Gravity power generation and energy storage system based on linear motor and deep learning multi-task model

Abstract

The invention provides a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model, and the system comprises a front load car and a rear load car which are respectively arranged in a front vertical track section and a rear vertical track section in a sliding manner. The front linear motor is in driving connection with the front load-carrying lift car, the rear linear motor is in driving connection with the rear load-carrying lift car, and the battery module is connected with the two linear motors. A central controller of the energy storage control unit is responsible for starting and operation control of the system, including stages of preheating, acceleration, steering, deceleration and the like. A multi-task module based on deep learning is introduced, a central controller dynamically adjusts the driving force of a linear motor, the car is prevented from inclining or deviating from a track, mistaken starting is avoided, the posture of the car is monitored and adjusted in real time, it is ensured that the car is kept stable in the movement process, and therefore the safety and stability of the system are improved. The invention aims to improve the efficiency, precision and safety of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of gravity energy storage, and particularly provides a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model. Background Art

[0002] Most existing gravity energy storage experimental systems are based on a lifting mechanical structure, and use a traditional rotary motor combined with gears, belts or hydraulic systems to achieve the vertical lifting of a mass block. The structural design of these systems is often relatively complex, including multiple mechanical conversion links, such as gearboxes and pulley systems, which work together to convert rotary motion into linear motion. And a large amount of energy loss will be brought in the multi-steps involved in this energy conversion, affecting the overall energy efficiency of the system. At the same time, due to the influence of mechanical delay and system inertia, the response speed of the model and the real-time performance of control are greatly limited. In addition, due to the mechanical structure transmission system involving a variety of complex mechanical components, the maintenance work requirements are relatively high, affecting the long-term sustainability of the system. Summary of the Invention

[0003] In order to achieve the above object and overcome the above problems, the present invention proposes a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model. The system includes a three-dimensional frame, a car unit, a heavy block transfer unit, an image acquisition unit, an image processing unit and an energy storage control unit. The three-dimensional frame is provided with a plurality of three-dimensional tracks at intervals along the width direction. The three-dimensional track includes a front vertical track section, a horizontal track section and a rear vertical track section;

[0004] The car unit, the front load car slides in the front vertical track section, the rear load car slides in the rear vertical track section, the front linear motor is drivingly connected to the front load car, the rear linear motor is drivingly connected to the rear load car, and the battery module is connected to the two linear motors;

[0005] The heavy block transfer unit, the conveyor belt slides in the three-dimensional track. The heavy blocks include a plurality of them and are evenly distributed along the length direction of the conveyor belt. The transfer driving mechanism is drivingly connected to the conveyor belt;

[0006] The image acquisition unit is placed above the conveyor belt track and includes an image acquisition module to ensure that the image acquisition module can acquire the track image and is connected to the image acquisition unit.

[0007] The image processing unit is connected to the central controller and includes an object detection module, an attitude detection module and a track wear detection module.

[0008] The energy storage control unit, the control process of the central controller includes an image preprocessing stage, a startup stage, an attitude detection stage and an operation stage:

[0009] Image preprocessing stage: It includes image acquisition, filtering out the mask of the track using traditional computer vision algorithms in the experiment, extracting image features using the Darknet backbone network, and performing real-time detection of the track using the object detection head. After detecting the mass block, the startup stage is carried out.

[0010] Startup stage: Before startup, the system automatically preheats and gradually accelerates the front linear motor and the rear linear motor to the operating speed according to the sensor data and preset parameters;

[0011] Image processing stage: Use the Darknet backbone network to extract image features, use the multi-task head to detect whether there are defective blocks on the track and the attitude data of the car, judge whether there is an abnormality in the movement of the car, and return the result to the central controller.

[0012] Operation stage: It includes the stages of accelerating upward, turning, and decelerating downward. The system adjusts the load and speed according to the energy demand.

[0013] Furthermore, the image processing unit first obtains the image through the image acquisition device, then obtains the edge information in the whole image through the edge detection algorithm, and then obtains the straight line fitted to the track boundary in the image through the Hough line transformation, and determines the middle area of the straight line as the mask.

[0014] Extract image features through Darknet, and then pass through the feature fusion network and then through the object detection head to train and detect the target block. The loss function in its training process is shown in Equation (1):

[0015] L = L loc + L conf + L cls + L structure (1)

[0016] L loc is the localization loss, which is used to measure the difference between the position and size of the predicted bounding box and the ground truth bounding box; L conf is the confidence loss, which is used to measure the difference between the probability that the predicted bounding box contains an object and the actual situation; L cls is the classification loss, which is used to measure the difference between the predicted class and the ground truth class, and L structure is the skeleton loss, which measures the key skeleton information of the attitude of the block.

[0017] Furthermore, L loc is calculated according to Equation (2):

[0018]

[0019] where λ_coord: weight coefficient, emphasizing the importance of the localization error; Sum over all grid cells, and S2 is the number of grids; Sum the number of bounding boxes B predicted for each grid cell; An indicator function that is 1 when the i-th grid cell and the j-th bounding box are responsible for predicting an object, and 0 otherwise; The center coordinates of the ground truth bounding box And the center coordinates x of the predicted bounding box i The mean squared error between them; The center coordinates y of the ground truth bounding box i And the center coordinates of the predicted bounding box The mean squared error between them; The width of the ground truth bounding box And the width of the predicted bounding box The mean squared error between them; The height of the ground truth bounding box And the height of the predicted bounding box The mean squared error between them.

[0020] Furthermore, L conf Is calculated according to Equation (3):

[0021]

[0022] Where c i : The confidence of the ground truth bounding box (1 if the box is responsible for predicting an object, 0 otherwise); The confidence of the predicted bounding box; The weight coefficient, an indicator function that is 1 when the i-th grid cell and the j-th bounding box are not responsible for predicting any object, and 0 otherwise.

[0023] Furthermore, L cls Is calculated according to Equation (4):

[0024]

[0025] P is the probability predicted by the model belonging to class c. By weighted summing the three loss functions, the loss function for object detection can be obtained.

[0026] Furthermore, for the pose detection task, in addition to the basic object detection loss function, a new loss function is involved, as shown in Equation (4):

[0027]

[0028] L structure Is the backbone loss of the model, used to judge the pose of the mass block, And Represent the true distance and the predicted distance between point k and point I respectively.

[0029] Furthermore, the energy storage control unit adjusts the operating parameters according to the real-time monitoring data of load changes, energy demands, and environmental conditions, and sets operations according to the energy demands to adjust the energy storage release speed to meet the demand fluctuations.

[0030] Furthermore, the energy storage control unit includes a preprocessing module that processes images, which can reduce the computational amount and cost. The energy storage control unit is provided with an anti-misoperation startup module that detects objects through a target detection algorithm and enters the startup phase only when a target mass block is detected, preventing misoperation. The energy storage control unit is set to maximize the energy capture and storage efficiency. During low-load periods, it automatically switches to an energy-saving mode and sets a refined energy feedback control logic to ensure the maximization of energy recovery and reuse at any operating stage.

[0031] Furthermore, the energy storage control unit is also provided with a fault detection module configured to diagnose the status of key components in real time, automatically identify potential faults and issue early warnings, and issue automatic alarms in case of overload, mechanical jamming, etc. The energy storage control unit is provided with a car status alarm module that will give an early warning when the attitude of the car changes abnormally to prevent accidents. The energy storage control unit is provided with a track defect detection module that will give an early warning and automatically stop the machine when there are damages or defects on the track, reducing losses and preventing accidents. The energy storage control unit is also provided with an emergency response module configured to automatically start an emergency shutdown procedure when detecting abnormal operations or performance degradation, and is equipped with an emergency stop button and a function of automatic shutdown in case of faults.

[0032] Furthermore, the acceleration time for the front linear motor and the rear linear motor to reach the best operating state is 0.2 s, and then they move at a constant speed. When they are 500 mm away from the target position, they enter low-speed operation, and when they reach the target position, they start the maximum progressive braking to stop running.

[0033] Furthermore, the front load car and the rear load car are respectively provided with height sensors, and the central controller is signal-connected to the height sensors and is configured to receive and control the operation of the linear motors according to the car height detected by the height sensors to perform the motion control of the front load car and the rear load car.

[0034] Furthermore, the front track section is provided with a front travel sensor for detecting the up and down travel of the front load car, and the rear track section is provided with a rear travel sensor for detecting the up and down travel of the rear load car. The central controller is signal-connected to the height sensors and is configured to receive and control the operation of the linear motors according to the car height detected by the height sensors to perform the motion control of the front load car and the rear load car.

[0035] Further, the front load car is provided with rollers that rollingly cooperate with the front vertical track, and the rear load car is provided with rollers that rollingly cooperate with the rear vertical track. Each roller is connected to the car through a wheel carrier; the conveyor belt is provided with rollers that cooperate with the front vertical track section, the horizontal track section, and the rear vertical track section, or the front vertical track section, the horizontal track section, and the rear vertical track section are provided with rollers that cooperate with the conveyor belt. The linear motor includes a stator and a mover that are slidably engaged with each other. The stator of the motor is arranged parallel to the front vertical track and the rear vertical track, and the mover of the motor is fixedly installed on the outer circumference of the car.

[0036] The system further includes a braking and regenerating unit that is braking-connected to the front load car and the rear load car. The braking and regenerating unit includes a mounting base plate, a rope pulley, a speed reducer, a generator, and a brake. The mounting base plate is fixedly installed directly above the front vertical track and the rear vertical track. The motor shaft of the generator is connected to the axle of the rope pulley through the speed reducer. The flexible rope end fixedly wound around the rope pulley extends vertically downward and is connected to the load car. The brake is braking-connected to the axle of the rope pulley. The conveying drive mechanism includes an electromagnetic push-pull device arranged inside the car. The electromagnetic push-pull device includes an electromagnet, a guiding track, and a polarity control circuit. When the weight moves to a specified height on the vertical track, the electromagnetic push-pull device is activated, and the electromagnet switches to a repulsive polarity to push the weight out of the car; when the weight moves to the end position on the mirror side through the conveyor belt, the electromagnetic push-pull device on the other side of the car is activated, and the electromagnet switches to an attractive polarity to pull the weight into the car on the mirror side.

[0037] Further, the conveying drive mechanism further includes an electric drive roller device arranged at the horizontal guide rail section, the front vertical track section, and the rear vertical track section. The electric drive roller device includes a conveying motor and a conveying wheel. The conveying motor is connected to the conveying wheel through a speed reduction transmission mechanism. The conveyor belt is frictionally driven by the conveying wheel. The conveying motor is configured to drive the conveying wheel to rotate, drive the conveyor belt to move along the three-dimensional track, and drive the weight to enter or leave the car.

[0038] Compared with the prior art, the technical advantages of the gravity energy storage system based on a linear motor provided by the present invention are at least reflected in:

[0039] 1. By directly connecting the linear motor to the load car, intermediate conversion mechanisms such as traditional gears and belts are removed, and electric energy is directly converted into linear power. While simplifying the structure, the energy conversion efficiency and the system response speed are significantly improved. Moreover, high-precision sensors are integrated to realize real-time monitoring and adjustment of the position and speed of the heavy object, ensuring precise operation and improving the efficiency and precision of the gravity energy storage system.

[0040] 2. Through regenerative braking technology, the kinetic energy during the descent of heavy objects is recovered and stored, reducing energy waste, enhancing the overall energy efficiency of the system. At the same time, the use of environmentally friendly battery modules reduces the impact on the environment, decreases energy consumption, reduces carbon emissions, and promotes the development of energy conservation and environmental protection.

[0041] 3. By using deep learning technology, the safety of the device is improved, which can effectively reduce energy waste, make reasonable safety warnings, and enhance the reliability of the system.

[0042] 4. The modular design enables each system component to be independently operated and maintained, reducing the complexity and cost of maintenance, improving the long-term sustainability of the system. It is applicable to actual application scenarios such as terminal shipping and bulk cargo storage, improving the efficiency of the bulk cargo storage and shipping process. The high-efficiency and precise control ability of the system enhances the operation efficiency, reduces the intensity of manual operation, and has significant social and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings forming a part of the specification depict embodiments of the present invention and, together with the specification, are used to explain the principles of the present invention. Referring to the drawings, the present invention can be more clearly understood according to the following detailed description, wherein:

[0044] Figure 1 is a schematic structural diagram of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to an embodiment of the present invention;

[0045] Figure 2 is a schematic structural diagram of the front load car according to an embodiment of the present invention; (split state)

[0046] Figure 3 is a schematic structural diagram of the electromagnetic push-pull device provided by an embodiment of the present invention; (split state)

[0047] Figure 4 is a schematic structural diagram of the braking regeneration unit according to an embodiment of the present invention;

[0048] Figure 5 is a schematic circuit system diagram of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to an embodiment of the present invention;

[0049] Figure 6 is the main system control interface of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to an embodiment of the present invention;

[0050] Figure 7It is an operation monitoring interface of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model provided by an embodiment of the present invention;

[0051] Figure 8 It is a total monitoring interface of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model provided by an embodiment of the present invention;

[0052] Figure 9 It is a fault alarm interface of a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model provided by an embodiment of the present invention;

[0053] Figure 10 It is a multi-task convolutional neural network flowchart provided by an embodiment of the present invention.

[0054] Explanation of the attached drawing reference numerals:

[0055] 11 - Stereo frame, 12 - Stereo track, 121 - Front vertical track section, 122 - Horizontal track section, 123 - Rear vertical track section;

[0056] 21 - Front load car, 22 - Rear load car, 23 - Linear motor, 24 - Roller, 25 - Wheel frame;

[0057] 31 - Conveyor belt, 32 - Heavy block, 33 - Electromagnet, 34 - Guide track;

[0058] 41 - Rope wheel, 42 - Reducer, 43 - Generator, 44 - Brake;

[0059] 51 - Central controller, 52 - Battery module.

[0060] It should be understood that the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship. In addition, the same or similar reference numerals represent the same or similar components. Detailed implementation manners

[0061] Due to the complex structure of the traditional gravity energy storage experimental model, there are problems such as poor energy efficiency, low control accuracy, poor sustainability, and complex maintenance. The present invention proposes a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model, which relates to a new linear motor model system for improving the system performance and experimental accuracy during gravity energy storage experiments, belongs to the field of gravity energy storage hardware systems, and is particularly applicable to the scenarios of terminal shipment and bulk cargo storage.

[0062] A gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model is provided. A plurality of three-dimensional tracks are spaced apart in a three-dimensional frame; an image acquisition device is placed on the three-dimensional track. The front linear motor of the car unit is arranged vertically on the front side of the three-dimensional frame and is drivingly connected to the adjacent front load-carrying car. The rear linear motor is arranged vertically on the rear side of the three-dimensional frame and is drivingly connected to the adjacent rear load-carrying car; the conveyor belt of the heavy block conveying unit slides in the track. The heavy blocks include a plurality of them and are evenly distributed along the length direction of the conveyor belt. The conveying drive mechanism drives the conveyor belt to move forward to drive the front heavy block to slide into the front load-carrying car and the rear heavy block to slide out of the rear load-carrying car, or drives the conveyor belt to move backward to drive the front heavy block to slide out of the front load-carrying car and the rear heavy block to slide into the rear load-carrying car; the central controller of the energy storage control unit is control-connected to the front linear motor, the rear linear motor and the conveying drive mechanism. The battery module is configured to store the electric energy generated by the linear motor and supply it for the system to use.

[0063] The specific algorithm flow of the image processing unit is as Figure 10 shown:

[0064] First, an image is obtained through an image acquisition device, and then the edge information in the whole image is obtained through an edge detection algorithm. Then, through the Hough line transformation, the straight lines fitted to the track boundaries in the image are obtained, and the middle area of the straight lines is recognized as a mask:

[0065] The image features are extracted through Darknet, and then after passing through a feature fusion network and then through a target detection head to train and detect the target block. The loss function in its training process is shown in Equation (1):

[0066] L = L loc + L conf + L cls + L structure (1)

[0067] L loc is the localization loss, which is used to measure the difference between the position and size of the predicted bounding box and the true bounding box. L conf is the confidence loss, which is used to measure the difference between the probability that the predicted bounding box contains an object and the actual situation. L cls is the classification loss, which is used to measure the difference between the predicted class and the true class. L structure is the skeleton loss, which measures the key skeleton information of the posture of the object block. Using this function, a multi-task model can be trained, as shown in Equation (2):

[0068]

[0069] where λ_coord: weight coefficient, emphasizing the importance of the localization error. Sum over all grid cells, and S2 is the number of grids. Sum the number of bounding boxes B predicted for each grid cell. An indicator function that is 1 if the i-th grid cell and the j-th bounding box are responsible for predicting an object, and 0 otherwise. True bounding box center coordinates And the predicted bounding box center coordinate x i The mean squared error between them. True bounding box center coordinate y i And the predicted bounding box center coordinate The mean squared error between them. True bounding box width And the predicted bounding box width The mean squared error between them. True bounding box height And the predicted bounding box height The mean squared error between them. L conf As shown in Equation (3):

[0070]

[0071] Where c i : The confidence of the true bounding box (1 if the box is responsible for predicting an object, 0 otherwise). The confidence of the predicted bounding box. The weight coefficient, an indicator function that is 1 if the i-th grid cell and the j-th bounding box are not responsible for predicting any object, 0 otherwise. L cls As shown in Equation (4):

[0072]

[0073] P is the probability of belonging to class c predicted by the model. By weighted summing the three loss functions, the loss function for object detection can be obtained.

[0074] For the pose detection task, in addition to the basic object detection loss function, a new loss function is involved, as shown in Equation (5).

[0075]

[0076] L structure Is the backbone loss of the model, used to judge the pose of the mass block, And Represent the true distance and the predicted distance between point k and point I, respectively.

[0077] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. The description of the exemplary embodiments is merely illustrative and in no way limits the present invention and its application or use. The present invention can be implemented in many different forms and is not limited to the embodiments herein. In the present invention, when it is described that a specific device is located between a first device and a second device, there may or may not be an intermediate device between the specific device and the first device or the second device.

[0078] Embodiment 1:

[0079] A gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model, a three-dimensional frame 11, with a plurality of three-dimensional tracks 12 arranged at intervals along the width direction. The three-dimensional tracks 12 include a front vertical track section 121, a horizontal track section 122, and a rear vertical track section 123; a car unit, a front load car 21 slides in the front vertical track section 121, a rear load car 22 slides in the rear vertical track section 123, a front linear motor 23 is drivingly connected to the front load car 21, a rear linear motor 23 is drivingly connected to the rear load car 22, and a battery module 52 is connected to the two linear motors 23; a heavy block conveying unit, a conveyor belt 31 slides in the three-dimensional track, and the heavy blocks 32 include a plurality of them and are evenly distributed along the length direction of the conveyor belt 31, and a conveying driving mechanism is drivingly connected to the conveyor belt 31; and

[0080] An energy storage control unit, the control process of the central controller includes a pneumatic stage and an operation stage: Start-up stage: Before starting, the system automatically preheats and gradually accelerates the front linear motor 23 and the rear linear motor 23 to the operating speed according to sensor data and preset parameters; Operation stage: including an accelerating upward, turning, and decelerating downward stage, and the system adjusts the load and speed according to the energy demand.

[0081] The modular design enables each system component to operate and be maintained independently, reducing the maintenance complexity and cost, improving the long-term sustainability of the system, being applicable to actual application scenarios such as terminal shipping and bulk cargo storage, improving the efficiency of the bulk cargo storage and shipping process, the high efficiency and precise control ability of the system improve the operation efficiency, reduce the manual operation intensity, and have significant social and economic benefits.

[0082] During the implementation process, the acceleration time for the front linear motor 23 and the rear linear motor 23 to reach the optimal operating state is 0.2 s, and then they move at a constant speed. When they are 500 mm away from the target position, they enter low-speed operation, and when they reach the target position, they start the maximum progressive braking to stop operating.

[0083] As Figures 6 to 9As shown, during implementation, the energy storage control unit adjusts operating parameters based on real-time monitoring data of load changes, energy demand, and environmental conditions, and sets operations according to energy demand to adjust the energy storage release speed to meet demand fluctuations.

[0084] In some preferred embodiments, the energy storage control unit is set to maximize energy capture and storage efficiency. During low-load periods, it automatically switches to an energy-saving mode and sets a refined energy feedback control logic to ensure maximum energy recovery and reuse at any operating stage.

[0085] In some preferred embodiments, the energy storage control unit is also provided with a fault detection module, which is configured to diagnose the status of key components in real time, automatically identify potential faults, and issue early warnings.

[0086] In some preferred embodiments, the energy storage control unit is also provided with an emergency response module for the response mechanism, which is configured to automatically start an emergency shutdown procedure when detecting abnormal operations or performance degradation, and is equipped with an emergency stop button and a function of automatic shutdown in case of faults.

[0087] Embodiment 2:

[0088] As Figures 1 to 5 shown, the present invention provides a gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model, including a three-dimensional framework, a car unit, a heavy block transfer unit, and an energy storage control unit.

[0089] The three-dimensional framework includes a plurality of vertical support beams and horizontal support beams. The three-dimensional framework 11 is provided with a plurality of three-dimensional tracks 12 at intervals along the width direction. The three-dimensional tracks 12 include a front vertical track section 121, a horizontal track section 122, and a rear vertical track section 123. The front vertical track section 121 and the rear vertical track section 123 are respectively arranged on the front and rear sides of the three-dimensional framework 11. The horizontal track section 122 is arranged longitudinally on the top of the three-dimensional framework 11, and the front and rear ends of the horizontal guide rail section are respectively connected to the upper ends of the front vertical track section 121 and the rear vertical track section 123.

[0090] During implementation, the three-dimensional framework 11 is composed of a plurality of vertical and horizontal support beams to form a strong matrix structure, providing the basic support and stability for the entire system, and ensuring that other components can operate safely and efficiently inside it.

[0091] The three-dimensional tracks 12 are composed of tracks in the horizontal and vertical directions, and the tracks are equipped with efficient rolling elements. It provides a moving path for the load car and the horizontal conveyor belt, reducing friction and energy loss. The track system ensures that the heavy object can move smoothly and accurately within the system. The vertical tracks are used in combination with the linear motor 23 to provide additional stability and safety, prevent lateral swing, and ensure precise positioning of the vertical movement of the heavy object.

[0092] The car unit includes a front load car 21, a rear load car 22, a front linear motor 23 and a rear linear motor 23. The front load car 21 slides in the front vertical track section 121, and the rear load car 22 slides in the rear vertical track section 123. The front linear motor 23 is arranged vertically on the front side of the three-dimensional frame 11 and is drivingly connected to the adjacent front load car 21. The rear linear motor 23 is arranged vertically on the rear side of the three-dimensional frame 11 and is drivingly connected to the adjacent rear load car 22. The load car is used to carry heavy blocks. High-precision sensors are equipped on the car, and it moves through high-efficiency rolling elements. The load car carries and transports heavy objects to various positions in the system, realizing the storage and release of heavy objects.

[0093] The heavy block transfer unit includes a conveyor belt 31, heavy blocks 32 and a transfer drive mechanism. The conveyor belt 31 slides in the horizontal track section 122, and its front end extends downward along the front vertical track section 121, and its rear end extends downward along the rear vertical track section 123. A plurality of heavy blocks 32 are included and are evenly distributed along the length direction of the conveyor belt 31. The transfer drive mechanism is drivingly connected to the conveyor belt 31 and is configured to drive the conveyor belt 31 to move forward to drive the front heavy block 32 to slide into the front load car 21 and the rear heavy block 32 to slide out of the rear load car 22, or drive the conveyor belt 31 to move backward to drive the front heavy block 32 to slide out of the front load car 21 and the rear heavy block 32 to slide into the rear load car 22.

[0094] The energy storage control unit includes a central controller 51 and a battery module 52. The central controller is control-connected to the front linear motor 23, the rear linear motor 23 and the transfer drive mechanism. The battery module 52 is configured to store the electric energy generated by the linear motor 23 and supply it for the system to use.

[0095] The system motor selects a linear motor suitable for the requirements of the system load, and the motor specifications match the mass of the heavy block and the operating speed of the system. The linear motor 23 includes a motor stator and a motor mover that are slidably engaged. The motor stator is arranged parallel to the front vertical track and the rear vertical track, and the motor mover is fixedly installed on the circumferential outer side of the car. The linear motor 23 is used to drive the vertical and horizontal movement of the heavy block. It directly converts electrical energy into linear power, driving the load car to rise and fall, as well as horizontal movement. The efficient energy conversion and precise control of the linear motor are the core of the system, ensuring efficient energy storage and release.

[0096] Among them, the battery energy storage unit is composed of multiple lithium iron phosphate battery modules and is equipped with a battery management system (BMS). It stores the electric energy generated by the linear motor and the regenerative braking system for the system to use. The battery energy storage unit provides reliable energy storage and supply, ensuring the energy demand of the system under different working conditions. The battery module is an environmentally friendly lithium iron phosphate battery, which reduces the impact on the environment, reduces energy consumption, reduces carbon emissions, and promotes the development of energy conservation and environmental protection.

[0097] A gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model provided by the present invention directly connects the linear motor with the load car, removing intermediate conversion mechanisms such as traditional gears and belts, directly converting electric energy into linear power, significantly improving the energy conversion efficiency and system response speed while simplifying the structure, and integrating high-precision sensors to achieve real-time monitoring and adjustment of the position and speed of the heavy object, ensuring precise operation and improving the efficiency and precision of the gravity energy storage experimental system.

[0098] The linear motor, the track system, the load car, and the battery energy storage unit are modularly designed to ensure that each module can operate independently, facilitating maintenance and possible upgrades.

[0099] Embodiment Three:

[0100] As Figure 1 and Figure 4 shown, on the basis of Embodiment One, the system is optimized by adding a braking regeneration unit to implement regenerative braking technology, converting the kinetic energy during the descent of the car into electric energy and storing it in the lithium iron phosphate battery.

[0101] Specifically, the system further includes a braking regeneration unit that is braking-connected to the front load car 21 and the rear load car 22. The braking regeneration unit includes a mounting base plate, a rope pulley 41, a speed reducer 42, a generator 43, and a brake 44. The mounting base plate is fixedly installed directly above the front vertical track and the rear vertical track. The motor shaft of the generator 43 is connected to the axle of the rope pulley 41 through the speed reducer 42. The flexible rope end fixedly wound around the rope pulley 41 extends vertically downward and is connected to the load car. The brake 44 is braking-connected to the axle of the rope pulley 41.

[0102] Through regenerative braking technology, the kinetic energy during the descent of the heavy object is recovered and stored, reducing energy waste, improving the overall energy efficiency of the system. At the same time, using environmentally friendly battery modules reduces the impact on the environment, reduces energy consumption, reduces carbon emissions, and promotes the development of energy conservation and environmental protection.

[0103] Embodiment Four:

[0104] As Figure 1 and Figure 3As shown in the figure, the conveying drive mechanism includes an electromagnetic push-pull device arranged inside the car. The electromagnetic push-pull device includes an electromagnet 33, a guiding track 34, and a polarity control circuit. When the weight 32 moves to a specified height on the vertical track, the electromagnetic push-pull device is activated, and the electromagnet 33 switches to a repulsive polarity to push the weight 32 out of the car. When the weight 32 moves to the end position on the mirror side through the conveyor belt 31, the electromagnetic push-pull device on the other side of the car is activated, and the electromagnet 33 switches to an attractive polarity to pull the weight 32 into the car on the mirror side.

[0105] An electromagnetic push-pull device is installed inside the car to realize the conversion of the heavy object from vertical movement to horizontal movement and to pull it into the car on the mirror side from the conveyor belt. The push-pull mechanism consists of an electromagnet, a polarity control circuit, and a guiding track. When the heavy object moves to a specified height on the vertical track, the electromagnetic push-pull device is activated, and the electromagnet switches to a repulsive polarity to push the heavy object out of the car and place it on the conveyor belt. When the heavy object moves to the end position on the mirror side through the conveyor belt, the electromagnetic push-pull device on the other side of the car is activated, and the electromagnet switches to an attractive polarity to pull the heavy object into the car on the mirror side.

[0106] Embodiment Five:

[0107] As Figure 1 shown in the figure, further, the conveying drive mechanism further includes an electric drive wheel device arranged at the horizontal guide rail section, the front vertical rail section 121, and the rear vertical rail section 123. The electric drive wheel device includes a conveying motor and a conveying wheel. The conveying motor is connected to the conveying wheel through a speed reduction transmission mechanism. The conveyor belt is frictionally driven by the conveying wheel. The conveying motor is configured to drive the conveying wheel to rotate, drive the conveyor belt 31 to move along the three-dimensional track 12, so as to drive the weight 32 to enter or leave the car.

[0108] Further, the conveyor belt 31 is a chain, the weights 32 are fixedly connected to different chain plates of the chain at equal intervals, the chain is slidably matched with the middle of the track, the weights 32 are slidably matched with both sides of the track, and the conveying wheel is a sprocket wheel matched with the chain.

[0109] Further, the conveyor belt 31 is a synchronous belt, the weights 32 are fixedly connected to the synchronous belt at equal intervals, the synchronous belt is slidably matched with the middle of the track, the weights 32 are slidably matched with both sides of the track, and the conveying wheel is a synchronous pulley matched with the synchronous belt.

[0110] Embodiment Six:

[0111] As Figures 1 to 4As shown, the sensor system consists of multiple sensors, including Hall sensors, position sensors, and speed sensors, which are installed at key positions. It monitors the status of various parts of the system in real time, including the position of heavy objects, speed, motor status, etc. The sensor system provides accurate feedback data to ensure the safe and efficient operation of the system. Sensors are integrated into the system to monitor the position and speed of the car and the heavy block, and the motor output is adjusted in real time to meet the requirements.

[0112] During implementation, the front load car 21 and the rear load car 22 are respectively provided with height sensors, and the central controller 51 is signal-connected to the height sensors and is configured to receive and control the operation of the linear motor 23 according to the car height detected by the height sensors to perform motion control of the front load car 21 and the rear load car 22.

[0113] The front track section is provided with a front travel sensor for detecting the up and down travel of the front load car 21, and the rear track section is provided with a rear travel sensor for detecting the up and down travel of the rear load car 22. The central controller 51 is signal-connected to the height sensors and is configured to receive and control the operation of the linear motor 23 according to the car height detected by the height sensors to perform motion control of the front load car 21 and the rear load car 22.

[0114] As Figure 2 shown, the front load car 21 is provided with rollers 24 that rollingly cooperate with the front vertical track, and the rear load car 22 is provided with rollers 24 that rollingly cooperate with the rear vertical track. Each roller 24 is connected to the car through a wheel frame 25.

[0115] The conveyor belt 31 is provided with rollers 24 that cooperate with the front vertical track section 121, the horizontal track section 122, and the rear vertical track section 123, or the front vertical track section 121, the horizontal track section 122, and the rear vertical track section 123 are provided with rollers 24 that cooperate with the conveyor belt 31.

[0116] Embodiment Seven:

[0117] As Figures 1 to 4 shown, the present invention provides a gravity energy storage system based on a linear motor and a deep learning multi-task model, including a three-dimensional frame, a car unit, a heavy block conveying unit, an image acquisition unit, an image processing unit, and an energy storage control unit. It monitors the status of various parts of the system in real time, including the position of heavy objects, speed, motor status, etc. The sensor system provides accurate feedback data to ensure the safe and efficient operation of the system. Sensors are integrated into the system to monitor the position and speed of the car and the heavy block, and the motor output is adjusted in real time to meet the requirements.

[0118] During implementation, height sensors are respectively provided on the front load car 21 and the rear load car 22. The central controller 51 is signal-connected to the height sensors and is configured to receive and control the operation of the linear motor 23 according to the car height detected by the height sensors to perform motion control of the front load car 21 and the rear load car 22.

[0119] During implementation, when the mass block has an abnormal pose change, the pose detection module detects the abnormality and issues a warning signal, effectively avoiding accidents and having significant social and economic benefits.

[0120] Embodiment VIII:

[0121] As Figure 1 shown, further, the conveying drive mechanism further includes electric drive wheel devices provided at the horizontal guide rail section, the front vertical rail section 121, and the rear vertical rail section 123. The electric drive wheel devices include conveying motors and conveying wheels. The conveying motors are connected to the conveying wheels through reduction drive mechanisms. The conveyor belt is friction-driven by the conveying wheels. The conveying motors are configured to drive the conveying wheels to rotate, drive the conveyor belt 31 to move along the three-dimensional track 12, and drive the heavy block 32 to enter or leave the car.

[0122] Further, during implementation, when there are foreign objects or damages on the conveyor belt, the conveyor belt will stop and a warning signal will be issued. Accidents are effectively avoided, and it has significant social and economic benefits.

[0123] The present invention not only improves the efficiency and accuracy of the gravity energy storage experimental system, but also reduces energy consumption, lowers carbon emissions, and promotes the development of energy conservation and environmental protection through applications in actual application scenarios such as terminal shipment and bulk cargo storage. At the same time, the high-efficiency and precise control ability of the system improves the operation efficiency, reduces the manual operation intensity, and has significant social and economic benefits.

[0124] So far, the embodiments of the present disclosure have been described in detail. To avoid obscuring the concept of the present disclosure, some details well known in the art have not been described.

[0125] Those skilled in the art can clearly understand how to implement the technical solutions disclosed here based on the above description. Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration and not for limiting the scope of the present invention. Those skilled in the art should understand that the above embodiments can be modified or some technical features can be equivalently replaced without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model, characterized in that, Comprising: A three-dimensional frame (11) is provided with a plurality of three-dimensional tracks (12) spaced along the width direction. The three-dimensional tracks (12) include a front vertical track section (121), a horizontal track section (122), and a rear vertical track section (123); A car unit, where the front load car (21) slides in the front vertical track section (121), the rear load car (22) slides in the rear vertical track section (123), the front linear motor (23) is drivingly connected to the front load car (21), the rear linear motor (23) is drivingly connected to the rear load car (22), and the battery module (52) is connected to the two linear motors (23); A weight transfer unit, where the conveyor belt (31) slides in the three-dimensional track, and the weights (32) include a plurality of them evenly distributed along the length direction of the conveyor belt (31), and the transfer driving mechanism is drivingly connected to the conveyor belt (31); An image acquisition unit is placed above the conveyor belt track and includes an image acquisition module to ensure that the image acquisition module can acquire the track image and is connected to the image acquisition unit; An image processing unit is connected to the central controller and includes an object detection module, an attitude detection module, and an orbit wear detection module; and An energy storage control unit, and the control process of the central controller includes an image preprocessing stage, a startup stage, an attitude detection stage, and an operation stage: Image preprocessing stage: It includes image acquisition, filtering out the mask of the track using traditional computer vision algorithms in the experiment, extracting image features using the Darknet backbone network, and using the object detection head to perform real-time detection on the track. After detecting the mass block, enter the startup stage; Startup stage: Before startup, the system automatically preheats and gradually accelerates the front linear motor (22) and the rear linear motor (23) to the operating speed according to the sensor data and preset parameters; Image processing stage: Use the Darknet backbone network to extract image features, use the multi-task head to detect whether there are defective blocks and the attitude data of the car on the track, judge whether there is any abnormality in the movement of the car, and return the result to the central controller; Operation stage: It includes an accelerating upward, turning, and decelerating downward stage, and the system adjusts the load and speed according to the energy demand.

2. The gravity power generation and energy storage system based on the linear motor and the deep learning multi-task model according to claim 1, wherein The image processing unit first obtains an image through an image acquisition device, then obtains the edge information in the whole image through an edge detection algorithm, and then obtains the straight line fitted to the track boundary in the figure through the Hough line transform, and determines the middle area of the straight line as the mask.

3. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 1 or 2, characterized in that Extract image features through Darknet, and then pass through a feature fusion network and then through an object detection head to train and detect the target block. The loss function of its training process is shown in formula (1): L = L loc + L conf + L cls + L structure (1) L loc is the localization loss, which is used to measure the difference between the position and size of the predicted bounding box and the ground truth bounding box; L conf is the confidence loss, which is used to measure the difference between the probability that the predicted bounding box contains an object and the actual situation; L cls is the classification loss, which is used to measure the difference between the predicted class and the ground truth class, L structure is the skeleton loss, which measures the key skeleton information of the object block pose.

4. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 3, characterized in that L loc Calculate according to formula (2): where λ_coord: weight coefficient, emphasizing the importance of localization error; Sum over all grid cells, where S2 is the number of grids; Sum over the number of bounding boxes B predicted for each grid cell; Indicator function, which is 1 when the i-th grid cell and the j-th bounding box are responsible for predicting an object, otherwise 0; True bounding box center coordinates And the predicted bounding box center coordinate x i The mean squared error between them; True bounding box center coordinate y i And the predicted bounding box center coordinate The mean squared error between them; True bounding box width And the predicted bounding box width The mean squared error between them; True bounding box height And the predicted bounding box height The mean squared error between them; L conf Calculate according to formula (3): where c i : Confidence of the true bounding box (1 if the box is responsible for predicting an object, otherwise 0); Confidence of the predicted bounding box; Weight coefficient, an indicator function that is 1 when the i-th grid cell and the j-th bounding box are not responsible for predicting any object, otherwise 0.

5. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 3, characterized in that L cls Calculate according to formula (4): P is the probability predicted by the model belonging to class c. By weighted summation of three loss functions, the loss function for object detection can be obtained.

6. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 3, characterized in that For the pose detection task, in addition to the basic object detection loss function, a new loss function is also involved, as shown in Equation (4): L structure is the skeleton loss of the model, which is used to judge the pose of the mass block, and represent the true distance and the predicted distance between point k and point I respectively.

7. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 1 or 2, characterized in that The energy storage control unit adjusts the operation parameters according to the real-time monitoring data of load changes, energy demands, and environmental conditions, sets operations according to the energy demands to adjust the energy storage release speed to meet demand fluctuations; the energy storage control unit is set to maximize the energy capture and storage efficiency, automatically switches to an energy-saving mode during low-load periods, and sets a refined control logic for energy feedback to ensure the maximum recovery and reuse of energy at any operation stage.

8. The gravity power generation and energy storage system based on the linear motor and the deep learning multi-task model according to claim 1, characterized in that, The front load-carrying car (21) and the rear load-carrying car (22) are respectively provided with height sensors. The central controller (51) is signal-connected to the height sensors and is configured to receive and control the operation of the linear motor (23) according to the car height detected by the height sensors to perform the motion control of the front load-carrying car (21) and the rear load-carrying car (22). The front track section is provided with a front travel sensor for detecting the up and down travel of the front load-carrying car (21), and the rear track section is provided with a rear travel sensor for detecting the up and down travel of the rear load-carrying car (22). The central controller (51) is signal-connected to the height sensors and is configured to receive and control the operation of the linear motor (23) according to the car height detected by the height sensors to perform the motion control of the front load-carrying car (21) and the rear load-carrying car (22).

9. The gravity power generation and energy storage system based on the linear motor and the deep learning multi-task model according to claim 8, characterized in that, The system further includes a braking and regenerating unit that is braking-connected to the front load-carrying car (21) and the rear load-carrying car (22). The braking and regenerating unit includes a mounting base plate, a rope pulley (41), a speed reducer (42), a generator (43), and a brake (44). The mounting base plate is fixedly installed directly above the front vertical track and the rear vertical track. The motor shaft of the generator (43) is connected to the axle of the rope pulley (41) through the speed reducer (42). The flexible rope end fixedly wound on the rope pulley (41) extends vertically downward and is connected to the load-carrying car. The brake (44) is braking-connected to the axle of the rope pulley (41).

10. The gravity power generation and energy storage system based on a linear motor and a deep learning multi-task model according to claim 9, characterized in that, The transmission driving mechanism includes an electromagnetic push-pull device arranged inside the car. The electromagnetic push-pull device includes an electromagnet (33), a guide track (34), and a polarity control circuit. When the weight (32) moves to the specified height on the vertical track, the electromagnetic push-pull device is activated, and the electromagnet (33) switches to a repulsive polarity to push the weight (32) out of the car; when the weight (32) moves to the end position on the mirror side through the conveyor belt (31), the electromagnetic push-pull device on the other side of the car is activated, and the electromagnet (33) switches to an attractive polarity to pull the weight (32) into the car on the mirror side.

Images