Unmanned parking control system and method

Abstract

The invention relates to the technical field of parking management, in particular to an unmanned parking control system and method, and provides the following scheme: the method comprises the following steps: constructing a multi-granularity fuzzy rule base, correcting a platform position error in combination with a vehicle parking error and a gravitational potential energy deviation, and forming a mechanical error; and dynamic updating is carried out based on the error distribution map, and the next transposition target position is intelligently planned. And an adaptive deformation reasoning weight network structure and a path intervention matrix are further introduced, so that the error correction precision and robustness are improved. The method effectively solves the problem of landing deviation in the platform transposition process, and is suitable for multiple types of parking scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of parking management, in particular to an unmanned parking control system and method. BACKGROUND

[0002] In the existing double-deck stereo parking system, the vehicle transposition process usually relies on a linkage device to vertically or horizontally move the vehicle between the upper layer parking space and the lower layer parking space to optimize the use of space. However, due to the limited positioning accuracy of the platform control, there is often a position error in the transposition and landing process of the vehicle, which will form mechanical fatigue or platform eccentric wear in the fixed docking area after long-term use, affecting the structural stability and operational safety.

[0003] In addition, the existing technology uses static preset control instructions for platform landing, lacks feedback mechanisms for actual parking state, vehicle center of gravity changes and historical error trends, making it difficult to dynamically correct each landing operation, resulting in insufficient transposition accuracy, frequent vehicle position deviation, and even causing misplacement, jamming or repeated calibration in severe cases, reducing system operation efficiency and user experience.

[0004] The above problems are particularly prominent in high-frequency parking scenarios with multiple transposition operations. To solve the above problems, the present application designs an unmanned parking control system and method. SUMMARY

[0005] The technical problem to be solved by the present application is to provide an unmanned parking control system and method to solve the problems of the prior art. The method corrects the platform position error by building a multi-granularity fuzzy rule base, combining vehicle parking error and gravitational potential deviation, and forming mechanical error. And based on the error distribution map, dynamically update the intelligent planning of the next transposition target position. Further introduce the reasoning weight network structure with adaptive deformation and the path intervention matrix to improve the error correction accuracy and robustness. This method effectively solves the landing deviation problem in the platform transposition process and is suitable for multiple types of parking scenarios.

[0006] To achieve the above purpose, the present application provides the following technical scheme:

[0007] An unmanned parking control method applied to a double-deck stereo parking system, the double-deck stereo parking system comprising an upper layer parking space and a lower layer parking space, when the lower layer parking space is in an occupied state, the idle parking spaces in the upper layer parking space are replaced one by one with the lower layer parking space through a linkage device, the method comprising:

[0008] When the parking space transposition operation is completed, the position error value of the platform controlled by the linkage device when reaching the transposition target position is calculated;

[0009] The position error value is corrected in combination with the recognized vehicle parking error and the gravity potential energy deviation, to obtain a mechanical error, wherein the correction comprises inputting the vehicle parking error and the gravity potential energy deviation as input variables of fuzzy reasoning into a fuzzy rule base comprising a coarse-grained rule subset, a medium-grained rule subset and a fine-grained rule subset for multi-scale error correction reasoning;

[0010] The preset error distribution map is updated according to the mechanical error, wherein the error distribution map is used to calculate a target position for the next position change, and a position change control instruction is generated according to the target position.

[0011] The position error value of the platform controlled by the linkage device when reaching the target position for the position change comprises:

[0012] An actual landing position after the parking space position change operation is completed, and a current physical position of the platform is generated;

[0013] A target position for the position change in the position change control instruction is generated, to generate a position change reference value;

[0014] State data of the platform during the position change is obtained, wherein the state data comprises platform acceleration, speed curve and transmission feedback information;

[0015] The position change reference value is compensated by a preset error estimation model based on the state data, to obtain a state-compensated theoretical landing position, wherein the error estimation model is constructed based on historical position change state data;

[0016] A difference between the current physical position of the platform and the state-compensated theoretical landing position is calculated, to obtain a position error value.

[0017] The position error value is corrected in combination with the recognized vehicle parking error and the gravity potential energy deviation, comprising:

[0018] The vehicle parking error and the gravity potential energy deviation are inputted as input variables of fuzzy reasoning into a fuzzy rule base according to a preset fuzzy rule base, and the position error value is corrected by the fuzzy rule base, wherein the fuzzy rule base comprises a first rule subset for coarse-grained error reasoning, a second rule subset for medium-grained error reasoning and a third rule subset for fine-grained error reasoning.

[0019] The fuzzy rule base further comprises an adaptive deformation reasoning weight network structure and a path intervention matrix, and the position error value is corrected by the fuzzy rule base, comprising:

[0020] inputting the input variables into the first rule subset, the second rule subset and the third rule subset respectively to obtain a plurality of fuzzy correction output sets corresponding to each granularity rule subset;

[0021] According to the input variables, the weight tensor nodes corresponding to each granularity rule subset are called in the inference weight network structure to calculate the activated weight coefficients, and the fuzzy correction output sets are weighted and calculated to obtain an initial inference result;

[0022] Obtain vehicle state data in the parking space transposition operation, determine the intervention variables associated with the input variables in the path intervention matrix, and select the corresponding intervention path according to the state value of the intervention variables;

[0023] According to the intervention rule of the intervention path, the initial inference result is de-fuzzified to obtain a fuzzy correction amount, and the fuzzy correction amount is combined with the position error value to obtain a mechanical error.

[0024] The inference weight network structure includes a tensor structure, the dimensions of the tensor structure correspond to the fuzzy levels of input variables and rule activation intensity, and the inference weight network structure is deformed by the current error distribution map, wherein:

[0025] According to the distribution trend of the mechanical error recorded in the historical transposition operation, the error high-incidence area and the error stable area in the error distribution map are determined;

[0026] For the error high-incidence area, the corresponding weight nodes are located in the tensor structure, and the corresponding weight nodes and adjacent nodes are stretched according to the entropy value of the error distribution map;

[0027] For the error stable area, the corresponding weight nodes are located in the tensor structure, and the corresponding weight nodes and adjacent nodes are compressed according to the entropy value of the error distribution map;

[0028] Wherein, the deformation is operated by adjusting the weight coefficient value, slope gradient and activation domain width of the weight node.

[0029] The path intervention matrix is used to adjust the cross-path regulation relationship between input variables in the fuzzy inference process, wherein the intervention path at least includes one of the following:

[0030] When there is a geometric mismatch state between the vehicle wheelbase and the platform limiting structure, the path intervention matrix controls the vehicle wheelbase variable to intervene in the inference path of the vehicle parking error, adjusts the activation probability, response amplitude and output priority of the corresponding rule;

[0031] When the gravity potential energy deviation during the transposition operation is greater than the preset potential energy threshold, the path intervention matrix controls the gravity potential energy deviation variable to intervene in the reasoning path of the gravity potential energy deviation, and adjusts the activation range of the corresponding rule, wherein the activation range adjustment includes compression or widening.

[0032] When an asymmetric load distribution state is detected during the transposition operation, the path intervention matrix controls the load distribution variable to simultaneously intervene in the reasoning path of the vehicle parking error and the gravity potential energy deviation, calculates the path intersection linkage rate, and activates the standby fuzzy rule.

[0033] The identification step of the vehicle parking error includes:

[0034] Obtain vehicle image data and platform state information before the start of the parking space transposition operation, wherein the vehicle image data includes an image of the parking position of the vehicle on the platform, and the platform state information includes a limiting structure, load sensing data, and platform attitude information;

[0035] According to the vehicle image data, the relative position between the vehicle tire and the limiting structure is detected by an image recognition algorithm, and the offset of the vehicle on the platform is calculated, wherein the offset includes a lateral offset and a longitudinal offset;

[0036] According to the load sensing data, the center of gravity position of the vehicle on the platform is determined and compared with the preset center point of the platform structure to obtain a center of gravity offset value;

[0037] According to the platform attitude information, the local torque generated by the vehicle on the platform is calculated;

[0038] The vehicle parking error is calculated in combination with the offset, the center of gravity offset value, and the local torque.

[0039] The preset error distribution map is updated according to the mechanical error, including:

[0040] Obtain the target parking space involved in the current transposition operation to obtain the error distribution map corresponding to the target parking space;

[0041] In the corresponding error distribution map, record the corresponding platform landing coordinates and the mechanical error of the current transposition operation to obtain a new error sample;

[0042] Fuse the new error sample with the historical error data in the error distribution map to update the error mean value, error fluctuation amplitude, and statistical density index corresponding to the target parking space;

[0043] The distribution density of the new error sample in the error distribution map is calculated according to the error mean, error fluctuation amplitude and statistical density index, and the error distribution map is regionally reconstructed according to the distribution density, wherein the regional reconstruction comprises adjusting the aggregation domain range and granularity density of error points.

[0044] The target position of the next transposition is calculated, comprising:

[0045] In the error distribution map corresponding to the target parking space, the position region is determined according to the aggregation domain range and granularity density;

[0046] The target position of the transposition is determined by a particle swarm algorithm according to the position region.

[0047] An unmanned parking control system, comprising:

[0048] A platform control module is configured to control a linkage device in a double-deck stereoscopic parking system to perform a parking space transposition operation, and to collect actual landing position data, target transposition position data and state data in the transposition process after the transposition operation is completed, and to calculate a position error value when the platform reaches the target transposition position;

[0049] An error correction module is configured to correct the position error value based on the identified vehicle parking error and gravitational potential energy deviation by fuzzy reasoning, to obtain a mechanical error, wherein the error correction module comprises a fuzzy rule base, a reasoning weight network structure and a path intervention matrix;

[0050] A map updating and control module is configured to update an error distribution map of a corresponding target parking space according to the mechanical error, and to calculate a target transposition position of the next transposition based on the error distribution map, and to generate a transposition control instruction for the next transposition operation.

[0051] Compared with the prior art, the application has the following beneficial effects:

[0052] The application constructs a closed-loop control mechanism based on error feedback, integrates vehicle parking error, gravitational potential energy deviation and platform state data, dynamically corrects platform landing error, and introduces an error distribution map to accumulate historical error data and learn distribution, thereby realizing intelligent optimization of the target transposition position. Further, through the synergistic effect of fuzzy reasoning logic, an adaptive deformation reasoning weight network structure and a path intervention matrix, the correction strategy in the transposition process can be adaptively adjusted according to the actual state, effectively avoiding mechanical fatigue accumulation points and reducing the risk of error repetition and superposition. BRIEF DESCRIPTION OF DRAWINGS

[0053] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments, made with reference to the accompanying drawings.

[0054] Figure 1 A schematic diagram of a parking space transposition principle of an embodiment of the application;

[0055] Figure 2 A schematic diagram of an error principle of an embodiment of the application;

[0056] Figure 3 A schematic diagram of a flow of a parking control method without people of an embodiment of the application;

[0057] Figure 4 A schematic diagram of a flow of a position error value correction of an embodiment of the application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the application will be apparently and completely described in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application.

[0059] In this document, the term “embodiment” means that a specific feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily independent or alternative embodiments to each other. It can be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0060] The application is applicable to an automatic three-dimensional parking device with a multi-platform cooperative transposition structure, a precise positioning requirement, and an error accumulation sensitive characteristic, and is particularly applicable to a parking system in a double-layer structure that needs to be transposed up and down on a platform to improve parking efficiency. Such a parking system is widely deployed in urban high-density areas, underground space utilization sites, and intelligent traffic facilities, and the repeated lifting and precise positioning process of the platform in the operation of the parking system puts extremely high requirements on the mechanical control precision.

[0061] Application scenarios include but are not limited to:

[0062] The upper parking space is exchanged with the lower parking space through a linkage mechanism, and the system needs to control the platform to micron-level precision in each transposition process;

[0063] The platform support structure is rigid linkage, and is affected by temperature drift, inertial loading, and long-period mechanical fatigue, and its landing error presents a slow accumulation and spatial concentration effect;

[0064] Due to the limited execution precision of the platform control system, there is a deviation between the actual position after the execution of each control instruction and the theoretical target, forming an error hot spot or a mechanical dead zone, which affects the stable operation of the system;

[0065] Multiple parking spaces reuse the same platform or path area, resulting in single-point impact accumulation, and the system needs to use a decentralized strategy to offset long-term concentrated loads.

[0066] The method provided in the application does not depend on specific mechanical arm types, parking assembly structures or linkage track forms, and does not require fixed vehicle size, mass or platform material, but is a general error dispersion and prediction control logic for an automated transposition platform with common operation bottlenecks.

[0067] The applicable scenarios have at least one of the following common characteristics:

[0068] The control target has long-period repeated motion, and the landing point is fixed or highly concentrated;

[0069] The control command cannot completely cover the inertial error, load disturbance or asymmetric response of the transmission chain;

[0070] The error itself has statistical, aggregation and path dependence, and cannot be eliminated by single correction;

[0071] The system needs to maintain millimeter-level positioning accuracy, but the platform operation is limited by mechanical aging or environmental disturbance.

[0072] Taking a double-layer three-dimensional parking system as an example, reference can be made to Figure 1 for understanding, Figure 1 which is a schematic diagram of the parking transposition principle of the embodiment of the application. The double-layer three-dimensional parking system is mainly applicable to the closed or semi-closed parking space scene of the upper and lower double-layer structure, and is especially suitable for dynamic allocation of limited parking resources in urban high-density three-dimensional parking garages.

[0073] Figure 1 The double-layer three-dimensional parking system is shown, which includes an upper parking space, a lower parking space and a transmission structure arranged between the two.

[0074] Figure 1 It is further shown that the upper parking space is arranged as a movable platform structure in an empty state, and the lower parking space is a movable platform structure for carrying a vehicle. The transmission structure is used to drive the upper parking space to move downward when it is detected that the lower parking space is in an occupied state, and to perform transposition operation with the lower parking space.

[0075] It can be understood that, Figure 1 which is only an example case. In specific actual applications, the upper parking space and the lower parking space can each include multiple movable platform structures. The transmission structure is not only used to transposition the upper parking space in the case of full occupancy of the lower parking space, but also includes transposition of the platform corresponding to the upper parking space with the platform in the lower parking space in an occupied state when a user needs to use the vehicle in the upper parking space.

[0076] Figure 1Further shown is that in the transposition process, the upper layer parking space is driven by the transmission structure to move downward in the vertical direction and complete the docking operation at the preset control position. The docking process requires that the platform edge and the parking space contact surface achieve mechanical limiting engagement to ensure the stability of the platform after transposition.

[0077] It can be understood that, Figure 1 The transmission structure in the above is only a schematic device for illustrating the linkage action in the parking space transposition process, and the application does not limit the specific implementation of the transmission structure in the implementation process. The linkage structure can adopt the following forms, but is not limited to them:

[0078] For example, vertical lifting transmission can be performed by a chain motor or a screw motor electric cylinder arranged on a garage upright column, and a hydraulic cylinder, a pneumatic cylinder or an electric sliding rail device can also be used as a driving execution component. In the transmission path design, various ways such as a scissors lifting structure, a guide rail lifting structure or a rotary arm lifting structure can be used to realize displacement control of the upper layer parking space according to the space layout.

[0079] In some optional embodiments, in a compact parking space arrangement, precise lifting and limiting docking in the transposition process can also be realized by a crank slider mechanism, a synchronous pulley system or a multi-point servo linkage structure. The application focuses on error identification, correction and control strategy optimization in the transposition control process. The specific mechanical implementation of the transmission structure as a basic execution mechanism can be flexibly selected according to the use scene, space condition and equipment configuration, which is a common means that can be understood and replaced by those skilled in the art, and will not be described here.

[0080] Taking docking error as an example, reference can be made to Figure 2 for understanding, Figure 2 which is an error principle schematic diagram of the embodiments of the application, used to illustrate the inconsistency of the docking position caused by structural error in the platform transposition process.

[0081] Figure 2 It is shown that in a double-layer three-dimensional parking system, two platform structures are provided for realizing vertical movement in the parking space transposition process.

[0082] Figure 2 Further shown is that in the transposition operation process, the transmission structure will perform precise docking to the target parking space under the system control instruction. The docking position refers to the design alignment point between the platform structure and the target parking space bottom limiting structure in the three-dimensional space, which usually includes horizontal positioning, vertical support and structural limiting and other mechanical cooperation elements. The docking relationship depends on the transmission accuracy and structural matching of the platform, which is usually realized in the form of mechanical guide rail, slot, conical limiting block or edge positioning block.

[0083] It can be understood that, limited to the response accuracy of the transmission system, the dynamic change of the platform load mass distribution, and the slight inertial deviation that the platform may generate during movement, when the platform finally reaches the transposition target position, there may be a slight error in the actual landing point. This error is the docking error, which is manifested as the relative misalignment between the platform and the parking space on the docking surface. For example, the platform fails to align with the slot or limiting structure of the parking space, resulting in edge overlap, skew or loose engagement, etc.

[0084] As can be appreciated by those skilled in the art, the docking error has long-term cumulative and position concentration, and if each transposition operation is performed with a fixed path and a fixed target point, the error will be repeatedly applied to the same structural node, which can easily cause local wear, structural deformation or loose contact, and reduce the long-term stability of the system.

[0085] Next, a kind of unmanned parking control method provided by the embodiment of the application will be introduced in combination with the drawings, Figure 3 The method shown is applied to a double-layer stereoscopic parking system, wherein the double-layer structure includes upper parking spaces and lower parking spaces. The upper parking spaces can realize position switching through a linkage platform structure. When it is detected that the lower parking spaces are all occupied, a control logic starts a transposition process to schedule idle spaces in the upper parking spaces to replace the lower parking spaces one by one through a linkage device. The transposition operation is completed through a lifting combined with horizontal movement, and the linkage platform serves as an execution mechanism, which needs to be accurately controlled in position during the whole process to avoid structural interference or vehicle offset caused by error accumulation. The method includes the following steps:

[0086] S1: calculating a position error value of the platform controlled by the linkage device when reaching a transposition target position after the parking space transposition operation is completed;

[0087] Specifically, the position error is not only affected by the error transmission of the mechanical mechanism itself, but also by speed disturbance, acceleration deviation and transmission system backlash during the execution process.

[0088] In this embodiment, in order to improve the accuracy of error identification, the platform running trajectory is fitted and compensated through fusion of state feedback. The state data includes but is not limited to acceleration change curve, limit feedback signal and horizontal displacement trajectory. Through the combination of state records of historical transposition operations, an error estimation model is constructed to establish a dynamic deviation relationship between the expected position and the actual response.

[0089] S2: correcting the position error value in combination with the identified vehicle parking error and gravitational potential energy deviation to obtain a mechanical error;

[0090] Specifically, in the transposition process, due to the actual parking position of the vehicle is not always strictly centered, the center of gravity distribution also exists a certain deviation, resulting in uneven load distribution of the platform structure in the running process, thereby causing the transmission path deformation or micro displacement response under the partial load.

[0091] In the embodiment, by fuzzy rule reasoning logic, the vehicle parking error and the gravity potential deviation are input as input variables into the fuzzy rule base including the coarse-grained, medium-grained and fine-grained rule subsets, and the correction suggestion for the position error is extracted through multi-layer fuzzy reasoning. The reasoning path is guided by the self-adaptive deformation weight network structure and the path intervention matrix, and the weight calculation is performed on the output results of each rule, so as to output the accurate error correction amount.

[0092] It can be understood that the foregoing correction logic can reduce the error uncertainty caused by the difference in vehicle state, and improve the pertinence and explainability of the correction output. In particular, in the repeated running process of the platform, the systematic deviation caused by the structural coupling error can be inhibited.

[0093] S3: updating a preset error distribution map according to the mechanical error, wherein the error distribution map is used to calculate a transposition target position of next transposition, and a transposition control instruction is generated according to the transposition target position;

[0094] In the embodiment, the error distribution map not only serves as a historical error recording tool, but also serves as a key basis for subsequent target point generation strategy. The system will actively call the map before transposition, judge whether the current target parking space is at an error high point according to the error distribution of the target parking space, and generate a micro deviation according to the map to adjust the target point by millimeter, so as to realize the active strategy of avoiding the mechanical dead zone or eccentric point. The structural life can be significantly prolonged, the error accumulation can be reduced, and the stability of long-term operation can be improved.

[0095] Before expanding the specific technical content of the present application, it needs to be further emphasized:

[0096] In the double-layer three-dimensional parking structure, the parking space transposition operation highly depends on the accurate docking of the platform structure. However, the actual landing point has an unavoidable micro error in each running process due to the limitations of the repeated motion accuracy of the actuator, the load disturbance factors and the external environmental influence. Although the error is usually controlled within millimeter level, the repetitive micro error will accumulate at certain spatial coordinate positions because the platform always docks at the same target point, which finally leads to the local force imbalance of the docking structure, resulting in the mechanical dead zone or error eccentric point. The structural imbalance will directly affect the structural stability, the limit assembly precision and the vehicle landing safety of the platform in the long-term running process.

[0097] It can be understood that the unmanned parking control method proposed in the embodiment does not attempt to compress the error range by improving the machining precision of the mechanical structure itself or the feedback frequency of the control unit, but actively identifies and avoids error-prone areas through dynamic adjustment logic and feedback learning logic, and realizes the error dispersion strategy of the transposition control logic. After the platform transposition is completed, the comprehensive landing error in the current operation process is identified according to the vehicle parking state and the transmission path disturbance, and the error is bound with the platform coordinates and written into the error distribution map of the corresponding target parking space, thereby forming a spatial error information map with historical traceability.

[0098] Further, the error distribution map relied on by the present application is not constructed based on ideal modeling or preset threshold, but is dynamically generated through the position offset results in the actual vehicle operation data. The core is not to establish a standard position template, but to master the density aggregation state and dynamic change trend of the error in the spatial coordinates, so as to deduce the load-sensitive area and structure adaptability area of the docking position. In a new round of transposition control, the control logic generates a transposition target position with a fine offset according to the spatial aggregation characteristics given by the error distribution map, so that the actual landing point of the platform each time is dispersed in space. This dispersion will not cause mechanical misalignment problems, but will help to evenly distribute the structural impact caused by micro errors to multiple points of the platform structure, slowing down the formation speed of local structure fatigue.

[0099] Next, the part of the method of the present application about the position error value is further expanded.

[0100] It can be understood that in the platform transposition process, the source of position deviation is multiple, including not only the long-term accumulation factors of stability such as geometric error, connection gap, assembly tolerance of linkage structure, but also the motion disturbance caused by instantaneous state fluctuation of the platform, such as inertia impact in acceleration and deceleration process, transmission lag, influence of environmental temperature on material expansion coefficient, etc. Such disturbance factors have certain randomness and change with each running state. If they are directly included in the error distribution map, the map will be distorted, and the subsequent transposition offset strategy will lose reference value.

[0101] That is, the purpose of compensating for the error is not to eliminate the mechanical error itself, but to strip off the non-structural deviation in the motion control process, so as to more accurately extract the mechanical error related to the transposition structure itself.

[0102] In one example, the specific steps of S1 are as follows:

[0103] S1.1: Obtain the actual landing position after the parking space transposition operation is completed, and generate the current physical position of the platform;

[0104] Specifically, the purpose of this step is to obtain the final physical position coordinates of the platform after the repositioning operation is completed, which is used for subsequent error comparison with the theoretical landing position.

[0105] In some optional embodiments, the current physical position of the platform is determined by combining the position encoder output, the boundary sensor trigger state, and the trajectory feedback signal. The position encoder can provide real-time feedback of the position scale of the linkage transmission structure, the boundary sensor is used to identify whether the platform is fully positioned, and the trajectory feedback signal provides the time sequence of the position change during the repositioning process. By integrating the three, the accurate position coordinates of the platform in the current cycle can be obtained. The position coordinates are absolute position data in a two-dimensional or three-dimensional structural coordinate system, and the unit can be set to millimeters or microns according to the platform control system.

[0106] S1.2: generating a repositioning reference value according to the repositioning target position in the repositioning control instruction;

[0107] In this embodiment, the repositioning reference value is based on the target landing position defined in the repositioning control instruction, specifically including the parking space structure coordinate center point, the platform motion end control amount, and the control time window parameter. The target landing position is obtained by modeling the parking space layout, which is a fixed node coordinate in a three-dimensional structure; the platform motion end control amount is given by the platform controller to ensure that the transmission structure completes the positioning action within the instruction time window; and the control time window parameter is used to limit the execution interval of the platform repositioning process, ensuring consistency of the target position in the control rhythm.

[0108] S1.3: obtaining state data of the platform during the repositioning process, wherein the state data includes platform acceleration, speed curve, and transmission feedback information;

[0109] Specifically, the purpose of this step is to provide dynamic data support for the subsequent error estimation model, and to evaluate the actual response state of the control execution by means of the kinematic parameters in the platform running process. The state change during the repositioning process will directly affect the stability and accuracy of the platform when it reaches the end point.

[0110] In this embodiment, the state data acquisition is composed of various sensors installed on the linkage device, the limit structure, and the platform body, including but not limited to three-axis acceleration sensors, speed closed-loop control feedback modules, load torque sensors, and limit position detectors.

[0111] Further, the acceleration data is used to judge the impact strength and dynamic stability of the platform during the repositioning process, the speed curve is used to evaluate the smoothness of the platform running track and whether there is overshoot or pullback phenomenon, and the transmission feedback information reflects the execution accuracy of the linkage device, such as motor angle, chain tension, etc.

[0112] In some optional embodiments, to avoid misjudgment caused by single data anomaly, multi-source fusion processing is performed on the state data, a weighted fusion-based method is used to perform synchronous analysis on different dimension data, so as to improve the reliability of the data.

[0113] S1.4: According to the state data, the transposition reference value is compensated through a preset error estimation model to obtain a state-compensated theoretical landing position, wherein the error estimation model is constructed based on historical transposition state data.

[0114] Specifically, the goal of this step is to include the actual running state in the transposition process into the adjustment process of the target position, dynamically compensate the deviation caused by factors such as control accuracy, load disturbance or path damping through the error estimation model, so that the theoretical landing position used for comparison is closer to the position that the platform should reach under the current state.

[0115] In this embodiment, the error estimation model is a fitting model constructed based on historical transposition state data, which establishes a mapping relationship between the state data and the landing deviation through a multivariate nonlinear regression method, and the input variables include the state data. The model can be trained based on historical transposition state data through least square method or gradient optimization strategy, which will not be repeated here.

[0116] S1.5: Calculate the difference between the current physical position of the platform and the state-compensated theoretical landing position to obtain a position error value.

[0117] Next, the part of the method of the present application related to the correction of the position error value will be further expanded.

[0118] Please refer to Figure 4 , Figure 4 for the flowchart of the position error value correction of the embodiments of the present application.

[0119] In this embodiment, the correction process is realized through fuzzy reasoning logic. The purpose of the correction process is to further analyze the position error generated in the transposition process into the deviation caused by controllable factors, and to identify and correct it through reasoning logic, so that the final mechanical error is more representative and repeatable. The introduction of fuzzy reasoning logic effectively solves the problems of inevitable error disturbance in actual operation, such as non-linear relationship, strong boundary fuzziness and unstable fluctuation range.

[0120] Further, the fuzzy inference logic includes a preset fuzzy rule base, which takes the vehicle parking error and the gravity potential deviation measured in advance as input variables, and corrects the position error value through the fuzzy rule base, wherein the vehicle parking error refers to the lateral offset, longitudinal offset, load center of gravity offset value and local torque generated thereby, relative to the platform limiting structure, when the vehicle is actually parked on the platform, which is usually obtained by image recognition, limiting sensing and load sensing, etc., and the gravity potential deviation refers to the gravity potential difference generated by the change in platform height during the transposition movement of the platform and the vehicle, which reflects the asymmetric deformation or load disturbance that the heavy load structure may bring during the docking process of the high and low platforms.

[0121] Still further, the fuzzy rule base includes a first rule subset for coarse-grained error inference, a second rule subset for medium-grained error inference, and a third rule subset for fine-grained error inference, and further includes an adaptive deformation inference weight network structure and a path intervention matrix, wherein the adaptive deformation inference weight network structure is used to adjust the weight coefficients of different inference paths in real time according to the error distribution trend in the historical error distribution map, so as to realize the improvement of inference sensitivity in the error high-occurrence area and the optimization of inference precision in the error stable area; and the path intervention matrix is used to dynamically regulate the cross relationship between the input variables, so as to solve the problem of coupled influence of some non-independent factors (such as load distribution and potential deviation linkage) in the inference path, and limit the influence boundary and avoid inference interference through path intervention rules.

[0122] In some optional embodiments, the construction of the fuzzy rule base includes the following aspects:

[0123] In the first aspect, the construction of each rule subset is completed by combining rule extraction and fuzzy classification. Specifically, first, based on a large number of historical error samples, the change trend between the input variables and the position offset is analyzed through a fuzzy clustering algorithm, and sample clusters with regular features are extracted, and then combined with the boundary conditions and weight constraints set by humans, they are inducted into “fuzzy IF-THEN rules”. Each rule contains the definition of fuzzy language value, corresponding membership function and the control range of output quantity.

[0124] For example, the coarse-grained rule may be: “when the vehicle lateral offset is large, the gravity deviation is moderate, and the position correction amount is large”, wherein “large”, “moderate” and “large” are fuzzy language items, which are modeled using triangular or trapezoidal membership functions. Each rule will be labeled with the corresponding applicable domain, confidence coefficient and priority identifier, so as to facilitate fast activation and fusion during subsequent inference.

[0125] In the second aspect, the fuzzy rule base further comprises intermediate inference logic as the key logic between rule triggering and output fusion. The intermediate inference logic is implemented by a fuzzy inference engine, including rule activation detection, fuzzy value combination, conflict rule processing and output fusion module.

[0126] It can be understood that, in order to avoid output conflicts between multiple rules, the weighted average method of weight fusion is used to dynamically calculate the activation strength of each rule in combination with the inference weight network structure.

[0127] In the embodiment, the inference weight network structure is in the form of a tensor, containing three dimensions: input variable level, rule subset number and error space index.

[0128] Further, the value of each node in the tensor represents the confidence of the output of a rule in the current context, and the node value is automatically adjusted according to the error distribution map, realizing adaptive deformation of inference strength in different parking spaces or use states, so that the fuzzy inference has the ability of historical memory and dynamic adjustment.

[0129] In the third aspect, in order to deal with the cross interference and causal inversion problem between input variables, a path intervention matrix is configured in the fuzzy rule base, which is used to dynamically manage the inference path control relationship between input variables. The path intervention matrix is a sparse connection structure, recording the intervention triggering conditions, path priority and cross linkage factors between input variables.

[0130] For example, when features such as abnormal vehicle wheelbase, severe gravity center deviation or platform posture asymmetry are detected, the path intervention matrix will activate a specific intervention path, guide the inference engine to enter the standby rule channel, and appropriately adjust the amplitude and response domain of the inference output. For example, when the gravity potential deviation is greater than the preset potential threshold, the path intervention matrix will activate the gravity deviation enhancement path, increase the activation probability of related rules and suppress other interference paths.

[0131] In one example, the specific steps of S2 are as follows:

[0132] S2.1: input the input variables into the first rule subset, the second rule subset and the third rule subset respectively, to obtain a plurality of fuzzy correction output sets corresponding to each granularity rule subset;

[0133] Specifically, the core purpose of this step is to design multi-level fuzzy reasoning channels for different scales and forms of deviation, so as to realize error correction with better resolution and context awareness. Since the vehicle parking error and the gravitational potential energy deviation may be jointly affected by factors such as platform attitude, load distribution, vehicle shape, etc. in actual scenarios, a single granularity rule set cannot comprehensively cover all reasoning paths, and is prone to identification jitter or correction fluctuations in boundary or critical states. Therefore, the fuzzy rules are divided into three categories: coarse-grained rule subset, medium-grained rule subset and fine-grained rule subset, which respectively undertake three types of functions: large error response, neutral range correction and fine disturbance adjustment. The three rule sets complement each other, can be executed in parallel and updated independently, forming a dynamic collaborative fuzzy response mechanism.

[0134] In this embodiment, the vehicle parking error and the gravitational potential energy deviation in the input variables are first subjected to fuzzy processing. The fuzzy processing can adopt the way of triangular membership function or Gaussian membership function, which is not described herein.

[0135] Further, the vehicle parking error is divided into fuzzy levels according to the longitudinal offset, lateral offset and tire limiting relationship, and the gravitational potential energy deviation is subjected to fuzzy processing according to the center of gravity height difference, platform inclination and load center calculation results. Each input variable is mapped to a membership space, each level in the space corresponds to a fuzzy label.

[0136] Further, the fuzzy input variables are combined as condition items and input into three fuzzy rule subsets. The rules in each rule subset are expressed in the form of classical "IF-THEN".

[0137] For example, the rule body is as follows:

[0138] The first rule subset example rule is: if the vehicle longitudinal offset is "high" and the gravitational potential energy deviation is "extremely high", the position error correction value is "strong backward pull bias";

[0139] The second rule subset example rule is: if the vehicle lateral offset is "medium" and the gravitational potential energy deviation is "medium", the position error correction value is "gentle right shift";

[0140] The third rule subset example rule is: if the vehicle longitudinal offset is "low" and the gravitational potential energy deviation is "slight", the position error correction value is "fine left correction".

[0141] It can be understood that each subset is executed by an independent fuzzy inference engine, and the minimum-maximum inference method is used for rule activation and output fuzzy set fusion. The activation degree of each rule is obtained by calculating the membership function of the fuzzy input condition, and the corresponding fuzzy correction output set is finally generated, which represents the possible distribution range of the correction value in the output space.

[0142] S2.2: According to the input variables, the weight tensor nodes corresponding to each granularity rule subset in the inference weight network structure are called, the activation weight coefficients are calculated, and the fuzzy correction output set is weighted to obtain an initial inference result;

[0143] Specifically, since the importance of rules of different granularities under different operating conditions has dynamic change characteristics, fixed weight fusion is easy to cause inference deviation. Therefore, the deformable weight tensor structure is introduced in the present application, and the outputs of each rule subset are dynamically weighted to obtain a more globally adaptive fusion result.

[0144] In the present embodiment, the weight tensor nodes form an index space according to the fuzzy level of the input variables, the regional weight in the historical error distribution map, and the response frequency of the subset. Through the error type and regional distribution recorded in the historical transposition operation, a group of weight entropy value distribution is formed to adjust the response slope and activation width of the nodes in the tensor structure. In the actual inference process, the current input variables will be matched to the corresponding activation nodes in the tensor network, and the fusion coefficients of each granularity subset are calculated through the output, and then the fuzzy correction output set is weighted.

[0145] Further, the inference weight network structure includes a tensor structure, the dimensions of the tensor structure correspond to the fuzzy level of the input variables and the rule activation strength, and the inference weight network structure is deformed by the current error distribution map.

[0146] In the present embodiment, in order to enhance the adaptability of the inference structure, the node weight values of the tensor structure can be deformed according to the current error distribution map. The error distribution map is composed of mechanical error samples in a plurality of historical transposition operations, and records the error amplitude and error change trend corresponding to each platform landing position. According to the map, the error high-incidence area and the error stable area can be identified, and based on this, the weight nodes associated with the error high-incidence area in the tensor structure are stretched, that is, the response weight is increased, and the nodes associated with the error stable area are compressed, that is, the weight amplitude is reduced or the activation domain is reduced. The weight deformation process can be realized by adjusting the weight value, response slope and width of the activation domain of the node, which can dynamically improve the accuracy of the fuzzy inference result without changing the original rule structure.

[0147] In some optional embodiments, the process of reasoning weight network structure morphing includes the following three dimensions:

[0148] In the first dimension, according to the distribution trend of mechanical errors recorded in the historical transposition operation, the error high-incidence area and the error stable area in the error distribution atlas are determined;

[0149] Specifically, the error distribution atlas takes the actual positioning coordinates of the parking platform in each transposition operation as the index, records the mechanical error value and its statistical changes of the corresponding position, such as error mean, variance, confidence interval, density and other information. Through spatial clustering and density analysis of historical error samples, the area where the mechanical error value frequently abnormally or fluctuates greatly can be marked in the coordinate graph, which is defined as the error high-incidence area. Relatively speaking, the area where the mechanical error value changes little, is distributed and stable in the long term can be identified as the error stable area.

[0150] In the second dimension, for the error high-incidence area, the corresponding weight node in the tensor structure is located, and the corresponding weight node and the adjacent node are stretched according to the entropy value of the error distribution atlas;

[0151] In this embodiment, the so-called stretching deformation refers to the enhancement operation on the weight value, activation function slope or action range of a certain weight node, so as to improve the expression ability of the rule in the reasoning process. For example, when a certain position in the error high-incidence area should be combined with a specific input fuzzy level, the node position mapped in the tensor structure can be identified, and the weight coefficient of the node or the activation interval can be widened to increase the fuzzy rule output weight of the area. The adjustment of the adjacent node can adopt the Gaussian adjacent weight decay mechanism, so that the weight changes moderately, avoiding the formation of sudden rules output, thereby enhancing the focusing and sensitivity of the rule response in the abnormal area, which is beneficial to the fine capture of extreme errors.

[0152] In the third dimension, for the error stable area, the corresponding weight node in the tensor structure is located, and the corresponding weight node and the adjacent node are compressed according to the entropy value of the error distribution atlas;

[0153] In this embodiment, the compression deformation aims to reduce the sensitivity and activation range of the weight node in the area, so as to avoid the over-intervention of the fuzzy rule in the stable area where no response is needed. The weight downscaling process can be realized by reducing the activation domain width of the node, reducing the response slope or directly reducing the weight coefficient. Gradient reduction mechanism can also be adopted for the adjacent nodes to converge the reasoning output fluctuation of the area as a whole. In the running process, unnecessary correction caused by small perturbations or measurement errors in the stable area can be effectively avoided, thereby reducing the risk of redundant calculation and error compensation, and improving the robustness and stability of the correction logic.

[0154] S2.3: obtaining vehicle state data in the parking space transposition operation, determining intervention variables associated with the input variables in the path intervention matrix, and selecting a corresponding intervention path according to the state values of the intervention variables;

[0155] Specifically, abnormal states such as structural mismatch, gravity center deviation and asymmetric load distribution may occur during the vehicle transposition process. Although such variables are not direct input variables, they can affect the context logic of error correction. To enhance the response capability of the reasoning process to the actual state, in the fuzzy reasoning process, part of the state data is embedded in the reasoning path through the path intervention matrix, and the active adjustment of the reasoning chain is realized.

[0156] In this embodiment, the path intervention matrix is pre-set with several intervention paths, each path is associated with one or more state variables, and when the intervention variable value meets the set condition, the path switching or additional correction logic is triggered.

[0157] In one example, the intervention path includes at least one of the following:

[0158] When there is a geometric mismatch between the vehicle wheelbase and the platform limiting structure, the path intervention matrix controls the vehicle wheelbase variable to intervene in the reasoning path of the vehicle parking error, adjusts the activation probability, response amplitude and output priority of the corresponding rule;

[0159] It can be understood that by increasing the activation probability of the vehicle wheelbase variable in the related fuzzy rule, the response amplitude of the fuzzy output is enhanced, and at the same time, the priority is increased to make it have higher decision-making power in rule conflict. The process is implemented based on a configurable criterion function, ensuring that the intervention behavior is only performed when necessary, avoiding invalid intervention.

[0160] When the gravity potential deviation during the transposition operation process is greater than the pre-set potential threshold, the path intervention matrix controls the gravity potential deviation variable to intervene in the reasoning path of the gravity potential deviation, and adjusts the activation range of the corresponding rule, wherein the activation range adjustment includes compression or widening;

[0161] When an asymmetric load distribution state is detected during the transposition operation process, the path intervention matrix controls the load distribution variable to intervene in the reasoning path of the vehicle parking error and the gravity potential deviation, calculates the path cross-linking rate, and activates the standby fuzzy rule.

[0162] It can be understood that the activation factor of the input membership degree corresponding to the variable is synchronously raised on the related fuzzy rules in the two paths, and the path cross linkage rate is solved by matrix, that is, the number and degree of rules in the two paths that are synchronously changed due to the state change of the variable are calculated, so as to identify the dominance of the variable in the current reasoning chain; if the linkage rate exceeds a preset threshold, a set of pre-provided fuzzy rules are activated, which are used to process the scene of asymmetric interference and existing composite deviation, so as to improve the coverage of the overall reasoning model.

[0163] S2.4: defuzzifying the initial reasoning result according to the intervention rule of the intervention path to obtain a fuzzy correction amount, and synthesizing the fuzzy correction amount and the position error value to obtain a mechanical error;

[0164] Specifically, the initial reasoning result is in a fuzzy set form, which needs to be converted into a specific numerical value through defuzzification to form a final correction amount and be used for subsequent control instruction generation.

[0165] In the embodiment, first, a correction surface is constructed according to the final output rule in the intervention path, the activation center and the weight boundary of the fuzzy amount are extracted, and the numerical value of the correction amount is calculated through the barycentric weighting method. In the defuzzification process, the fuzzy level of the input variable is combined to adjust the correction amplitude and direction, so that the correction amount is structurally consistent with the original position error in direction, so as to prevent overcompensation or offset problems. Subsequently, the correction amount is synthesized with the position error value calculated in S1, and finally the mechanical error corresponding to the current operation of the platform is obtained.

[0166] Among them, the synthesis method can adopt vector superposition or nonlinear fusion, which is configured according to the differences between the platform motion characteristics and the error manifestation forms.

[0167] In one example, the vehicle parking error identification step includes:

[0168] Obtaining vehicle image data and platform state information before the vehicle parking position operation is initiated, wherein the vehicle image data includes a parking position image of the vehicle on the platform, and the platform state information includes a limiting structure, load sensing data and platform attitude information;

[0169] According to the vehicle image data, the relative position between the vehicle tire and the limiting structure is detected by an image recognition algorithm, and the offset amount of the vehicle on the platform is calculated, wherein the offset amount includes a lateral offset amount and a longitudinal offset amount;

[0170] According to the load sensing data, the center of gravity position of the vehicle on the platform is determined and compared with the center point of the platform structure to obtain a center of gravity offset value;

[0171] According to the platform attitude information, the local torque generated by the vehicle on the platform is calculated.

[0172] The vehicle parking error is calculated in combination with the offset, the center of gravity offset value and the local torque.

[0173] Specifically, the identification of the vehicle parking error is to accurately evaluate the spatial offset between the actual parking state of the vehicle on the platform and the ideal design state, and to provide a data basis for subsequent position error correction and transposition control. As a passive load, the parking state of the vehicle will change randomly due to driving behavior, performance of the limiting device and rigid response of the platform. If such errors cannot be identified in advance, it will directly lead to the misalignment of the platform transposition action reference, produce cumulative mechanical offset, and affect the overall precision and stability of the unmanned parking control.

[0174] In the embodiment, first, the vehicle image data before the parking space transposition operation is initiated and the platform state information are acquired, wherein the image data is collected by a top or lateral visual sensing module, including a sequence of original image frames composed of the four-wheel landing state of the vehicle and the platform boundary features. The platform state information is acquired by the built-in attitude sensor, the limiting detector and the load sensor, including but not limited to the contact state signal of the limiting structure, the load distribution vector measured by the sensor node and the platform surface inclination data. The image frames and the state data are synchronized by time stamp to ensure the construction of a complete initial parking state description vector.

[0175] Further, for the vehicle image data, a contour recognition algorithm based on deep learning is used for processing, combining a convolutional neural network and an image edge enhancement technology to separate the vehicle tire outer contour from the image, and extract the minimum distance value between the tire boundary and the limiting structure. The lateral offset and the longitudinal offset of the vehicle in the platform coordinate system are calculated by the way of bidirectional vector distance measurement.

[0176] In the embodiment, the load sensing data is collected by the multi-point piezoelectric pressure sensor arranged below the platform. The sensor array can sense the distribution center of gravity of the vehicle load on the platform plane in real time. The platform structure center point is a reference point automatically calculated and pre-set according to the platform geometric boundary parameters in advance. By inputting the load distribution matrix fed back by the sensor into the bidirectional cosine center of gravity calculation model, the offset of the center of gravity of the vehicle relative to the center of the platform can be determined. The result is defined as the center of gravity offset value, which is used for subsequent mechanical analysis and inference correction.

[0177] Further, the platform attitude information is collected by a high-precision IMU module, including a pitch angle, a roll angle and a local vibration frequency characteristic. In combination with a gravity center offset value, a local torque caused by an asymmetric load distribution or a deviation of a limiting support of the vehicle can be derived. A torque model is constructed by using a simplified model of an equivalent rigid body eccentric force, small dynamic disturbances are ignored, and only the torque contribution under the stable distribution condition is considered. The direction and amplitude of the torque value are used to determine whether there is structural asymmetric compression of the vehicle on the platform structure, and are used as auxiliary variables in the fuzzy reasoning logic to participate in rule activation.

[0178] In one example, the preset error distribution atlas is updated according to the mechanical error, including:

[0179] S3.1: Obtain a target parking space involved in the current transposition operation, and obtain an error distribution atlas corresponding to the target parking space;

[0180] S3.2: In the corresponding error distribution atlas, record the corresponding platform landing coordinates and the mechanical error of the current transposition operation to obtain a new error sample;

[0181] Specifically, the purpose of this step is to construct a structured new error sample to be included in the error distribution atlas for incremental updating. The new error sample includes the actual landing coordinates of the platform in the transposition action (obtained through position sensors and image recognition feedback) and the mechanical error value corrected according to the fuzzy reasoning.

[0182] In this embodiment, to ensure sample computability, the error sample is saved in a structure, including spatial three-coordinate, error vector, collection timestamp and transposition state label. The error vector includes the numerical amplitude and direction component of the combined error, which is used for subsequent distribution evolution modeling.

[0183] S3.3: Fuse the new error sample with historical error data in the error distribution atlas, update the error mean value, error fluctuation amplitude and statistical density index corresponding to the target parking space;

[0184] In this embodiment, the update of the error mean value adopts a time decay weighted average strategy, giving higher weight to recent samples; the error fluctuation amplitude is dynamically calculated based on the standard deviation or the maximum-minimum difference, to reflect the uncertainty of the error distribution; the statistical density index is estimated by the number of samples per unit area, to reflect the error aggregation degree of the region in history.

[0185] S3.3: Calculate the distribution density of the new error sample in the error distribution atlas according to the error mean value, error fluctuation amplitude and statistical density index, and perform regional reconstruction on the error distribution atlas according to the distribution density, wherein the regional reconstruction includes adjusting the aggregation domain range and granularity density of the error points.

[0186] S3.4: In the error distribution map corresponding to the target parking space, determine the position region according to the aggregation domain range and the granularity density;

[0187] Specifically, the purpose of this step is to adjust the division method of the error region according to the distribution state of the new error sample in the error distribution map, so as to classify and mark the stability region and the abnormal region subsequently. The distribution density reflects the error aggregation, and high density often means the risk of repetitive error or structural error.

[0188] In this embodiment, by performing spatial density clustering on the map grid, a dynamic region sliding window statistical strategy is used to detect high error point aggregation area, and the original grid aggregation domain range is adjusted accordingly. If the local error fluctuation is significant, the granularity density is reduced to achieve more detailed data representation; otherwise, adjacent low-density regions are merged to reduce the model complexity.

[0189] S3.5: According to the position region, determine the transposition target position by particle swarm algorithm;

[0190] Specifically, the last step is to use the constructed map structure as a search space, and use intelligent algorithms to predict and re-plan the next transposition target position. The particle swarm algorithm combines the error aggregation domain, directional trend and low error region probability weight in the search process to ensure that the selected target point has a higher landing success rate.

[0191] In this embodiment, each position region in the error distribution map is regarded as the corresponding position of the search particle, and the historical error mean is used as the fitness function basis. The particle velocity and position direction are updated by iteration, and finally the local optimal target point is determined. The fitness evaluation standard not only considers the minimum position error, but also comprehensively evaluates the error fluctuation risk and the region stability, so as to output the optimal transposition control target point coordinates.

[0192] In one example, the embodiment of the present application provides an unmanned parking control system, which comprises:

[0193] A platform control module is configured to control a linkage device in a double-deck stereoscopic parking system to perform a parking space transposition operation, and to collect actual landing position, transposition target position and state data in the transposition process after the transposition operation is completed, and to calculate a position error value when the platform reaches the transposition target position;

[0194] An error correction module is configured to perform fuzzy reasoning correction on the position error value based on the identified vehicle parking error and the gravitational potential energy deviation, to obtain a mechanical error, wherein the error correction module comprises a fuzzy rule base, an inference weight network structure and a path intervention matrix;

[0195] The error distribution atlas of the corresponding target parking space is updated according to the mechanical error, and a target position for the next position switching is calculated based on the error distribution atlas, and a position switching control instruction is generated for the next position switching operation.

[0196] Although the embodiments of the present application have been shown and described above, it should be understood by those skilled in the art that the above embodiments are exemplary and cannot be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for unmanned parking control, applied to a double-decked stereoscopic parking system, characterized in that, The double-layer stereoscopic parking system includes upper-layer parking spaces and lower-layer parking spaces, and when the lower-layer parking spaces are in an occupied state, idle parking spaces in the upper-layer parking spaces are replaced with the lower-layer parking spaces one by one through linkage devices, and the method comprises the following steps: When the parking space replacement operation is completed, a position error value of a platform controlled by the linkage device reaching a replacement target position is calculated; The position error value is corrected in combination with a recognized vehicle parking error and a gravitational potential energy deviation to obtain a mechanical error, wherein the correction comprises taking the vehicle parking error and the gravitational potential energy deviation as input variables of fuzzy reasoning and inputting them into a fuzzy rule base comprising a coarse-grained rule subset, a medium-grained rule subset and a fine-grained rule subset for multi-scale error correction reasoning; The mechanical error is used to update a preset error distribution map, wherein the error distribution map is used to calculate a replacement target position of the next replacement, and a replacement control instruction is generated according to the replacement target position.

2. The unmanned parking control method according to claim 1, wherein The position error value of the platform controlled by the linkage device reaching the replacement target position comprises the following steps: An actual landing position after the parking space replacement operation is completed is obtained to generate a current physical position of the platform; A replacement reference value is generated according to the replacement target position in the replacement control instruction; State data of the platform in the replacement process is obtained, wherein the state data comprises platform acceleration, speed curve and transmission feedback information; The replacement reference value is compensated by a preset error estimation model based on the state data to obtain a state-compensated theoretical landing position, wherein the error estimation model is constructed based on historical replacement state data; A difference between the current physical position of the platform and the state-compensated theoretical landing position is calculated to obtain the position error value.

3. The unmanned parking control method according to claim 1, wherein The position error value is corrected in combination with the recognized vehicle parking error and the gravitational potential energy deviation, which comprises the following steps: The vehicle parking error and the gravitational potential energy deviation are taken as input variables of fuzzy reasoning according to a preset fuzzy rule base, and the position error value is corrected by the fuzzy rule base, wherein the fuzzy rule base comprises a first rule subset for coarse-grained error reasoning, a second rule subset for medium-grained error reasoning and a third rule subset for fine-grained error reasoning.

4. The unmanned parking control method according to claim 3, wherein The fuzzy rule base further comprises an adaptive morphing reasoning weight network structure and a path intervention matrix, and the position error value is corrected by the fuzzy rule base, which comprises the following steps: The input variables are input into the first rule subset, the second rule subset and the third rule subset respectively to obtain a plurality of fuzzy correction output sets corresponding to each rule subset; According to the input variables, a weight tensor node corresponding to each rule subset is called in the reasoning weight network structure to calculate an activated weight coefficient, and the fuzzy correction output sets are weighted and calculated to obtain an initial reasoning result; Vehicle state data in the parking space replacement operation is obtained, an intervention variable associated with the input variables is determined in the path intervention matrix, and a corresponding intervention path is selected according to a state value of the intervention variable; The initial inference result is de-blurred according to an intervention rule of the intervention path, to obtain a blur correction amount, and the blur correction amount is combined with the position error value to obtain a mechanical error.

5. The unmanned parking control method according to claim 4, wherein The inference weight network structure includes a tensor structure, dimensions of the tensor structure correspond to blur levels of input variables and rule activation strengths, and the inference weight network structure is deformed by a current error distribution map, wherein: According to the distribution trend of the mechanical error recorded in the historical transposition operation, the error high-incidence area and the error stable area in the error distribution map are determined; For the error high-incidence area, corresponding weight nodes are located in the tensor structure, and the corresponding weight nodes and adjacent nodes are stretched according to the entropy value of the error distribution map; For the error stable area, corresponding weight nodes are located in the tensor structure, and the corresponding weight nodes and adjacent nodes are compressed according to the entropy value of the error distribution map; The deformation is operated by adjusting the weight coefficient value, slope gradient and activation domain width of the weight nodes.

6. The unmanned parking control method according to claim 4, wherein The path intervention matrix is used to adjust the cross-path regulation relationship between input variables in the fuzzy inference process, wherein the intervention path at least includes one of the following: When there is a geometric mismatch state between the vehicle wheelbase and the platform limiting structure, the path intervention matrix controls the vehicle wheelbase variable to intervene in the inference path of the vehicle parking error, adjusts the activation probability, response amplitude and output priority of the corresponding rule; When the gravity potential deviation in the transposition operation process is greater than a preset potential threshold, the path intervention matrix controls the gravity potential deviation variable to intervene in the inference path of the gravity potential deviation, and adjusts the activation range of the corresponding rule, wherein the activation range adjustment includes compression or widening; When an asymmetric load distribution state is detected in the transposition operation process, the path intervention matrix controls the load distribution variable to simultaneously intervene in the inference paths of the vehicle parking error and the gravity potential deviation, calculates the path cross-linkage rate, and activates the standby fuzzy rule.

7. The unmanned parking control method according to claim 1, wherein The vehicle parking error identification step includes: Obtaining vehicle image data and platform state information before the transposition operation of the parking space is initiated, wherein the vehicle image data includes a parking position image of the vehicle on the platform, and the platform state information includes limiting structure, load sensing data and platform attitude information; According to the vehicle image data, the relative position between the vehicle tire and the limiting structure is detected by an image recognition algorithm, and the offset amount of the vehicle on the platform is calculated, wherein the offset amount includes a lateral offset amount and a longitudinal offset amount; According to the load sensing data, the center of gravity position of the vehicle on the platform is determined and compared with the center point of the platform structure to obtain a center of gravity offset value; According to the platform attitude information, the local torque generated by the vehicle on the platform is calculated; The vehicle parking error is calculated in combination with the offset amount, the center of gravity offset value and the local torque.

8. The unmanned parking control method of claim 1, wherein, The preset error distribution map is updated according to the mechanical error, including: Obtaining a target parking space involved in the current transposition operation to obtain an error distribution map corresponding to the target parking space; In the corresponding error distribution map, record the mechanical error corresponding to the platform landing coordinate and the current transposition operation, and obtain a new error sample; Fuse the new error sample with the historical error data in the error distribution map, and update the error mean value, error fluctuation amplitude and statistical density index corresponding to the target parking space; According to the error mean value, error fluctuation amplitude and statistical density index, calculate the distribution density of the new error sample in the error distribution map, and according to the distribution density, reconstruct the region of the error distribution map, wherein the region reconstruction includes adjusting the aggregation domain range and granularity density of error points.

9. The unmanned parking control method according to claim 8, wherein The calculation of the transposition target position of the next transposition includes: In the error distribution map corresponding to the target parking space, determine the position region according to the aggregation domain range and granularity density; According to the position region, determine the transposition target position by particle swarm algorithm.

10. An unmanned parking control system for implementing an unmanned parking control method according to any one of claims 1 to 9, characterized in that, The system includes: A platform control module is configured to control the linkage device in the double-layer stereoscopic parking system to perform a parking space transposition operation, and to collect the actual landing position of the platform, the transposition target position and the state data in the transposition process after the transposition operation is completed, and to calculate the position error value when the platform reaches the transposition target position; An error correction module is configured to correct the position error value based on the identified vehicle parking error and the gravitational potential energy deviation, and to obtain a mechanical error, wherein the error correction module includes a fuzzy rule base, an inference weight network structure and a path intervention matrix; A map updating and control module is configured to update the error distribution map of the corresponding target parking space according to the mechanical error, and to calculate the transposition target position of the next transposition based on the error distribution map, and to generate a transposition control instruction for the next round of transposition operation.

Images