Adaptive WKNN indoor positioning method, system, device and medium

Abstract

The invention discloses a self-adaptive WKNN indoor positioning method, system, device and medium, and belongs to the technical field of wireless localization, and the method comprises the steps: collecting RSSI data of all beacons at each reference point to construct a fingerprint database, and obtaining the confidence of the beacons at each reference point; collecting RSSI data of each beacon in real time by a to-be-positioned point; according to the difference degree of the to-be-positioned point and the reference points to the beacon visibility, screening the reference points, and obtaining the comprehensive similarity of each screened reference point and the to-be-positioned point; and adaptively selecting a set number of target reference points, performing weighted fusion on the positions of all the target reference points, and obtaining the position of the to-be-positioned point. According to the invention, equipment difference can be overcome, differentiated processing is carried out on RSSI data, interference of overall bias / unified gain on matching is reduced, and the amplification effect of unstable signals is inhibited, so that the overall positioning precision and robustness are improved in a complex indoor environment.

Description

Technical Field

[0001] The present invention belongs to the field of wireless positioning technology, and in particular relates to an adaptive WKNN indoor positioning method, system, device and medium. Background Art

[0002] Indoor positioning, a key foundational capability for scenarios such as smart cities, smart factories, commercial navigation, and security inspections, has developed alongside multiple technology paths. Fingerprint positioning based on the Received Signal Strength Indicator (RSSI) has gained widespread adoption due to its low deployment cost, minimal network modification, and compatibility with multiple device brands. A typical process involves collecting RSSIs from multiple beacons (such as Wi-Fi APs and BLE Beacon) at a reference point during the offline phase to build a fingerprint database. During the online phase, the RSSI vector of the device being located is obtained, matched against the fingerprint database for similarity, and outputs a location estimate.

[0003] Among many fingerprint algorithms, weighted K-nearest neighbor (WKNN) is widely adopted due to its simplicity and superior robustness to KNN. Its basic concept is to calculate the similarity between the fingerprint to be located and the fingerprints of each reference point (usually based on a magnitude metric such as Euclidean distance or Manhattan distance), select the K reference points with the highest similarity, and sum them up weighted by the similarity to obtain a location estimate. However, as the scale of applications and environmental complexity increase, several inherent assumptions and engineering simplifications of traditional WKNN have gradually exposed limitations, primarily in the following aspects: 1. Amplitude metrics are sensitive to overall bias / unity gain; Factors such as different terminal models, transmit power control, temperature and battery status, and front-end gain differences can cause RSSI to shift or scale uniformly. When measured solely based on amplitude differences, this global shift systematically amplifies the distances between previously similar fingerprints, reversing the order of reference points and leading to position estimation errors.

[0004] 2. Beacon stability has significant heteroscedasticity and time-varying properties; Affected by factors such as occlusion, crowds, multipath, and antenna pointing, the RSSI fluctuations of different beacons vary significantly and change over time. If a uniform weight is used or a metric that does not distinguish between stability differences, unstable beacons can easily have a disproportionate adverse impact on the distance and weighting process, reducing positioning accuracy and robustness.

[0005] 3. The visibility mode is not pre-constrained; In real-world environments, the visible beacon sets between the target location and the reference point often differ due to differences in floors, rooms, and walls. Direct distance calculations in amplitude space can easily introduce false neighbors that span floors or penetrate walls. The lack of consistent gating based on visibility patterns fundamentally reduces the reliability of candidate neighbors.

[0006] 4. The selection of K value often relies on fixed experience or heuristic threshold; A fixed K is difficult to take into account different distribution forms such as dense or sparse fingerprints, sharp or flat similarities; the heuristic dynamic K based on the mean or threshold is also difficult to characterize the similarity distribution coverage or risk level, resulting in alternating overfitting (K is too small) or underfitting (K is too large).

[0007] 5. A single amplitude channel lacks the ability to exploit order invariance; Traditional similarity measures only compare RSSI values ​​themselves, failing to leverage the order of strength, a structural information that is inherently insensitive to overall translation / scaling. The discriminative power of a single amplitude channel is significantly reduced when there is device bias or uniform gain variation.

[0008] 6. Over-reliance on a single strong beacon and lack of robustness assessment; In some scenarios, individual high-power / close-range beacons will dominate the similarity evaluation; once the beacon is temporarily abnormal or blocked, the matching is easily affected by single-point distortion, and the traditional process lacks a robust correction mechanism for beacon-by-beacon sensitivity.

[0009] In summary, the existing WKNN system still has structural shortcomings in accuracy and stability when facing device differences, beacon heteroscedasticity, NLOS / occlusion, multimodality and time-varying environments. Summary of the Invention

[0010] In response to the deficiencies in the prior art, the present invention provides an adaptive WKNN indoor positioning method, system, device, and medium, which can overcome device differences, perform differentiated processing on RSSI data, reduce the interference of overall bias / unified gain on matching, and suppress the amplification effect of unstable signals, thereby improving the accuracy and robustness of overall positioning in complex indoor environments.

[0011] The present invention provides the following technical solutions: In a first aspect, an adaptive WKNN indoor positioning method is provided, wherein a plurality of beacons and reference points are deployed indoors; the method comprises the following steps: S1: Collect RSSI data of all beacons at each reference point to build a fingerprint library, and obtain the confidence of the beacon at each reference point based on the stability of each beacon at the reference point; S2: The positioning point collects RSSI data of each beacon in real time; S3: Based on the difference in beacon visibility between the target point and the reference point, the reference points are screened, and for each screened reference point, the comprehensive similarity between the reference point and the target point is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the target point, the similarity between the reference point and the target point is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the target point is obtained after fusion. S4: Adaptively select a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be located, weightedly fuse the positions of all target reference points to obtain the position of the points to be located.

[0012] Optionally, the fingerprint library includes: fingerprint data samples of all reference points; wherein the fingerprint data samples of the reference points include: position coordinates of the reference points and RSSI vectors of the reference points; Constructing the RSSI vector of the reference point: Specifically, the reference point collects RSSI data of all beacons multiple times, and obtains the RSSI vector of the reference point after performing sliding average and denoising filtering.

[0013] Optionally, in step S1, obtaining the confidence level of the beacon at each reference point includes obtaining the confidence level of the beacon b At the reference point r Confidence, specifically: based on the reference point r Multiple beacon collection b The variance of the RSSI data , the confidence level is obtained according to the following formula Then normalize; ; in, A constant to prevent the denominator from being zero.

[0014] Optionally, in step S3, the reference point is selected based on the difference in beacon visibility between the point to be located and the reference point, specifically: S31: Define the point to be located x The visibility vector and reference points r The visibility vector ; ; ; in, , M is the total number of beacons, and Reference points r and the point to be located xRSSI vector of beacon b RSSI value; S32: Obtain the point to be located by calculating the Hamming distance between the two visibility vectors x and reference points r Visibility difference , if the visibility difference If it is greater than the set threshold, the reference point will be eliminated r Otherwise, keep the reference point r , complete the screening; ; in, is the Hamming distance between the visibility vectors of the point to be located and the reference point.

[0015] Optionally, in step S3, for each reference point after screening, the comprehensive similarity between the reference point and the point to be located is obtained, including obtaining the point to be located x With reference point r The comprehensive similarity is: S33: Constructing points to be located x With reference point r The set of commonly visible beacons ,in, and Points to be positioned x With reference point r Visibility vectors for commonly visible beacons p The weight; S34: Calculate the reference point based on the RSSI data of all beacons received at the reference point and the point to be positioned and the confidence of the beacon at the reference point. r With the point to be positioned x The weighted Euclidean distance , and map it in the interval [0,1] to obtain the amplitude channel similarity ; S35: Set the points to be positioned x and reference points r China's visible beacon collection P The corresponding RSSI values ​​are sorted in descending order to obtain the values ​​of the points to be located. x and reference points r The two corresponding orders; using the Kendall rank correlation coefficient Measure the consistency between two orders and map them in the interval [0,1] to obtain the order channel similarity ; S36: Based on preset fusion weights , obtain the points to be located by weighting x With reference point rThe comprehensive similarity ; .

[0016] Optionally, in step S34, the reference point r With the point to be positioned x The weighted Euclidean distance for: ,in, and Points to be positioned x and reference points r RSSI vector of commonly visible beacons p RSSI value, Beacon p At the reference point r Normalized confidence of the amplitude channel similarity for: .

[0017] Optionally, in step S4, specifically: S41: sorting the comprehensive similarities between the point to be located and each of the screened reference points in descending order to obtain a descending sequence of comprehensive similarities; S42: Set is the cumulative sum of the first K elements of the descending sequence of comprehensive similarity, For the sum of all elements of the descending sequence of comprehensive similarity, find the minimum integer K that satisfies ,in, To set the coverage weight value; S43: Preset the boundary of the K value. If the obtained K is within the boundary, the number of target reference points is determined to be K; if the obtained K is outside the boundary, the number of target reference points is determined to be boundary points; if the total number of elements in the descending order of the comprehensive similarity is less than the minimum boundary value of K, the number of target reference points is the total number of elements in the descending order of the comprehensive similarity; S44: Weight the positions of all target reference points to obtain the position of the point to be located ; ; in, is the normalized target reference point k With the point to be positioned x The comprehensive similarity value of Target reference point k The location coordinates of .

[0018] In a second aspect, an adaptive WKNN indoor positioning system is provided, comprising: Several beacons are deployed in the environment where positioning is required according to the set spatial intervals; Several reference points are evenly distributed in the positioning environment according to the path nodes or grid density of the target area; Fingerprint library construction module: collects RSSI data of all beacons at each reference point to build a fingerprint library, and obtains the confidence of the beacon at each reference point based on the stability of each beacon at the reference point; Real-time acquisition module: The positioning point collects RSSI data of each beacon in real time; Comprehensive similarity acquisition module: Based on the difference in beacon visibility between the target point and the reference point, the reference point is screened and the comprehensive similarity between each screened reference point and the target point is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the target point, the similarity between the reference point and the target point is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the target point is obtained after fusion. Positioning module: Adaptively selects a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be positioned, weightedly fuses the positions of all target reference points to obtain the position of the points to be positioned.

[0019] In a third aspect, a computer device is provided, comprising a processor and a memory; wherein, when the processor executes a computer program stored in the memory, the steps of the adaptive WKNN indoor positioning method described in any one of the first aspects are implemented.

[0020] In a fourth aspect, a computer-readable storage medium is provided for storing a computer program; when the computer program is executed by a processor, the steps of the adaptive WKNN indoor positioning method described in any one of the first aspects are implemented.

[0021] Compared with the prior art, the present invention has the following beneficial effects: On the one hand, this application introduces the use of strong and weak order invariance in similarity construction to reduce the interference of overall bias / uniform gain on matching; on the other hand, in the candidate neighborhood screening and weighting process, it combines beacon stability confidence and visibility mode gating, and suppresses the amplification effect of unstable signals through mechanisms such as adaptive K value and robustness correction, thereby improving the overall positioning accuracy and robustness in complex indoor environments. Therefore, this application can overcome device differences, perform differentiated processing on RSSI data, and reduce the interference of overall bias / uniform gain on matching, thereby improving indoor positioning accuracy and stability, and is particularly suitable for complex and changeable indoor environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a flow chart of the adaptive WKNN indoor positioning method of the present invention; Figure 2 It is a structural block diagram of the adaptive WKNN indoor positioning system of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below with reference to the accompanying drawings. The following examples are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. It should be noted that the term "comprising" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0024] Example 1 like Figure 1 As shown, an adaptive WKNN indoor positioning method is provided, and the indoor positioning method includes the following steps: S1: Collect RSSI data of all beacons at each reference point to build a fingerprint library, and obtain the confidence of the beacon at each reference point based on the stability of each beacon at the reference point.

[0025] In indoor or semi-outdoor environments where positioning is required, several Bluetooth beacon nodes (Beacons) are deployed at regular intervals. The number and location of beacons can be flexibly set based on the site area, the distribution of obstructions, and the required positioning accuracy. All beacons are set to broadcast periodically, with consistent operating frequency and transmit power to ensure comparable and reproducible RSSI values.

[0026] Multiple reference points are manually set in the target area. Each reference point has a clear two-dimensional position coordinate. The reference points can be evenly distributed according to grid density, path nodes or areas of interest.

[0027] The fingerprint library includes: fingerprint data samples of all reference points; wherein the fingerprint data samples of the reference points include: position coordinates of the reference points and RSSI vectors of the reference points.

[0028] S11: Construct the RSSI vector of the reference point, specifically: At each reference point, RSSI data is collected using a positioning terminal device (such as a smartphone). To enhance data stability and representativeness, multiple continuous RSSI samples are collected at each point (once per second for a total of 30 seconds), and the timestamp, beacon number, and reception strength value corresponding to each RSSI vector are recorded.

[0029] The reference point collects RSSI data from all beacons multiple times, performs sliding averaging and denoising filtering, and obtains the RSSI vector of the reference point. Missing values ​​in the RSSI vector can be filled with -100dBm.

[0030] S12: Get the confidence level of the beacon at each reference point, including getting the beacon b At the reference point r The confidence level is: Based on reference point r Multiple beacon collection b The variance of the RSSI data , the confidence level is obtained according to the following formula Then normalize; ; in, To prevent the denominator from being zero, a constant of 0.1dB is optionally used. 2 .

[0031] That is, in order to improve the ability to characterize the stability of each beacon at the reference point in the offline stage, the statistical RSSI variance is used in the offline stage. The higher the confidence score, the better the beacon. b At the reference point r The more reliable the signal.

[0032] For the convenience of calculation, the confidence Perform normalization transformation to obtain normalized confidence weight : ; in, is the total number of beacons.

[0033] S2: The location point collects RSSI data of each beacon in real time.

[0034] During the actual positioning process, the mobile terminal first uses the Bluetooth module to broadcast and monitor multiple beacons deployed in the environment, collecting the received signal strength indicator (RSSI) value of each beacon at the current moment. Assuming that there are M Bluetooth beacons deployed in the environment, any point to be positioned can form an RSSI vector of dimension M, which is recorded as: ; in Indicates that the point to be positioned receives the b RSSI value of each beacon, in dBm.

[0035] Considering that Bluetooth signals are susceptible to interference from factors such as multipath fading and obstruction in indoor environments, the present invention adopts a time window sampling mechanism for preprocessing to improve the stability and availability of the RSSI vector. The specific method is as follows: continuously collect n sets of RSSI data in a short period of time (such as 1-2 seconds); perform denoising on the n RSSI samples of each beacon and take the average; construct the final input vector , used for similarity calculation and positioning inference with fingerprint library samples; the default RSSI value of uncollected beacons is -100dBm.

[0036] S3: Based on the difference in beacon visibility between the point to be located and the reference point, the reference points are screened, and for each screened reference point, the comprehensive similarity between it and the point to be located is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the point to be located, the similarity between the reference point and the point to be located is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the point to be located is obtained after fusion.

[0037] Step S3 specifically includes the following sub-steps: S31: Define the point to be located x The visibility vector and reference points r The visibility vector ; ; ; in, , M is the total number of beacons, and Reference points r and the point to be located x RSSI vector of beacon b RSSI value; Since the RSSI value not collected will be assigned to -100dBm, and In the absence of beacons b The value is -100 dBm; of course, other values ​​can be assigned according to actual conditions.

[0038] S32: Obtain the point to be located by calculating the Hamming distance between the two visibility vectors x and reference points r Visibility difference , if the visibility difference Greater than the set threshold , then remove the reference point r Otherwise, keep the reference point r, complete the screening; ; in, is the Hamming distance between the visibility vector of the point to be located and the reference point, that is, is the number of different elements in the two vectors.

[0039] Alternatively, the threshold , generally takes a value of 0.3; if the visibility difference Greater than the set threshold , it proves that the point to be located is not connected to the reference point, and the reference point is eliminated. r Otherwise, it proves that the point to be located is connected to the reference point, and the reference point is retained.

[0040] By defining the visibility vector, the visibility consistency check is performed on the positioning point and the reference point, and only the spatially connected reference points are retained, thereby suppressing the "false neighbor" problem of cross-region mismatching.

[0041] S33: Constructing points to be located x With reference point r The set of commonly visible beacons ,in, and Points to be positioned x With reference point r Visibility vectors for commonly visible beacons p The weight.

[0042] S34: Calculate the reference point based on the RSSI data of all beacons received at the reference point and the point to be positioned and the confidence of the beacon at the reference point. r With the point to be positioned x The weighted Euclidean distance , and map it in the interval [0,1] to obtain the amplitude channel similarity .

[0043] Specifically, the reference point r With the point to be positioned x The weighted Euclidean distance for: ; in, and Points to be positioned x and reference points r RSSI vector of commonly visible beacons p RSSI value, Beacon p At the reference point r Normalized confidence of ; Amplitude channel similarity for: .

[0044] By calculating the amplitude channel similarity, it is possible to suppress the noise interference of high-fluctuation beacons and improve the stability of signal values.

[0045] S35: Set the points to be positioned x and reference points r China's visible beacon collection P The corresponding RSSI values ​​are sorted in descending order to obtain the values ​​of the points to be located. x and reference points r The two corresponding orders; using the Kendall rank correlation coefficient Measure the consistency between two orders and map them in the interval [0,1] to obtain the order channel similarity .

[0046] Calculation and points to be located x and reference points r The kendall rank correlation coefficient of the two corresponding orders The method can refer to the existing technology, specifically: Step 1: Generate order: and The set of commonly visible beacons P Sort the RSSI values ​​in descending order to get the order , If the interpolation between two beacons is less than the threshold (e.g. 2dB), they are considered tied and the average rank is used for calculation.

[0047] Step 2: Calculate the Kendall rank correlation coefficient :For two ordered sequences The relative order of each element in is determined to obtain the number of same-order pairs and reverse-order pairs, and the Kendall rank correlation coefficient is calculated to measure the consistency of the two sequences.

[0048] ; in, Represented as a same-order pair, the two beacons are The relative order in is consistent, Represented as a reverse order pair, the two beacons are The relative order in is inconsistent; Expressed as the total number of beacon pairs, it is calculated as: ,in is the number of visible beacons of the point to be positioned and the reference point.

[0049] Step 3: Kendall rank correlation coefficient The similarity range [-1,1] is mapped to [0,1] to obtain the order channel similarity ; .

[0050] S36: Based on preset fusion weights , obtain the points to be located by weighting x With reference point r The comprehensive similarity ; .

[0051] Fusion weight Indicates the proportion of the order channel in the comprehensive similarity, which serves as an optional fusion weight The value is 0.4. Of course, in some other embodiments, the global bias is used to perform fusion weighting. Adaptive adjustment.

[0052] S4: Adaptively select a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be located, weightedly fuse the positions of all target reference points to obtain the position of the points to be located.

[0053] Step S4 specifically includes the following sub-steps: S41: sorting the comprehensive similarities between the point to be located and each filtered reference point in descending order to obtain a descending sequence of comprehensive similarities.

[0054] S42: Set is the cumulative sum of the first K elements of the descending sequence of comprehensive similarity, For the sum of all elements of the descending sequence of comprehensive similarity, find the minimum integer K that satisfies ,in, To set the coverage weight value.

[0055] That is, the cumulative similarity of the first K reference points is required to reach the total similarity times, generally, Take the value as 0.85. Substitute K from 1 to N into the sum one by one and iterate. When the condition is met, K at this time is the minimum K that meets the condition.

[0056] ; ; S43: A boundary of K is preset. If the obtained K is within the boundary, the number of target reference points is determined to be K; if the obtained K is outside the boundary, the number of target reference points is determined to be boundary points. If the total number of elements in the descending order of comprehensive similarity is less than the minimum boundary value of K, the number of target reference points is the total number of elements in the descending order of comprehensive similarity. When the K value is too small, that is, when When , K is adjusted to .

[0057] When the K value is too large, that is, when When , K is adjusted to .

[0058] In the case of insufficient candidate points (the total number of elements in the descending sequence of comprehensive similarity is less than the minimum boundary value of K), K is taken as the number of all available points.

[0059] S44: Weight the positions of all target reference points to obtain the position of the point to be located ; ; in, is the normalized target reference point k With the point to be positioned x The comprehensive similarity value of Target reference point k The location coordinates of .

[0060] Compared with the traditional WKNN algorithm that directly uses the inverse of the Euclidean distance of K reference points as weights, the weight fusion mechanism adopted by the present invention simultaneously introduces the RSSI confidence factor and signal difference measurement, which can better adapt to the uncertainty brought about by differences in channel quality in different environments and improve positioning robustness.

[0061] Example 2 like Figure 2 As shown, an adaptive WKNN indoor positioning system includes: Several beacons are deployed in the environment where positioning is required according to the set spatial intervals; Several reference points are evenly distributed in the positioning environment according to the path nodes or grid density of the target area; Fingerprint library construction module: collects RSSI data of all beacons at each reference point to build a fingerprint library, and obtains the confidence of the beacon at each reference point based on the stability of each beacon at the reference point; Real-time acquisition module: The positioning point collects RSSI data of each beacon in real time; Comprehensive similarity acquisition module: Based on the difference in beacon visibility between the target point and the reference point, the reference point is screened and the comprehensive similarity between each screened reference point and the target point is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the target point, the similarity between the reference point and the target point is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the target point is obtained after fusion. Positioning module: Adaptively selects a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be positioned, weightedly fuses the positions of all target reference points to obtain the position of the points to be positioned.

[0062] For more specific processes of the above system, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0063] Example 3 The present invention provides a computer device, comprising a processor and a memory; wherein the processor implements the steps of the above-mentioned adaptive WKNN indoor positioning method when executing a computer program stored in the memory.

[0064] For more specific details about the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be described again here.

[0065] Example 4 The present invention provides a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, the steps of the above-mentioned adaptive WKNN indoor positioning method are implemented.

[0066] For more specific details about the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be described again here.

[0067] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. References to the same or similar parts between the various embodiments are sufficient. The systems, devices, and storage media disclosed in the embodiments are described briefly because they correspond to the methods disclosed in the embodiments. For relevant details, refer to the method description.

[0068] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software and a necessary general-purpose hardware platform. Based on this understanding, the technical solutions in the embodiments of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium such as ROM / RAM, a magnetic disk, or an optical disk, and includes instructions for enabling a computer device (such as a personal computer, server, or network device) to execute the methods described in various embodiments of the present invention, or portions thereof.

[0069] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. An adaptive WKNN indoor positioning method, wherein a plurality of beacons and reference points are deployed indoors; characterized in that: The following steps are involved: S1: Collect RSSI data of all beacons at each reference point to build a fingerprint library, and obtain the confidence of the beacon at each reference point based on the stability of each beacon at the reference point; S2: The positioning point collects RSSI data of each beacon in real time; S3: Based on the difference in beacon visibility between the target point and the reference point, the reference points are screened, and for each screened reference point, the comprehensive similarity between the reference point and the target point is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the target point, the similarity between the reference point and the target point is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the target point is obtained after fusion. S4: Adaptively select a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be located, weightedly fuse the positions of all target reference points to obtain the position of the points to be located.

2. The adaptive WKNN indoor positioning method according to claim 1, characterized in that: In step S1, the fingerprint library includes: fingerprint data samples of all reference points; wherein the fingerprint data samples of the reference points include: position coordinates of the reference points and RSSI vectors of the reference points; Constructing the RSSI vector of the reference point: Specifically, the reference point collects RSSI data of all beacons multiple times, and performs sliding average and denoising filtering processing to obtain the RSSI vector of the reference point.

3. The adaptive WKNN indoor positioning method according to claim 2, characterized in that: In step S1, the confidence level of the beacon at each reference point is obtained, including obtaining the beacon b At the reference point r Confidence, specifically: based on the reference point r Multiple beacon collection b The variance of the RSSI data , the confidence level is obtained according to the following formula Then normalize; ； in, A constant to prevent the denominator from being zero.

4. The adaptive WKNN indoor positioning method according to claim 1, characterized in that In step S3, the reference point is selected based on the difference in beacon visibility between the point to be located and the reference point, specifically: S31: Define the point to be located x The visibility vector and reference points r The visibility vector ； ； ； in, , M is the total number of beacons, and Reference points r and the point to be located x RSSI vector of beacon b RSSI value; S32: Obtain the point to be located by calculating the Hamming distance between the two visibility vectors x and reference points r Visibility difference , if the visibility difference If it is greater than the set threshold, the reference point will be eliminated r Otherwise, keep the reference point r , complete the screening; ； in, is the Hamming distance between the visibility vectors of the point to be located and the reference point.

5. The adaptive WKNN indoor positioning method according to claim 4, characterized in that: In step S3, for each reference point after screening, the comprehensive similarity between it and the point to be located is obtained, including obtaining the point to be located x With reference point r The comprehensive similarity is: S33: Constructing points to be located x With reference point r The set of commonly visible beacons ,in, and Points to be positioned x With reference point r Visibility vectors for commonly visible beacons p The weight; S34: Calculate the reference point based on the RSSI data of all beacons received at the reference point and the point to be positioned and the confidence of the beacon at the reference point. r With the point to be positioned x The weighted Euclidean distance , and map it in the interval [0,1] to obtain the amplitude channel similarity ; S35: Set the points to be positioned x and reference points r China's visible beacon collection P The corresponding RSSI values ​​are sorted in descending order to obtain the values ​​of the points to be located. x and reference points r The two corresponding orders; using the Kendall rank correlation coefficient Measure the consistency between two orders and map them in the interval [0,1] to obtain the order channel similarity ; S36: Based on preset fusion weights , obtain the points to be located by weighting x With reference point r The comprehensive similarity ; 。 6. The adaptive WKNN indoor positioning method according to claim 5, characterized in that: In step S34, the reference point r With the point to be positioned x The weighted Euclidean distance for: ,in, and Points to be positioned x and reference points r RSSI vector of commonly visible beacons p RSSI value, Beacon p At the reference point r Normalized confidence of the amplitude channel similarity for: .

7. The adaptive WKNN indoor positioning method according to claim 1, characterized in that: In step S4, specifically: S41: sorting the comprehensive similarities between the point to be located and each of the screened reference points in descending order to obtain a descending sequence of comprehensive similarities; S42: Set is the cumulative sum of the first K elements of the descending sequence of comprehensive similarity, For the sum of all elements of the descending sequence of comprehensive similarity, find the minimum integer K that satisfies ,in, To set the coverage weight value; S43: Preset the boundary of the K value. If the obtained K is within the boundary, the number of target reference points is determined to be K; if the obtained K is outside the boundary, the number of target reference points is determined to be boundary points; if the total number of elements in the descending order of the comprehensive similarity is less than the minimum boundary value of K, the number of target reference points is the total number of elements in the descending order of the comprehensive similarity; S44: Weight the positions of all target reference points to obtain the position of the point to be located ; ； in, is the normalized target reference point k With the point to be positioned x The comprehensive similarity value of Target reference point k The location coordinates of .

8. An adaptive WKNN indoor positioning system, characterized in that: include: Several beacons are deployed in the environment where positioning is required according to the set spatial intervals; Several reference points are evenly distributed in the positioning environment according to the path nodes or grid density of the target area; Fingerprint library construction module: collects RSSI data of all beacons at each reference point to build a fingerprint library, and obtains the confidence of the beacon at each reference point based on the stability of each beacon at the reference point; Real-time acquisition module: The positioning point collects RSSI data of each beacon in real time; Comprehensive similarity acquisition module: Based on the difference in beacon visibility between the target point and the reference point, the reference point is screened and the comprehensive similarity between each screened reference point and the target point is obtained. Specifically, based on the confidence of the beacon at the reference point and the RSSI data of all beacons received by the reference point and the target point, the similarity between the reference point and the target point is obtained from the amplitude and order dual channels respectively, and the comprehensive similarity between the reference point and the target point is obtained after fusion. Positioning module: Adaptively selects a set number of target reference points, and based on the comprehensive similarity between the target reference points and the points to be positioned, weightedly fuses the positions of all target reference points to obtain the position of the points to be positioned.

9. A computer device, characterized in that: The method comprises a processor and a memory; wherein, when the processor executes the computer program stored in the memory, the steps of the adaptive WKNN indoor positioning method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that Used to store a computer program; when the computer program is executed by a processor, the steps of the adaptive WKNN indoor positioning method according to any one of claims 1 to 7 are implemented.

Images