An indoor floor positioning method, system, electronic device and storage medium

Abstract

The present invention relates to an indoor floor positioning method, system, electronic device, and storage medium, belonging to the technical field of indoor floor positioning. In the offline phase, a fingerprint database is used to train a wavelet neural network to obtain an indoor floor position recognition model. In the online phase, the indoor Wi-Fi signal strength of the missing signal in the signal fingerprint matrix of the positioning point is dynamically compensated. The dynamically compensated signal fingerprint matrix is then input into the trained wavelet neural network, and the building floor position is output. Based on signal dynamic compensation and wavelet neural networks, the present invention reduces the impact of missing signals on floor positioning accuracy, effectively improving the accuracy and stability of building floor position recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor floor positioning, and in particular to an indoor floor positioning method, system, electronic equipment and storage medium. Background Art

[0002] With the rapid development and widespread adoption of mobile internet technology, the positioning service industry is booming, and demand for location-based services is increasing dramatically. For modern buildings, accurately locating a user's vertical and horizontal position is equally important. In high-rise buildings, floor identification is the prerequisite and foundation for indoor two-dimensional positioning. Accurate floor determination effectively reduces the search space in the matching phase, thereby improving positioning accuracy and reducing computational overhead. Therefore, achieving precise indoor floor positioning has significant research significance and practical value.

[0003] Wireless LAN fingerprinting is a promising and widely adopted method for indoor floor positioning, where the fingerprint database is crucial for fingerprint positioning. However, when establishing a fingerprint database, an excessive number of wireless access points can reduce positioning efficiency. Traditional fingerprint algorithms employ complex filtering and time-consuming manual parameter adjustment, which is very time-consuming and unsuitable for large-scale indoor environments. In recent years, researchers have adopted deep learning methods for indoor positioning. Considering the difficulties in collecting indoor environmental data and the impact of RSSI missingness on fingerprint positioning, some scholars have proposed three different RSSI data enhancement methods based on multi-output Gaussian processes for multi-building and multi-floor indoor positioning. These methods also comprehensively process the correlation between RSSI observations from multiple access points, improving positioning accuracy and stability to a certain extent.

[0004] However, traditional fingerprint algorithms require complex data pre / post-processing and a large number of parameter adjustments, and the positioning accuracy and stability will be significantly reduced when the signal is missing. Summary of the Invention

[0005] The purpose of the present invention is to provide an indoor floor positioning method, system, electronic device and storage medium to reduce the impact of signal loss on floor positioning accuracy and effectively improve the accuracy and stability of building floor position recognition.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An indoor floor positioning method, comprising:

[0008] Constructing a fingerprint database; the fingerprint database includes a spatial position vector of each positioning sampling point and a signal fingerprint matrix of each positioning sampling point; the spatial position vector includes the building number, floor number, and specific location of the positioning sampling point; the signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points;

[0009] The fingerprint database is used to train the wavelet neural network to obtain the indoor floor position recognition model;

[0010] Obtain all indoor Wi-Fi signal strengths collected at the point to be located and construct a signal fingerprint matrix of the point to be located;

[0011] The fingerprint database is used to compensate for the indoor Wi-Fi signal strengths of the wireless access points that are missing from the signal fingerprint matrix of the target location point, to obtain a compensated signal fingerprint matrix; the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n;

[0012] The compensated signal fingerprint matrix is input into the indoor floor position recognition model, and the building number, floor number and specific location of the building where the point to be located is located are output.

[0013] Optionally, the building of a fingerprint database specifically includes:

[0014] At each positioning sampling point, each wireless access point is sampled 10 times, and the average of the 10 indoor Wi-Fi signal strengths obtained by sampling is calculated;

[0015] Eliminate the indoor Wi-Fi signal strengths whose difference from the average value is greater than 10% of the average value, and average the remaining indoor Wi-Fi signal strengths as the indoor Wi-Fi signal strength value of each wireless access point;

[0016] The spatial position vector of each positioning sampling point and the indoor Wi-Fi signal strength values of all wireless access points collected at each positioning sampling point form a fingerprint record;

[0017] The fingerprint records of all positioning sampling points constitute the fingerprint database D = (Lp, Fp), p = 1, 2, ..., q; q is the number of positioning sampling points, Lp is the spatial position vector of the p-th positioning sampling point, and Fp is the signal fingerprint matrix of the p-th positioning sampling point; MAC r is the rth physical address of the pth positioning sampling point, MAC collected multiple times r The corresponding average indoor Wi-Fi signal strength, r = 1, 2, ..., s, where s is the number of indoor Wi-Fi signals collected at the p-th positioning sampling point.

[0018] Optionally, the wavelet neural network includes: an input layer, a hidden layer and an output layer;

[0019] The number of neurons in the input layer is n, the number of neurons in the hidden layer is k, and the number of neurons in the output layer is m;

[0020] The Morlet wavelet function is selected as the transfer function of the hidden layer nodes;

[0021] The mathematical expression of the input function of the hidden layer is Where h in (k) is the input value of the hidden layer, ω ik is the network weight from the input layer to the hidden layer, x i is the indoor WIFI signal strength input by the i-th neuron in the input layer, i≤n;

[0022] The input value h of the hidden layer in (k) The mathematical expression after translation and scaling is: Where h is the value after translation and scaling, b k is the translation factor of the wavelet basis function, a k is the scaling factor of the wavelet basis function;

[0023] The mathematical expression of the output value of the hidden layer is Where h out (k) is the output value of the hidden layer;

[0024] The mathematical expression of the output prediction value of the output layer is: Where y j is the j-th predicted value output by the output layer, ω kj is the network weight from the hidden layer to the output layer, h out (l) is the output value of the lth neuron in the hidden layer.

[0025] Optionally, the signal fingerprint matrix of the point to be located is expressed as

[0026] Where f is the signal fingerprint matrix of the point to be located, mac1, mac2, and mact are the 1st, 2nd, and tth physical addresses of the point to be located, respectively; rss1, rss2, and rsst are the indoor Wi-Fi signal strengths corresponding to the 1st, 2nd, and tth physical addresses collected at the point to be located, respectively.

[0027] Optionally, using the fingerprint database to compensate for the indoor Wi-Fi signal strength of the wireless access point that is missing from the signal fingerprint matrix of the target location point to obtain a compensated signal fingerprint matrix specifically includes:

[0028] The positioning sampling point closest to the point to be positioned is determined as the compensation point;

[0029] Determine the missing physical address of the point to be located relative to the compensation point and the indoor Wi-Fi signal strength corresponding to the physical address, and add them to the signal fingerprint matrix of the point to be located to obtain the compensated signal fingerprint matrix; the physical address corresponds one-to-one with the wireless access point.

[0030] Optionally, the compensated signal fingerprint matrix is input into the indoor floor position recognition model to output the building number, floor number and specific location of the point to be located, and then the following steps are further included:

[0031] Determine the union of the physical addresses in the signal fingerprint matrix of the compensation point and the signal fingerprint matrix of the point to be located, and use the signal fingerprint matrix of the compensation point as the reference fingerprint and the signal fingerprint matrix of the point to be located as the test fingerprint;

[0032] Select the minimum indoor Wi-Fi signal strength RSSmin in the reference fingerprint and the minimum indoor Wi-Fi signal strength rssmin in the test fingerprint respectively;

[0033] Initialize w to 1;

[0034] If the reference fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint does not change;

[0035] If the reference fingerprint does not contain the w-th physical address in the union, assign rssmin the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint;

[0036] If the test fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint will not be changed;

[0037] If the test fingerprint does not contain the w-th physical address in the union, use RSSmin to assign the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint;

[0038] Increase the value of w by 1 and return to step "If the reference fingerprint contains the w-th physical address in the union, do not change the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint" until the value of w is greater than the total number of physical addresses in the union. Then stop the loop and obtain the updated fingerprint database.

[0039] The updated fingerprint database is used to replace the fingerprint database, and the process returns to step "using the fingerprint database to train the wavelet neural network and obtain an indoor floor position recognition model".

[0040] An indoor floor positioning system, comprising:

[0041] A database construction module is used to construct a fingerprint database; the fingerprint database includes a spatial position vector of each positioning sampling point and a signal fingerprint matrix of each positioning sampling point; the spatial position vector includes the building number, floor number and specific location of the positioning sampling point; the signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points;

[0042] Model training module, used to train the wavelet neural network using the fingerprint database to obtain the indoor floor position recognition model;

[0043] The real-time acquisition module is used to obtain the strength of all indoor Wi-Fi signals collected at the point to be located and form a signal fingerprint matrix of the point to be located;

[0044] a compensation module, configured to compensate for the indoor Wi-Fi signal strengths of the wireless access points missing from the signal fingerprint matrix of the target location point using the fingerprint database, to obtain a compensated signal fingerprint matrix; wherein the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n;

[0045] The recognition module is used to input the compensated signal fingerprint matrix into the indoor floor position recognition model, and output the building number, floor number and specific location of the building where the point to be located is located.

[0046] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the indoor floor positioning method as described above is implemented.

[0047] A computer-readable storage medium stores a computer program, which implements the aforementioned indoor floor positioning method when executed.

[0048] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0049] The present invention discloses a method, system, electronic device, and storage medium for indoor floor positioning. In an offline phase, a fingerprint database is used to train a wavelet neural network to obtain an indoor floor location recognition model. In an online phase, the signal fingerprint matrix of the positioning point is dynamically compensated for the indoor Wi-Fi signal strength at locations where the signal is missing. The dynamically compensated signal fingerprint matrix is then input into the trained wavelet neural network, which outputs the building floor location. Based on dynamic signal compensation and a wavelet neural network, the present invention reduces the impact of missing signals on floor positioning accuracy, effectively improving the accuracy and stability of building floor location recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 A flowchart of an indoor floor positioning method provided by an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of an indoor floor positioning method provided by an embodiment of the present invention;

[0053] Figure 3 This is a structural diagram of a wavelet neural network provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] The purpose of the present invention is to provide an indoor floor positioning method, system, electronic device and storage medium to reduce the impact of signal loss on floor positioning accuracy and effectively improve the accuracy and stability of building floor position recognition.

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0057] To eliminate the complex pre- and post-processing of data and the need for fewer parameter adjustments in traditional fingerprint algorithms, the present invention provides an indoor floor positioning method that addresses signal loss. This method is divided into offline and online phases. In the offline phase, a wavelet neural network is trained using actual fingerprint data, such as indoor Wi-Fi signal strength. This training generates an indoor floor location recognition model, which is then used to identify indoor floors. In the online phase, the RSS of missing signals in the fingerprint data is dynamically compensated. The compensated fingerprint data is then used to update the original fingerprint library data. The compensated test data is then input into the trained wavelet neural network, and the building floor location is obtained from the wavelet neural network output.

[0058] Reference Figure 1As shown, the indoor floor positioning method provided by the present invention specifically includes the following steps:

[0059] Step S1: constructing a fingerprint database.

[0060] During the offline phase, a fingerprint database is constructed. The fingerprint database includes the spatial position vector of each positioning sampling point and the signal fingerprint matrix of each positioning sampling point. The spatial position vector includes the building number, floor number, and specific floor location of the positioning sampling point. The signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points.

[0061] For example, Figure 2 The four steps of constructing a fingerprint database are to obtain RSS (Received Signal Strength) fingerprint data, process data, match spatially, and build a fingerprint library. The specific construction process is as follows: at each positioning sampling point, sample each wireless access point 10 times and calculate the average value of the 10 indoor WIFI signal strengths obtained by sampling; eliminate the indoor WIFI signal strengths whose difference from the average value is greater than 10% of the average value, and average the remaining indoor WIFI signal strengths as the indoor WIFI signal strength value of each wireless access point; combine the spatial position vector of each positioning sampling point and the indoor WIFI signal strength values of all wireless access points collected by each positioning sampling point into a fingerprint record; the fingerprint records of all positioning sampling points constitute the fingerprint database. The fingerprint database constructed in the offline stage is represented as

[0062] D=(Lp,Fp), p=1,2,...,q

[0063] Where q is the number of positioning sampling points, Lp is the spatial position vector of the p-th positioning sampling point, and Fp is the signal fingerprint matrix of the p-th positioning sampling point;

[0064]

[0065] MAC r is the rth physical address of the pth positioning sampling point, MAC collected multiple times r The corresponding average indoor Wi-Fi signal strength, r = 1, 2, ..., s, where s is the number of indoor Wi-Fi signals collected at the p-th positioning sampling point.

[0066] Table 1 shows an example of a specific format of a fingerprint database. X in Table 1 indicates that the AP is not connected and has no RSS value.

[0067] Table 1 Fingerprint database

[0068]

[0069]

[0070] Step S2: Using the fingerprint database to train the wavelet neural network to obtain an indoor floor position recognition model.

[0071] The basic structure of wavelet neural network is as follows Figure 3 As shown in the figure, it consists of three layers: input layer, hidden layer (middle layer) and output layer. The input layer has n neurons, and the input signal is the characteristic fingerprint data x1, x2, ... x of the indoor WIFI signal strength (RSS). n , which correspond to the RSS values of each WIFI in the building collected at the sampling location. The number of WIFI used for positioning can be selected according to the building volume. The output layer has m neurons, and the output signal is the building number, floor number and specific location y1, y2, ... y m Preferably, m=5, y1-y2 correspond to the building number and floor number, and y3-y5 correspond to the spatial coordinates. ik is the network weight from the input layer to the middle layer, ω kj are the network weights from the middle layer to the output layer. i = 1, 2, ... n; j = 1, 2, ... m. The number of neurons in the middle layer is k.

[0072] The present invention selects Morlet wavelet function as the transfer function of the middle layer node, and selects wavelet function. After the network weight calculation, the mathematical expression of the input function of the middle layer is formula (1), where the input value h of the middle layer is in The mathematical expression of (k) after translation and expansion is formula (2). k is the translation factor of the wavelet basis function, a k is the scaling factor of the wavelet basis function. The mathematical expression of the output value of the middle layer (hidden layer) is formula (3), and the function is the Morlet wavelet function. The output value h of the middle layer (hidden layer) is out (k) is used as the input signal and input to the output layer. According to the output prediction value mathematical expression (4), the floor position information y1, y2, ... y m .

[0073]

[0074]

[0075]

[0076]

[0077] In formula (1), hin (k) is the input value of the middle layer, ω ik is the network weight from the input layer to the middle layer, x i is the indoor WIFI signal strength input by the i-th neuron in the input layer, i≤n. In formula (4), y j is the j-th predicted value output by the output layer, h out (l) is the output value of the lth neuron in the hidden layer, ω lj is the network weight from the lth neuron in the hidden layer to the output layer.

[0078] A large amount of measured fingerprint data is input into the neural network for training to obtain an indoor floor position recognition model.

[0079] Step S3: Obtain all indoor Wi-Fi signal strengths collected at the point to be located and form a signal fingerprint matrix of the point to be located.

[0080] The fingerprint data f of the point to be located collected during the online phase is called the test fingerprint. As shown in the following formula, mac and rss represent the physical address and received signal strength, respectively, and the number of collected AP (Wireless Access Point) signals is t.

[0081]

[0082] Where f is the signal fingerprint matrix of the point to be located, mac1, mac2, and mact are the 1st, 2nd, and tth physical addresses of the point to be located, respectively; rss1, rss2, and rsst are the indoor Wi-Fi signal strengths corresponding to the 1st, 2nd, and tth physical addresses collected at the point to be located, respectively.

[0083] Step S4: Using the fingerprint database, the indoor Wi-Fi signal strengths of the wireless access points that are missing from the signal fingerprint matrix of the target location are compensated to obtain a compensated signal fingerprint matrix; the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n.

[0084] Since the input layer of the wavelet neural network has n neurons, the online stage (real-time positioning) needs to collect the RSS values of n APs at the point to be located. However, some APs are not connected and have no RSS values. It is necessary to use the data in the fingerprint database to compensate for each other and complete the RSS values of the n APs in the signal fingerprint matrix of the point to be located.

[0085] In one example, the compensation process is as follows: the positioning sampling point closest to the point to be positioned is determined as the compensation point; the physical address of the point to be positioned relative to the compensation point and the indoor Wi-Fi signal strength corresponding to the physical address are determined, and added to the signal fingerprint matrix of the point to be positioned to obtain the compensated signal fingerprint matrix; the physical address corresponds one-to-one to the wireless access point.

[0086] Step S5: input the compensated signal fingerprint matrix into an indoor floor position recognition model, and output the building number, floor number, and specific location of the floor where the point to be located is located.

[0087] After step S5, the fingerprint database data is also dynamically updated. The detailed process of dynamic update is as follows:

[0088] ① Determine the union of the physical addresses in the signal fingerprint matrix of the compensation point and the signal fingerprint matrix of the point to be located, and use the signal fingerprint matrix of the compensation point as the reference fingerprint and the signal fingerprint matrix of the point to be located as the test fingerprint. Let Umac be the union of the physical addresses in the reference fingerprint and the test fingerprint, the number of physical addresses is u, and the value range is [max(m, t), m+t];

[0089] ② Select the minimum indoor Wi-Fi signal strength RSSmin in the reference fingerprint and the minimum indoor Wi-Fi signal strength rssmin in the test fingerprint respectively;

[0090] ③ Initialize w to 1;

[0091] ④ If the reference fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint does not change;

[0092] ⑤ If the reference fingerprint does not contain the w-th physical address in the union, assign rssmin the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint;

[0093] ⑥ If the test fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint will not be changed;

[0094] ⑦ If the test fingerprint does not contain the w-th physical address in the union, use RSSmin to assign the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint;

[0095] ⑧Increase the value of w by 1 and return to step ④ until the value of w is greater than the total number of physical addresses in the union. Then stop the loop and obtain the updated fingerprint database.

[0096] ⑨ Replace the fingerprint database with the updated fingerprint database and return to step S2.

[0097] The physical addresses in Umac are traversed to update the fingerprint database, and the wavelet neural network is retrained using the updated fingerprint database.

[0098] The indoor floor positioning method of RSS minimum value dynamic compensation and wavelet neural network proposed in the present invention reduces the influence of signal loss on floor positioning accuracy, and effectively improves the accuracy and stability of building floor position recognition.

[0099] In addition, corresponding to the indoor floor positioning method provided above, the present invention also provides an indoor floor positioning system, comprising:

[0100] A database construction module is used to construct a fingerprint database; the fingerprint database includes a spatial position vector of each positioning sampling point and a signal fingerprint matrix of each positioning sampling point; the spatial position vector includes the building number, floor number and specific location of the positioning sampling point; the signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points;

[0101] Model training module, used to train the wavelet neural network using the fingerprint database to obtain the indoor floor position recognition model;

[0102] The real-time acquisition module is used to obtain the strength of all indoor Wi-Fi signals collected at the point to be located and form a signal fingerprint matrix of the point to be located;

[0103] a compensation module, configured to compensate for the indoor Wi-Fi signal strengths of the wireless access points missing from the signal fingerprint matrix of the target location point using the fingerprint database, to obtain a compensated signal fingerprint matrix; wherein the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n;

[0104] The recognition module is used to input the compensated signal fingerprint matrix into the indoor floor position recognition model, and output the building number, floor number and specific location of the building where the point to be located is located.

[0105] The multi-indoor floor positioning system provided by the embodiment of the present invention has similar working principles and beneficial effects to the indoor floor positioning method described in the above embodiment, so it will not be described in detail here. For specific details, please refer to the introduction of the above method embodiment.

[0106] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the aforementioned indoor floor positioning method when executing the computer program.

[0107] In addition, when the computer program in the above-mentioned memory is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.

[0108] Furthermore, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the aforementioned indoor floor positioning method when the computer program is executed.

[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0110] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. An indoor floor positioning method, characterized in that: include: Constructing a fingerprint database; the fingerprint database includes a spatial position vector of each positioning sampling point and a signal fingerprint matrix of each positioning sampling point; the spatial position vector includes the building number, floor number, and specific location of the positioning sampling point; the signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points; The fingerprint database is used to train the wavelet neural network to obtain the indoor floor position recognition model; Obtain all indoor Wi-Fi signal strengths collected at the point to be located and construct a signal fingerprint matrix of the point to be located; The fingerprint database is used to compensate for the indoor Wi-Fi signal strengths of the wireless access points that are missing from the signal fingerprint matrix of the target location point, to obtain a compensated signal fingerprint matrix; the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n; Input the compensated signal fingerprint matrix into the indoor floor position recognition model, and output the building number, floor number and specific location of the building where the point to be located is located; The wavelet neural network comprises: an input layer, a hidden layer and an output layer; The number of neurons in the input layer is n, the number of neurons in the hidden layer is k, and the number of neurons in the output layer is m; The Morlet wavelet function is selected as the transfer function of the hidden layer nodes; The mathematical expression of the input function of the hidden layer is Where h in (k) is the input value of the hidden layer, ω ik is the network weight from the input layer to the hidden layer, x i is the indoor WIFI signal strength input by the i-th neuron in the input layer, i≤n; The input value h of the hidden layer in (k) The mathematical expression after translation and scaling is: Where h is the value after translation and scaling, b k is the translation factor of the wavelet basis function, a k is the scaling factor of the wavelet basis function; The mathematical expression of the output value of the hidden layer is Where h out (k) is the output value of the hidden layer; The mathematical expression of the output prediction value of the output layer is: Where y j is the j-th predicted value output by the output layer, ω lj is the network weight from the lth neuron in the hidden layer to the output layer, h out (l) is the output value of the lth neuron in the hidden layer.

2. The indoor floor positioning method according to claim 1, characterized in that: The construction of the fingerprint database specifically includes: At each positioning sampling point, each wireless access point is sampled 10 times, and the average of the 10 indoor Wi-Fi signal strengths obtained by sampling is calculated; Eliminate the indoor Wi-Fi signal strengths whose difference from the average value is greater than 10% of the average value, and average the remaining indoor Wi-Fi signal strengths as the indoor Wi-Fi signal strength value of each wireless access point; The spatial position vector of each positioning sampling point and the indoor Wi-Fi signal strength values of all wireless access points collected at each positioning sampling point form a fingerprint record; The fingerprint records of all positioning sampling points constitute the fingerprint database D = (Lp, Fp), p = 1, 2, ..., q; q is the number of positioning sampling points, Lp is the spatial position vector of the p-th positioning sampling point, and Fp is the signal fingerprint matrix of the p-th positioning sampling point; MAC r is the rth physical address of the pth positioning sampling point, MAC collected multiple times r The corresponding average indoor Wi-Fi signal strength, r = 1, 2, ..., s, where s is the number of indoor Wi-Fi signals collected at the p-th positioning sampling point.

3. The indoor floor positioning method according to claim 1, characterized in that: The signal fingerprint matrix of the point to be located is expressed as Where f is the signal fingerprint matrix of the point to be located, mac1, mac2 and mac t are the 1st, 2nd and tth physical addresses of the point to be located, rss1, rss2 and rss t are the indoor WIFI signal strengths corresponding to the 1st, 2nd and tth physical addresses collected at the point to be located.

4. The indoor floor positioning method according to claim 1, characterized in that: The fingerprint database is used to compensate for the indoor Wi-Fi signal strength of the wireless access point that is missing from the signal fingerprint matrix of the location point to obtain a compensated signal fingerprint matrix, specifically including: The positioning sampling point closest to the point to be positioned is determined as the compensation point; Determine the missing physical address of the point to be located relative to the compensation point and the indoor Wi-Fi signal strength corresponding to the physical address, and add them to the signal fingerprint matrix of the point to be located to obtain the compensated signal fingerprint matrix; the physical address corresponds one-to-one with the wireless access point.

5. The indoor floor positioning method according to claim 4, characterized in that: The compensated signal fingerprint matrix is input into the indoor floor position recognition model to output the building number, floor number and specific location of the point to be located, and then further includes: Determine the union of the physical addresses in the signal fingerprint matrix of the compensation point and the signal fingerprint matrix of the point to be located, and use the signal fingerprint matrix of the compensation point as the reference fingerprint and the signal fingerprint matrix of the point to be located as the test fingerprint; Select the minimum indoor Wi-Fi signal strength RSSmin in the reference fingerprint and the minimum indoor Wi-Fi signal strength rssmin in the test fingerprint respectively; Initialize w to 1; If the reference fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint does not change; If the reference fingerprint does not contain the w-th physical address in the union, assign rssmin the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint; If the test fingerprint contains the w-th physical address of the union, the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint will not be changed; If the test fingerprint does not contain the w-th physical address in the union, use RSSmin to assign the indoor Wi-Fi signal strength corresponding to the w-th physical address in the test fingerprint; Increase the value of w by 1 and return to step "If the reference fingerprint contains the w-th physical address in the union, do not change the indoor Wi-Fi signal strength corresponding to the w-th physical address in the reference fingerprint" until the value of w is greater than the total number of physical addresses in the union. Then stop the loop and obtain the updated fingerprint database. Replace the fingerprint database with the updated fingerprint database and return to step "Train the wavelet neural network using the fingerprint database to obtain an indoor floor position recognition model." 6. An indoor floor positioning system, characterized in that: include: A database construction module is used to construct a fingerprint database; the fingerprint database includes a spatial position vector of each positioning sampling point and a signal fingerprint matrix of each positioning sampling point; the spatial position vector includes the building number, floor number and specific location of the positioning sampling point; the signal fingerprint matrix includes n indoor Wi-Fi signal strengths collected from n wireless access points; Model training module, used to train the wavelet neural network using the fingerprint database to obtain the indoor floor position recognition model; The real-time acquisition module is used to obtain the strength of all indoor Wi-Fi signals collected at the point to be located and form a signal fingerprint matrix of the point to be located; a compensation module, configured to compensate for the indoor Wi-Fi signal strengths of the wireless access points missing from the signal fingerprint matrix of the target location point using the fingerprint database, to obtain a compensated signal fingerprint matrix; wherein the number of indoor Wi-Fi signal strengths in the compensated signal fingerprint matrix is n; An identification module is used to input the compensated signal fingerprint matrix into the indoor floor position recognition model and output the building number, floor number and specific location of the building where the point to be located is located; The wavelet neural network comprises: an input layer, a hidden layer and an output layer; The number of neurons in the input layer is n, the number of neurons in the hidden layer is k, and the number of neurons in the output layer is m; The Morlet wavelet function is selected as the transfer function of the hidden layer nodes; The mathematical expression of the input function of the hidden layer is Where h in (k) is the input value of the hidden layer, ω ik is the network weight from the input layer to the hidden layer, x i is the indoor WIFI signal strength input by the i-th neuron in the input layer, i≤n; The input value h of the hidden layer in (k) The mathematical expression after translation and scaling is: Where h is the value after translation and scaling, b k is the translation factor of the wavelet basis function, a k is the scaling factor of the wavelet basis function; The mathematical expression of the output value of the hidden layer is Where h out (k) is the output value of the hidden layer; The mathematical expression of the output prediction value of the output layer is: Where y j is the j-th predicted value output by the output layer, ω lj is the network weight from the lth neuron in the hidden layer to the output layer, h out (l) is the output value of the lth neuron in the hidden layer.

7. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the indoor floor positioning method according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed, the indoor floor positioning method according to any one of claims 1 to 5 is implemented.

Images