Method for improving mine positioning precision by using signal confidence

Abstract

The invention relates to the technical field of wireless positioning, and particularly provides a method for improving mine positioning precision by using signal confidence, which comprises the following steps of: receiving a signal of a wireless carrier communication tag; calculating a signal confidence coefficient and a distance measurement value according to the signal of the wireless carrier communication tag; performing state prediction and state updating on the distance measurement value by adopting a filter; a first distance prediction value is obtained through state prediction, whether the deviation between the distance measurement value and the first distance prediction value meets a preset condition or not is judged, if yes, the distance measurement value is adopted as an observation value of a filter to conduct state updating and output a positioning result, and if not, a second distance prediction value obtained through calculation of a neural network model is obtained; and using the second distance prediction value as an observation value of a filter to carry out state updating and outputting a positioning result. According to the technical scheme of the invention, the problem of poor positioning precision caused by easy reflection of signals in a mine laneway environment can be effectively solved.

Description

Technical Field

[0001] The invention belongs to the field of wireless positioning, and specifically relates to a method for improving mine positioning accuracy by utilizing signal confidence. Background Art

[0002] The commonly used method in traditional coal mines is to achieve positioning based on technologies such as Bluetooth, radio frequency identification (RFID), Zigee and wireless hotspots (WIFI). However, these traditional positioning systems cannot meet the current industry standard requirement of system static positioning error less than 0.3 meters. With the continuous development and improvement of ultra-wideband wireless carrier communication technology (UWB), wireless carrier communication technology has the advantages of strong signal anti-interference ability, high positioning accuracy and low power consumption. Wireless carrier communication technology has gradually replaced the above traditional positioning solutions and become the mainstream precise positioning solution, and has been widely used in coal mines.

[0003] Coal mines are mostly long tunnels, and usually use the time-of-flight one-dimensional ranging method to accurately locate underground personnel. However, the tunnel is a special structure of narrow and long, so it is easy for the wireless carrier communication signal emitted by the positioning identification card to be detected by the positioning base station after multiple reflections, which greatly affects the positioning accuracy and even makes it impossible to locate.

[0004] At the same time, with the construction of intelligent mines, higher requirements are placed on the accuracy and stability of the positioning system, and the traditional one-dimensional positioning method can no longer meet current requirements. Summary of the invention

[0005] The present invention provides a method for improving the positioning accuracy of mines by using signal confidence, which can effectively improve the positioning accuracy and reduce the adverse effects of signal reflection, and is particularly suitable for accurate positioning of underground personnel in coal mines. It can effectively solve the problem of poor positioning accuracy due to easy signal reflection in mine tunnel environments.

[0006] The technical solution adopted by the present invention is:

[0007] A method for improving mine positioning accuracy by using signal confidence comprises the following steps:

[0008] receiving a signal from a wireless carrier communication tag;

[0009] Calculating signal confidence and distance measurement values ​​based on the signal of the wireless carrier communication tag;

[0010] Using filters to predict and update the state of distance measurements;

[0011] The state prediction obtains a first distance prediction value, and determines whether the deviation between the distance measurement value and the first distance prediction value meets a preset condition. If so, the distance measurement value is used as the observation value of the filter to update the state and output the positioning result. If not, the second distance prediction value calculated by the neural network model is obtained, and the second distance prediction value is used as the observation value of the filter to update the state and output the positioning result.

[0012] Further, the calculating of signal confidence according to the signal of the wireless carrier communication tag includes:

[0013] Read the channel impulse response index and the channel impulse response maximum amplitude index corresponding to the direct path in the wireless carrier communication chip register;

[0014] The judgment index is calculated based on the channel impulse response index of the direct path and the channel impulse response maximum index:

[0015] diff=|dp_idx-mp_idx|

[0016]

[0017] Wherein, Pr is a judgment index, diff is the absolute value of the difference between the channel impulse response index of the direct path and the detected channel impulse response maximum index, dp_idx is the channel impulse response index of the direct path, and mp_idx is the channel impulse response maximum index;

[0018] Calculate the amplitude ratio:

[0019]

[0020] Wherein, Mc is the amplitude ratio, dp_ampl1 is the maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ampl2 is the second maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ample3 is the third maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, and mp_ampl is the maximum amplitude of the channel impulse response;

[0021] The signal confidence is obtained based on Mc and Pr.

[0022] Furthermore, Mc, Pr and signal confidence satisfy:

[0023]

[0024] Where Δ is the confidence of the signal.

[0025] Further, the signal confidence is calculated when the wireless carrier communication signal transmitted by the wireless carrier communication tag for the first time is received.

[0026] Further, the calculating of the distance measurement value according to the signal of the wireless carrier communication tag includes:

[0027] Use the time-of-flight method to measure the distance and obtain the time-of-flight TOF;

[0028] Get the distance measurement based on the flight time and signal confidence:

[0029] S=TOF*c*Δ

[0030] Among them, S is the distance measurement value between the wireless carrier communication tag and the wireless carrier communication positioning base station, c is the speed of light, and Δ is the confidence of the signal.

[0031] Further, the flight time method is used for ranging, and the flight time is obtained including:

[0032] Get the timestamp of the wireless carrier communication signal and calculate the flight time based on the timestamp:

[0033]

[0034] Among them, TOF is the air flight time of the wireless carrier communication signal, T1 is the sending timestamp of the wireless carrier communication tag request frame, T2 is the receiving timestamp of the wireless carrier communication tag request frame, T3 is the sending timestamp of the wireless carrier communication positioning base station response frame, T4 is the receiving timestamp of the wireless carrier communication positioning base station response frame, T5 is the sending timestamp of the wireless carrier communication tag positioning frame, and T6 is the receiving timestamp of the wireless carrier communication tag positioning frame.

[0035] Furthermore, the filter is one of a Kalman filter, an extended Kalman filter, and a particle filter.

[0036] Further, the using a filter to perform state prediction and state update on the distance measurement value includes:

[0037] State initialization: The state vector is represented as:

[0038] x = [s, v]

[0039] The s is the distance, and the v is the speed;

[0040] The observation matrix is:

[0041] H = [1, 0]

[0042] State prediction: Calculate the first distance prediction value at the next moment based on the following formula:

[0043] S n+1 =S n +v*t

[0044] Status update: Get the distance measurement value at the current moment, determine whether the deviation between the distance measurement value and the first distance prediction value meets the preset conditions, if so, use the distance measurement value as the observation value of the filter to update the status and output the positioning result, if not, get the second distance prediction value calculated by the neural network model, use the second distance prediction value as the observation value of the filter to update the status and output the positioning result.

[0045] Furthermore, the neural network model is a back propagation neural network, comprising an input layer, an output layer and a hidden layer;

[0046] The input vector X of the input layer is:

[0047] X=[w, h, CIR1, CIR2, CIR3, rssi, S]

[0048] Where: w is the width of the lane, h is the height, CIR1, CIR2, CIR3 are the first three channel impulse response amplitudes of wireless carrier communication from large to small, rssi is the wireless carrier communication signal strength, and S is the distance measurement value;

[0049] The output vector Y of the output layer is:

[0050] Y=[y]

[0051] Where y is the second distance prediction value;

[0052] The neural network model uses a Sigmoid function as the excitation function:

[0053]

[0054] Furthermore, it also includes: model training steps:

[0055] The handheld wireless carrier communication tag moves from near to far within the predetermined range of the wireless carrier communication positioning base station, collects data once for every N meters of distance from the wireless carrier communication positioning base station, collects data corresponding to the input vector multiple times within the predetermined range of the same position, and calculates multiple distance measurement values ​​at the same position using the flight time method and takes the average value;

[0056] Construct training sets and test sets based on the collected data, and train the neural network model through the training sets and test sets;

[0057] The mean square error function is used to modify the network weights and thresholds of the neural network model;

[0058] When the mean square error is less than the preset value, the iteration is stopped and the neural network model is output.

[0059] The technical solution of the present invention has the following beneficial effects:

[0060] 1. By adopting a neural network model and training it through an offline model, a more targeted neural network model is obtained, which overcomes the problem that the existing neural network model will be affected by signal reflection when facing the positioning work of coal mine shafts, resulting in insufficient output result accuracy. This greatly improves the targeting and adaptability of the neural network model to the occasions of use, and improves the stability and accuracy of the system.

[0061] 2. Due to the introduction of confidence parameters, the deviation value of the data measured by the flight time method is adjusted by confidence, which overcomes the problem that the existing distance measurement method is affected by the reflected signal in the complex environment of coal mines and cannot accurately measure the actual distance, thereby greatly improving the stability and accuracy of the data.

[0062] 3. Due to the use of a neural network model and a confidence algorithm, the accuracy of the data is judged through a combined algorithm, and the data is adjusted and compared to ensure the accuracy of the data. This overcomes the problem that the existing single method of ranging can only continuously optimize signals and adjust parameters in the face of interference but cannot debug the data. The output data accuracy and stability are more reliable, avoiding errors caused by complex environmental reflections. The process is simple and the real-time performance of data detection is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a flow chart of a method for improving mine positioning accuracy by using signal confidence of the present invention;

[0064] Figure 2 is a training flow chart of a neural network in an embodiment of the present invention;

[0065] Figure 3 is a schematic diagram of the network structure of a neural network model in an embodiment of the present invention;

[0066] Figure 4 is a channel impulse response information diagram in an embodiment of the present invention;

[0067] Figure 5 It is a flow chart of the time-of-flight ranging method in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0069] A method of improving mine positioning accuracy by using signal confidence in this embodiment is as follows: Figure 1-5 As shown, the following steps are included:

[0070] The wireless carrier communication tag transmits the signal, and the wireless carrier communication positioning base station receives the signal of the wireless carrier communication tag; the wireless carrier communication adopts ultra-wideband technology. Ultra-wideband technology makes the channel fading insensitive and significantly improves the positioning accuracy.

[0071] Signal confidence and distance measurements are calculated based on the signal from the wireless carrier communication tag.

[0072] Specifically, Figure 4 and Figure 5 As shown in the figure, each time the tag performs one-dimensional positioning using the time-of-flight method, it needs to send two ultra-wideband signals to the base station. The ultra-wideband positioning base station uses the ultra-wideband signal sent by the ultra-wideband tag for the first time to calculate the current signal confidence, and uses this as the criterion for judging whether the current ultra-wideband signal has been reflected and reached the ultra-wideband positioning base station. The specific process of signal confidence calculation is as follows:

[0073] Read the channel impulse response index and the channel impulse response maximum amplitude index corresponding to the direct path in the wireless carrier communication chip register;

[0074] The judgment index is calculated based on the channel impulse response index of the direct path and the channel impulse response maximum index:

[0075] diff=|dp_idx-mp_idx|

[0076]

[0077] Wherein, Pr is a judgment index, diff is the absolute value of the difference between the channel impulse response index of the direct path and the detected channel impulse response maximum index, dp_idx is the channel impulse response index of the direct path, and mp_idx is the channel impulse response maximum index;

[0078] Calculate the amplitude ratio:

[0079]

[0080] Wherein, Mc is the amplitude ratio, dp_ampl1 is the maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ampl2 is the second maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ample3 is the third maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, and mp_ampl is the maximum amplitude of the channel impulse response;

[0081] The signal confidence is obtained based on Mc and Pr. Mc, Pr and signal confidence satisfy:

[0082]

[0083] Where Δ is the confidence of the signal.

[0084] The details are shown in Table 1:

[0085] Table 1 Signal confidence discrimination table

[0086] Mc Pr Signal confidence Any value 1.0 1.0 <0.9 ＞0 Pr ≥0.9 Any value 1.0 Any value 0.0 0.0

[0087] The distance measurement between the UWB tag and the UWB positioning base station is measured using the time-of-flight method and adjusted in combination with the signal confidence. One distance measurement requires sending the UWB signal three times, generating a total of six timestamps. The air flight time formula of the time-of-flight confidence method is:

[0088]

[0089] Wherein, TOF is the air flight time of the UWB signal, T1 is the sending timestamp of the UWB tag request frame, T2 is the receiving timestamp of the UWB tag request frame, T3 is the sending timestamp of the UWB positioning base station response frame, T4 is the receiving timestamp of the UWB positioning base station response frame, T5 is the sending timestamp of the UWB tag positioning frame, and T6 is the receiving timestamp of the UWB tag positioning frame.

[0090] The measurement distance formula of the flight time confidence method is:

[0091] S=TOF*c*Δ

[0092] Where S is the distance measurement between the UWB tag and the UWB positioning base station, c is the speed of light, and Δ is the signal confidence.

[0093] A filter is used to perform state prediction and state update on the distance measurement value; the state prediction obtains a first distance prediction value, and determines whether the deviation between the distance measurement value and the first distance prediction value meets a preset condition. If so, the distance measurement value is used as the observation value of the filter to perform state update and output the positioning result; if not, a second distance prediction value calculated by a neural network model is obtained, and the second distance prediction value is used as the observation value of the filter to perform state update and output the positioning result. The filter is one of a Kalman filter, an extended Kalman filter, and a particle filter.

[0094] Specifically, in this implementation, a Kalman filter is used to predict and update the state of the distance measurement value, and the chi-square deviation between the predicted value and the measured value is calculated using a Kalman filter algorithm that can filter outliers and data noise, which can further improve the positioning accuracy, thereby further determining whether the current time-of-flight distance measurement result is reliable. Figure 1 As shown, including:

[0095] State initialization: The state vector is represented as:

[0096] x = [s, v]

[0097] The s is the distance, and the v is the speed;

[0098] The observation matrix is:

[0099] H = [1, 0]

[0100] State prediction: Calculate the first distance prediction value at the next moment based on the following formula:

[0101] S n+1 =S n +v*t

[0102] Status update: Get the distance measurement value at the current moment, determine whether the deviation between the distance measurement value and the first distance prediction value meets the preset conditions, if so, use the distance measurement value as the observation value of the filter to update the status and output the positioning result, if not, get the second distance prediction value calculated by the neural network model, use the second distance prediction value as the observation value of the filter to update the status and output the positioning result.

[0103] like Figure 3 As shown, the neural network model in this embodiment is a back propagation neural network, and the neural network model is trained offline by cooperating with an ultra-wideband positioning base station and a handheld ultra-wideband tag. The model of the back propagation neural network is set to 3 layers, namely, an input layer, a hidden layer, and an output layer; the method for training the offline model of the back propagation neural network is as follows Figure 2 As shown; the specific steps are:

[0104] The handheld ultra-wideband tag moves from near to far within the predetermined range of the ultra-wideband positioning base station, and collects data every N meters of the distance to the ultra-wideband positioning base station. In this embodiment, data is collected every 5 meters, and the sample output value is the distance from the identification card to the ultra-wideband positioning base station measured by the laser rangefinder. The input vector X of the input layer is:

[0105] X=[w, h, CIR1, CIR2, CIR3, rssi, S]

[0106] Where: w is the width of the lane, h is the height of the lane, CIR1, CIR2, CIR3 are the first three values ​​of the UWB channel impulse response amplitude from large to small, rssi is the UWB signal strength, and S is the distance measured by the time-of-flight method;

[0107] The output layer output vector Y is:

[0108] Y=[y]

[0109] Where y is the second distance prediction value;

[0110] The input vector data is collected 10 times within a predetermined range at the same position, and the distance at the same position is calculated 3 times using the flight time method and the average value is taken; the sampled data is preprocessed, including but not limited to data limiting, data transformation and data normalization.

[0111] The number of hidden layer neurons used to balance computational complexity and output accuracy is determined to be 10, and the Sigmoid function is used as the excitation function, that is:

[0112]

[0113] Construct training sets and test sets based on the collected data, and train the neural network model through the training sets and test sets;

[0114] Set the mean square error function used to correct the weights and thresholds of the entire network, and use the mean square error function to calculate the error;

[0115] When the mean square error is less than 0.01, the iteration is stopped and the neural network model is output.

[0116] The output results of the technical solution of this embodiment improve the pertinence and adaptability to the occasion, thereby improving the overall stability and accuracy of the system. In addition, this method introduces a confidence parameter, which overcomes the problem that the existing distance measurement method is easily interfered by the reflected signal in the complex environment of a coal mine, so that the system can accurately measure the actual distance, greatly improving the stability of the system. It also uses two algorithms to complement each other, thereby avoiding the problem that a single algorithm encounters interference that cannot be avoided. Through algorithm comparison and optimization, the signal is closer to the true value, and the interference of the reflected signal is avoided, thereby improving the detection accuracy and real-time performance.

[0117] The directional terms mentioned in the present invention, such as "up", "down", "left", "right", etc., are only for better and clearer explanation and understanding of the present invention, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore should not be understood as a limitation of the present invention.

[0118] The above is an explanation of the preferred embodiments of the present invention, but it cannot be understood as a limitation of the claims. The present invention is not limited to the above embodiments, and its specific structure is allowed to be changed. All changes made within the scope of protection of the independent claims of the present invention are within the scope of protection of the present invention.

Claims

1. A method for improving mine positioning accuracy by using signal confidence, characterized in that: The following steps are involved: receiving a signal from a wireless carrier communication tag; Calculating signal confidence and distance measurement values ​​based on the signal of the wireless carrier communication tag; Using filters to predict and update the state of distance measurements; The state prediction obtains a first distance prediction value, and determines whether the deviation between the distance measurement value and the first distance prediction value meets a preset condition. If so, the distance measurement value is used as the observation value of the filter to update the state and output the positioning result. If not, the second distance prediction value calculated by the neural network model is obtained, and the second distance prediction value is used as the observation value of the filter to update the state and output the positioning result.

2. The method for improving mine positioning accuracy by using signal confidence according to claim 1, characterized in that: The calculating of signal confidence according to the signal of the wireless carrier communication tag comprises: Read the channel impulse response index and the channel impulse response maximum amplitude index corresponding to the direct path in the wireless carrier communication chip register; The judgment index is calculated based on the channel impulse response index of the direct path and the channel impulse response maximum index: diff=|dp_idx-mp_idx| Wherein, Pr is a judgment index, diff is the absolute value of the difference between the channel impulse response index of the direct path and the detected channel impulse response maximum index, dp_idx is the channel impulse response index of the direct path, and mp_idx is the channel impulse response maximum index; Calculate the amplitude ratio: Wherein, Mc is the amplitude ratio, dp_ampl1 is the maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ampl2 is the second maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, dp_ample3 is the third maximum amplitude around the direct path channel impulse response of the wireless carrier communication chip, and mp_ampl is the maximum amplitude of the channel impulse response; The signal confidence is obtained based on Mc and Pr.

3. The method for improving mine positioning accuracy by using signal confidence according to claim 2, characterized in that: Mc, Pr and signal confidence satisfy: Where Δ is the confidence of the signal.

4. The method for improving mine positioning accuracy by using signal confidence according to claim 2, characterized in that: The signal confidence is calculated when the wireless carrier communication signal transmitted by the wireless carrier communication tag for the first time is received.

5. The method for improving mine positioning accuracy by using signal confidence according to claim 2, characterized in that: Calculating the distance measurement value according to the signal of the wireless carrier communication tag includes: Use the time-of-flight method to measure the distance and obtain the time-of-flight TOF; Get the distance measurement based on the flight time and signal confidence: s=TOF*c*Δ Among them, S is the distance measurement value between the wireless carrier communication tag and the wireless carrier communication positioning base station, c is the speed of light, and Δ is the confidence of the signal.

6. The method for improving mine positioning accuracy by using signal confidence according to claim 5, characterized in that: Using the time-of-flight method to measure distance, the flight time includes: Get the timestamp of the wireless carrier communication signal and calculate the flight time based on the timestamp: Among them, TOF is the air flight time of the wireless carrier communication signal, T1 is the sending timestamp of the wireless carrier communication tag request frame, T2 is the receiving timestamp of the wireless carrier communication tag request frame, T3 is the sending timestamp of the wireless carrier communication positioning base station response frame, T4 is the receiving timestamp of the wireless carrier communication positioning base station response frame, T5 is the sending timestamp of the wireless carrier communication tag positioning frame, and T6 is the receiving timestamp of the wireless carrier communication tag positioning frame.

7. The method for improving mine positioning accuracy by using signal confidence according to claim 1, characterized in that: The filter is one of a Kalman filter, an extended Kalman filter and a particle filter.

8. The method for improving mine positioning accuracy by using signal confidence according to claim 1, characterized in that: The use of a filter to perform state prediction and state update on the distance measurement value comprises: State initialization: The state vector is represented as: x=[s,v] The s is the distance, and the v is the speed; The observation matrix is: H＝[1,0] State prediction: Calculate the first distance prediction value at the next moment based on the following formula: S n+1 =S n +v*t Status update: Get the distance measurement value at the current moment, determine whether the deviation between the distance measurement value and the first distance prediction value meets the preset conditions, if so, use the distance measurement value as the observation value of the filter to update the status and output the positioning result, if not, get the second distance prediction value calculated by the neural network model, use the second distance prediction value as the observation value of the filter to update the status and output the positioning result.

9. The method for improving mine positioning accuracy by using signal confidence according to claim 1, characterized in that: The neural network model is a back propagation neural network, comprising an input layer, an output layer and a hidden layer; The input vector X of the input layer is: X=[w,h,CIR1,CIR2,CIR3,rssi,S] Where: w is the width of the lane, h is the height, CIR1, CIR2, CIR3 are the first three channel impulse response amplitudes of wireless carrier communication from large to small, rssi is the wireless carrier communication signal strength, and S is the distance measurement value; The output vector Y of the output layer is: Y=[y] Where y is the second distance prediction value; The neural network model uses a Sigmoid function as the excitation function:

10. A method for improving mine positioning accuracy by using signal confidence according to claim 9, characterized in that: Also includes: Model training steps: The handheld wireless carrier communication tag moves from near to far within the predetermined range of the wireless carrier communication positioning base station, collects data once for every N meters of distance from the wireless carrier communication positioning base station, collects data corresponding to the input vector multiple times within the predetermined range of the same position, and calculates multiple distance measurement values ​​at the same position using the flight time method and takes the average value; Construct training sets and test sets based on the collected data, and train the neural network model through the training sets and test sets; The mean square error function is used to modify the network weights and thresholds of the neural network model; When the mean square error is less than the preset value, the iteration is stopped and the neural network model is output.

Images