An Adaptive Indoor Fusion Location Method Based on Multiple Fingerprints

Abstract

The present invention discloses an adaptive indoor fusion positioning method based on multiple fingerprints. The method comprises: dividing an indoor environment to be positioned into a plurality of grid points at equal intervals, and arranging a plurality of base stations at equal intervals in the indoor environment; collecting the received signal strength and the corresponding coordinates at each grid point to form a received signal strength offline fingerprint database, collecting the channel impulse response and the corresponding coordinates at each grid point to form a channel impulse response offline fingerprint database, and collecting the image information and coordinates taken at each grid point to form an image offline fingerprint database; performing time domain filtering and dimensionality reduction processing on the channel impulse response data; screening the received signal strength fingerprint library according to the received signal strength data of the point to be measured, and using the received signal strength fingerprint, the channel impulse response fingerprint and the image feature fingerprint to perform positioning to obtain the position of the point to be measured.

Description

Technical Field

[0001] The present invention belongs to the field of indoor positioning, and in particular relates to an adaptive indoor fusion positioning method based on multiple fingerprints. Background Art

[0002] With the rapid development of information technology and the internet, location-based services are widely used in areas such as smart homes, car navigation, and mobile phone tracking. Global Navigation Satellite Systems (GNSS) provide users with highly accurate positioning services, largely meeting their needs for location-based services in outdoor scenarios. However, indoors, due to obstruction by buildings, GNSS signals attenuate rapidly, making them inaccurate for indoor positioning. Therefore, precise positioning in indoor environments has become a hot topic in positioning research. The research findings not only bring significant economic benefits but also enable innovation in other fields.

[0003] Existing indoor positioning technologies include Bluetooth, UWB, infrared, geomagnetic, and ZigBee. However, these technologies all have varying shortcomings and struggle to maintain accuracy in changing indoor environments. Indoor positioning based on CSI and RSS fingerprinting relies on a database comparison method, but the collected database may not encompass all indoor conditions, leading to database comparison errors. While CSI-based methods offer better overall positioning performance than RSS-based methods, this is not always the case in practical applications. Therefore, combining CSI and RSS fingerprints is necessary to improve positioning accuracy. Furthermore, both technologies are susceptible to indoor variations, reducing positioning accuracy. Here, we employ an image-based positioning approach to improve robustness. With the advancement of communication technology and the demand for wireless communication speeds, communication networks have accelerated, making it easier to achieve coordinated positioning with base stations. Fusion positioning methods have also been developed. However, most indoor positioning methods fail to effectively mitigate the impact of indoor environmental variations on positioning, lacking environmental adaptability. Furthermore, single indoor positioning methods often exhibit lower accuracy than fusion positioning.

[0004] In summary, most existing indoor positioning technologies are unable to select the best positioning method for indoor environmental changes, and the performance of their single indoor positioning method is worse than that of fusion positioning. Summary of the Invention

[0005] Purpose of the invention: The present invention proposes an adaptive indoor fusion positioning method based on multiple fingerprints. On the one hand, the use of fusion positioning can achieve higher positioning accuracy; on the other hand, the use of multiple fingerprints for adaptive positioning has better environmental adaptability.

[0006] Technical solution: The present invention proposes an adaptive indoor fusion positioning method based on multiple fingerprints, which includes the following steps:

[0007] (1) First, the scene to be located is divided into several grid points at equal intervals. The scene image information of each grid point is captured by the camera and the corresponding coordinates are sent to the server to form an image offline fingerprint database; the received signal strength of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a received signal strength offline fingerprint database; the channel impulse response of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a channel impulse response offline fingerprint database;

[0008] (2) using a terminal device to collect the received signal strength of the target point to be located, screening the received signal strength offline fingerprint database based on the received signal strength information of the point, and retaining offline fingerprint data whose correlation with the target point to be located is greater than a first threshold;

[0009] (3) Based on the received signal strength fingerprint, the K nearest neighbor algorithm is used to locate the target to be measured, and three reference points that are most likely to be the target to be measured are selected;

[0010] (4) Locate the target based on the channel impulse response fingerprint positioning method and select three reference points that are most likely to be the target points;

[0011] (5) Locate the target based on the image feature fingerprint positioning method and select three reference points that are most likely to be the target points;

[0012] (6) After completing (3)(4)(5), calculate the weights of each of the nine points according to the weight calculation method;

[0013] (7) Select the three points with the largest weights among the nine points and perform weighted averaging to obtain the final position of the test point.

[0014] Furthermore, the step (1) includes the following steps:

[0015] (1-1) Use a camera to capture image data of the scene to be located. When collecting images, four pictures are taken from the four angles of east, south, west and north at the same grid point. The coordinates of each point and the four pictures are uploaded to the server as an offline image fingerprint database.

[0016] (1-2) The receiver collects the received signal strength of each grid point and sends it to the server, which together with the coordinates of each grid point constitutes a received signal strength offline fingerprint database;

[0017] (1-3) The channel impulse response of each grid point is collected by the receiver and subjected to time domain filtering and dimensionality reduction processing, and then sent to the server together with the corresponding reference point coordinates to form a channel impulse response offline fingerprint database.

[0018] Furthermore, the step (2) includes the following steps:

[0019] (2-1) Use the terminal device to collect the time, received signal strength, channel impulse response and image parameters of the target point to be measured;

[0020] (2-2) The offline database is filtered using the received signal strength of the test point to remove redundant information in the offline database.

[0021] Furthermore, the specific method of step (2-2) is as follows:

[0022] (2-2-1) Assume that there are a total of L base stations detected during positioning, and there are n grid points. o There are n target points to be located. T , definition and They represent the received signal strength of the i-th grid point and the received signal strength of the j-th target point to be located respectively; define and They represent the base station distribution detected by the i-th grid point and the base station distribution detected by the j-th target point to be located, respectively. represents the detection index of the lth base station in the i-th measured grid point, when When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected; Represents the detection index of the lth base station in the jth target point to be located. When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected;

[0023] (2-2-2) Define two comparison indicators to filter the received signal strength offline fingerprint database. The first indicator is to calculate the received signal strength of the i-th grid point and the received signal strength of the jth target point to be located The similarity between them is defined as The second indicator is also a similarity, defined as in, Indicates the base station number that can be detected by the i-th grid point and the j-th target point to be located at the same time;

[0024] (2-2-3) The final similarity is s(i,j)=s1(i,j)s2(i,j). When locating the jth target point to be located, sort the similarities of all grid points from high to low, and select the received signal strength and corresponding coordinates of the first n grid points with the highest similarity as the refined fingerprint database, where n <n o .

[0025] Furthermore, the implementation process of step (3) is as follows:

[0026] For the lth target point to be located, calculate the Euclidean distance between its received signal strength vector and the received signal strength vector of each grid point in the fingerprint library:

[0027] Among them, n o is the number of grid points, rss i with RSS l They represent the received signal strength vector of the i-th grid point and the received signal strength vector of the l-th target point to be located, and the three grid points with the smallest Euclidean distance are selected as candidate reference points.

[0028] Furthermore, the implementation process of step (4) is as follows:

[0029] The channel impulse response is expressed as:

[0030]

[0031] Among them, τ n 、a n ,θ n Respectively represent the delay, phase, and amplitude values of the nth link path, N represents the total number of link propagation paths, and δ(τ) is the Dirac function;

[0032] The channel impulse response fingerprint library includes l i and F i {τ n ,p n ,θ n};

[0033] Among them, l i =(x i ,y i ) represents the position of the i-th grid point. The channel impulse response of each grid point is divided into N links. i is the channel impulse response fingerprint corresponding to the i-th grid point, which consists of delay, power and amplitude values, p n Indicates the transmit power of the nth link;

[0034] During positioning, the channel impulse response fingerprint feature F of the lth target point to be positioned is first collected. l {τ n ,p n ,θ n}, calculate the similarity between the channel impulse response fingerprint features of the current position and the channel impulse response fingerprint features of each grid point, and use the three grid points with the highest similarity as candidate reference points.

[0035] Furthermore, the implementation process of step (5) is as follows:

[0036] First, define the image parameters f(color, shape, angle), where color, shape, and angle represent the color, shape, and angle features of the image, respectively. When locating the target point to be located, use Yolo to identify people and objects in the image captured by the target point to be located, extract these three features from the image, and calculate the similarity Δ between the image captured by the lth target point to be located and the image of the i-th grid point:

[0037]

[0038]

[0039]

[0040] Among them, color l 、shape l and angle l are the color feature, shape feature and angle feature of the lth target point to be located respectively; n o is the total number of grid points; Δ1 is the color similarity, Δ2 is the shape similarity, and Δ3 is the angle similarity; The three grid points with the largest similarity Δ and their corresponding coordinates are selected as candidate reference points.

[0041] Furthermore, the specific method of step (6) is as follows:

[0042] (6-1) Assume that the scene to be positioned is divided into n equally spaced o grid points, and the interval between each grid point is a. When calculating the weight of each candidate reference point, a circle is drawn with the point as the center and R as the radius. The number of candidate reference points falling within the circle is calculated. The initial weight is set to 0. If the reference point falls within the circle, the weight of the point is increased by 1. If it falls on the boundary, the weight of the point is increased by 0.5. In this way, the weights of the 9 candidate reference points are obtained. The three candidate reference points with the largest weights are selected to calculate the position of the target point to be located.

[0043] (6-2) If the number of candidate reference points with the largest weight exceeds three, expand the radius of the circle, set R = R + offset, filter out points whose weights no longer increase, and recalculate the weights of each point using the method in (6-1), where offset sets the following principle: When , the candidate reference points adjacent to the candidate reference point fall into the circle, and the offset is set to when When , the candidate reference point adjacent to the candidate reference point or opposite to the rectangular corner of the candidate reference point falls into the circle, and the offset is set to when When the candidate reference point within two adjacent grids of the candidate reference point or the candidate reference point opposite to the rectangular corner of the candidate reference point falls into the circle, the offset is set to

[0044] Furthermore, the implementation process of step (7) is as follows: P1, P2, and P3 are the candidate reference point positions with the largest weights in the three positioning methods, and their coordinates are (x1, y1), (x2, y2), and (x3, y3), respectively. The estimated position P4 (x4, y4) is obtained by weighted average of the above three points:

[0045]

[0046]

[0047] Beneficial effects: Compared with the prior art, the present invention can adaptively obtain the best positioning method in different indoor environments or variable indoor environments by using multiple different types of fingerprint positioning methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a framework for adaptive indoor fusion positioning method based on multiple fingerprints;

[0049] Figure 2 Schematic diagram of weight distribution when fusion of results;

[0050] Figure 3 This is the position estimation diagram in the positioning stage. DETAILED DESCRIPTION

[0051] The technical solution of the invention is described in detail below with reference to the accompanying drawings:

[0052] The present invention proposes an adaptive indoor fusion positioning method based on multiple fingerprints, which includes the following steps:

[0053] (1) First, the scene to be located is divided into several grid points at equal intervals. The scene image information of each grid point is captured by the camera and the corresponding coordinates are sent to the server to form an image offline fingerprint database; the received signal strength of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a received signal strength offline fingerprint database; the channel impulse response of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a channel impulse response offline fingerprint database;

[0054] (2) using a terminal device to collect the received signal strength of the target point to be located, screening the received signal strength offline fingerprint database based on the received signal strength information of the point, and retaining offline fingerprint data whose correlation with the target point to be located is greater than a first threshold;

[0055] (3) Based on the received signal strength fingerprint, the K nearest neighbor algorithm is used to locate the target to be measured, and three reference points that are most likely to be the target to be measured are selected;

[0056] (4) Locate the target based on the channel impulse response fingerprint positioning method and select three reference points that are most likely to be the target points;

[0057] (5) Locate the target based on the image feature fingerprint positioning method and select three reference points that are most likely to be the target points;

[0058] (6) After completing (3)(4)(5), calculate the weights of each of the nine points according to the weight calculation method;

[0059] (7) Select the three points with the largest weights among the nine points and perform weighted averaging to obtain the final position of the test point.

[0060] The step (1) comprises the following steps:

[0061] (1-1) Use a camera to capture image data of the scene to be located. When collecting images, four pictures are taken from the four angles of east, south, west and north at the same grid point. The coordinates of each point and the four pictures are uploaded to the server as an offline image fingerprint database.

[0062] (1-2) The receiver collects the received signal strength of each grid point and sends it to the server, which together with the coordinates of each grid point constitutes a received signal strength offline fingerprint database;

[0063] (1-3) The channel impulse response of each grid point is collected by the receiver and subjected to time domain filtering and dimensionality reduction processing, and then sent to the server together with the corresponding reference point coordinates to form a channel impulse response offline fingerprint database.

[0064] The step (2) comprises the following steps:

[0065] (2-1) Use the terminal device to collect the time, received signal strength, channel impulse response and image parameters of the target point to be measured;

[0066] (2-2) The offline database is filtered using the received signal strength of the test point to remove redundant information in the offline database.

[0067] The specific method of step (2-2) is as follows:

[0068] (2-2-1) Assume that there are a total of L base stations detected during positioning, and there are n grid points. o There are n target points to be located. T , definition and They represent the received signal strength of the i-th grid point and the received signal strength of the j-th target point to be located respectively; define and They represent the base station distribution detected by the i-th grid point and the base station distribution detected by the j-th target point to be located, respectively. represents the detection index of the lth base station in the i-th measured grid point, when When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected; Represents the detection index of the lth base station in the jth target point to be located. When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected;

[0069] (2-2-2) Define two comparison indicators to filter the received signal strength offline fingerprint database. The first indicator is to calculate the received signal strength of the i-th grid point and the received signal strength of the jth target point to be located The similarity between them is defined as The second indicator is also a similarity, defined as in, Indicates the base station number that can be detected by the i-th grid point and the j-th target point to be located at the same time;

[0070] (2-2-3) The final similarity is s(i,j)=s1(i,j)s2(i,j). When locating the jth target point to be located, sort the similarities of all grid points from high to low, and select the received signal strength and corresponding coordinates of the first n grid points with the highest similarity as the refined fingerprint database, where n <no .

[0071] The implementation process of step (3) is as follows:

[0072] For the lth target point to be located, calculate the Euclidean distance between its received signal strength vector and the received signal strength vector of each grid point in the fingerprint library:

[0073] Among them, n o is the number of grid points, rss i with RSS l They represent the received signal strength vector of the i-th grid point and the received signal strength vector of the l-th target point to be located, and the three grid points with the smallest Euclidean distance are selected as candidate reference points.

[0074] The implementation process of step (4) is as follows:

[0075] The channel impulse response is expressed as:

[0076]

[0077] Among them, τ n 、a n ,θ n Respectively represent the delay, phase, and amplitude values of the nth link path, N represents the total number of link propagation paths, and δ(τ) is the Dirac function;

[0078] The channel impulse response fingerprint library includes l i and F i {τ n ,p n ,θ n};

[0079] Among them, l i =(x i ,y i ) represents the position of the i-th grid point. The channel impulse response of each grid point is divided into N links. i is the channel impulse response fingerprint corresponding to the i-th grid point, which consists of delay, power and amplitude values, p n Indicates the transmit power of the nth link;

[0080] During positioning, the channel impulse response fingerprint feature F of the lth target point to be positioned is first collected. l {τ n ,p n ,θ n}, calculate the similarity between the channel impulse response fingerprint features of the current position and the channel impulse response fingerprint features of each grid point, and use the three grid points with the highest similarity as candidate reference points.

[0081] The implementation process of step (5) is as follows:

[0082] First, define the image parameters f(color, shape, angle), where color, shape, and angle represent the color, shape, and angle features of the image, respectively. When locating the target point to be located, use Yolo to identify people and objects in the image captured by the target point to be located, extract these three features from the image, and calculate the similarity Δ between the image captured by the lth target point to be located and the image of the i-th grid point:

[0083]

[0084]

[0085]

[0086] Among them, color l 、shape l and angle l are the color feature, shape feature and angle feature of the lth target point to be located respectively; n o is the total number of grid points; Δ1 is the color similarity, Δ2 is the shape similarity, and Δ3 is the angle similarity; The three grid points with the largest similarity Δ and their corresponding coordinates are selected as candidate reference points.

[0087] The specific method of step (6) is as follows:

[0088] (6-1) Assume that the scene to be positioned is divided into n equally spaced o grid points, and the interval between each grid point is a. When calculating the weight of each candidate reference point, a circle is drawn with the point as the center and R as the radius. The number of candidate reference points falling within the circle is calculated. The initial weight is set to 0. If the reference point falls within the circle, the weight of the point is increased by 1. If it falls on the boundary, the weight of the point is increased by 0.5. In this way, the weights of the 9 candidate reference points are obtained. The three candidate reference points with the largest weights are selected to calculate the position of the target point to be located.

[0089] (6-2) If the number of candidate reference points with the largest weight exceeds three, expand the radius of the circle, set R = R + offset, filter out points whose weights no longer increase, and recalculate the weights of each point using the method in (6-1), where offset sets the following principle: When , the candidate reference points adjacent to the candidate reference point fall into the circle, and the offset is set to when When , the candidate reference point adjacent to the candidate reference point or opposite to the rectangular corner of the candidate reference point falls into the circle, and the offset is set to when When the candidate reference point within two adjacent grids of the candidate reference point or the candidate reference point opposite to the rectangular corner of the candidate reference point falls into the circle, the offset is set to

[0090] The implementation process of step (7) is as follows: P1, P2, and P3 are the candidate reference point positions with the largest weights in the three positioning methods, and their coordinates are (x1, y1), (x2, y2), and (x3, y3), respectively. The estimated position P4 (x4, y4) is obtained by weighted average of the above three points:

[0091]

[0092]

[0093] The above is a detailed introduction to an adaptive indoor fusion positioning method based on multiple fingerprints provided in an embodiment of the present invention. For those skilled in the art, according to the ideas of the embodiments of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. An adaptive indoor fusion positioning method based on multiple fingerprints, characterized in that: The method comprises the following steps: (1) First, the scene to be located is divided into several grid points at equal intervals. The scene image information of each grid point is captured by the camera and the corresponding coordinates are sent to the server to form an image offline fingerprint database; the received signal strength of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a received signal strength offline fingerprint database; the channel impulse response of each grid point is collected by the receiver and the corresponding coordinates are sent to the server to form a channel impulse response offline fingerprint database; (2) using a terminal device to collect the received signal strength of the target point to be located, screening the received signal strength offline fingerprint database based on the received signal strength information of the point, and retaining offline fingerprint data whose correlation with the target point to be located is greater than a first threshold; (3) Based on the received signal strength fingerprint, the K nearest neighbor algorithm is used to locate the target to be measured, and three reference points that are most likely to be the target to be measured are selected; (4) Locate the target based on the channel impulse response fingerprint positioning method and select three reference points that are most likely to be the target points; (5) Locate the target based on the image feature fingerprint positioning method and select three reference points that are most likely to be the target points; (6) After completing (3)(4)(5), calculate the weights of each of the nine points according to the weight calculation method; (7) Select the three points with the largest weights among the nine points and perform weighted averaging to obtain the final position of the test point.

2. The adaptive indoor fusion positioning method based on multiple fingerprints according to claim 1 is characterized in that: The step (1) comprises the following steps: (1-1) Use a camera to capture image data of the scene to be located. When collecting images, four pictures are taken from the four angles of east, south, west and north at the same grid point. The coordinates of each point and the four pictures are uploaded to the server as an offline image fingerprint database. (1-2) The receiver collects the received signal strength of each grid point and sends it to the server, which together with the coordinates of each grid point constitutes a received signal strength offline fingerprint database; (1-3) The channel impulse response of each grid point is collected by the receiver and subjected to time domain filtering and dimensionality reduction processing, and then sent to the server together with the corresponding reference point coordinates to form a channel impulse response offline fingerprint database.

3. The adaptive indoor fusion positioning method based on multiple fingerprints according to claim 1 is characterized in that: The step (2) comprises the following steps: (2-1) Use the terminal device to collect the time, received signal strength, channel impulse response and image parameters of the target point to be measured; (2-2) The offline database is filtered using the received signal strength of the test point to remove redundant information in the offline database.

4. The adaptive indoor fusion positioning method based on multiple fingerprints according to claim 3 is characterized in that: The specific method of step (2-2) is as follows: (2-2-1) Assume that there are a total of L base stations detected during positioning, and there are n grid points. o There are n target points to be located. T , definition and They represent the received signal strength of the i-th grid point and the received signal strength of the j-th target point to be located respectively; define and They represent the base station distribution detected by the i-th grid point and the base station distribution detected by the j-th target point to be located, respectively. represents the detection index of the lth base station in the i-th measured grid point, when When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected; Represents the detection index of the lth base station in the jth target point to be located. When there is a value, detect the indicator If it is 1, it means the first base station is detected. Otherwise, If it is 0, it means that the first base station is not detected; (2-2-2) Define two comparison indicators to filter the received signal strength offline fingerprint database. The first indicator is to calculate the received signal strength of the i-th grid point and the received signal strength of the jth target point to be located The similarity between them is defined as The second indicator is also a similarity, defined as in, Indicates the base station number that can be detected by the i-th grid point and the j-th target point to be located at the same time; (2-2-3) The final similarity is s(i,j)=s1(i,j)s2(i,j). When locating the jth target point to be located, sort the similarities of all grid points from high to low, and select the received signal strength and corresponding coordinates of the first n grid points with the highest similarity as the refined fingerprint database, where n <n o .

5. The adaptive indoor fusion positioning method based on multiple fingerprints according to claim 1 is characterized in that: The implementation process of step (3) is as follows: For the jth target point to be located, calculate the Euclidean distance between its received signal strength vector and the received signal strength vector of each grid point in the fingerprint library: Among them, n o is the number of grid points, rss i with RSS j They represent the received signal strength vector of the i-th grid point and the received signal strength vector of the j-th target point to be located, and the three grid points with the smallest Euclidean distance are selected as candidate reference points.

6. The adaptive indoor fusion positioning method based on multiple fingerprints according to claim 1, characterized in that: The implementation process of step (4) is as follows: The channel impulse response is expressed as: Among them, τ n 、a n ,θ n Respectively represent the delay, phase, and amplitude of the nth link path, N represents the total number of link propagation paths, δ(t-τ n ) is the Dirac function; The channel impulse response fingerprint library includes l i and F i {τ n ,p n ,θ n }; Among them, l i =(x i ,y i ) represents the position of the i-th grid point. The channel impulse response of each grid point is divided into N links. i is the channel impulse response fingerprint corresponding to the i-th grid point, which consists of delay, power and amplitude values, p n Indicates the transmit power of the nth link; During positioning, the channel impulse response fingerprint feature F of the jth target point to be positioned is first collected. j {τ n ,p n ,θ n }, calculate the similarity between the channel impulse response fingerprint features of the current position and the channel impulse response fingerprint features of each grid point, and use the three grid points with the highest similarity as candidate reference points.

7. The adaptive indoor fusion positioning method based on dynamic environment according to claim 1, characterized in that: The implementation process of step (5) is as follows: First, define the image parameters f(color, shape, angle), where color, shape, and angle represent the color, shape, and angle features of the image, respectively. When locating the target point to be located, use Yolo to identify people and objects in the image captured by the target point to be located, extract these three features from the image, and calculate the similarity Δ between the image captured by the jth target point to be located and the image of the i-th grid point: Among them, color j 、shape j and angle j are the color feature, shape feature and angle feature of the jth target point to be located; n o is the total number of grid points; Δ1 is the color similarity, Δ2 is the shape similarity, and Δ3 is the angle similarity; The three grid points with the largest similarity Δ and their corresponding coordinates are selected as candidate reference points.

8. The adaptive indoor fusion positioning method based on dynamic environment according to claim 7, characterized in that: The specific method of step (6) is as follows: (6-1) Assume that the interval between each grid point in the scene to be located is a. When calculating the weight of each candidate reference point, draw a circle with the point as the center and R as the radius. Count the number of candidate reference points that fall within the circle. The initial weight is set to 0. If the reference point falls within the circle, the weight of the point is increased by 1. If it falls on the boundary, the weight of the point is increased by 0.

5. In this way, the weights of the 9 candidate reference points are obtained. The three candidate reference points with the largest weights are selected to calculate the position of the target point to be located. (6-2) If the number of candidate reference points with the largest weight exceeds three, expand the radius of the circle, set R = R + offset, filter out points whose weights no longer increase, and recalculate the weights of each point using the method in (6-1), where offset sets the following principle: When , the candidate reference points adjacent to the candidate reference point fall into the circle, and the offset is set to when When , the candidate reference point adjacent to the candidate reference point or opposite to the rectangular corner of the candidate reference point falls into the circle, and the offset is set to when When the candidate reference point within two adjacent grids of the candidate reference point or the candidate reference point opposite to the rectangular corner of the candidate reference point falls into the circle, the offset is set to 9. The adaptive indoor fusion positioning method based on dynamic environment according to claim 1, characterized in that: The implementation process of step (7) is as follows: P1, P2, and P3 are the candidate reference point positions with the largest weights in the three positioning methods, and their coordinates are (x 11 ,y 11 )、(x 22 ,y 22 )、(x 33 ,y 33 ), then the position to be estimated P4(x 44 ,y 44 ) is obtained by weighted average of the above three points:

Images