Ship unloader anti-adhesion cooperative system for viscous bulk materials
Through multimodal detection and dynamic strategy optimization, integrated near-infrared spectroscopy and capacitive viscosity detection, and combined with deep learning algorithms, the problems of low detection accuracy and poor adaptability of traditional ship unloader anti-adhesion systems have been solved, and the unloading efficiency of viscous bulk materials and the reliability of equipment have been improved.
Patent Information
- Application Number
- CN202511088154.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-05
- Publication Date
- 2025-09-12
AI Technical Summary
The anti-adhesion system of traditional ship unloaders has low detection accuracy, control lag, and poor adaptability, resulting in low unloading efficiency of sticky bulk materials and large equipment losses, making it difficult to meet the efficient and intelligent operation needs of modern ports.
It adopts multimodal detection, dynamic strategy optimization and environmental adaptive design, integrates multi-source sensing technologies such as near-infrared spectroscopy detection, capacitive viscosity detection, temperature and humidity sensing, and combines deep learning algorithms with reinforcement learning control strategies to achieve real-time quantification of material viscosity parameters and dynamic adjustment of anti-adhesion actions.
It significantly improves the unloading efficiency and equipment reliability of viscous bulk materials, realizes the simultaneous quantification of multi-dimensional parameters, dynamically adapts to changes in material properties, reduces equipment losses, and improves the environmental adaptability and energy efficiency of the system.
Smart Images
Figure CN120622153A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of port machinery automation, and in particular to an anti-adhesion coordination system for ship unloaders targeting at viscous bulk materials. Background Art
[0002] During bulk terminal operations, sticky bulk materials such as coal, iron ore, and mineral fines are prone to sticking to the inner walls of the grab bucket due to their high humidity and fine particle size, resulting in reduced unloading efficiency and increased equipment wear. Traditional ship unloader anti-sticking systems suffer from numerous technical deficiencies: Their detection methods are limited, relying solely on capacitive or infrared sensors. These sensors fail to fully quantify multi-dimensional parameters such as viscosity, humidity, and particle size, resulting in incomplete data. Anti-sticking measures are passive, relying solely on fixed-frequency vibration or constant temperature control, unable to dynamically adapt to changes in material properties and prone to failure when sticky materials vary. Control strategies lag, lacking real-time data fusion and optimization capabilities, with strategy adjustment times exceeding 300ms, making them difficult to respond to sudden sticking events. Environmental adaptability is weak, failing to consider the secondary effects of ambient temperature and humidity on material viscosity. Consequently, anti-sticking effectiveness is significantly diminished under extreme operating conditions (such as high humidity and low temperatures). Furthermore, traditional systems operate independently, lacking coordinated control mechanisms. This results in high energy consumption and significant equipment wear, making them unable to meet the demands of modern ports for efficient, low-cost, and intelligent operations.
[0003] To address these issues, this invention achieves full automation of the ship unloader's anti-adhesion system through multimodal detection, dynamic strategy optimization, and environmentally adaptive design. The system integrates multi-source sensing technologies, including near-infrared spectroscopy, capacitive viscosity detection, and temperature and humidity sensing. Combined with deep learning algorithms and reinforcement learning control strategies, it can quantify material viscosity parameters in real time and dynamically adjust anti-adhesion actions. This significantly improves the unloading efficiency and equipment reliability of viscous bulk materials, making it suitable for intelligent operations involving various types of materials in ports, mines, and other locations. Summary of the Invention
[0004] In response to the problems in the prior art, the present invention provides a ship unloader anti-adhesion collaborative system for sticky bulk materials, including a viscosity detection module, which is installed on the inside of the ship unloader grab and is used to detect the viscosity parameters of the material in real time. It contains a near-infrared spectrum detection submodule and a capacitive viscosity detection submodule, and realizes the real-time collection of viscosity parameters through the fusion of non-contact and contact sensing; an anti-adhesion execution module, which is connected to the viscosity detection module via CAN bus data and dynamically adjusts the anti-adhesion strategy based on the detection results. It contains a high-frequency vibration submodule and a coating temperature control submodule to perform dynamic anti-adhesion actions; a collaborative control module, which is integrated into the ship unloader electrical control cabinet and is connected via EtherCAT The bus is connected to the anti-adhesion execution module, which includes a multi-source data fusion submodule and a strategy optimization submodule. The environmental perception module is installed on the outside of the cantilever and connected to the collaborative control module via 5G wireless communication. It includes a temperature and humidity detection submodule and a material humidity compensation submodule. The data acquisition and processing module is integrated into the collaborative control module and includes an edge computing unit and a preprocessing algorithm library to achieve raw data cleaning and feature extraction. The human-computer interaction module is connected to the collaborative control module via the TCP / IP protocol for strategy verification and parameter correction. The energy management module has a master-slave architecture and includes a main power supply and a supercapacitor backup power supply, which supplies power to each module through redundant power supply lines. Each module achieves closed-loop control through real-time data streams, and the total system response time does not exceed 120ms. The viscosity detection module adopts an improved partial least squares regression algorithm, and the viscosity detection accuracy reaches ±2%FS.
[0005] Furthermore, the near-infrared spectrum detection submodule includes two miniature near-infrared spectrometers with a wavelength range of 900-2500nm and a spectral resolution better than 2nm. They are installed in symmetrical positions on the inner wall of the grab bucket. The optical path length is dynamically adjusted within the range of 5-15mm through a precision screw mechanism with an adjustment accuracy of 0.1mm; the light source adopts a wide-band luminous body with a power density of ≥5W / cm² and a color temperature range of 2800-3200K, and is equipped with an optical feedback stabilization circuit to ensure spectral stability; the detector adopts a linear array structure with a responsivity of ≥0.7A / W and a dark current of ≤1.2nA. The signal is sampled by a 24-bit analog-to-digital converter with a sampling rate of ≥1kHz.
[0006] Furthermore, the capacitive viscosity detection submodule uses a parallel plate capacitance sensor with an electrode area of 20mm×15mm, an electrode spacing of 0.5mm, and a polytetrafluoroethylene film coated on the electrode surface to prevent material adhesion; the detection circuit has an operating frequency range of 1kHz-10kHz, a sensitivity of 0.1pF / Pa·s, a built-in temperature compensation network, a compensation range of -20℃~80℃, and a compensation accuracy of ±0.3%FS; the output signal is differentially amplified and low-pass filtered, the filter cutoff frequency is 500Hz, and the signal-to-noise ratio is ≥60dB.
[0007] Furthermore, the vibration source of the high-frequency vibration submodule adopts a piezoelectric ceramic laminate structure with an external dimension of 50mm×10mm, a resonance frequency of 8kHz±200Hz, and a vibration amplitude that can be steplessly adjusted within the range of 0.1-1.5mm through pulse width modulation; the drive circuit adopts LLC resonant topology, with an input voltage range of 24V±10%, an output power density ≥5W / cm³, and a conversion efficiency ≥90%; the vibration disk is connected to the side wall of the grab bucket through an elastic damping member with a damping coefficient of 0.2-0.5, a vibration transmission rate ≤15%, and a protection level of IP68.
[0008] Furthermore, the inner wall of the coating temperature control submodule grab is coated with a nano-silica hydrophobic coating with a coating thickness of 50-100 μm, a water contact angle ≥160°, and a rolling angle ≤5°; the bottom is integrated with a semiconductor refrigeration element with a cooling capacity ≥50W, a temperature control range of -10°C to 60°C, a control accuracy of ±0.5°C, and is driven by a proportional-integral-differential algorithm with a response time of ≤30s; a temperature sensor array is set on the coating surface with a sensor spacing of 20 mm and a measurement accuracy of ±0.2°C. The data is used for feedback control of the coating temperature.
[0009] Furthermore, the input data of the multi-source data fusion submodule includes near-infrared spectral data, capacitance value, and ambient temperature and humidity; the fusion algorithm adopts an improved deep belief network, which contains 3 hidden layers with 512, 256, and 128 nodes respectively, and the activation function adopts a rectified linear unit; the training data set contains ≥8000 groups of viscous bulk material samples, covering a viscosity range of 10-1000mPa·s and a humidity range of 5-90%RH; the fusion output viscosity prediction value error is ≤2%FS, and the humidity compensation coefficient ΔRH is ≤1%.
[0010] Furthermore, the strategy optimization submodule is based on a deep Q-network algorithm. The state space includes real-time viscosity value, grab load, ambient temperature and humidity, and the state representation is normalized. The action space defines vibration frequency, coating temperature, and grab opening. The action selection adopts a greedy strategy with an initial value of 0.9 and a decay rate of 0.995. The reward function is designed with an anti-adhesion efficiency weight of 0.6, an energy consumption weight of 0.3, and an equipment loss weight of 0.1, where the anti-adhesion efficiency is evaluated by the amount of residual material in the grab.
[0011] Furthermore, the temperature and humidity detection submodule adopts a digital temperature and humidity sensor with a temperature measurement range of -40℃~125℃, an accuracy of ±0.2℃, and a humidity measurement range of 0-100%RH, with an accuracy of ±1.5%RH; the sensor is encapsulated in a waterproof and breathable membrane, the membrane material is polytetrafluoroethylene, the pore size is 0.2μm, and it is connected to the electrical cabinet through an M12 aviation plug. The protection level is IP67, and the response time based on temperature conditions is ≤5s, and the response time based on humidity conditions is ≤8s.
[0012] Furthermore, the material humidity compensation submodule has a built-in humidity-viscosity correction model, the model form is a cubic polynomial regression, the input is the ambient humidity and the detected viscosity, and the output is the corrected viscosity value; the model training data correction error is ≤1.5%FS, the model update frequency is every 1000 sets of new data or every 24 hours, and the update method adopts the recursive least squares method.
[0013] Furthermore, the edge computing unit adopts an embedded computing platform, equipped with an 8-core processor, a graphics processing unit, and a memory ≥16GB; deploys a preprocessing algorithm library, including Savitsky-Golay filtering, baseline correction polynomial fitting order ≤5, and first-order derivative spectral processing; data storage uses a solid-state hard drive with a capacity ≥1TB, supports H.265 video encoding compression, a compression ratio ≥50:1, and a processing delay ≤20ms.
[0014] The beneficial effects of the present invention are as follows: through multimodal detection, dynamic strategy optimization and environmental adaptive design, the problems of low detection accuracy, control lag and poor adaptability in traditional technologies are solved, and the efficiency and reliability of viscous bulk material unloading operations are significantly improved. Multimodal detection realizes synchronous quantification of viscosity, humidity and particle size. The polytetrafluoroethylene coating prevents material adhesion and ensures the long-term stability of capacitive detection; high-frequency vibration and low-temperature coating work together to dynamically adjust the vibration amplitude and coating temperature to adapt to different viscous materials, monitor the ambient temperature and humidity in real time, and eliminate the interference of humidity on viscosity detection by correcting the model; dynamic compensation makes the calculated viscosity value closer to the true value, thereby improving the accuracy of the strategy. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] The present invention will be further described below with reference to the accompanying drawings and examples.
[0016] Figure 1 This is a system architecture diagram of the present invention.
[0017] Figure 2 The figure is a flowchart of the system working process of the present invention.
[0018] In the figure: 100, viscosity detection module; 110, near-infrared spectrum detection submodule; 120, capacitive viscosity detection submodule; 200, anti-adhesion execution module; 210, high-frequency vibration submodule; 220, coating temperature control submodule; 300, collaborative control module; 310, multi-source data fusion submodule; 320, strategy optimization submodule; 400, environmental perception module; 410, temperature and humidity detection submodule; 420, material humidity compensation submodule; 500, data acquisition and processing module; 510, edge computing unit; 520, preprocessing algorithm library; 600, human-computer interaction module; 700, energy management module; 710, main power supply; 720, supercapacitor backup power supply. DETAILED DESCRIPTION
[0019] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.
[0020] like Figure 1-Figure 2 As shown, the present invention provides a ship unloader anti-adhesion collaborative system for sticky bulk materials, including a viscosity detection module 100, which is installed on the inside of the ship unloader grab and is used to detect the material viscosity parameters in real time, including a near-infrared spectrum detection submodule 110 and a capacitive viscosity detection submodule 120; an anti-adhesion execution module 200, which is connected to the viscosity detection module 100 via CAN bus data and dynamically adjusts the anti-adhesion strategy based on the detection results, including a high-frequency vibration submodule 210 and a coating temperature control submodule 220; a collaborative control module 300, which is integrated into the ship unloader electrical control cabinet, is connected to the anti-adhesion execution module 200 via the EtherCAT bus, and includes a multi-source data fusion submodule 310 and strategy optimization submodule 320; environmental perception module 400, installed on the outside of the cantilever, connected to the collaborative control module 300 via 5G wireless communication, including temperature and humidity detection submodule 410 and material humidity compensation submodule 420; data acquisition and processing module 500, integrated into the collaborative control module 300, including edge computing unit 510 and preprocessing algorithm library 520; human-computer interaction module 600, connected to the collaborative control module 300 via TCP / IP protocol for strategy verification and parameter correction; energy management module 700, with a master-slave architecture, including a main power supply 710 and a supercapacitor backup power supply 720, which supplies power to each module through redundant power supply lines; Each module implements closed-loop control through real-time data flow, and the total response time of the system does not exceed 120ms. The viscosity detection module 100 adopts an improved partial least squares regression algorithm, and the viscosity detection accuracy reaches ±2%FS.
[0021] The near-infrared spectrum detection submodule 110 includes two miniature near-infrared spectrometers with a wavelength range of 900-2500nm and a spectral resolution better than 2nm. They are installed in symmetrical positions on the inner wall of the grab bucket, and the optical path length is dynamically adjusted within the range of 5-15mm via a precision screw mechanism. The light source adopts a broadband luminous body with a power density of ≥5W / cm² and a color temperature range of 2800-3200K. It is equipped with an optical feedback stabilization circuit to ensure spectral stability. The detector adopts a linear array structure with a responsivity of ≥0.7A / W and a dark current of ≤1.2nA. The signal is sampled by a 24-bit analog-to-digital converter with a sampling rate of ≥1kHz.
[0022] Two miniature near-infrared spectrometers were symmetrically mounted on the inner wall of the grab bucket, 1.5 m from the centerline of the grab bucket bottom, with an optical axis spacing of 120 mm. The spectrometers were fixed via an L-shaped aluminum alloy bracket, which was connected to the inner wall of the grab bucket using M10 × 30 mm stainless steel bolts with a bolt preload torque of 20 N·m. A rubber shock-absorbing pad was installed at the bottom of the bracket. A polytetrafluoroethylene calibration plate was inserted, and the lead screw mechanism was adjusted to ensure an optical path length error of ≤0.1 mm. The wavelength ranges of the two spectrometers were matched using software, with a wavelength repeatability of ≤0.2 nm. The spectrometer entrance was shielded, and the measured dark current value was ≤1.2 nA, with automatic compensation when the threshold was exceeded.
[0023] The capacitive viscosity detection submodule 120 uses a parallel plate capacitance sensor with a plate area of 20 mm × 15 mm and a plate spacing of 0.5 mm. The plate surface is coated with a polytetrafluoroethylene film to prevent material adhesion. The detection circuit has an operating frequency range of 1 kHz to 10 kHz, a sensitivity of 0.1 pF / Pa·s, and a built-in temperature compensation network with a compensation range of -20°C to 80°C and a compensation accuracy of ±0.3% FS. The output signal is differentially amplified and low-pass filtered with a filter cutoff frequency of 500 Hz and a signal-to-noise ratio of ≥60 dB.
[0024] The parallel plate capacitance sensor is embedded in the bottom of the grab bucket, 200 mm from the edge of the grab bucket and flush with the material contact surface. The sensor plate is bonded to the grab bucket base with epoxy resin glue. The glue layer has a thickness of 0.5 mm and is cured at 80°C for 2 hours. The surface of the plate is covered with a polytetrafluoroethylene film with a thickness of 15 μm and a roughness of Ra 0.3 μm. The film is bonded by hot pressing at a temperature of 200°C and a pressure of 0.5 MPa. In the absence of material, the no-load capacitance value is measured and the circuit zero point is adjusted to 0 pF ± 0.05 pF. A standard viscosity liquid (viscosity value 500 mPa·s) is injected, and the capacitance change is measured by 5 pF. The sensitivity is calculated to be 0.1 pF / (Pa·s). The compensation coefficient is verified at 20°C, 40°C, and 60°C, with an error of ≤ 0.3% FS.
[0025] The vibration source of the high-frequency vibration submodule 210 adopts a piezoelectric ceramic laminate structure with an external dimension of 50mm×10mm, a resonance frequency of 8kHz±200Hz, and a vibration amplitude that can be steplessly adjusted within the range of 0.1-1.5mm through pulse width modulation. The drive circuit adopts an LLC resonant topology with an input voltage range of 24V±10%, an output power density ≥5W / cm³, and a conversion efficiency ≥90%. The vibration disk is connected to the side wall of the grab bucket through an elastic damping member with a damping coefficient of 0.2-0.5, a vibration transmission rate ≤15%, and a protection level of IP68.
[0026] The piezoelectric ceramic vibration disk is installed on the side wall of the grab bucket, 300 mm from the centerline of the grab bucket bottom, and the vibration direction is perpendicular to the direction of grab bucket movement. The vibration disk is connected to the grab bucket side wall via an elastic damping member (made of silicone rubber, Shore hardness 70A), with a damping coefficient of 0.3. The connecting bolts are M8×25 mm stainless steel bolts with a preload torque of 15 N·m. The vibration disk frequency response is scanned using an impedance analyzer, and the resonant frequency is determined to be 8kHz±200Hz. A PWM signal (duty cycle 50%) is input, and the vibration amplitude is measured to be 1.0 mm with an error of ≤0.01 mm. Vibration is initiated in coal material, and the amount of residual material is measured to be reduced by 80%, with a vibration efficiency of ≥90%.
[0027] The inner wall of the coating temperature control submodule 220 grab bucket is coated with a nano-silica hydrophobic coating with a coating thickness of 50-100 μm, a water contact angle ≥160°, and a rolling angle ≤5°; the bottom is integrated with a semiconductor refrigeration element with a cooling capacity ≥50W, a temperature control range of -10°C to 60°C, a control accuracy of ±0.5°C, and is driven by a proportional-integral-differential algorithm with a response time of ≤30s; a temperature sensor array is set on the coating surface with a sensor spacing of 20 mm and a measurement accuracy of ±0.2°C. The data is used for feedback control of the coating temperature.
[0028] The nano-silica hydrophobic coating covers the entire working surface of the grab's inner wall, with a thickness of 80μm. The semiconductor cooling chip is embedded in the bottom of the grab, and the contact surface with the coating has a flatness of ≤0.1mm. The coating is deposited by a plasma spraying process with a spraying distance of 100mm and a spraying angle of 90°. The cooling chip is bonded to the grab base with thermal grease (thermal conductivity coefficient 1.5W / (m·K)), and the grease layer is 0.2mm thick.
[0029] The input data of the multi-source data fusion submodule 310 includes near-infrared spectral data, capacitance value, and ambient temperature and humidity. The fusion algorithm adopts an improved deep belief network, which includes three hidden layers with 512, 256, and 128 nodes respectively, and the activation function adopts a rectified linear unit. The training data set contains ≥8000 groups of viscous bulk material samples, covering a viscosity range of 10-1000mPa·s and a humidity range of 5-90%RH. The error of the fusion output viscosity prediction value is ≤2%FS, and the humidity compensation coefficient ΔRH is ≤1%.
[0030] The industrial computer is deployed in an electrical control cabinet, 1.8 meters above the ground, and connected to the anti-adhesion execution unit via the EtherCAT bus. The industrial computer is equipped with an octa-core processor (main frequency 2.5GHz), 32GB of DDR4 memory, and a 512GB SSD. The EtherCAT bus coupler is set to a synchronization period of 250μs and a data refresh rate of 1kHz. The timestamps of the near-infrared spectrum, capacitance value, and temperature and humidity data are matched with a delay of ≤10ms. 8000 sets of sample data are input, and the DBN network is trained to a loss value of ≤0.05 and a validation set accuracy of ≥95%. The input viscosity value is 500mPa·s and the humidity is 70%RH. The predicted value is 495mPa·s with an error of ≤2%FS.
[0031] The strategy optimization submodule 320 is based on a deep Q-network algorithm. The state space includes real-time viscosity value, grab load, ambient temperature and humidity, and the state representation is normalized. The action space defines vibration frequency, coating temperature, and grab opening. The action selection adopts a greedy strategy with an initial value of 0.9 and a decay rate of 0.995. The reward function is designed with an anti-adhesion efficiency weight of 0.6, an energy consumption weight of 0.3, and an equipment loss weight of 0.1, where the anti-adhesion efficiency is evaluated by the amount of residual material in the grab.
[0032] The DQN algorithm was deployed on an industrial computer and connected to the human-computer interaction unit via TCP / IP. The system set the viscosity value (10-1000 mPa·s), grab load (0-10 kN), ambient temperature and humidity; vibration frequency (0-8 kHz), coating temperature (-10°C to 60°C), grab opening angle (0-90°); anti-adhesion efficiency (weight 0.6), energy consumption (weight 0.3), and equipment loss (weight 0.1). The system ran 5000 simulated ship unloading cycles until the reward value converged to R ≥ 0.8.
[0033] The temperature and humidity detection submodule 410 uses a digital temperature and humidity sensor with a temperature measurement range of -40°C to 125°C, an accuracy of ±0.2°C, and a humidity measurement range of 0-100%RH, with an accuracy of ±1.5%RH. The sensor is encapsulated in a waterproof and breathable membrane made of polytetrafluoroethylene with a pore size of 0.2μm. It is connected to the electrical cabinet through an M12 aviation plug, with a protection level of IP67. The response time based on temperature conditions is ≤5s, and the response time based on humidity conditions is ≤8s.
[0034] The material humidity compensation submodule 420 has a built-in humidity-viscosity correction model in the form of a cubic polynomial regression, with the input being the ambient humidity and the detected viscosity, and the output being the corrected viscosity value; the correction error of the model training data is ≤1.5%FS, the model is updated every 1,000 sets of new data or every 24 hours, and the update method adopts the recursive least squares method.
[0035] The digital temperature and humidity sensor is installed on the outside of the cantilever, 5m away from the grab's motion trajectory, and is connected to the electrical cabinet via an M12 aviation plug. The sensor is encapsulated in a waterproof and breathable polytetrafluoroethylene membrane with a pore size of 0.2μm, and the membrane material is fixed by ultrasonic welding. The aviation plug connector adopts IP67 protection grade and has an insertion and removal force of 20N±5N. Verification in a constant temperature and humidity chamber shows that the temperature error is ≤0.2°C and the humidity error is ≤1.5%RH. The response time for rapid changes in temperature and humidity (20°C→30°C, 50%RH→80%RH) is ≤5s (temperature) and ≤8s (humidity). After 72 hours of continuous operation, the data drift is ≤0.5%FS.
[0036] The edge computing unit 510 adopts an embedded computing platform, equipped with an 8-core processor, a graphics processing unit, and a memory ≥16GB; deploys a preprocessing algorithm library, including Savitsky-Golay filtering, baseline correction polynomial fitting order ≤5, and first-order derivative spectral processing; data storage uses a solid-state hard drive with a capacity ≥1TB, supports H.265 video encoding compression, a compression ratio ≥50:1, and a processing delay ≤20ms.
[0037] The humidity-viscosity correction model is deployed on an industrial computer and connected to the data acquisition unit via a USB interface. The coefficients are updated using the recursive least squares method and are triggered every 1,000 sets of data or every 24 hours.
[0038] The specific workflow of the present invention is as follows: Initialization phase
[0039] The main power supply supplies power to each module, the supercapacitor backup power supply enters standby mode, and after the redundant power supply line passes the self-test, the system completes power-on initialization; the viscosity detection unit (near-infrared spectrometer, capacitive sensor), anti-adhesion execution unit (vibration disk, cooling plate), and environmental perception unit (temperature and humidity sensor) are started in sequence to complete the hardware status check and communication link test; the collaborative control unit reads the preset parameters (such as vibration frequency range, coating target temperature, and strategy optimization weight) from the memory, and the human-computer interaction unit displays the initialization progress bar.
[0040] Data collection phase
[0041] The near-infrared spectroscopy detection component scans the material at a sampling rate of 1kHz, acquiring spectral data in the 900-2500nm band and optimizing the optical signal intensity through an optical path adjustment mechanism. The capacitive viscosity detection component synchronously collects changes in the parallel plate capacitance and, in conjunction with a temperature compensation network, corrects for the effects of ambient temperature and humidity on the capacitance value. The data acquisition and processing unit pre-processes the raw spectrum and capacitance value (e.g., Savitsky-Golay filtering and baseline correction) to extract characteristic peak intensities and capacitance change rates.
[0042] The temperature and humidity detection component updates the environmental data outside the cantilever at a frequency of 10Hz and transmits it to the collaborative control unit through the 5G network; the material humidity compensation component calculates the corrected viscosity value based on the ambient humidity and the detected viscosity value through a cubic polynomial regression model.
[0043] Data processing and strategy generation stage
[0044] The collaborative control unit receives preprocessed spectral data, capacitance values, and ambient temperature and humidity, and uses an improved deep belief network (3 hidden layers, 512-256-128 nodes) for feature fusion, outputting real-time viscosity values (error ≤ 2% FS) and humidity compensation coefficients (ΔRH ≤ 1%). The fusion results are further processed by the edge computing unit to generate a state space vector containing the viscosity value, ambient temperature and humidity, and grab load.
[0045] Based on the deep Q-network algorithm, the state space vector is used as input and the optimal action (vibration frequency, coating temperature, grab opening angle) is selected through the ε-greedy strategy. The reward function calculates the immediate benefit based on the anti-adhesion efficiency (weight 0.6), energy consumption (weight 0.3), and equipment loss (weight 0.1), and updates the Q-value table to optimize the long-term strategy.
[0046] Anti-adhesion action execution phase
[0047] The piezoelectric ceramic vibration disk adjusts the PWM duty cycle according to the vibration frequency (0-8kHz) output by the strategy, generating a vibration with an amplitude of 0.1-1.5mm, which destroys the adhesion between the material and the inner wall of the grab bucket; the vibration disk is connected to the side wall of the grab bucket through elastic damping parts to ensure efficient transmission of vibration energy (transmission rate ≥85%) while suppressing the impact of resonance on the equipment.
[0048] The semiconductor refrigeration chip adjusts the cooling capacity according to the target temperature (-10℃~60℃) output by the strategy, and provides real-time feedback of the coating temperature through the temperature sensor array (spacing 20mm, accuracy ±0.2℃); the nano-silica hydrophobic coating (water contact angle ≥160°, rolling angle ≤5°) further reduces material residue and reduces the cohesion of viscous materials in conjunction with the low temperature environment.
[0049] Feedback adjustment and closed-loop control stage
[0050] The viscosity detection unit continuously collects material parameters after anti-adhesion actions are performed, and compares the expected values of the strategy with the actual measured values (such as the amount of residual material and the viscosity change rate). The environmental perception unit updates the ambient temperature and humidity data to evaluate the accuracy of the humidity compensation model.
[0051] If the actual anti-adhesion efficiency is lower than the threshold (such as the amount of residual material > 20% of the grab capacity), the collaborative control unit will readjust the strategy weight or action range; the energy management unit will dynamically switch the main power supply and the supercapacitor backup power supply according to the power requirements of the actuator to ensure power supply for high-energy consumption actions (such as high-frequency vibration).
[0052] Task completion and system standby phase
[0053] After unloading is completed, the grab opening angle is adjusted to 0°, the vibrating plate and refrigeration plate stop working, and the coating temperature returns to ambient temperature; the system records the viscosity test data, strategy adjustment records, and energy consumption statistics of this operation to form a traceable operation log; the collaborative control unit enters standby mode, retaining only the basic monitoring functions of the temperature and humidity detection components and the energy management unit, waiting for the next operation instruction.
[0054] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.
[0055] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.
Claims
1. A ship unloader anti-adhesion coordination system for sticky bulk materials, characterized by: The invention comprises a viscosity detection module (100), which is installed inside the grab bucket of the ship unloader and is used for real-time detection of material viscosity parameters, and comprises a near-infrared spectrum detection submodule (110) and a capacitive viscosity detection submodule (120); an anti-adhesion execution module (200), which is connected to the viscosity detection module (100) via CAN bus data and dynamically adjusts the anti-adhesion strategy based on the detection results, and comprises a high-frequency vibration submodule (210) and a coating temperature control submodule (220); a collaborative control module (300), which is integrated into the electrical control cabinet of the ship unloader and is connected to the anti-adhesion execution module (200) via the EtherCAT bus and comprises a multi-source data fusion submodule (310) and a strategy optimization submodule (320); an environmental monitoring system. An environmental perception module (400) is installed on the outside of the cantilever and is connected to the collaborative control module (300) via 5G wireless communication, and includes a temperature and humidity detection submodule (410) and a material humidity compensation submodule (420); a data acquisition and processing module (500) is integrated into the collaborative control module (300) and includes an edge computing unit (510) and a pre-processing algorithm library (520); a human-computer interaction module (600) is connected to the collaborative control module (300) via the TCP / IP protocol and is used for strategy verification and parameter correction; an energy management module (700) is a master-slave architecture and includes a main power supply (710) and a supercapacitor backup power supply (720), and supplies power to each module via a redundant power supply line; Each module realizes closed-loop control through real-time data flow, and the total response time of the system does not exceed 120ms. Among them, the viscosity detection module (100) adopts an improved partial least squares regression algorithm, and the viscosity detection accuracy reaches ±2%FS.
2. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The near-infrared spectrum detection submodule (110) comprises two miniature near-infrared spectrometers with a wavelength range of 900-2500 nm and a spectral resolution better than 2 nm. The near-infrared spectrometers are installed at symmetrical positions on the inner wall of the grab bucket, and the optical path length is dynamically adjusted within the range of 5-15 mm by a precision screw mechanism. The light source adopts a wide-band luminous body with a power density of ≥5 W / cm² and a color temperature range of 2800-3200 K. The near-infrared spectrum detection submodule (110) is equipped with an optical feedback stabilization circuit to ensure spectral stability. The detector adopts a linear array structure with a responsivity of ≥0.7 A / W and a dark current of ≤1.2 nA. The signal is sampled by a 24-bit analog-to-digital converter with a sampling rate of ≥1 kHz.
3. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The capacitive viscosity detection submodule (120) adopts a parallel plate capacitance sensor with an electrode area of 20 mm×15 mm, an electrode spacing of 0.5 mm, and a polytetrafluoroethylene film coated on the electrode surface to prevent material adhesion; the detection circuit has an operating frequency range of 1 kHz-10 kHz, a sensitivity of 0.1 pF / Pa·s, a built-in temperature compensation network, a compensation range of -20°C to 80°C, and a compensation accuracy of ±0.3% FS; the output signal is differentially amplified and low-pass filtered, the filter cutoff frequency is 500 Hz, and the signal-to-noise ratio is ≥60 dB.
4. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The vibration source of the high-frequency vibration submodule (210) adopts a piezoelectric ceramic laminated structure, has an external dimension of 50 mm×10 mm, a resonance frequency of 8 kHz±200 Hz, and a vibration amplitude that is steplessly adjusted within a range of 0.1-1.5 mm through pulse width modulation; the driving circuit adopts an LLC resonant topology, has an input voltage range of 24 V±10%, an output power density ≥5 W / cm³, and a conversion efficiency ≥90%; the vibration disk is connected to the side wall of the grab bucket through an elastic damping member, has a damping coefficient of 0.2-0.5, a vibration transmission rate ≤15%, and a protection level of IP68.
5. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The inner wall of the coating temperature control submodule (220) is coated with a nano-silicon dioxide hydrophobic coating with a coating thickness of 50-100 μm, a water contact angle of ≥160°, and a rolling angle of ≤5°; a semiconductor refrigeration element is integrated at the bottom with a cooling capacity of ≥50W, a temperature control range of -10°C to 60°C, a control accuracy of ±0.5°C, a proportional-integral-differential algorithm drive, and a response time of ≤30s; a temperature sensor array is set on the coating surface with a sensor spacing of 20 mm and a measurement accuracy of ±0.2°C, and the data is used for feedback control of the coating temperature.
6. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The input data of the multi-source data fusion submodule (310) include near-infrared spectral data, capacitance value, and ambient temperature and humidity; the fusion algorithm adopts an improved deep belief network, which includes three hidden layers with 512, 256, and 128 nodes respectively, and the activation function adopts a rectified linear unit; the training data set includes ≥8000 groups of viscous bulk material samples, covering a viscosity range of 10-1000 mPa·s and a humidity range of 5-90%RH; the fusion output viscosity prediction value error is ≤2%FS, and the humidity compensation coefficient ΔRH is ≤1%.
7. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The strategy optimization submodule (320) is based on a deep Q network algorithm, the state space includes real-time viscosity value, grab load, ambient temperature and humidity, and the state representation adopts normalization processing; The action space defines vibration frequency, coating temperature, and grab opening. The action selection adopts a greedy strategy with an initial value of 0.9 and a decay rate of 0.
995. The reward function is designed with an anti-adhesion efficiency weight of 0.6, an energy consumption weight of 0.3, and an equipment loss weight of 0.
1. The anti-adhesion efficiency is evaluated by the amount of residual material in the grab.
8. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The temperature and humidity detection submodule (410) adopts a digital temperature and humidity sensor with a temperature measurement range of -40°C to 125°C and an accuracy of ±0.2°C, and a humidity measurement range of 0-100%RH and an accuracy of ±1.5%RH. The sensor is encapsulated in a waterproof and breathable membrane made of polytetrafluoroethylene with a pore size of 0.2μm. It is connected to the electrical cabinet via an M12 aviation plug, with a protection level of IP67, a response time based on temperature conditions of ≤5s, and a response time based on humidity conditions of ≤8s.
9. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The material humidity compensation submodule (420) has a built-in humidity-viscosity correction model, the model form is a cubic polynomial regression, the input is the ambient humidity and the detected viscosity, and the output is the corrected viscosity value; the correction error of the model training data is ≤1.5%FS, the model update frequency is every 1000 sets of new data or every 24 hours, and the update method adopts the recursive least squares method.
10. The ship unloader anti-adhesion coordination system for sticky bulk materials according to claim 1, characterized in that: The edge computing unit (510) adopts an embedded computing platform, is equipped with an 8-core processor, a graphics processing unit, and a memory ≥16GB; deploys a preprocessing algorithm library, including Savitsky-Golay filtering, baseline correction polynomial fitting order ≤5, and first-order derivative spectrum processing; data storage adopts a solid-state hard disk with a capacity ≥1TB, supports H.265 video encoding compression, a compression ratio ≥50:1, and a processing delay ≤20ms.
Citation Information
Cited By
Synthesis reaction device and synthesis method of sulfur-free odorant
CN121571079A