Intelligent locomotive positioning method, device and equipment based on multiple sensors and medium

Abstract

The invention provides an intelligent locomotive positioning method, device and equipment based on multiple sensors and a medium, and the method comprises the steps: collecting data of each sensor, including GPS position information, RFID tag information, quick transmission position information and inertial measurement unit data; setting a corresponding positioning calculation strategy according to a locomotive operation scene, and realizing automatic switching of the positioning calculation strategy through state machine logic; acquiring sensor data based on the current operation scene and the positioning calculation strategy of the locomotive; and hardware-level time synchronization and space coordinate calibration are carried out based on the acquired sensor data, so that the operation position of the locomotive is determined. According to the method, the self-adaptive error model is constructed, the speed transmission error growth rate is predicted according to the historical data and is compensated in real time, positioning offset after long-term operation is avoided, and the positioning accuracy and stability are improved; and in each mode, through cooperative work and information complementation of multiple sensors and in combination with an adaptive weight distribution algorithm, the adaptive capacity of the system in a complex environment is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of locomotive positioning technology, and in particular to a multi-sensor-based intelligent locomotive positioning method, device, equipment and medium. BACKGROUND

[0002] Currently, locomotive positioning technology mainly relies on single or limited sensor combinations. Common solutions include: satellite positioning technology, which achieves centimeter-level high-precision positioning through real-time kinematic (RTK) technology, is suitable for open environments, relies on satellite signals, and is easily affected by sheltered environments such as tunnels, viaducts, and urban canyons, resulting in signal loss or drift. Radio frequency identification (RFID) assisted positioning, which deploys RFID tags along the track, allows the locomotive to calibrate its position by reading the tag information, requires pre-deployment of infrastructure, has limited coverage, and cannot achieve continuous positioning. Speed sensor integral positioning uses speed sensors to collect speed information and integrate to calculate displacement, but the cumulative error increases with time and distance, and the long-term positioning accuracy decreases significantly. Some systems attempt to combine the above technologies, but usually use simple weight distribution or switching logic. Existing multi-sensor systems lack dynamic fusion algorithms, and each sensor data is processed independently, failing to fully utilize the complementarity. For example, when the RTK signal is lost, it switches to RFID or odometry, but does not solve the error coupling problem. SUMMARY

[0003] To overcome the problems in the related art, the present disclosure provides a multi-sensor-based intelligent locomotive positioning method, device, equipment and medium to solve the technical problems in the related art.

[0004] One or more embodiments of the present specification provide a multi-sensor-based intelligent locomotive positioning method, comprising the following steps:

[0005] Collecting sensor data, including GPS position information, RFID tag information, speed sensor position information, and inertial measurement unit data;

[0006] Setting corresponding positioning solution strategies according to the locomotive operating scenario, and automatically switching the positioning solution strategies through state machine logic; based on the current operating scenario of the locomotive and the positioning solution strategy, collecting sensor data;

[0007] The positioning solution strategy is to select a single sensor to determine the locomotive position information according to different running scenes, or to determine the locomotive position information based on a preset weight according to the positioning data determined by multiple sensors; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning result according to the GPS accuracy factor model; in a tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in a transition scene, a mixed weight and linear combination of three sensor data are used; in a fault scene, only the integral information of the speed transmission device is used for positioning, and a trained error model is used for dynamic prediction of the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device; and

[0008] Based on the collected sensor data, hardware-level time synchronization and spatial coordinate calibration are performed to determine the running position of the locomotive.

[0009] One or more embodiments of the present specification provide an intelligent locomotive positioning device based on multiple sensors, comprising:

[0010] A data acquisition module is configured to acquire sensor data, including GPS position information, RFID tag information, speed transmission position information, and inertial measurement unit data.

[0011] A positioning solution strategy switching module is configured to set a corresponding positioning solution strategy according to the running scene of the locomotive, and automatically switch the positioning solution strategy through a state machine logic; based on the current running scene of the locomotive and the positioning solution strategy, sensor data is acquired through the data acquisition module;

[0012] The positioning solution strategy is to select a single sensor to determine the locomotive position information according to different running scenes, or to determine the locomotive position information based on a preset weight according to the positioning data determined by multiple sensors; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning result according to the GPS accuracy factor model; in a tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in a transition scene, a mixed weight and linear combination of three sensor data are used; in a fault scene, only the integral information of the speed transmission device is used for positioning, and a trained error model is used for dynamic prediction of the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device;

[0013] A position calculation module is configured to perform hardware-level time synchronization and spatial coordinate calibration based on the collected sensor data to determine the running position of the locomotive.

[0014] The one or more embodiments of the specification provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the multi-sensor-based intelligent locomotive positioning method as described above when executing the computer program.

[0015] The one or more embodiments of the specification provide a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the multi-sensor-based intelligent locomotive positioning method as described above.

[0016] The multi-sensor-based intelligent locomotive positioning method, device, equipment and medium provided by the disclosure have the advantages that by integrating multi-source sensor data such as RFID, RTK high-precision positioning, speed transmission equipment (speed integration), the problems of insufficient positioning accuracy, poor environmental adaptability, uncontrollable cumulative error and the like in the prior art are solved, and by setting multi-modal switching logic, automatic switching of open mode, tunnel mode and fault mode is realized in combination with real-time environment detection (such as satellite signal-to-noise ratio and RFID signal strength), RFID absolute position calibration and RTK dynamic correction mechanism are adopted to suppress the drift of the speed transmission equipment in real time, and the cumulative error problem is effectively eliminated. By constructing an adaptive error model, the speed transmission error growth rate is predicted according to historical data and is compensated in real time, the positioning deviation after long-term operation is avoided, and the accuracy and stability of positioning are significantly improved; and in each mode, the adaptive weight distribution algorithm is combined to enhance the adaptability of the system in complex environments through the cooperative work and information complementation of multiple sensors. In harsh environments such as underground tracks, electromagnetic interference areas and dense building groups, stable positioning can still be maintained, and the limitations of a single sensor in complex environments are effectively overcome. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the one or more embodiments of the specification or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the specification, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 A flowchart of a multi-sensor-based intelligent locomotive positioning method provided by the one or more embodiments of the specification is provided.

[0019] Figure 2 A block diagram of a multi-sensor-based intelligent locomotive positioning device provided by the one or more embodiments of the specification is provided.

[0020] Figure 3A structural diagram of a computer device provided for one or more embodiments of the present specification. DETAILED DESCRIPTION

[0021] In order for those skilled in the art to better understand the technical solutions in one or more embodiments of the present specification, the technical solutions in one or more embodiments of the present specification will be clearly and completely described below in conjunction with the accompanying drawings of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on one or more embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should fall within the protection scope of the present application.

[0022] The present application will be described in detail below in conjunction with the specific embodiments and the accompanying drawings of the specification.

[0023] Method embodiment

[0024] According to the embodiment of the present application, a multi-sensor-based intelligent locomotive positioning method is provided, as shown in the figure, the multi-sensor-based intelligent locomotive positioning method provided by the present embodiment, according to the multi-sensor-based intelligent locomotive positioning method of the embodiment of the present application, comprising the steps of: Figure 1

[0025] Step S1, collecting sensor data, including GPS position information, RFID tag information, speed position information and inertial measurement unit data;

[0026] Step S2, setting corresponding positioning solution strategy according to the locomotive running scene, and realizing automatic switching of the positioning solution strategy through the state machine logic; based on the current running scene of the locomotive and the positioning solution strategy, collecting sensor data;

[0027] Among them, the positioning solution strategy is to select a single sensor to determine the position information according to different running scenes, or to determine the position information of the locomotive based on the preset weight according to the positioning data determined by multiple sensors; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning results according to the GPS accuracy factor model; in the tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in the transition scene, the mixed weight and linear combination of the three kinds of sensor data are used; in the fault scene, only the integral information of the speed transmission device is used for positioning, and the trained error model is used for dynamic prediction of the cumulative error of the speed transmission device to realize the correction of the cumulative error of the speed transmission device.

[0028] Step S3, based on the collected sensor data, performing hardware-level time synchronization and space coordinate calibration to determine the running position of the locomotive. ​

[0029] In an embodiment, the positioning solution strategies corresponding to different running scenarios are as follows:

[0030] Open scenario: activated when RTK signal-to-noise ratio ≥ 35 dB and the number of satellites ≥ 5, switched to RTK + fast transmission positioning data as the locomotive position;

[0031] Tunnel scenario: activated when RTK signal is lost (signal-to-noise ratio < 20 dB for 1 second) and RFID signal is valid, switched to RFID + fast transmission positioning information fusion to determine the locomotive position;

[0032] Transition scenario: activated when RTK signal is unstable (such as signal-to-noise ratio 20-35 dB), RTK + fast transmission + RFID hybrid positioning information fusion is used to determine the locomotive position;

[0033] Fault scenario: activated when both RFID and RTK fail, relying only on fast transmission equipment to determine the locomotive position, at this time the trained error model is called to dynamically predict the cumulative error of the fast transmission equipment to correct the cumulative error of the fast transmission equipment.

[0034] In a specific embodiment, the weight configuration of each positioning sensor data in the positioning solution strategy corresponding to different running scenarios is as follows:

[0035] Open mode: RTK: 0.8-1.0, fast transmission: 0-0.2, RFID: 0; for example, set RTK weight to 0.8 and fast transmission weight to 0.2, then locomotive position = 0.8 x RTK position value + 0.2 fast transmission position value.

[0036] Tunnel mode: RFID: 0.5, fast transmission: 0.5;

[0037] Transition mode RTK: 0.3-0.7, RFID: 0.2-0.4, fast transmission: 0.1-0.3.

[0038] The embodiment also includes, under each running scenario, setting a real-time confidence factor to score the stability of RTK and RFID positioning data, and dynamically correcting the weights of RTK and RFID.

[0039] In an embodiment, the stability of RTK and RFID data throughout the running process is scored by the epoch solution success rate of RTK and the reading accuracy rate of RFID, and the weights of RTK and RFID are corrected according to the standard value, wherein,

[0040] Epoch solution success rate = (number of successfully solved epochs / total number of epochs) x 100%;

[0041] Reading accuracy rate = (number of correctly read RFID tags / actual number of tags) x 100%;

[0042] For example, the epoch solution success rate standard of RTK position data is 95%, the reading accuracy standard of RFID position data is 90%, and under the condition of the initial weight set, when the calculated epoch solution success rate is higher than the standard value, the weight is increased; when the calculated epoch solution success rate is lower than the standard value, the weight is reduced. For example, when the difference is -5% (i.e. the epoch solution success rate is 5% lower than the standard score), the RTK data weight is reduced by 5% in the next calculation; when the difference is +3% (i.e. the stability score is 3% higher than the standard score), the weight is increased by 3%. The RFID dynamic weight correction is the same.

[0043] The initial error model of the embodiment determines the cumulative error of the speed transmission device (i.e. the difference between the "real position provided by RFID / RTK" and the "position calculated by speed transmission integration") and the locomotive running parameters in the corresponding period through the absolute position of the locomotive determined by RFID or the position of the locomotive determined by RTK collected in each period of multiple running scenarios, including real-time speed, cumulative running distance since the last calibration, cumulative working time of the speed transmission device and historical error trend, which is determined by the difference between the real train position coordinates provided by the latest N times of RFID / RTK and the position coordinates determined by the speed transmission device. The training set is constructed based on the collected running parameter data to train the error model, so as to realize the prediction of the cumulative error of the speed transmission device according to the real-time speed of the locomotive, the cumulative running distance since the last calibration and the cumulative working time of the speed transmission device. In addition, the model is fine-tuned through the absolute position of the locomotive determined by RFID or the position of the locomotive determined by RTK, the cumulative error of the speed transmission device and the running parameter data collected in the process of locomotive running.

[0044] In the embodiment, the automatic switching of modes is realized through state machine logic, and the weighted smoothing algorithm (such as gradually reducing the weight of the original mode and increasing the weight of the new mode in the first 100 ms) is used to determine the position of the locomotive during the switching process to avoid positioning jump.

[0045] In the embodiment, the automatic switching of positioning solution strategies is realized through state machine logic, which can be realized through the following technologies:

[0046] Based on the constructed electronic track running map, the track position coordinate points of different running scenario switching are marked, and the corresponding positioning strategies to be switched are set correspondingly, which are used to guide and realize the confirmation of the locomotive running scenario and the switching of the locomotive positioning solution strategy.

[0047] The preferred embodiment of the present application uses improved Kalman filtering technology to dynamically adjust the weight of sensor data in the positioning solution strategy according to environmental changes, realizes the adjustment of the weight of multi-sensor data, dynamically fuses multi-source data, significantly improves the real-time performance and dynamic response capability of the system, and can quickly adapt to environmental changes. Specifically, the following steps are included:

[0048] In step S21, under each positioning solution strategy, an adaptive Kalman filter is used, the displacement calculated from the speed output by the speed transmission device is used as the predicted value, and the absolute position of the RTK and / or RFID is used as the observation value. The trustworthiness of prediction and observation is dynamically balanced through weight to realize the fusion calculation of multi-source data to determine the current position of the locomotive. The specific steps are as follows:

[0049] In step S211, based on the speed value collected by the speed transmission device, the real-time position coordinate value is calculated by integration;

[0050] In step S212, the absolute position coordinate value of RTK or RFID is used as the observation value of Kalman filtering;

[0051] In step S213, the initial observation noise covariance R0 is set, and based on the RTK and / or RFID weight value preset in the previous step, the new observation noise covariance R is calculated K =R0 / w k , wherein w k =1-speed weight; wherein the weight w k is greater → R K is smaller, and the observation is more trusted (Kalman gain K increases); the weight w k is smaller → R K is greater, and the prediction is more trusted (Kalman gain K decreases).

[0052] The advantages of the dynamic adjustment of the present embodiment are as follows: based on the carrier phase signal-to-noise ratio (SNR) of RTK and the signal strength of RFID, during the operation of the locomotive, the scene changes, when the RTK / RFID signal is good (w k is large), the integral drift is corrected with high-precision observation, and when the RTK / RFID signal is poor (w k is small), the position information is determined by relying on short-term more reliable speed integration, avoiding the influence of wild values. The traditional Kalman filter fixes the covariance R, which cannot adapt to the change of observation noise. The present embodiment dynamically adjusts the observation noise covariance R, dynamically balances the trustworthiness of prediction and observation through weight, and makes the position positioning more reliable.

[0053] In an embodiment, an error model is used to dynamically predict the cumulative error of the speed transmission device, and the formula is:

[0054] Etotal(v, d, t, H) = Ebase(d) + Erate(v, d) + Eaging(t) + Etrend(H);

[0055] Eaging(t) = Kaging x t, Kaging = 0.005% / h;

[0056] wherein Etotal is the cumulative error prediction value of the speed transmission device; v is the real-time speed of the locomotive (km / h); d is the cumulative running distance since the last calibration (km); t is the cumulative working time of the speed transmission device; H is the historical calibration error set (for example, the cumulative error value and the change rate of the last 5 calibrations); Ebase(d) is the basic error based on the cumulative running distance; Erate(v, d) is the error growth rate based on the speed interval and the cumulative distance; Eaging(t) is the aging error based on the cumulative working time of the device; Etrend(H) is the error adjustment amount based on the historical error trend H.

[0057] In this embodiment, the cumulative error Etotal of the speed transmission will be corrected in the fault scenario, and the corrected position coordinate Pcorrected is obtained by deducting the cumulative error Etotal from the position coordinate Pspeed determined by the speed transmission, i.e., Pcorrected = Pspeed - Etotal.

[0058] In an embodiment, the error growth rate Erate(v, d) is used to quantify the increase in the cumulative error of the speed transmission per preset distance (such as 1 km) in a specific speed interval. In this embodiment, different error growth rate calculation formulas are set according to different speed intervals, as follows:

[0059] Low speed interval (low speed < 50 km / h): Erate_low(d) = klow x d;

[0060] wherein klow is the error growth rate coefficient of the low speed interval (% / km), and d is the cumulative running distance (km).

[0061] Medium speed interval (50-120 km / h): Erate_mid(d) = kmid x d;

[0062] wherein kmid is the error growth rate coefficient of the medium speed interval (% / km).

[0063] High speed interval (> 120 km / h): Erate_high(d) = khigh x d;

[0064] wherein khigh is the error growth rate coefficient of the high speed interval (% / km).

[0065] The higher the interval speed is, the greater the error growth rate coefficient value is.

[0066] In an embodiment, since the accumulated error grows with time and distance, the long-term positioning accuracy decreases significantly, and therefore the following step is further included in the embodiment:

[0067] The basic error is compensated by accumulating the distance, i.e., the basic error rate is dynamically adjusted according to the accumulated running distance (reflecting the wheel diameter wear) for correcting the basic error, and the calculation formula of the basic error Ebase(d) is:

[0068] Ebase(d) = rbase(d) x d;

[0069] wherein rbase(d) is the basic error rate (%) / km dynamically corrected based on the accumulated running distance, and d is the accumulated running distance (km).

[0070] For example, the basic error rate is increased by 0.02% / km for each 1000 km of accumulated running (based on the linear relationship between the wheel diameter wear and the error), and the calculation formula of the basic error rate rbase(d) is: rbase(d) = r0 + 0.02% / km x d, and r0 = 0.1%.

[0071] In the embodiment, the following step is further included: the overall deviation of the error adjustment Etrend(H) is corrected by statistically determining the mean value of the residual errors of the last N times, e.g., if the mean value of the residual errors is positive, it indicates that the error model underestimates the error, and the error prediction value is increased by a as a whole, and the specific scheme is as follows:

[0072] The mean value of the residual errors is assumed to be the average value of the residual error values of the last N calibrations, and the residual error values of the last N calibrations are r1, r2, …, ri, N = 1-i, wherein ri is the difference between the cumulative error ereali determined by the RFID / RTK calibration speed and the cumulative error epredi predicted by the error model, i.e., the residual error ri = ereali-epredi, and the mean value of the residual errors is rP = (r1+r2+r3+…+ri) / N, and N can be 5-10 in the embodiment, wherein,

[0073] If rP is positive, it indicates that the error prediction value of the error model is low, and Etrend(H) = epred_old x (1+ a), i.e., the error adjustment of the error model is increased by (1+ a) times;

[0074] If rP is negative, it indicates that the error prediction value of the error model is high, and Etrend(H) = epred_old x (1- a), i.e., the error adjustment of the error model is increased by (1- a) times;

[0075] In other cases, Etrend(H) = epred_old, without adjustment.

[0076] The embodiment also includes evaluation of the deviation of the error model prediction result, and determination of a model optimization strategy according to the evaluation result, with the specific steps being as follows:

[0077] In step S40, based on the speed interval and the cumulative distance interval of the speed interval, each evaluation interval is determined, and the residual value of the corresponding running scene of each evaluation interval is obtained, wherein,

[0078] The setting rule of the evaluation interval is that each evaluation interval of the scene is formed by dividing the speed interval and the cumulative distance interval of the speed interval. For example, the speed is divided into three speed intervals of <50km / h, 50-120km / h, and >120km / h, and the cumulative distance is divided into multiple distances of 0-5km, 5-10km, and >10km. When the locomotive running speed is 80-100km / h, the intervals of the cumulative distance of 0-5km, 5-10km, and >10km corresponding to the running are three evaluation intervals, and when the locomotive running speed is 450km / h, the intervals of the cumulative distance of 0-5km, 5-10km, and >10km corresponding to the running are also three evaluation intervals.

[0079] In step S41, the residual value in the evaluation interval and the residual mean value of the continuous multiple evaluation intervals are determined, and if the residual value is greater than the set threshold value, the error coefficient of the interval is fine-tuned by the least square method; if the residual value is greater than the set threshold value for two consecutive times in the same speed interval, the average value of the cumulative error values of the last 3 effective calibrations is used to replace the prediction value; if the residual value exceeds the threshold value for 3 consecutive times, the parameters of the error model are rolled back to use the model parameters of the previous Nth version, and feedback is given to the train operation; if the absolute value of the residual mean value is > 2 times the overall residual mean value in the running process, it indicates that the prediction accuracy of the prediction model in the scene is low, and the error growth rate coefficient of the corresponding evaluation interval is adjusted.

[0080] In the embodiment, through the established error model and combined with the real-time calibration mechanism, accurate monitoring and control of the cumulative error of the speed transmission equipment are realized, and when the residual error of single calibration or multiple calibrations exceeds the threshold value, the abnormal correction mechanism can be triggered in time, the deviation of the positioning result is effectively prevented from spreading, and the stability and reliability of the system are improved.

[0081] In the embodiment, in order to improve the prediction accuracy of the model, iteration and backup of the error model are also included, with the specifics being as follows:

[0082] In step S50, after 10 calibrations are completed, a model parameter summary update is automatically performed once to generate a new error model version, and the specific process is as follows:

[0083] The cumulative error of each completed 1 time express transmission is used to train the error model and fine-tune the parameters using the cumulative error value and the locomotive operation parameters when the acquired RFID / RTK signal is effective. The system temporarily records the cumulative 10 times of calibration, obtains new model parameters by averaging each parameter, and updates all parameters of the error model.

[0084] The multi-sensor-based intelligent locomotive positioning method provided in the embodiment solves the problems of insufficient positioning accuracy, poor environmental adaptability, uncontrollable cumulative error, etc. in the prior art by integrating multi-source sensor data such as RFID, RTK high-precision positioning, express transmission equipment (speed integration), etc. Through the set multi-modal switching logic, combined with real-time environment detection (such as satellite signal-to-noise ratio, RFID signal strength), the open mode, tunnel mode, and fault mode are automatically switched. The RFID absolute position calibration and RTK dynamic correction mechanism are used to suppress the drift of the express transmission equipment in real time, effectively eliminating the cumulative error problem. By constructing an adaptive error model, the express transmission error growth rate is predicted according to historical data and real-time compensation is performed, avoiding the positioning deviation after long-term operation, and significantly improving the accuracy and stability of positioning. In each mode, through the cooperative work and information complementation of multiple sensors, combined with an adaptive weight distribution algorithm, the adaptability of the system in complex environments is enhanced. In harsh environments such as underground tracks, electromagnetic interference areas, and dense building groups, stable positioning can still be maintained, effectively overcoming the limitations of single sensors in complex environments.

[0085] Device embodiment

[0086] According to the embodiment of the application, a multi-sensor-based intelligent locomotive positioning device is provided, as shown in the accompanying drawings. Figure 2 The multi-sensor-based intelligent locomotive positioning device provided in the embodiment is a block diagram, and the multi-sensor-based intelligent locomotive positioning device according to the embodiment of the application comprises:

[0087] The data acquisition module 10 is used to acquire sensor data, including GPS position information, RFID tag information, express transmission position information, and inertial measurement unit data.

[0088] The positioning solution strategy switching module 20 is used to set the corresponding positioning solution strategy according to the locomotive operation scene, and to realize automatic switching of the positioning solution strategy through state machine logic. Based on the current locomotive operation scene and the positioning solution strategy, sensor data is acquired through the data acquisition module 10.

[0089] The positioning solution strategy is to select a single sensor to determine the position information of the locomotive according to different running scenes, or to fuse the positioning data determined by multiple sensors based on a preset weight to determine the position information of the locomotive; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning result according to the GPS accuracy factor model; in a tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in a transition scene, a mixed weight and linear combination of the data of the three sensors are used; in a failure scene, only the integral information of the speed transmission device is used for positioning, and a trained error model is used for dynamic prediction of the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device.

[0090] The position calculation module 30 performs hardware-level time synchronization and spatial coordinate calibration based on the collected sensor data, thereby determining the running position of the locomotive.

[0091] In an embodiment, the positioning solution strategies corresponding to different running scenes are as follows:

[0092] Open scene: activated when the RTK signal-to-noise ratio is greater than or equal to 35 dB and the number of satellites is greater than or equal to 5, and switched to RTK+speed transmission positioning data as the position of the locomotive;

[0093] Tunnel scene: activated when the RTK signal is lost (signal-to-noise ratio less than 20 dB for 1 second) and the RFID signal is valid, and switched to RFID+speed transmission positioning information fusion to determine the position of the locomotive;

[0094] Transition scene: activated when the RTK signal is unstable (such as signal-to-noise ratio 20-35 dB), and RTK+speed transmission+RFID hybrid positioning information fusion is used to determine the position of the locomotive;

[0095] Failure scene: activated when both RFID and RTK are invalid, and only relying on the speed transmission device to determine the position of the locomotive, at this time a trained error model is called to dynamically predict the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device.

[0096] In a specific embodiment, the weight configuration of each positioning sensor data in the positioning solution strategy corresponding to different running scenes is as follows:

[0097] Open mode: RTK: 0.8-1.0, speed transmission: 0-0.2, RFID: 0; for example, set the RTK weight to 0.8 and the speed transmission weight to 0.2, then the position of the locomotive = 0.8 x RTK position value + 0.2 speed transmission position value.

[0098] Tunnel mode: RFID: 0.5, speed transmission: 0.5;

[0099] Transition mode RTK: 0.3-0.7, RFID: 0.2-0.4, express: 0.1-0.3.

[0100] The embodiment also includes, under each operating scenario, setting a real-time confidence factor to score the stability of RTK and RFID positioning data, and dynamically correcting the weights of RTK and RFID.

[0101] In an embodiment, the stability of RTK and RFID data is scored by the success rate of RTK epoch solution and the reading accuracy of RFID, and the weights of RTK and RFID are corrected according to the standard value, wherein,

[0102] Epoch solution success rate = (number of successfully solved epochs / total number of epochs) x 100%;

[0103] Reading accuracy = (number of correctly read RFID tags / actual number of tags) x 100%.

[0104] The embodiment also sets a data fusion module for dynamically adjusting the weights of sensor data in the positioning solution strategy according to environmental changes using improved Kalman filtering technology, which adjusts the weights of multi-sensor data, dynamically fuses multi-source data, significantly improves the real-time performance and dynamic response capability of the system, and quickly adapts to environmental changes. Specifically, it is used to implement the following process:

[0105] Under each positioning solution strategy, adaptive Kalman filtering is used, the displacement calculated by the speed output by the express device is used as the predicted value, and the absolute position of RTK and / or RFID is used as the observed value. The trustworthiness of prediction and observation is dynamically balanced by weight to realize the fusion calculation of multi-source data to determine the current position of the locomotive.

[0106] In this embodiment, the error model is used to dynamically predict the cumulative error of the express device, and the formula is:

[0107] Etotal(v, d, t, H) = Ebase(d) + Erate(v, d) + Eaging(t) + Etrend(H);

[0108] Eaging(t) = Kaging x t, Kaging = 0.005% / h;

[0109] Wherein, Etotal is the cumulative error prediction value of the speed transmission device; v is the real-time speed of the locomotive (km / h); d is the cumulative running distance since the last calibration (km); t is the cumulative working time of the speed transmission device; H is the historical error trend (for example, the cumulative error value and the change rate of the last 5 calibrations); Ebase(d) is the basic error based on the cumulative running distance; Erate(v,d) is the error growth rate based on the speed interval and the cumulative distance; Eaging(t) is the aging error based on the cumulative working time of the device; Etrend(H) is the error adjustment amount based on the historical error trend H.

[0110] In an embodiment, an error growth rate adaptive selection module is also provided, which is used to select different error growth rate calculation formulas according to different speed intervals, as follows respectively:

[0111] Low speed interval (low speed < 50 km / h): Erate_low(d) = klow x d;

[0112] Wherein, klow is the error growth rate coefficient of the low speed interval (% / km), and d is the cumulative running distance (km).

[0113] Medium speed interval (50-120 km / h): Erate_mid(d) = kmid x d;

[0114] Wherein, kmid is the error growth rate coefficient of the medium speed interval (% / km).

[0115] High speed interval (> 120 km / h): Erate_high(d) = khigh x d;

[0116] Wherein, khigh is the error growth rate coefficient of the high speed interval (% / km).

[0117] The basic error compensation module is used to compensate the basic error through the cumulative distance compensation, that is, according to the cumulative running distance (reflecting the wheel diameter wear), the basic error rate is dynamically adjusted to correct the basic error, and the calculation formula of the basic error Ebase(d) is:

[0118] Ebase(d) = rbase(d) x d;

[0119] Wherein, rbase(d) is the basic error rate (% / km) dynamically corrected based on the cumulative running distance, and d is the cumulative running distance (km).

[0120] The error adjustment amount correction module is also provided, which is used to correct the overall offset of the error adjustment amount Etrend(H) by statistics of the mean value of the residual error determined for the last N times, for example, if the mean value of the residual error is continuously positive, it means that the error model underestimates the error, and the overall error prediction value is adjusted by α%.

[0121] The embodiment also provides a model optimization module for evaluating deviation of the error model prediction result and determining a model optimization strategy according to an evaluation result, wherein the model optimization module comprises a residual value determination submodule and an optimization strategy selection submodule.

[0122] The residual value determination submodule is configured to accumulate distances, determine each evaluation interval based on a speed interval and each accumulated distance interval of the speed interval, and obtain residual values of corresponding running scenarios in each evaluation interval, wherein the evaluation interval is set according to the speed interval and the accumulated distance interval of the speed interval.

[0123] The optimization strategy selection submodule is configured to determine residual values in the evaluation interval and residual mean values of continuous multiple evaluation intervals, judge whether the residual values are greater than a set threshold value, and adjust error coefficients of the interval by a least square method if the residual values are greater than the set threshold value, replace the prediction value with an average value of cumulative error values of the last three effective calibrations if the residual values of the same speed interval are greater than the set threshold value for two times in succession, use model parameters of the Nth version before the error model parameters are rolled back if the residual values are all greater than the threshold value for three times in succession, and feed back to driving if an absolute value of the residual mean value is greater than twice of an overall residual mean value, which indicates that the prediction model has low prediction result accuracy in the scenario.

[0124] The embodiment of the present application is a device embodiment corresponding to the above-mentioned method embodiment, and the specific operation of each module processing step can be understood with reference to the description of the method embodiment, which will not be repeated here.

[0125] As shown in Figure 3 The present application also provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the intelligent locomotive positioning method based on multiple sensors in the above-mentioned embodiment when executing the computer program.

[0126] The present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executable on the processor to implement the error training method in the above-mentioned embodiment or executable on the processor to implement the intelligent locomotive positioning based on multiple sensors in the above-mentioned embodiment.

[0127] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of each method can be included. Any reference to memory, storage, database or other medium used in each embodiment provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0128] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, it is described more simply, and the relevant part can be referred to the part of the method embodiment. The above-described device and system embodiments are only illustrative, and the units described as separate components can be or can not be physically separated, and the components displayed as units can be or can not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to the actual needs. Those skilled in the art can understand and implement without creative labor.

[0129] In addition, each functional module in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software functional modules.

[0130] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can still be modified, or some or all of the technical features can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and the contents not described in detail in the specification of the present application are the known technology of those skilled in the art.

Claims

1. A multi-sensor based intelligent locomotive positioning method, characterized in that, The method comprises the following steps: Collecting sensor data, including GPS position information, RFID tag information, speed transmission position information and inertial measurement unit data; Setting corresponding positioning solution strategies according to the running scene of the locomotive, and realizing automatic switching of the positioning solution strategies through state machine logic; Collecting sensor data based on the current running scene of the locomotive and the positioning solution strategy; The positioning solution strategy is to select a single sensor to determine the position information of the locomotive according to different running scenes, or to fuse the positioning data determined by multiple sensors based on a preset weight; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning results according to the GPS accuracy factor model; in a tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in a transition scene, a mixed weight and linear combination of the data of the three sensors are used; in a fault scene, only the integral information of the speed transmission device is used for positioning, and a trained error model is used for dynamic prediction of the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device; and Based on the collected sensor data, hardware-level time synchronization and spatial coordinate calibration are performed to determine the running position of the locomotive.

2. The multi-sensor based intelligent locomotive positioning method as claimed in claim 1, wherein, The positioning solution strategies corresponding to the different running scenes are as follows: Open scene: activated when the RTK signal-to-noise ratio is greater than or equal to 35 dB and the number of satellites is greater than or equal to 5, and switched to RTK + speed transmission positioning data as the position of the locomotive; Tunnel scene: activated when the RTK signal is lost and the RFID signal is valid, and switched to RFID + speed transmission positioning information fusion to determine the position of the locomotive; Transition scene: activated when the RTK signal is unstable, and RTK + speed transmission + RFID hybrid positioning information fusion is used to determine the position of the locomotive; Fault scene: activated when both RFID and RTK are invalid, then rely on the speed transmission device to determine the position of the locomotive, at this time, a trained error model is used to dynamically predict the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device.

3. The multi-sensor based intelligent locomotive positioning method as claimed in claim 2, wherein, The weight configuration of each positioning sensor data in the positioning solution strategies corresponding to the different running scenes is as follows: Open mode: RTK: 0.8-1.0, speed transmission: 0-0.2, RFID: 0; Tunnel mode: RFID: 0.5, speed transmission: 0.5; Transition mode: RTK: 0.3-0.7, RFID: 0.2-0.4, speed transmission: 0.1-0.

3.

4. The multi-sensor based intelligent locomotive positioning method as claimed in claim 1, wherein, It also includes the following steps: In each running scene, the stability of the RTK and RFID positioning data is scored by setting a real-time confidence factor, and the weights of the RTK and RFID are dynamically corrected.

5. The multi-sensor based intelligent locomotive positioning method as claimed in claim 1, wherein, The error model is used to dynamically predict the cumulative error of the speed transmission device, and the error model is as follows: Etotal(v, d, t, H) = Ebase(d) + Erate(v, d) + Eaging(t) + Etrend(H); Eaging(t) = Kaging x t, Kaging = 0.005% / h; Wherein, Etotal is the cumulative error prediction value of the speed transmission device; v is the real-time speed of the locomotive; d is the cumulative running distance since the last calibration; t is the cumulative working time length of the speed transmission device; H is the historical calibration error set; Ebase(d) is the basic error based on the cumulative running distance; Erate(v, d) is the error growth rate based on the speed interval and the cumulative distance; Eaging(t) is the aging error based on the cumulative working time length of the device; Etrend(H) is the error adjustment amount based on the historical error trend H.

6. The multi-sensor based intelligent locomotive positioning method as claimed in claim 1, wherein, Further comprising steps of: The improved Kalman filtering technology is adopted to dynamically adjust the weight of the sensor data in the positioning solution strategy according to the environmental changes, and the weight of the multi-sensor data is adjusted to dynamically fuse the multi-source data.

7. The multi-sensor based intelligent locomotive positioning method as claimed in claim 5, wherein, The error growth rate is determined according to different speed intervals, and is specifically as follows: Low speed interval: Erate_low(d) = klow x d; Wherein, klow is the error growth rate coefficient (%) of the low speed interval, and d is the cumulative running distance; Medium speed interval: Erate_mid(d) = kmid x d; Wherein, kmid is the error growth rate coefficient of the medium speed interval; High speed interval: Erate_high(d) = khigh x d; Wherein, khigh is the error growth rate coefficient of the high speed interval.

8. A multi-sensor based intelligent locomotive positioning apparatus, comprising: Comprise: A data acquisition module for acquiring sensor data, including GPS position information, RFID tag information, speed transmission position information and inertial measurement unit data; A positioning solution strategy switching module for setting the corresponding positioning solution strategy according to the locomotive running scene, and realizing the automatic switching of the positioning solution strategy through the state machine logic; Based on the current running scene of the locomotive and the positioning solution strategy, the sensor data is collected through the data acquisition module; Wherein, the positioning solution strategy is to select a single sensor to determine the position information of the locomotive according to different running scenes, or to fuse and determine the position information of the locomotive based on the preset weight of the positioning data determined by multiple sensors; that is, in an open scene, only the position coordinates determined by RTK are used to correct the positioning results according to the GPS accuracy factor model; in a tunnel scene, the absolute position information of the RFID tag and the integral information of the speed transmission device are used for fusion positioning; in a transition scene, a mixed weight and linear combination of three kinds of sensor data are used; in a fault scene, only the integral information of the speed transmission device is used for positioning, and a trained error model is used for dynamically predicting the cumulative error of the speed transmission device to correct the cumulative error of the speed transmission device; A position calculation module for hardware-level time synchronization and spatial coordinate calibration based on the collected sensor data, thereby determining the running position of the locomotive.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the multi-sensor based intelligent locomotive positioning method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to realize the multi-sensor based intelligent locomotive positioning method of any one of claims 1 to 7.

Images