Method for reducing energy consumption of intelligent sorting system

Abstract

The invention discloses a method for reducing energy consumption of an intelligent sorting system. The method comprises the following steps that S1, material physical characteristic parameters and equipment operation energy consumption data are collected in real time through a sensing device; s2, constructing a multi-objective optimization model based on the collected data, and solving through an optimization algorithm to obtain cooperative control parameters; s3, according to the cooperative control parameters, adjusting the output power of an execution mechanism by adopting a power adjusting technology, and synchronously controlling the operation parameters of an identification unit; and S4, monitoring the deviation between the actual energy consumption value and the theoretical optimization value of the execution mechanism in real time, and when the deviation exceeds a preset deviation threshold value, triggering to recalculate the cooperative control parameter. According to the method, a sensing-decision-execution-feedback closed-loop control process is constructed, real-time collection of material characteristics and energy consumption data, multi-objective optimization algorithm solving, dynamic power adjustment and feedback correction are organically combined, and collaborative optimization of energy consumption and efficiency in the sorting process is achieved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control and energy-saving optimization technology, and specifically relates to a method for reducing energy consumption in an intelligent sorting system. Background Technology

[0002] Currently, intelligent sorting systems are mainly used in express logistics, renewable resource recycling, and food processing, and generally adopt a three-stage open-loop structure of "identification-sorting-conveyance". In the renewable resource recycling scenario, traditional sorting systems mostly rely on a combination of conveyor belts and robotic arms driven by fixed power. The equipment operating parameters are preset and cannot be dynamically adjusted according to the characteristics of the materials, resulting in high energy consumption even under no-load or light-load conditions. There is a lack of standardized communication interfaces between sorting equipment from different manufacturers, and core modules such as vision recognition systems, robotic arm controllers, and conveyor belt drivers cannot achieve real-time sharing of energy consumption data, forming information silos.

[0003] Existing technologies overemphasize improving sorting accuracy, neglecting the energy consumption characteristics of the algorithms themselves, resulting in low overall system energy efficiency. Current sorting systems generally adopt a "rated power + fixed cycle time" operating mode. The motor drive module always maintains its maximum designed power output, unable to dynamically adjust according to real-time material flow. Even during periods of sparse material flow, it remains under high load, resulting in wasted energy. The vision recognition unit uses a continuous inference mode, maintaining high load even when no material is passing through, generating significant wasted energy. Furthermore, the lack of standardized communication interfaces between sorting equipment from different manufacturers prevents real-time sharing of energy consumption data. Module control strategies are independent; for example, an emergency stop of the robotic arm does not trigger a deceleration command on the conveyor belt, leading to instantaneous power surges, increased equipment wear, and low collaborative efficiency.

[0004] Existing technologies cannot differentiate energy consumption control based on the physical characteristics of materials. They still operate with uniform power parameters for materials of different specifications. When dealing with changes in material mixing, parameters need to be manually recalibrated, which is time-consuming and puts the system in a high-energy-consumption state. Existing technologies focus too much on improving sorting accuracy and neglect energy consumption optimization of algorithm inference. The model input resolution is not dynamically adjusted according to material size, and the feature extraction network is not lightweighted, resulting in wasted computing resources and high energy consumption per unit sorting volume. The braking energy of traditional sorting systems is not effectively recovered. The kinetic energy generated during the movement of the robotic arm and the deceleration of the conveyor belt is converted into heat energy and wasted through braking resistors. At the same time, the system is not equipped with energy storage devices and cannot cope with instantaneous power peaks, resulting in increased reactive power loss on the grid side.

[0005] Therefore, the present invention provides a method for reducing energy consumption in an intelligent sorting system to solve the problems existing in the background art. Summary of the Invention

[0006] In view of the problems mentioned in the background technology above, the purpose of this invention is to provide a method for reducing energy consumption in an intelligent sorting system, thereby achieving synergistic optimization of energy consumption and efficiency during the sorting process.

[0007] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows: A method for reducing energy consumption in an intelligent sorting system includes the following steps: S1: Real-time collection of material physical characteristic parameters and equipment operating energy consumption data through sensing devices; S2: Construct a multi-objective optimization model based on the collected data, and obtain the cooperative control parameters by solving the optimization algorithm; S3: Based on the aforementioned collaborative control parameters, power regulation technology is used to adjust the output power of the actuator, and the operating parameters of the identification unit are controlled synchronously. S4: Monitor the deviation between the actual energy consumption value and the theoretical optimized value of the actuator in real time. When the deviation exceeds the preset deviation threshold, trigger the recalculation of the collaborative control parameters.

[0008] Further specified, the energy consumption data acquisition in S1 uses a signal processing algorithm to remove noise; the sensing device includes a vibration sensor, which is arranged at the joint of the robotic arm.

[0009] Furthermore, the optimization algorithm in S2 introduces an adaptive crossover probability operator and adopts a fitness function based on energy consumption sensitivity.

[0010] Furthermore, the power regulation technology in S3 employs space vector pulse width modulation technology, switching carrier modes according to material characteristics.

[0011] Furthermore, the preset deviation threshold in S4 is adjusted using a dynamic adjustment mechanism.

[0012] Furthermore, the window size of the signal processing algorithm is dynamically adjusted according to the material throughput speed.

[0013] Furthermore, the energy consumption sensitivity function incorporates a temperature correction coefficient to prioritize reducing energy consumption output under specific operating conditions.

[0014] Further, it also includes an offline model pre-training step: using transfer learning to transfer energy consumption model parameters from other scenarios to the target scenario; the offline model pre-training process uses knowledge distillation technology to transfer knowledge from the high-performance model to the lightweight model.

[0015] Furthermore, S2 also includes an energy consumption prediction submodule, which uses a neural network to predict future energy consumption trends and initiate adjustment commands in advance; the input features of the energy consumption prediction submodule include multi-dimensional parameters, and the humidity parameter is collected by a humidity sensor.

[0016] Further specifying, the sensing module includes a lidar, which is used to acquire three-dimensional spatial information of the material; the decision module includes an edge computing unit and an industrial control computer, the edge computing unit is used to run a lightweight recognition model, and the industrial control computer is used to execute an energy consumption optimization algorithm; The execution module includes a servo motor driver and a robotic arm controller. The servo motor driver is used to adjust the output power of the execution mechanism, and the robotic arm controller is used to control the movement of the robotic arm.

[0017] Further specifying, the communication module employs time-sensitive networking technology to achieve time synchronization between modules; the energy management module includes an energy recovery unit for recovering energy generated during the braking process of the actuator; and the decision-making module includes a coprocessing unit for accelerating the calculation process of the optimization algorithm.

[0018] Furthermore, the lidar and vision sensor in the perception module are calibrated using an external parameter calibration method.

[0019] An energy-saving system for intelligent sorting systems includes: The sensing module is used to collect material characteristic data and equipment operating energy consumption data; The decision module is connected to the sensing module via a data communication link, and is used to receive data collected by the sensing module, run energy consumption optimization algorithms, and generate collaborative control parameters. The execution module is connected to the decision module via a real-time communication network and is used to receive control parameters output by the decision module and drive the execution mechanism to perform sorting actions. The communication module is used to realize data interaction and time synchronization between the perception module, decision-making module, and execution module; The energy management module is connected to the sensing module, decision-making module, and execution module respectively, and is used to collect energy consumption data from each module to realize energy recovery and energy consumption metering. The sensing module, decision-making module, execution module, communication module, and energy management module form a closed-loop control system through the communication module, realizing dynamic optimization of energy consumption in the sorting process.

[0020] The beneficial effects of this invention are: The method described in this invention constructs a closed-loop control process of perception-decision-execution-feedback, organically combining real-time acquisition of material characteristics and energy consumption data, multi-objective optimization algorithm solving, dynamic power adjustment, and feedback correction to achieve synergistic optimization of energy consumption and efficiency during the sorting process. This method overcomes the limitations of fixed operating parameters in traditional open-loop sorting systems. By adaptively adjusting the control strategy, it effectively avoids ineffective energy consumption output under idling or light-load conditions, while ensuring that sorting accuracy is not affected. It addresses the technological gap in existing technologies that prioritize efficiency over energy consumption, balancing the dual needs of reducing energy consumption and improving sorting efficiency. By dynamically adjusting the parameters of the actuators and the operating strategy of the identification unit, it achieves energy efficiency management throughout the entire process.

[0021] The system described in this invention, through modular design and standardized communication interfaces, achieves collaborative operation of the sensing module, decision-making module, execution module, communication module, and energy management module, breaking down information silos between different devices. The multi-dimensional data acquisition of the sensing module provides accurate basis for decision-making; the hierarchical computing architecture of the decision-making module balances rapid local processing with global optimization; the dynamic power adjustment of the execution module enables on-demand energy supply; the energy recovery function of the energy management module further improves energy utilization; and efficient communication between modules ensures real-time transmission and execution of control commands, significantly improving the overall energy efficiency of the system. The system's redundancy design and protection mechanisms enhance operational reliability, and the introduction of an environmental adaptation module enables the system to cope with complex operating conditions, reducing energy waste and equipment failures caused by environmental factors.

[0022] The technical solution of this invention possesses excellent cross-scenario adaptability and can be seamlessly integrated into existing sorting lines, reducing the complexity and cost of system transformation. It is not only applicable to renewable resource recycling scenarios but can also be extended to automated production lines in logistics warehousing, food processing, and other fields, achieving energy efficiency improvements in different scenarios through a unified energy consumption control logic. Simultaneously, the system's lightweight algorithm design and edge computing application reduce algorithm inference energy consumption and improve response speed while ensuring recognition accuracy. Through data linkage and user interaction functions, the technical solution enhances the synergy of each link and user participation, comprehensively improving the economic and social benefits of the sorting system and providing a complete solution for energy saving and efficiency improvement in the field of intelligent sorting. Attached Figure Description

[0023] The present invention can be further illustrated by the non-limiting embodiments given in the accompanying drawings; Figure 1 This is a flowchart illustrating the steps of an embodiment of the energy consumption reduction method for an intelligent sorting system according to the present invention. Figure 2 This is a system block diagram of an embodiment of an energy consumption reduction method for an intelligent sorting system according to the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments. The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0025] like Figure 1 As shown, a method for reducing energy consumption in an intelligent sorting system according to the present invention includes the following steps: S1: Real-time collection of material physical characteristic parameters and equipment operating energy consumption data through sensing devices; S2: Construct a multi-objective optimization model based on the collected data, and obtain the cooperative control parameters by solving the optimization algorithm; S3: Based on the aforementioned collaborative control parameters, power regulation technology is used to adjust the output power of the actuator, and the operating parameters of the identification unit are controlled synchronously. S4: Monitor the deviation between the actual energy consumption value and the theoretical optimized value of the actuator in real time. When the deviation exceeds the preset deviation threshold, trigger the recalculation of the collaborative control parameters.

[0026] Specifically, current intelligent sorting systems are mainly used in express logistics, renewable resource recycling, and food processing, and generally adopt a three-stage open-loop structure of "identification-sorting-conveyance". In the renewable resource recycling scenario, traditional sorting systems mostly rely on a combination of conveyor belts and robotic arms driven by fixed power. The equipment operating parameters are preset and cannot be dynamically adjusted according to the characteristics of the materials, resulting in high energy consumption even under no-load or light-load conditions. There is a lack of standardized communication interfaces between sorting equipment from different manufacturers. Core modules such as vision recognition systems, robotic arm controllers, and conveyor belt drivers cannot achieve real-time sharing of energy consumption data, forming information silos. Energy consumption data cannot be shared in real time. Existing technologies focus too much on improving sorting accuracy and ignore the energy consumption characteristics of the algorithm itself, resulting in low overall system energy efficiency. The motor drive module maintains rated power output, and the vision recognition unit continuously infers, causing a large amount of ineffective energy consumption. Braking energy is not effectively recovered, and instantaneous power peaks lead to increased reactive power loss in the power grid.

[0027] This invention utilizes a multispectral visual sensor and a vibration energy consumption sensor to collect real-time physical characteristic parameters of materials and energy consumption data of equipment operation. The physical characteristic parameters include material volume, weight, density, and surface texture. Based on the collected data, a multi-objective optimization model is constructed, and a collaborative control parameter is solved using an optimization algorithm. Then, pulse width modulation (PWM) technology is used to adjust the output power of the actuator according to the collaborative control parameter, while simultaneously controlling the operating parameters of the identification unit. Furthermore, the deviation between actual energy consumption and theoretical optimized values ​​is monitored in real time. When the deviation exceeds a threshold (when the deviation exceeds a preset threshold ±5%), the collaborative control parameter is recalculated. This invention achieves dynamic energy consumption optimization through a closed-loop logic of "collection-optimization-regulation-feedback," which helps break down information silos, dynamically adjust operating parameters, reduce ineffective energy consumption, and improve system energy efficiency.

[0028] In the practical application of this embodiment, the energy consumption data acquisition in S1 uses a signal processing algorithm to remove noise; the sensing device includes a vibration sensor, which is arranged at the joint of the robotic arm.

[0029] Specifically, the energy consumption data acquisition of this invention employs a 16-bit AD conversion module with a sampling frequency of 2kHz, and removes high-frequency noise using a sliding window filtering algorithm. The sensing device includes a piezoelectric vibration sensor for acquiring vibration energy consumption signals. These vibration energy consumption signals reflect the equipment's operating status, and the filtering algorithm improves data accuracy, thereby reducing noise interference, accurately acquiring energy consumption data, and providing reliable input for subsequent optimization.

[0030] In the practical application of this embodiment, the optimization algorithm in S2 introduces an adaptive crossover probability operator and adopts a fitness function based on energy consumption sensitivity.

[0031] Specifically, the optimization algorithm in S2 is an improved NSGA-III multi-objective optimization algorithm, which introduces an adaptive crossover probability operator. For example, when the population convergence is >0.85, the crossover probability linearly decreases from 0.7 to 0.4. An energy-sensitive fitness function is adopted, defined as the ratio of the rate of change in energy consumption to the rate of change in sorting efficiency. By dynamically adjusting the operator to balance exploration and utilization capabilities, the fitness function takes into account both energy consumption and efficiency; thus avoiding local optima, improving optimization efficiency, and achieving coordinated optimization of energy consumption and sorting efficiency.

[0032] In the practical application of this embodiment, the power regulation technology in S3 adopts space vector pulse width modulation technology and switches the carrier mode according to the material characteristics.

[0033] Specifically, the power regulation technology in S3 employs Space Vector Pulse Width Modulation (SVPWM) with a carrier frequency dynamic range of 5kHz-15kHz. When the material weight is less than a preset threshold, it switches to a low-frequency carrier mode. By dynamically adjusting the carrier frequency, switching losses are reduced, thereby reducing the ineffective energy consumption of the motor drive module and extending the equipment's lifespan.

[0034] In the practical application of this embodiment, the preset deviation threshold in S4 is adjusted using a dynamic adjustment mechanism.

[0035] Specifically, the preset deviation threshold in S4 is dynamically adjusted by a fuzzy PID controller; when the system is stable for three consecutive sampling cycles, the threshold is relaxed to ±8%. This reduces unnecessary optimization calculations, lowers algorithm energy consumption, and improves system response speed.

[0036] In the practical application of this embodiment, the window size of the signal processing algorithm is dynamically adjusted according to the material throughput speed.

[0037] Specifically, the signal processing algorithm uses a sliding window filtering algorithm, with the window size dynamically adjusted according to the material's speed. When the conveyor belt speed is >1m / s, the window size is reduced from 512 points to 256 points. By balancing real-time performance and filtering effect, it ensures data real-time performance at high speeds and improves filtering accuracy at low speeds.

[0038] In the practical application of this embodiment, the energy consumption sensitivity function introduces a temperature correction coefficient to prioritize reducing energy consumption output under specific operating conditions.

[0039] Specifically, the energy consumption sensitivity function incorporates a temperature correction coefficient; for example, when the motor winding temperature is >65℃, the sensitivity weight is increased by 30%. By prioritizing the reduction of energy consumption under high load conditions, the equipment is protected, and the increase in energy consumption caused by overheating is reduced.

[0040] In practical applications of this embodiment, an offline model pre-training step is also included: using transfer learning to transfer energy consumption model parameters from other scenarios to the target scenario; the offline model pre-training process uses knowledge distillation technology to transfer knowledge from the high-performance model to the lightweight model.

[0041] Specifically, transfer learning is used to transfer the energy consumption model parameters from the industrial sorting scenario to the target scenario; knowledge distillation technology is employed during the pre-training process to transfer knowledge from the high-performance model to the lightweight model. This involves using existing data to shorten convergence time, and using a lightweight model to reduce inference energy consumption; by reducing online training time, the energy consumption of the recognition unit is reduced while maintaining recognition accuracy.

[0042] In the practical application of this embodiment, S2 also includes an energy consumption prediction submodule, which uses a neural network to predict future energy consumption trends and initiate adjustment commands in advance; the input features of the energy consumption prediction submodule include multi-dimensional parameters, and the humidity parameter is collected by a humidity sensor.

[0043] Specifically, the energy consumption prediction submodule receives input features including historical energy consumption sequences, ambient temperature, material humidity, and grid voltage fluctuations; humidity parameters are collected via a capacitive humidity sensor. By supplementing these multi-dimensional parameters, prediction accuracy is improved, thereby optimizing energy consumption prediction precision and further reducing energy consumption.

[0044] An energy-saving system for intelligent sorting systems includes: The sensing module is used to collect material characteristic data and equipment operating energy consumption data; The decision module is connected to the sensing module via a data communication link, and is used to receive data collected by the sensing module, run energy consumption optimization algorithms, and generate collaborative control parameters. The execution module is connected to the decision module via a real-time communication network and is used to receive control parameters output by the decision module and drive the execution mechanism to perform sorting actions. The communication module is used to realize data interaction and time synchronization between the perception module, decision-making module, and execution module; The energy management module is connected to the sensing module, decision-making module, and execution module respectively, and is used to collect energy consumption data from each module to realize energy recovery and energy consumption metering. The sensing module, decision-making module, execution module, communication module, and energy management module form a closed-loop control system through the communication module, realizing dynamic optimization of energy consumption in the sorting process.

[0045] Specifically, the sensing module uses piezoelectric vibration sensors and capacitive humidity sensors to collect material characteristics and equipment energy consumption data. The decision module uses an embedded processor, which is connected to the sensing module via an industrial Ethernet network and runs an improved NSGA-III algorithm to generate collaborative control parameters. The execution module uses an SVPWM driver and is connected to the decision module via a real-time communication network to drive the robotic arm and conveyor belt to perform sorting actions. The communication module uses an industrial Ethernet switch to realize data interaction and time synchronization between modules. The energy management module uses an energy recovery device connected to each module to collect energy consumption data and realize braking energy recovery. All modules form a closed-loop control system through the communication module.

[0046] By performing the following steps in sequence: the sensing module collects data, the decision-making module performs optimization calculations, the execution module controls the process, and the energy management module provides feedback, dynamic energy consumption optimization is achieved; thereby breaking down information silos, coordinating various modules to reduce energy consumption, and improving system energy efficiency.

[0047] In the practical application of this embodiment, the sensing module includes a lidar, which is used to acquire three-dimensional spatial information of the material; the decision module includes an edge computing unit and an industrial control computer, the edge computing unit is used to run a lightweight recognition model, and the industrial control computer is used to execute an energy consumption optimization algorithm. In the practical application of this embodiment, the lidar and vision sensor in the sensing module are calibrated using a calibration method to ensure external parameters are corrected. Specifically, the sensing module includes a 16-line lidar with a horizontal resolution of 0.2°, a scanning frequency of 10Hz, and an effective detection distance of 0.1-10m. It is installed 1.8m above the sorting conveyor belt, 50cm horizontally from the multispectral camera, to acquire three-dimensional point cloud data of the material. Its working principle involves emitting a laser beam and receiving reflected signals to calculate the material's spatial coordinates, volume, and shape information. This information is then fused with the surface texture features acquired by the multispectral camera to form a multi-dimensional feature vector. The beneficial effects include accurately identifying the material's spatial location and physical properties, avoiding ineffective gripping actions by the robotic arm, reducing idle energy consumption, improving sorting accuracy, and reducing additional energy consumption during secondary sorting.

[0048] The decision-making module includes an edge computing unit and an industrial control computer. The edge computing unit uses an NVIDIA Jetson AGX Xavier processor to run a lightweight recognition model, while the industrial control computer is equipped with an Intel Core i7-10750H processor to run an improved NSGA-III optimization algorithm. The working principle is that the edge computing unit processes image data locally and outputs the material category, while the industrial control computer generates collaborative control parameters based on the recognition results and energy consumption data, forming a hierarchical architecture of local fast processing + global optimization. The beneficial effects are that edge computing reduces transmission latency and energy consumption, while industrial control ensures optimal overall energy efficiency, improves response speed, and reduces energy consumption while waiting for materials.

[0049] The execution module includes a servo motor driver and a robotic arm controller. The servo motor driver is used to adjust the output power of the execution mechanism, and the robotic arm controller is used to control the movement of the robotic arm.

[0050] Specifically, the execution module includes a servo motor driver and a robotic arm controller. The servo motor driver uses an EtherCAT bus to adjust the output power, and the robotic arm controller controls the joint angle and timing. The working principle is that the driver receives PWM signals and dynamically adjusts the motor voltage and frequency through SVPWM technology, while the controller plans the optimal path based on three-dimensional coordinates to achieve on-demand power supply. The beneficial effects are to avoid rated power output under light load and no-load conditions, reduce motor copper and iron losses, reduce the stroke and time of movement, improve sorting efficiency, and reduce operating energy consumption.

[0051] In the practical application of this embodiment, the communication module adopts time-sensitive networking technology to achieve time synchronization between modules; the energy management module includes an energy recovery unit to recover energy generated during the braking process of the actuator; and the decision module includes a coprocessing unit to accelerate the calculation process of the optimization algorithm.

[0052] Specifically, the communication module uses a TSN switch (supporting the IEEE 802.1AS protocol) to achieve time synchronization with a synchronization accuracy of ±1μs. The working principle is to ensure low-latency transmission of real-time signals through priority queues and time scheduling, and clock synchronization ensures that the action sequence of each module is consistent. The beneficial effects are to avoid action conflicts, reduce the energy consumption of repeated operations, improve collaborative efficiency, and reduce waiting energy consumption.

[0053] Specifically, the energy management module includes an energy recovery unit containing a 500F supercapacitor (24-72V) and a bidirectional DC / DC converter. The working principle is that the motor generates electricity when the actuator brakes, and the electrical energy is stored in the supercapacitor through the converter. It is released to supply power when the actuator accelerates. The beneficial effect is that the braking energy is recovered and recycled, reducing the dependence on the power grid, avoiding heat energy waste and reducing heat dissipation energy consumption.

[0054] Specifically, the overlap rate of the field of view of the lidar and the multispectral camera is ≥30%, and it is calibrated by Zhang Zhengyou calibration method (error ≤0.5 pixels); the working principle is to capture chessboard image and calculate rotation matrix and translation vector, and unify coordinate system to ensure data space consistency; the beneficial effect is to improve the accuracy of multi-sensor fusion, reduce the energy consumption of invalid actions caused by recognition errors, and reduce the energy consumption of secondary sorting.

[0055] Specifically, the decision module includes an improved NSGA-III algorithm accelerated by an FPGA coprocessor. The working principle is that the FPGA implements the parallel computing part (population initialization, crossover mutation) through hardware logic, which shortens the optimization time. The beneficial effect is that the computing time is reduced from 200ms to 85ms, the dynamic response speed is improved to adapt to material changes, and the load on the industrial control computer and its own energy consumption are reduced.

[0056] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for reducing energy consumption of an intelligent sorting system, characterized in that, The method comprises the following steps: S1: collecting physical characteristic parameters of materials and equipment operation energy consumption data in real time through a sensing device; S2: constructing a multi-objective optimization model based on the collected data, and obtaining collaborative control parameters by solving the optimization algorithm; S3: adjusting the output power of the actuator by power regulation technology according to the collaborative control parameters, and synchronously controlling the operating parameters of the identification unit; S4: monitoring the deviation between the actual energy consumption value and the theoretical optimized value of the actuator in real time, and triggering the recalculation of the collaborative control parameters when the deviation exceeds the preset deviation threshold. 2.The method for reducing energy consumption of an intelligent sorting system according to claim 1, characterized in that: The energy consumption data collection in S1 adopts a signal processing algorithm to remove noise; the sensing device includes a vibration sensor arranged at the joint of the mechanical arm; The optimization algorithm in S2 introduces an adaptive crossover probability operator and uses an energy consumption sensitivity-based fitness function. 3.The method of Claim 1, wherein: The power regulation technology in S3 uses space vector pulse width modulation technology and switches the carrier mode according to the material characteristics; The preset deviation threshold in S4 is adjusted by a dynamic adjustment mechanism. 4.The energy consumption reduction method of an intelligent sorting system according to claim 1, characterized in that: The window size of the signal processing algorithm is dynamically adjusted according to the material passing speed; The energy consumption sensitivity function introduces a temperature correction coefficient to preferentially reduce energy consumption output under specific working conditions.

5. The method of claim 1, wherein the intelligent sorting system is a system for sorting mail. It also includes an offline model pre-training step: using a transfer learning method to migrate energy consumption model parameters from other scenarios to the target scenario; the offline model pre-training process uses a knowledge distillation technology to migrate the knowledge of high-performance models to lightweight models.

6. The method of claim 1, wherein the intelligent sorting system is a system for sorting mail. The S2 also includes an energy consumption prediction submodule that uses a neural network to predict future energy consumption trends and initiate adjustment instructions in advance; the input features of the energy consumption prediction submodule include multi-dimensional parameters, and the humidity parameter is collected by a humidity sensor.

7. An intelligent sorting system energy reduction system, characterized by, It comprises: a perception module for collecting material characteristic data and equipment operation energy consumption data; a decision module connected to the perception module through a data communication link, for receiving the data collected by the perception module, running an energy consumption optimization algorithm and generating collaborative control parameters; an execution module connected to the decision module through a real-time communication network, for receiving the control parameters output by the decision module and driving the actuator to perform sorting actions; a communication module for realizing data interaction and time synchronization among the perception module, decision module, execution module; an energy management module connected to the perception module, decision module, execution module respectively, for collecting energy consumption data of each module and realizing energy recovery and energy consumption metering; The perception module, decision module, execution module, communication module and energy management module form a closed-loop control system through the communication module to realize dynamic optimization of energy consumption in the sorting process.

8. The energy reduction system for an intelligent sorting system of claim 7, wherein: The perception module includes a laser radar for obtaining three-dimensional spatial information of materials; the decision module includes an edge computing unit and an industrial control computer, the edge computing unit is used to run a lightweight identification model, and the industrial control computer is used to execute an energy consumption optimization algorithm; The execution module includes a servo motor driver and a mechanical arm controller, the servo motor driver is used to adjust the output power of the actuator, and the mechanical arm controller is used to control the action of the mechanical arm.

9. The energy reduction system for an intelligent sorting system of claim 7, wherein: The communication module adopts a time-sensitive network technology and is used for realizing time synchronization among modules; the energy management module comprises an energy recovery unit and is used for recovering energy generated in a braking process of an actuator; and the decision module comprises a coprocessor unit and is used for accelerating a calculation process of an optimization algorithm.

10. The energy reduction system for an intelligent sorting system of claim 8, wherein: The laser radar and the visual sensor in the perception module are calibrated by a calibration method.

Images