based on the time series; the driving data includes vehicle speed, travel direction and acquisition time; The communication transmission module is used to transmit the vehicle initial position and the driving component analysis table to the monitoring display terminal in time sequence; In this embodiment, the communication transmission module uses the NetBEUI / TCP network protocol to transmit the driving component analysis table and the initial position of the vehicle to the supervision platform end; The monitoring display terminal includes an information storage module, a trajectory fusion module and a monitoring display module; In this embodiment, the monitoring display terminal can be a mobile phone APP terminal, a web page terminal or a smart screen; The trajectory fusion module is used to fuse the received vehicle initial position and driving component analysis table to generate a vehicle positioning trajectory; Monitoring and display module, used to display the vehicle positioning trajectory and movement status on the positioning map; An information storage module, used to store the positioning trajectory, driving data, initial position and driving component analysis table of the industrial vehicle; In the present invention, the acquisition processing module first obtains the vehicle speed and the travel direction respectively measured by the speed sensor and the steering sensor of the industrial vehicle at regular intervals, and records the acquisition time of the speed sensor and the steering sensor; then uses the data deduplication algorithm to deduplicate the driving data according to the acquisition time, and uses the filtering algorithm to denoise the deduplicated driving data; then uses the association fitting algorithm to associate the denoised driving data according to the time series combined with the spatial interpolation algorithm to obtain a dynamic driving sequence table; finally, uses the component resolution algorithm to decompose the vehicle speed and the travel direction of the dynamic driving sequence table according to the initial position of the vehicle and the direction position, and constructs a driving component analysis table; In the present invention, the trajectory fusion module uses the data extraction layer of the route fusion model to extract the azimuth angle component data and the speed component data in the driving component analysis table, and integrates the azimuth angle component data and the speed component data according to the time series to construct a simultaneous sequence vector; uses the feature extraction layer of the route fusion model to extract vector features and obstacle features from the simultaneous sequence vector and the indoor distribution map to construct a vector trajectory feature matrix; uses the learning and training layer of the route fusion model to call the bidirectional wolf pack algorithm, and performs several iterative training on the vector trajectory feature matrix in combination with the initial position of the vehicle to obtain the vector weight feature matrix; uses the update recognition layer of the route fusion model to call the multi-objective whale optimization algorithm, and optimizes and updates the vector weight feature matrix in combination with the obstacle features to identify and output several segments of simulated trajectories; uses the trajectory reconciliation layer of the route fusion model to call the fitting curve algorithm, and performs data fusion on several segments of simulated trajectories to generate a vehicle positioning trajectory; In the present invention, the feature extraction layer includes 4 3X3 convolutional layers, 3 3X3 maximum pooling layers and 2 normalization layers; the learning and training layer includes 4 combined convolution blocks consisting of 1X3 convolutional layers and 3X1 convolutional layers, 1 7X7 upsampling layer, 3 5X5 downsampling layers, 1 3X3 hole convolutional layer, 1 3X3 upsampling layer, 1 3X3 downsampling layer, 4 shuffled blocks of different dimensions, 6 residual blocks of different dimensions, RELU activation function and Tanh activation function; the shuffled block includes 1 1X1 upsampling group layer, 1 1X1 downsampling group layer, 2 3X3 depth separation convolutions, 1 3X3 convolutional layer, 3 normalization layers, 1 3X3 average pooling layer and RELU activation function; the update recognition layer includes 3 average pooling layers, 2 fully connected layers, 1 random dropout layer and cross entropy-KL divergence loss function; like Figure 3 As shown, in the present invention, the learning training layer calls the bidirectional wolf pack algorithm to train the vector trajectory feature matrix, and the steps of obtaining the vector weight feature matrix are: A: Initialize the position, number, maximum number of iterations, wolf detection ratio factor, update ratio factor, maximum number of wandering times, maximum number of running times, distance determination factor, step size factor and preset accuracy value of the artificial wolves in the wolf pack; B: The learning and training layer uses 4 combined convolution blocks consisting of 1X3 convolution layers and 3X1 convolution layers, 1 7X7 upsampling layer and 3 5X5 downsampling layers to expand the vector trajectory feature matrix to form a feature polynomial; C: Select the best artificial wolf as the alpha wolf, and divide the artificial wolves other than the alpha wolf into scout wolves and fierce wolves. The scout wolves perform wandering behavior within the characteristic polynomial until the concentration of prey odor detected by a scout wolf is greater than the concentration of prey odor perceived by the alpha wolf or the maximum wandering times is reached, and the first optimized characteristic formula is obtained; D: The fierce wolf rushes into the learning and training layer. If the concentration of prey odor perceived by the fierce wolf is greater than the concentration of prey odor perceived by the alpha wolf, the fierce wolf and the alpha wolf simultaneously summon the pack of wolves to initiate a siege behavior, that is, the fierce wolf combines the network layer architectures in the learning and training layer, and iterates the first optimized feature formula reached by the alpha wolf several times to obtain the second optimized feature formula; if the concentration of prey odor perceived by the fierce wolf is less than the concentration of prey odor perceived by the alpha wolf, the fierce wolf continues to rush in both directions in the learning and training layer until the distance between the fierce wolf and the alpha wolf is less than or equal to the judgment distance or the maximum number of rushes is reached; E: Update the positions of all artificial wolves participating in the siege behavior. Update the position of the alpha wolf according to the "winner takes all" rule, and then update the group according to the "survival of the fittest" wolf pack update mechanism. F: Determine the accuracy value or the number of iterations. If the accuracy value and / or the number of iterations reach the preset value, output the position of the leader, that is, the vector weight feature matrix; if the accuracy value or the number of iterations does not reach the preset value, repeat steps C to F; Embodiment 1: After the industrial vehicle is started indoors, the switching execution module of the on-board positioning terminal executes the initialization command sent by the monitoring and display terminal to initialize the vehicle identity, real-time position and motion state of the industrial vehicle, and obtain the initial position of the vehicle; the acquisition and processing module of the on-board positioning terminal first obtains the vehicle speed and travel direction respectively measured by the speed sensor and steering sensor of the industrial vehicle at regular intervals, and records the acquisition time of the speed sensor and steering sensor; then uses the data deduplication algorithm to deduplicate the driving data according to the acquisition time, uses the filtering algorithm to denoise the deduplicated driving data, and uses the association fitting algorithm to associate the denoised driving data according to the time series combined with the spatial interpolation algorithm to obtain a dynamic driving sequence table; uses the component solution algorithm to decompose the vehicle speed and travel direction of the dynamic driving sequence table according to the initial position of the vehicle and the direction position, and constructs a driving component analysis table; the communication transmission module transmits the vehicle initial position and the driving component analysis table to the monitoring and display terminal according to the time series; The trajectory fusion module of the monitoring display terminal uses the data extraction layer of the route fusion model to extract the azimuth angle component data and speed component data in the driving component analysis table, and integrates the azimuth angle component data and the speed component data according to the time series to construct a simultaneous sequence vector; through the feature extraction layer of the route fusion model, four 3X3 convolutional layers, three 3X3 maximum pooling layers and two normalization layers are used to extract vector features and obstacle features from the simultaneous sequence vector and the indoor distribution map to construct a vector trajectory feature matrix; through the learning and training layer of the route fusion model, four combined convolution blocks consisting of 1X3 convolutional layers and 3X1 convolutional layers, one 7X7 upsampling layer, three 5X5 downsampling layers, one 3X3 hole convolutional layer, one 3X3 upsampling layer, one 3X3 downsampling layer, four shuffled blocks of different dimensions, six residual blocks of different dimensions, RE The LU activation function and the Tanh activation function are combined with the bidirectional wolf pack algorithm to perform several iterative training on the vector trajectory feature matrix according to the initial position of the vehicle to obtain the vector weight feature matrix; the update recognition layer of the route fusion model uses 3 average pooling layers, 2 fully connected layers, 1 random dropout layer and the cross entropy-KL divergence loss function combined with the multi-objective whale optimization algorithm to optimize and update the vector weight feature matrix in combination with obstacle features to identify and output several segments of simulated trajectories; the trajectory reconciliation layer of the route fusion model is used to call the fitting curve algorithm to fuse the data of several segments of simulated trajectories to generate the vehicle positioning trajectory; the monitoring display module of the monitoring display end displays the vehicle positioning trajectory and motion status on the positioning map; the information storage module of the monitoring display end stores the positioning trajectory, driving data, initial position and driving component analysis table of the industrial vehicle; like Figure 2 As shown, the present invention also includes a simple industrial vehicle indoor positioning method, which is applied to an industrial vehicle indoor positioning device and includes the following steps: S1: Install the vehicle-mounted positioning terminal of the indoor positioning device on an industrial vehicle; S2: The industrial vehicle starts, and the speed sensor, steering sensor and vehicle positioning terminal on the vehicle start working; S3: The monitoring display terminal is connected to the switching execution module of the vehicle positioning terminal, and sends an initialization command to initialize the positioning information of the industrial vehicle to obtain the initial position of the vehicle; S4: The acquisition and processing module of the vehicle positioning terminal regularly collects and cleans the driving data of the industrial vehicle and obtains the speed and angle of the industrial vehicle; S5: The acquisition and processing module calculates the angle and speed and obtains the relative position of the vehicle. Quantity and Component, build driving component analysis table; S6: The communication transmission module transmits the driving component analysis table to the trajectory fusion module of the monitoring display terminal; S7: The trajectory fusion module receives the initial position of the vehicle and driving component analysis table and , Indicates the acquisition time, and uses the route fusion model to perform data fusion and obtain the coordinates of the simulated trajectory ( , ), Indicates the current moment; S8: The route fusion model calls the fitting curve algorithm to fuse the data of several simulated trajectories to generate a vehicle positioning trajectory, and displays it in the positioning map through the monitoring display module and stores it through the information storage module; S9: When the industrial vehicle drives outdoors, the signal switching command is executed from the monitoring display end to the satellite positioning end to update the positioning information, and the information is displayed on the monitoring display module of the monitoring display end, thereby continuously updating the vehicle position; Embodiment 2: The vehicle-mounted positioning end of the indoor positioning device is installed on the industrial vehicle; the industrial vehicle is started, and the speed sensor, steering sensor and vehicle-mounted positioning end on the vehicle start working; the monitoring display end is connected to the switching execution module of the vehicle-mounted positioning end, and an initialization instruction is sent to initialize the positioning information of the industrial vehicle and obtain the initial position of the vehicle; the acquisition and processing module of the vehicle-mounted positioning end regularly collects and cleans the driving data of the industrial vehicle and obtains the speed and angle of the industrial vehicle; the acquisition and processing module solves the angle and speed and obtains the relative initial position of the vehicle. Quantity and Components, build a driving component analysis table; the communication transmission module transmits the driving component analysis table to the trajectory fusion module of the monitoring and display end; the trajectory fusion module uses the route fusion model to perform data fusion on the received vehicle initial position and driving component analysis table to obtain a simulated trajectory; the route fusion model calls the fitting curve algorithm to perform data fusion on several segments of simulated trajectories to generate a vehicle positioning trajectory, and displays it in the positioning map through the monitoring and display module and stores it through the information storage module; when the industrial vehicle drives outdoors, the execution signal switching command is converted from the monitoring and display end to the satellite positioning end to update the positioning information, and is displayed on the monitoring and display module of the monitoring and display end, thereby continuously updating the vehicle position.

Claims

1. A simple indoor positioning device for industrial vehicles, characterized in that: It includes the vehicle positioning terminal and the monitoring display terminal; among which, The vehicle-mounted positioning terminal is used to initialize the positioning information of the industrial vehicle by switching the execution module according to the initialization instruction to obtain the initial position of the vehicle; it is also used to obtain the driving data of the industrial vehicle through the acquisition and processing module, and perform data cleaning and data solving on the driving data, and then construct a driving component analysis table according to the time series; it is also used to transmit the initial position of the vehicle and the driving component analysis table to the monitoring display terminal through the communication transmission module according to the time series to display the vehicle's forward route; The monitoring and display end is used to generate a vehicle positioning trajectory by fusing the received vehicle initial position and driving component analysis table using a trajectory generation algorithm or a route fusion model through a trajectory fusion module; it is also used to display the vehicle positioning trajectory on a positioning map through the monitoring and display module; it is also used to store the vehicle positioning trajectory through an information storage module.

2. The simple indoor positioning device for industrial vehicles according to claim 1, characterized in that: The vehicle-mounted positioning terminal includes a switching execution module, a collection processing module and a communication transmission module; The switching execution module is used to execute the initialization command sent by the monitoring display terminal to initialize the positioning information of the industrial vehicle after the industrial vehicle is started indoors, and obtain the initial position of the vehicle; the positioning information includes the vehicle identity, real-time position and movement status; The acquisition and processing module is used to collect the driving data of industrial vehicles at regular intervals, clean and solve the driving data, and build a driving component analysis table based on the time series; the driving data includes vehicle speed, travel direction and acquisition time; The communication transmission module is used to transmit the vehicle initial position and the driving component analysis table to the monitoring display terminal in time sequence; The monitoring display terminal includes an information storage module, a trajectory fusion module and a monitoring display module; The trajectory fusion module is used to fuse the received vehicle initial position and driving component analysis table to generate a vehicle positioning trajectory; Monitoring and display module, used to display the vehicle positioning trajectory and movement status on the positioning map; The information storage module is used to store the positioning trajectory, driving data, initial position and driving component analysis table of the industrial vehicle.

3. The simple indoor positioning device for industrial vehicles according to claim 2, characterized in that: The acquisition processing module first periodically acquires the vehicle speed and the travel direction respectively measured by the speed sensor and the steering sensor of the industrial vehicle, and records the acquisition time of the speed sensor and the steering sensor; Then, the data deduplication algorithm is used to deduplicate the driving data according to the collection time, and the filtering algorithm is used to denoise the deduplicated driving data; then the association fitting algorithm is used to associate the denoised driving data according to the time series combined with the spatial interpolation algorithm to obtain a dynamic driving sequence table; finally, the component solution algorithm is used to decompose the dynamic driving sequence table according to the vehicle initial position and the direction position to decompose the vehicle speed and travel direction, and construct a driving component analysis table.

4. The simple indoor positioning device for industrial vehicles according to claim 2, characterized in that: The trajectory fusion module uses the data extraction layer of the route fusion model to extract the azimuth angle component data and the speed component data in the driving component analysis table, and integrates the azimuth angle component data and the speed component data according to the time series to construct a simultaneous sequence vector; uses the feature extraction layer of the route fusion model to extract vector features and obstacle features from the simultaneous sequence vector and the indoor distribution map to construct a vector trajectory feature matrix; uses the learning and training layer of the route fusion model to call the bidirectional wolf pack algorithm, and performs several iterative training on the vector trajectory feature matrix in combination with the initial position of the vehicle to obtain the vector weight feature matrix; uses the update recognition layer of the route fusion model to call the multi-objective whale optimization algorithm, and optimizes and updates the vector weight feature matrix in combination with the obstacle features to identify and output several segments of simulated trajectories; uses the trajectory reconciliation layer of the route fusion model to call the fitting curve algorithm, and performs data fusion on several segments of simulated trajectories to generate a vehicle positioning trajectory.

5. The simple indoor positioning device for industrial vehicles according to claim 4, characterized in that: The feature extraction layer includes 4 3X3 convolutional layers, 3 3X3 maximum pooling layers and 2 normalization layers; the learning and training layer includes 4 combined convolution blocks consisting of 1X3 convolutional layers and 3X1 convolutional layers, 1 7X7 upsampling layer, 3 5X5 downsampling layers, 1 3X3 hole convolutional layer, 1 3X3 upsampling layer, 1 3X3 downsampling layer, 4 shuffled blocks of different dimensions, 6 residual blocks of different dimensions, RELU activation function and Tanh activation function; the shuffled block includes 1 1X1 upsampling group layer, 1 1X1 downsampling group layer, 2 3X3 depth separation convolutions, 1 3X3 convolutional layer, 3 normalization layers, 1 3X3 average pooling layer and RELU activation function; the update recognition layer includes 3 average pooling layers, 2 fully connected layers, 1 random dropout layer and cross entropy-KL divergence loss function.

6. The simple indoor positioning device for industrial vehicles according to claim 4, characterized in that: The learning and training layer calls the bidirectional wolf pack algorithm to train the vector trajectory feature matrix, and the steps of obtaining the vector weight feature matrix are: A: Initialize the position, number, maximum number of iterations, wolf detection ratio factor, update ratio factor, maximum number of wandering times, maximum number of running times, distance determination factor, step size factor and preset accuracy value of the artificial wolves in the wolf pack; B: The learning and training layer uses 4 combined convolution blocks consisting of 1X3 convolution layers and 3X1 convolution layers, 1 7X7 upsampling layer and 3 5X5 downsampling layers to expand the vector trajectory feature matrix to form a feature polynomial; C: Select the best artificial wolf as the alpha wolf, and divide the artificial wolves other than the alpha wolf into scout wolves and fierce wolves. The scout wolves perform wandering behavior within the characteristic polynomial until the concentration of prey odor detected by a scout wolf is greater than the concentration of prey odor perceived by the alpha wolf or the maximum wandering times is reached, and the first optimized characteristic formula is obtained; D: The fierce wolf rushes into the learning and training layer. If the concentration of prey odor perceived by the fierce wolf is greater than the concentration of prey odor perceived by the alpha wolf, the fierce wolf and the alpha wolf simultaneously summon the pack of wolves to initiate a siege behavior, that is, the fierce wolf combines the network layer architectures in the learning and training layer, and iterates the first optimized feature formula reached by the alpha wolf several times to obtain the second optimized feature formula; if the concentration of prey odor perceived by the fierce wolf is less than the concentration of prey odor perceived by the alpha wolf, the fierce wolf continues to rush in both directions in the learning and training layer until the distance between the fierce wolf and the alpha wolf is less than or equal to the judgment distance or the maximum number of rushes is reached; E: Update the positions of all artificial wolves participating in the siege behavior. Update the position of the alpha wolf according to the "winner takes all" rule, and then update the group according to the "survival of the fittest" wolf pack update mechanism. F: Judge the accuracy value or the number of iterations. If the accuracy value and / or the number of iterations reach the preset value, output the position of the leader, that is, the vector weight feature matrix; if the accuracy value or the number of iterations does not reach the preset value, repeat steps C to F.

7. The simple indoor positioning device for industrial vehicles according to claim 1, characterized in that: The communication transmission module uses the NetBEUI / TCP network protocol to transmit the driving component analysis table and the vehicle initial position to the supervision platform end.

8. The simple indoor positioning device for industrial vehicles according to claim 1, characterized in that: The monitoring display terminal is a mobile phone APP terminal.

9. A simple industrial vehicle indoor positioning method, applied to the industrial vehicle indoor positioning device according to any one of claims 1 to 8, characterized in that: The following steps are involved: S1: Install the vehicle-mounted positioning terminal of the indoor positioning device on an industrial vehicle; S2: The industrial vehicle starts, and the speed sensor, steering sensor and vehicle positioning terminal on the vehicle start working; S3: The monitoring display terminal is connected to the switching execution module of the vehicle positioning terminal, and sends an initialization command to initialize the positioning information of the industrial vehicle to obtain the initial position of the vehicle; S4: The acquisition and processing module of the vehicle positioning terminal regularly collects and cleans the driving data of the industrial vehicle and obtains the speed and angle of the industrial vehicle; S5: The acquisition and processing module calculates the angle and speed and obtains the relative position of the vehicle. Quantity and Component, build driving component analysis table; S6: The communication transmission module transmits the driving component analysis table to the trajectory fusion module of the monitoring display terminal; S7: The trajectory fusion module receives the initial position of the vehicle and driving component analysis table and , Indicates the acquisition time, and uses the route fusion model to perform data fusion and obtain the coordinates of the simulated trajectory ( , ), Indicates the current moment; S8: The route fusion model calls the fitting curve algorithm to fuse the data of several simulated trajectories to generate a vehicle positioning trajectory, and displays it in the positioning map through the monitoring display module and stores it through the information storage module; S9: When the industrial vehicle drives outdoors, the signal switching command is executed by the monitoring display terminal to be converted to the satellite positioning terminal to update the positioning information, and the positioning information is displayed on the monitoring display module of the monitoring display terminal, thereby continuously updating the vehicle position.

Images