Intelligent transfer bed boundary crossing alarm method and system

Abstract

The application provides a kind of intelligent transfer bed boundary alarm method and system, the method includes obtaining alternative AP signal data set, according to signal strength, AP physical position and historical signal strength statistical characteristics calculate the source selection weight of each alternative AP, and the confidence of last monitoring period positioning result is adjusted AP subset scale, constructs current AP subset;When meeting update trigger condition, select replacement AP and set observation period, in observation period and parallel computing include and do not include replacement AP positioning coordinates, compare two kinds of positioning coordinates with the extended prediction position obtained based on historical trajectory, and adjust the contribution weight of replacement AP according to this;Based on the signal fingerprint of updated AP subset and contribution weight, calculate the real-time positioning coordinate and positioning result confidence of transfer bed;Again, combined with real-time positioning coordinate and historical movement trajectory, generate comprehensive boundary judgment index, and use alarm threshold associated with positioning result confidence to make boundary alarm judgment.

Description

Technical Field

[0001] This application belongs to the field of alarms, and in particular relates to a method and system for alarming intelligent transfer beds to cross boundaries. Background Technology

[0002] In the daily operations of modern medical and rehabilitation institutions, location management and activity range monitoring of mobile medical equipment such as transport beds are crucial aspects of asset management. Existing systems typically employ indoor positioning technology based on wireless access point (AP) signals. By receiving signals from surrounding APs and calculating coordinates, they determine whether the equipment has strayed outside the designated activity area. However, indoor environments are susceptible to interference from obstructions, personnel movement, and other radio frequency devices, which can easily cause multipath effects and fading in wireless signals, leading to fluctuations in positioning results. As the transport bed moves, the subset of APs used for positioning also needs to be updated. Directly replacing positioning nodes can easily cause abrupt changes in positioning coordinates or trajectory drift, affecting the continuity of the positioning trajectory.

[0003] Indoor positioning accuracy varies with region and time, and the confidence level of each positioning result is not the same. If a fixed alarm threshold is used, false alarms or missed alarms are likely to occur when the positioning error is large. In addition, judging boundary crossings based solely on a single instantaneous coordinate point is insufficient to reflect the continuous movement trajectory and trend of the transport bed, resulting in insufficient alarm reliability. Therefore, it is necessary to propose a boundary crossing alarm mechanism. This mechanism improves the smoothness and stability of positioning calculations by optimizing the AP node update process, and dynamically adjusts the alarm judgment conditions based on movement trajectory characteristics and positioning confidence level, thereby achieving reasonable monitoring of the transport bed's boundary crossing behavior. Summary of the Invention

[0004] To address the problems in existing technologies, such as direct updates of AP nodes leading to sudden changes in positioning trajectories, fixed alarm thresholds being unable to adapt to fluctuations in positioning reliability, and insufficient reliability of instantaneous coordinate boundary crossing judgments, this invention provides an intelligent transfer bed boundary crossing alarm method and system.

[0005] In a first aspect, the present invention provides an intelligent transfer bed boundary crossing alarm method, comprising the following steps: Obtain candidate AP signal dataset; calculate source selection weight for each candidate AP based on the signal strength in the signal dataset, the preset physical location of each AP, and the stored historical signal strength statistical characteristics; adjust the number N of members in the current AP subset based on the confidence level of the positioning result obtained in the previous monitoring cycle to form an AP subset; An update trigger condition is set. When the update trigger condition is met, an AP to be replaced is determined from the current AP subset, and a replacement AP is selected from the candidate APs outside the AP subset. The replacement AP and the reserved APs in the current AP subset (excluding the AP to be replaced) are combined into an observation subset, and an observation period is set. During the observation period, a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP are calculated. The first and second positioning coordinates are compared with the extended predicted positions obtained based on historical movement trajectories, and the contribution weight of the replacement AP is determined based on the comparison results. After the observation period ends, it is determined whether the replacement AP should be formally included in the AP subset to replace the AP to be replaced based on the contribution weight. Using the updated AP signal fingerprint data within the AP subset and combining it with the contribution weight, the real-time positioning coordinates and positioning result confidence level of the transfer bed are calculated; a comprehensive boundary crossing judgment index is generated based on the real-time positioning coordinates and historical movement trajectory of the transfer bed; an alarm threshold associated with the positioning result confidence level is set, and an alarm is triggered when the alarm threshold is exceeded.

[0006] Optionally, the step of calculating the source selection weight for each candidate AP based on the signal strength in the signal dataset, the preset physical location of each AP, and the stored historical signal strength statistical characteristics includes: Read the current received signal strength value of the candidate APs in the signal data set; Extract the historical average intensity of the corresponding candidate AP from the stored historical signal strength statistical features; Extract the historical positioning coordinates of one monitoring cycle on the transport bed, and calculate the straight-line spatial distance between the historical positioning coordinates and the preset corresponding candidate AP physical location; The first value is obtained by adding the absolute value of the historical average signal strength to the absolute value of the difference between the current received signal strength and the historical average signal strength. Then add the linear spatial distance to a preset first distance buffer constant with a unit of length to obtain the second value; For each candidate AP, the first value and the second value are multiplied to obtain a comprehensive denominator, and one is divided by the comprehensive denominator to obtain the source selection weight of the corresponding candidate AP.

[0007] Optionally, adjusting the number N of members in the current AP subset based on the confidence level of the positioning result calculated in the previous monitoring cycle includes: A first fixed confidence level lower limit parameter and a second fixed confidence level lower limit parameter are preset, wherein the first fixed confidence level lower limit parameter is less than the second fixed confidence level lower limit parameter; When the confidence value of the positioning result calculated in the previous monitoring cycle is less than the first fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 7. When the confidence value of the positioning result calculated in the previous monitoring cycle is not less than the first fixed confidence lower limit parameter and is less than the second fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 5. When the confidence value of the positioning result calculated in the previous monitoring cycle is not less than the second fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 4.

[0008] Optionally, when the update triggering condition is met, determining the APs to be replaced from the current AP subset and selecting replacement APs from the candidate APs outside the AP subset includes: Obtain all remaining candidate APs that are not included in the AP subset, and establish a replacement candidate set; The APs to be replaced are determined based on the source selection weights of each AP in the current AP subset. The AP with the lowest source selection weight is determined as the AP to be replaced, and the APs in the current AP subset other than the AP to be replaced are reserved APs. Extract the physical location coordinate parameters of each candidate AP in the replacement candidate set; The physical location coordinate parameters of each candidate AP in the replacement candidate set are combined with the physical location coordinate parameters of the retained AP to form multiple independent test geometry matrices; Read the historical positioning coordinates of the transfer bed for one monitoring cycle and use them as the reference position for geometric calculation; Based on the reference position and the two-dimensional planar physical position coordinates of each AP, the test geometry matrix is ​​linearized to obtain the corresponding geometric accuracy attenuation factor; Extract the minimum value among all calculated geometric precision attenuation factors, and designate the candidate AP associated with the test geometry matrix corresponding to the minimum value as the replacement AP.

[0009] Optionally, the step of forming a review subset by combining the replacement AP with the retained APs in the current AP subset excluding the AP to be replaced, and setting a review period; during the review period, calculating a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP, comparing the first and second positioning coordinates with the extended predicted positions obtained based on historical movement trajectories, and determining the contribution weight of the replacement AP based on the comparison results, including: The observation period is defined as the time span of three consecutive data sampling periods. The coordinates of the five most recent historical movement trajectories of the transfer bed are processed by calling the first-order Kalman filter equation to calculate the specific coordinates of the extended predicted position. Calculate the first Euclidean distance from the first positioning coordinates generated by the replacement AP to the extended predicted position, and the second Euclidean distance from the second positioning coordinates not generated by the replacement AP to the extended predicted position. The distance improvement ratio is calculated based on the first Euclidean distance and the second Euclidean distance, wherein the distance improvement ratio is the reduction ratio of the first Euclidean distance to the second Euclidean distance; When the distance improvement ratio is greater than zero, the contribution weight of the replacement AP is calculated based on the initial fixed constant weight of the replacement AP and the distance improvement ratio; When the distance improvement ratio is not greater than zero, the contribution weight of the replaced AP is set to a preset lower limit value.

[0010] Optionally, the step of using the AP signal fingerprint data within the updated AP subset and combining it with the contribution weight to calculate the real-time positioning coordinates and positioning result confidence of the transfer bed includes: The signal strength of the corresponding AP is extracted using the updated AP subset to form an online fingerprint feature vector, and the contribution weight of each AP in the subset is obtained. Call the offline fingerprint database to calculate the normalized weighted Manhattan distance between the online fingerprint feature vector and the corresponding AP signal strength component in the fingerprint vector of each pre-entered reference point. The normalized weighted Manhattan distance is obtained by normalizing the absolute value of the signal strength difference of the corresponding AP, multiplying it by the corresponding contribution weight, and summing the results. The contribution weight of the non-replacement AP is a preset constant value. Extract the first four reference points and their grid coordinates in ascending order of the normalized weighted Manhattan distance; The position weight of each reference point is obtained by normalizing the inverse of the distances corresponding to the first four reference points, and the grid coordinates of each reference point are weighted and summed to obtain the real-time positioning coordinates of the transfer bed. Normalization mapping is performed based on the reciprocal of the root mean square of the distances to the first four reference points, and the mapping result is restricted to between 0 and 1 to obtain the confidence level of the positioning result.

[0011] Optionally, the step of generating a comprehensive boundary crossing judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed includes: Retrieve the polyline coordinate dataset of the closed boundary of the limited activity area of ​​the transfer bed, and determine whether the transfer bed is located within the limited activity area based on its real-time positioning coordinates; if it is located outside the limited activity area, the boundary crossing alarm condition is met. If it is located within a limited activity area, a distance risk item is generated based on the positive difference between the maximum allowable distance of the preset alarm limit and the shortest distance from the real-time positioning coordinates to the nearest closed boundary line. A trend velocity vector is generated based on the historical movement trajectory coordinates of the transfer bed and the latest real-time positioning coordinates, and a trend risk term is generated based on the positive value of the direction cosine between the trend velocity vector and the outer normal vector of the nearest boundary. The distance risk item and the trend risk item are weighted and summed according to preset weights to obtain a comprehensive boundary judgment index.

[0012] Optionally, setting the alarm threshold associated with the confidence level of the positioning result includes: Recall the preset warning values ​​stored within the control system; Read the actual measured value of the confidence level of the acquired positioning result, and add a preset small positive number to the measured value to obtain the adjustment factor; Multiply the preset warning value by the adjustment factor to obtain the real-time alarm threshold corresponding to the comprehensive boundary judgment index, and use the real-time alarm threshold to update the fixed value trigger limit in the main control judgment execution logic.

[0013] Secondly, the present invention also provides an intelligent transfer bed boundary crossing alarm system, comprising the following modules: The calculation module is used to acquire candidate AP signal datasets; calculate source selection weights for each candidate AP based on the signal strength in the signal dataset, the preset physical locations of each AP, and the stored historical signal strength statistical characteristics; and adjust the number N of members in the current AP subset based on the confidence level of the positioning results calculated in the previous monitoring cycle to form an AP subset. An adjustment module is used to set update trigger conditions. When the update trigger conditions are met, APs to be replaced are determined from the current AP subset, and replacement APs are selected from the candidate APs outside the AP subset. The replacement APs are combined with the reserved APs in the current AP subset (excluding the APs to be replaced) to form an observation subset, and an observation period is set. During the observation period, a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP are calculated. The first and second positioning coordinates are compared with the extended predicted positions obtained based on historical movement trajectories, and the contribution weight of the replacement AP is determined based on the comparison results. After the observation period ends, it is determined whether to formally include the replacement AP in the AP subset to replace the APs to be replaced based on the contribution weight. The generation module is used to calculate the real-time positioning coordinates and positioning result confidence of the transfer bed by using the AP signal fingerprint data in the updated AP subset and combining it with the contribution weight; generate a comprehensive boundary judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed; and set an alarm threshold associated with the positioning result confidence, triggering an alarm when the alarm threshold is exceeded.

[0014] Preferably, the step of calculating the source selection weight for each candidate AP based on the signal strength in the signal dataset, the preset physical location of each AP, and the stored historical signal strength statistical characteristics includes: Read the current received signal strength value of the candidate APs in the signal data set; Extract the historical average intensity of the corresponding candidate AP from the stored historical signal strength statistical features; Extract the historical positioning coordinates of one monitoring cycle on the transport bed, and calculate the straight-line spatial distance between the historical positioning coordinates and the preset corresponding candidate AP physical location; The first value is obtained by adding the absolute value of the historical average signal strength to the absolute value of the difference between the current received signal strength and the historical average signal strength. Then add the linear spatial distance to a preset first distance buffer constant with a unit of length to obtain the second value; For each candidate AP, the first value and the second value are multiplied to obtain a comprehensive denominator, and one is divided by the comprehensive denominator to obtain the source selection weight of the corresponding candidate AP.

[0015] The present invention has the following beneficial effects: The source selection weight is calculated by combining signal strength, AP physical location and historical signal statistical characteristics, and the AP subset size is adjusted according to the positioning reliability of the previous cycle, so as to balance positioning reliability and computational efficiency. When updating the AP subset, a replacement AP observation period is introduced. During the observation period, the positioning coordinates including and not including the replacement AP are calculated in parallel and compared with the predicted positions obtained based on historical trajectories. The contribution weight of the replacement AP is adjusted accordingly to reduce the impact of unstable new signal sources on the positioning results and improve the continuity of the positioning trajectory. By combining real-time coordinates and historical movement trajectories, a comprehensive boundary crossing judgment index is generated, and the alarm threshold is dynamically adjusted based on the location reliability to reduce false alarms or missed alarms caused by environmental signal fluctuations and improve the reliability of the transfer bed boundary crossing alarm. Attached Figure Description

[0016] Figure 1 A flowchart of an intelligent transfer bed boundary crossing alarm method; Figure 2 Flowchart for selecting a subset of candidate APs. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] Example 1 In Embodiment 1 of the present invention, a method for providing an intelligent transfer bed boundary crossing alarm is provided, such as... Figure 1 As shown, it includes the following steps: S1, obtain candidate AP signal dataset; calculate source selection weight for each candidate AP based on the signal strength in the signal dataset, the preset physical location of each AP, and the stored historical signal strength statistical characteristics; adjust the number N of members in the current AP subset based on the confidence level of the positioning result obtained in the previous monitoring cycle to form an AP subset.

[0019] The wireless communication module on the transport bed scans surrounding wireless access points according to a preset sampling frequency, collecting the medium access control address and received signal strength indication value of each available wireless access point to form a candidate access point signal dataset, i.e., a candidate AP signal dataset. AP stands for wireless access point, and AP signal refers to the wireless signal transmitted by the wireless access point and received by the device. The signal strength can be used to help determine the location of the device.

[0020] The server extracts the signal strength values ​​of each access point in the dataset, and combines them with the historical signal strength statistical characteristics of the corresponding access point and the pre-stored two-dimensional plane physical location coordinates to calculate the source selection weight for evaluating the current positioning applicability of the access point.

[0021] Then, the confidence level of the positioning results from the previous monitoring period is extracted. The number N of members in the current AP subset is determined based on the first and second fixed confidence level parameters. When the positioning result confidence level is less than the first fixed confidence level parameter, N is set to 7; when the positioning result confidence level is not less than the first fixed confidence level parameter and is less than the second fixed confidence level parameter, N is set to 5; when the positioning result confidence level is not less than the second fixed confidence level parameter, N is set to 4. Based on the calculated source selection weights, all candidate access points are sorted in descending order, and the top N candidate access points are selected to form the access point subset for the current period. The candidate AP subset selection process is as follows: Figure 2 As shown.

[0022] In one embodiment, the step of calculating the source selection weight for each candidate AP based on the signal strength in the signal dataset, the preset physical locations of each AP, and the stored historical signal strength statistical characteristics includes: Read the current received signal strength value of the candidate APs in the signal data set; Extract the historical average intensity of the corresponding candidate AP from the stored historical signal strength statistical features; Extract the historical positioning coordinates of one monitoring cycle on the transport bed, and calculate the straight-line spatial distance between the historical positioning coordinates and the preset corresponding candidate AP physical location; The first value is obtained by adding the absolute value of the historical average signal strength to the absolute value of the difference between the current received signal strength and the historical average signal strength. Then add the linear spatial distance to a preset first distance buffer constant with a unit of length to obtain the second value; For each candidate AP, the first value and the second value are multiplied to obtain a comprehensive denominator, and one is divided by the comprehensive denominator to obtain the source selection weight of the corresponding candidate AP.

[0023] The received signal strength value S of a candidate AP in the current monitoring period is obtained by real-time scanning. It is assumed that the signal strength S read after smoothing filtering is -68dBm. Extract the historical intensity mean μ of the candidate AP from the stored historical statistical feature data table. For example, the extracted μ value is -60dBm.

[0024] The historical two-dimensional positioning coordinates (12.5, 8.4) meters of the transfer bed in the previous monitoring cycle were retrieved. Combined with the pre-entered physical location (15.0, 10.0) meters of the candidate AP, the straight-line spatial distance d from the historical positioning coordinates to the target candidate AP was calculated to be approximately 2.96 meters according to the Euclidean distance formula.

[0025] The absolute value of the extracted historical average signal strength -60dBm is taken to obtain a constant of 60. The absolute value of the difference between the current received signal strength S and the historical average μ is calculated, resulting in 8. Adding 60 and 8 yields the first value of 68. The linear spatial distance d, 2.96 meters, is then compared with the preset first constant parameter. Assuming Setting the buffer distance to 1.0 meter and adding the effects of zero-depletion risk and near-field abrupt changes, the second value is 3.96. Multiplying the calculated first value 68 and the second value 3.96, the denominator of the comprehensive effect is 269.28. Dividing 1 by 269.28, the source selection weight W of the candidate AP is 0.0037.

[0026] In one embodiment, adjusting the number N of members in the current AP subset based on the confidence level of the positioning result calculated in the previous monitoring cycle includes: A first fixed confidence level lower limit parameter and a second fixed confidence level lower limit parameter are preset, wherein the first fixed confidence level lower limit parameter is less than the second fixed confidence level lower limit parameter; When the confidence value of the positioning result calculated in the previous monitoring cycle is less than the first fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 7. When the confidence value of the positioning result calculated in the previous monitoring cycle is not less than the first fixed confidence lower limit parameter and is less than the second fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 5. When the confidence value of the positioning result calculated in the previous monitoring cycle is not less than the second fixed confidence lower limit parameter, the number of members N of the current AP subset is set to the value 4.

[0027] In the initial state, the first fixed confidence level lower limit parameter is preset to 0.6 in the program configuration, and the second fixed confidence level lower limit parameter is preset to 0.85.

[0028] When starting the positioning calculation in each monitoring cycle, the confidence value of the positioning result calculated in the previous monitoring cycle is read first. Assuming that the confidence value of the positioning result read in the previous monitoring cycle is 0.42, since this value is less than the first fixed confidence lower limit parameter of 0.6, it is diagnosed that the current environmental interference is strong, and the number of members N of the current AP subset is set to the value of 7.

[0029] If the confidence level of the location result read from the previous monitoring period is 0.75, which satisfies the condition 0.6≤0.75<0.85, that is, not less than the first fixed confidence level lower limit parameter and less than the second fixed confidence level lower limit parameter, the number of members N of the current AP subset is set to the value 5.

[0030] If the confidence level of the positioning result read from the previous monitoring period is 0.92, which is not less than the second fixed confidence level lower limit parameter of 0.85, it indicates that the positioning environment signal is good, and the number of members N of the current AP subset is set to the value 4.

[0031] S2, set update trigger conditions. When the update trigger conditions are met, determine the APs to be replaced from the current AP subset and select a replacement AP from the candidate APs outside the AP subset. Combine the replacement AP with the reserved APs in the current AP subset excluding the AP to be replaced to form an observation subset, and set an observation period. During the observation period, calculate the first positioning coordinates containing the replacement AP and the second positioning coordinates not containing the replacement AP. Compare the first and second positioning coordinates with the extended predicted positions obtained based on historical movement trajectories, and determine the contribution weight of the replacement AP based on the comparison results. After the observation period ends, determine whether to formally include the replacement AP in the AP subset to replace the AP to be replaced based on the contribution weight.

[0032] The update trigger condition is set as follows: the number of APs in the current AP subset whose current received signal strength is lower than a set lower limit exceeds a proportional threshold, or the confidence level of the positioning results in the most recent monitoring periods shows a downward trend; wherein, the set lower limit is preferably -75dBm, the proportional threshold is preferably half the number of members in the current AP subset, and the most recent monitoring periods are preferably the most recent three monitoring periods; when the confidence level of the positioning results in the most recent three monitoring periods decreases sequentially, it is determined that the confidence level of the positioning results shows a downward trend.

[0033] When any update trigger condition is met, the AP with the lowest source selection weight in the current AP subset is determined as the AP to be replaced, and the APs other than the AP to be replaced are reserved APs; then, from the remaining candidate APs outside the current AP subset, a replacement AP is selected according to the geometric precision attenuation factor, and the replacement AP and the reserved AP form an examination subset, and a series of consecutive monitoring periods are set as the examination period.

[0034] During each monitoring cycle within the observation period, a weighted nearest neighbor algorithm is used to perform two-way positioning calculations in parallel: the first path calculates the first positioning coordinates based on the observed subset containing the replaced APs, and the second path calculates the second positioning coordinates based on the retained APs that do not contain the replaced APs. Simultaneously, a Kalman filter algorithm is used to process the historical movement trajectory coordinates of the transport bed from multiple past cycles to obtain the extended predicted position coordinates for the current cycle.

[0035] Calculate the Euclidean distances from the first and second positioning coordinates to the extended predicted position coordinates, respectively. If the Euclidean distance corresponding to the first positioning coordinate is less than the Euclidean distance corresponding to the second positioning coordinate, it is determined that the replacement AP can improve the positioning result, and the contribution weight of the replacement AP is calculated based on the degree of improvement of the distances between the two. If the Euclidean distance corresponding to the first positioning coordinate is greater than or equal to the Euclidean distance corresponding to the second positioning coordinate, it is determined that the replacement AP has not improved the positioning result, and the contribution weight of the replacement AP is set to a preset lower limit value.

[0036] After the observation period ends, the average contribution weight of the replacement AP in each monitoring cycle during the observation period is calculated, and it is determined whether the average value is greater than a preset inclusion weight threshold. If it is greater, the replacement AP is determined to meet the formal inclusion conditions, and the replacement AP is formally included in the AP subset to replace the AP to be replaced. The preset inclusion weight threshold is preferably 1.0. If the average value is not greater than the preset inclusion weight threshold, the replacement AP is not formally included in the AP subset, and the original AP subset is maintained or the replacement AP selection process is restarted.

[0037] In one embodiment, when the update triggering condition is met, determining the APs to be replaced from the current AP subset and selecting replacement APs from the candidate APs outside the AP subset includes: Obtain all remaining candidate APs that are not included in the AP subset, and establish a replacement candidate set; The APs to be replaced are determined based on the source selection weights of each AP in the current AP subset. The AP with the lowest source selection weight is determined as the AP to be replaced, and the APs in the current AP subset other than the AP to be replaced are reserved APs. Extract the physical location coordinate parameters of each candidate AP in the replacement candidate set; The physical location coordinate parameters of each candidate AP in the replacement candidate set are combined with the physical location coordinate parameters of the retained AP to form multiple independent test geometry matrices; Read the historical positioning coordinates of the transfer bed for one monitoring cycle and use them as the reference position for geometric calculation; Based on the reference position and the two-dimensional planar physical position coordinates of each AP, the test geometry matrix is ​​linearized to obtain the corresponding geometric accuracy attenuation factor; Extract the minimum value among all calculated geometric precision attenuation factors, and designate the candidate AP associated with the test geometry matrix corresponding to the minimum value as the replacement AP.

[0038] Suppose there are a total of 12 AP devices in the current deployment of the backup network. When it is determined that there is a node in the AP subset containing 5 backup APs that needs to be removed, first determine one AP to be replaced according to the source selection weight of each AP in the current AP subset, and keep the remaining 4 APs as the reserved APs.

[0039] Obtain the remaining 7 candidate APs that are not included in the original subset and establish a replacement candidate set; then extract the two-dimensional planar physical location coordinate parameters of the 7 candidate APs in the replacement candidate set. Sequentially extract the two-dimensional planar physical location coordinate data of the 4 retained APs in the current AP subset, and combine the data with the two-dimensional planar physical location coordinate parameters of the 1st to 7th candidate APs in the replacement candidate set to establish 7 test geometry matrices.

[0040] When calculating the geometric accuracy attenuation factor for each test geometry matrix, the historical positioning coordinates of the transport bed for one monitoring cycle are first read and used as the reference position for geometric calculation. Then, the direction vectors from the reference position to the two-dimensional planar physical position coordinates of each AP in the test geometry matrix are calculated, and each direction vector is normalized. The normalized direction vector corresponding to each AP is used as a row in a Jacobian matrix, and these rows are written sequentially according to the order of the APs in the test geometry matrix to form the Jacobian matrix. Subsequently, the inverse matrix of the product of the transpose of the Jacobian matrix and the Jacobian matrix is ​​calculated, and the geometric accuracy attenuation factor of the corresponding test geometry matrix is ​​obtained based on the square root of the sum of the diagonal elements of the inverse matrix. This process is repeated for each candidate AP's corresponding test geometry matrix, with calculated values ​​typically ranging from 1.5 to 2.8.

[0041] Extract the minimum value among all calculated geometric precision attenuation factors, such as 1.5, and designate the candidate AP associated with the test geometry matrix corresponding to this minimum value as the replacement AP.

[0042] In one embodiment, the replacement AP is combined with the retained APs in the current AP subset excluding the AP to be replaced to form an observation subset, and an observation period is set. During the observation period, a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP are calculated respectively. The first and second positioning coordinates are compared with the extended predicted positions obtained based on historical movement trajectories, and the contribution weight of the replacement AP is determined according to the comparison results, including: The observation period is defined as the time span of three consecutive data sampling periods. The coordinates of the five most recent historical movement trajectories of the transfer bed are processed by calling the first-order Kalman filter equation to calculate the specific coordinates of the extended predicted position. Calculate the first Euclidean distance from the first positioning coordinates generated by the replacement AP to the extended predicted position, and the second Euclidean distance from the second positioning coordinates not generated by the replacement AP to the extended predicted position. The distance improvement ratio is calculated based on the first Euclidean distance and the second Euclidean distance, wherein the distance improvement ratio is the reduction ratio of the first Euclidean distance to the second Euclidean distance; When the distance improvement ratio is greater than zero, the contribution weight of the replacement AP is calculated based on the initial fixed constant weight of the replacement AP and the distance improvement ratio; When the distance improvement ratio is not greater than zero, the contribution weight of the replaced AP is set to a preset lower limit value.

[0043] Within the scheduling configuration, a time span of three consecutive data sampling periods is set as the observation period. When the periodic calculation is started within the observation period, a preset first-order Kalman filter module is invoked to recursively process the coordinates of the five most recent historical movement trajectories arranged in chronological order, such as [(6.1, 2.0), (6.4, 2.2), (6.7, 2.0), (7.0, 2.1), (7.3, 2.0)]. The planar position and velocity of the transfer bed are used as the filter state variables, and the coordinates of each historical movement trajectory are used as the observation variables. The recursion is performed in the order of state prediction, observation update, and state extrapolation for the next cycle to obtain the predicted planar position for the next monitoring cycle. The specific coordinate values ​​of the extended predicted position used for benchmark comparison are calculated, such as (7.6, 2.1).

[0044] Using a dual-thread concurrent calculation mode, the first positioning coordinates (7.5, 2.1) are calculated using the signal generated by the replacement AP. The first Euclidean distance parameter from this point to the extended predicted position (7.6, 2.1) is calculated to be 0.1 meters. Simulation verification is performed using the signal not generated by the replacement AP to calculate the second positioning coordinates (7.2, 2.5). The second Euclidean distance parameter from this point to (7.6, 2.1) is calculated to be 0.56 meters. A ratio calculation process is executed, subtracting the first Euclidean distance (0.1) from the second Euclidean distance (0.56), and dividing this difference by the second Euclidean distance (0.56) yields a distance improvement ratio of 0.82.

[0045] Since the distance improvement ratio of 0.82 is greater than zero, it indicates that the positioning coordinates including the replacement AP are closer to the extended predicted location. Therefore, the contribution weight of the replacement AP is calculated based on this distance improvement ratio, using its initial fixed constant weight. Assuming the initial fixed constant weight of the replacement AP is 1.0, multiplying the initial fixed constant weight of 1.0 by the distance improvement ratio of 0.82 yields a contribution weight of 1.82 for the replacement AP. If the first Euclidean distance is greater than or equal to the second Euclidean distance, it indicates that the replacement AP has not improved the positioning result, and its contribution weight is set to a preset lower limit value, preferably zero. This completes the calculation and assignment of the replacement AP's contribution weight.

[0046] S3. Using the AP signal fingerprint data in the updated AP subset and combining it with the contribution weight, calculate the real-time positioning coordinates and positioning result confidence of the transfer bed; generate a comprehensive boundary crossing judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed; set an alarm threshold associated with the positioning result confidence, and trigger an alarm when the alarm threshold is exceeded.

[0047] During the location fingerprinting stage, the signal strength values ​​measured in real time for each access point in the updated access point subset are extracted to form an online fingerprint vector. This online fingerprint vector is then matched with the fingerprint vectors of each reference point in the offline fingerprint database. During the matching calculation, the contribution weight of each access point is applied to the signal strength difference in the corresponding dimension. Based on the normalized weighted Manhattan distance, multiple reference points closest to the online fingerprint vector are determined, and the weighted average of the physical coordinates of these reference points is taken as the real-time positioning coordinates of the transport bed. Simultaneously, a normalized mapping is performed based on the reciprocal of the root mean square of the distances corresponding to the multiple reference points to obtain the confidence level of the positioning result for the current period.

[0048] The coordinates of the pre-established polygonal boundary of the hospital department's safety area are extracted. The shortest distance from the real-time positioning coordinates to the nearest boundary line is calculated, and a distance risk item is generated based on the positive difference between the preset alarm limit's maximum allowable distance and the shortest distance. Simultaneously, the direction cosine value of the historical movement trajectory of the transfer bed towards the outside of the nearest boundary line is calculated, and a trend risk item is generated based on the positive value of the direction cosine value. The distance risk item and the trend risk item are weighted and summed according to preset weights to obtain a comprehensive boundary crossing judgment index. When the real-time positioning coordinates are located outside the limited activity area, the boundary crossing alarm condition is directly determined to be met.

[0049] A baseline warning value is set, an adjustment factor is generated based on the confidence level of the current positioning result, and an alarm threshold corresponding to the comprehensive boundary crossing judgment index is obtained. When the calculated comprehensive boundary crossing judgment index is greater than or equal to the alarm threshold, it indicates that the transport bed is at risk of leaving the safe area. An boundary crossing alarm signal is sent to the nurse station monitoring terminal via a wired network, and the buzzer on the transport bed is activated.

[0050] In one embodiment, the step of using the updated AP signal fingerprint data within the AP subset, combined with the contribution weight, to calculate the real-time positioning coordinates and positioning result confidence of the transfer bed includes: The signal strength of the corresponding AP is extracted using the updated AP subset to form an online fingerprint feature vector, and the contribution weight of each AP in the subset is obtained. Call the offline fingerprint database to calculate the normalized weighted Manhattan distance between the online fingerprint feature vector and the corresponding AP signal strength component in the fingerprint vector of each pre-entered reference point. The normalized weighted Manhattan distance is obtained by normalizing the absolute value of the signal strength difference of the corresponding AP, multiplying it by the corresponding contribution weight, and summing the results. The contribution weight of the non-replacement AP is a preset constant value. Extract the first four reference points and their grid coordinates in ascending order of the normalized weighted Manhattan distance; The position weight of each reference point is obtained by normalizing the inverse of the distances corresponding to the first four reference points, and the grid coordinates of each reference point are weighted and summed to obtain the real-time positioning coordinates of the transfer bed. Normalization mapping is performed based on the reciprocal of the root mean square of the distances to the first four reference points, and the mapping result is restricted to between 0 and 1 to obtain the confidence level of the positioning result.

[0051] The received signal strength of a total of 5 APs in the updated AP subset is read and constructed as an online fingerprint feature vector containing index values ​​such as [-62, -58, -50, -74, -80] dBm; at the same time, the contribution weight of each AP is obtained, where the contribution weight of the 4 non-replacement APs is set to a preset constant value of 1.0, and the contribution weight of the 5th replacement AP is the updated generated weight, for example, 1.82.

[0052] The offline fingerprint database is invoked, and the fingerprint vectors of the pre-stored reference points are extracted one by one. For example, the fingerprint vector of a certain reference point is [-60, -60, -52, -70, -78] dBm. The offline fingerprint database is constructed by pre-collection. Specifically, the activity area of ​​the transfer bed is divided into multiple reference points according to a preset grid spacing. The received signal strength of each AP is collected at each reference point, and the two-dimensional grid coordinates of the reference point, the AP identifier, and the corresponding received signal strength are associated and stored to form the offline fingerprint database.

[0053] Before calculating the distance, the absolute value of the signal strength difference of each corresponding AP is normalized according to the preset signal strength difference range. Then, the normalized difference is multiplied by the corresponding contribution weight and summed to obtain the normalized weighted Manhattan distance.

[0054] Sort all normalized weighted Manhattan distances in ascending order of numerical value, and extract the top four normalized weighted Manhattan distances and their corresponding reference point grid coordinates. Introduce a preset minimal positive number parameter, such as 0.01, to prevent division by zero. Calculate the reciprocal of the sum of the corresponding normalized weighted Manhattan distance and this 0.01, divide it by the sum of the reciprocals of the sum of the four normalized weighted Manhattan distances and 0.01, and then perform normalization to obtain the position weights of each reference point.

[0055] The four position weights are multiplied by their corresponding grid coordinates and then summed to generate a unique real-time positioning coordinate for the current period. The root mean square of the normalized weighted Manhattan distances of the top four positions is extracted, and a preset minimum positive number of 0.01 is added to prevent division by zero. The reciprocal is then taken and mapped to the range of 0 to 1 using a preset normalization function to obtain the positioning result confidence score, which is then submitted as output.

[0056] In one embodiment, generating a comprehensive boundary violation judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed includes: Retrieve the polyline coordinate dataset of the closed boundary of the limited activity area of ​​the transfer bed, and determine whether the transfer bed is located within the limited activity area based on its real-time positioning coordinates; if it is located outside the limited activity area, the boundary crossing alarm condition is met. If it is located within a limited activity area, a distance risk item is generated based on the positive difference between the maximum allowable distance of the preset alarm limit and the shortest distance from the real-time positioning coordinates to the nearest closed boundary line. A trend velocity vector is generated based on the historical movement trajectory coordinates of the transfer bed and the latest real-time positioning coordinates, and a trend risk term is generated based on the positive value of the direction cosine between the trend velocity vector and the outer normal vector of the nearest boundary. The distance risk item and the trend risk item are weighted and summed according to preset weights to obtain a comprehensive boundary judgment index.

[0057] Retrieve the coordinate dataset of the closed boundary polyline of the limited activity area of ​​the transfer bed. Using the real-time positioning coordinates of the transfer bed obtained in this cycle as the positioning point, for example, coordinates (18.5, 9.2), first determine whether the positioning point is located within the limited activity area enclosed by the closed boundary polyline. If it is located outside the limited activity area, the boundary crossing alarm condition is directly met and an boundary crossing alarm is triggered; if the positioning point is located within the limited activity area, calculate the shortest Euclidean distance from the positioning point to the nearest closed boundary polyline, for example, the shortest distance is determined to be 0.6 meters.

[0058] Set a maximum permissible alarm distance, for example, 1.5 meters. When the shortest Euclidean distance is less than the maximum permissible alarm distance, subtract the shortest Euclidean distance from the maximum permissible alarm distance to obtain the deviation value; when the shortest Euclidean distance is not less than the maximum permissible alarm distance, set the deviation value to zero. In this embodiment, the shortest Euclidean distance of 0.6 meters is less than the maximum permissible alarm distance of 1.5 meters, therefore, subtracting 0.6 meters from 1.5 meters yields a deviation value of 0.9 meters.

[0059] Extract the third-to-last historical trajectory coordinates from the historical trajectory cache, for example (16.5, 8.2), and use this coordinate as the starting point and the latest real-time positioning coordinates (18.5, 9.2) as the ending point to generate a directional velocity vector, for example [2.0, 1.0].

[0060] Combining the boundary normal vector closest to the current real-time positioning coordinates and pointing towards the outside of the defined activity area, for example, the standardized outer normal vector [0.8, 0.6], the direction cosine value between the trend velocity vector and the boundary normal vector is calculated by vector dot product. Assuming the calculation result is +0.85, since this direction cosine value is greater than zero, it is used as a trend risk term; if the direction cosine value is not greater than zero, it indicates that the transfer bed has not moved towards the outside of the defined activity area, and the trend risk term is set to zero.

[0061] Based on the first preset weighting coefficient of 0.6 and the second preset weighting coefficient of 0.4, the comprehensive boundary judgment index is calculated as follows: the deviation value of 0.9 is divided by the maximum allowable distance of 1.5 under alarm limit to obtain the distance risk item of 0.6; the distance risk item of 0.6 is multiplied by the first preset weighting coefficient of 0.6 to obtain 0.36; the trend risk item of 0.85 is multiplied by the second preset weighting coefficient of 0.4 to obtain 0.34; the two are added together to obtain the comprehensive boundary judgment index of 0.70.

[0062] In one embodiment, setting the alarm threshold associated with the confidence level of the positioning result includes: Recall the preset warning values ​​stored within the control system; Read the actual measured value of the confidence level of the acquired positioning result, and add a preset small positive number to the measured value to obtain the adjustment factor; Multiply the preset warning value by the adjustment factor to obtain the real-time alarm threshold corresponding to the comprehensive boundary judgment index, and use the real-time alarm threshold to update the fixed value trigger limit in the main control judgment execution logic.

[0063] For example, the preset warning value is 0.75. Read the actual measured value of the confidence level of the positioning result generated by the upstream positioning calculation cycle. Assuming that the current environmental interference is strong, this value is 0.25. Then add the preset small positive number to prevent overflow, such as 0.05, to get the adjustment factor of 0.30.

[0064] Multiplying the preset warning value of 0.75 by the adjustment factor of 0.30 yields a real-time alarm threshold of 0.225, which is then used to update the comparison limit in the trigger judgment logic. Subsequently, when the system obtains a comprehensive boundary violation judgment index of 0.4, although this index does not exceed the original preset warning value of 0.75, it exceeds the current real-time alarm threshold of 0.225, thus triggering a boundary violation alarm, and the processing chip drives the buzzer to issue a safety warning. Therefore, by lowering the alarm trigger threshold when the confidence level of the positioning result is low, the sensitivity of alarm judgment is improved, and the possibility of missed alarms due to positioning fluctuations is reduced.

[0065] Example 2 Embodiment 2 of the present invention proposes an intelligent transfer bed boundary crossing alarm system, comprising the following modules: The calculation module is used to acquire candidate AP signal datasets; calculate source selection weights for each candidate AP based on the signal strength in the signal dataset, the preset physical locations of each AP, and the stored historical signal strength statistical characteristics; and adjust the number N of members in the current AP subset based on the confidence level of the positioning results calculated in the previous monitoring cycle to form an AP subset. An adjustment module is used to set update trigger conditions. When the update trigger conditions are met, APs to be replaced are determined from the current AP subset, and replacement APs are selected from the candidate APs outside the AP subset. The replacement APs are combined with the reserved APs in the current AP subset (excluding the APs to be replaced) to form an observation subset, and an observation period is set. During the observation period, a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP are calculated. The first and second positioning coordinates are compared with the extended predicted positions obtained based on historical movement trajectories, and the contribution weight of the replacement AP is determined based on the comparison results. After the observation period ends, it is determined whether to formally include the replacement AP in the AP subset to replace the APs to be replaced based on the contribution weight. The generation module is used to calculate the real-time positioning coordinates and positioning result confidence of the transfer bed by using the AP signal fingerprint data in the updated AP subset and combining it with the contribution weight; generate a comprehensive boundary judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed; and set an alarm threshold associated with the positioning result confidence, triggering an alarm when the alarm threshold is exceeded.

[0066] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A smart transfer bed boundary crossing alarm method, characterized in that, The method comprises the following steps: acquiring a candidate AP signal dataset; calculating a source selection weight for each candidate AP according to signal strength in the signal dataset, preset physical positions of each AP, and stored historical signal strength statistical characteristics; adjusting the number N of members of the current AP subset according to the positioning result confidence obtained in the last monitoring period to form an AP subset; setting an update trigger condition, and when the update trigger condition is met, determining an AP to be replaced from the current AP subset and selecting a replacement AP from the candidate APs outside the AP subset; forming an investigation subset by combining the replacement AP and the remaining APs in the current AP subset except the AP to be replaced, and setting an investigation period; in the investigation period, calculating a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP, comparing the first positioning coordinate and the second positioning coordinate with an extended predicted position obtained based on a historical movement trajectory, and determining a contribution weight of the replacement AP according to the comparison result; after the investigation period, determining whether to formally include the replacement AP in the AP subset to replace the AP to be replaced according to the contribution weight; calculating the real-time positioning coordinate and the positioning result confidence of the transfer bed by using the AP signal fingerprint data in the updated AP subset and combining the contribution weight; generating a comprehensive out-of-boundary judgment index according to the real-time positioning coordinate of the transfer bed and the historical movement trajectory; setting an alarm threshold value associated with the positioning result confidence, and triggering an alarm when the alarm threshold value is exceeded.

2. The method of claim 1, wherein, The method of calculating the source selection weight for each candidate AP according to signal strength in the signal dataset, preset physical positions, and stored historical signal strength statistical characteristics comprises the following steps: reading the current received signal strength value of the candidate AP in the signal dataset; extracting the historical strength mean value of the corresponding candidate AP in the stored historical signal strength statistical characteristics; extracting the historical positioning coordinate of the transfer bed in the last monitoring period, and calculating the straight-line spatial distance between the historical positioning coordinate and the preset physical position of the corresponding candidate AP; adding the absolute value of the historical strength mean value to the absolute value of the difference between the current received signal strength value and the historical strength mean value to obtain a first value; adding the straight-line spatial distance to a preset first distance buffer constant with a length unit to obtain a second value; for each candidate AP, multiplying the corresponding first value and the second value to obtain a comprehensive denominator, and dividing one by the comprehensive denominator to obtain the source selection weight of the corresponding candidate AP.

3. The method of claim 1, wherein, The method of adjusting the number N of members of the current AP subset according to the positioning result confidence obtained in last monitoring period comprises the following steps: presetting a first fixed confidence lower limit parameter and a second fixed confidence lower limit parameter, and the first fixed confidence lower limit parameter is smaller than the second fixed confidence lower limit parameter; when the positioning result confidence value obtained in the last monitoring period is read and is less than the first fixed confidence lower limit parameter, setting the number N of members of the current AP subset to 7. when the positioning result confidence value obtained in the last monitoring period is read and is not less than the first fixed confidence lower limit parameter and is less than the second fixed confidence lower limit parameter, the number N of members of the current AP subset is set to 5; when the positioning result confidence value obtained in the last monitoring period is read and is not less than the second fixed confidence lower limit parameter, the number N of members of the current AP subset is set to 4.

4. The method of claim 2, wherein, The determining of the AP to be replaced from the current AP subset and the selection of the replacement AP from the candidate AP outside the AP subset when the update trigger condition is met, comprises: all the remaining candidate APs not included in the AP subset are acquired to establish a replacement candidate set; the AP to be replaced is determined according to the source selection weight of each AP in the current AP subset, the AP with the lowest source selection weight is determined as the AP to be replaced, and the APs in the current AP subset except the AP to be replaced are determined as the reserved APs; the physical position coordinate parameters of each candidate AP in the replacement candidate set are extracted; the physical position coordinate parameters of each candidate AP in the replacement candidate set are combined with the physical position coordinate parameters of the reserved APs respectively to form a plurality of independent test geometric matrices; the historical positioning coordinates of the trolley in the last monitoring period are read and are taken as the reference position for geometric calculation; the test geometric matrices are linearly calculated based on the reference position and the two-dimensional plane physical position coordinates of each AP to obtain corresponding geometric dilution of precision factors; the minimum value of all the geometric dilution of precision factors obtained by calculation is extracted, and the candidate AP associated with the minimum value of the test geometric matrix is designated as the replacement AP.

5. The method of claim 1, wherein, The replacement AP and the reserved APs in the current AP subset except the AP to be replaced are combined to form an investigation subset, and an investigation period is set; in the investigation period, a first positioning coordinate containing the replacement AP and a second positioning coordinate not containing the replacement AP are calculated respectively, the first positioning coordinate and the second positioning coordinate are compared with the extended predicted position obtained based on the historical movement trajectory respectively, and the contribution weight of the replacement AP is determined according to the comparison result, comprising: a time span of 3 consecutive data sampling periods is set as the investigation period; a first-order Kalman filtering equation is called to process the last 5 historical movement trajectory coordinates of the trolley to calculate the specific coordinate value of the extended predicted position; a first Euclidean distance from the first positioning coordinate generated by the replacement AP to the extended predicted position and a second Euclidean distance from the second positioning coordinate not containing the replacement AP to the extended predicted position are calculated respectively; a distance improvement ratio is calculated according to the first Euclidean distance and the second Euclidean distance, the distance improvement ratio being the reduction ratio of the first Euclidean distance relative to the second Euclidean distance; when the distance improvement ratio is greater than zero, the contribution weight of the replacement AP is calculated based on the initial fixed constant weight of the replacement AP and the distance improvement ratio; when the distance improvement ratio is not greater than zero, the contribution weight of the replacement AP is set to a preset lower limit value.

6. The method of claim 1, wherein, The step of using the updated AP signal fingerprint data within the AP subset, combined with the contribution weights, to calculate the real-time positioning coordinates and positioning result confidence of the transfer bed includes: The signal strength of the corresponding AP is extracted using the updated AP subset to form an online fingerprint feature vector, and the contribution weight of each AP in the subset is obtained. Call the offline fingerprint database to calculate the normalized weighted Manhattan distance between the online fingerprint feature vector and the corresponding AP signal strength component in the fingerprint vector of each pre-entered reference point. The normalized weighted Manhattan distance is obtained by normalizing the absolute value of the signal strength difference of the corresponding AP, multiplying it by the corresponding contribution weight, and summing the results. The contribution weight of the non-replacement AP is a preset constant value. Extract the first four reference points and their grid coordinates in ascending order of the normalized weighted Manhattan distance; The position weight of each reference point is obtained by normalizing the inverse of the distances corresponding to the first four reference points, and the grid coordinates of each reference point are weighted and summed to obtain the real-time positioning coordinates of the transfer bed. Normalization mapping is performed based on the reciprocal of the root mean square of the distances to the first four reference points, and the mapping result is restricted to between 0 and 1 to obtain the confidence level of the positioning result.

7. The method of claim 1, wherein, The step of generating a comprehensive boundary violation judgment index based on the real-time positioning coordinates and historical movement trajectory of the transfer bed includes: Retrieve the polyline coordinate dataset of the closed boundary of the limited activity area of ​​the transfer bed, and determine whether the transfer bed is located within the limited activity area based on its real-time positioning coordinates; if it is located outside the limited activity area, the boundary crossing alarm condition is met. If it is located within a limited activity area, a distance risk item is generated based on the positive difference between the maximum allowable distance of the preset alarm limit and the shortest distance from the real-time positioning coordinates to the nearest closed boundary line. A trend velocity vector is generated based on the historical movement trajectory coordinates of the transfer bed and the latest real-time positioning coordinates, and a trend risk term is generated based on the positive value of the direction cosine between the trend velocity vector and the outer normal vector of the nearest boundary. The distance risk item and the trend risk item are weighted and summed according to preset weights to obtain a comprehensive boundary judgment index.

8. The method of claim 1, wherein, The setting of the alarm threshold associated with the confidence level of the positioning result includes: Recall the preset warning values ​​stored within the control system; Read the actual measured value of the confidence level of the acquired positioning result, and add a preset small positive number to the measured value to obtain the adjustment factor; Multiply the preset warning value by the adjustment factor to obtain the real-time alarm threshold corresponding to the comprehensive boundary judgment index, and use the real-time alarm threshold to update the fixed value trigger limit in the main control judgment execution logic.

9. An intelligent transfer bed out-of-boundary alarm system, characterized in that, Includes the following modules: The calculation module is used to acquire candidate AP signal datasets; calculate source selection weights for each candidate AP based on the signal strength in the signal dataset, the preset physical locations of each AP, and the stored historical signal strength statistical characteristics; and adjust the number N of members in the current AP subset based on the confidence level of the positioning results calculated in the previous monitoring cycle to form an AP subset. An adjusting module is configured to set an update trigger condition, determine an AP to be replaced from a current AP subset when the update trigger condition is met, and select a replacement AP from candidate APs outside the AP subset; The replacement AP and reserved APs in the current AP subset except the AP to be replaced form an observation subset, and an observation period is set; During the observation period, a first positioning coordinate including the replacement AP and a second positioning coordinate excluding the replacement AP are calculated respectively, the first and second positioning coordinates are compared with an extended predicted position based on a historical moving trajectory, and a contribution weight of the replacement AP is determined according to a comparison result; after the observation period, it is determined whether to formally include the replacement AP in the AP subset to replace the AP to be replaced according to the contribution weight; A generating module is configured to calculate real-time positioning coordinates and a positioning result confidence of the stretcher by using AP signal fingerprint data in the updated AP subset and combining the contribution weight; A comprehensive out-of-boundary judgment index is generated according to the real-time positioning coordinates of the stretcher and the historical moving trajectory; an alarm threshold associated with the positioning result confidence is set, and an alarm is triggered when the alarm threshold is exceeded.

10. The system of claim 9, wherein, The source selection weight of each candidate AP is calculated according to the signal strength in the signal data set, the preset physical position of each AP, and the stored historical signal strength statistical characteristics, including: Reading the current received signal strength value of the candidate AP in the signal data set; Extracting the historical strength mean value of the corresponding candidate AP in the stored historical signal strength statistical characteristics; Extracting the historical positioning coordinates of the stretcher in a monitoring period, and calculating the straight-line spatial distance between the historical positioning coordinates and the preset physical position of the corresponding candidate AP; Adding the absolute value of the historical strength mean value to the absolute value of the difference between the current received signal strength value and the historical strength mean value to obtain a first value; Then, adding the straight-line spatial distance to a preset first distance buffer constant with a length unit to obtain a second value; For each candidate AP, the corresponding first value and the second value are multiplied to obtain a comprehensive denominator, and the source selection weight of the corresponding candidate AP is obtained by dividing one by the comprehensive denominator.

Images