A production park personnel positioning method and system

Abstract

The present invention provides a personnel positioning method and system in a production park. The method trains a personnel behavior model by acquiring historical movement trajectories and historical behavior data of personnel in the production park. The method also trains an environment prediction model by acquiring historical environmental dynamic change data in the production park. Multi-source data is acquired through various sensors and pre-processed. The multi-source data includes Wi-Fi signal strength indication data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park. The pre-processed multi-source data is fused according to a Kalman filter algorithm to obtain a first position. A Bayesian positioning algorithm is used to combine the first position and the output results of the personnel behavior model and the environment prediction model to calculate the probability distribution of the personnel's position in the production park and determine the target position, thereby effectively realizing accurate and real-time positioning of personnel in the production park in a complex environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of personnel positioning, and in particular relates to a personnel positioning method and system for a production park. Background Art

[0002] In production parks, accurate positioning of personnel is crucial for safe production and efficient management. Traditional positioning technologies, such as GPS (Global Positioning System), suffer from severe degradation or even inoperability in indoor or obstructed environments. Emerging technologies like UWB (UltraWideBand), while offering higher accuracy, are costly and complex to deploy. Furthermore, most existing positioning methods fail to fully account for the dynamic nature of production park environments and the behavioral patterns of personnel. This makes it difficult to achieve stable and accurate positioning in complex scenarios such as rapidly moving personnel, multi-story buildings, and densely obstructed areas (such as shelves and pipelines). Summary of the Invention

[0003] Based on this, an embodiment of the present invention provides a production park personnel positioning method and system, which aims to achieve accurate and real-time positioning of personnel in the production park by integrating multiple positioning technologies and intelligent algorithms.

[0004] A first aspect of an embodiment of the present invention provides a method for locating personnel in a production park, the method comprising:

[0005] Obtain historical movement trajectories and historical behavior data of personnel in the production park to train personnel behavior models;

[0006] Obtain historical environmental dynamic change data within the production park and train environmental prediction models;

[0007] Acquire multi-source data through various sensors and pre-process the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indicator data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park;

[0008] According to the Kalman filter algorithm, the pre-processed multi-source data is fused to obtain the first position;

[0009] A Bayesian positioning algorithm is used to calculate the probability distribution of personnel positions in the production park and determine the target position by combining the first position, the result obtained by inputting the trained personnel behavior model into the first position, and the result obtained by inputting the current environment dynamic change data into the trained environment prediction model.

[0010] Furthermore, the step of fusing the pre-processed multi-source data according to the Kalman filter algorithm to obtain the first position includes:

[0011] Defining a state vector of personnel movement and establishing a prediction model, wherein the prediction model is represented by a state transition equation;

[0012] Establish observation models for different types of sensors respectively, and merge the observation values ​​of each sensor into a unified observation vector to form an overall observation equation. The overall observation equation is represented by the observation equation corresponding to each sensor and the observation noise vector;

[0013] In the state prediction process, the state estimation at the previous moment and the state transition equation are used to predict the current state, and at the same time, the state covariance matrix is ​​updated;

[0014] In the update phase, the observation equation is expanded by the first order Taylor and the Jacobian matrix is ​​calculated;

[0015] Calculate the observation residual based on the actual observation value and the predicted observation value;

[0016] Calculating the Kalman gain based on the covariance matrix, the Jacobian matrix, and the noise covariance;

[0017] The predicted state and the observed residual are fused using the Kalman gain to obtain the first position.

[0018] Furthermore, in the step of fusing the preprocessed multi-source data according to the Kalman filter algorithm to obtain the first position, when the sampling frequencies of the sensors are different, the timestamps are aligned by interpolation or zero-order hold; when there are multiple sensors of the same type, the source corresponding to the observation value is determined by nearest neighbor matching.

[0019] Furthermore, in the step of obtaining the historical movement trajectory and historical behavior data of personnel in the production park and training the personnel behavior model, the architecture of the personnel behavior model is an LSTM network, the historical movement trajectory includes at least historical timestamps and historical positions, and the historical behavior data includes at least basic personnel information, inertial measurement data, work calendar and task work order. The personnel behavior model is used to output the position probability distribution within a preset time based on the input timestamp, position, basic personnel information, inertial measurement data, work calendar and task work order.

[0020] Furthermore, in the step of obtaining historical environmental dynamic change data within the production park and training the environmental prediction model, the architecture of the environmental prediction model is a graph neural network, and the historical environmental dynamic change data includes environmental parameters, equipment start / stop status, and the intensity fluctuation range of Wi-Fi / radio frequency signals in various areas, wherein the environmental prediction model is used to output the expected intensity fluctuation range of Wi-Fi / radio frequency signals in various areas based on the input environmental parameters and equipment start / stop status.

[0021] Furthermore, the step of using a Bayesian positioning algorithm to calculate the probability distribution of the personnel's position within the production park and determine the target position by combining the first position, the result obtained by inputting the trained personnel behavior model into the first position, and the result obtained by inputting the current environment dynamic change data into the trained environment prediction model includes:

[0022] Obtaining initial weights for various sensors based on the first location, and adjusting the initial weights based on inertial measurement data and an expected intensity fluctuation range of Wi-Fi / radio frequency signals in an area corresponding to the first location to obtain target weights;

[0023] Calculating a first probability of the person appearing at the first location based on historical data, and revising the first probability based on an output result of the person behavior model to obtain a second probability;

[0024] According to the target weight and the observation probability density function, a sensor joint likelihood model is constructed, and credibility weighted fusion is performed to obtain a likelihood result;

[0025] The posterior probability is calculated according to the second probability and the likelihood result, and the maximum lag estimate is determined according to the posterior probability to obtain the target position.

[0026] Furthermore, the noise covariance is adjusted according to the target weight of the sensor.

[0027] A second aspect of an embodiment of the present invention provides a production park personnel positioning system, which is used to implement the production park personnel positioning method provided by the first aspect of the embodiment of the present invention. The system includes:

[0028] The first training module is used to obtain historical movement trajectories and historical behavior data of personnel in the production park and train personnel behavior models;

[0029] The second training module is used to obtain historical environmental dynamic change data within the production park and train the environmental prediction model;

[0030] An acquisition module, configured to acquire multi-source data through various sensors and pre-process the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indicator data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel within the production park;

[0031] A fusion module, configured to fuse the pre-processed multi-source data according to a Kalman filter algorithm to obtain a first position;

[0032] The calculation module is used to use a Bayesian positioning algorithm, combined with the first position, the result obtained after the trained personnel behavior model is input into the first position, and the result obtained after the current environment dynamic change data is input into the trained environment prediction model, to calculate the probability distribution of the personnel's position in the production park and determine the target position.

[0033] A third aspect of an embodiment of the present invention provides a computer-readable storage medium, including:

[0034] The readable storage medium stores one or more programs, which, when executed by the processor, implement the production park personnel positioning method as described in the first aspect.

[0035] A fourth aspect of an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein:

[0036] The memory is used to store computer programs;

[0037] When the processor is used to execute the computer program stored in the memory, the production park personnel positioning method as described in the first aspect is implemented.

[0038] A production park personnel positioning method and system provided in an embodiment of the present invention trains a personnel behavior model by acquiring historical movement trajectories and historical behavior data of personnel in the production park; acquires historical environmental dynamic change data in the production park and trains an environmental prediction model; acquires multi-source data through various sensors and pre-processes the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indication data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park; fuses the pre-processed multi-source data according to the Kalman filter algorithm to obtain a first position; adopts a Bayesian positioning algorithm, combines the first position, the result obtained after inputting the trained personnel behavior model into the first position, and the result obtained after inputting the current environmental dynamic change data into the trained environmental prediction model, calculates the probability distribution of the personnel's position in the production park, and determines the target position, thereby effectively realizing accurate and real-time positioning of personnel in the production park in a complex environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flowchart of a method for locating personnel in a production park provided in Example 1 of the present invention;

[0040] Figure 2 This is a structural block diagram of a production park personnel positioning system provided by Example 3 of the present invention;

[0041] Figure 3 This is a structural block diagram of an electronic device provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0042] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate several embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0043] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0045] Example 1

[0046] The first embodiment of the present invention provides a method for locating personnel in a production park. Figure 1 , is a flowchart for implementing a method for locating personnel in a production park, specifically including steps S01 to S05.

[0047] Step S01: Acquire historical movement trajectories and historical behavior data of personnel in the production park and train a personnel behavior model.

[0048] Specifically, the architecture of the personnel behavior model is an LSTM network, the historical motion trajectory includes at least historical timestamps and historical positions, and the historical behavior data includes at least basic personnel information, inertial measurement data, work calendar, and task work order. The personnel behavior model is used to output the position probability distribution within a preset time based on the input timestamp, position, basic personnel information, inertial measurement data, work calendar, and task work order. According to the position probability distribution within the preset time, the high-frequency activity areas of personnel in future time periods can be understood, wherein the inertial measurement data can be obtained through the built-in nine-axis sensor (accelerometer, gyroscope, magnetometer) of the work badge worn by the personnel.

[0049] Step S02: Obtain historical environmental dynamic change data within the production park and train an environmental prediction model.

[0050] Specifically, the architecture of the environmental prediction model is a graph neural network, with nodes representing campus grids and edges representing signal propagation paths. The historical environmental dynamic change data includes environmental parameters, device start / stop status, and the intensity fluctuation range of Wi-Fi / RF signals in each region. The environmental prediction model is used to output the expected intensity fluctuation range of Wi-Fi / RF signals in each region based on the input environmental parameters and device start / stop status. The environmental parameters include at least temperature, humidity, and electromagnetic interference intensity, and the device start / stop status includes forklift movement trajectory, machine operating power, etc. It is understandable that in subsequent steps, since the first position can be determined, the environmental dynamic change data at that first position can be obtained when the first position is known. In addition, dynamic environmental changes generally do not affect visual data and inertial measurement data. Therefore, the output of the environmental prediction model is only the intensity fluctuation range of Wi-Fi / RF signals in each region, and does not include the intensity fluctuation range of visual and inertial signals in each region.

[0051] Step S03: Acquire multi-source data through various sensors and pre-process the multi-source data. The multi-source data includes Wi-Fi signal strength indication data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park.

[0052] It should be noted that for Wi-Fi signal strength indication data, Wi-Fi access points (APs) are deployed in a grid layout (e.g., every 20 meters × 20 meters) within the production park. Terminal devices (e.g., work badges) scan AP signals at a frequency of 5Hz, and record MAC addresses, signal strength (dBm), and timestamps. Data preprocessing includes removing invalid data with signal strength below a threshold (e.g., -90dBm), averaging five consecutive sampling values ​​of the same AP to reduce random noise, and converting Wi-Fi signal strength indication data into a grid coordinate probability distribution through a fingerprint library matching algorithm (e.g., KNN, SVM). For radio frequency data, RFID readers can be deployed at key locations such as entrances and exits, boundaries of hazardous areas, and equipment workstations. Personnel wear active tags, and when the reader is triggered, the tag ID, reading time, and reader coordinates are recorded. Data preprocessing includes filtering out duplicate triggered data (e.g., the same Tags are read multiple times within 1 second), and personnel identity and position information are associated with the tag ID. Historical data is combined to calculate the threshold for the length of stay in each area. For visual data, industrial cameras are used to collect data, and the YOLOv8 algorithm is used to detect personnel targets in real time, outputting pixel coordinates and confidence levels. Data preprocessing includes using the camera's intrinsic parameter matrix (K) and extrinsic parameter matrix (R, t) to convert pixel coordinates into world coordinates. For occluded scenes (such as people being partially obscured by shelves), the DeepSORT algorithm is used to track trajectories, and missing data is supplemented by historical coordinate interpolation. Preprocessing of inertial measurement data includes using the Mahony complementary filtering algorithm to fuse accelerometer and gyroscope data, outputting attitude angles (pitch, roll, and heading), suppressing accelerometer integral drift through the zero-speed correction (ZUPT) algorithm, and resetting the displacement error when the person is stationary (speed ≤ 0.1m / s for 1 second).

[0053] Step S04: fusing the pre-processed multi-source data according to the Kalman filter algorithm to obtain a first position.

[0054] In the embodiment of the present invention, first, when the sampling frequencies of the sensors are different, the timestamps are aligned by interpolation or zero-order keeping. When there are multiple sensors of the same type, the signal source corresponding to the observation value is determined by nearest neighbor matching. Furthermore, the state vector of the personnel movement is defined, and a prediction model is established. The prediction model is represented by a state transfer equation, wherein the state vector is defined according to the actual two-dimensional or three-dimensional requirements. Taking two-dimensional as an example, the state vector is , x represents the horizontal axis, y represents the vertical coordinate, v x represents the velocity on the horizontal axis, v y represents the velocity on the ordinate. Assuming that the personnel moves at a uniform speed, the dynamic equation is used to describe the state transition from time k−1 to k. That is, the state transition equation is:

[0055] ;

[0056] Among them, the state transfer matrix F is:

[0057] ;

[0058] Δt is the time interval, is the process noise (obeying Gaussian distribution , is the covariance matrix);

[0059] Observation models for different types of sensors are established separately, and the observation values ​​of each sensor are merged into a unified observation vector to form an overall observation equation. The overall observation equation is represented by the observation equation corresponding to each sensor and the observation noise vector. Specifically, the observation model of Wi-Fi signal strength indicator data is expressed as:

[0060] ;

[0061] A is the reference signal strength, n is the path loss index, For the AP location, in practice, it is necessary to jointly solve the Wi-Fi signal strength indicator data of multiple APs;

[0062] The observation model of RF data is expressed as:

[0063] ;

[0064] is the reader position, r is the effective detection radius of RF;

[0065] The observation model of visual data is expressed as:

[0066] ;

[0067] is the position in the camera coordinate system, is the observation noise;

[0068] In the inertial measurement data observation model, the acceleration a and angular velocity ω are directly measured, and the position and velocity are updated by integration;

[0069] Then according to the observation equation corresponding to each sensor , forming the overall observation equation:

[0070] ;

[0071] is the observation noise vector;

[0072] In the state prediction process, the state estimation of the previous moment is used And the state transition equation predicts the current state, expressed as:

[0073] ;

[0074] is the predicted state;

[0075] At the same time, update the state covariance matrix , expressed as:

[0076] ;

[0077] In the update phase, the observation equation predicts the state Perform a first-order Taylor expansion and calculate the Jacobian matrix, which is expressed as:

[0078] ;

[0079] is the Jacobian matrix. According to the actual observation value and the predicted observation value, the observation residual is calculated and expressed as:

[0080] ;

[0081] is the observation residual, is the actual observed value, is the predicted observation value;

[0082] The Kalman gain is calculated based on the covariance matrix, the Jacobian matrix and the noise covariance, which is expressed as:

[0083] ;

[0084] is the Kalman gain, is the noise covariance;

[0085] The predicted state and the observed residual are fused using the Kalman gain to obtain the first position, which is expressed as:

[0086] ;

[0087] For the first position, additionally, the covariance update is expressed as:

[0088] .

[0089] In step S05, a Bayesian positioning algorithm is used to calculate the probability distribution of the personnel's position in the production park and determine the target position by combining the first position, the result obtained by inputting the trained personnel behavior model into the first position, and the result obtained by inputting the current environment dynamic change data into the trained environment prediction model.

[0090] Specifically, according to the first position, the initial weights of various sensors are obtained, and the initial weights are adjusted according to the inertial measurement data and the expected intensity fluctuation range of the Wi-Fi / radio frequency signal in the area corresponding to the first position to obtain the target weight. It should be noted that, according to the inertial measurement data and the corresponding mapping relationship, the behavior pattern matching factor of the personnel is determined. For example, when the personnel is in the inspection mode, the behavior pattern matching factor is 1.2. Then, according to the size of the expected intensity fluctuation range and the corresponding mapping relationship, the environmental impact factor is determined. For example, the larger the expected intensity fluctuation range, the smaller the environmental impact factor. Finally, the initial weight, the behavior pattern matching factor and the environmental impact factor are multiplied, the initial weight is adjusted, and normalization is performed to obtain the target weight, ensuring that the sum of the weights is 1;

[0091] Based on historical data, a first probability of the person appearing at the first location is calculated. This can be understood as dividing the person's historical stay time at the first location by the total historical stay time. Based on the output of the person behavior model, the first probability is modified to obtain a second probability. This is the probability of the person being at the first location output by the LSTM network, multiplied by the first probability to obtain the second probability.

[0092] According to the target weight and observation probability density function, a sensor joint likelihood model is constructed, which is expressed as:

[0093] ;

[0094] is the target weight, is the observation probability density function of sensor i at position s, and n is the total number of sensors;

[0095] Then the credibility weighted fusion is performed to obtain the likelihood result, which is expressed as:

[0096] ;

[0097] According to the second probability and the likelihood result, the posterior probability is calculated, which is expressed as:

[0098] ;

[0099] is the second probability, is the likelihood result, S is the set of all discrete grid points in the production park;

[0100] According to the posterior probability, the maximum delay estimate is determined to obtain the target position, which is expressed as:

[0101] ;

[0102] For the target position, it is understandable that we traverse all grid points and find the position with the largest posterior probability.

[0103] In summary, an embodiment of the present invention proposes a method for positioning personnel in a production park. The method trains a personnel behavior model by obtaining historical movement trajectories and historical behavior data of personnel in the production park; obtains historical environmental dynamic change data in the production park to train an environmental prediction model; obtains multi-source data through various sensors and pre-processes the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indication data, radio frequency data, visual data, and inertial measurement data for representing the location information of personnel in the production park; according to the Kalman filtering algorithm, the pre-processed multi-source data is fused to obtain a first position; a Bayesian positioning algorithm is used to calculate the probability distribution of the personnel's position in the production park by combining the first position, the result obtained after inputting the trained personnel behavior model into the first position, and the result obtained after inputting the current environmental dynamic change data into the trained environmental prediction model, and the target position is determined, thereby effectively realizing accurate and real-time positioning of personnel in the production park in a complex environment.

[0104] Example 2

[0105] The second embodiment of the present invention also provides a method for locating personnel in a production park. The difference from the first embodiment of the present invention is that the noise covariance is adjusted according to the target weight of the sensor. In this embodiment of the present invention, a mapping relationship between the weight and the noise covariance is established. The higher the sensor credibility (the greater the weight), the smaller its observation noise; conversely, the lower the credibility (the smaller the weight), the greater the noise.

[0106] It should be noted that since the target weight is determined after the first position is obtained, and in the process of determining the first position, the noise covariance corresponding to the preset initial weight is applied to the Kalman filtering algorithm, in order to optimize the first position, the noise covariance corresponding to the target weight is re-processed according to the Kalman filtering algorithm, and the preprocessed multi-source data is fused to obtain the first position operation, and a new first position is obtained. Based on the new first position, the corresponding target weight is determined, and the above operation is repeated until the first position no longer changes, or the cycle stops after a preset number of times, and the subsequent target position determination operation is performed with the first position obtained after stopping.

[0107] Example 3

[0108] The third embodiment of the present invention provides a production park personnel positioning system 200, see Figure 2 , is a structural block diagram of a production park personnel positioning method system, the production park personnel positioning system 200 includes:

[0109] A first training module 21 is used to obtain historical movement trajectories and historical behavior data of personnel within the production park and train a personnel behavior model. The architecture of the personnel behavior model is an LSTM network. The historical movement trajectories include at least historical timestamps and historical locations. The historical behavior data includes at least basic personnel information, inertial measurement data, work calendar, and task work orders. The personnel behavior model is used to output a position probability distribution within a preset time based on the input timestamp, location, basic personnel information, inertial measurement data, work calendar, and task work orders.

[0110] A second training module 22 is configured to acquire historical environmental dynamic change data within the production park and train an environmental prediction model. The environmental prediction model is structured as a graph neural network. The historical environmental dynamic change data includes environmental parameters, device startup and shutdown status, and the intensity fluctuation range of Wi-Fi / RF signals in various areas. The environmental prediction model is configured to output the expected intensity fluctuation range of Wi-Fi / RF signals in various areas based on the input environmental parameters and device startup and shutdown status.

[0111] An acquisition module 23 is configured to acquire multi-source data through various sensors and pre-process the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indicator data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park;

[0112] A fusion module 24 is configured to fuse the pre-processed multi-source data using a Kalman filter algorithm to obtain a first position. When the sampling frequencies of the sensors are different, the timestamps are aligned by interpolation or zero-order hold. When there are multiple sensors of the same type, the source corresponding to the observation value is determined by nearest neighbor matching.

[0113] The calculation module 25 is used to use the Bayesian positioning algorithm, combined with the first position, the result obtained after the trained personnel behavior model is input into the first position, and the result obtained after the current environment dynamic change data is input into the trained environment prediction model, to calculate the probability distribution of the personnel's position in the production park and determine the target position.

[0114] Furthermore, in some other embodiments of the present invention, the fusion module 24 includes:

[0115] A first establishing unit is used to define a state vector of personnel movement and establish a prediction model, wherein the prediction model is represented by a state transition equation;

[0116] a second establishing unit, configured to establish observation models for different types of sensors respectively, and merge the observation values ​​of each sensor into a unified observation vector to form an overall observation equation, wherein the overall observation equation is represented by the observation equation corresponding to each sensor and the observation noise vector;

[0117] The prediction unit is used to predict the current state using the previous state estimate and the state transition equation during the state prediction process, and at the same time, update the state covariance matrix;

[0118] The first calculation unit is used to perform a first-order Taylor expansion on the observation equation and calculate the Jacobian matrix in the update phase;

[0119] The second calculation unit is used to calculate the observation residual according to the actual observation value and the predicted observation value;

[0120] a third calculation unit, configured to calculate a Kalman gain according to the covariance matrix, the Jacobian matrix, and the noise covariance, and to adjust the noise covariance according to a target weight of the sensor;

[0121] The fourth computing unit is configured to fuse the predicted state and the observed residual using the Kalman gain to obtain the first position.

[0122] Furthermore, in some other embodiments of the present invention, the calculation module 25 includes:

[0123] an adjustment unit, configured to obtain initial weights of various sensors based on the first location, and adjust the initial weights based on inertial measurement data and an expected intensity fluctuation range of Wi-Fi / radio frequency signals in an area corresponding to the first location to obtain target weights;

[0124] a correction unit, configured to calculate a first probability of a person appearing at the first location based on historical data, and correct the first probability based on an output result of the person behavior model to obtain a second probability;

[0125] A fusion unit is used to construct a sensor joint likelihood model based on the target weight and the observation probability density function, and perform credibility weighted fusion to obtain a likelihood result;

[0126] A fifth calculation unit is used to calculate a posterior probability according to the second probability and the likelihood result, and determine a maximum lag estimate according to the posterior probability to obtain the target position.

[0127] Example 4

[0128] The fourth embodiment of the present invention provides an electronic device, see Figure 3 , is a structural block diagram of an electronic device, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, the production park personnel positioning method as described above is implemented.

[0129] In some embodiments, the processor 10 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, used to run program codes or process data stored in the memory 20, such as executing access restriction programs.

[0130] The memory 20 includes at least one type of readable storage medium, including flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 20 may be an internal storage unit of the electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 may also be an external storage device of the electronic device, such as a plug-in hard disk equipped on the electronic device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Furthermore, the memory 20 may include both an internal storage unit of the electronic device and an external storage device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or is about to be output.

[0131] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned production park personnel positioning method.

[0132] Those skilled in the art will appreciate that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device), or in conjunction with such instruction execution system, apparatus, or device. For purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by an instruction execution system, apparatus, or device, or in conjunction with such instruction execution system, apparatus, or device.

[0133] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting, or processing it in another suitable manner as necessary, and then storing it in a computer memory.

[0134] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the aforementioned embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following technologies known in the art may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0135] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0136] The above embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A production park personnel positioning method, characterized in that: The method comprises: Obtain historical movement trajectories and historical behavior data of personnel in the production park to train personnel behavior models; Obtain historical environmental dynamic change data within the production park and train environmental prediction models; Acquire multi-source data through various sensors and pre-process the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indicator data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel in the production park; According to the Kalman filter algorithm, the pre-processed multi-source data is fused to obtain the first position; Using a Bayesian positioning algorithm, the first position, the result obtained by inputting the first position into a trained personnel behavior model, and the result obtained by inputting the current environment dynamic change data into a trained environment prediction model are combined to calculate the probability distribution of the personnel's position within the production park and determine the target position; The architecture of the personnel behavior model is an LSTM network. The historical motion trajectory includes at least historical timestamps and historical locations. The historical behavior data includes at least basic personnel information, inertial measurement data, work calendar, and task work orders. The personnel behavior model is used to output the location probability distribution within a preset time based on the input timestamp, location, basic personnel information, inertial measurement data, work calendar, and task work orders. The environment prediction model is structured as a graph neural network. The historical environmental dynamic change data includes environmental parameters, device startup and shutdown status, and the intensity fluctuation range of Wi-Fi / RF signals in various areas. The environment prediction model is configured to output the expected intensity fluctuation range of Wi-Fi / RF signals in various areas based on the input environmental parameters and device startup and shutdown status. The step of using a Bayesian positioning algorithm to calculate the probability distribution of the personnel's position within the production park and determine the target position by combining the first position, the result obtained by inputting a trained personnel behavior model into the first position, and the result obtained by inputting the current environment dynamic change data into a trained environment prediction model, includes: Obtaining initial weights for various sensors based on the first location, and adjusting the initial weights based on inertial measurement data and an expected intensity fluctuation range of Wi-Fi / radio frequency signals in an area corresponding to the first location to obtain target weights; Calculating a first probability of the person appearing at the first location based on historical data, and revising the first probability based on an output result of the person behavior model to obtain a second probability; According to the target weight and the observation probability density function, a sensor joint likelihood model is constructed, and credibility weighted fusion is performed to obtain a likelihood result; Calculating a posterior probability based on the second probability and the likelihood result, and determining a maximum lag estimate based on the posterior probability to obtain the target position; Among them, the noise covariance is adjusted according to the target weight of the sensor. Specifically, in order to optimize the first position, the noise covariance corresponding to the target weight is re-processed according to the Kalman filter algorithm to fuse the preprocessed multi-source data to obtain the first position, and a new first position is obtained. Then, based on the new first position, the corresponding target weight is determined, and the above operation is repeated until the first position no longer changes, or the cycle is stopped after a preset number of times, and the subsequent target position determination operation is performed with the first position obtained after stopping.

2. The production park personnel positioning method according to claim 1, characterized in that: The step of fusing the pre-processed multi-source data according to the Kalman filter algorithm to obtain the first position includes: Defining a state vector of personnel movement and establishing a prediction model, wherein the prediction model is represented by a state transition equation; Establish observation models for different types of sensors respectively, and merge the observation values ​​of each sensor into a unified observation vector to form an overall observation equation. The overall observation equation is represented by the observation equation corresponding to each sensor and the observation noise vector; In the state prediction process, the state estimation at the previous moment and the state transition equation are used to predict the current state, and at the same time, the state covariance matrix is ​​updated; In the update phase, the observation equation is expanded by the first order Taylor and the Jacobian matrix is ​​calculated; Calculate the observation residual based on the actual observation value and the predicted observation value; Calculating the Kalman gain based on the covariance matrix, the Jacobian matrix, and the noise covariance; The predicted state and the observed residual are fused using the Kalman gain to obtain the first position.

3. The production park personnel positioning method according to claim 2, characterized in that: In the step of fusing the preprocessed multi-source data according to the Kalman filter algorithm to obtain the first position, when the sampling frequencies of the sensors are different, the timestamps are aligned by interpolation or zero-order hold; when there are multiple sensors of the same type, the source corresponding to the observation value is determined by nearest neighbor matching.

4. A production park personnel positioning system, characterized in that: For implementing the production park personnel positioning method according to any one of claims 1 to 3, the system comprises: The first training module is used to obtain historical movement trajectories and historical behavior data of personnel in the production park and train personnel behavior models; The second training module is used to obtain historical environmental dynamic change data within the production park and train the environmental prediction model; An acquisition module, configured to acquire multi-source data through various sensors and pre-process the multi-source data, wherein the multi-source data includes Wi-Fi signal strength indicator data, radio frequency data, visual data, and inertial measurement data used to represent the location information of personnel within the production park; A fusion module, configured to fuse the pre-processed multi-source data according to a Kalman filter algorithm to obtain a first position; The calculation module is used to use a Bayesian positioning algorithm, combined with the first position, the result obtained after the trained personnel behavior model is input into the first position, and the result obtained after the current environment dynamic change data is input into the trained environment prediction model, to calculate the probability distribution of the personnel's position in the production park and determine the target position.

5. A computer-readable storage medium, characterized in that include: The readable storage medium stores one or more programs, which, when executed by a processor, implement the production park personnel positioning method according to any one of claims 1 to 3.

6. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the production park personnel positioning method described in any one of claims 1-3.

Images