CN104748735A - Intelligent terminal-based indoor positioning method and equipment thereof - Google Patents

Abstract

The invention discloses an intelligent terminal-based indoor positioning method and equipment thereof, and relates to the technical field of mobile communication. The method comprises the following steps: an indoor positioning nerve network is trained by using an indoor travel data sample; when indoor positioning is needed, whether one step travel from an initial position is carried out or not is determined according to acceleration data acquired from a positioning device at an intelligent terminal; when one step travel is completed is determined, the length and the offset angle of the step travel are obtained through calculating according to acceleration data and pace frequency in the step by the indoor positioning nerve network; and the current position is positioned according to the length and the offset angle of the step travel. An intelligent nerve network algorithm adopted in the embodiment of the invention combines the average reference step length of different motion individuals with measurement of the acceleration data through using a sensor, so scientific and accurate real-time pace displacement calculation is realized, thereby accurate indoor positioning and a course updating solution scheme can be realized through the positioning device at the intelligent terminal.

Description

Based on indoor orientation method and the equipment of intelligent terminal

Technical field

The present invention relates to mobile communication technology field, particularly a kind of indoor orientation method based on intelligent terminal and equipment.

Background technology

At present, for the place that effectively cannot receive gps signal, as indoor environment, the indoor positioning solution realized based on intelligent terminal is also little.Only several solution, nearly all there is positioning precision inadequate, accuracy is limited, locates coarse problem.In the urgent need to a kind of new solution, a kind of accurate indoor positioning solution realized based on intelligent terminal newly can be provided.

Summary of the invention

In view of the above problems, the embodiment of the present invention provides a kind of indoor orientation method based on intelligent terminal and equipment, can carry out rational adjustment algorithm realize accurate indoor positioning and course more new solution by the locating device acquisition data of intelligent terminal.

The embodiment of the present invention have employed following technical scheme:

One embodiment of the invention provides a kind of indoor orientation method based on intelligent terminal, and described method comprises:

Indoor data sample of advancing is utilized to train indoor positioning neural network;

When needs carry out indoor positioning:

From initial position, judge whether that row further according to the acceleration information that the locating device from intelligent terminal obtains;

When judging to have advanced a step, obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle;

According to described walking progress long value and described deviation angle, current location is positioned;

Using current location as initial position, continue to perform described basis and judge whether that row further from the acceleration information that the locating device of intelligent terminal obtains, terminate to this location.

Described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value;

Described utilization indoor data sample of advancing carries out training to indoor positioning neural network and comprises:

For often going further data in multiple sample as a training data, perform respectively: using the paces frequency in this training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in this training data as the output neuron of indoor positioning neural network;

According to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron;

Described connection weights are revised according to repeatedly training data.

Describedly obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing and comprise:

Using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network;

According to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value.

From the acceleration information that the locating device of intelligent terminal obtains, described basis judges whether that row comprises further:

Judge whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

If meet described paces Rule of judgment, then determine a step of having advanced.

Describedly also to comprise after indoor positioning neural computing obtains this walking progress long value according to acceleration information in this step and paces frequency:

Judge whether described walking progress long value is less than the 3rd prevalue;

If be greater than the 3rd prevalue, then perform and describedly according to described walking progress long value and described deviation angle, current location to be positioned;

If be less than the 3rd prevalue, then re-execute and judge whether that row further according to the acceleration information obtained from the locating device of intelligent terminal.

In addition, the embodiment of the present invention additionally provides a kind of indoor positioning device based on intelligent terminal, and described equipment comprises:

Training module, trains indoor positioning neural network for utilizing indoor data sample of advancing;

Indoor positioning module, for when needs carry out indoor positioning, carries out indoor positioning;

Described indoor positioning module comprises:

According to the acceleration information that the locating device from intelligent terminal obtains, paces judging unit, for from initial position, judges whether that row further;

Acquiring unit, during for judging to have advanced a step when described paces judging unit, obtains this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle;

Positioning unit, positions current location for described walking progress long value obtaining according to acquiring unit and described deviation angle;

Cycling element, for using current location as initial position, continue start described paces judging unit, to this location terminate.

Described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value;

Described training module comprises:

Assignment unit, for for often going further data in multiple sample as a training data, using the paces frequency in each training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in each training data as the output neuron of indoor positioning neural network;

Connect weight calculation unit, for according to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron;

Amending unit, revises according to repeatedly training data for described connection weights.

Described acquiring unit comprises step-length acquiring unit and deviation angle acquiring unit:

Described step-length acquiring unit comprises,

Subelement is determined in input, during for judging to have advanced a step when described paces judging unit, using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network;

Step size computation subelement, for according to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value;

Described deviation angle acquiring unit, during for judging to have advanced a step when described paces judging unit, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle.

Described paces judging unit comprises:

Condition judgment subelement, for judging whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

Determine subelement, if for the judged result of described condition judgment subelement for meeting described paces Rule of judgment, then determine a step of having advanced.

Described positioning unit also comprises:

Location determination subelement, for according to acceleration information in this step and paces frequency after indoor positioning neural computing obtains this walking progress long value, judge whether described walking progress long value is less than the 3rd prevalue;

Continuing locator unit, for the judged result when described location determination subelement for being greater than the 3rd prevalue, then according to described walking progress long value and described deviation angle, current location being positioned;

Again paces judgment sub-unit, for the judged result when described location determination subelement for being less than the 3rd prevalue, then restarts described paces judging unit.

Visible, the embodiment of the present invention provides a kind of indoor orientation method based on intelligent terminal and equipment, utilizes indoor data sample of advancing to train indoor positioning neural network; When needs carry out indoor positioning: from initial position, judge whether that row further according to the acceleration information that the locating device from intelligent terminal obtains; When judging to have advanced a step, obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle; According to described walking progress long value and described deviation angle, current location is positioned; Using current location as initial position, continue to perform described basis and judge whether that row further from the acceleration information that the locating device of intelligent terminal obtains, terminate to this location.Because the embodiment of the present invention adopts Intelligent Neural Network algorithm, and the average reference step-length embodying motion unit difference is combined with employing sensor acceleration analysis data, thus achieve the science calculating of paces displacement in real time accurately more, then can by the locating device combination of intelligent terminal to the accurate indoor positioning of accurate adjustment algorithm realization of step-length and course more new solution.

Accompanying drawing explanation

A kind of indoor orientation method process flow diagram based on intelligent terminal that Fig. 1 provides for the embodiment of the present invention;

Fig. 2 utilizes intelligent terminal direct of travel schematic diagram for the one that the embodiment of the present invention provides;

A kind of indoor positioning neural network internal arithmetic schematic diagram that Fig. 3 provides for the embodiment of the present invention;

A kind of indoor positioning device structured flowchart based on intelligent terminal that Fig. 4 provides for the embodiment of the present invention.

Embodiment

For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.

See Fig. 1, the embodiment of the present invention provides a kind of indoor orientation method based on intelligent terminal, specifically comprises:

S101: utilize indoor data sample of advancing to train indoor positioning neural network.

It should be noted that, advancing in indoor, to be multiple motion unit (i.e. the actual tester of multidigit) to position by hand-held terminal device (intelligent terminal) data sample that test process obtains to data sample in indoor.

It should be noted that, in the embodiment of the present invention, advanced in indoor by hand-held intelligent terminal, obtain locating information by the locating device (such as gyroscope etc.) of intelligent terminal, in conjunction with actual test data, jointly obtain data sample.

In the embodiment of the present invention, intelligent terminal refers to and can obtain locating information and hand-holdable equipment, if namely can obtain locating information and can movement all within the scope of intelligent terminal described in the embodiment of the present invention, can be such as smart mobile phone, PAD etc.

Preferably, described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value.

The embodiment of the present invention relates to the MEMS(Micro-Electro-Mechanical Systems MEMS (micro electro mechanical system) in application Portable intelligent terminal (i.e. mobile terminal)) in locating device (as position transducer etc.) carry out indoor location location.

The place of gps signal effectively cannot be received, as indoor environment, the sensors such as the compass in MEMS, gravitometer, gyroscope, accelerometer can be adopted to position.Common position sensing location technology employing magnetometer, gyroscope differentiation course (namely obtaining deviation angle before relatively advancing after advancing) in intelligent terminal, carry out speed with acceleration transducer and displacement judges.The locating device in intelligent terminal is utilized merely to carry out in speed and displacement judgement, usually only the measurement data of consideration and speed itself carries out speed calculating and offset estimation, thus, the embodiment of the present invention also considers the difference of motion unit simultaneously: as the height of individual, body weight, fitness can bring the difference of motion unit average reference step-length, namely by reference to step value (the average reference step value of each motion unit, in practice can actual measurement obtain).To improve the accuracy of location further.

It should be noted that in addition, see Fig. 2, direct of travel, as shown in y-axis direction in Fig. 2, vertically travels direction as shown in z-axis direction in Fig. 2.Direct of travel is y-axis direction, before producing in advancing (backward acceleration); X-axis direction can produce the left rightward acceleration of level when change in location; Z-axis direction can produce the acceleration of vertical direction in advancing.

Preferably, described utilization indoor data sample of advancing carries out training to indoor positioning neural network and comprises:

For often going further data in multiple sample as a training data, perform respectively: using the paces frequency in this training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in this training data as the output neuron of indoor positioning neural network.

According to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron.

Then, above-mentioned connection weights are revised according to repeatedly training data.

Concrete, as shown in Figure 3, indoor positioning neural network input neuron is 4, and correspondence is cut down frequency, direct of travel acceleration variance, vertically travelled directional acceleration variance and reference step value respectively.Hidden layer neuron number is also 4, output neuron , be step value.

Wherein, the algorithm of paces frequency is:

f _step(n)＝1/(t _n-t _n-1)

T _k-t _k-1be a step holding time cycle, that is, often go further data as a training data, t _k-t _k-1for the time cycle of each training data.

The computing formula of direct of travel acceleration variance is:

$\mathrm{Var}\left[a_y\left(n\right)\right]=\underset{t=t_k-1}{\overset{t_k}{\mathrm{\Σ}}}\frac{{[a_y\left(t\right)-{\stackrel{\‾}{a}}_y\left(t_n\right)]}^2}{N}$

The computing formula vertically travelling directional acceleration variance is:

$\mathrm{Var}\left[a_z\left(n\right)\right]=\underset{t=t_k-1}{\overset{t_k}{\mathrm{\Σ}}}\frac{{[a_z\left(t\right)-{\stackrel{\‾}{a}}_z\left(t_n\right)]}^2}{N}$

A kind of preferred embodiment, t _k-1to t _kin one step in the time cycle, the number of samples that (such as the locating device of I9023 embedded in mobile phone) accelerometer exports is minimum may only have two.In practice, the acceleration and vertically travel directional acceleration and carry out suitable interpolation processing, to improve accuracy further of advancing to obtaining can be considered.First consider linear compensation algorithm (the acceleration bound threshold values that Y and Z is axially different is set, may be thought of as axially different | a| _max).

The each neuronic input of hidden layer is as follows:

$I_{h\left(i\right)}\left(n\right)=\underset{j=1}{\overset{M}{\mathrm{\Σ}}}{w}_{h\left(i\right),\mathrm{in}\left(j\right)}^2\×{\mathrm{in}}_j\left(n\right)+b_{h\left(i\right)}^2$

Indoor positioning neural network in the embodiment of the present invention specifically adopts Sigmoid excitation function:

$f\left(x\right)=\frac{1}{1+e^{-x}}$

That is, implicit each neuronal layers exports and is specially:

$y_i\left(n\right)=f\left[I_{h\left(i\right)}\left(n\right)\right]=\frac{1}{{1+e}^{-\left[I_{h\left(i\right)}\left(n\right)\right]}}$

Network output function:

$z_1\left(n\right)=\underset{j=1}{\overset{M}{\mathrm{\Σ}}}{w}_{z\left(1\right),h\left(j\right)}^3\×y_j\left(n\right)+b_{z\left(i\right)}^3$

Output Shaping equation: $l_{\mathrm{step}}\left(n\right)=\left\{\begin{array}{cc}0:& z_1\left(n\right)<l_{\mathrm{down}}\\ z_1\left(n\right):& l_{\mathrm{down}}<z_1\left(n\right)<l_{\mathrm{up}}\\ l_{\mathrm{up}}:& z_1\left(n\right)>l_{\mathrm{up}}\end{array}\right.$

Wherein l _downfor step-length lower limit (preferably gets l _down=0.3m, can adjust), l _upfor the step-length upper limit (preferably gets l _up=1.5m, can adjust).

In addition, in the embodiment of the present invention, further, the step to each connection modified weight is also comprised.Be described in detail as follows:

By hidden layer neuron and the connection weights of input layer and the connection weights composition of vector W=[v of output layer and hidden layer neuron _ki, θ _i, w _ij], or $W=[w_{z\left(1\right),h\left(i\right)}^3{b}_{h\left(i\right)}^2,w_{h\left(i\right),\mathrm{in}\left(j\right)}^2].$

Backward learning process:

Being provided with learning sample is { in ₁(n), in ₂(n), in ₃(n), x ₄(n); l _{step_ref}(n) }, n=1,2 ..., P; P is learning sample sum.For certain sample { in ₁(n), in ₂(n), in ₃(n), x ₄(n); l _{step_ref}(n) }, after given network connects weight vector W, the output valve l of network can be calculated _stepn (), the error that definition exports is d (n)=l _{step_ref}(n)-l _step(n), and the error function defining sample p is:

$e\left(n\right)=\frac{1}{2}{\left[d\left(n\right)\right]}^2=\frac{1}{2}{[l_{\mathrm{step}\_\mathrm{ref}}\left(n\right)-l_{\mathrm{step}}\left(n\right)]}^2$

Wherein, l _{step_ref}n () is the n-th sample reference value, l _stepn () is that the n-th step-length judges output valve. for hidden layer neuron i and input layer x _jconnection weights, θ _ifor the threshold values of hidden layer neuron i, hidden layer neuron number is M, M=4 in this algorithm.Output layer neuron number is L, L=1 in this algorithm.

By adjusting the value of W, progressively reduce error d (n), to improve the computational accuracy of network.In opposite direction communication process, be that the negative gradient direction changed with W along error function e (n) is revised W.Get in formula, α is learning rate, value between (0 ~ 1).

$\mathrm{\&Delta;W}=\mathrm{\&alpha;}\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}d\left(n\right)\&times;\frac{{\&PartialD;l}_{\mathrm{step}}\left(n\right)}{\&PartialD;W}$

Wherein, when sample is n, the element in Δ W is:

$\begin{array}{c}\mathrm{\&Delta;}{w}_{z\left(1\right),h\left(i\right)}^3\left(n\right)=\mathrm{\&alpha;d}\left(n\right)a_i^2\left(n\right)\\ \mathrm{\&Delta;}{b}_{h\left(i\right)}^2=\mathrm{\&alpha;}[1-a_i^2\left(n\right)]d\left(n\right)w_{z\left(1\right),h\left(i\right)}^3\\ \mathrm{\&Delta;}{w}_{h\left(i\right),\mathrm{in}\left(j\right)}^2=\mathrm{\&alpha;}{a}_{\mathrm{ip}}^2\left(n\right)[1-a_i^2\left(n\right)]{\mathrm{in}}_j\left(n\right)d\left(n\right)w_{z\left(1\right),h\left(i\right)}^3\end{array}$

Preferably, adjusted by difference feedback see Fig. 3, in each step length data, judge that neural network algorithm basic parameter is as follows:

M＝4,α＝0.01,γ＝0.02,ρ＝0.993,num _ite＝200

Be wherein hidden layer neuron number, α is learning rate, and γ is factor of momentum, and ρ is the Learning Process Control factor (reducing learning rate in learning process), num _itefor iterations.

A kth hidden layer is to the weights learning adjustment algorithm of output layer:

$w_{1,k}^3(n+1)=w_{1,k}^3\left(n\right)-\mathrm{\&gamma;\&Delta;}{w}_{1,k}^3\left(n\right)+(1-\mathrm{\&gamma;})2\mathrm{\&alpha;d}\left(n\right)a_k^2$

A jth input layer is calculated to i-th hidden layer weights learning adjustment

Method: $w_{i,j}^2(n+1)=w_{i,j}^2\left(n\right)-\mathrm{\&gamma;\&Delta;}{w}_{i,j}^2\left(n\right)+(1-\mathrm{\&gamma;})2\left(\mathrm{\&rho;\&alpha;}\right)d\left(n\right)\&times;w_{1,i}^3\&times;e^{-M}\&times;a_i^2\&times;{\mathrm{in}}_j$

Hidden layer threshold values study adjustment algorithm:

$b_{h\left(i\right)}^2(n+1)=b_{h\left(i\right)}^2\left(n\right)-\mathrm{\&gamma;\&Delta;}{b}_{h\left(i\right)}^2\left(n\right)$

Needing the collection carrying out various sample data, adjustment is optimized to each layer weights and bias parameter in software, obtaining a set of parameter for Practical Project test, for accurately judging paces and single step displacement.Reference step in algorithm varies with each individual, the individual parameter of Test Engineer's input when being each test beginning, to improve the accuracy that step-length judges.

When needs carry out indoor positioning, perform following operation:

According to the acceleration information that the locating device from intelligent terminal obtains, S102: from initial position, judges whether that row further.

S103: when judging to have advanced a step, obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle.

Wherein, obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing to comprise:

Using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network;

According to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value.

Preferably, the acceleration information obtained according to the locating device from intelligent terminal judges whether that row comprises further:

Judge whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

If meet described paces Rule of judgment, then determine a step of having advanced.

It should be noted that, preferably, the first prevalue is 1, and the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and the behavior of taking a step vertically travelling and direction collects tester's above-below direction is described; Preferably, second prevalue is 0.2, because under normal circumstances, acceleration straw rope is [-0.1, + 0.1] between, the overwhelming majority drops between [-0.08 ,+0.08], here, second prevalue is preferably chosen for 0.2, be the erroneous judgement in order to avoid causing due to noise, direct of travel acceleration information comprises the negative slope that is greater than the second prevalue, and the acceleration and deceleration behavior before and after direct of travel that collects is described.

In practical operation, the embodiment of the present invention after the acceleration information obtained from the locating device of intelligent terminal, can also carry out effective filter operation of noise, to improve accuracy further to acceleration information.

As preferably, also comprise after indoor positioning neural computing obtains this walking progress long value according to acceleration information in this step and paces frequency:

Judge whether described walking progress long value is less than the 3rd prevalue;

If be greater than the 3rd prevalue, then perform and describedly according to described walking progress long value and described deviation angle, current location to be positioned;

If be less than the 3rd prevalue, then re-execute and judge whether that row further according to the acceleration information obtained from the locating device of intelligent terminal.

3rd prevalue can be chosen to be 0.3 meter, that is, if the step-length of current judgement is actual be less than 0.3 meter, then thinks to cause due to noise, abandons this result, do not do location and use.

Preferably, after the acceleration information that can obtain the locating device from intelligent terminal carries out noise filtering operation, the acceleration information re-executed according to obtaining from the locating device of intelligent terminal judges whether row further step.

S104: current location is positioned according to described walking progress long value and described deviation angle.

In actual applications, the embodiment of the present invention can also, according to often walking positioning result, be implemented to play up, and constantly updates course, realizes the location based on indoor.

S105: using current location as initial position, continues to perform described basis and judges whether that row further from the acceleration information that the locating device of intelligent terminal obtains, terminate to this location.

Visible, the embodiment of the present invention provides a kind of indoor orientation method based on intelligent terminal, because the embodiment of the present invention adopts Intelligent Neural Network algorithm, and the average reference step-length embodying motion unit difference is combined with employing sensor acceleration analysis data, thus achieve the science calculating of paces displacement in real time accurately more, then can by the locating device combination of intelligent terminal to the accurate indoor positioning of accurate adjustment algorithm realization of step-length and course more new solution.

It should be noted that, the indoor orientation method based on intelligent terminal that the embodiment of the present invention provides, can be applied in plurality of application scenes, greatly can expand the various application of field of locating technology, promote correlation technique development further, the indoor orientation method that the such as embodiment of the present invention provides, can be applied in Indoor measurement scene, also can be applied in location-based service scene, can also be applied in wireless network test scene, etc.

See Fig. 4, the embodiment of the present invention also provides a kind of indoor positioning device based on intelligent terminal, and described equipment comprises: training module 400 and indoor positioning module 500.

Training module 400, trains indoor positioning neural network for utilizing indoor data sample of advancing.

Indoor positioning module 500, for when needs carry out indoor positioning, carries out indoor positioning.

Concrete, described indoor positioning module 500 comprises:

According to the acceleration information that the locating device from intelligent terminal obtains, paces judging unit 501, for from initial position, judges whether that row further;

Acquiring unit 502, during for judging to have advanced a step when described paces judging unit, obtains this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle;

Positioning unit 503, positions current location for described walking progress long value obtaining according to acquiring unit and described deviation angle;

Cycling element 504, for using current location as initial position, continue start described paces judging unit, to this location terminate.

Preferably, described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value.

Accordingly, described training module comprises:

Assignment unit, for for often going further data in multiple sample as a training data, using the paces frequency in each training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in each training data as the output neuron of indoor positioning neural network.

Connect weight calculation unit, for according to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron.

With, amending unit, revise according to repeatedly training data for described connection weights.

Described acquiring unit comprises step-length acquiring unit and deviation angle acquiring unit:

Wherein, described step-length acquiring unit comprises,

Subelement is determined in input, during for judging to have advanced a step when described paces judging unit, using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network.

With, step size computation subelement, for according to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value.

Described deviation angle acquiring unit, during for judging to have advanced a step when described paces judging unit, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle.

Preferably, described paces judging unit comprises:

Condition judgment subelement, for judging whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

Determine subelement, if for the judged result of described condition judgment subelement for meeting described paces Rule of judgment, then determine a step of having advanced.

Further, described positioning unit also comprises:

Location determination subelement, for according to acceleration information in this step and paces frequency after indoor positioning neural computing obtains this walking progress long value, judge whether described walking progress long value is less than the 3rd prevalue;

Continuing locator unit, for the judged result when described location determination subelement for being greater than the 3rd prevalue, then according to described walking progress long value and described deviation angle, current location being positioned;

Again paces judgment sub-unit, for the judged result when described location determination subelement for being less than the 3rd prevalue, then restarts described paces judging unit.

It should be noted that, the indoor positioning device based on intelligent terminal that the embodiment of the present invention provides can be built in intelligent terminal, also can be separately in intelligent terminal, can communicate with intelligent terminal, obtain the information such as acceleration information from the locating device of intelligent terminal.

It should be noted that, the modules in present system embodiment or the principle of work of submodule and processing procedure see the associated description in embodiment of the method shown in above-mentioned Fig. 1-Fig. 3, can repeat no more herein.

Visible, the embodiment of the present invention provides a kind of indoor positioning device based on intelligent terminal, because the embodiment of the present invention adopts Intelligent Neural Network algorithm, and the average reference step-length embodying motion unit difference is combined with employing sensor acceleration analysis data, thus achieve the science calculating of paces displacement in real time accurately more, then can by the locating device combination of intelligent terminal to the accurate indoor positioning of accurate adjustment algorithm realization of step-length and course more new solution.

For the ease of the technical scheme of the clear description embodiment of the present invention, in inventive embodiment, have employed the printed words such as " first ", " second " to distinguish the substantially identical identical entry of function and efficacy or similar item, it will be appreciated by those skilled in the art that the printed words such as " first ", " second " do not limit quantity and execution order.

One of ordinary skill in the art will appreciate that, the all or part of step realized in above-described embodiment method is that the hardware that can carry out instruction relevant by program has come, described program can be stored in a computer read/write memory medium, this program is when performing, comprise the steps: (step of method), described storage medium, as: ROM/RAM, magnetic disc, CD etc.

The foregoing is only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention.All any amendments done within the spirit and principles in the present invention, equivalent replacement, improvement etc., be all included in protection scope of the present invention.

Claims (10)

1. based on an indoor orientation method for intelligent terminal, it is characterized in that, described method comprises:

Indoor data sample of advancing is utilized to train indoor positioning neural network;

When needs carry out indoor positioning:

From initial position, judge whether that row further according to the acceleration information that the locating device from intelligent terminal obtains;

When judging to have advanced a step, obtain this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle;

According to described walking progress long value and described deviation angle, current location is positioned;

Using current location as initial position, continue to perform described basis and judge whether that row further from the acceleration information that the locating device of intelligent terminal obtains, terminate to this location.

2. method according to claim 1, is characterized in that, described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value;

Described utilization indoor data sample of advancing carries out training to indoor positioning neural network and comprises:

For often going further data in multiple sample as a training data, perform respectively: using the paces frequency in this training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in this training data as the output neuron of indoor positioning neural network;

According to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron;

Described connection weights are revised according to repeatedly training data.

3. method according to claim 2, is characterized in that, describedly obtains this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing and comprises:

Using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network;

According to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value.

4. method according to claim 1, is characterized in that, from the acceleration information that the locating device of intelligent terminal obtains, described basis judges whether that row comprises further:

Judge whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

If meet described paces Rule of judgment, then determine a step of having advanced.

5. method according to claim 1, is characterized in that, describedly also comprises after indoor positioning neural computing obtains this walking progress long value according to acceleration information in this step and paces frequency:

Judge whether described walking progress long value is less than the 3rd prevalue;

If be greater than the 3rd prevalue, then perform and describedly according to described walking progress long value and described deviation angle, current location to be positioned;

If be less than the 3rd prevalue, then re-execute and judge whether that row further according to the acceleration information obtained from the locating device of intelligent terminal.

6. based on an indoor positioning device for intelligent terminal, it is characterized in that, described equipment comprises:

Training module, trains indoor positioning neural network for utilizing indoor data sample of advancing;

Indoor positioning module, for when needs carry out indoor positioning, carries out indoor positioning;

Described indoor positioning module comprises:

According to the acceleration information that the locating device from intelligent terminal obtains, paces judging unit, for from initial position, judges whether that row further;

Acquiring unit, during for judging to have advanced a step when described paces judging unit, obtains this walking progress long value according to acceleration information in this step and paces frequency through indoor positioning neural computing; Further, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle;

Positioning unit, positions current location for described walking progress long value obtaining according to acquiring unit and described deviation angle;

Cycling element, for using current location as initial position, continue start described paces judging unit, to this location terminate.

7. equipment according to claim 6, is characterized in that, described indoor data sample of advancing comprises paces frequency, direct of travel acceleration variance, vertically travels directional acceleration variance and reference step value;

Described training module comprises:

Assignment unit, for for often going further data in multiple sample as a training data, using the paces frequency in each training data, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as the input neuron of indoor positioning neural network; Using the value of the actual step size value in each training data as the output neuron of indoor positioning neural network;

Connect weight calculation unit, for according to the value of each input neuron of described indoor positioning neural network and the value of output neuron, calculate the connection weights between each hidden layer neuron in each input neuron and indoor positioning neural network respectively, and each hidden layer neuron is connected weights with between output neuron;

Amending unit, revises according to repeatedly training data for described connection weights.

8. equipment according to claim 7, is characterized in that, described acquiring unit comprises step-length acquiring unit and deviation angle acquiring unit:

Described step-length acquiring unit comprises,

Subelement is determined in input, during for judging to have advanced a step when described paces judging unit, using paces frequency in this step, direct of travel acceleration variance, vertically travel directional acceleration variance and the reference step value value as each input neuron of described indoor positioning neural network;

Step size computation subelement, for according to the connection weights between each input neuron and each hidden layer neuron, and each hidden layer neuron is connected weights with between output neuron, calculates the value of output neuron, as this walking progress long value;

Described deviation angle acquiring unit, during for judging to have advanced a step when described paces judging unit, from the locating device of intelligent terminal obtain this walking enter rear course relatively this walking enter before deviation angle.

9. equipment according to claim 6, is characterized in that, described paces judging unit comprises:

Condition judgment subelement, for judging whether the acceleration information obtained from the locating device of intelligent terminal meets paces Rule of judgment;

Described paces Rule of judgment is: the acceleration information vertically travelling direction comprises the negative slope that is greater than the first prevalue, and direct of travel acceleration information comprises the negative slope that is greater than the second prevalue;

Determine subelement, if for the judged result of described condition judgment subelement for meeting described paces Rule of judgment, then determine a step of having advanced.

10. equipment according to claim 6, is characterized in that, described positioning unit also comprises:

Location determination subelement, for according to acceleration information in this step and paces frequency after indoor positioning neural computing obtains this walking progress long value, judge whether described walking progress long value is less than the 3rd prevalue;

Continuing locator unit, for the judged result when described location determination subelement for being greater than the 3rd prevalue, then according to described walking progress long value and described deviation angle, current location being positioned;

Again paces judgment sub-unit, for the judged result when described location determination subelement for being less than the 3rd prevalue, then restarts described paces judging unit.