A method and system for height positioning

Abstract

The application discloses a height positioning method and system. The method comprises the following steps: acquiring horizontal positioning of a terminal, determining a grid where the terminal is located, acquiring a grid air pressure model corresponding to the grid based on an ID of the grid, the grid air pressure model comprising an environmental semantic label corresponding to the grid and an air pressure reference baseline; adjusting a weight of a corresponding air pressure calibration parameter based on the environmental semantic label, combining the air pressure calibration parameter after the weight adjustment with the air pressure reference baseline to obtain a calibrated grid air pressure reference value, and combining the calibrated grid air pressure reference value with a height-air pressure change rate after the terminal sampling air pressure is subtracted from the calibrated grid air pressure reference value to vertically position the terminal. The method improves the accuracy of vertical positioning, and enhances the anti-interference ability and measurement stability in a complex environment through environmental semantic label identification and dynamic parameter weight calibration.

Description

Technical Field

[0001] This invention relates to the field of high-precision positioning, and more particularly to a height positioning method and system. Background Technology

[0002] Vertical positioning using barometric pressure sensors is an important auxiliary method in indoor and outdoor navigation. Its basic principle is to invert altitude differences by measuring changes in atmospheric pressure. However, the accuracy of this method is highly susceptible to various environmental disturbances, such as wind pressure, temperature variations, heat sources, and micro-environmental pressure distortion caused by local terrain or building structures. These disturbances can cause fluctuations in barometric pressure readings that are unrelated to actual altitude changes, thus introducing significant positioning errors.

[0003] Currently, most calibration methods in related technologies employ a single filtering algorithm or rely on fixed environmental parameters for global compensation, failing to fundamentally distinguish and quantify the differentiated impact mechanisms of different types of environmental disturbances on air pressure. For example, the instantaneous high-frequency fluctuations caused by gusts are physically completely different from the continuous air pressure offsets generated inside buildings, but existing technologies lack a clear understanding of environmental characteristics, making it impossible to achieve accurate and adaptive differentiation and calibration compensation. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application aims to provide a height positioning method and system that improves the accuracy of vertical positioning through environmental semantic label recognition and dynamic parameter weight calibration.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] Firstly, this application proposes a height positioning method, which includes:

[0007] The horizontal positioning of the terminal is obtained, the grid where the terminal is located is determined, and the grid pressure model corresponding to the grid is obtained based on the grid ID. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid.

[0008] Based on environmental semantic tags, the corresponding air pressure calibration parameters are weighted and adjusted. The weighted air pressure calibration parameters are then combined with the air pressure reference baseline to obtain the calibrated grid air pressure reference value. The terminal sampled air pressure is then subtracted from the grid air pressure reference value and combined with the altitude-air pressure change rate to perform vertical positioning of the terminal.

[0009] In one embodiment, the altitude positioning method further includes constructing a grid pressure model by:

[0010] Load the historical spatiotemporal database, which includes historical sampling data from the sampling device and the coordinate information of the sampling device at the time of sampling. The sampling device samples one or more of the following data types: air pressure, temperature, and humidity.

[0011] Based on a historical spatiotemporal database, an environmental semantic label is generated for at least one specified region. The environmental semantic label is used to characterize the probability of a specific environment existing within the specified region.

[0012] The barometric fingerprint coverage area is rasterized. Based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, the historical sampling data is mapped to the corresponding raster, and the specified area with semantic labels is mapped to the rasterized barometric fingerprint coverage area.

[0013] Based on historical sampling data mapped to the grid, a grid pressure model is constructed for each grid. The grid pressure model includes the environmental semantic label corresponding to the grid and the pressure reference baseline calculated based on the historical sampling data mapped to the grid.

[0014] In one embodiment, environmental semantic labels are generated for at least one specified region based on a historical spatiotemporal database, including:

[0015] Based on historical spatiotemporal databases, classifier processing or Bayesian posterior processing is performed to obtain the probability of a specified region for any type of environmental semantic label, and the one with the highest probability is selected as the environmental semantic label of the specified region.

[0016] In one embodiment, the height localization method further includes a writing stability coefficient for environmental semantic tags, including:

[0017] The variance of the location change of the environmental semantic tag is calculated, and the variance is mapped by an inverse function to obtain the stability coefficient of the environmental semantic tag. The stability coefficient is used to influence the semantic trajectory formed by the environmental semantic tag.

[0018] In one embodiment, the method further includes:

[0019] Construct a mapping table between barometric calibration parameters and environmental semantic labels, where each environmental semantic label corresponds to at least one barometric calibration parameter.

[0020] Based on the mapping table between air pressure calibration parameters and environmental semantic tags, the weights of environmental semantic tags on the corresponding air pressure calibration parameters are adjusted.

[0021] The air pressure calibration parameters include: temperature correction factor, humidity correction factor, wind disturbance term, structural disturbance term, and sensor bias. Among them, the wind disturbance term represents the influence of wind on air pressure, and the structural disturbance term represents the influence of the condition of the structure or facility on air pressure.

[0022] In one embodiment, the method further includes performing wind disturbance discrimination, including:

[0023] If the air pressure fluctuation is detected, and the air pressure fluctuation amplitude is greater than or equal to the preset air pressure threshold within a preset time, and there is a high frequency pulsation, then it is determined that there is wind disturbance in the air pressure data sequence.

[0024] When wind disturbance is detected in the air pressure data sequence, power spectrum or variance detection is performed on the air pressure data sequence sampled from the terminal. The equivalent amplitude is estimated by matching the reference wind speed and wind direction. This equivalent amplitude is then exponentially decayed and used as the wind disturbance term.

[0025] In one embodiment, the method further includes:

[0026] Based on the horizontal positioning coordinates, the horizontal positioning standard deviation is calculated. When the horizontal positioning standard deviation is greater than the grid accuracy, the grid pressure model corresponding to the first grid where the terminal is located is obtained, as well as the grid pressure model corresponding to the second grid adjacent to the first grid. Based on the environmental semantic labels of the first grid and the second grid, semantic trajectory constraints are generated.

[0027] When the terminal's sampled air pressure is subtracted from the grid air pressure reference value and combined with the altitude-pressure change rate, the vertical positioning result is calibrated based on semantic trajectory constraints.

[0028] In one embodiment, the method further includes:

[0029] The pressure calibration value is obtained by subtracting the grid pressure reference value from the terminal sampling pressure. Multiple pressure calibration values ​​are compared to calculate the residual. The pressure calibration value calculated by comparing the residuals satisfies the following condition: the distance between the terminal sampling positions corresponding to any two pressure calibration values ​​is within a preset range.

[0030] Based on the environmental semantic labels of the terminal sampling location, a residual threshold is set, and the residual is compared with the residual threshold to generate a quality index for each air pressure calibration value;

[0031] Based on quality indicators, the various air pressure calibration values ​​are weighted and fused to update the air pressure reference baseline of the corresponding grid and recalculate the probability of environmental semantic labels in the grid.

[0032] Secondly, this application also provides an altitude positioning system applied to a terminal, the system comprising:

[0033] The sampling module is configured to sample at least air pressure data;

[0034] The data processing module is configured to acquire the horizontal positioning of the terminal;

[0035] The data processing module is also configured to determine the grid where the terminal is located based on the horizontal positioning results of the terminal, and obtain the grid pressure model corresponding to the grid based on the grid ID. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid.

[0036] The data processing module adjusts the weights of the corresponding air pressure calibration parameters based on environmental semantic tags. It then combines the weighted air pressure calibration parameters with the air pressure reference baseline to obtain the calibrated grid air pressure reference value. Finally, it subtracts the grid air pressure reference value from the terminal's sampled air pressure and combines it with the altitude-air pressure change rate to perform vertical positioning of the terminal.

[0037] In one embodiment, the altitude positioning system is applied to a cloud server, and the altitude positioning system includes:

[0038] The historical spatiotemporal database includes historical sampling data from the sampling equipment and the coordinate information of the sampling equipment at the time of sampling. The sampling equipment samples one or more of the following types of data: air pressure, temperature, and humidity.

[0039] The data processing module is configured to generate environmental semantic labels for at least one specified area based on a historical spatiotemporal database. The environmental semantic labels are used to characterize the probability of a specific environment existing in the specified area. The data processing module rasterizes the barometric fingerprint coverage area, maps the historical sampling data to the corresponding raster based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, and maps the specified area with semantic labels to the rasterized barometric fingerprint coverage area.

[0040] The grid pressure model generation module is configured to construct a grid pressure model for each grid based on historical sampling data mapped to the grid. The grid pressure model includes the environmental semantic label corresponding to the grid and the pressure reference baseline calculated based on the historical sampling data mapped to the grid.

[0041] The data transmission module is used to send the grid pressure model. The grid pressure model is used to: adjust the weights of the corresponding pressure calibration parameters through environmental semantic tags, so that the terminal can combine the weighted pressure calibration parameters with the pressure reference baseline to obtain the calibrated grid pressure reference value; and the terminal can combine the sampled pressure minus the grid pressure reference value with the altitude-pressure change rate to perform vertical positioning of the terminal.

[0042] The aforementioned altitude positioning method determines the grid cell to which the terminal belongs by acquiring its horizontal positioning and obtains a pre-set grid pressure model for that grid. This grid pressure model includes environmental semantic tags and a pressure reference baseline. Further, based on the environmental semantic tags, the weights of the corresponding pressure calibration parameters are dynamically adjusted. The weighted parameters are then fused with the pressure reference baseline to generate a calibrated grid pressure benchmark value. Finally, vertical positioning is achieved by calculating the difference between the terminal's sampled pressure and this benchmark value, and combining this with an altitude-pressure change rate model. This method improves the accuracy of vertical positioning and enhances the anti-interference capability and measurement stability in complex environments through environmental semantic tag recognition and dynamic parameter weight calibration. Attached Figure Description

[0043] Figure 1 Here is a flowchart of a height positioning method in one embodiment;

[0044] Figure 2 This is a flowchart illustrating the construction of a grid pressure model in one embodiment;

[0045] Figure 3 This is a structural diagram of a height positioning system in one embodiment. Detailed Implementation

[0046] To enable those skilled in the art to better understand the present application, the technical solutions in specific embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0047] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0048] In one embodiment, such as Figure 1 As shown, a height positioning method is provided, which includes the following steps:

[0049] Step 101: Obtain the horizontal positioning of the terminal, determine the grid where the terminal is located, and obtain the grid pressure model corresponding to the grid based on the grid ID. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid.

[0050] Specifically, the terminal can perform horizontal positioning by connecting to a base station. For example, the terminal can use cellular triangulation for coarse positioning (the accuracy of cellular triangulation is in the tens to hundreds of meters), thereby determining which pre-divided geographic grid the terminal is located in (such as a 20m x 20m square). Horizontal positioning information can also be obtained through positioning methods such as Bluetooth, GNSS, and radio.

[0051] It should be noted that each grid cell has a unique ID. Terminals can use this ID to request or retrieve the corresponding grid pressure model from the server. The grid pressure model can be a data file, which includes at least environmental semantic labels and a pressure reference baseline. The environmental semantic labels can be things like wind vents, water bodies, or shaded areas, describing inherent environmental characteristics. The pressure reference baseline represents the historical average pressure value of that grid cell under standard conditions.

[0052] For example, when a mobile terminal locates a grid (ID: 6075) and obtains its grid pressure model, the grid pressure model can obtain the corresponding environmental semantic label and pressure reference baseline based on the ID. For instance, the environmental semantic label for the grid with ID 6075 is "wind vent," and the baseline pressure is 1013.2 hPa.

[0053] Step 102: Adjust the weights of the corresponding air pressure calibration parameters based on environmental semantic tags, combine the weighted air pressure calibration parameters with the air pressure reference baseline to obtain the calibrated grid air pressure reference value, subtract the grid air pressure reference value from the terminal sampled air pressure and combine it with the altitude-air pressure change rate to perform vertical positioning of the terminal.

[0054] Specifically, after obtaining the environmental semantic label and air pressure reference baseline in the grid pressure model corresponding to its grid ID, the terminal begins to continuously collect sampling data such as air pressure and temperature for that grid. For example, when a pressure fluctuation is detected, it first determines whether the pressure fluctuation is related to the features indicated by the current environmental semantic label. If a high-frequency, short-term, severe pressure fluctuation is detected in a grid labeled "wind vent," then the fluctuation is highly correlated with this environmental semantic label. Further, if the correlation with the environmental semantic label is confirmed, the weights of the air pressure calibration parameters corresponding to that environmental semantic label can be adjusted (for example, in the case of "wind vent," the weight or confidence of the wind disturbance parameter can be increased, while the weights of other parameters can be decreased). If the correlation between the air pressure fluctuation and the features indicated by the current environmental semantic label is low, the weight of the corresponding air pressure calibration parameter can also be decreased.

[0055] Furthermore, by combining the weighted calibration parameters (such as wind disturbance terms and temperature correction coefficients) with a stable air pressure reference baseline, a calibrated grid air pressure reference value that better reflects the current instantaneous actual situation is calculated, which is the air pressure value that should theoretically exist at this moment and place.

[0056] Finally, by subtracting the calibrated grid pressure reference value from the original sampled air pressure collected by the terminal, the net air pressure difference caused by the height difference can be obtained. Then, by substituting this net air pressure difference into the standard height-pressure change rate formula (such as the air pressure decreasing by about 1 hPa for every 8 meters of ascent), the precise relative height change can be calculated, thereby achieving accurate vertical positioning, such as determining whether a user has walked from the first floor to the third floor.

[0057] In this embodiment, the method uses a base station to perform horizontal positioning of the terminal to determine its assigned grid and obtains a pre-set grid pressure model for that grid. The grid pressure model includes environmental semantic tags and a pressure reference baseline. Further, based on the environmental semantic tags, the weights of the corresponding pressure calibration parameters are dynamically adjusted, and the weighted parameters are fused with the pressure reference baseline to generate a calibrated grid pressure benchmark value. Finally, vertical positioning is achieved by calculating the difference between the terminal's sampled pressure and this benchmark value, and combining this with an altitude-pressure change rate model. This method improves the accuracy of vertical positioning and enhances anti-interference capabilities and measurement stability in complex environments through environmental semantic tag recognition and dynamic parameter weight calibration.

[0058] In one embodiment, such as Figure 2 As shown, the altitude positioning method also includes constructing a grid pressure model by means of the following steps:

[0059] Step 201: Load the historical spatiotemporal database. The historical spatiotemporal database includes historical sampling data from the sampling device and the coordinate information of the sampling device at the time of sampling. The sampling device samples one or more of the following data types: air pressure, temperature, and humidity.

[0060] Specifically, a historical spatiotemporal database can refer to a multi-dimensional sensor dataset pre-accumulated by sampling devices, containing timestamps and geographic coordinates. This database can contain data on air pressure, temperature, humidity, etc., collected at different times and locations by various sampling devices (such as smartphones, IoT terminals, or fixed weather stations). It can also record the coordinates (such as latitude, longitude, and altitude) and time information corresponding to each data point, thus forming a spatiotemporal data foundation that reflects long-term environmental change patterns and spatial distribution characteristics.

[0061] Step 202: Based on the historical spatiotemporal database, generate environmental semantic labels for at least one specified region. The environmental semantic labels are used to characterize the probability of a specific environment existing within the specified region.

[0062] Specifically, generating environmental semantic labels based on historical spatiotemporal databases can refer to utilizing long-term accumulated spatiotemporally labeled data such as air pressure, temperature, and humidity in a historical spatiotemporal database of a specified area. Machine learning algorithms (such as classifiers or Bayesian inference) are then used to extract features and recognize patterns from the time-series data within the specified area, thereby calculating a probability estimate that the area belongs to a specific environmental type (such as a windy area, shaded area, water body, or heat source). This probability estimate is the environmental semantic label, which quantifies the confidence level of the existence of a certain stable environmental characteristic in the area.

[0063] Step 203: Rasterize the barometric fingerprint coverage area. Based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, map the historical sampling data to the corresponding grid, and map the specified area with semantic labels to the rasterized barometric fingerprint coverage area.

[0064] Specifically, gridding the barometric fingerprint coverage area can refer to dividing a city or park into regular spatial unit grids with a fixed geometric precision (e.g., 20m x 20m). Based on the device coordinate information of each sampling data record in the historical spatiotemporal database, discrete historical barometric pressure, temperature, humidity, and other sampling data are aggregated into the corresponding grid units through a spatial matching algorithm.

[0065] Furthermore, through spatial overlay analysis, regions with probabilistic environmental semantic labels (e.g., wind gap probability of 0.9) generated in advance through machine learning are registered with the units of the grid network, so that each grid inherits the environmental semantic labels of the area it covers, ultimately forming a barometric fingerprint coverage area where each grid ID uniquely corresponds to a set of historical data statistical features and one or more environmental semantic labels.

[0066] Step 204: Based on the historical sampling data mapped to the grid, construct a grid pressure model for each grid. The grid pressure model includes the environmental semantic label corresponding to the grid and the pressure reference baseline calculated based on the historical sampling data mapped to the grid.

[0067] Specifically, environmental semantic labels can be directly associated with the corresponding grid as prior knowledge (e.g., a grid is labeled as having a wind gap probability of 0.85). Secondly, the pressure reference baseline can be calculated by performing denoising (e.g., removing outliers), weighting (e.g., based on data quality scores and time decay coefficients), and regression analysis on historical pressure data from multiple years and time periods within the grid. This allows for the calculation of the expected stable pressure under standard conditions (e.g., after sea level conversion and removal of typical disturbances). Finally, each grid can generate a grid pressure model containing environmental semantic labels and the pressure reference baseline.

[0068] In one embodiment, based on a historical spatiotemporal database, environmental semantic labels are generated for at least one specified region, including:

[0069] Based on historical spatiotemporal databases, classifier processing or Bayesian posterior processing is performed to obtain the probability of a specified region for any type of environmental semantic label, and the one with the highest probability is selected as the environmental semantic label of the specified region.

[0070] Specifically, multidimensional features of meteorological data within a specified area are obtained (e.g., spectral energy of air pressure sequences, daily temperature variation, humidity stability, etc.). These features are then inferred using pre-trained classifiers (such as random forests or neural networks) or Bayesian probabilistic models (based on historical prior distributions and likelihood calculations) to calculate the probability value of the area belonging to each environmental category (e.g., wind gaps, water bodies, heat sources, shaded areas, etc.). Finally, the category with the highest probability is selected as the final environmental semantic label for the area.

[0071] For example, after analyzing the historical air pressure data of a specified area, it was found that the high-frequency fluctuation energy was significantly higher than the average level (wind gap characteristic) and the daily temperature range was small (water body or shadow characteristic). However, the Bayesian posterior calculation showed that the probability of wind gap reached 0.88, which was much higher than 0.07 for water body and 0.05 for shadow area. Therefore, the environmental semantic label of this area was marked as wind gap.

[0072] In one embodiment, the height localization method further includes a writing stability coefficient for environmental semantic tags, including:

[0073] The variance of the location change of the environmental semantic tag is calculated, and the variance is mapped by an inverse function to obtain the stability coefficient of the environmental semantic tag. The stability coefficient is used to influence the semantic trajectory formed by the environmental semantic tag.

[0074] Specifically, the system continuously monitors the changes in the geographical area covered by each environmental semantic tag over time, calculating the variance of its location boundary coordinates (a measure of the degree of fluctuation in the area's range). Further, the variance is converted into a stability coefficient between 0 and 1 using an inverse function mapping (such as using a reciprocal or negative exponential function). For example, a larger variance results in a higher stability coefficient closer to 0, indicating that the spatial range of the environmental feature tag (such as a wind gap) is unstable and highly time-varying; conversely, a coefficient closer to 1 indicates that the feature spatial distribution of the environmental feature tag is stable. This stability coefficient can be used to weight and adjust the contribution of semantic tags in forming semantic trajectories, thus concatenating environmental semantic tags from consecutive locations into a semantic trajectory as the terminal moves.

[0075] In one embodiment, the method further includes:

[0076] Construct a mapping table between barometric calibration parameters and environmental semantic labels, where each environmental semantic label corresponds to at least one barometric calibration parameter.

[0077] Based on the mapping table between air pressure calibration parameters and environmental semantic tags, the weights of environmental semantic tags on the corresponding air pressure calibration parameters are adjusted.

[0078] The air pressure calibration parameters include relevant data that affect air pressure. Specifically, the air pressure calibration parameters include at least: temperature correction factor, humidity correction factor, wind disturbance term, structural disturbance term, and sensor bias. Among them, the wind disturbance term represents the influence of wind on air pressure, and the structural disturbance term represents the influence of the condition of the structure or facility on air pressure.

[0079] Specifically, the mapping table can be a knowledge base or lookup table that predefines the associations between different types of environmental semantic labels and one or more barometric calibration parameters. The purpose of constructing this mapping table is to transform the qualitative description of the environment (environmental semantic labels) into quantitative physical correction instructions (which parameters should be adjusted). It is necessary to know that the environmental semantic label for a certain location is a wind vent, and also to know which parameter the wind vent primarily affects, and in what way.

[0080] This mapping table can be predefined based on physical principles and historical data statistical analysis. For example, fluid dynamics shows that wind causes dynamic fluctuations in air pressure, and historical data has repeatedly verified that in areas marked as wind gaps, high-frequency fluctuations in air pressure (wind disturbance term) are highly correlated with wind speed. Therefore, the wind disturbance term parameter is associated with the environmental semantic label of the wind gap in the mapping table. When the terminal determines that the current air pressure fluctuation is related to the environmental semantic label of the current grid (e.g., high-frequency fluctuations are detected in the wind gap label area), a weight adjustment is performed. First, the mapping table is queried to find one or more air pressure calibration parameters associated with the current environmental semantic label, and then the weight or confidence of the air pressure calibration parameters is increased. Here, weight adjustment means that in subsequent comprehensive calibration calculations, changes in the weighted air pressure calibration parameter values ​​have a greater impact on the final calibration result.

[0081] The air pressure calibration equation is as follows:

[0082] Predicted air pressure = air pressure reference baseline + a × temperature correction factor * temperature deviation + b × humidity correction factor * humidity deviation + c × wind disturbance term + d × structural disturbance term + sensor bias.

[0083] It should be noted that the temperature correction factor quantifies the impact of local temperature deviation on barometric pressure sensor readings. For example, for every 1°C increase in temperature compared to the reference temperature, the barometric pressure reading might be artificially inflated by a certain number of Pascals. Its function is to eliminate the influence of temperature on the barometric pressure reading, allowing the barometric pressure reading to more accurately reflect changes in altitude.

[0084] The humidity correction factor can quantify the impact of local humidity deviation on barometer readings. Its function is to eliminate the influence of humidity on barometer readings, so that barometer readings can more accurately reflect changes in altitude.

[0085] The wind disturbance term is a dynamically changing value (unit: Pa), representing the instantaneous dynamic pressure effect of wind on air pressure. The greater the wind speed, the larger the absolute value of this term. It is usually positive (wind pressure causes a temporary increase in air pressure reading).

[0086] The structural disturbance term is a relatively stable bias value (unit: Pa), representing the continuous impact of permanent or semi-permanent structures (such as buildings and ventilation shafts) on local atmospheric circulation, causing the air pressure in this area to deviate from the theoretical value of the surrounding area in a long-term and stable manner.

[0087] Sensor bias is the inherent zero-point error (unit: Pa) of the terminal device's own barometric pressure sensor. It is a device-related parameter that needs to be eliminated through calibration.

[0088] It should be noted that a, b, c, and d are the weights of the corresponding air pressure calibration parameters, which are affected by the environmental semantic label, and a + b + c + d = 1. For example, if the environmental semantic label is "water body or reservoir", then the weight "b" of the humidity-related air pressure calibration parameter needs to be increased accordingly, and other weights need to be adjusted accordingly; if the environmental semantic label is "heat source", then the weight "a" of the temperature-related air pressure calibration parameter needs to be increased accordingly, and other weights need to be adjusted accordingly; if the environmental semantic label is "wind vent", then the weight "c" of the wind disturbance term needs to be increased accordingly, and other weights need to be adjusted accordingly; if the environmental semantic label is "shaded area", then the weight "d" of the structural disturbance needs to be increased accordingly, and other weights need to be adjusted accordingly.

[0089] For example, suppose there exists an area with the environmental semantic label "wind gap," which corresponds to the wind disturbance item. The weight adjustment process is as follows: the terminal obtains the environmental semantic label of this area as "[wind gap, probability 0.95]." Further, the terminal detects high-frequency, large-amplitude fluctuations in air pressure within a short period (e.g., a change of 15 Pa within 3 seconds). This fluctuation is determined to be highly correlated with the wind gap label. After querying the mapping table, it is decided to increase the weight of the wind disturbance item parameter. When calculating the final calibration value, most of the observed air pressure fluctuations are preferentially attributed to changes in the wind disturbance item, and a larger wind disturbance value (e.g., +12 Pa) is calculated to explain this fluctuation, thereby avoiding severe jumps in vertical positioning.

[0090] In one embodiment, the method further includes wind disturbance discrimination, including:

[0091] If the air pressure fluctuation is detected, and the air pressure fluctuation amplitude is greater than or equal to the preset air pressure threshold (e.g., 10 Pa) within a preset time (e.g., within 3 seconds), and there is high-frequency pulsation, then it is determined that there is wind disturbance in the air pressure data sequence.

[0092] When wind disturbance is detected in the air pressure data sequence, power spectrum or variance detection is performed on the air pressure data sequence sampled from the terminal. The equivalent amplitude is estimated by matching the reference wind speed and wind direction. This equivalent amplitude is then exponentially decayed and used as the wind disturbance term.

[0093] Specifically, a trigger condition is set: when a fluctuation amplitude of ≥10Pa in the air pressure sequence is detected within a 3-second time window, accompanied by high-frequency pulsations (identified through short-time Fourier transform or wavelet analysis), wind disturbance is immediately determined to exist. After confirming the existence of wind disturbance, power spectral density analysis is performed on the air pressure time series data (to locate the frequency band where energy is concentrated) or the sliding variance is calculated (to quantify the fluctuation intensity). This data is then combined with reference wind speed and direction data provided by the base station (such as measurements from an ultrasonic weather station) for cross-sensor matching verification to eliminate equipment noise or other interference.

[0094] The equivalent disturbance amplitude (unit: Pa) generated by wind pressure is further calculated based on the spectral energy or variance amplitude, and an exponential decay model is assigned to this amplitude (e.g., setting the decay time constant to 10 seconds) to simulate the natural dissipation process of the wind pressure effect. Finally, the dynamically decayed disturbance amplitude is output as the wind disturbance term and substituted into the air pressure calibration equation.

[0095] In one embodiment, the method further includes:

[0096] Based on the horizontal positioning coordinates obtained from base station positioning, the horizontal positioning standard deviation is calculated. When the horizontal positioning standard deviation is greater than the grid accuracy, the grid pressure model corresponding to the first grid where the terminal is located and the grid pressure model corresponding to the second grid adjacent to the first grid are obtained. Based on the environmental semantic labels of the first grid and the second grid, semantic trajectory constraints are generated.

[0097] When the terminal's sampled air pressure is subtracted from the grid air pressure reference value and combined with the altitude-pressure change rate, the vertical positioning result is calibrated based on semantic trajectory constraints.

[0098] Specifically, the terminal can obtain horizontal coordinates through base station positioning, but these horizontal coordinates have errors, the accuracy of which is quantified by the horizontal positioning standard deviation (1σ). Since the map has been divided into grids with fixed precision (e.g., 20m x 20m), when the horizontal positioning standard deviation is greater than the grid precision (e.g., a positioning error of 35 meters while the grid side length is 20 meters), it means that it is impossible to be certain which grid the terminal is located in.

[0099] At this point, instead of acquiring only the grid pressure model of a single grid cell, the system acquires the grid pressure models of the first grid cell where the terminal is most likely located and its surrounding adjacent second grid cells (usually a neighboring cell).

[0100] By analyzing the environmental semantic labels (such as water bodies and shaded areas) of the first and second grids, semantic trajectory constraints (e.g., movement from water bodies to shaded areas) are generated based on the terminal's movement trends. During vertical positioning calculations, these semantic trajectory constraints are used to determine which grid's barometric pressure model yields the most logically consistent results.

[0101] For example, assuming the terminal's horizontal positioning error reaches 30 meters (grid accuracy is 20 meters), and two grid pressure models are loaded: the first grid represents the labeled water body with a baseline pressure of 1012.0 hPa, and the second grid represents the labeled shaded area with a baseline pressure of 1012.2 hPa. If the terminal's measured pressure is 1011.8 hPa, calculating the height reveals that the first grid descends by 1.7 meters and the second grid descends by 3.3 meters. At this point, semantic trajectory constraints indicate that the terminal has moved from the water body area to the shaded area, both of which are flat terrain. Weighted fusion of the two results ultimately calibrates the vertical height to a decrease of approximately 2.5 meters, preventing positioning jumps.

[0102] In one embodiment, the method further includes:

[0103] The pressure calibration value is obtained by subtracting the grid pressure reference value from the terminal sampling pressure. Multiple pressure calibration values ​​are compared to calculate the residual. The pressure calibration value calculated by comparing the residuals satisfies the following condition: the distance between the terminal sampling positions corresponding to any two pressure calibration values ​​is within a preset range.

[0104] Based on the environmental semantic labels of the terminal sampling location, a residual threshold is set, and the residual is compared with the residual threshold to generate a quality index for each air pressure calibration value;

[0105] Based on quality indicators, the various air pressure calibration values ​​are weighted and fused to update the air pressure reference baseline of the corresponding grid and recalculate the probability of environmental semantic labels in the grid.

[0106] Specifically, the sampled air pressure reported by each terminal is subtracted from the air pressure reference baseline of its corresponding grid to obtain a single air pressure calibration value (reflecting the deviation between the measured and theoretical values ​​at that point). When multiple terminals report calibration values ​​at similar locations (within a preset range, such as within 20 meters), these calibration values ​​are compared and the residuals between them are calculated. Further, differentiated residual judgment thresholds are set according to the environmental semantic label type of these sampling points. For example, a wider threshold (such as ±5Pa) is set in the wind vent environmental semantic label area, and a stricter threshold (such as ±1Pa) is set in the water body environmental semantic label area. After comparing the actual residuals with the thresholds, a quality index (such as a confidence level between 0 and 1) is generated for each calibration value. Based on the quality index, multiple calibration values ​​are weighted and fused (high confidence data have higher weights), and the air pressure reference baseline of the corresponding grid is updated with the fusion result, while the environmental semantic label probability of that grid is reassessed simultaneously.

[0107] For example, suppose that within the same water body labeling grid, two terminals 15 meters apart report air pressure calibration values ​​almost simultaneously: Terminal A reports +0.8 Pa, and Terminal B reports -0.3 Pa. The calculated residual is 1.1 Pa. Since the environmental semantic label for this area is water, a strict threshold (e.g., ±1 Pa) can be used. Because the residual of 1.1 Pa slightly exceeds the threshold, a moderate quality index (e.g., 0.7) needs to be generated for the two data points. After weighted fusion, the calibration value of +0.3 Pa is obtained. Based on this calibration value, the air pressure reference baseline for this grid is fine-tuned (e.g., adjusting the original baseline of 1012.0 hPa to 1012.03 hPa). Simultaneously, since the residual performance is better than the characteristic of the environmental semantic label being a wind gap, the probability of the water body environmental semantic label for this grid can be increased from 0.85 to 0.88.

[0108] As an alternative implementation, machine learning can be used to improve the accuracy of the grid-based barometric pressure model by leveraging data uploaded from the terminal. Specifically, a sampling device is used as a reference, with known temporal and spatial information, including at least air pressure and temperature, and then including but not limited to humidity, wind direction, wind speed, rainfall, evaporation, and ground temperature. The sampling device uploads sampled data at a certain frequency. The uploaded data is used for fitting to derive preliminary semantic labels. These labels and various thresholds can be adjusted based on expert experience to confirm their accuracy, thus establishing a preliminary grid-based barometric pressure model. Subsequently, the model can be continuously iterated, adjusted, and learned from the data uploaded from the terminal to improve its accuracy.

[0109] It should be noted that when generating environmental semantic labels for a raster, there can be multiple environmental semantic labels. Whether to integrate these environmental semantic labels into the raster barometric pressure model is determined by the confidence level (i.e., the probability of the environmental semantic labels). A confidence threshold can be set. When the confidence level is higher than the confidence threshold, the corresponding environmental semantic label is integrated into the raster corresponding to the raster barometric pressure model.

[0110] Based on the same concept, such as Figure 3 As shown, this application also provides an altitude positioning system, which is applied to a terminal. The altitude positioning system includes:

[0111] Sampling module 301 is configured to sample at least air pressure data;

[0112] Communication module 302 is configured to communicate with a base station;

[0113] The data processing module 303 is configured to perform base station positioning based on the communication data of the base station to determine the horizontal positioning of the terminal;

[0114] The data processing module 303 is also configured to determine the grid where the terminal is located based on the horizontal positioning result of the terminal, and request the grid pressure model corresponding to the grid from the base station based on the grid ID. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid.

[0115] The data processing module 303 adjusts the weights of the corresponding air pressure calibration parameters based on environmental semantic tags, combines the weighted air pressure calibration parameters with the air pressure reference baseline to obtain the calibrated grid air pressure reference value, and combines the terminal sampled air pressure with the grid air pressure reference value and the altitude-air pressure change rate to perform vertical positioning of the terminal.

[0116] In one embodiment, such as Figure 3 As shown, the altitude positioning system is applied to the cloud server, and the altitude positioning system includes:

[0117] Historical spatiotemporal database 304. The historical spatiotemporal database includes historical sampling data from the sampling equipment and the coordinate information of the sampling equipment at the time of sampling. The sampling equipment samples one or more of the following types of data: air pressure, temperature, and humidity.

[0118] The data processing module 303 is configured to generate environmental semantic labels for at least one specified area based on a historical spatiotemporal database. The environmental semantic labels are used to characterize the probability of a specific environment existing in the specified area. The data processing module rasterizes the barometric fingerprint coverage area, maps the historical sampling data to the corresponding raster based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, and maps the specified area with semantic labels to the rasterized barometric fingerprint coverage area.

[0119] The grid pressure model generation module 305 is configured to construct a grid pressure model for each grid based on historical sampling data mapped to the grid. The grid pressure model includes the environmental semantic label corresponding to the grid and the pressure reference baseline calculated based on the historical sampling data mapped to the grid.

[0120] The data transmission module 306 is used to send the grid pressure model. The grid pressure model is used to: adjust the weights of the corresponding pressure calibration parameters through environmental semantic tags, so that the terminal can combine the weighted pressure calibration parameters with the pressure reference baseline to obtain the calibrated grid pressure reference value; and the terminal can combine the sampled pressure minus the grid pressure reference value with the altitude-pressure change rate to perform vertical positioning of the terminal.

[0121] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The above are merely preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application are included within the scope of protection of this application.

Claims

1. A height positioning method, characterized in that, The height positioning method includes: The horizontal positioning of the terminal is obtained, the grid where the terminal is located is determined, and the grid pressure model corresponding to the grid is obtained based on the grid ID. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid. Based on the environmental semantic tags, the corresponding air pressure calibration parameters are weighted and adjusted. The weighted air pressure calibration parameters are combined with the air pressure reference baseline to obtain the calibrated grid air pressure reference value. The terminal sampled air pressure is subtracted from the grid air pressure reference value and combined with the altitude-air pressure change rate to perform vertical positioning of the terminal. The altitude positioning method also includes constructing the grid pressure model using the following method: Load a historical spatiotemporal database, which includes historical sampling data from the sampling device and the coordinate information of the sampling device at the time of sampling. The sampling device samples one or more of the following types of data: air pressure, temperature, and humidity. Based on the historical spatiotemporal database, an environmental semantic label is generated for at least one specified region. The environmental semantic label is used to characterize the probability that a specific environment exists within the specified region. The barometric fingerprint coverage area is rasterized. Based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, the historical sampling data is mapped to the corresponding raster, and the specified area with the semantic label is mapped to the rasterized barometric fingerprint coverage area. Based on the historical sampling data mapped to the grid, a grid pressure model is constructed for each grid. The grid pressure model includes the environmental semantic label corresponding to the grid and the pressure reference baseline calculated based on the historical sampling data mapped to the grid.

2. The height positioning method according to claim 1, characterized in that, Based on the historical spatiotemporal database, environmental semantic labels are generated for at least one specified region, including: Based on the historical spatiotemporal database, classifier processing or Bayesian posterior processing is performed to obtain the probability of the specified region with respect to any type of environmental semantic label, and the one with the highest probability is selected as the environmental semantic label of the specified region.

3. The height positioning method according to claim 2, characterized in that, The height localization method also includes a writing stability coefficient for the environmental semantic tags, including: Calculate the variance of the location change of the environmental semantic tag, perform an inverse function mapping on the variance to obtain the stability coefficient of the environmental semantic tag, and the stability coefficient is used to influence the semantic trajectory formed by the environmental semantic tag.

4. The height positioning method according to claim 1, characterized in that, The method further includes: Construct a mapping table between barometric calibration parameters and environmental semantic labels, where each environmental semantic label corresponds to at least one barometric calibration parameter. Based on the mapping table between the air pressure calibration parameters and environmental semantic tags, the weights of the environmental semantic tags on the corresponding air pressure calibration parameters are adjusted. The air pressure calibration parameters include: temperature correction coefficient, humidity correction coefficient, wind disturbance term, structural disturbance term, and sensor bias, wherein the wind disturbance term represents the influence of wind on air pressure, and the structural disturbance term represents the influence of the state of structure or facility on air pressure.

5. The height positioning method according to claim 4, characterized in that, The method also includes wind disturbance discrimination, including: If the air pressure fluctuation is detected, and the air pressure fluctuation amplitude is greater than or equal to the preset air pressure threshold within a preset time, and there is a high frequency pulsation, then it is determined that there is wind disturbance in the air pressure data sequence. When wind disturbance is detected in the air pressure data sequence, power spectrum or variance detection is performed on the air pressure data sequence sampled by the terminal. The equivalent amplitude is estimated by matching the reference wind speed and wind direction. The equivalent amplitude is then exponentially decayed and used as the wind disturbance term.

6. The height positioning method according to claim 1, characterized in that, The method further includes: Based on the horizontal positioning coordinates, the horizontal positioning standard deviation is calculated. When the horizontal positioning standard deviation is greater than the grid accuracy, the grid pressure model corresponding to the first grid where the terminal is located is obtained, as well as the grid pressure model corresponding to the second grid adjacent to the first grid. Based on the environmental semantic labels of the first grid and the second grid, semantic trajectory constraints are generated. When the terminal's sampled air pressure is subtracted from the grid air pressure reference value and combined with the altitude-pressure change rate to perform vertical positioning of the terminal, the vertical positioning result is calibrated based on the semantic trajectory constraint.

7. The height positioning method according to claim 1, characterized in that, The method further includes: The air pressure calibration value is obtained by subtracting the grid air pressure reference value from the terminal sampling air pressure. The residual is calculated by comparing multiple air pressure calibration values. The air pressure calibration value calculated by comparing the residual satisfies the following condition: the distance between the terminal sampling positions corresponding to any two air pressure calibration values ​​is within a preset range. Based on the environmental semantic label of the terminal sampling location, a residual threshold is set, and the residual is compared with the residual threshold to generate a quality index for each air pressure calibration value; Based on the quality indicators, the pressure calibration values ​​are weighted and fused to update the pressure reference baseline of the corresponding grid and recalculate the probability of environmental semantic labels in the grid.

8. A height positioning system, characterized in that, The altitude positioning system is applied to the terminal, and the altitude positioning system applies the altitude positioning method as described in any one of claims 1 to 7, wherein the altitude positioning system comprises: The sampling module is configured to sample at least air pressure data; The data processing module is configured to acquire the horizontal positioning of the terminal; The data processing module is also configured to determine the grid where the terminal is located based on the horizontal positioning result of the terminal, and obtain the grid pressure model corresponding to the grid based on the ID of the grid. The grid pressure model includes the environmental semantic label and pressure reference baseline corresponding to the grid. The data processing module adjusts the weights of the corresponding air pressure calibration parameters based on the environmental semantic tags, combines the weighted air pressure calibration parameters with the air pressure reference baseline to obtain the calibrated grid air pressure reference value, and combines the terminal sampled air pressure with the grid air pressure reference value and the altitude-air pressure change rate to perform vertical positioning of the terminal.

9. A height positioning system, characterized in that, The altitude positioning system is applied to a cloud server, and the altitude positioning system uses the altitude positioning method as described in any one of claims 1 to 7, wherein the altitude positioning system comprises: A historical spatiotemporal database, which includes historical sampling data collected by the sampling device and the coordinate information of the sampling device at the time of sampling. The sampling device samples one or more of the following types of data: air pressure, temperature, and humidity. The data processing module is configured to generate environmental semantic labels for at least one specified area based on the historical spatiotemporal database. The environmental semantic labels are used to characterize the probability of a specific environment existing in the specified area. The data processing module rasterizes the barometric fingerprint coverage area, maps the historical sampling data to the corresponding raster based on the coordinate information of the sampling device during sampling in the historical spatiotemporal database, and maps the specified area with the semantic labels to the rasterized barometric fingerprint coverage area. The grid pressure model generation module is configured to construct a grid pressure model for each grid based on the historical sampling data mapped to the grid. The grid pressure model includes an environmental semantic label corresponding to the grid and a pressure reference baseline calculated based on the historical sampling data mapped to the grid. The data transmission module is used to send the grid pressure model, which is used to: adjust the weights of the corresponding pressure calibration parameters through the environmental semantic tags, so that the terminal combines the weighted pressure calibration parameters with the pressure reference baseline to obtain the calibrated grid pressure reference value; and the terminal combines the sampled pressure minus the grid pressure reference value with the altitude-pressure change rate to perform vertical positioning of the terminal.

Images