An automatic positioning method and system for portal crane based on multi-sensing fusion

Abstract

The present application discloses an automatic positioning method and system for a gantry crane based on multi-sensory fusion, the method comprising: using two types of sensors to respectively obtain the first and second position data of each type of operating mechanism of the gantry crane at the current moment; collecting real-time environmental data, integrating the real-time environmental data with the first and second position data respectively, and inputting the corresponding matching error compensation models respectively to obtain the first and second position data after error compensation; using GNSS positioning technology to collect GNSS signal data, inputting the data into a GNSS positioning fault judgment module together with the real-time environmental data, calculating and obtaining the GNSS RTK position according to the judgment result, and using the GNSS RTK position to calibrate the third position generated by the calibration of the first and second position data to obtain the final position or based on the GNSS RTK position received at the previous moment as a base point, combining the third position to obtain the final position; the present application can achieve more accurate and stable positioning.

Description

Technical Field

[0001] The present application relates to the technical field of automatic positioning of gantry cranes, and specifically to an automatic positioning method and system for gantry cranes based on multi-sensory fusion. Background Art

[0002] As an important equipment for port operations, gantry cranes are widely used in shipyards, dock loading and unloading and other places. In the process of automation and intelligence of modern ports, the high-precision positioning technology of cranes has become an important support for improving operation efficiency and safety. High-precision positioning not only improves operation efficiency, but also greatly reduces potential safety hazards, and promotes the modernization and digital development of port management.

[0003] At present, in order to achieve high-precision positioning of gantry cranes, the industry generally adopts a variety of technical means, mainly including laser rangefinders, encoders, barcode positioning technology and global navigation satellite systems (GNSS). Among them, laser rangefinders can provide instantaneous distance measurements, but the stability and accuracy of their data are easily affected by vibration; encoders are often used to provide position feedback, but due to errors in mechanical processing and installation, the actual accuracy is difficult to guarantee; barcode positioning technology is easily affected by reflections under strong light or direct sunlight, resulting in reduced recognition efficiency; and GNSS positioning can provide continuous positioning services with high accuracy, but the signal is easily interfered in complex environments.

[0004] There are also some methods in the existing technology that use a combination of multiple positioning technologies to achieve precise positioning, but there are still problems in practical applications. Since environmental factors have different degrees of influence on each positioning technology, these influences cannot be simply calibrated by using the position determined by the positioning technology with less influence. Instead, compensation is required based on actual conditions to minimize the impact of environmental factors on the accuracy of gantry crane positioning. Once a positioning failure occurs, further research is needed on how to set up redundant control strategies to improve the stability and reliability of gantry crane positioning. Summary of the invention

[0005] In order to enhance the environmental adaptability of the automatic positioning process of a gantry crane and achieve more accurate and stable positioning, the present application provides a method and system for automatic positioning of a gantry crane based on multi-sensory fusion.

[0006] In a first aspect, the present application discloses an automatic positioning method for a portal crane based on multi-sensor fusion, comprising: for each type of operating mechanism in the portal crane, two types of sensors are used to respectively obtain position data of the operating mechanism, which are recorded as first position data and second position data respectively; the types of the operating mechanism include: a lifting mechanism, a luffing mechanism, a rotating mechanism and a walking mechanism;

[0007] Real-time environmental data is collected by using an environmental sensor installed on a portal crane, and the real-time environmental data is respectively integrated with the first position data and the second position data, and the integrated data is respectively input into a first error compensation model that matches a corresponding type of sensor and a corresponding type of operating mechanism to obtain the first position data after error compensation and the second position data after error compensation; the first position data after error compensation is calibrated using the second position data after error compensation to generate third position data; the first error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data; GNSS positioning technology is used to collect GNSS signal data in real time, and the real-time collected GNSS signal data and real-time environmental data are input into a GNSS positioning fault judgment module to obtain the result of whether the GNSS positioning fault occurs at the current moment; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated using historical environmental data and GNSS signal data marked with faults; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is obtained by calculating the GNSS signal data, and the GNSS The RTK position is calibrated to the third position to obtain the final position of each type of operating mechanism; if the GNSS positioning fails at the current moment, the relative displacement and final position of each type of operating mechanism after the GNSS failure are calculated based on the GNSS RTK position received at the previous moment as the base point and combined with the third position.

[0008] By adopting the above scheme, two types of sensors are used to obtain the initial position information of the gantry crane, and the neural network model is used to compensate the error of the initial position information affected by the environment, and then the multi-sensor positions are calibrated with each other to improve the accuracy of positioning; by collecting GNSS signal data to achieve further calibration, the influence of the environment that may cause GNSS positioning failure is considered to ensure that when the GNSS signal fails, the accurate final position can still be calculated through the position information of the previous moment and the multi-sensory fusion positioning result, thereby enhancing the robustness and adaptability of the system.

[0009] Preferably, it also includes:

[0010] The dimension data of each type of operating mechanism in the gantry crane is collected, and in the process of integrating the real-time environmental data with the first position data and the second position data respectively, the dimension data is integrated together to generate new integrated data; the integrated data is replaced by the new integrated data, and the data are input into the second error compensation model matched with the corresponding type of sensor and the corresponding type of operating mechanism respectively to obtain the first position data after error compensation and the second position data after error compensation; the second error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical dimension data of the corresponding type of operating mechanism, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data.

[0011] By adopting the above scheme, taking into account that the operating mechanisms of gantry cranes of different sizes are affected to different degrees by the same environmental factors, there are differences in the corresponding error compensation required. By using the neural network model, the size of the gantry crane is further considered to compensate for environmental factors, so as to obtain more accurate error compensation, thereby improving the accuracy and robustness of the positioning system.

[0012] Preferably, the method of using the error-compensated second position data to calibrate the position of the error-compensated first position data includes: there are multiple second position data, which are respectively collected and acquired by multiple sensors of the same type; for each type of operating mechanism in the gantry crane, the first position data are respectively calculated with the multiple second position data, and the calculated errors are clustered to obtain second position data corresponding to the error calculation result that is more than a preset distance away from the cluster center, the obtained second position data is identified as abnormal second position data, and any second position data except the abnormal second position data is selected to calibrate the first position data to generate third position data.

[0013] By adopting the above scheme, considering that the sensor may fail during the positioning process using the sensor, the positioning data is obtained by multiple sensors for error comparison to determine the faulty sensor, and the position data collected by the faulty sensor is eliminated, thereby improving the calibration accuracy and reliability.

[0014] Preferably, it also includes:

[0015] The real-time collected GNSS signal data and real-time environmental data are input into the GNSS positioning fault risk assessment judgment module to obtain an assessment result of whether the GNSS signal at the current moment has a fault risk; the GNSS positioning fault risk assessment module adopts a neural network model, and is trained and generated using historical environmental data and GNSS signal data marked with whether there is a fault risk; if the current GNSS signal has a fault risk, a high-gain antenna is selected at the next moment to receive the GNSS signal and the received GNSS signal is enhanced and filtered; otherwise, no processing is performed.

[0016] By adopting the above solution, while monitoring the GNSS signal in real time, it is timely evaluated whether there is a risk of failure, and based on the evaluation results, it is decided whether to enable the high-gain antenna for signal enhancement and filtering, thereby effectively improving the quality of the GNSS signal and the reliability of positioning.

[0017] Preferably, it also includes:

[0018] Statistics are taken of the final position and actual position located within a period of time including the current moment. If the ratio of the similarity between the final position and the real-time position is greater than the preset similarity number and the ratio of the similarity between the final position and the real-time position is not greater than the preset similarity number and is less than the preset ratio, then the first error compensation model is optimized and trained, and the frequency of environmental data collection and the frequency of position data collection of the operating mechanism are increased accordingly.

[0019] By adopting the above scheme, the acquired position is compared with the actual position, and the first error compensation model is optimized according to the comparison result to improve the accuracy of error compensation. The frequency of environmental data collection and the frequency of position data collection of the operating mechanism are increased to increase the number of calibrations, thereby improving the accuracy and reliability of the positioning system.

[0020] Preferably, it also includes:

[0021] The current operation cycle of the gantry crane is obtained, and the operating mechanisms in the gantry crane that are in operation and those that are in a stationary state within the current operation cycle are determined; for the operating mechanisms that are in operation, two types of sensors are selected to respectively obtain position data of the operating mechanisms, which are recorded as first position data and second position data respectively; for the operating mechanisms that are in a stationary state, the final position at the previous moment in the stationary state is selected to be used as the first position data and the second position data instead of using two types of sensors to respectively obtain the position data of the operating mechanisms.

[0022] By adopting the above scheme, considering the different states of the operating mechanisms in different operation cycles, for the operating mechanisms in the operating state, the position data is obtained in real time through at least two types of sensors to calibrate and improve the positioning accuracy; while for the operating mechanisms in the static state, the final position information of the previous moment is directly used, which simplifies the data processing process, saves computing resources, and improves the overall efficiency of the system.

[0023] Preferably, the adopting two types of sensors to respectively acquire the position data of the operating mechanism comprises:

[0024] The first absolute encoder and the incremental encoder are used to obtain the lifting height value of the lifting mechanism; the second absolute encoder and the inclinometer are used to obtain the working amplitude value of the variable amplitude mechanism; the third absolute encoder and the QR code reader are used to obtain the rotation angle value of the rotating mechanism; the fourth absolute encoder and the RFID magnetic nail read-write sensor are used to obtain the walking distance of the walking mechanism.

[0025] By adopting the above scheme, the characteristics of each operating mechanism of the gantry crane are taken into consideration, and the combination of the two types of sensors is adaptively planned to improve the positioning accuracy and robustness of each operating mechanism of the gantry crane.

[0026] In a second aspect, the present application provides an automatic positioning system for a portal crane based on multi-sensor fusion, including: a portal crane initial position acquisition module, which is used to use two types of sensors to respectively acquire position data of the operating mechanism for each type of operating mechanism in the portal crane, and record them as first position data and second position data respectively; the types of the operating mechanism include: a lifting mechanism, a luffing mechanism, a rotating mechanism and a walking mechanism;

[0027] The initial position compensation module of the gantry crane is used to collect real-time environmental data by using the environmental sensor installed on the gantry crane, integrate the real-time environmental data with the first position data and the second position data respectively, input the integrated data into the first error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism respectively, and obtain the first position data after error compensation and the second position data after error compensation; use the second position data after error compensation to calibrate the first position data after error compensation, and calibrate to generate third position data; the first error compensation model is a neural network model, which has multiple types, each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data;

[0028] The final position acquisition module of the portal crane is used to collect GNSS signal data in real time by adopting GNSS positioning technology, input the real-time collected GNSS signal data and real-time environmental data into the GNSS positioning fault judgment module, and obtain the result of whether the GNSS positioning fault occurs at the current moment; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated by using historical environmental data and GNSS signal data marked with faults or not; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is calculated by using the GNSS signal data, and the third position data is calibrated by using the GNSS RTK position to obtain the final position of each type of operating mechanism; if the GNSS positioning fails at the current moment, the relative displacement and final position of each type of operating mechanism after the GNSS failure are calculated based on the GNSS RTK position received at the previous moment as the base point and combined with the third position data.

[0029] By adopting the above solution, automatic positioning of the gantry crane in various complex environments can be achieved, and more accurate and stable positioning results can be obtained.

[0030] In a third aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method as described above.

[0031] In a fourth aspect, the present application provides a computer device, comprising a memory, a processor, and a program stored and executable on the memory, wherein the program implements the steps of the above method when executed by the processor.

[0032] In summary, this application has the following beneficial effects:

[0033] 1. The position data of different operating mechanisms of the gantry crane are collected through various types of sensors, which are convenient for mutual calibration and effectively reduce the error caused by a single positioning technology. Environmental data are collected in real time using environmental sensors, and then integrated with the size data and position data of the gantry crane and input into the error compensation model to achieve error compensation of the position data and further improve the positioning accuracy and robustness. The GNSS positioning technology and fault judgment mechanism are used to further calibrate the positioning results using the position data obtained by GNSS technology, and to ensure the continuity and reliability of positioning in the event of a GNSS positioning failure.

[0034] 2. Further consider the situation of sensor failure, use the error comparison results of multiple sensors to determine the faulty sensor, and combine with the multi-sensor redundant control strategy. When some sensors fail or are interfered with by the outside world, reliable positioning can still be maintained through the data provided by other sensors, and the position data obtained by the faulty sensor can be eliminated to enhance the robustness and stability of the system; 3. When it is detected that the GNSS signal may have a risk of failure, by selecting a high-gain antenna and enhancing and filtering the signal, the signal-to-noise ratio of the received signal can be significantly improved, noise interference can be reduced, and signal strength can be improved, further ensuring the accuracy and stability of positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of an automatic positioning method for a portal crane based on multi-sensing fusion described in a specific embodiment;

[0036] Figure 2 It is a diagram of obtaining position data by using multiple sensors in a method for automatic positioning of a portal crane based on multi-sensor fusion described in a specific embodiment;

[0037] Figure 3 It is a schematic diagram of GNSSS positioning in a portal crane automatic positioning method based on multi-sensor fusion described in a specific embodiment;

[0038] Figure 4 It is a structural schematic diagram of an automatic positioning system for a portal crane based on multi-sensing fusion described in a specific embodiment;

[0039] Figure 5 It is a diagram of the hardware structure of an automatic positioning system for a gantry crane based on multi-sensor fusion described in a specific embodiment. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0041] like Figure 1 As shown, the embodiment of the present application discloses an automatic positioning method for a portal crane based on multi-sensing fusion, and the specific steps include:

[0042] S1. Use various types of sensors to collect and obtain the position data of each type of operating mechanism in the gantry crane.

[0043] like Figure 2As shown in the figure, the types of operating mechanisms in the portal crane include: lifting mechanism, luffing mechanism, rotating mechanism and traveling mechanism. Considering that the different types of operating mechanisms are in different states during the operation cycle of the portal crane, such as: during the operation preparation stage, the traveling mechanism is in operation; during the loading and unloading operation stages, the lifting mechanism, luffing mechanism and rotating mechanism are in operation; in order to save positioning resources, the current operation cycle of the portal crane is obtained, and the operating mechanisms in the portal crane that are in operation and the operating mechanisms that are in a static state during the current operation cycle are determined.

[0044] For the operating mechanism in the operating state, two types of sensors are selected to respectively obtain position data of the operating mechanism, which are respectively recorded as first position data and second position data.

[0045] In this embodiment, the first absolute encoder and the incremental encoder are used to obtain the lifting height value of the lifting mechanism. Specifically, the first absolute encoder is installed on the output shaft of the lifting drum. abs and the absolute code h of the current position encoder collected g_abs , combined with the absolute code value h during calibration 0_abs The number of turns of the drum is calculated from the drum diameter d, and converted into the lifting height value h by calculation. abs , recorded as the first position data of the lifting mechanism, the calculation formula is as follows:

[0046]

[0047] Similarly, the incremental encoder is used to obtain the lifting height value of the lifting mechanism. Specifically, the incremental encoder is installed on the output shaft of the lifting motor, and its A and B phase signals are transmitted to the inverter; the counting processing module inside the inverter processes these signals and outputs the pulse number to the PLC, which is then collected by the edge controller to analyze the single-turn pulse number n of the encoder. in The number of pulses h output by the current position encoder g_in , combined with the encoder pulse number h during calibration 0_in Reduction ratio i with lifting gearbox hoist , calculate the number of revolutions of the lifting motor h in , and convert it into the lifting height value accordingly, recorded as the second position data of the lifting mechanism. The calculation formula is as follows:

[0048]

[0049] The second absolute encoder and inclinometer are used to obtain the working amplitude value of the luffing mechanism. Specifically, the second absolute encoder is installed at the pinion shaft end of the luffing mechanism to detect the rotation angle of the pinion. The angle zero point of the absolute encoder is when the gantry crane is at the maximum working amplitude. At this time, the straight-line distance from the center of the luffing pinion to the center of the hinge between the boom and the luffing rack is S. max When the pinion rotates, the absolute encoder detects the pinion rotation angle α, converts it into the running distance value of the luffing rack S = π × D × α / 360, and then combines the relevant dimensions of the boom system structure in the luffing mechanism of the portal crane to calculate the horizontal distance from the front end of the boom to the rotation center, that is, the working amplitude value, which is recorded as the first position data of the luffing mechanism. Similarly, the inclinometer is installed at the root of the boom to detect the angle between the boom and the plane, and combined with the dimensions of the luffing mechanism, the working amplitude value is calculated to calculate the second position data of the luffing mechanism.

[0050] The third absolute encoder and the QR code reader are used to obtain the rotation angle value of the rotating mechanism; specifically, the third absolute encoder is installed on the driven pinion shaft end outside the rotating large bearing, and the single turn bit number n of the encoder is ... abs And the absolute code of the current position encoder collected Calculate the rotation angle by combining the absolute encoder value during calibration and the transmission ratio i between the fixed main gear ring and the driven pinion rotate , calculate the rotation angle value and record it as the first position data of the rotating mechanism. The formula is:

[0051]

[0052] Similarly, QR code tags are installed at specific positions on the rotating platform. Each tag contains preset position information. When the rotating mechanism reaches the reading range of the QR code, the QR code reader scans and identifies these tags and converts the image data into machine-readable codes. These codes are then matched with the rotation angles stored in the system to obtain the rotation angles, which are recorded as the second position data of the rotating mechanism.

[0053] The fourth absolute encoder and RFID magnetic nail read-write sensor are used to collect the travel distance of the walking mechanism. Specifically, the fourth absolute encoder is installed on the driven wheel of the trolley. If the position value output by the absolute encoder is P, this value directly reflects the current position of the crane. The travel distance D of the crane is calculated by the formula: D = PP 初始, recorded as the first position data of the walking mechanism. Similarly, considering the harsh on-site environment, dust and other impurities fall on the track, which may adhere to the wheel surface after repeated rolling by the wheels, resulting in inaccurate measurement values ​​during operation. Therefore, a set of RFID magnetic nails is added, and the RFID magnetic nails are used as data code carriers. The RFID magnetic nail sensor is used to reversely calculate the position of the encoder according to the content of the induction magnetic nails, and calculate the walking distance, which is recorded as the second position data of the walking mechanism.

[0054] For the operating mechanism in a stationary state, the above two types of sensors can be used to obtain the position data of the operating mechanism respectively, which are recorded as the first position data and the second position data respectively. However, considering that in a stationary state, the operating parameter data collected by the sensor in real time are mostly 0, it is not meaningful to perform position calculation. In order to avoid wasting computing resources, the final position at the previous moment in the stationary state can be selected as the first position data and the second position data instead of using two types of sensors to obtain the position data of the operating mechanism respectively.

[0055] S2. Collect real-time environmental data, use a neural network model to perform error compensation on the acquired position data of the operating mechanism, and obtain the position data after error compensation.

[0056] Since environmental data has an impact on the position data of the operating mechanism collected by sensors, environmental sensors installed on the gantry crane are used to collect real-time environmental data, and the environmental data include: temperature, humidity, mechanical vibration, etc.; based on the real-time acquisition of environmental data, a neural network model is constructed to intelligently adjust the encoder reading according to the data of environmental factors to eliminate errors caused by temperature drift, humidity influence, mechanical vibration, etc. The specific error compensation steps are as follows: the real-time environmental data is integrated with the first position data and the second position data respectively, and first integrated data including real-time environmental data and the first position data, and second integrated data including real-time environmental data and the second position data are generated accordingly; in order to better perform error compensation, data preprocessing such as data cleaning and data compensation can be performed during the integration process.

[0057] The integrated data, i.e., the first integrated data and the second integrated data, are respectively input into the first error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism to obtain the first position data after error compensation and the second position data after error compensation. Wherein, the first error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated through the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data. For example, the first integrated data containing the first position data and environmental data of the lifting mechanism is input into the first error compensation model matching the first absolute encoder and the lifting mechanism to obtain the first position data after the lifting mechanism error compensation; the second integrated data containing the second position data and environmental data of the lifting mechanism is input into the first error compensation model matching the incremental encoder and the lifting mechanism to obtain the second position data after the lifting mechanism error compensation; wherein, the first error compensation model matching the first absolute encoder and the lifting mechanism is generated through the historical environmental data, the historical position data of the lifting mechanism obtained by the first absolute encoder, and the real position data.

[0058] S3. Perform position calibration using the error-compensated position data obtained based on multi-sensor acquisition to generate position data based on multi-type sensor calibration.

[0059] Specifically, the first position data after error compensation is calibrated using the second position data after error compensation, and the third position data is generated by calibration; wherein each type of operating mechanism is matched with a position calibration method suitable for itself.

[0060] For example: take the first position data after a specific type of error compensation as the calibration object, use the second position data after the corresponding type of error compensation for calibration, adjust the first position data according to the second position data, generate the third position data, thereby realizing the calibration of the absolute encoder data.

[0061] In addition to direct comparison calibration, complex calibration can also be used. Taking the calibration of the lifting mechanism as an example, the reading of the incremental encoder can be used as the independent variable, and the parameters of the linear model can be adjusted through the recursive least squares method (RLS) and the Jacobian matrix to minimize the linear model prediction value. The Jacobian matrix provides information on the sensitivity of the model parameters to input changes, while the RLS algorithm minimizes the prediction error by recursively updating the parameters. Through the combination of the Jacobian matrix and the recursive least squares method, real-time data calibration can be achieved, effectively eliminating system errors and random noise. Or taking the calibration of the walking mechanism as an example, the real-time position of the trolley = (trolley encoder value - encoder value of the last magnetic nail checkpoint) × encoder code value coefficient + trolley position corresponding to the last checkpoint.

[0062] S4. Use GNSS positioning technology to obtain the position data of each operating mechanism in the gantry crane, and use the position data to further calibrate the position data based on multi-type sensor calibration to obtain the final position data.

[0063] Specifically, the GNSS positioning technology used in this embodiment uses the real-time kinematic (RTK) technology of the global navigation satellite system (GNSS) to obtain the positioning and attitude data of each operating mechanism in the portal crane. Figure 3 As shown, the RTK position is obtained by at least one base station with a known position and a mobile station. The base station is installed at a fixed location (usually on the roof of the engineering department office building) and includes a satellite receiver, a transmission unit and an antenna, which is used to provide real-time differential reference positioning data to improve the accuracy of the mobile station positioning data;

[0064] The mobile station is installed in the electrical room of the gantry crane and consists of a GNSS RTK positioning electrical equipment box installed in the electrical room, two GNSS RTK satellite antennas, two receiving units, a data transmission radio and antenna, and two coaxial cables. The mobile station outputs the position data of each mechanism of the gantry crane according to the accuracy and effectiveness of the satellite positioning data.

[0065] The GNSS RTK position acquisition steps include: using the receiving unit to capture the satellites to be measured selected according to a certain satellite cut-off angle, and tracking the movement of these satellites. After the tracked satellite signals (GNSS signals) are captured by the GNSS RTK antenna in real time, the pseudo-range and distance change rate from the receiving antenna to the satellite can be measured, and the satellite orbit parameters and other data can be demodulated. Then, the positioning calculation is performed according to the positioning solution method, and the latitude and longitude, altitude, speed, time, heading angle, pitch angle, roll angle and other information of the geographical location corresponding to each operating mechanism of the portal crane are calculated.

[0066] Specifically, according to the acquired GNSS RTK position, the absolute encoder reading (third position data) after sensor calibration is calibrated, and the final position of each major operating mechanism is obtained in combination with the size structure of the portal crane. The calibration process includes: selecting a fixed reference point on the mechanism, which will be used as the basis for subsequent distance calculation and encoder calibration, and converting the absolute position data acquired by GNSS RTK into three-dimensional coordinates relative to the selected reference point; using the converted relative coordinates, through spatial vector operations, calculate the straight-line distance between the current position of the operating mechanism in the portal crane and the reference point, and compare the calculated straight-line distance with the reading of the absolute encoder. If there is a deviation, the third position data needs to be calibrated to keep it consistent with the GNSS RTK data;

[0067] Alternatively, according to the dual-antenna direction finding method, the angle between the baseline direction and the true north direction (i.e., the azimuth / heading angle) and the angle between the baseline direction and the horizontal plane (i.e., the pitch angle) are used. The azimuth angle is solved and converted through coordinate calculation to obtain the rotation angle of the rotating mechanism. After the pitch angle coordinates are settled, the working amplitude of the variable amplitude mechanism is calculated in combination with the mechanical model of the gantry crane. The calculated value is compared with the reading of the absolute encoder. If there is a deviation, the third position data needs to be calibrated to keep it consistent with the GNSS RTK data.

[0068] The above methods of using GNSS positioning technology for further calibration and final position determination do not take into account the impact of environmental factors on GNSS signal data. Once the GNSS signal data is abnormal, it will cause GNSS positioning failure, making it impossible to perform real-time calibration. Therefore, the following solution is adopted:

[0069] Collecting GNSS signal data in real time, and inputting the real-time collected GNSS signal data and real-time environmental data into a GNSS positioning fault judgment module; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated using historical environmental data and GNSS signal data marked with faults or not;

[0070] GNSS signal data is collected in real time using GNSS positioning technology, and the real-time collected GNSS signal data and real-time environmental data are input into the GNSS positioning fault judgment module to obtain the result of whether the GNSS positioning fault occurs at the current moment; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is obtained by calculating the GNSS signal data in the above manner, and the third position is calibrated using the GNSS RTK position to obtain the final position of each type of operating mechanism;

[0071] If the GNSS positioning fails at the current moment, the GNSS RTK position received at the previous moment is used as the base point. This base point provides the accurate position of the gantry crane just before the failure occurs. Combined with the third position, the relative displacement and final position of each type of operating mechanism after the GNSS failure are calculated. For example: if the encoder based on the sensor calibration shows that the gantry crane moved 10 meters to the east and 5 meters to the south after the failure, then the actual position of the gantry crane can be determined based on the base point position and these relative displacement data.

[0072] In a specific embodiment, considering that portal cranes of different sizes are affected to different degrees by the same environmental data, the size of the portal crane is further considered to perform more accurate error compensation, thereby achieving more precise positioning. The method further includes:

[0073] Collecting dimension data of each type of operating mechanism in the portal crane, and integrating the dimension data together in the process of integrating the real-time environment data with the first position data and the second position data, to generate new integrated data; that is, new first integrated data including the real-time environment data, the dimension data of the corresponding type of operating mechanism, and the first position data of the corresponding type, or new second integrated data including the real-time environment data, the dimension data of the corresponding type of operating mechanism, and the second position data of the corresponding type;

[0074] The newly integrated data replaces the integrated data, that is, the new first integrated data and the new second integrated data,

[0075] The second error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism is input respectively to obtain the first position data after error compensation and the second position data after error compensation; the second error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical size data of the corresponding type of operating mechanism, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data.

[0076] For example, the first integrated data including the first position data, dimensional data and environmental data of the lifting mechanism is input into the second error compensation model matched with the first absolute encoder and the lifting mechanism to obtain the first position data after the lifting mechanism error compensation; the second integrated data including the second position data, dimensional data and environmental data of the lifting mechanism is input into the second error compensation model matched with the incremental encoder and the lifting mechanism to obtain the second position data after the lifting mechanism error compensation.

[0077] In a specific embodiment, in addition to considering GNSS positioning failures, it is also necessary to consider failure conditions in the process of positioning calibration using different types of sensors to avoid calibration errors caused by sensor failures. The method further includes: using the error-compensated second position data to calibrate the error-compensated first position data includes: the number of the second position data is multiple, and they are respectively collected and obtained by multiple sensors of the same type;

[0078] For each type of operating mechanism in the gantry crane, the error calculation is performed between the first position data and multiple second position data respectively, and the calculated errors are clustered to obtain the second position data corresponding to the error calculation result that is more than a preset distance away from the cluster center and identified as abnormal second position data, indicating that the sensor corresponding to the abnormal second position data may be faulty, and any second position data except the abnormal second position data is selected to calibrate the first position data to generate the third position data.

[0079] In a specific embodiment, in addition to considering the situation due to GNSS positioning failure, it is also necessary to consider the possible failure risk. When the possible failure risk is predicted, the GNSS signal is enhanced to further ensure the accuracy of positioning. The method also includes:

[0080] The real-time collected GNSS signal data and real-time environmental data are input into the GNSS positioning fault risk assessment judgment module to obtain an assessment result of whether the GNSS signal has a fault risk at the current moment; the GNSS positioning fault risk assessment module adopts a neural network model, which is trained and generated using historical environmental data and GNSS signal data marked with whether there is a fault risk.

[0081] If there is a risk of failure in acquiring the current GNSS signal, a high-gain antenna is selected to receive the GNSS signal at the next moment and the received GNSS signal is enhanced and filtered, otherwise no processing is performed.

[0082] In a specific embodiment, by comparing the difference between the acquired final position and the actual position, further adjusting the model to optimize the error compensation, and adjusting the corresponding compensation frequency to complete the error compensation more timely, the method further includes:

[0083] Statistics are collected on the final position and actual position located within a period of time including the current moment. If the ratio of the similarity between the final position and the real-time position is greater than the preset similarity number and the ratio of the similarity between the final position and the real-time position is not greater than the preset similarity number and is less than the preset ratio, the first error compensation model or the second error compensation model is optimized and trained, and the frequency of environmental data collection and the frequency of position data collection of the operating mechanism are increased accordingly.

[0084] In addition, in order to complete the error compensation more timely, the method further includes:

[0085] Analyze the environmental data acquired in real time to determine whether there is any sudden change in the environment acquired in real time within a period of time including the current moment, that is, whether there is any type of environmental data with a change rate greater than the preset change rate; if a sudden change occurs, adjust the environmental data collection frequency and the position data collection frequency of the operating mechanism accordingly.

[0086] like Figure 4 As shown, the embodiment of the present application discloses that the present application provides an automatic positioning system for a gantry crane based on multi-sensory fusion.

[0087] Among them, the hardware structure corresponding to the system is divided into perception layer, network layer and application layer, such as Figure 5As shown. The perception layer is the data source of the entire system, including various sensors (absolute encoder, incremental encoder, inclinometer, RFID, QR code reader, GNSS, etc.) to collect information on the operating status of each mechanism in the gantry crane, thereby ensuring that the system can obtain high-precision, real-time data. The network layer uses switches and edge controllers to transmit and distribute data, acting as a bridge between the perception layer and the application layer to ensure that data can flow efficiently and reliably between various system components. The industrial computer in the application layer processes and analyzes the transmitted data, executes data processing algorithms, and ultimately achieves precise positioning of each structure of the gantry crane.

[0088] The software structure corresponding to the system includes a data acquisition layer, a positioning algorithm layer and an output result layer. Specifically, it includes: a portal crane initial position acquisition module 101, which is used to use two types of sensors to respectively acquire the position data of each type of operating mechanism in the portal crane, and record them as first position data and second position data respectively; the types of the operating mechanisms include: lifting mechanism, luffing mechanism, rotating mechanism and walking mechanism.

[0089] The portal crane initial position compensation module 102 is used to collect real-time environmental data using environmental sensors installed on the portal crane, integrate the real-time environmental data with the first position data and the second position data respectively, input the integrated data into the first error compensation model that matches the corresponding type of sensor and the corresponding type of operating mechanism respectively, and obtain the first position data after error compensation and the second position data after error compensation; perform position calibration on the first position data after error compensation using the second position data after error compensation, and calibrate to generate third position data; the first error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data;

[0090] The final position acquisition module 103 of the portal crane is used to collect GNSS signal data in real time using GNSS positioning technology, input the real-time collected GNSS signal data and real-time environmental data into the GNSS positioning fault judgment module, and obtain the result of whether the GNSS positioning fault occurs at the current moment; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated using historical environmental data and GNSS signal data marked with faults or not; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is calculated using the GNSS signal data, and the third position is calibrated using the GNSS RTK position to obtain the final position of each type of operating mechanism; if the GNSS positioning fails at the current moment, the relative displacement and final position of each type of operating mechanism after the GNSS failure is calculated based on the GNSS RTK position received at the previous moment as the base point and combined with the third position.

[0091] The system also includes: a portal crane final position optimization module 104, which is used to input the real-time collected GNSS signal data and real-time environmental data into the GNSS positioning fault risk assessment and judgment module to obtain an assessment result of whether the GNSS signal at the current moment has a fault risk; the GNSS positioning fault risk assessment module adopts a neural network model, and is trained and generated using historical environmental data and GNSS signal data marked with whether there is a fault risk; if the current GNSS signal is found to have a fault risk, a high-gain antenna is selected at the next moment to receive the GNSS signal and the received GNSS signal is enhanced and filtered; otherwise, no processing is performed.

[0092] The portal crane initial position acquisition optimization module 105 is used to count the final position and the actual position located within a period of time including the current moment. If the ratio of the similarity between the final position and the real-time position is greater than the number of preset similarities and the ratio of the similarity between the final position and the real-time position is not greater than the number of preset similarities and is less than the preset ratio, the first error compensation model is optimized and trained, and the environmental data collection frequency and the position data collection frequency of the operating mechanism are increased accordingly.

[0093] The portal crane initial position acquisition module 101 in the system is also used to obtain the current operation cycle of the portal crane, and determine the operating mechanisms in the portal crane that are in operation and the operating mechanisms that are in a stationary state within the current operation cycle; wherein, for the operating mechanisms that are in operation, two types of sensors are selected to respectively obtain the position data of the operating mechanisms, which are recorded as the first position data and the second position data respectively; for the operating mechanisms that are in a stationary state, the final position at the previous moment in the stationary state is selected to be used as the first position data and the second position data instead of using two types of sensors to respectively obtain the position data of the operating mechanisms.

[0094] In addition, the gantry crane initial position compensation module 102 in the system is also used to collect dimensional data of each type of operating mechanism in the gantry crane, and in the process of integrating the real-time environmental data with the first position data and the second position data, the dimensional data is integrated together to generate new integrated data; the integrated data is replaced by the new integrated data, and the data are input into the second error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism, respectively, to obtain the error-compensated first position data and the error-compensated second position data; the second error compensation model is a neural network model, which has multiple types, and each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training through the collected historical environmental data, the historical dimensional data of the corresponding type of operating mechanism, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data.

[0095] The embodiment of the present application also discloses a computer-readable storage medium.

[0096] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed such as the above-mentioned automatic positioning method of a gantry crane based on multi-sensory fusion. The computer-readable storage medium includes, for example: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0097] The embodiment of the present application also discloses a computer device.

[0098] Specifically, the computer device includes a memory and a processor, and the memory stores a computer program that can be loaded by the processor and execute the above-mentioned automatic positioning method of the gantry crane based on multi-sensory fusion.

[0099] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. An automatic positioning method for a portal crane based on multi-sensor fusion, characterized in that: include: For each type of operating mechanism in the portal crane, two types of sensors are used to obtain position data of the operating mechanism, which are recorded as first position data and second position data respectively; The types of operating mechanisms include: lifting mechanism, luffing mechanism, rotating mechanism and traveling mechanism; Using an environmental sensor installed on the portal crane to collect real-time environmental data, and integrating the real-time environmental data with the first position data and the second position data to obtain integrated data; In the process of integrating the real-time environment data with the first position data and the second position data, the dimension data is integrated to generate new integrated data, and the new integrated data replaces the integrated data, and is input into the second error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism, respectively, to obtain the error-compensated first position data and the error-compensated second position data to replace the input of the integrated data into the first error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism, respectively, to obtain the error-compensated first position data and the error-compensated second position data; The first error compensation model is a neural network model, which has multiple types. Each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data; The second error compensation model is a neural network model, which has multiple types. Each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated through training through the collected historical environmental data, the historical dimension data of the corresponding type of operating mechanism, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data; The first position data after error compensation is calibrated by using the second position data after error compensation, and the third position data is generated by calibration; the GNSS positioning technology is used to collect GNSS signal data in real time, and the real-time collected GNSS signal data and real-time environmental data are input into the GNSS positioning fault judgment module to obtain whether the GNSS positioning fault occurs at the current moment; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated by using historical environmental data and GNSS signal data marked with faults; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is calculated by using the GNSS signal data, and the third position data is calibrated by using the GNSS RTK position to obtain the final position of each type of operating mechanism; if the GNSS positioning fails at the current moment, the relative displacement and final position of each type of operating mechanism after the GNSS failure are calculated based on the GNSS RTK position received at the previous moment as a base point and combined with the third position data.

2. The automatic positioning method for a portal crane based on multi-sensing fusion according to claim 1 is characterized in that: The performing position calibration on the error-compensated first position data using the error-compensated second position data comprises: There are multiple second position data, which are collected and acquired by multiple sensors of the same type respectively; For each type of operating mechanism in the gantry crane, the error calculation is performed between the first position data and multiple second position data respectively, and the calculated errors are clustered to obtain the second position data corresponding to the error calculation result that is more than a preset distance away from the cluster center. The obtained second position data is identified as abnormal second position data, and any second position data except the abnormal second position data is selected to calibrate the first position data to generate third position data.

3. The automatic positioning method for a portal crane based on multi-sensing fusion according to claim 1 is characterized in that: Also includes: Input the real-time collected GNSS signal data and real-time environmental data into the GNSS positioning fault risk assessment and judgment module to obtain an assessment result of whether the GNSS signal has a fault risk at the current moment; The GNSS positioning fault risk assessment and judgment module adopts a neural network model, which is trained and generated using historical environmental data and GNSS signal data marked with whether there is a fault risk or not; If there is a risk of failure in acquiring the current GNSS signal, a high-gain antenna is selected at the next moment to receive the GNSS signal and the received GNSS signal is enhanced and filtered; Otherwise no processing is performed.

4. The automatic positioning method for a portal crane based on multi-sensing fusion according to claim 1 is characterized in that: Also includes: Statistics are taken of the final position and actual position located within a period of time including the current moment. If the ratio of the similarity between the final position and the real-time position is greater than the preset similarity number and the ratio of the similarity between the final position and the real-time position is not greater than the preset similarity number and is less than the preset ratio, then the first error compensation model is optimized and trained, and the frequency of environmental data collection and the frequency of position data collection of the operating mechanism are increased accordingly.

5. The automatic positioning method for a portal crane based on multi-sensing fusion according to claim 1 is characterized in that: Also includes: The current operation cycle of the gantry crane is obtained, and the operating mechanisms in the gantry crane that are in operation and those that are in a stationary state within the current operation cycle are determined; for the operating mechanisms that are in operation, two types of sensors are selected to respectively obtain position data of the operating mechanisms, which are recorded as first position data and second position data respectively; for the operating mechanisms that are in a stationary state, the final position at the previous moment in the stationary state is selected to be used as the first position data and the second position data instead of using two types of sensors to respectively obtain the position data of the operating mechanisms.

6. The automatic positioning method for a portal crane based on multi-sensing fusion according to claim 1 is characterized in that: The use of two types of sensors to respectively obtain the position data of the operating mechanism includes: The first absolute encoder and incremental encoder are used to obtain the lifting height value of the lifting mechanism; the second absolute encoder and inclinometer are used to obtain the working amplitude value of the variable amplitude mechanism; the third absolute encoder and QR code reader are used to obtain the rotation angle value of the rotating mechanism; the fourth absolute encoder and RFID magnetic nail read-write sensor are used to obtain the walking distance of the walking mechanism.

7. An automatic positioning system for a portal crane based on multi-sensor fusion, characterized in that: include: The portal crane initial position acquisition module is used for acquiring the position data of each type of operating mechanism in the portal crane by using two types of sensors, which are recorded as first position data and second position data respectively; The types of operating mechanisms include: lifting mechanism, luffing mechanism, rotating mechanism and traveling mechanism; The portal crane initial position compensation module is used to collect real-time environmental data using an environmental sensor installed on the portal crane, and integrate the real-time environmental data with the first position data and the second position data to obtain integrated data; In the process of integrating the real-time environment data with the first position data and the second position data, the dimension data is integrated to generate new integrated data, and the new integrated data replaces the integrated data, and is input into the second error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism, respectively, to obtain the error-compensated first position data and the error-compensated second position data to replace the input of the integrated data into the first error compensation model matching the corresponding type of sensor and the corresponding type of operating mechanism, respectively, to obtain the error-compensated first position data and the error-compensated second position data; The first error compensation model is a neural network model, which has multiple types. Each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated by training with the collected historical environmental data, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data; The second error compensation model is a neural network model, which has multiple types. Each type of error compensation model matches a type of sensor and a type of operating mechanism, and is generated through training through the collected historical environmental data, the historical dimension data of the corresponding type of operating mechanism, the historical position data of the corresponding type of operating mechanism collected by the corresponding type of sensor, and the historical actual position data; Using the error-compensated second position data to calibrate the first position data after error compensation, so as to generate third position data; The final position acquisition module of the portal crane is used to collect GNSS signal data in real time by adopting GNSS positioning technology, input the real-time collected GNSS signal data and real-time environmental data into the GNSS positioning fault judgment module, and obtain the result of whether the GNSS positioning fault occurs at the current moment; the GNSS positioning fault judgment module adopts a neural network model, and is trained and generated by using historical environmental data and GNSS signal data marked with faults or not; if the GNSS positioning does not fail at the current moment, the GNSS RTK position is calculated by using the GNSS signal data, and the third position data is calibrated by using the GNSS RTK position to obtain the final position of each type of operating mechanism; if the GNSS positioning fails at the current moment, the relative displacement and final position of each type of operating mechanism after the GNSS failure is calculated based on the GNSS RTK position received at the previous moment as the base point and combined with the third position data.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A computer device, characterized in that: The computer device comprises a memory, a processor and a program stored and executable on the memory, and the program implements the steps of the method according to any one of claims 1 to 6 when executed by the processor.

Images