Micropower geomagnetic parking space detection method and device

Abstract

The invention relates to the technical field of parking space detection and vehicle control, and discloses a micropower geomagnetic parking space detection method and device, and the method comprises the steps: fusing the vector intensity and direction angle according to pre-obtained empty parking space geomagnetic vector data and real-time geomagnetic vector data, constructing a geomagnetic reference map, and calculating a real-time geomagnetic rotation angle; comparing and analyzing the geomagnetic reference map and the real-time geomagnetic rotation angle, generating a high-frequency sampling signal, acquiring high-frequency geomagnetic vector data of high-frequency sampling, and constructing a dynamic feature set; the adjacent vehicle interference condition is analyzed, a dynamic parking space state monitoring mechanism is configured, and the real-time state of a parking space is obtained; in response to a real-time vehicle moving task, planning and optimizing a vehicle moving path in real time to obtain an optimized path, and controlling the vehicle to move; the parking space state is adjusted in real time in combination with the adjacent vehicle interference condition, the accuracy of the parking space state detection result is improved, the optimal vehicle moving path is planned, the vehicle locating time is shortened, and the vehicle moving efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of parking space detection and vehicle control technology, and more specifically to a low-power geomagnetic parking space detection method and device. Background Technology

[0002] Current parking space detection technologies mainly rely on ultrasonic waves, infrared sensors, cameras, or geomagnetic sensors. Among them, geomagnetic detection technology has attracted attention due to its advantages such as low power consumption and resistance to environmental interference. However, traditional geomagnetic methods are mostly based on scalar data and cannot fully capture the vector characteristics of the geomagnetic field, resulting in low accuracy of parking space detection results in complex scenarios.

[0003] The existing technology has the following problems: it relies on the intensity information of geomagnetic sensors, ignores changes in azimuth angle, and is easily affected by the movement of neighboring vehicles, temperature drift, or electromagnetic interference, resulting in low data accuracy; it uses a fixed sampling rate, which cannot be adaptively adjusted according to the vehicle's speed, and has a response delay when high-speed vehicles enter or exit, and the parking space detection process cannot reflect the real-time status of the parking space; it ignores the influence of changes in the status of neighboring vehicles on the parking space status and lacks neighboring vehicle interference separation, resulting in low accuracy of the parking space status judgment result; in order to solve at least one of the above problems, this application proposes a low-power geomagnetic parking space detection method and device. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide a low-power geomagnetic parking space detection method and device, which can effectively solve the problems in the background technology. The specific technical solution of this application is as follows:

[0005] A low-power geomagnetic parking space detection method includes:

[0006] Based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, the vector intensity and direction angle are fused to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle;

[0007] By comparing and analyzing the geomagnetic reference map and the real-time geomagnetic rotation angle, a high-frequency sampling signal is generated, high-frequency geomagnetic vector data is obtained from the high-frequency sampling, and the corresponding features are extracted to construct a dynamic feature set.

[0008] Based on the dynamic feature set, the interference from neighboring vehicles is analyzed, a dynamic parking space status monitoring mechanism is configured, and the parking space status is monitored in real time to obtain the real-time parking space status.

[0009] In response to real-time vehicle movement tasks, and combined with the real-time status of parking spaces, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement.

[0010] Specifically, the step of fusing vector intensity and direction angle based on pre-acquired empty parking space geomagnetic vector data and real-time geomagnetic vector data to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle includes:

[0011] The pre-acquired geomagnetic vector data of empty parking spaces is aligned with coordinates, and the data is compressed and encoded using a preset data encoding model to obtain the data encoding.

[0012] Based on the data encoding, the geomagnetic vector sequence is reconstructed by fusing vector intensity and orientation angle, and a geomagnetic reference map is constructed.

[0013] The real-time geomagnetic vector data is matched with the geomagnetic reference map to calculate the real-time geomagnetic rotation angle.

[0014] Specifically, based on the data encoding, the geomagnetic vector sequence is reconstructed by fusing vector intensity and orientation angle, and a geomagnetic reference map is constructed, including:

[0015] By using a pre-defined data reconstruction model, the data encoding is reconstructed to obtain a geomagnetic vector sequence;

[0016] Based on the geomagnetic vector sequence, vector features are extracted using a preset feature extraction model to construct a feature vector;

[0017] By combining the loss of eigenvectors with vector intensity and orientation angle analysis, loss compensation is performed on the geomagnetic vector sequence to construct a geomagnetic reference map.

[0018] Specifically, the process involves comparing and analyzing the geomagnetic reference map and the real-time geomagnetic rotation angle to generate a high-frequency sampling signal, acquiring high-frequency geomagnetic vector data from the high-frequency samples, extracting the corresponding features, and constructing a dynamic feature set, including:

[0019] The real-time geomagnetic rotation angle is matched and analyzed against the geomagnetic reference map to calculate the rate of change of the rotation angle.

[0020] Based on the rotation angle change rate, the rotation angle change trend is predicted by a preset rotation prediction model, and a high-frequency sampling signal is generated;

[0021] In response to the high-frequency sampling signal, high-frequency geomagnetic vector data of high frequency sampling is acquired;

[0022] The corresponding features are extracted from the high-frequency geomagnetic vector data to construct a dynamic feature set.

[0023] Specifically, in response to the high-frequency sampling signal, acquiring high-frequency sampled high-frequency geomagnetic vector data includes:

[0024] In response to high-frequency sampling signals, the energy distribution of real-time geomagnetic vector data is analyzed, and the highest frequency of the data is calculated.

[0025] Based on the highest frequency of the data, the initial sampling rate of the data is adjusted in real time through a preset dynamic data sampling model to obtain an optimized sampling rate;

[0026] A dynamic sampling mechanism is configured to use an initial sampling rate for data sampling in the portion where the rate of change of rotation angle is less than a preset rate of change threshold, and to use an optimized sampling rate for data sampling in the portion where the rate of change of rotation angle is greater than or equal to the preset rate of change threshold, thereby obtaining high-frequency geomagnetic vector data.

[0027] Specifically, based on the aforementioned dynamic feature set, the interference from neighboring vehicles is analyzed, a dynamic parking space status monitoring mechanism is configured, and the parking space status is monitored in real time to obtain the real-time parking space status, including:

[0028] Based on a dynamic feature set, each parking space is treated as a node, and connection edges are established by combining the location relationship and feature association relationship of parking spaces to construct a parking space association graph;

[0029] The interference from neighboring vehicles is analyzed in the parking space association diagram, and a dynamic parking space status monitoring mechanism is configured to monitor the parking space status in real time and obtain the real-time status of the parking space.

[0030] Specifically, the interference from neighboring vehicles is analyzed in the parking space association diagram, a dynamic parking space status monitoring mechanism is configured, and the real-time status of the parking spaces is monitored to obtain the real-time status of the parking spaces, including:

[0031] Analyze the interference from neighboring vehicles based on the parking space association diagram, calculate the interference degree of each node, and identify the interference propagation path;

[0032] Based on the aforementioned interference propagation path, the node signals are separated through a dynamic parking space status monitoring mechanism to calculate the first vacancy probability of each parking space.

[0033] Analyze the state interference between parking spaces, correct the first vacancy probability, and obtain the second vacancy probability;

[0034] Based on the second vacancy probability of each parking space, and combined with the correlation and collaborative analysis between parking spaces, the real-time status of the parking spaces is obtained.

[0035] Specifically, in response to real-time vehicle movement tasks, and in conjunction with the real-time status of parking spaces, the vehicle movement path is planned and optimized in real-time using a preset path optimization model to obtain an optimized path, thereby controlling vehicle movement, including:

[0036] In response to real-time vehicle movement tasks, and combined with the real-time status of parking spaces, features reflecting the spatial, temporal, and dynamic characteristics of parking spaces are extracted to construct a parking space status feature vector.

[0037] Based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time through a preset path optimization model to obtain an optimized path and control vehicle movement.

[0038] Specifically, based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement, including:

[0039] Based on the parking space status feature vector, a moving path is planned for each vehicle to be moved through a preset path optimization model, resulting in a first set of moving paths.

[0040] Filter and identify path conflicts in the first set of movement paths, and identify the conflict points;

[0041] Based on the conflict points, the first movement path is optimized to obtain an optimized path, and the vehicle movement is controlled.

[0042] A low-power geomagnetic parking space detection device, used to implement the aforementioned low-power geomagnetic parking space detection method, includes:

[0043] The geomagnetic data analysis module, based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, fuses vector intensity and direction angle to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle;

[0044] The feature extraction module compares and analyzes the geomagnetic reference map and the real-time geomagnetic rotation angle, generates a high-frequency sampling signal, acquires high-frequency geomagnetic vector data from the high-frequency sampling, extracts the corresponding features, and constructs a dynamic feature set.

[0045] The parking space status recognition module analyzes the interference from neighboring vehicles based on the dynamic feature set, configures a dynamic parking space status monitoring mechanism, monitors the parking space status in real time, and obtains the real-time status of the parking space.

[0046] The vehicle control module responds to real-time vehicle movement tasks, combines the real-time status of parking spaces, and plans and optimizes the vehicle movement path in real time using a preset path optimization model to obtain an optimized path and control vehicle movement.

[0047] The beneficial effects of this application are as follows: Based on the geomagnetic vector intensity and stationary direction angle, a geomagnetic reference map is constructed. The rate of change is calculated through real-time geomagnetic rotation angle to generate a high-frequency sampling signal. The data acquisition process is adaptively adjusted. A dynamic feature set is constructed based on the high-frequency sampling data. Neighboring vehicle interference is analyzed in conjunction with the parking space association map. Neighboring vehicle interference is removed during the parking space status identification process to identify the real-time status of the parking space. Based on the real-time status of the parking space, the vehicle movement path is planned, and the vehicle movement process is controlled in real time. By constructing a geomagnetic reference map and adaptive data acquisition, magnetic field offset can be avoided. The acquired data can reflect the real-time status of the parking space. The real-time adjustment of the parking space status in conjunction with the neighboring vehicle interference can reduce misjudgment and improve the accuracy of parking space status detection results. Thus, the optimal vehicle movement path is planned, the vehicle search time is reduced, and the vehicle movement efficiency is improved. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the process of a low-power geomagnetic parking space detection method according to an embodiment of this application.

[0049] Figure 2 This is a schematic diagram of the geomagnetic reference map in the embodiments of this application;

[0050] Figure 3 This is a schematic diagram of the parking space association diagram in the embodiments of this application;

[0051] Figure 4 This is a schematic diagram of the structure of a low-power geomagnetic parking space detection device in an embodiment of this application. Detailed Implementation

[0052] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. Identical components are indicated by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0053] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0054] Hereinafter, the terms "first," "second," and other generic terms are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0055] refer to Figure 1 The image shows a specific implementation of a low-power geomagnetic parking space detection method according to this application, including:

[0056] S101. Based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, fuse the vector intensity and direction angle to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle.

[0057] S102. Compare and analyze the geomagnetic reference map and the real-time geomagnetic rotation angle, generate high-frequency sampling signals, obtain high-frequency geomagnetic vector data from high-frequency sampling, extract the corresponding features, and construct a dynamic feature set.

[0058] S103. Based on the dynamic feature set, analyze the interference from neighboring vehicles, configure a dynamic parking space status monitoring mechanism, monitor the parking space status in real time, and obtain the real-time status of the parking space.

[0059] S104. In response to real-time vehicle movement tasks, combined with the real-time status of parking spaces, the vehicle movement path is planned and optimized in real time through a preset path optimization model to obtain an optimized path and control vehicle movement.

[0060] During vehicle parking, a significant amount of time is spent searching for available parking spaces, resulting in low parking efficiency and traffic congestion. This embodiment preprocesses the pre-acquired geomagnetic vector data of available parking spaces. A preset data encoding model is used to compress and encode the preprocessed geomagnetic vector data, constructing an encoding that efficiently represents the characteristics of the geomagnetic vector data. This encoding is then used for sparse encoding of the geomagnetic vector data of available parking spaces, reducing the data volume while preserving key geomagnetic vector features. Based on the data encoding, the intensity and orientation information of the geomagnetic vectors are fused, and the data encoding is reconstructed using a preset data reconstruction model. A preset feature extraction model is used to extract vector features from the reconstructed geomagnetic vector sequence, obtaining feature vectors that represent the core characteristics of the geomagnetic vector sequence. A geomagnetic reference map is constructed based on the feature vectors. Real-time geomagnetic vector data is matched against the geomagnetic reference map to calculate the real-time geomagnetic rotation angle.

[0061] It should be noted that by compressing the data through data encoding and retaining key features, the constructed geomagnetic reference map can accurately reflect the true geomagnetic characteristics of empty parking spaces, avoiding the distortion of the reference map caused by data deviation and feature loss. The real-time geomagnetic rotation angle obtained by matching real-time geomagnetic vector data with the geomagnetic reference map can accurately reflect the change in the real-time geomagnetic state relative to the geomagnetic state of empty parking spaces, providing accurate angle data for parking space status detection and improving the accuracy and reliability of parking space status detection results.

[0062] Specifically, the real-time geomagnetic rotation angle is compared and analyzed with the reference rotation angle at the corresponding position in the geomagnetic reference map. The rate of change of the real-time geomagnetic rotation angle over time is calculated. A preset rotation prediction model is used to predict the trend of the rotation angle over a future period. If the prediction result shows that the rotation angle will change significantly, it indicates that there is dynamic behavior such as vehicles entering or leaving the parking space. A high-frequency sampling signal is generated, and the initial sampling rate of the data is adjusted in real time to obtain an optimized sampling rate. The data sampling process is adjusted using the optimized sampling rate to obtain high-frequency geomagnetic vector data. Corresponding features are extracted from the high-frequency geomagnetic vector data to construct a dynamic feature set.

[0063] It is important to emphasize that analyzing the dynamic trend of geomagnetic changes through the rotation angle change rate and rotation prediction model to generate high-frequency sampling signals and acquire high-frequency sampling data can improve the accuracy of parking space status recognition results and avoid the resource waste caused by single high-frequency sampling. Adaptive optimization of the sampling rate can reduce unnecessary sampling operations and lower equipment energy consumption while ensuring the acquisition of high-frequency geomagnetic vector data that reflects key information on geomagnetic dynamic changes. By extracting multi-dimensional features to construct a dynamic feature set, dynamic change information reflecting geomagnetic state can be obtained, providing an accurate data foundation for parking space status recognition and detection, and improving the accuracy and comprehensiveness of parking space status detection results.

[0064] Simultaneously, using each parking space as a node, connection edges are established based on the geomagnetic dynamic features corresponding to each parking space extracted from the dynamic feature set, constructing a parking space association graph. The interference from neighboring vehicles is analyzed within the parking space association graph, and the interference degree of each node is calculated. Based on the connection relationships between nodes and the transmission of interference degree, the interference propagation path is identified. Node signals are separated through a dynamic parking space status monitoring mechanism, and the real-time status of each parking space is analyzed based on the separated target parking space geomagnetic signal. By constructing the parking space association graph, the location and feature relationships between parking spaces are reflected, providing data support for neighboring vehicle interference analysis. Calculating the interference degree and identifying the interference propagation path based on the association graph allows for the location of interference sources and transmission paths, providing data support for interference signal separation. Effectively separating the interference signal from the target parking space signal improves the accuracy of real-time parking space status monitoring, especially in scenarios with dense parking spaces and severe neighboring vehicle interference, reducing the problem of misjudgment of parking space status caused by neighboring vehicle interference.

[0065] Specifically, by combining the real-time status of parking spaces, features reflecting the spatial, temporal, and dynamic characteristics of parking spaces are extracted to construct parking space status feature vectors. A pre-defined path optimization model is then used to plan and optimize vehicle movement paths in real time, resulting in optimized paths. Based on these optimized paths, vehicle control commands are generated to control vehicles to travel along them, achieving precise control over vehicle movement. By extracting multi-dimensional parking space status features to construct feature vectors, comprehensive and accurate data support is provided for the path optimization model, making the planned movement paths more consistent with the actual conditions of the parking lot and the needs of vehicle movement. Controlling vehicle movement based on optimized paths guides vehicles to travel efficiently and orderly, shortening vehicle movement time and improving the overall traffic efficiency of the parking lot. Simultaneously, it reduces ineffective vehicle movement caused by unreasonable path planning, reduces vehicle search time, and improves overall vehicle traffic efficiency.

[0066] This application constructs a geomagnetic reference map based on geomagnetic vector intensity and stationary direction angle. It calculates the rate of change of the geomagnetic rotation angle in real time to generate a high-frequency sampling signal, adaptively adjusting the data acquisition process. A dynamic feature set is constructed based on the high-frequency sampling data. Interference from neighboring vehicles is analyzed using a parking space association map, and this interference is removed during parking space status identification to determine the real-time status of the parking space. Based on this real-time status, vehicle movement paths are planned, and the vehicle movement process is controlled in real time. By constructing a geomagnetic reference map and using adaptive data acquisition, magnetic field offset can be avoided, and the acquired data reflects the real-time status of the parking space. Real-time adjustment of the parking space status based on neighboring vehicle interference reduces misjudgments and improves the accuracy of parking space status detection results, thereby planning the optimal vehicle movement path, reducing vehicle search time, and improving vehicle movement efficiency.

[0067] Furthermore, based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, the vector intensity and direction angle are fused to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle, including:

[0068] S201. Align the coordinates of the pre-acquired empty parking space geomagnetic vector data, and compress and encode the data using a preset data encoding model to obtain the data encoding.

[0069] S202. Based on the data encoding, the vector intensity and direction angle are fused to reconstruct the geomagnetic vector sequence and construct a geomagnetic reference map;

[0070] S203. Match the real-time geomagnetic vector data with the geomagnetic reference map and calculate the real-time geomagnetic rotation angle.

[0071] In this embodiment, the pre-acquired empty parking space geomagnetic vector data suffers from inconsistent coordinate references. Based on the equipment deployment location coordinates and angle information recorded during data acquisition, a coordinate transformation algorithm maps the pre-acquired empty parking space geomagnetic vector data to a unified three-dimensional coordinate system, eliminating coordinate deviations between different acquired data. A pre-set data encoding model is used for compression encoding. This model includes, but is not limited to, a geomagnetic data-specific encoding model based on sparse coding. The geomagnetic data-specific encoding model is trained using a large amount of empty parking space geomagnetic vector sample data to obtain a pre-trained model. The empty parking space geomagnetic vector data is then input into this pre-trained model, which compresses and encodes the data to obtain the encoded data. Coordinate alignment eliminates the inconsistency in data coordinate references, avoiding misjudgments of geomagnetic features caused by coordinate deviations. By compressing and encoding the data, the core features of the geomagnetic vector data are preserved during compression, reducing data volume while avoiding the loss of key geomagnetic information during compression, thus improving the accuracy of geomagnetic vector sequence reconstruction.

[0072] like Figure 2 As shown, the data encoding only retains the core feature parameters of the geomagnetic vector. A complete geomagnetic vector sequence needs to be reconstructed. Based on the data encoding, the vector intensity and orientation angle of the geomagnetic vectors are fused to reconstruct a vector sequence that accurately reflects the geomagnetic distribution of empty parking spaces, and a geomagnetic reference map is constructed. By reconstructing the data encoding, the geomagnetic vector sequence can be accurately recovered, avoiding sequence distortion caused by data loss. Fusing the two key features of vector intensity and orientation angle provides a more comprehensive and accurate representation of the geomagnetic state, making the reconstructed geomagnetic vector sequence more consistent with the actual geomagnetic distribution of empty parking spaces. By constructing a geomagnetic reference map, a benchmark is provided for the real-time data analysis and matching process, improving the matching accuracy of the benchmark map and the accuracy and efficiency of the real-time geomagnetic vector data analysis process.

[0073] Specifically, real-time geomagnetic vector data is matched with a geomagnetic reference map, and the directional difference is calculated to obtain the real-time geomagnetic rotation angle. This allows for the quantification of the change in the real-time geomagnetic state relative to the reference state of an empty parking space. The geomagnetic distribution is relatively stable in an empty parking space. When a vehicle enters the space, the vehicle's metal components interfere with the geomagnetic field, causing a change in the direction of the real-time geomagnetic vector. By calculating the angular difference between the real-time vector and the reference vector, the degree of interference is quantified, thus determining whether the parking space is occupied. By matching and analyzing real-time geomagnetic vector data with the geomagnetic reference map, the calculated rotation angle provides quantitative data support for parking space status judgment, improving the accuracy of parking space detection results, enhancing the stability and reliability of rotation angle calculation, and adapting to scenarios where different areas within a parking space have varying degrees of geomagnetic interference.

[0074] Furthermore, based on the data encoding, the geomagnetic vector sequence is reconstructed by fusing vector intensity and orientation angle, and a geomagnetic reference map is constructed, including:

[0075] S301. The data encoding is reconstructed using a preset data reconstruction model to obtain a geomagnetic vector sequence;

[0076] S302. Based on the geomagnetic vector sequence, extract vector features using a preset feature extraction model to construct a feature vector;

[0077] S303. Analyze the loss of eigenvectors by combining vector intensity and orientation angle, perform loss compensation on the geomagnetic vector sequence, and construct a geomagnetic reference map.

[0078] In this embodiment, the data reconstruction model includes, but is not limited to, a geomagnetic data-specific reconstruction model based on the inverse process of sparse coding. The geomagnetic data-specific reconstruction model is trained using a large amount of encoded data to obtain a pre-trained model. The encoded data is then input into the pre-trained model, which reconstructs the data encoding and outputs a complete and accurate geomagnetic vector sequence. By reconstructing the encoded data, the spatial distribution information of the geomagnetic vectors can be completely recovered, enabling the reconstructed geomagnetic vector sequence to reflect the actual geomagnetic state of empty parking spaces, thus improving the reliability and accuracy of the geomagnetic reference map.

[0079] Specifically, the preset feature extraction model includes, but is not limited to, a convolutional neural network (CNN) model. This CNN model is trained using a large amount of historical geomagnetic vector sequence data to obtain a pre-trained model. The geomagnetic vector sequence is then input into the pre-trained CNN model, which extracts vector features that accurately reflect the core characteristics of the magnetic field, constructing feature vectors. This feature extraction model can quickly and accurately extract feature vectors reflecting the characteristics of the geomagnetic benchmark, providing accurate data support for the construction of geomagnetic benchmark maps.

[0080] Simultaneously, a standard feature library of empty parking spaces' geomagnetic data was established using a large number of interference-free empty parking space geomagnetic samples. The feature vectors were compared with this standard feature library, and the loss value for each feature dimension was calculated. For vector intensity features, the deviation rate between the actual and standard feature values ​​was calculated; for azimuth features, the absolute deviation between the actual and standard azimuth angles was calculated. Based on the importance of intensity and azimuth angle in the geomagnetic detection scenario, corresponding weights were assigned. Weighted compensation was used to correct the lost vector data, resulting in a compensated geomagnetic vector sequence. This compensated geomagnetic vector sequence was then mapped onto a two-dimensional space to construct a two-dimensional geomagnetic reference map. By compensating for the loss and deviation of the geomagnetic vector sequence, it is ensured that the reference map accurately reflects the spatial distribution pattern of the geomagnetic data in empty parking spaces. The constructed geomagnetic reference map reflects the status of empty parking spaces, providing a reference standard for real-time parking space status detection.

[0081] Furthermore, a comparative analysis is performed on the geomagnetic reference map and the real-time geomagnetic rotation angle to generate high-frequency sampling signals, acquire high-frequency geomagnetic vector data, extract corresponding features, and construct a dynamic feature set, including:

[0082] S401. Perform matching analysis on the real-time geomagnetic rotation angle in the geomagnetic reference map and calculate the rate of change of rotation angle;

[0083] S402. Based on the rotation angle change rate, predict the rotation angle change trend through a preset rotation prediction model and generate a high-frequency sampling signal;

[0084] S403. In response to the high-frequency sampling signal, acquire high-frequency geomagnetic vector data sampled at high frequency;

[0085] S404. Extract the corresponding features from the high-frequency geomagnetic vector data and construct a dynamic feature set.

[0086] In this embodiment, the geomagnetic reference map stores the reference geomagnetic rotation angles for each spatial location under the condition of an empty parking space. The real-time geomagnetic rotation angle is the actual magnetic field direction angle at each location at the current moment. By comparing the differences in the real-time geomagnetic rotation angles on the geomagnetic reference map, the degree of magnetic field disturbance can be reflected, thereby determining the parking space occupancy status. According to the spatial location, the real-time geomagnetic rotation angle is matched on the geomagnetic reference map to locate the corresponding parking space position, and the corresponding reference geomagnetic rotation angle is extracted. Using the current moment as the end point of the window, a time sliding window is set to select the previous N consecutive sampling moments. N can be set according to the device sampling cycle and response speed requirements. The difference between the real-time geomagnetic rotation angle and the corresponding reference rotation angle is calculated to obtain a rotation angle difference sequence. Based on the rotation angle difference sequence, a linear regression algorithm is used to fit the trend line, and the slope of the trend line is calculated to obtain the rotation angle change rate. By matching the real-time rotation angle with the reference rotation angle, the overall trend of rotation angle change is analyzed. The calculated rotation angle change rate can reflect the real geomagnetic dynamics, improving the accuracy of the parking space status detection results.

[0087] Specifically, the rotation angle change rate reflects the current dynamics. A pre-set rotation prediction model predicts the trend of rotation angle changes, allowing for early detection of when the magnetic field enters a period of rapid change and timely high-frequency sampling. The rotation prediction model includes, but is not limited to, an LSTM model. A pre-trained LSTM model is trained using a large amount of historical rotation angle change rate data. The rotation angle change rate is then input into the pre-trained LSTM model, which outputs a predicted change rate. A high-frequency sampling signal is generated based on the maximum increase in the predicted change rate. By predicting the trend of the rotation angle change rate, key data on rapid changes in the magnetic field can be collected, providing complete and accurate data support for feature extraction and parking space status detection.

[0088] In response to the high-frequency sampling signal, the energy distribution of real-time data is analyzed to determine the optimal sampling rate. Dynamic sampling is then performed in different regions by dynamically adjusting the sampling rate to acquire high-frequency geomagnetic vector data. This regional dynamic sampling limits high-frequency sampling to segments of rapidly changing magnetic fields, reducing the amount of data sampled, improving data acquisition efficiency, and providing accurate data support for parking space status detection.

[0089] Specifically, corresponding features are extracted from high-frequency geomagnetic vector data. These features include, but are not limited to, time-domain features, frequency-domain features, and correlation features. Time-domain features include, but are not limited to, the mean, variance, and maximum rate of change of vector intensity. Frequency-domain features include, but are not limited to, frequency components, the proportion of dominant frequency energy, and the frequency centroid. Correlation features include, but are not limited to, the Pearson correlation coefficient of vector intensity between different acquisition points and the synchronicity coefficient of azimuth angle changes. The extracted multidimensional features are combined to construct a dynamic feature set. By extracting time-domain, frequency-domain, and correlation features, multidimensional features reflecting different dynamic changes in the magnetic field can be obtained, avoiding the one-sidedness of analysis caused by single-dimensional features. This provides multidimensional feature data for the parking space status analysis process, thereby improving the accuracy of parking space status detection results.

[0090] Furthermore, in response to the high-frequency sampling signal, acquiring high-frequency sampled high-frequency geomagnetic vector data includes:

[0091] S501: Responding to high-frequency sampling signals, analyze the energy distribution of real-time geomagnetic vector data and calculate the highest frequency of the data;

[0092] S502. Based on the highest frequency of the data, the initial sampling rate of the data is adjusted in real time using a preset dynamic data sampling model to obtain an optimized sampling rate.

[0093] S503. Configure a dynamic sampling mechanism. When the rotation angle change rate is less than the preset change rate threshold, the initial sampling rate is used for data sampling. When the rotation angle change rate is greater than or equal to the preset change rate threshold, the optimized sampling rate is used for data sampling to obtain high-frequency geomagnetic vector data.

[0094] In this embodiment, in response to the high-frequency sampling signal, the currently acquired real-time geomagnetic vector data is frequency-domain transformed to obtain the frequency domain energy of each segment. The highest frequency of the data is calculated, and the maximum value among all effective frequency components is extracted to obtain the real-time highest frequency. Extracting the highest frequency through Fourier transform allows for accurate analysis of the frequency domain characteristics of the geomagnetic vector data. The calculated highest frequency reflects the upper frequency limit of effective magnetic field changes, providing data support for sampling rate adjustment and avoiding signal aliasing due to insufficient sampling rate or power waste caused by sampling rate redundancy.

[0095] Specifically, based on the highest frequency of the data, the initial sampling rate is adjusted in real time using a pre-set dynamic data sampling model to obtain an optimized sampling rate. This dynamic sampling model includes, but is not limited to, an adaptive adjustment model based on the Nyquist theorem. The adaptive adjustment model is trained using a large amount of highest frequency data to obtain a pre-trained model. The highest frequency of the data is then input into the pre-trained model, which calculates the optimal sampling rate that satisfies the Nyquist theorem while also considering the computing power and storage limitations of low-power devices. By calculating the optimized sampling rate, the basic principles of signal sampling are met, ensuring no aliasing signals and avoiding missed sampling due to sudden frequency increases. This ensures the optimized sampling rate matches the actual capabilities of the device, preventing overload due to excessively high sampling rates. The calculated optimized sampling rate satisfies the detection requirements of dynamic magnetic field changes while adapting to the hardware limitations of low-power devices, improving the accuracy and reliability of the acquired data.

[0096] Furthermore, a dynamic sampling mechanism is configured. By using regional dynamic sampling, the optimized sampling rate is applied to key areas where the magnetic field changes rapidly, while the initial sampling rate is maintained in areas where the magnetic field is flat. The dynamic changes of geomagnetic vector data are not uniform throughout the entire time period. Only areas with a high rate of change of rotation angle require high-frequency sampling, while areas with a low rate of change of rotation angle continue to use the initial sampling rate. Regional dynamic sampling can reduce the energy consumption of low-power devices while ensuring that critical data is not lost.

[0097] Based on a large amount of geomagnetic data from empty parking spaces with no vehicle activity and no interference from neighboring vehicles, the maximum value of the rotation angle change rate is calculated, and 1.5 times this value is set as the rotation angle change rate threshold. The rotation angle change rate is compared with the threshold. If the rotation angle change rate is less than the threshold, the current magnetic field is in a calm period, and data acquisition continues at the initial sampling rate. If the rotation angle change rate is greater than or equal to the threshold, the current magnetic field is in a rapid change period, and data acquisition is switched to an optimized sampling rate. The geomagnetic vector data collected at different sampling rates are stitched together in time-stamp order to obtain complete high-frequency geomagnetic vector data. By using regional dynamic sampling, high-frequency sampling can be applied only to key areas, reducing equipment energy consumption and minimizing invalid high-frequency sampling. The high-frequency geomagnetic vector data stitched together by time-stamp can retain complete key information about rapid magnetic field changes, providing complete and accurate data support for feature extraction and parking space status judgment.

[0098] Furthermore, based on the aforementioned dynamic feature set, the interference from neighboring vehicles is analyzed, a dynamic parking space status monitoring mechanism is configured, and the parking space status is monitored in real time to obtain the real-time parking space status, including:

[0099] S601. Based on a dynamic feature set, each parking space is treated as a node, and connection edges are established by combining the parking space location relationship and feature association relationship to construct a parking space association graph;

[0100] S602. Analyze the interference of neighboring vehicles in the parking space association diagram, configure a dynamic parking space status monitoring mechanism, monitor the parking space status in real time, and obtain the real-time status of the parking space.

[0101] like Figure 3 As shown, based on a dynamic feature set, each parking space is treated as an independent node, with node information including dynamic features and spatial coordinates. The spatial Euclidean distance between two parking space nodes is calculated, and a distance threshold is set according to the accuracy requirements of parking space status detection. Spatial associations are established between nodes whose spatial Euclidean distance is less than the distance threshold. The feature similarity between the dynamic feature sets of two parking spaces is calculated, and the linear correlation of key features is calculated using the Pearson correlation coefficient to obtain the association degree. An association threshold is set according to the feature similarity in an interference-free scenario, and feature associations are established between nodes whose association degree is less than the association threshold. When two nodes have both spatial and feature associations, a connection edge is established between the corresponding nodes, and the weight of the edge is set according to the product of feature similarity and the inverse of spatial distance to construct a parking space association graph.

[0102] It should be noted that by combining spatial association and feature association to establish connecting edges, misjudgments based on a single dimension can be avoided, thus improving the accuracy of the association. The constructed parking space association graph provides data support for interference association analysis, enabling rapid location of interference sources. Compared with the traditional method of analyzing individual parking spaces in isolation, it can quickly identify the interference effects between parking spaces, providing a data foundation for anti-interference processing and thereby improving the accuracy of parking space status detection results.

[0103] Specifically, based on the parking space association graph, neighboring vehicle interference is quantified, and a dynamic monitoring mechanism is configured to avoid misjudgments of parking space status caused by interference. The interference level of different parking spaces is analyzed, and the monitoring strategy is dynamically adjusted to monitor the parking space status in real time, thus obtaining the real-time parking space status. By analyzing the neighboring vehicle interference and adjusting the parking space status detection process in real time, the accuracy and reliability of the real-time parking space status detection results can be improved, ensuring accurate monitoring of parking space status even in complex neighboring vehicle interference scenarios, and enhancing the environmental adaptability of the parking space status detection process.

[0104] Furthermore, the interference from neighboring vehicles is analyzed in the parking space association diagram, and a dynamic parking space status monitoring mechanism is configured to monitor the parking space status in real time, obtaining the real-time parking space status, including:

[0105] S701. Analyze the interference situation of neighboring vehicles based on the parking space association diagram, calculate the interference degree of each node, and identify the interference propagation path;

[0106] S702. Combining the interference propagation path, the node signals are separated through the dynamic parking space status monitoring mechanism, and the first vacancy probability of each parking space is calculated.

[0107] S703. Analyze the state interference between parking spaces, correct the first idle probability, and obtain the second idle probability.

[0108] S704. Based on the second vacancy probability of each parking space, and combined with the correlation and collaborative analysis between parking spaces, the parking space status is obtained in real time.

[0109] In this embodiment, the interference from neighboring vehicles is analyzed based on the parking space association graph. The characteristic fluctuation amplitude of each node is extracted from the dynamic feature set, including but not limited to the variance of vector intensity and the maximum fluctuation range of the direction angle, reflecting the activity level of the corresponding parking space's magnetic field. For a target node, all its associated edges are traversed, and the interference contribution value of the connected node is calculated by weighting the characteristic fluctuation amplitude of the connected node with the corresponding edge weight. The interference contribution values ​​of all associated edges are summed to obtain the interference degree of the target node. A higher interference degree indicates stronger interference from neighboring vehicles. An interference source threshold is set based on the maximum characteristic fluctuation amplitude in an interference-free scenario. Nodes with interference contribution values ​​greater than the interference source threshold are searched from each node to obtain the interference propagation path. By calculating the interference degree, the degree of interference to each parking space is quantified, providing data support for anti-interference analysis of parking space status. By identifying the interference propagation path, the source and transmission path of interference can be quickly located, avoiding blindly checking all parking spaces and improving the efficiency of interference analysis.

[0110] Specifically, based on the dynamic characteristic signals of the target parking space and the dynamic characteristic signals in the interference propagation path, node signal separation is performed, decomposing the mixed signal into its own characteristic signal and interference characteristic signal, and removing the interference characteristic signal; based on the own characteristic signal, the first idle probability is calculated using a logistic regression model pre-trained with a large amount of characteristic signal data. By separating the interference signal, the influence of neighboring vehicle interference on the probability can be eliminated, improving the accuracy of the calculation of the first idle probability.

[0111] Furthermore, the influence of parking space status on each other is analyzed. A status influence coefficient matrix is ​​constructed based on the influence coefficients between parking spaces. Elements in the matrix represent the influence coefficients of the target parking space's first vacancy probability when adjacent parking spaces are vacant or occupied. The corresponding influence coefficients are calculated based on the deviations between the target parking space's first vacancy probability and the actual status when adjacent parking spaces are vacant or occupied, according to historical data. The first vacancy probability is corrected based on the influence coefficients of adjacent parking spaces with edges connected to the target node in the status influence coefficient matrix: the average influence coefficient of all adjacent parking spaces is calculated using a weighted average of the associated edge weights. The target parking space's first vacancy probability is then multiplied by this average influence coefficient to obtain the second vacancy probability. By correcting the first vacancy probability, the deviation caused by the status of adjacent parking spaces can be eliminated, improving the matching degree between the second vacancy probability and the actual parking space status. This provides accurate probabilistic support for the parking space status detection process and improves the accuracy of the parking space status detection results.

[0112] Specifically, based on the parking lot scenario requirements, vacancy thresholds and occupancy thresholds are set. If the vacancy threshold is greater than the occupancy threshold, and the second vacancy probability of the target parking space is greater than the vacancy threshold, the status is determined as candidate vacancy; if it is less than the occupancy threshold, the status is determined as candidate occupancy; if it is between the two thresholds, the status is determined as candidate uncertain. For a target parking space with a status of candidate vacancy, the status determination results of all adjacent parking spaces with associated edges are counted, and the proportion of adjacent parking spaces initially determined as candidate vacancy is calculated to obtain the adjacent vacancy proportion. If the adjacent vacancy proportion is greater than the vacancy threshold, the target parking space is confirmed to be vacant; if the adjacent vacancy proportion is less than or equal to the vacancy threshold, the characteristics of the target parking space are re-examined. If the characteristics are normal, the candidate vacancy is maintained; if the characteristics are abnormal, the candidate uncertain is corrected.

[0113] For a target parking space in the candidate occupied state, the status determination results of all adjacent parking spaces with associated edges are statistically analyzed. The proportion of adjacent parking spaces initially determined to be candidate occupied is calculated to obtain the adjacent occupied proportion. If the adjacent occupied proportion is less than the occupancy threshold, the target parking space is confirmed as occupied. If the adjacent vacant proportion is greater than or equal to the vacant threshold, the characteristics of the target parking space are re-examined. If the characteristics are normal, the candidate occupied status is maintained; if the characteristics are abnormal, the candidate status is corrected to uncertain. For target parking spaces with uncertain status, the characteristics of the target parking space are re-examined to determine the corresponding status. The real-time status of the parking space is obtained through analysis. The accuracy and reliability of parking space detection results are improved by analyzing the status trends of associated parking spaces. By combining collaborative analysis to judge the status of associated parking spaces, the status determination can be adapted to complex scenarios. Compared with independent determination of a single parking space, this can reduce the false judgment rate of real-time status, improve the accuracy of real-time parking space status judgment results, and improve the operational efficiency of the parking lot.

[0114] Furthermore, in response to real-time vehicle movement tasks, and combined with the real-time status of parking spaces, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement, including:

[0115] S801. In response to real-time vehicle movement tasks, combined with the real-time status of parking spaces, features reflecting the spatial characteristics, temporal characteristics, and dynamic characteristics of parking spaces are extracted to construct a parking space status feature vector.

[0116] S802. Based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time through a preset path optimization model to obtain an optimized path and control the vehicle movement.

[0117] In this embodiment, in response to real-time vehicle movement tasks, features reflecting the spatial, temporal, and dynamic characteristics of parking spaces are extracted based on the real-time status of the parking spaces. Spatial characteristics include, but are not limited to, spatial coordinates, spatial constraint parameters, and adjacency relationships. Temporal characteristics include, but are not limited to, continuous idle duration, historical occupancy frequency, and expected idle duration. Dynamic characteristics include, but are not limited to, real-time state confidence, interference residuals, and state change rate. The extracted features are arranged sequentially to construct a parking space state feature vector. By extracting multi-dimensional features, the decision-making bias caused by single-dimensional features can be avoided. The fusion of multi-dimensional features enables path planning to adapt to both static constraints and dynamic changes, providing a data foundation for path planning and optimization, and improving the accuracy and environmental adaptability of path planning results.

[0118] Specifically, based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time through a preset path optimization model to obtain an optimized path and control vehicle movement; the movement path is optimized in real time to improve the safety of the movement path; and the vehicle movement efficiency is improved. It is suitable for scenarios with high requirements for path dynamics and safety, such as smart parking lots and unmanned delivery.

[0119] Furthermore, based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement, including:

[0120] S901. Based on the parking space status feature vector, a moving path is planned for each vehicle to be moved through a preset path optimization model to obtain the first set of moving paths;

[0121] S902. Filter and identify path conflicts in the first set of movement paths, and identify the conflict points;

[0122] S903. Based on the conflict points, optimize the first moving path to obtain an optimized path and control the vehicle movement.

[0123] In this embodiment, based on the parking space status feature vector, a pre-defined path optimization model is used to plan a movement path for each vehicle to be moved, resulting in a first set of movement paths. The path optimization model includes, but is not limited to, a path scoring model based on the A* algorithm, which selects the target parking space with the highest overall score. Starting from the vehicle's current position and ending at the entrance to the target parking space, process nodes are continuously searched until the endpoint is reached, generating an initial movement path. This process is repeated for each vehicle to be moved, forming a first set of movement paths containing the initial paths of all vehicles. The path optimization model can quickly match a corresponding initial movement path for each vehicle, improving the matching degree between vehicles and movement paths and enhancing path planning efficiency.

[0124] Specifically, path conflicts are identified and conflict points are determined from the first set of movement paths. The spatiotemporal trajectory of each path is obtained from the first set of movement paths. The spatiotemporal trajectories of any two vehicles are compared pairwise. If the two trajectories contain the same location and their time intervals at that location overlap, that location is identified as a conflict point, resulting in a set of conflict points. By identifying conflict points, the path optimization process can directly optimize and adjust conflict points without retracing all paths, improving path optimization efficiency and path safety.

[0125] Furthermore, based on the conflict points, a pre-defined path optimization model is used to re-search for local paths, prioritizing alternative paths with the shortest detour distance that do not create new conflicts with other paths. This optimizes the first moving path to obtain the optimized path, controlling vehicle movement. Path optimization based on conflict points can improve path optimization efficiency, enhance the safety of vehicle movement paths, and thus improve vehicle traffic efficiency.

[0126] like Figure 4 As shown, a low-power geomagnetic parking space detection device is used to implement a low-power geomagnetic parking space detection method, comprising:

[0127] The geomagnetic data analysis module, based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, fuses vector intensity and direction angle to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle;

[0128] The feature extraction module compares and analyzes the geomagnetic reference map and the real-time geomagnetic rotation angle, generates a high-frequency sampling signal, acquires high-frequency geomagnetic vector data from the high-frequency sampling, extracts the corresponding features, and constructs a dynamic feature set.

[0129] The parking space status recognition module analyzes the interference from neighboring vehicles based on the dynamic feature set, configures a dynamic parking space status monitoring mechanism, monitors the parking space status in real time, and obtains the real-time status of the parking space.

[0130] The vehicle control module responds to real-time vehicle movement tasks, combines the real-time status of parking spaces, and plans and optimizes the vehicle movement path in real time using a preset path optimization model to obtain an optimized path and control vehicle movement.

[0131] In this embodiment, the geomagnetic data analysis module receives pre-acquired three-dimensional vector information of the magnetic field when there is no vehicle interference in an empty parking space, along with real-time collected geomagnetic vector data. Through data preprocessing and integrating multi-dimensional information such as vector intensity and direction angle, a geomagnetic reference map is constructed. By matching real-time data with the reference map, a real-time geomagnetic rotation angle reflecting the current magnetic field direction change is output, quantifying the degree of deflection of the current magnetic field relative to the empty parking space reference. The reference map constructed by fusing vector intensity and direction angle provides a more comprehensive reflection of the spatial distribution characteristics of the magnetic field in an empty parking space compared to a reference map relying solely on a single intensity feature, avoiding reference failure caused by environmental magnetic field drift. This provides a judgment benchmark for parking space detection, improving the accuracy of parking space detection.

[0132] The feature extraction module, based on the geomagnetic reference map and real-time geomagnetic rotation angle output by the geomagnetic data analysis module, calculates the rate of change of the rotation angle over time to determine whether the magnetic field has entered a period of rapid change, generating a high-frequency sampling signal. Based on this signal, the sampling rate is adjusted to acquire high-frequency geomagnetic vector data that can fully capture the dynamic changes of the magnetic field. Then, through multi-dimensional feature extraction, the high-dimensional raw data is transformed into a structured dynamic feature set. Triggering high-frequency sampling by the rate of change of the rotation angle avoids the power waste caused by fixed high-frequency sampling, ensuring that key data on rapid changes in the magnetic field can be collected. Constructing a dynamic feature set can eliminate environmental noise and retain features related to parking space status, providing accurate feature data for adjacent vehicle interference analysis and parking space status identification, thus improving the accuracy of parking space status identification results.

[0133] The parking space status recognition module receives the dynamic feature set output by the feature extraction module, treats each parking space as a node, and constructs a parking space association graph by combining the spatial adjacency relationship and feature association relationship of parking spaces to identify interference from neighboring vehicles. Based on the interference analysis results, a dynamic parking space status monitoring mechanism is configured to output accurate real-time parking space status by separating its own magnetic field characteristics from the interference characteristics of neighboring vehicles. Through the parking space association graph and dynamic monitoring mechanism, the interference source and interference propagation path can be identified and located, and the interference characteristics can be separated from the self-characteristics, improving the accuracy of the parking space status judgment results. Through the dynamic monitoring mechanism, it adapts to the interference scenarios of different parking spaces in the parking lot, ensuring that the device can accurately identify the parking space status even in complex scenarios with dense parking spaces and frequent vehicle flow, improving the environmental adaptability of the parking space recognition process.

[0134] The vehicle control module receives real-time vehicle movement tasks and, combined with the real-time parking space status output by the parking space status recognition module, completes real-time planning and optimization of the vehicle movement path through a preset path optimization model. It then outputs the optimized path command and converts it into vehicle control signals via the vehicle-to-everything (V2X) interface, achieving precise control of vehicle movement. By planning and optimizing vehicle movement paths in real time, inefficient vehicle movement can be avoided, improving vehicle movement efficiency.

[0135] The above description is merely a preferred embodiment of this application. The scope of protection of this application is not limited to the above embodiments. All technical solutions falling within the scope of this application's concept are within the scope of protection of this application. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this application should also be considered within the scope of protection of this application.

Claims

1. A low-power geomagnetic parking space detection method, characterized in that, include: Based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, the vector intensity and direction angle are fused to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle; By comparing and analyzing the geomagnetic reference map and the real-time geomagnetic rotation angle, a high-frequency sampling signal is generated, high-frequency geomagnetic vector data is obtained from the high-frequency sampling, and the corresponding features are extracted to construct a dynamic feature set. Based on the dynamic feature set, the interference from neighboring vehicles is analyzed, a dynamic parking space status monitoring mechanism is configured, and the parking space status is monitored in real time to obtain the real-time parking space status. In response to real-time vehicle movement tasks, and combined with the real-time status of parking spaces, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement.

2. The low-power geomagnetic parking space detection method according to claim 1, characterized in that, The process of constructing a geomagnetic reference map and calculating the real-time geomagnetic rotation angle by fusing vector intensity and direction angle based on pre-acquired empty parking space geomagnetic vector data and real-time geomagnetic vector data includes: The pre-acquired geomagnetic vector data of empty parking spaces is aligned with coordinates, and the data is compressed and encoded using a preset data encoding model to obtain the data encoding. Based on the data encoding, the geomagnetic vector sequence is reconstructed by fusing vector intensity and orientation angle, and a geomagnetic reference map is constructed. The real-time geomagnetic vector data is matched with the geomagnetic reference map to calculate the real-time geomagnetic rotation angle.

3. The low-power geomagnetic parking space detection method according to claim 2, characterized in that, Based on the data encoding, the geomagnetic vector sequence is reconstructed by fusing vector intensity and orientation angle, and a geomagnetic reference map is constructed, including: By using a pre-defined data reconstruction model, the data encoding is reconstructed to obtain a geomagnetic vector sequence; Based on the geomagnetic vector sequence, vector features are extracted using a preset feature extraction model to construct a feature vector; By combining the loss of eigenvectors with vector intensity and orientation angle analysis, loss compensation is performed on the geomagnetic vector sequence to construct a geomagnetic reference map.

4. The low-power geomagnetic parking space detection method according to claim 1, characterized in that, The process involves comparing and analyzing the geomagnetic reference map and the real-time geomagnetic rotation angle to generate a high-frequency sampling signal, acquiring high-frequency geomagnetic vector data, extracting corresponding features, and constructing a dynamic feature set, including: The real-time geomagnetic rotation angle is matched and analyzed against the geomagnetic reference map to calculate the rate of change of the rotation angle. Based on the rotation angle change rate, the rotation angle change trend is predicted by a preset rotation prediction model, and a high-frequency sampling signal is generated; In response to the high-frequency sampling signal, high-frequency geomagnetic vector data of high frequency sampling is acquired; The corresponding features are extracted from the high-frequency geomagnetic vector data to construct a dynamic feature set.

5. The low-power geomagnetic parking space detection method according to claim 4, characterized in that, In response to the high-frequency sampling signal, high-frequency geomagnetic vector data of the high-frequency sampling is acquired, including: In response to high-frequency sampling signals, the energy distribution of real-time geomagnetic vector data is analyzed, and the highest frequency of the data is calculated. Based on the highest frequency of the data, the initial sampling rate of the data is adjusted in real time through a preset dynamic data sampling model to obtain an optimized sampling rate; A dynamic sampling mechanism is configured to use an initial sampling rate for data sampling in the portion where the rate of change of rotation angle is less than a preset rate of change threshold, and to use an optimized sampling rate for data sampling in the portion where the rate of change of rotation angle is greater than or equal to the preset rate of change threshold, thereby obtaining high-frequency geomagnetic vector data.

6. The low-power geomagnetic parking space detection method according to claim 1, characterized in that, Based on the aforementioned dynamic feature set, the interference from neighboring vehicles is analyzed, a dynamic parking space status monitoring mechanism is configured, and the parking space status is monitored in real time to obtain the real-time parking space status, including: Based on a dynamic feature set, each parking space is treated as a node, and connection edges are established by combining the location relationship and feature association relationship of parking spaces to construct a parking space association graph; The interference from neighboring vehicles is analyzed in the parking space association diagram, and a dynamic parking space status monitoring mechanism is configured to monitor the parking space status in real time and obtain the real-time status of the parking space.

7. The low-power geomagnetic parking space detection method according to claim 6, characterized in that, The interference from neighboring vehicles is analyzed in the parking space association diagram, and a dynamic parking space status monitoring mechanism is configured to monitor the parking space status in real time, obtaining the real-time parking space status, including: Analyze the interference from neighboring vehicles based on the parking space association diagram, calculate the interference degree of each node, and identify the interference propagation path; Based on the aforementioned interference propagation path, the node signals are separated through a dynamic parking space status monitoring mechanism to calculate the first vacancy probability of each parking space. Analyze the state interference between parking spaces, correct the first vacancy probability, and obtain the second vacancy probability; Based on the second vacancy probability of each parking space, and combined with the correlation and collaborative analysis between parking spaces, the real-time status of the parking spaces is obtained.

8. The low-power geomagnetic parking space detection method according to claim 1, characterized in that, The response to real-time vehicle movement tasks, combined with the real-time status of parking spaces, utilizes a preset path optimization model to plan and optimize vehicle movement paths in real time, obtaining optimized paths and controlling vehicle movement, including: In response to real-time vehicle movement tasks, and combined with the real-time status of parking spaces, features reflecting the spatial, temporal, and dynamic characteristics of parking spaces are extracted to construct a parking space status feature vector. Based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time through a preset path optimization model to obtain an optimized path and control vehicle movement.

9. The low-power geomagnetic parking space detection method according to claim 8, characterized in that, Based on the parking space status feature vector, the vehicle movement path is planned and optimized in real time using a preset path optimization model to obtain an optimized path and control vehicle movement, including: Based on the parking space status feature vector, a moving path is planned for each vehicle to be moved through a preset path optimization model, resulting in a first set of moving paths. Filter and identify path conflicts in the first set of movement paths, and identify the conflict points; Based on the conflict points, the first movement path is optimized to obtain an optimized path, and the vehicle movement is controlled.

10. A low-power geomagnetic parking space detection device, characterized in that, A method for implementing a low-power geomagnetic parking space detection method as described in any one of claims 1 to 9 includes: The geomagnetic data analysis module, based on the pre-acquired geomagnetic vector data of empty parking spaces and real-time geomagnetic vector data, fuses vector intensity and direction angle to construct a geomagnetic reference map and calculate the real-time geomagnetic rotation angle; The feature extraction module compares and analyzes the geomagnetic reference map and the real-time geomagnetic rotation angle, generates a high-frequency sampling signal, acquires high-frequency geomagnetic vector data from the high-frequency sampling, extracts the corresponding features, and constructs a dynamic feature set. The parking space status recognition module analyzes the interference from neighboring vehicles based on the dynamic feature set, configures a dynamic parking space status monitoring mechanism, monitors the parking space status in real time, and obtains the real-time status of the parking space. The vehicle control module responds to real-time vehicle movement tasks, combines the real-time status of parking spaces, and plans and optimizes the vehicle movement path in real time using a preset path optimization model to obtain an optimized path and control vehicle movement.

Images