Virtual space boundary early warning method

Abstract

The invention relates to a virtual space boundary early warning method, which comprises the following steps of: 1, defining a virtual boundary by taking an electric power facility as a center; 2, tracking the position of the monitored object in the monitored area; 3, based on a boundary detection algorithm, the shortest distance between the monitored object and the virtual boundary is judged; and step 4, performing early warning and warning for the shortest distance between the monitored object and the virtual boundary. According to the invention, through real-time monitoring, intelligent analysis and rapid response, a layer of additional safety guarantee is provided for operation and maintenance of electric power facilities, human errors and accidents are reduced, and stable operation of an electric power system and life safety of personnel are ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of power supply equipment safety and discloses a virtual space boundary early warning alarm method. Background Art

[0002] In the electric power industry, especially during the operation and maintenance of power facilities, operators face many safety risks, including but not limited to electric shock, falling from heights, equipment failure, etc.

[0003] At present, virtual space safety boundaries are being widely used and studied. This field urgently needs to use virtual space safety boundaries to implement a warning alarm method to improve operational safety, reduce human errors and accidents, and help ensure the safety of power facilities and personnel. Summary of the invention

[0004] The present invention discloses a virtual space boundary early warning alarm method, which comprises the following steps:

[0005] Step 1: Define the virtual boundary with the power facility as the center;

[0006] Step 2: Tracking the location of the monitored object in the monitoring area;

[0007] Step 3: Based on the boundary detection algorithm, determine the shortest distance between the monitored object and the virtual boundary;

[0008] Step 4: Issue an early warning and an alarm based on the shortest distance between the monitored object and the virtual boundary.

[0009] Furthermore, in step 1, the following steps are also included:

[0010] Step 11, using high-precision GPS and RTK technology to determine the center point of the power facility;

[0011] Step 12, according to safety regulations and technical requirements, the virtual boundary is set to a basic circular boundary with a radius of 5 meters;

[0012] Step 13, for substations with complex shapes, use computer-aided design tools to draw composite boundaries;

[0013] Step 14, digitize the boundary data and store it in the system database.

[0014] Furthermore, in step 2, the following steps are also included:

[0015] Step 21, equipping the monitored object with an RFID tag;

[0016] Step 22, monitoring the monitoring area based on laser radar and high-definition camera;

[0017] Step 23, using Kalman filtering algorithm to fuse multiple sensor information to establish a nonlinear dynamic model;

[0018] Step 24, according to the nonlinear dynamics model, the state at the next moment is predicted using the current state and the control input to calculate the shortest distance between the monitored object and the virtual boundary.

[0019] Furthermore, in step 23, the following steps are also included:

[0020] Step 231, obtaining relative position information provided by the RFID tag, distance information provided by the laser radar, and visual positioning information provided by the image captured by the high-definition camera;

[0021] Step 232, defining a system state vector X, wherein the state vector X includes the position coordinates of the person and the speed caused by the change in position;

[0022] Step 233, establish a nonlinear dynamic model f describing the movement of personnel as follows:

[0023] x k+1 =f(x k ,u k )+w k

[0024] Where k is a time step, which represents a discrete time point at which the system estimates and updates its state. At each time step k, the estimate of the person's position is updated based on the state estimate at the previous time step k-1 and the current measurement data; u k represents the change in velocity, i.e., acceleration, which serves as the control input of the model; w k represents the process noise caused by the position change of personnel; x k represents the position coordinates and velocity of the person in k time step; x k+1 Indicates the position coordinates and velocity of the person in k+1 time step;

[0025] Step 234, establish corresponding nonlinear models for RFID, laser radar and high-definition camera respectively, wherein the model corresponding to RFID is:

[0026] z RFID =h RFID (x k )+v RFID

[0027] z RFID is the actual measurement value of RFID, v RFID is the RFID measurement noise, h RFID (x k ) is the RFID nonlinear measurement model;

[0028] The model corresponding to the laser radar is:

[0029] z Lidar =h Lidar (x k )+v Lidar

[0030] z Lidar is the actual measurement value of the laser radar, v Lidar is the lidar measurement noise, h Lidar (x k ) is the nonlinear measurement model of laser radar;

[0031] The model corresponding to the HD camera is:

[0032] z Camera =h Camera (x k )+v Camera

[0033] z Camera is the actual measurement value of the high-definition camera, v Camera Measuring noise for high-definition cameras, h Camera (x k ) is the nonlinear measurement model of the high-definition camera.

[0034] Furthermore, in step 234, the RFID nonlinear measurement model h RFID (x k ) is determined as follows:

[0035] Deploy multiple RFID tags at known locations, record the RSSI values ​​of wireless signal strength at different locations, and fit the formula between the location of the target person and RSSI;

[0036] Lidar nonlinear measurement model h Lidar (x k ) is determined as follows:

[0037] Use triangulation to fit the distance between the target person's position and the lidar;

[0038] Nonlinear measurement model of high-definition camera Camera (x k ) is determined as follows:

[0039] First, the camera's internal and external parameters are calibrated; the internal parameters include focal length and principal point, and the external parameters include position and posture relative to the world coordinate system;

[0040] Secondly, extract the key feature points of the target person from the image captured by the camera;

[0041] Again, deep learning is used to identify and track the key feature points of the target person to determine the nonlinear measurement model of the target person's position and the image captured by the camera.

[0042] Furthermore, in step 24, the following steps are also included:

[0043] Step 241, predict the time step k=1 according to the nonlinear dynamic model and the initial state x0 Right now as follows:

[0044]

[0045] Among them, u0 is the control input at the first time step, that is, k = 0, w0 is the process noise at the first time step, that is, k = 0, which is assumed to be zero-mean Gaussian white noise; x0 represents the initial state;

[0046] Process noise w k The noise covariance matrix is ​​denoted as Q k , where the noise covariance matrix is ​​used to describe the uncertainty of the nonlinear dynamics model;

[0047] When k = 0, the corresponding noise covariance matrix is ​​Q0;

[0048] State vector X k The state covariance matrix of k , where the state covariance matrix is ​​used to describe the uncertainty of state estimation;

[0049] Step 242, further update the state covariance matrix P 1|0 :

[0050] P 1|0 =F0P 0|0 F0 T +Q0

[0051] Among them, P 0|0 is the initial state covariance matrix, F0 is the Jacobian matrix of f to the state vector X, and we have:

[0052]

[0053] Step 243, obtain the first measurement value Z1 when k=1, and update the following variables:

[0054] Known Kalman gain K k The following formula:

[0055] K k =P k|k-1 H k T(H k P k|k-1 H k T +R k ) -1

[0056] Among them, R k is the covariance matrix of the measurement noise, which is used to describe the uncertainty of the measurement;

[0057] P k|k-1 represents the state covariance matrix of the prediction phase from time step k-1 to k, and has:

[0058] P k|k-1 =F k P k-1|k-1 F k T +Q k-1

[0059] Among them, P k-1|k-1 is the optimal state covariance matrix of the previous time step k-1, Q k-1 is the covariance matrix of the process noise at k-1; F k is the Jacobian matrix of f for the state vector X, which is:

[0060]

[0061] in, is the optimal state estimate at the previous time step k-1;

[0062] H k is the Jacobian matrix of h for the state vector X, which is:

[0063]

[0064] For the optimal state estimate at step k, it is calculated as follows:

[0065]

[0066] Among them, z k Represents the actual measurement value of the current k-th step;

[0067] According to the predicted state Calculated expected measurements;

[0068] The update formula of the known state covariance matrix is ​​as follows:

[0069] P k|k =(IK k H k ) k|k-1

[0070] Where I represents the identity matrix, and the Kalman gain K1 is further calculated:

[0071] K1=P 1|0 H1 T (H1P 1|0 H1 T +R1) -1

[0072] And further update the state estimate:

[0073]

[0074] Further update the state covariance matrix:

[0075] P 1|1 =(I-K1H1)P 1|0

[0076] Step 244, continue iterative calculation based on the above, use the Kalman gain, optimal state estimate, and state covariance matrix of the previous time step to continuously predict the state of the next time step, and continuously update the Kalman gain, state estimate, and state covariance matrix, so as to iteratively update the estimate of the position of the monitored object at each time step, thereby achieving high-precision position tracking.

[0077] Furthermore, in step 241, for the initial state X0, its corresponding initial state covariance matrix P 0|0 , where the initial state covariance matrix P 0|0 represents a measure of the uncertainty in the initial state estimate and is set as follows:

[0078] When reliable prior information is available, the reliable prior information is directly used as X0; and the initial state covariance matrix P 0|0 The values ​​in are set to smaller values ​​to indicate lower uncertainty;

[0079] If there is no reliable prior information, the predicted value of the initial state is based on the empirical value. And use this to set the initial state X0; and the initial state covariance matrix P 0|0 The values ​​in are set to larger values ​​to indicate higher uncertainty.

[0080] Furthermore, the virtual space boundary early warning alarm method further comprises the following steps:

[0081] Step 5: Display a real-time map in the on-site control room and on mobile devices, marking the location of the operators and the virtual boundary; the operator in the control room can adjust the virtual boundary parameters through the interface.

[0082] Furthermore, the virtual space boundary early warning alarm method further comprises the following steps:

[0083] Step 6: Record the time, location, operator ID and event details of each alarm trigger;

[0084] Analyze alarm frequency and trigger conditions, optimize virtual boundary parameters and alarm mechanisms, and improve system performance.

[0085] Furthermore, the virtual space boundary early warning alarm method further comprises the following steps:

[0086] Step 7: The data transmission adopts AES-256 encryption standard to ensure the security of data during transmission.

[0087] The beneficial effects achieved by the present invention are:

[0088] Through real-time monitoring, intelligent analysis and rapid response, it provides an additional layer of safety for the operation and maintenance of power facilities, helps reduce human errors and accidents, and ensures the stable operation of the power system and the safety of personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 It is a flowchart of a virtual space boundary early warning alarm method proposed in the first embodiment of the present invention. DETAILED DESCRIPTION

[0090] The present invention will be further described below in conjunction with specific embodiments, and the advantages and features of the present invention will become clearer as the description proceeds. However, these embodiments are exemplary only and do not constitute any limitation to the scope of the present invention. It should be understood by those skilled in the art that the details and forms of the technical solution of the present invention may be modified or replaced without departing from the spirit and scope of the present invention, but these modifications and replacements all fall within the scope of protection of the present invention.

[0091] like Figure 1 As shown, in this embodiment, a virtual space boundary early warning alarm method is provided, and the method includes the following steps:

[0092] Step 1: Define virtual boundaries

[0093] Location data: Using GPS and RTK technology, combined with ground base stations, centimeter-level positioning accuracy is achieved, and a circular virtual safety boundary with a radius of 5 meters is defined with the power facility as the center. Shape and size of the virtual boundary: For substations, according to their layout and safety regulations, a composite shape boundary containing all key facilities is defined, such as a combination of multiple connected rectangles and circles, to ensure that all high-voltage areas are covered.

[0094] The specific steps include:

[0095] Step 11: Use high-precision GPS and RTK technology to determine the center point of the power facility.

[0096] Step 12, set the basic circular boundary (radius 5 meters) according to safety regulations and technical requirements.

[0097] Step 13: For substations with complex shapes, a computer-aided design (CAD) tool is used to draw a composite boundary.

[0098] Step 14, digitize the boundary data and store it in the system database.

[0099] Compared with traditional static or roughly delineated safety areas, the present invention achieves a more precise and customizable virtual boundary definition, which can reach the centimeter level.

[0100] This significantly improves the safety of the working environment and reduces the risk of accidents caused by accidentally entering a danger zone.

[0101] Step 2: Position tracking of the monitored object

[0102] Sensor technology: Each worker is equipped with a safety helmet with an RFID tag, and lidar and high-definition cameras are deployed around the substation to achieve all-round monitoring.

[0103] RFID tags are used for near-field positioning, lidar provides precise distance measurement, and cameras are used for visual confirmation.

[0104] Data fusion: Use Kalman filters to integrate multi-source data from RFID, LiDAR, and HD cameras to improve the accuracy and reliability of position estimation.

[0105] The specific steps include:

[0106] Step 21, equip all operators with RFID tags.

[0107] Step 22, deploy laser radar and high-definition cameras to cover the entire working area.

[0108] Step 23, using the Kalman filter algorithm to fuse multiple sensor information to improve positioning accuracy, specifically includes the following sub-steps:

[0109] Step 231, obtaining relative position information provided by the RFID tag, distance information provided by the laser radar, and visual positioning information provided by the image captured by the high-definition camera;

[0110] Step 232, defining a system state vector X, wherein the state vector X includes the position coordinates of the person (e.g., three-dimensional coordinates in a spatial coordinate system); optionally, it may also include speeds caused by changes in position (e.g., different speeds in different directions, i.e., x, y, and z directions in the spatial coordinate system);

[0111] Step 233, establish a nonlinear dynamic model describing the movement of personnel as follows:

[0112] x k+1 =f(x k ,u k )+w k

[0113] Where k is a time step, which represents a discrete time point at which the system estimates and updates its state. At each time step k, the estimate of the person's position is updated based on the state estimate at the previous time step k-1 and the current measurement data.

[0114] u k represents the change in velocity, i.e., acceleration, which serves as the control input of the model;

[0115] w k Represents process noise caused by position changes of personnel;

[0116] x0 represents the initial state, which includes the initial position coordinates and speed of the person;

[0117] Step 234, establish a corresponding nonlinear dynamic model for each sensor, wherein:

[0118] The model corresponding to RFID is:

[0119] z RFID =h RFID (x k )+v RFID

[0120] The model corresponding to the laser radar is:

[0121] z Lidar =h Lidar (x k )+v Lidar

[0122] The model corresponding to the HD camera is:

[0123] z Camera =h Camera (x k )+v Camera

[0124] Among them, z with different subscripts represents the corresponding actual measurement value, V with different subscripts represents the corresponding measurement noise, h with different subscripts represents the corresponding nonlinear measurement model; z with different subscripts is combined to form a measurement vector Z;

[0125] Among them, the nonlinear measurement model h RFID (x k ) is determined by: deploying multiple RFID tags at known locations, recording the RSSI values ​​of the wireless signal strength at different locations, and fitting a formula between the location of the target person and the RSSI;

[0126] Nonlinear measurement model h Lidar (x k ) is determined by: fitting the distance between the position of the target person and the laser radar using a triangulation method;

[0127] Nonlinear measurement model h Camera (x k ) is determined by: first, calibrating the internal parameters (including focal length, principal point) and external parameters (position and posture relative to the world coordinate system) of the camera; second, extracting the key feature points of the target person from the image captured by the camera; third, using deep learning to identify and track the key feature points of the target person to determine a nonlinear measurement model of the position of the target person and the image captured by the camera;

[0128] Among them, the above different subscripts v represent the measurement noise and w k The process noise is determined as follows: the noise is assumed to be zero-mean Gaussian white noise and is estimated from experimental data;

[0129] Step 24, predicting the state at the next moment using the current state and control input according to the nonlinear dynamic model described in step 233, specifically includes:

[0130] Step 241, predict the time step k=1 based on the nonlinear dynamic model and the initial state x0 Right now as follows:

[0131]

[0132] Where u0 is the control input at the first time step, i.e., k=0, and w0 is the process noise at the first time step, i.e., k=0, which is assumed to be zero-mean Gaussian white noise;

[0133] Process noise w k The noise covariance matrix is ​​denoted as Q k , where the noise covariance matrix is ​​used to describe the uncertainty of the nonlinear dynamics model;

[0134] When k = 0, the corresponding noise covariance matrix is ​​Q0;

[0135] State vector X k The state covariance matrix of k , where the state covariance matrix is ​​used to describe the uncertainty of state estimation;

[0136] For the initial state X0, the corresponding initial state covariance matrix P 0|0 , where the initial state covariance matrix P 0|0 represents a measure of the uncertainty in the initial state estimate and is set as follows:

[0137] When there is reliable prior information (such as initial position data obtained by sensors or manually input), the reliable prior information is directly used as X0; and the initial state covariance matrix P 0|0 The values ​​in are set to smaller values ​​to indicate lower uncertainty;

[0138] If there is no reliable prior information, the predicted value of the initial state is based on the empirical value. The initial state X0 is set based on this (for example, assuming that the person starts from a known position and is stationary); the initial state covariance matrix P 0|0 The values ​​in are set to larger values ​​to indicate higher uncertainty;

[0139] Step 242, further update the state covariance matrix P 1|0 :

[0140] P 1|0 =F0P 0|0 F0 T +Q0

[0141] Among them, F0 is the Jacobian matrix of f to the state vector X, which is:

[0142]

[0143] Step 243, obtain the first measurement value Z1 when k=1, and update the following variables:

[0144] Known Kalman gain K k The following formula:

[0145] K k =P k|k-1 H k T (H k P k|k-1 H k T +Rk ) -1

[0146] Among them, R k is the covariance matrix of the measurement noise, which is used to describe the uncertainty of the measurement; it should be noted that, similar to the various noises mentioned above in the present invention, the measurement noise is also assumed to be zero-mean Gaussian white noise and is estimated through experimental data;

[0147] P k|k-1 represents the state covariance matrix of the prediction phase from time step k-1 to k, and has:

[0148] P k|k-1 =F k P k-1|k-1 F k T +Q k-1

[0149] Among them, P k-1|k-1 is the optimal state covariance matrix of the previous time step k-1,

[0150] Q k-1 is the covariance matrix of the process noise at k-1;

[0151] F k is the Jacobian matrix of f for the state vector X, which is:

[0152]

[0153] in, is the optimal state estimate at the previous time step k-1;

[0154] H k is the Jacobian matrix of h for the state vector X, which is:

[0155]

[0156] It should be noted that the optimal state estimate for the kth step is calculated as follows:

[0157]

[0158] And there is,

[0159] Among them, z k Represents the actual measurement value of the current k-th step;

[0160] According to the predicted state Calculated expected measurements;

[0161] The update formula of the known state covariance matrix is ​​as follows:

[0162] P k|k =(IK k H k ) k|k-1

[0163] Where I represents the identity matrix,

[0164] According to the above formula, the Kalman gain K1 is further calculated:

[0165] K1=P 1|0 H1 T (H1P 1|0 H1 T +R1) -1

[0166] And further update the state estimate:

[0167]

[0168] And further update the state covariance matrix:

[0169] P 1|1 =(I-K1H1)P 1|0

[0170] Step 244, continue iterative calculation based on the above, use the Kalman gain, optimal state estimate, and state covariance matrix of the previous time step to continuously predict the state of the next time step, and continuously update the Kalman gain, state estimate, and state covariance matrix, so as to iteratively update the estimate of the personnel position at each time step, thereby achieving high-precision position tracking.

[0171] The above specific steps combine different types of sensors and perform data fusion through a fusion algorithm, which improves the accuracy of position tracking, reduces the possibility of false alarms, and increases the reliability of the system compared to a single technology.

[0172] Step 3, boundary detection algorithm, includes the following specific steps:

[0173] Distance calculation: Calculate the shortest distance between the operator and the virtual boundary in real time according to step 2 above.

[0174] Threshold judgment and warning: Judgment and warning are made based on two pre-set warning thresholds: for example, 10 meters is used as the first threshold for issuing a preliminary warning, and 5 meters is used as the second threshold for issuing an emergency warning. It is judged whether the calculated shortest distance reaches the threshold range and different levels of warnings (such as the preliminary warning and emergency warning) are issued.

[0175] Among them, the threshold can be adjusted dynamically: according to the real-time weather data and the operating status of the substation, the warning threshold is dynamically adjusted, for example, the threshold is increased under severe weather conditions.

[0176] Step 3 above innovatively introduces the function of dynamically adjusting the warning threshold, which enhances the system's ability to adapt to environmental changes and makes safety management measures more flexible and effective;

[0177] Step 4: Early warning alarm mechanism

[0178] Warning level: divided into yellow warning (preliminary warning) and red warning (emergency warning).

[0179] Alarm triggering: When the monitored object enters the warning area, the system automatically triggers an alarm of the corresponding level.

[0180] Feedback method: through the vibration of the smart bracelet worn by the operator, the warning sound played by the on-site speaker, and the push notification of information on the mobile device.

[0181] The specific steps include:

[0182] Step 41, based on the result obtained in step 3, triggering an alarm of a corresponding level when the staff approaches the warning area;

[0183] Step 42, notifying relevant personnel through wristband vibration, sound prompts, etc.;

[0184] Step 43: Display the alarm information in the control center and record the event details.

[0185] The above ensures the effectiveness of information communication through various forms of instant feedback mechanisms, thereby quickly responding to potential threats and preventing dangerous behaviors from occurring in a timely manner.

[0186] Step 5, user interaction, includes the following specific steps:

[0187] Visual display: Display real-time maps in the field control room and on mobile devices, marking the location of operators and virtual boundaries;

[0188] Operational Control: Operators in the control room can adjust virtual boundary parameters through the interface, such as temporarily expanding or contracting the boundary and temporarily disabling alarms.

[0189] Step 6: Data recording and analysis

[0190] Event log: Record the time, location, operator ID and event details of each alarm trigger.

[0191] Data analysis: Analyze alarm frequency and trigger conditions, optimize virtual boundary parameters and alarm mechanisms, and improve system performance.

[0192] Step 7, Security and Privacy

[0193] Data encryption: All transmitted data uses the AES-256 encryption standard to ensure the security of data during transmission.

[0194] Privacy protection: When processing personal location data, strictly comply with data protection regulations such as GDPR to ensure that personal information is not abused.

[0195] The present invention is not limited to the above-mentioned specific implementation modes. A person skilled in the art may implement the present invention in various other specific implementation modes according to the embodiments and the disclosure of the drawings. Therefore, any design that adopts the design structure and concept of the present invention and makes some simple transformations or changes shall fall within the scope of protection of the present invention.

Claims

1. A virtual space boundary early warning alarm method, characterized in that: The virtual space boundary early warning alarm method comprises the following steps: Step 1: Define the virtual boundary with the power facility as the center; Step 2: Tracking the location of the monitored object in the monitoring area; Step 3: Based on the boundary detection algorithm, determine the shortest distance between the monitored object and the virtual boundary; Step 4: Issue an early warning and an alarm based on the shortest distance between the monitored object and the virtual boundary.

2. The virtual space boundary early warning alarm method according to claim 1, characterized in that: In step 1, the following steps are also included: Step 11, using high-precision GPS and RTK technology to determine the center point of the power facility; Step 12, according to safety regulations and technical requirements, the virtual boundary is set to a basic circular boundary with a radius of 5 meters; Step 13, for substations with complex shapes, use computer-aided design tools to draw composite boundaries; Step 14, digitize the boundary data and store it in the system database.

3. The virtual space boundary early warning alarm method according to claim 1, characterized in that: In step 2, the following steps are also included: Step 21, equipping the monitored object with an RFID tag; Step 22, monitoring the monitoring area based on laser radar and high-definition camera; Step 23, using Kalman filter algorithm to fuse multiple sensor information to establish a nonlinear dynamic model; Step 24, according to the nonlinear dynamics model, the state at the next moment is predicted using the current state and the control input to calculate the shortest distance between the monitored object and the virtual boundary.

4. The virtual space boundary early warning alarm method according to claim 1, characterized in that: In step 23, the following steps are also included: Step 231, obtaining relative position information provided by the RFID tag, distance information provided by the laser radar, and visual positioning information provided by the image captured by the high-definition camera; Step 232, defining a system state vector X, wherein the state vector X includes the position coordinates of the person and the speed caused by the change in position; Step 233, establish a nonlinear dynamic model f describing the movement of personnel as follows: x k+1 =f(x k ,u k )+w k Where k is a time step, which represents a discrete time point at which the system estimates and updates its state. At each time step k, the estimate of the person's position is updated based on the state estimate at the previous time step k-1 and the current measurement data; u k represents the change in velocity, i.e., acceleration, which serves as the control input of the model; w k represents the process noise caused by the position change of personnel; x k represents the position coordinates and velocity of the person in k time step; x k+1 Indicates the position coordinates and velocity of the person in k+1 time step; Step 234, establish corresponding nonlinear models for RFID, laser radar and high-definition camera respectively, wherein the model corresponding to RFID is: z RFID =h RFID (x k )+v RFID z RFID is the actual measurement value of RFID, v RFID is the RFID measurement noise, h RFID (x k ) is the RFID nonlinear measurement model; The model corresponding to the laser radar is: z Lidar =h Lidar (x k )+v Lidar z Lidar is the actual measurement value of the laser radar, v Lidar is the lidar measurement noise, h Lidar (x k ) is the nonlinear measurement model of laser radar; The model corresponding to the HD camera is: z Camera =h Camera (x k )+v Camera z Camera is the actual measurement value of the high-definition camera, v Camera Measuring noise for high-definition cameras, h Camera (x k ) is the nonlinear measurement model of the high-definition camera.

5. The virtual space boundary early warning alarm method according to claim 4, characterized in that: In step 234, the RFID nonlinear measurement model h RFID (x k ) is determined as follows: Deploy multiple RFID tags at known locations, record the RSSI values ​​of wireless signal strength at different locations, and fit the formula between the location of the target person and RSSI; Lidar nonlinear measurement model h Lidar (x k ) is determined as follows: Use triangulation to fit the distance between the target person's position and the lidar; Nonlinear measurement model of high-definition camera Camera (x k ) is determined as follows: First, the camera's internal and external parameters are calibrated; the internal parameters include focal length and principal point, and the external parameters include position and posture relative to the world coordinate system; Secondly, extract the key feature points of the target person from the image captured by the camera; Again, deep learning is used to identify and track the key feature points of the target person to determine the nonlinear measurement model of the target person's position and the image captured by the camera.

6. The virtual space boundary early warning alarm method according to claim 4, characterized in that: In step 24, the following steps are also included: Step 241, predict the time step k=1 according to the nonlinear dynamic model and the initial state x0 Right now as follows: Among them, u0 is the control input at the first time step, that is, k = 0, w0 is the process noise at the first time step, that is, k = 0, which is assumed to be zero-mean Gaussian white noise; x0 represents the initial state; Process noise w k The noise covariance matrix is ​​denoted as Q k , where the noise covariance matrix is ​​used to describe the uncertainty of the nonlinear dynamics model; When k = 0, the corresponding noise covariance matrix is ​​Q0; State vector X k The state covariance matrix of k , where the state covariance matrix is ​​used to describe the uncertainty of state estimation; Step 242, further update the state covariance matrix P 1|0 : P 1|0 =F0P 0|0 F0 T +Q0 Among them, P 0|0 is the initial state covariance matrix, F0 is the Jacobian matrix of f to the state vector X, and we have: Step 243, obtain the first measurement value Z1 when k=1, and update the following variables: Known Kalman gain K k The following formula: K k =P k|k-1 H k T (H k P k|k-1 H k T +R k ) -1 Among them, R k is the covariance matrix of the measurement noise, which is used to describe the uncertainty of the measurement; P k|k-1 represents the state covariance matrix of the prediction phase from time step k-1 to k, and has: P k|k-1 =F k P k-1|k-1 F k T +Q k-1 Among them, P k-1|k-1 is the optimal state covariance matrix of the previous time step k-1, Q k-1 is the covariance matrix of the process noise at k-1; F k is the Jacobian matrix of f for the state vector X, which is: in, is the optimal state estimate at the previous time step k-1; H k is the Jacobian matrix of h for the state vector X, which is: For the optimal state estimate at step k, it is calculated as follows: Among them, z k Represents the actual measurement value of the current k-th step; According to the predicted state Calculated expected measurements; The update formula of the known state covariance matrix is ​​as follows: P k|k =(I-K k H k )P k|k-1 Where I represents the identity matrix, and the Kalman gain K1 is further calculated: K1=P 1|0 H1 T (H1P 1|0 H1 T +R1) -1 And further update the state estimate: Further update the state covariance matrix: P 1|1 =(I-K1H1)P 1|0 Step 244, continue iterative calculation based on the above, use the Kalman gain, optimal state estimate, and state covariance matrix of the previous time step to continuously predict the state of the next time step, and continuously update the Kalman gain, state estimate, and state covariance matrix, so as to iteratively update the estimate of the position of the monitored object at each time step, thereby achieving high-precision position tracking.

7. The virtual space boundary early warning alarm method according to claim 6, characterized in that: In step 241, for the initial state X0, its corresponding initial state covariance matrix P 0|0 , where the initial state covariance matrix P 0|0 represents a measure of the uncertainty in the initial state estimate and is set as follows: When reliable prior information is available, the reliable prior information is directly used as X0; and the initial state covariance matrix P 0|0 The values ​​in are set to smaller values ​​to indicate lower uncertainty; If there is no reliable prior information, the predicted value of the initial state is based on the empirical value. And use this to set the initial state X0; and the initial state covariance matrix P 0|0 The values ​​in are set to larger values ​​to indicate higher uncertainty.

8. The virtual space boundary early warning alarm method according to claim 1, characterized in that: The virtual space boundary early warning alarm method further comprises the following steps: Step 5: Display a real-time map in the on-site control room and on mobile devices, marking the location of the operators and the virtual boundary; the operator in the control room can adjust the virtual boundary parameters through the interface.

9. The virtual space boundary early warning alarm method according to claim 1, characterized in that: The virtual space boundary early warning alarm method further comprises the following steps: Step 6: Record the time, location, operator ID and event details of each alarm trigger; Analyze alarm frequency and trigger conditions, optimize virtual boundary parameters and alarm mechanisms, and improve system performance.

10. The virtual space boundary early warning alarm method according to claim 1, characterized in that: The virtual space boundary early warning alarm method further comprises the following steps: Step 7: The data transmission adopts AES-256 encryption standard to ensure the security of data during transmission.

Images