Train positioning method and device based on kalman filtering algorithm, equipment and medium

Abstract

This application provides a train positioning method, apparatus, device, and medium based on the Kalman filter algorithm. The method involves acquiring the first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by tags attached to the target train. Initial position observations of the target train in each positioning subsystem are determined based on the first, second, and third position information. The observability of each positioning subsystem is determined based on the state vector and observation noise of the initial position observations. Candidate position observations of the target train are determined based on the observability of each positioning subsystem. The candidate position observations of the target train in the train positioning system are filtered using an unscented Kalman filter algorithm to obtain the target position information. This method achieves high-precision train positioning.

Description

Technical Field

[0001] This invention relates to the field of vehicle positioning technology, and more specifically, to a train positioning method, apparatus, equipment, and medium based on the Kalman filter algorithm. Background Technology

[0002] With the development of maglev technology, suspended permanent magnet maglev rail transit systems are becoming increasingly mature. Permanent magnet maglev trains have significant advantages such as being green and energy-saving, intelligent driving, safe and reliable, and economical and practical. Because maglev trains are susceptible to various factors during operation, accurate positioning is crucial for real-time monitoring of their location and operating status to ensure their safety.

[0003] In existing technologies, a single sensor or a single positioning system is typically used for train positioning. However, due to the numerous control variables, low level of automation, high integration of the control system, large hardware complexity, and strict real-time requirements of the suspended permanent magnet maglev rail transit system, for example, when the permanent magnet maglev train passes through special sections such as trees or tunnels, some sensors may be interfered with, or a single system may malfunction and fail to collect effective data. This leads to problems such as poor robustness of sensor information fusion, low positioning accuracy, and poor stability when using a single system and a single sensor for positioning, making it difficult to achieve high-precision train positioning. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a train positioning method, apparatus, device and medium based on the Kalman filter algorithm to achieve high-precision positioning of trains.

[0005] In a first aspect, embodiments of this application provide a train positioning method based on the Kalman filter algorithm, the method comprising:

[0006] The first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the Tag tag set on the target train are obtained at each time point.

[0007] Based on the first position information and the second position information at each time moment, the first initial position observation value of the target train at each time moment in the first positioning subsystem is determined using the Kalman filter algorithm. The first positioning subsystem consists of the inertial navigation system and the global navigation satellite system.

[0008] Based on the first position information and the third position information at each time moment, the second initial position observation value of the target train at each time moment in the second positioning subsystem is determined using the Kalman filter algorithm. The second positioning subsystem consists of the inertial navigation system and the tag.

[0009] The first observability of the first positioning subsystem at each time moment is determined based on the state vector of the first initial position observation value of the target train at each time moment and the observation noise of the first initial position observation value of the target train at each time moment; and the second observability of the second positioning subsystem at each time moment is determined based on the state vector of the second initial position observation value of the target train at each time moment and the observation noise of the second initial position observation value of the target train at each time moment.

[0010] The candidate position observation values ​​of the target train at each time moment are determined based on the first observability of the first positioning subsystem at each time moment and the second observability of the second positioning subsystem at each time moment. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag.

[0011] The unscented Kalman filter algorithm is used to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step.

[0012] Optionally, determining the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step includes:

[0013] The first target position observation value of the first positioning subsystem at each time is determined based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time.

[0014] The second target position observation value of the second positioning subsystem at each time is determined based on the second observability of the second positioning subsystem at each time and the second initial position observation value of the target train at each time.

[0015] The candidate position observations of the target train at each time are determined based on the first target position observations of the first positioning subsystem at each time and the second target position observations of the second positioning subsystem at each time.

[0016] Optionally, determining the first target position observation value of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation value of the target train at each time step includes:

[0017] For each time moment, the product of the first observability of the first positioning subsystem at that time moment and the first initial position observation value of the target train at that time moment is determined as the first target position observation value of the first positioning subsystem at that time moment.

[0018] The step of determining the second target position observation value of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation value of the target train at each time step includes:

[0019] For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation value of the target train at that time moment is determined as the second target position observation value of the second positioning subsystem at that time moment.

[0020] The step of determining the candidate position observation value of the target train at each time step based on the first target position observation value of the first positioning subsystem at each time step and the second target position observation value of the second positioning subsystem at each time step includes:

[0021] For each time moment, the sum of the first target position observation value of the first positioning subsystem and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

[0022] Optionally, after using the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step, the method further includes:

[0023] The trajectory of the target train is generated based on the target position information and time information of the target train at each time.

[0024] Secondly, embodiments of this application provide a train positioning device based on a Kalman filter algorithm, the device comprising:

[0025] The position information determination module is used to acquire the first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the Tag tag set on the target train at each time moment.

[0026] The first initial position observation determination module is used to determine the first initial position observation value of the target train in the first positioning subsystem at each time step based on the first position information and the second position information at each time step using a Kalman filter algorithm. The first positioning subsystem consists of the inertial navigation system and the global navigation satellite system.

[0027] The second initial position observation determination module determines the second initial position observation value of the target train at each time step using a Kalman filter algorithm based on the first position information and the third position information at each time step. The second positioning subsystem consists of the inertial navigation system and the tag.

[0028] An observability determination module is used to determine the first observability of the first positioning subsystem at each time step based on the state vector of the first initial position observation value of the target train at each time step and the observation noise of the first initial position observation value of the target train at each time step, and to determine the second observability of the second positioning subsystem at each time step based on the state vector of the second initial position observation value of the target train at each time step and the observation noise of the second initial position observation value of the target train at each time step;

[0029] The candidate position observation determination module is used to determine the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag.

[0030] The target location information determination module is used to filter the candidate location observations of the target train at each time step in the train positioning system using an unscented Kalman filter algorithm to obtain the target location information of the target train at each time step.

[0031] Optionally, when the candidate position observation determination module determines the candidate position observation of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step, it is specifically used for:

[0032] The first target position observation value of the first positioning subsystem at each time is determined based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time.

[0033] The second target position observation value of the second positioning subsystem at each time is determined based on the second observability of the second positioning subsystem at each time and the second initial position observation value of the target train at each time.

[0034] The candidate position observations of the target train at each time are determined based on the first target position observations of the first positioning subsystem at each time and the second target position observations of the second positioning subsystem at each time.

[0035] Optionally, when the candidate position observation determination module determines the first target position observation of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation of the target train at each time step, it is specifically used for:

[0036] For each time moment, the product of the first observability of the first positioning subsystem at that time moment and the first initial position observation value of the target train at that time moment is determined as the first target position observation value of the first positioning subsystem at that time moment.

[0037] When the candidate position observation determination module determines the second target position observation of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation of the target train at each time step, it is specifically used for:

[0038] For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation value of the target train at that time moment is determined as the second target position observation value of the second positioning subsystem at that time moment.

[0039] When the candidate position observation determination module determines the candidate position observation of the target train at each time step based on the first target position observation of the first positioning subsystem at each time step and the second target position observation of the second positioning subsystem at each time step, it is specifically used for:

[0040] For each time moment, the sum of the first target position observation value of the first positioning subsystem and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

[0041] Optionally, the device further includes:

[0042] The trajectory determination module is used to generate the trajectory of the target train based on the target position information and the time information of the target train at each time after the target position information determination module uses the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time in the train positioning system.

[0043] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the train positioning method based on the Kalman filter algorithm described in any of the optional embodiments of the second aspect above.

[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the train positioning method based on the Kalman filter algorithm described in any of the optional embodiments of the second aspect above.

[0045] The technical solution provided in this application includes, but is not limited to, the following beneficial effects:

[0046] The first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the tag set on the target train are obtained at each time moment. Through the above steps, the position information of the target train collected by multiple different positioning systems can be obtained.

[0047] Based on the first and second position information at each time point, a Kalman filter algorithm is used to determine the first initial position observation value of the target train at each time point in the first positioning subsystem, wherein the first positioning subsystem consists of the inertial navigation system and the global navigation satellite system; based on the first and third position information at each time point, a Kalman filter algorithm is used to determine the second initial position observation value of the target train at each time point in the second positioning subsystem, wherein the second positioning subsystem consists of the inertial navigation system and the tag; through the above steps, multiple positioning systems can be combined to obtain different positioning subsystems, and the fused position observation value of the target train under each positioning subsystem can be determined, so as to achieve the fusion of the target train's position information for the first time.

[0048] The first observability of the first positioning subsystem at each time step is determined based on the state vector of the first initial position observation value of the target train at each time step and the observation noise of the first initial position observation value of the target train at each time step. The second observability of the second positioning subsystem at each time step is determined based on the state vector of the second initial position observation value of the target train at each time step and the observation noise of the second initial position observation value of the target train at each time step. Through the above steps, the observability of the target train in each positioning subsystem can be obtained.

[0049] Based on the first observability of the first positioning subsystem at each time and the second observability of the second positioning subsystem at each time, the candidate position observations of the target train in the train positioning system at each time are determined. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag. Through the above steps, the position observations of the target train under each positioning subsystem can be fused to obtain the candidate position observations of the target train under the overall positioning system, thereby realizing the fusion of the position information of the target train in the second time.

[0050] The unscented Kalman filter algorithm is used to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step. Through the above steps, the position information of the target train obtained after fusion can be filtered to obtain more accurate target position information.

[0051] Using the above method, the position information of the target train is collected from multiple systems respectively. Then, the position information of each pair of systems is fused using the Kalman filter algorithm to obtain the position information of the target train in the subsystem composed of each pair of systems. Then, the position information of the target train in each subsystem is fused again according to the observability of each subsystem to obtain the position information of the target train in the entire positioning system. Finally, the position information of the target train in the entire positioning system is filtered to obtain more accurate position information of the target train, so as to achieve high-precision positioning of the train.

[0052] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 A flowchart of a train positioning method based on the Kalman filter algorithm provided in Embodiment 1 of the present invention is shown;

[0055] Figure 2 The flowchart of a method for determining candidate location observations provided in Embodiment 1 of the present invention is shown;

[0056] Figure 3 A schematic diagram of a train positioning device based on the Kalman filter algorithm provided in Embodiment 2 of the present invention is shown.

[0057] Figure 4 A schematic diagram of the structure of the second type of train positioning device based on the Kalman filter algorithm provided in Embodiment 2 of the present invention is shown;

[0058] Figure 5 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0060] Example 1

[0061] To facilitate understanding of this application, the following is combined with... Figure 1 The flowchart illustrating a train positioning method based on the Kalman filter algorithm provided in Embodiment 1 of the present invention will be described in detail for Embodiment 1 of this application.

[0062] See Figure 1 As shown, Figure 1 The flowchart illustrates a train positioning method based on the Kalman filter algorithm provided in Embodiment 1 of the present invention, wherein the method includes steps S101 to S106:

[0063] S101: Acquire the first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train obtained by the Tag tag set on the target train at each time moment.

[0064] Specifically, the INS inertial navigation system is based on the principle of Newtonian mechanics and uses accelerometers and gyroscopes to acquire the acceleration and angular velocity information of the launch vehicle. It calculates the launch vehicle's position information by integrating the acceleration twice, making it a highly accurate navigation system. Based on differences in their structural principles, INS systems are mainly divided into GINS platform-type inertial navigation systems, composed of physical inertial platforms, and SINS strapdown inertial navigation systems, composed of mathematical computation platforms. SINS does not rely on external information; it can achieve continuous, all-weather, three-dimensional orientation and positioning globally, without relying on external information. It has strong anti-interference capabilities and does not require data exchange with the external carrier, exhibiting strong autonomy and good concealment.

[0065] GNSS (Global Navigation Satellite System) is a navigation and positioning system that relies on artificial positioning satellites and is based on wireless communication technology, enabling global coverage. GNSS can obtain the latitude and longitude of a target, and its position information can be obtained by iteratively calculating the latitude and longitude between two points using a formula. It can continuously transmit radio signals of different frequencies and, working in conjunction with ground receiving stations, can provide trains with vehicle information such as heading, elevation, and speed 24 / 7. It updates the acquired measurement data in real time, offering advantages such as high positioning accuracy and continuous operation.

[0066] Tag tags are primarily used to detect the accuracy of train positioning. Tag tags are installed on both sides of the train, and readers are installed on the track. During train operation, as the train passes over the track with the tag readers, the readers read the data from the tags. After data filtering and conversion, the real-time train data is sent to the information fusion control center. This data is used to test the accuracy of the information fusion control center's GNSS and SINS positioning data. Because the positioning data from the train's tag tags is definitively accurate, the ground information fusion control center also uses this data to correct the train positioning data fused from the positioning methods, thereby improving the positioning accuracy of the permanent magnet maglev train.

[0067] S102: Based on the first position information and the second position information at each time moment, the first initial position observation value of the target train at each time moment in the first positioning subsystem is determined using the Kalman filter algorithm, wherein the first positioning subsystem consists of the inertial navigation system and the global navigation satellite system.

[0068] S103: Based on the first position information and the third position information at each time moment, the second initial position observation value of the target train at each time moment in the second positioning subsystem is determined using the Kalman filter algorithm, wherein the second positioning subsystem consists of the inertial navigation system and the tag.

[0069] Specifically, this invention selects sensors such as SINS, GNSS, and Tag tags. SINS is used as the basic navigation system, while GNSS and Tag tags are used as external auxiliary navigation systems. They are combined into two subsystems: SINS / GNSS and SINS / Tag tags. The SINS / GNSS subsystem is the first positioning subsystem mentioned above, and the SINS / Tag tag subsystem is the second subsystem mentioned above.

[0070] When individual sensors in a multi-sensor system measure the same characteristic parameter of a target with different confidence levels, the multiple sensors provide redundant information. Since the observation noise of individual sensors in a multi-sensor system is uncorrelated, the redundant information obtained from the measurements can be fused to reduce system uncertainty and improve system accuracy and reliability. Furthermore, due to the existence of redundant information, when one or more sensors fail, the system can still maintain normal operation using information acquired by other sensors.

[0071] The SINS / GNSS / Tag-based positioning system for maglev trains employs a two-layer structure: First, in the first layer, SINS is combined with GNSS and Tag tags to form two simultaneously operating sub-filters: SINS / GNSS and SINS / Tag tags. Optimal estimation results are obtained through synchronous processing of measurement data. Further, in the second layer, the data is processed by a main filter. When the satellite signal of the maglev train positioning system is lost, SINS and Tag tags can operate independently, using the measurement information from the Tag tags to correct the positioning output of SINS. Conversely, when the satellite signal is unstable, SINS, GNSS, and Tag tags operate simultaneously, using the Tag tags to correct the output of SINS / GNSS.

[0072] The dynamic model is designed based on a single-particle train model. The maglev train is considered as a single particle, and the interactions between vehicles are not taken into account. The equations of motion for the dynamic maglev train model are as follows.

[0073]

[0074] v k =v k-1 +a×Δk (2)

[0075] Where, x k and x k-1 v represents the position of the maglev train at times k and k-1, respectively. k and v k-1 Let represent the speeds of the maglev train at times k and k-1, respectively, where Δk is the time interval and a is the acceleration of the maglev train.

[0076] For SINS / GNSS integrated navigation systems, SINS / GNSS measurement vectors The error is composed of the differences between the position coordinates and velocity calculated by SINS and the corresponding values ​​measured by GNSS. Therefore, the error measurement equation of the SINS / GNSS integrated navigation system is as follows:

[0077]

[0078] in, For SINS / GNSS subsystem measurement vectors, Let x be the observation matrix. k+1 For the SINS / GNSS subsystem state vector, To observe noise.

[0079] Similarly, the error measurement equation for the SINS / Tag integrated navigation system can be derived as follows:

[0080]

[0081] in, For the SINS / Tag subsystem measurement vector, Let x be the observation matrix. k+1 For the SINS / Tag subsystem state vector, To observe noise.

[0082] The aforementioned navigation error model, error state, and measurement equations play a crucial role in the processing and fusion of navigation information in the proposed adaptive unscented Kalman filter.

[0083] After completing the above steps, the data measured by multiple sensors are preprocessed for redundancy. The complementarity of multi-source information is used to improve the quality of information. The observability (DOO) of each combined subsystem is calculated. The information quality and performance of the subsystem are evaluated based on the DOO value, and the observation values ​​are weighted. A mathematical model of maglev train positioning based on multi-sensor information fusion is established. Adaptive filtering is used for information fusion. During the entire information fusion process, when there is a mismatch between the observation noise covariance matrix and the noise that appears in the current filtering process, covariance matching technology is used to correct the observation noise covariance matrix in real time. Finally, the position information of the permanent magnet maglev train is obtained, revealing the interrelationship of multi-source information for permanent magnet maglev train positioning, and further improving the spatial positioning accuracy of the train.

[0084] S104: Determine the first observability of the first positioning subsystem at each time step based on the state vector of the first initial position observation value of the target train at each time step and the observation noise of the first initial position observation value of the target train at each time step, and determine the second observability of the second positioning subsystem at each time step based on the state vector of the second initial position observation value of the target train at each time step and the observation noise of the second initial position observation value of the target train at each time step.

[0085] Specifically, observability is a crucial concept in error state estimation within integrated navigation, as it determines the existence and nature of the navigation solution. While observability analysis for linear time-invariant systems is typically straightforward, analysis for linear time-varying systems is more complex, requiring evaluation of the observability Gramian. The Gramian criterion only provides "yes-no" answers and cannot describe the relative observability of all state variables. Therefore, analysis based on the DOO (Domain of Observability) value is essential for more accurately describing the observability of each navigation error.

[0086] For integrated navigation systems, the navigation error model of each subsystem has a significant impact on the accuracy and performance of the entire system. Due to the diversity of sensor types, the error models of integrated navigation systems are combined into a model complex, which includes error models for SINS, GNSS, and tags.

[0087] Let the discrete form of the error state be represented as:

[0088]

[0089] Among them, X k Let X represent the position state value of the train at time k, f(·) represent the nonlinear state transition function, and X represent the position state value of the train at time k. k-1 w represents the position state value of the train at time k-1. k-1 Z represents process noise. k H represents the observed position of the train at time k, h(*) represents the nonlinear measurement function, and H k Let ν represent the observation matrix at time k. k This indicates observation noise.

[0090] Based on the discrete form of the error state in (5), a new measurement equation can be introduced:

[0091]

[0092] in, For the sensor's observations, O k Let X represent the observation matrix. k Let k represent the state vector of the sensor at time k. This indicates observation noise.

[0093] Considering the pseudo-inverse of the observability matrix, we can obtain the relationship between the system error state and the measurement:

[0094]

[0095] Where Θ denotes the pseudo-inverse of the matrix, Represents the observation matrix O k The false rebellion.

[0096] Assumption This can be expressed in scalar form:

[0097]

[0098] in, Represents vector Γ k The i-th component, Represents Ψ k The i-th component, Representation matrix Let i be the row vector of the j-th state variable, where j = 1, 2, ..., n, and n is the number of state variables. For the observations of the k-th sensor, For the observation value of the (k+1)th sensor, For the observation value of the (k+n-1)th sensor, The observation noise of the k-th sensor, The observation noise of the (k+1)th sensor, This represents the observation noise of the (k+n-1)th sensor.

[0099] Therefore, the vector form of the observability of a linear time-varying system The calculation is as follows:

[0100]

[0101] in,

[0102] Among them, E k [(X i ) 2 ] represents the variance of the i-th state variable, E k [(Γ i ) 2 [] represents the variance of the measured values, k represents time, l represents the number of state variables, X i Let E represent the i-th state variable. k This indicates the calculation of the standard deviation. Let i be the state variable at time j. This is the i-th measurement value at time j.

[0103] Convert to scalar form as follows:

[0104]

[0105] in,

[0106] Among them, e k This indicates the standard deviation, x. iE represents the value of the i-th state variable. k [(x i ) 2 ] represents the variance of the i-th state variable, E k [(Γ i ) 2 ] represents the variance of the measured values. This indicates that the j-th state variable is in the matrix The i-th row vector, where n represents the number of state variables. Let be the observability of the i-th localization subsystem at time k. When i = 1, Let be the observability of the first localization subsystem at time k, i.e., the first observability at time k; when i = 2, Let be the observability of the second localization subsystem at time k, i.e., the second observability at time k.

[0107] S105: Based on the first observability of the first positioning subsystem at each time and the second observability of the second positioning subsystem at each time, the candidate position observation value of the target train in the train positioning system at each time is determined, wherein the train positioning system consists of the inertial navigation system, the global navigation satellite system and the tag.

[0108] Specifically, in integrated navigation systems, the performance of each sensor may differ due to variations in the external environment. In such cases, some sensors operate under favorable conditions, while others operate under harsh conditions. Therefore, it is necessary to evaluate the performance of each sensor. This invention proposes a scalar method to measure the observability of linear time-varying systems. Based on the DOO criterion, a novel numerical evaluation method for sensor performance is proposed. The degree of observability helps us obtain information on the observability of state variables and can be further used as an indicator to evaluate the performance of each navigation sensor and the autonomous management of sensors in an integrated navigation system. Therefore, the new algorithm can help provide relatively high-precision navigation solutions in a continuous manner. This invention contributes to the development of observability theory and provides developers with numerical indicators for optimizing control and navigation system design. Based on the calculated DOO value, corresponding weights are assigned to the observations of each subsystem, thereby filtering and fusing the weighted observations.

[0109] because Therefore, the DOO value calculated for each subsystem can be used as the weight of the observation to obtain the new train position observation:

[0110]

[0111] in, Let be the observability of the i-th localization subsystem at time k. When i = 1, Let be the observability of the first localization subsystem at time k, i.e., the first observability at time k; when i = 2, Let k be the observability of the second localization subsystem at time k, i.e., the second observability at time k. The candidate position observations at time k. This represents the measurement vector of the train position in the i-th positioning subsystem at time k.

[0112] S106: The unscented Kalman filter algorithm is used to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step.

[0113] Specifically, the standard unscented Kalman filter algorithm is as follows:

[0114] Considering a general nonlinear discrete-time dynamic system, the process and measurement models can be described as follows:

[0115]

[0116] Where X k ∈R n×1 Z represents the train's position and state vector at time k. k ∈R m×1 Let H represent the measurement vector of the train position observation at time k, where f(·) and h(·) represent the known nonlinear state transition and measurement functions, respectively. k Let X represent the observation matrix. k Let X represent the train's position and state vector at time k. k-1 Let w represent the train's position and state vector at time k-1. k-1 For process noise, v k To observe noise.

[0117] Assume w k-1 and v k Uncorrelated zero-mean Gaussian white noise, w k-1 The process noise is zero-mean Gaussian white noise, v k The observation noise is zero-mean Gaussian white noise, w k-1 The covariance is Q k-1 v k The covariance is R k ,satisfy:

[0118]

[0119]

[0120]

[0121] in, For process noise covariance, To measure the noise covariance, Let w be the cross-covariance of process noise and observation noise. k For process noise, v k Measure noise, Q k It is a process noise covariance and non-negative definite matrix, R k It is a positive definite matrix that measures the noise covariance.

[0122] The UKF algorithm is based on the concept that estimating a nonlinear distribution should be easier than approximating a nonlinear function. In standard UKF, an unscented transformation is implemented to generate the sigma points to be transformed nonlinearly, and the first two matrices of the transformation set are computed.

[0123] The general steps are as follows:

[0124] Step 1: Initialization:

[0125]

[0126] in, E{X0} represents the initial mean, X0 is the initial value, T is the transpose matrix symbol, and P0 is the initial error covariance.

[0127] Step 2: Sigma point calculation:

[0128]

[0129]

[0130]

[0131] in, The mean of the first sigma point at time k-1. Predict the mean at time k-1. Let P be the mean of the i-th sigma point at time k-1. k-1 Let represent the error covariance at time k-1, n be the dimension of the state, λ be the composite scaling factor, and:

[0132] λ=α 2 (n+k)-n (18)

[0133] Where α is the first adjustment parameter and κ is the second adjustment parameter. Generally, parameter α is set to 0≤α≤1, and k is generally set to 0 by default.

[0134] Step 3: State Prediction

[0135]

[0136]

[0137]

[0138] P k =P XX +Q k-1 (twenty two)

[0139] in, Let f represent the predicted mean of the i-th sigma point at time k, and f be a known nonlinear state transition function. This represents the predicted measurement value of the i-th sigma point at time k-1. k P represents the prediction error covariance at time k. XX For state measurement covariance, Q k-1 Indicates process noise. and These are the weight vectors for the mean and covariance, respectively, where n is the dimension of the state. The predicted mean at time k is defined as:

[0140]

[0141]

[0142]

[0143] Where the subscript indicates the sampling point number, m represents the mean, and c represents the covariance. The initial mean weights, For the initial covariance weights, the optimal setting for β is β = 2 for a Gaussian distribution, λ is the composite scaling factor, n is the dimension of the state, and α is the first adjustment parameter.

[0144] Step 4: Measurement and Prediction

[0145]

[0146]

[0147] in, Let h(·) represent the measurement prediction for the i-th sigma point at time k, and h(·) be a known nonlinear measurement function. Let k represent the predicted mean of the i-th sigma point at time k, where k represents time. These are the measured values ​​of the train's position. It is Z k The new train position measurement obtained by multiplying by the weight.

[0148] Step 5: Kalman gain calculation:

[0149]

[0150]

[0151]

[0152] Among them, P XZ For the cross-covariance of state measurements, P ZZ For the new information covariance, K k For Kalman gain, R k To observe the noise covariance.

[0153] Step Six: Filter Update

[0154]

[0155]

[0156] in, For posterior estimation, P k+1 This indicates the update error covariance. For K k The transpose of .

[0157] Step 7: Repeat steps 2-6 until all sample points have been calculated.

[0158] In the standard filtering algorithm described above, the noise covariance matrix needs to be initialized. However, as the number of iterations increases, if the noise covariance matrix does not match the noise encountered during the current filtering process, the standard UKF algorithm does not correct the noise covariance matrix in real time. This makes the algorithm unsuitable for time-varying noise, leading to larger errors in the final estimation result or even non-convergence. Therefore, real-time adaptive correction of the noise covariance matrix is ​​essential. Calculating the noise covariance at every moment would increase the algorithm's complexity and computational load, ultimately affecting navigation and positioning accuracy. In summary, correcting the noise covariance only when filtering anomalies significantly reduces workload and improves positioning accuracy.

[0159] The improved unscented Kalman filter algorithm used in this invention specifically includes the following steps:

[0160] Step 1: Initialize as in a standard UKF file:

[0161]

[0162] in, E{X0} represents the initial mean, X0 is the initial value, T is the transpose matrix symbol, and P0 is the initial error covariance.

[0163] Step 2: Calculate the sigma point according to formula (17):

[0164] Step 3: Determine if the filtering is abnormal:

[0165] Using covariance matching to determine the target state observation value obtained by the sensor at time k Are there any mismatches in the filtering states? This allows us to introduce the concept of innovation, let's assume the innovation is ζ. k This represents the real-time estimation error information, sum of squared innovations. Then the covariance matrix M of the innovation theory is:

[0166]

[0167] Among them, H k For the observation matrix, This represents the covariance at time k. This represents the transpose of the observation matrix. This represents the observation noise covariance.

[0168] In standard unscented Kalman filtering, the information form of Kalman filtering is used, defining the difference between the actual measured value and its predicted value as the innovation sequence:

[0169]

[0170] Among them, Z k This is the measured value of the train's position, H. k For the observation matrix, This represents the train's position and state vector at time k.

[0171] The difference between the actual measured value and its estimated value is defined as the residual sequence η. k :

[0172]

[0173] in, This represents the train's position and state vector at time k+1.

[0174] The criteria for determining if the filtering is abnormal are:

[0175]

[0176] in, For the sum of squares of new information, H k For the observation matrix, Let k be the covariance at time k. For the transpose of the observation matrix, Observation noise covariance.

[0177] If equation (37) is not true, it means that there is no abnormality in the current filtering. If the observed noise covariance is adapted, the algorithm proceeds to step five; otherwise, it indicates that the filtering is abnormal, the observed noise covariance matrix and the current filtering are no longer matched, and more attention should be paid to the current observed data. The current observed noise covariance matrix needs to be updated to adapt to the current filtering process, and the algorithm proceeds to step four.

[0178] Step 4: Real-time correction of the observation noise covariance matrix:

[0179] Substituting formula (12) into (35) yields:

[0180]

[0181] definition:

[0182]

[0183] Where, ζ k For new information, Z k The measurement vector representing the train's position at time k. H represents the predicted state vector of the train's position at time k. k Let X be the observation matrix. k Let v be the true state vector of the train's position at time k. k To observe the noise, X k+1 This represents the true state vector of the train's position at time k+1. This represents the predicted train position state vector at time k+1. Let be the deviation value of the train's position state vector at time k+1.

[0184] Therefore, based on the definitions of prediction covariance and error covariance, we can obtain:

[0185]

[0186]

[0187] in, This represents the covariance of the prediction error at time k. This represents the covariance of the prediction error at time k+1. This represents the prediction error covariance. This represents the prediction error covariance. for The transpose of the matrix, for The transpose of the matrix, is the deviation value of the train position state vector at time k.

[0188] Due to the orthogonality between the observation noise and the state estimation error, the innovation covariance can be calculated from (35) as:

[0189]

[0190]

[0191] where is the sum of the squares of the innovations, is the observation noise covariance, is the innovation covariance, and P k represents the prediction error covariance at time k.

[0192] Considering the smoothness of the covariance estimation, a recursive estimation formula is used. Finally, the measurement noise covariance is calculated as follows:

[0193]

[0194] q k =(1 - b)(1 - b k+1 ) -1 (46)

[0195] where q k is the ratio for adjusting the noise in real time according to the system and external conditions, so as to enhance the reliability and credibility of the system, reduce the system noise and external interference, improve the ability to respond to state mutations, and enhance the tracking performance. b is the genetic factor and 0 < b < 1. The size of the genetic factor can change the memory length of the filter. The appropriate value of b is an important guarantee for the convergence of the filtering algorithm error. If the noise covariance changes rapidly, more attention should be paid to the impact of the current observation data on the estimation performance at this moment, and b should take a larger value; is the adaptive observation noise covariance, H k is the observation matrix, P k is the error covariance at time k, is the transpose of the observation matrix, ε k is the innovation, and k is the time.

[0196] The value of the optimal forgetting factor b is:

[0197]

[0198] where represents the train position prediction measurement vector at time k, tr represents the trace, is the transpose of the observation matrix, Q k is the process noise covariance, A is the state transition matrix, Hk For the observation matrix, Let be the covariance matrix at time k. This is the adaptive observation noise covariance.

[0199] Step 5: According to formula (27), we can obtain:

[0200]

[0201] in, This represents the predicted observation value of the train's position. Represents the mean weight. This represents the total observed position of the train in both subsystems at time k. This represents the DOO value of the subsystem. The measurement vector represents the position of the train at time k.

[0202] Step 6: Calculate Kalman gain and filter update;

[0203]

[0204]

[0205] K k ′ =P X ′ Z (P Z ′ Z ) -1 (51)

[0206]

[0207] P k ′ +1 =P k -K k ′ P Z ′ z (K k ′ ) T (53)

[0208] Among them, P X ′ Z P represents the cross-covariance of state measurements. Z ′ Z Indicates the new information covariance. P represents the posterior estimate. k ′+1 This represents the error covariance at time k+1. Weighted by mean, This represents the predicted mean of the i-th sigma point at time k. This represents the measurement prediction of the i-th sigma point at time k. This represents the predicted observation value of the train's position. To adapt the observation noise covariance, K represents the total observed position of the train in the two subsystems at time k. k ′ P represents the Kalman gain. k Let k represent the error covariance at time k.

[0209] 7. Repeat steps 2-6 until the end.

[0210] The covariance matching technique allows for real-time updates of the observation noise covariance, reducing its impact on the filtering algorithm. The newly weighted measurements are then subjected to adaptive Kalman filtering. Even with time-varying information, the filtering algorithm still achieves satisfactory results, overcoming the interference of noise covariance errors on positioning and improving the positioning accuracy of maglev trains.

[0211] In one feasible implementation plan, see Figure 2 As shown, Figure 2 The flowchart illustrates a method for determining candidate position observations according to Embodiment 1 of the present invention. The method for determining the candidate position observations of the target train in the train positioning system at each time step, based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step, includes steps S201 to S203:

[0212] S201: Determine the first target position observation value of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation value of the target train at each time step.

[0213] S202: Determine the second target position observation value of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation value of the target train at each time step.

[0214] S203: Determine the candidate position observation value of the target train at each time step based on the first target position observation value of the first positioning subsystem at each time step and the second target position observation value of the second positioning subsystem at each time step.

[0215] Specifically, refer to the specific algorithm in formula (11).

[0216] In one feasible implementation, determining the first target position observation value of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation value of the target train at each time step includes:

[0217] For each moment, the product of the first observability of the first positioning subsystem at that moment and the first initial position observation of the target train at that moment is determined as the first target position observation of the first positioning subsystem at that moment.

[0218] The step of determining the second target position observation value of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation value of the target train at each time step includes:

[0219] For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation of the target train at that time moment is determined as the second target position observation of the second positioning subsystem at that time moment.

[0220] The step of determining the candidate position observation value of the target train at each time step based on the first target position observation value of the first positioning subsystem at each time step and the second target position observation value of the second positioning subsystem at each time step includes:

[0221] For each time moment, the sum of the first target position observation value of the first positioning subsystem and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

[0222] In a feasible implementation, after using the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step, the method further includes:

[0223] The trajectory of the target train is generated based on the target position information and time information of the target train at each time.

[0224] Specifically, for each moment, the target position information of the target train at that moment is used to generate coordinate points based on the time of that moment, and curve fitting is performed on each coordinate point to obtain the running trajectory of the target train.

[0225] In addition, the above process will be explained again here.

[0226] To overcome the shortcomings of traditional maglev train speed measurement and positioning methods, such as relying on a single sensor, susceptibility to interference, and low positioning accuracy, this invention proposes an adaptive unscented Kalman filter method based on observability. First, error state modeling is performed on the integrated navigation system. Based on the derived observability formula, the observability of each subsystem is calculated, and the information quality and performance of each subsystem are evaluated. Second, a covariance matching method is used to correct the observation noise covariance matrix in real time. Finally, the proposed adaptive Kalman filter algorithm is used for information fusion. Comparison with the standard unscented Kalman filter algorithm demonstrates the improved algorithm's positioning advantages.

[0227] The acceleration and rotation speed of the SINS sensor are integrated once to calculate velocity and attitude. Then, the velocity is integrated again to obtain position. The navigation equations require initial values ​​for position, velocity, and attitude. This process is called position and velocity initialization and attitude calibration.

[0228] I. Processing of Measurement Data from Combined Systems

[0229] For SINS, the goal is simple: to obtain its navigation results (position, velocity, attitude) from the raw angular velocity (acquired by gyroscope) and acceleration (acquired by accelerometer) data from the IMU.

[0230] An inertial navigation system (INS) is a navigation parameter calculation system that uses gyroscopes and accelerometers as sensing devices. This system establishes a navigation coordinate system based on the gyroscope output and calculates the velocity and position of the vehicle within that coordinate system based on the accelerometer output. To a certain extent, INS is a type of deductive navigation, meaning it calculates the position of the next point from the position of a known point based on continuously measured heading angles and velocities of the moving object, thus continuously measuring the current position of the moving object. In an INS, the gyroscope forms a navigation coordinate system, stabilizes the accelerometer's measurement axis within this system, and provides the heading and attitude angles. The accelerometer measures the acceleration of the moving object; the velocity is obtained by integrating the velocity once over time, and the distance is obtained by integrating the velocity once over time.

[0231] The basis for inertial navigation systems to achieve navigation and positioning is:

[0232]

[0233]

[0234] Where x(t) k () represents the displacement at time k, x0 represents the initial displacement, and t kLet k represent time, t0 represent the initial time, and v(t) represent the velocity equation. k Let v(t0) represent the velocity at time k, v(t0) represent the initial velocity, a(t) represent the acceleration equation, and t represent time.

[0235] According to formula (55), the inertial navigation system at time t can be obtained. k The speed can be calculated, and the position at time t can be solved according to formula (54).

[0236] In this invention, a SINS / GNSS chip is installed on the train, and the train's latitude and longitude location information is directly obtained by analyzing the received SINS / GNSS data. The mileage of the maglev train should be calculated using latitude and longitude information. In this invention, the Haversin formula is used to calculate the distance between adjacent latitudes and longitudes per unit time, and the total mileage is considered as the mileage of the permanent magnet maglev train.

[0237]

[0238]

[0239] Where represents the radius of the Earth. and θ represents the latitude of points A and B. A and θ B The longitudes of points A and B are represented by d, the distance between points A and B is represented by d, and Hav is an abbreviation for the semi-versus function.

[0240] Compared to SINS and GNSS, tags provide absolute positioning, allowing direct access to the location data of maglev trains without further data processing.

[0241] In the DOO (Depth of Observability) calculation of an integrated navigation subsystem, for a given system, a higher DOO value generally indicates higher estimation accuracy and faster convergence of the estimation error under the same conditions. Each error state in an integrated navigation system possesses unique observability measures and convergence rates. Therefore, the DOO value standard can be applied to measure the observability of each error state.

[0242] Observability refers to a state component at a specific time. It is the ratio of the standard deviation of the initial setup error of a state component at time 0 to the standard deviation of the filtering error of the same state component at time k. Observability is a dimensionless quantity; a larger value indicates a more significant reduction in the estimation error of the corresponding state component after Kalman filtering, or a more obvious improvement in accuracy.

[0243] The observability D of the state components is determined by a human-defined approximation. k(j) Strength and weakness are:

[0244] D k(j) ≤1 Unobservable

[0245] 1< k(j) ≤2 weak

[0246] 2< k(j) ≤10 Medium

[0247] D k(j) Top 10 (58)

[0248] The calculated DOO is a scalar. This characteristic distinguishes it from other existing forms by allowing direct computation of the DOO for each state variable without eigenvalues ​​and singular values. Furthermore, the DOO can be easily computed using the Kalman filter algorithm, and its computationally intensive usage is simpler than other methods.

[0249] To calculate the observability of the integrated navigation system, the SINS error model is used as the system model, while positioning and velocity errors are measured values. The integrated navigation system consists of two independent parts: the SINS / GNSS subsystem and the SINS / Tag subsystem. In most cases, each sensor operates independently, and its operating conditions may vary due to environmental changes. To evaluate the performance of these three subsystems, DOO is used as an indicator and calculated in scalar form using (10), thereby evaluating the information quality and performance indicators of each subsystem based on the calculated DOO values.

[0250] Based on a rough assessment of the observability of state components using human-defined parameters, if the calculated DOO value of a state component is less than or equal to 1, it indicates that the state component is unobservable. In this case, to avoid reducing positioning accuracy and causing the unscented Kalman filter to diverge, this state component can be directly removed and not included in the filtering. When the observability of a state component is weak or moderate, its proportion in the filtering algorithm can be reduced to decrease its impact on the filtering positioning accuracy and reduce the probability of filter divergence. If the observability of the currently calculated state component is strong, it proves that the positioning error of this state component is small and the positioning accuracy is high. In the entire filtering process, this state component can be used as the main part of the filtering algorithm, thereby reducing the filtering error and improving the positioning accuracy and fault tolerance of the permanent magnet maglev train.

[0251] Based on (10), the DOO value of the subsystem is obtained. The calculated DOO value of the subsystem is assigned the corresponding weight and substituted into (11) to obtain the new observation value. The new observation value is then filtered.

[0252] In adaptive UKF information fusion, the performance of the standard UKF unscented Kalman filter depends on the accurate statistical characteristics of the system noise. If the noise distribution of the inertial navigation system and the global navigation satellite system receiver is not corrected in real time, the standard UKF filtering effect will be greatly reduced or even diverge. This invention proposes an adaptive UKF that corrects the observation noise covariance in real time to overcome the limitations of the standard UKF. Based on the covariance matching technique, the covariance matrix of the observation noise is determined using the innovation sequence. The proposed algorithm can estimate and adjust the system noise statistics online, thereby enhancing the adaptive capability of the standard UKF.

[0253] The new observations obtained in (11) and the corresponding data are fused using the adaptive UKF algorithm. During the fusion process, if the observation noise covariance matrix and the current filter do not match, the current observation data should be given more attention. The current observation noise covariance matrix needs to be updated to adapt to the current filtering process. In addition, the optimal forgetting factor is introduced when recalculating the process noise covariance matrix. The noise statistics are estimated and updated online, thereby reducing the error of time-varying noise in the system state estimation and improving the positioning accuracy of the permanent magnet maglev train system.

[0254] Compared with standard UKF and strong tracking UKF, the proposed adaptive UKF calculates the DOO value of each combined subsystem based on observability, evaluates its information quality and performance, and assigns weights to the observations to prevent errors from increasing or even diverging during the filtering process. Furthermore, the adaptive UKF reduces the mean error (ME), mean relative error (MRE), and root mean square error (RMSE). Compared with the actual trajectory of the maglev train, the trajectory predicted by the proposed adaptive UKF is closer and has smaller errors, indicating that the filtering method used in this invention reduces positioning errors and improves the positioning accuracy of permanent magnet maglev trains.

[0255] Example 2

[0256] See Figure 3 As shown, Figure 3 The diagram shows a schematic of a train positioning device based on a Kalman filter algorithm according to Embodiment 2 of the present invention, wherein the device includes:

[0257] The position information determination module 301 is used to acquire the first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the Tag tag set on the target train at each time moment.

[0258] The first initial position observation determination module 302 is used to determine the first initial position observation value of the target train in the first positioning subsystem at each time step based on the first position information and the second position information at each time step using a Kalman filter algorithm. The first positioning subsystem consists of the inertial navigation system and the global navigation satellite system.

[0259] The second initial position observation determination module 303 determines the second initial position observation value of the target train at each time step using a Kalman filter algorithm based on the first position information and the third position information at each time step. The second positioning subsystem consists of the inertial navigation system and the tag.

[0260] The observability determination module 304 is used to determine the first observability of the first positioning subsystem at each time step based on the state vector of the first initial position observation value of the target train at each time step and the observation noise of the first initial position observation value of the target train at each time step, and to determine the second observability of the second positioning subsystem at each time step based on the state vector of the second initial position observation value of the target train at each time step and the observation noise of the second initial position observation value of the target train at each time step;

[0261] The candidate position observation determination module 305 is used to determine the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag.

[0262] The target position information determination module 306 is used to filter the candidate position observations of the target train at each time step in the train positioning system using an unscented Kalman filter algorithm to obtain the target position information of the target train at each time step.

[0263] In a feasible implementation, the candidate position observation determination module, when determining the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step, is specifically used for:

[0264] The first target position observation value of the first positioning subsystem at each time is determined based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time.

[0265] The second target position observation value of the second positioning subsystem at each time is determined based on the second observability of the second positioning subsystem at each time and the second initial position observation value of the target train at each time.

[0266] The candidate position observations of the target train at each time are determined based on the first target position observations of the first positioning subsystem at each time and the second target position observations of the second positioning subsystem at each time.

[0267] In a feasible implementation, the candidate position observation determination module, when determining the first target position observation of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation of the target train at each time step, is specifically used for:

[0268] For each time moment, the product of the first observability of the first positioning subsystem at that time moment and the first initial position observation value of the target train at that time moment is determined as the first target position observation value of the first positioning subsystem at that time moment.

[0269] When the candidate position observation determination module determines the second target position observation of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation of the target train at each time step, it is specifically used for:

[0270] For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation value of the target train at that time moment is determined as the second target position observation value of the second positioning subsystem at that time moment.

[0271] When the candidate position observation determination module determines the candidate position observation of the target train at each time step based on the first target position observation of the first positioning subsystem at each time step and the second target position observation of the second positioning subsystem at each time step, it is specifically used for:

[0272] For each time moment, the sum of the first target position observation value of the first positioning subsystem and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

[0273] In one feasible implementation plan, see Figure 4 As shown, Figure 4The diagram illustrates the structure of a second type of train positioning device based on the Kalman filter algorithm provided in Embodiment 2 of the present invention, wherein the device further includes:

[0274] The trajectory determination module 401 is used to generate the trajectory of the target train based on the target position information and the time information of the target train at each time after the target position information determination module uses the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time in the train positioning system.

[0275] Example 3

[0276] Based on the same application concept, see [link / reference] Figure 5 As shown, Figure 5 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown, wherein, as Figure 5 As shown, the computer device 500 provided in Embodiment 3 of this application includes:

[0277] The computer device 500 includes a processor 501, a memory 502, and a bus 503. The memory 502 stores machine-readable instructions that can be executed by the processor 501. When the computer device 500 is running, the processor 501 communicates with the memory 502 through the bus 503. When the machine-readable instructions are executed by the processor 501, the steps of the train positioning method based on the Kalman filter algorithm shown in Embodiment 2 above are executed.

[0278] Example 4

[0279] Based on the same application concept, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, performs the steps of the train positioning method based on the Kalman filter algorithm described in any of the above embodiments.

[0280] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0281] The computer program product for train positioning based on the Kalman filter algorithm provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0282] The train positioning device based on the Kalman filter algorithm provided in this embodiment of the invention can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this embodiment of the invention are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiments can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0283] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0284] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0285] In addition, the functional units in the embodiments provided by the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0286] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0287] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A train positioning method based on the Kalman filter algorithm, characterized in that, The method includes: The first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the Tag tag set on the target train are obtained at each time point. Based on the first position information and the second position information at each time moment, the first initial position observation value of the target train at each time moment in the first positioning subsystem is determined using the Kalman filter algorithm. The first positioning subsystem consists of the inertial navigation system and the global navigation satellite system. Based on the first position information and the third position information at each time moment, the second initial position observation value of the target train at each time moment in the second positioning subsystem is determined using the Kalman filter algorithm. The second positioning subsystem consists of the inertial navigation system and the tag. The first observability of the first positioning subsystem at each time moment is determined based on the state vector of the first initial position observation value of the target train at each time moment and the observation noise of the first initial position observation value of the target train at each time moment; and the second observability of the second positioning subsystem at each time moment is determined based on the state vector of the second initial position observation value of the target train at each time moment and the observation noise of the second initial position observation value of the target train at each time moment. The candidate position observation values ​​of the target train at each time moment are determined based on the first observability of the first positioning subsystem at each time moment and the second observability of the second positioning subsystem at each time moment. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag. The unscented Kalman filter algorithm is used to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step.

2. The method according to claim 1, characterized in that, The step of determining the candidate position observation values ​​of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step includes: The first target position observation value of the first positioning subsystem at each time is determined based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time. The second target position observation value of the second positioning subsystem at each time is determined based on the second observability of the second positioning subsystem at each time and the second initial position observation value of the target train at each time. The candidate position observations of the target train at each time are determined based on the first target position observations of the first positioning subsystem at each time and the second target position observations of the second positioning subsystem at each time.

3. The method according to claim 2, characterized in that, The step of determining the first target position observation value of the first positioning subsystem at each time based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time includes: For each time moment, the product of the first observability of the first positioning subsystem at that time moment and the first initial position observation value of the target train at that time moment is determined as the first target position observation value of the first positioning subsystem at that time moment. The step of determining the second target position observation value of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation value of the target train at each time step includes: For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation of the target train at that time moment is determined as the second target position observation of the second positioning subsystem at that time moment. The step of determining the candidate position observation value of the target train at each time step based on the first target position observation value of the first positioning subsystem at each time step and the second target position observation value of the second positioning subsystem at each time step includes: For each time moment, the sum of the first target position observation value of the first positioning subsystem at that time moment and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

4. The method according to claim 1, characterized in that, After using the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time step in the train positioning system to obtain the target position information of the target train at each time step, the method further includes: The trajectory of the target train is generated based on the target position information and time information of the target train at each time.

5. A train positioning device based on the Kalman filter algorithm, characterized in that, The device includes: The position information determination module is used to acquire the first position information of the target train in the SINS inertial navigation system, the second position information of the target train in the GNSS global navigation satellite system, and the third position information of the target train collected by the Tag tag set on the target train at each time moment. The first initial position observation determination module is used to determine the first initial position observation value of the target train in the first positioning subsystem at each time step based on the first position information and the second position information at each time step using a Kalman filter algorithm. The first positioning subsystem consists of the inertial navigation system and the global navigation satellite system. The second initial position observation determination module determines the second initial position observation value of the target train at each time step using a Kalman filter algorithm based on the first position information and the third position information at each time step. The second positioning subsystem consists of the inertial navigation system and the tag. An observability determination module is used to determine the first observability of the first positioning subsystem at each time step based on the state vector of the first initial position observation value of the target train at each time step and the observation noise of the first initial position observation value of the target train at each time step, and to determine the second observability of the second positioning subsystem at each time step based on the state vector of the second initial position observation value of the target train at each time step and the observation noise of the second initial position observation value of the target train at each time step; The candidate position observation determination module is used to determine the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step. The train positioning system consists of the inertial navigation system, the global navigation satellite system, and the tag. The target location information determination module is used to filter the candidate location observations of the target train at each time step in the train positioning system using an unscented Kalman filter algorithm to obtain the target location information of the target train at each time step.

6. The apparatus according to claim 5, characterized in that, When the candidate position observation determination module determines the candidate position observations of the target train in the train positioning system at each time step based on the first observability of the first positioning subsystem at each time step and the second observability of the second positioning subsystem at each time step, it is specifically used for: The first target position observation value of the first positioning subsystem at each time is determined based on the first observability of the first positioning subsystem at each time and the first initial position observation value of the target train at each time. The second target position observation value of the second positioning subsystem at each time is determined based on the second observability of the second positioning subsystem at each time and the second initial position observation value of the target train at each time. The candidate position observations of the target train at each time are determined based on the first target position observations of the first positioning subsystem at each time and the second target position observations of the second positioning subsystem at each time.

7. The apparatus according to claim 6, characterized in that, When the candidate position observation determination module determines the first target position observation of the first positioning subsystem at each time step based on the first observability of the first positioning subsystem at each time step and the first initial position observation of the target train at each time step, it is specifically used for: For each time moment, the product of the first observability of the first positioning subsystem at that time moment and the first initial position observation value of the target train at that time moment is determined as the first target position observation value of the first positioning subsystem at that time moment. When the candidate position observation determination module determines the second target position observation of the second positioning subsystem at each time step based on the second observability of the second positioning subsystem at each time step and the second initial position observation of the target train at each time step, it is specifically used for: For each time moment, the product of the second observability of the second positioning subsystem at that time moment and the second initial position observation of the target train at that time moment is determined as the second target position observation of the second positioning subsystem at that time moment. When the candidate position observation determination module determines the candidate position observation of the target train at each time step based on the first target position observation of the first positioning subsystem at each time step and the second target position observation of the second positioning subsystem at each time step, it is specifically used for: For each time moment, the sum of the first target position observation value of the first positioning subsystem at that time moment and the second target position observation value of the second positioning subsystem at that time moment is determined as the candidate position observation value of the target train at each time moment.

8. The apparatus according to claim 5, characterized in that, The device further includes: The trajectory determination module is used to generate the trajectory of the target train based on the target position information and the time information of the target train at each time after the target position information determination module uses the unscented Kalman filter algorithm to filter the candidate position observations of the target train at each time in the train positioning system.

9. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the train positioning method based on the Kalman filter algorithm as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the train positioning method based on the Kalman filter algorithm as described in any one of claims 1 to 4.

Images