A new energy electric drive system of a novel green intelligent carrying frame

Abstract

The application discloses a new energy electric drive system of a novel green intelligent carrying frame, which comprises a multi-source energy supply module, an intelligent driving module, an energy efficiency control module, an energy recovery module and a communication interaction module.The multi-source energy supply module is used for supplying energy under different working conditions for the carrying frame.The intelligent driving module is used for driving walking mechanisms arranged on both sides of the carrying frame.The energy efficiency control module is used for generating a dynamic energy consumption adjustment decision by using an energy consumption optimization algorithm and sending the decision to the multi-source energy supply module and the intelligent driving module.The energy recovery module is used for monitoring the running state of the carrying frame in real time to start an energy recovery process and output electric energy to the multi-source energy supply module.The communication interaction module is used for establishing an information transmission channel between the intelligent driving module, the energy efficiency control module and a warehouse intelligent scheduling system.The novel green intelligent carrying frame has the remarkable effects that the green environmental protection, energy utilization efficiency and intelligent operation capacity of the carrying frame are remarkably improved through multi-element new energy collaborative power supply, full-working-condition high-efficiency driving, intelligent energy efficiency control and precise energy recovery.

Description

Technical Field

[0001] This invention relates to the field of electric drive technology for material handling equipment, and specifically to a new energy electric drive system for a novel green and intelligent material handling rack. Background Technology

[0002] As logistics warehousing, industrial production and other fields accelerate their transformation towards green and intelligent directions, the environmental performance, energy utilization efficiency and intelligent management level of the power system of the material handling rack, as the core equipment for material transfer, have become key factors restricting technological upgrades. Existing electric drive systems for transport racks mostly adopt a single lithium battery power supply combined with a traditional motor drive architecture, which has obvious technical shortcomings: First, the energy source is singular, relying excessively on lithium battery energy storage. The range is rigidly limited by the battery capacity, and the charge-discharge cycle life of lithium batteries is limited. Long-term large-scale use can easily lead to waste battery pollution problems, which contradicts the concept of green development. Second, the motor drive efficiency is low. The energy conversion efficiency of traditional AC motors or ordinary permanent magnet synchronous motors is generally below 90%, and there is a lack of dynamic power adjustment mechanisms for the variable load conditions of the transport rack, resulting in a large amount of energy waste. Third, the intelligent control capability is weak. Existing systems can only realize basic start-stop and speed control, and cannot implement dynamic energy efficiency optimization based on the working scenario, load changes, and energy status. It is also difficult to deeply coordinate with the intelligent scheduling system of the warehouse, which seriously restricts the intelligent operation efficiency of the transport rack. Fourth, the energy recovery mechanism is imperfect. Only simple energy recovery can be achieved during the braking phase. The recovery efficiency is low, and a collaborative control system with the energy storage system has not been formed, which further reduces the overall energy utilization rate.

[0003] Therefore, developing a new type of electric drive system that integrates diversified green energy supply, efficient intelligent drive, and energy optimization under all operating conditions has become an urgent need to address the shortcomings of existing technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this invention is to provide a new type of green intelligent handling rack with a new energy electric drive system. This system aims to achieve coordinated power supply from multiple new energy sources, efficient drive under all operating conditions, intelligent energy efficiency management and precise energy recovery, significantly improving the green environmental protection, energy utilization efficiency and intelligent operation capabilities of the handling rack, and meeting the high-efficiency and low-carbon operation requirements of modern logistics and warehousing.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A novel green intelligent handling rack's new energy electric drive system, its key features include a multi-source energy supply module, an intelligent drive module, an energy efficiency management module, an energy recovery module, and a communication interaction module, wherein: The multi-source energy supply module is used to provide energy for the transport rack under different operating scenarios, and to provide working power for the intelligent drive module, energy efficiency management module, energy recovery module and communication interaction module. The intelligent drive module is used to achieve stepless speed change and precise torque output by using a vector control algorithm to drive the walking mechanism located on both sides of the transport frame. The energy efficiency management module is used to collect data on the load weight and lifting torque of the transport rack, and, in conjunction with the operation scenario map of the transport rack, uses an energy consumption optimization algorithm to generate dynamic energy consumption adjustment decisions and send them to the multi-source energy supply module and the intelligent drive module. The energy recovery module is used to monitor the operating status of the transport rack in real time to start the energy recovery process and output electrical energy to the multi-source energy supply module. The communication interaction module is used to establish an information transmission channel between the intelligent drive module, the energy efficiency management module, and the warehouse intelligent scheduling system.

[0006] Furthermore, the multi-source energy supply module includes: A flexible photovoltaic array is installed on top of the transport frame to provide power for low-power operation scenarios of the transport frame; Lithium-ion battery packs are used to provide the main driving force for the transport rack; An energy coordination controller is used to dynamically allocate the output ratio of each energy unit using a fuzzy control algorithm.

[0007] Furthermore, the step of the energy co-controller dynamically allocating the output ratio of each energy unit using a fuzzy control algorithm includes: Step A1: Parameter Acquisition; Step A2: Fuzzification processing, converting the collected parameters into fuzzy variables; Step A3: Establish core fuzzy rules, infer fuzzy variables based on Mamdani inference method, and output fuzzy sets of power supply ratios of each energy unit; Step A4: The fuzzy set is converted into a precise power supply ratio using the center of gravity method. The output power of each energy unit is adjusted by the PWM controller, and the adjustment is fed back in real time to ensure that the power supply ratio is dynamically matched with the operating conditions.

[0008] Furthermore, the intelligent drive module includes a permanent magnet synchronous motor, a power control unit, a distributed drive controller, and a transmission mechanism; wherein: The permanent magnet synchronous motor unit adopts a dual-motor distributed layout, which drives the walking mechanisms on both sides of the transport frame through the transmission mechanism. The distributed drive controller is used to dynamically adjust the control strategy of the permanent magnet synchronous motor set. The power control unit uses a vector control algorithm based on the control strategy to achieve stepless speed change and precise torque output of the permanent magnet synchronous motor.

[0009] Furthermore, the power control unit employs a vector control algorithm to achieve stepless speed regulation and precise torque output of the permanent magnet synchronous motor, including the following steps: Step B1: Collect the three-phase stator current, rotor position and speed of the permanent magnet synchronous motor using a high-precision Hall sensor, and filter the collected current signal. Step B2: Coordinate transformation decoupling, transforming the complex three-phase AC control into independent DC control on the dq axis; Step B3: Receive the speed command and torque command output by the energy efficiency management module, and construct speed closed-loop and torque closed-loop control respectively; Step B4: Voltage limiting and SVPWM modulation to achieve stepless speed change and precise torque output.

[0010] Furthermore, the energy efficiency management module includes a core control unit, a load detection unit, and a scene perception unit, wherein: The load detection unit is used to collect data on the load weight and lifting torque of the transport frame through the coordinated use of pressure and torque sensors. The scene perception unit is used to collect the positioning data of the transport frame in collaboration with the lidar and the inertial measurement unit, and to build a high-precision operation scene map in combination with the SLAM algorithm. The core control unit is used to dynamically adjust decisions based on changes in the work scenario, load, and energy status according to the data collected by the load detection unit and the scene perception unit, and optimize the unit energy consumption efficiency under different working conditions.

[0011] Furthermore, the core control unit dynamically adjusts its decisions based on changes in the work scenario, load, and energy status according to an energy consumption optimization algorithm. The steps for optimizing the unit energy consumption efficiency under different working conditions include: Step C1: Initialize the parameters of the improved DQN neural network model and simultaneously collect multi-dimensional state parameters; Step C2: The improved DQN neural network outputs three types of decision instructions by fitting the improved action value function: operation mode, motor power adjustment coefficient, and path planning correction amount, while calling the built-in working condition rule library to correct the decision. Step C3: The core control unit sends decision commands to the energy co-controller and intelligent drive module to perform corresponding power adjustment, mode switching and path correction, while collecting energy consumption data after execution in real time; Step 4: Calculate the reward value based on the unit energy consumption operation efficiency. If the efficiency meets the standard, a positive reward is given; if the equipment is abnormal, a negative reward is given. Backpropagate the reward value and the state vector of the next time step to the neural network to update the weight parameters. Step C5: Continuously cycle through steps 1-4, dynamically adjusting decisions based on changes in the work scenario, load, and energy status.

[0012] Furthermore, the formula for calculating the reward value is as follows: Reward value = k1 × (carrying weight × distance / power consumption) - k2 × (motor overheating time + battery degradation rate) Where k1 and k2 are weighting coefficients.

[0013] Furthermore, the energy efficiency management module also includes a fault prediction unit, which is used to predict potential faults of the transport rack in advance.

[0014] Furthermore, the energy recovery module includes a full-condition energy recovery controller, a braking recovery unit, a lifting recovery unit, and a charging / discharging regulation unit, wherein: The full-condition energy recovery controller is used to monitor the operating status of the transport frame in real time and automatically start the energy recovery process under conditions such as braking, downhill, lifting and lowering. The braking recovery unit is used to convert the kinetic energy recovered during braking into electrical energy through a motor reverse drag power generation method; The lifting and recovery unit is used to convert the gravitational potential energy released during the lowering of the heavy object into electrical energy through a hydraulic-electrical energy conversion mechanism. The charge / discharge regulation unit is used to preferentially store the recovered electrical energy into the supercapacitor energy storage unit, and the excess electrical energy is used to replenish the lithium battery pack.

[0015] The significant effects of this invention are: 1. The system described in this invention adopts a multi-energy collaborative power supply architecture of photovoltaic, lithium battery and supercapacitor, which realizes energy self-sufficiency under low power conditions, greatly reduces the dependence on grid power supply and lithium battery energy storage, and reduces carbon emissions; at the same time, with the combination of tiered charging and discharging and battery health management technology, the cycle life of lithium battery is extended, effectively reducing waste battery pollution, which is in line with the concept of green development.

[0016] 2. The system described in this invention uses a high-efficiency permanent magnet synchronous motor driven by a silicon carbide power module to improve the energy conversion efficiency to over 95%; combined with a full-condition energy recovery mechanism covering braking, downhill, and heavy load lowering scenarios, the overall energy utilization rate of the system is improved by more than 30% compared with the existing technology, and the range of the transport rack is extended by 20%-30%.

[0017] 3. The system described in this invention is based on a reinforcement learning-based energy consumption optimization algorithm and high-precision scene perception technology, which realizes dynamic power allocation and optimal path planning under multiple working conditions. It can flexibly adapt to the operational needs of different loads and different warehousing scenarios. At the same time, it supports deep collaboration with the intelligent warehousing scheduling system to realize multi-vehicle collaborative operation and dynamic task allocation, and the operation efficiency is improved by more than 35% compared with the traditional electric drive handling rack.

[0018] 4. The system described in this invention integrates distributed dual-motor drive, triple braking protection, liquid cooling and heat dissipation and fault prediction technology, which not only realizes early warning and can accurately predict multiple potential faults, but also effectively improves the system's operational stability under heavy load and continuous operation scenarios, reducing the failure rate by more than 23%; moreover, the blockchain log recording and dual-link communication mechanism ensure that the operation process and fault handling are traceable and controllable, fully meeting industrial-grade safety standards. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a schematic diagram of the structure of a multi-source energy supply module; Figure 3 This is a schematic diagram of the structure of the intelligent drive module; Figure 4 This is a structural schematic diagram of the energy efficiency management module. Figure 5 This is a schematic diagram of the energy recovery module. Detailed Implementation

[0020] The specific embodiments and working principles of the present invention will be further described in detail below with reference to the accompanying drawings.

[0021] Example: like Figure 1 As shown, a new type of green intelligent handling rack's new energy electric drive system includes a multi-source energy supply module, an intelligent drive module, an energy efficiency management module, an energy recovery module, and a communication interaction module, wherein: The multi-source energy supply module is used to provide energy for the transport rack under different operating scenarios, and to provide working power for the intelligent drive module, energy efficiency management module, energy recovery module and communication interaction module. The intelligent drive module is used to achieve stepless speed change and precise torque output by using a vector control algorithm to drive the walking mechanism located on both sides of the transport frame. The energy efficiency management module is used to collect data on the load weight and lifting torque of the transport rack, and, in conjunction with the operation scenario map of the transport rack, uses an energy consumption optimization algorithm to generate dynamic energy consumption adjustment decisions and send them to the multi-source energy supply module and the intelligent drive module. The energy recovery module is used to monitor the operating status of the transport rack in real time to start the energy recovery process and output electrical energy to the multi-source energy supply module. The communication interaction module is used to establish an information transmission channel between the intelligent drive module, the energy efficiency management module, and the warehouse intelligent scheduling system.

[0022] See appendix Figure 2 In specific implementation, the multi-source energy supply module includes: A flexible photovoltaic array is installed on top of the transport frame to provide power for low-power operation scenarios of the transport frame; Lithium-ion battery packs are used to provide the main driving force for the transport rack; Supercapacitor energy storage unit, used to achieve rapid charge and discharge buffering; An energy coordination controller is used to dynamically allocate the output ratio of each energy unit using a fuzzy control algorithm.

[0023] Preferably, the flexible photovoltaic array uses high-efficiency heterojunction photovoltaic cells with a daily power generation of 5-8 kWh to meet the energy self-sufficiency requirements of the transport rack in low-power operation scenarios; the lithium battery pack uses ternary lithium batteries with a rated capacity of not less than 100 Ah; the peak power density of the supercapacitor energy storage unit can reach 5 kW / L, which can effectively alleviate the current surge of the lithium battery pack under heavy load start-up, rapid acceleration and other conditions, and extend the service life of the lithium battery.

[0024] In some preferred embodiments, the step of the energy coordination controller dynamically allocating the output ratio of each energy unit using a fuzzy control algorithm includes: Step A1: Real-time acquisition of four core parameters, including flexible photovoltaic array output power / voltage, lithium battery pack SOC and SOH, supercapacitor voltage / remaining capacity, and real-time load and operating conditions of the transport frame. Among them, the operating conditions include start-up / constant speed / acceleration / braking. The acquisition frequency is uniformly 50Hz to ensure data timeliness and multi-module synchronization. Step A2: Fuzzy processing. The collected parameters are converted into fuzzy variables. Photovoltaic power is divided into three levels: low (<2kW), medium (2-5kW), and high (>5kW). Lithium battery SOC is divided into three levels: low (<30%), medium (30%-70%), and high (>70%). Load is divided into three levels: light load (<500kg), medium load (500-1500kg), and heavy load (>1500kg). The precise values ​​are mapped to fuzzy subsets by using a triangular membership function. Step A3: Establish 50 core fuzzy rules. The core fuzzy rules can be: when photovoltaic power is high, lithium battery SOC is high, and under light load, photovoltaic power supply accounts for 70%-80%, lithium battery is on standby (0%), and supercapacitor is on standby; or when photovoltaic power is low, lithium battery SOC is medium, and under heavy load startup, lithium battery power supply accounts for 60%, supercapacitor power supply accounts for 40%, and photovoltaic supplementary power supply accounts for ≤10%. Based on the Mamdani inference method, infer the fuzzy variables and output the fuzzy set of power supply proportion of each energy unit. It's important to note that the Mamdani inference method is one of the most commonly used inference methods in fuzzy control, particularly suitable for designing fuzzy controllers. It simulates the human fuzzy reasoning process, converting precise input quantities into fuzzy linguistic variables, and then inferring using a set of "if-then" rules to ultimately output a precise control quantity. The core process of this method includes four main steps: fuzzification, establishing an application rule base, inference, and defuzzification, which transforms the synthesized fuzzy output set into a final, precise output value. The most commonly used defuzzification method in the Mamdani inference method is the centroid method, which determines the final output by calculating the centroid of the fuzzy output set. This is a weighted average method that fully utilizes information from all rules.

[0025] Step A4: The fuzzy set is converted into a precise power supply ratio using the center of gravity method. The output power of each energy unit is adjusted by the PWM controller, and the adjustment is fed back in real time to ensure that the power supply ratio is dynamically matched with the operating conditions.

[0026] During implementation, the multi-source energy supply module is also equipped with a modular liquid cooling system, which uses microchannel temperature equalization technology to control the operating temperature of the lithium battery pack and supercapacitor energy storage unit to within 120°C, significantly improving the operational stability and service life of the energy unit.

[0027] like Figure 3 As shown, the intelligent drive module includes a permanent magnet synchronous motor, a power control unit, a distributed drive controller, and a transmission mechanism; wherein: The permanent magnet synchronous motor unit adopts a dual-motor distributed layout, which drives the walking mechanisms on both sides of the transport frame through the transmission mechanism. The distributed drive controller is used to dynamically adjust the control strategy of the permanent magnet synchronous motor set. The power control unit uses a vector control algorithm based on the control strategy to achieve stepless speed change and precise torque output of the permanent magnet synchronous motor.

[0028] In the specific implementation process, the power control unit uses a vector control algorithm to achieve stepless speed change and precise torque output of the permanent magnet synchronous motor, including the following steps: Step B1: Acquire the three-phase stator current, rotor position and speed of the permanent magnet synchronous motor using a high-precision Hall sensor at a frequency of 5kHz. The acquired current signal is then filtered using a second-order Butterworth low-pass filter to eliminate high-frequency interference, ensure data accuracy, and provide reliable basic data for vector control. Step B2, coordinate transformation and decoupling: First, the three-phase stator current is transformed from the ABC stationary coordinate system to the αβ two-phase stationary coordinate system through Clark transformation (3 / 2 transformation), eliminating the coupling relationship of the three-phase current. The zero-sequence component is ignored during the transformation to simplify the calculation. Then, the current in the αβ coordinate system is transformed into the dq synchronous rotating coordinate system (d-axis is the direction of rotor flux linkage, and q-axis is perpendicular to d-axis) through Park transformation (synchronous rotation transformation). The complex three-phase AC control is transformed into independent DC control on the dq axis, realizing the decoupled regulation of flux linkage and torque. Step B3: Receive the speed command and torque command output by the energy efficiency management module, and construct speed closed-loop and torque closed-loop control respectively: The speed loop inputs the deviation between the actual speed and the command value into the dq-axis PI regulator and outputs the d-axis current command; the torque loop adjusts the q-axis current command according to the torque deviation. The proportional coefficient of the speed loop PI regulator is set to 0.8 and the integral coefficient is set to 0.05, and the proportional coefficient of the torque loop PI regulator is set to 1.2 and the integral coefficient is set to 0.1. At the same time, an anti-integral saturation mechanism is added to avoid overshoot and ensure system stability. Step B4: Voltage Limitation and SVPWM Modulation. Based on the rated voltage of the silicon carbide power module, the amplitude of the dq axis voltage command is limited to prevent overvoltage damage to the module. Then, the limited dq axis voltage command is generated into 6 drive pulse signals through space vector pulse width modulation to control the on / off state of the 6 switches of the silicon carbide power module. By adjusting the pulse duty cycle, the stator voltage of the motor is continuously and smoothly adjusted, thereby achieving stepless speed change and precise torque output. Step B5: Dynamic compensation and safety protection. Real-time monitoring of motor temperature (°C), stator current (A), and rotor speed. When the temperature exceeds 120°C or the current exceeds 1.2 times the rated value, the torque output is dynamically reduced through a vector control algorithm, with the reduction magnitude adjusted linearly according to the degree of abnormality. At the same time, the modular liquid cooling system and triple braking protection mechanism are triggered. For special working conditions such as heavy-load start-up and turning, a cross-coupling compensation term is added to offset the load difference between the two motors, ensuring that a smooth acceleration and deceleration performance of 0.1m / s can be maintained under a full load of 2000kg. When turning, the speed difference between the two motors is adjusted through dual-motor differential vector control.

[0029] See appendix Figure 4The energy efficiency management module includes a core control unit, a load detection unit, and a scene perception unit, wherein: The load detection unit is used to collect data on the load weight and lifting torque of the transport frame through the coordinated use of pressure and torque sensors. The scene perception unit is used to collect the positioning data of the transport frame through the collaborative acquisition of lidar and inertial measurement unit, to achieve centimeter-level environment modeling and sub-centimeter-level positioning, and to build a high-precision operation scene map by combining SLAM algorithm; The core control unit adopts an industrial-grade embedded processor, which supports real-time parallel processing of data from multiple modules. It is used to dynamically adjust decisions based on changes in the work scenario, load, and energy status according to the data collected by the load detection unit and the scene perception unit, and optimize the unit energy consumption efficiency under different working conditions.

[0030] In this example, the energy consumption optimization algorithm takes "maximizing the efficiency of work per unit of energy consumption (material handling weight × distance completed per unit kWh, i.e., kg·m / kWh)" as its core optimization objective and constructs a three-layer architecture: state layer, decision layer, and feedback layer. In specific implementation, the core control unit dynamically adjusts its decisions based on changes in the work scenario, load, and energy status according to the energy consumption optimization algorithm. The steps for optimizing the efficiency of work per unit of energy consumption under different operating conditions include: Step C1: Initialize the parameters of the improved DQN neural network model and synchronously collect multi-dimensional state parameters, including load weight, path slope, real-time photovoltaic power, lithium battery SOC, supercapacitor voltage, task priority (high / medium / low), real-time motor power / speed, and path information between the current position and the target position, to form a 12-dimensional state vector; In this example, the improved DQN neural network makes three core improvements to the traditional DQN for the energy consumption optimization scenario of the transport rack. First, it introduces a dual experience replay mechanism, constructing a "normal working condition experience pool" and an "extreme working condition (heavy load climbing, low power operation) experience pool" respectively, and samples them for training at a 7:3 ratio to avoid data dilution in extreme scenarios and improve model robustness. Second, it optimizes the target network update strategy, adopting an adaptive update cycle, updating once every 100 steps in normal working conditions and once every 50 steps in extreme working conditions, while adding a momentum term (momentum coefficient set to 0.9) to accelerate convergence and solve the update lag problem of traditional DQN. Third, it improves the action value function, introducing an energy consumption penalty term and an efficiency incentive term, correcting the bias of the traditional Q(s,a) function, so that the output decision is more in line with the core objective of "optimal energy consumption". The neural network contains 3 hidden layers, with 64, 32 and 16 neurons in each layer, respectively, and the activation function is the ReLU function, with an iteration step size of 0.001. Step C2: The improved DQN neural network outputs three types of decision instructions by fitting the improved action value function: operation mode (heavy load high efficiency / light load energy saving / standby hibernation), motor power adjustment coefficient (0.4-1.2 times rated power), and path planning correction amount (±0.5 meters path offset). At the same time, it calls the built-in working condition rule library covering eight typical scenarios such as no-load return trip and heavy load climbing to correct the decision and improve reliability. Step C3: The core control unit sends decision commands to the energy co-controller and intelligent drive module to execute corresponding power adjustment, mode switching and path correction, while collecting energy consumption data, operation efficiency data and equipment operating parameters in real time after execution; Step C4: Calculate the reward value based on the unit energy consumption operation efficiency. If the efficiency meets the standard, a positive reward is given; if the equipment is abnormal, a negative reward is given. Backpropagate the reward value and the state vector of the next time step to the neural network to update the weight parameters and realize the model self-optimization. The rule base is iteratively updated once every 10 operation tasks are completed. In this example, the formula for calculating the reward value is: Reward value = k1 × (carrying weight × distance / power consumption) - k2 × (motor overheating time + battery degradation rate) Where k1 and k2 are weighting coefficients, set to 0.6 and 0.4 respectively.

[0031] Step C5: Continuously cycle through steps 1-4, dynamically adjusting decisions based on changes in the work scenario, load, and energy status to ensure optimal unit energy consumption efficiency under all working conditions.

[0032] In this example, the energy efficiency management module also includes a fault prediction unit, which is used to predict potential faults in the handling rack in advance. The prediction process of the fault prediction unit is as follows: Step D1: Relying on the parallel processing capability of the core control unit, three types of core data are collected simultaneously, including motor operation data such as stator current, rotor speed, vibration spectrum, and winding temperature, which are consistent with the vector control algorithm; lithium battery SOC decay rate, charging and discharging voltage fluctuation, and internal resistance change, which are synchronized with the energy consumption optimization algorithm; and vibration amplitude of precision transmission mechanism, hydraulic system pressure fluctuation, and brake gap change, which are transmission and structural data. At the same time, the historical fault database is accessed to construct a sample dataset. Step D2: Perform noise reduction filtering on the collected data. Wavelet threshold noise reduction is used for motor vibration data, and moving average filtering is used for battery data to remove abnormal interference data. Subsequently, multi-dimensional fault features were extracted. On the motor side, features such as current harmonic components, vibration peak factor / kurtosis, and temperature gradient were extracted. On the battery side, features such as voltage ripple coefficient and internal resistance growth rate were extracted. On the transmission side, features such as vibration main frequency offset and pressure fluctuation coefficient were extracted, ultimately forming a 29-dimensional fault feature vector. Step D3: Use a CNN+LSTM fusion machine learning model. The CNN module is used to extract spatial features such as vibration spectrum distribution and current waveform features from the feature vector, while the LSTM module is used to mine temporal features such as the changing trend of parameters with the duration of operation. The model input is a 29-dimensional feature vector, and the output is the fault type and fault development cycle. Using historical fault data and real-time collected data, we constructed a training set (70%), a validation set (20%), and a test set (10%). We iteratively trained and optimized the model parameters, setting the learning rate to 0.001 and the number of iterations to 500. We also added a dropout mechanism to prevent overfitting. Step D4: The model receives the preprocessed feature data at a frequency of 50Hz, outputs the failure probability and remaining safe operating time in real time, and sets a three-level early warning threshold. Level 1 warning (failure probability 30%-50%), predicting 72 hours of remaining safe operating time, prompting preparations for inspection; Level 2 warning (fault probability 50%-80%) predicts the remaining safe operating time of 24 hours and triggers local parameter adjustments such as reducing motor load and optimizing charging and discharging strategies; Level 3 warning (fault probability > 80%): If the remaining safe operating time is predicted to be ≤ 6 hours, an immediate shutdown and maintenance order will be issued. Step D5: The early warning information is uploaded to the warehouse intelligent scheduling system in real time through the communication interaction module, and is also written into the blockchain log recording unit, marking the fault characteristics, predicted duration and handling suggestions, to ensure full traceability and effectively reduce the probability of unplanned downtime.

[0033] like Figure 5 As shown, the energy recovery module includes a full-condition energy recovery controller, a braking recovery unit, a lifting recovery unit, and a charging / discharging regulation unit, wherein: The full-condition energy recovery controller is used to monitor the operating status of the transport frame in real time and automatically start the energy recovery process under conditions such as braking, downhill, lifting and lowering. The braking recovery unit is used to convert the kinetic energy recovered during braking into electrical energy through a motor reverse drag power generation method; The lifting and recovery unit is used to convert the gravitational potential energy released during the lowering of the heavy object into electrical energy through a hydraulic-electrical energy conversion mechanism. The charge and discharge regulation unit is used to preferentially store the recovered electrical energy into the supercapacitor energy storage unit, and use the excess electrical energy to replenish the lithium battery pack, while effectively suppressing charging harmonics.

[0034] In this example, the communication interaction module adopts a dual-link communication architecture of 5G slicing technology and industrial Ethernet. This module supports standardized API interfaces and can realize two-way interaction of job task reception, operation status upload, remote debugging and fault early warning information. At the same time, it integrates a blockchain-style log recording unit to store the operation trajectory of energy scheduling, drive control, safety instructions and other operations in an immutable manner, providing reliable support for accident tracing.

[0035] Application example: The new energy electric drive system of the novel green intelligent handling rack described in the embodiment is applied to the heavy-load handling scenario of a large-scale warehousing and logistics center. It is designed with a load capacity of 2000kg and can meet the needs of continuous 24-hour operation. Specifically: In the multi-source energy supply module, the flexible photovoltaic array uses heterojunction photovoltaic cells with a laying area of ​​8㎡ and an average daily power generation of 6.5kWh, which can meet the energy needs of low-power conditions such as standby and light-load translation of the transport rack; the lithium battery pack uses 120Ah ternary lithium batteries with a rated voltage of 48V, and is equipped with a 500F supercapacitor energy storage unit; under heavy-load start-up conditions, the energy coordination controller adjusts the output ratio of the supercapacitor to 40%, effectively mitigating the current surge of the lithium battery, and after startup, it switches to a coordinated mode of lithium battery main power supply (70%) and photovoltaic supplementary power (30%); the liquid cooling heat dissipation system stabilizes the battery pack operating temperature at 80-100℃ through a microchannel structure, ensuring the stability of continuous system operation.

[0036] The intelligent drive module employs two 15kW permanent magnet synchronous motors, distributed on both sides of the handling rack. A silicon carbide power module, combined with a vector control algorithm, achieves precise drive. Specifically, the system works as follows: Motor operating parameters are acquired via Hall effect sensors at a 5kHz frequency. After current decoupling through Clark and Park transformations, a PI regulator controls the dq-axis current in a closed loop. SVPWM modulation outputs drive pulses (switching frequency 20kHz) to achieve stepless speed regulation from 0.05 to 1.5 m / s. In a narrow storage aisle with a width of 3 meters, dual-motor differential vector control dynamically adjusts the speed difference between the two motors (maximum 50 r / min), and a cross-coupling compensation term is added to offset load differences, enabling flexible steering with a 1.5-meter turning radius. Torque fluctuations during turning are controlled within ±5 N·m. The resampling triple braking mechanism has a response time of only 80 milliseconds in dangerous scenarios such as sudden personnel approach. At this time, the vector control algorithm synchronously triggers the motor's anti-drag braking, which, combined with mechanical and hydraulic braking, controls the full-load braking distance to 2.8 meters, fully complying with ISO3691-4 safety standards.

[0037] The energy efficiency management module collects load weight in real time through pressure sensors, and LiDAR and IMU work together to construct a 3D map of the warehouse. The aforementioned deep reinforcement learning energy consumption optimization algorithm is based on four core parameters: "load weight - path distance - remaining energy - path slope", combined with a preset working condition rule base to dynamically optimize the operation plan. The specific implementation is as follows: The algorithm uses an improved DQN neural network, which accurately adapts to different working scenarios through a dual experience playback mechanism. In extreme working conditions such as heavy load (1500-2000kg) or climbing (slope >5°), it prioritizes sampling the trained parameters from the extreme working condition experience pool to output decisions, with a response delay ≤10ms. Through the vector control algorithm, the motor torque output is increased by 20% (corresponding to a power adjustment coefficient of 1.2 and a rated power of 15kW), and instructions are simultaneously sent to the energy co-controller to make it operate according to the mode. The fuzzy control rules allocate full power supply to photovoltaic and lithium batteries (70% lithium battery, 30% photovoltaic) to ensure sufficient power for operation. Under normal working conditions with light load (below 500kg) or horizontal straight path (slope within ±1°), the algorithm reduces the motor power adjustment coefficient to 0.4-0.6 based on conventional experience pool parameters. The control vector control algorithm reduces the motor speed and torque, and at the same time instructs the energy co-controller to prioritize the allocation of photovoltaic clean energy to extend its lifespan. When the standby time exceeds 5 minutes, the algorithm triggers the standby sleep mode, controls the power cut-off of non-core components, and only retains the low-power operation of sensors and communication modules, reducing energy consumption by 80% compared to the working state. During the sleep period, task instructions are collected every 100ms, and the response wake-up delay is ≤300ms. After waking up, the algorithm initialization and status acquisition steps are quickly executed to resume normal operation.

[0038] The energy recovery module achieves a recovery efficiency of 86% under braking conditions. In a scenario where a 1000kg heavy object is lowered, the gravitational potential energy is recovered through a hydraulic-electric energy conversion mechanism, and the converted electrical energy can meet the needs of a 30-meter horizontal movement operation of the transport frame. The recovered electrical energy is preferentially stored in a supercapacitor for rapid release during the next startup, and excess electrical energy is used to replenish the lithium battery pack. The system can recover an average of 2.2kWh of electrical energy per day.

[0039] The communication module connects to the warehouse intelligent scheduling system via 5G slicing and industrial Ethernet dual links, receiving real-time operational tasks such as shelf numbers and target locations, and simultaneously uploading operational status data such as energy consumption, load, and location, as well as fault prediction and early warning information. The scheduling system dynamically allocates tasks based on the operational status of multiple vehicles, effectively avoiding path conflicts, and prioritizes maintenance tasks for equipment with level 2 and above early warnings. The blockchain log recording unit stores daily energy scheduling, fault information, operation instructions, and fault prediction trajectories, providing data support for subsequent fault analysis and model optimization.

[0040] Based on actual working condition tests, the electric drive system described in this embodiment reduces overall energy consumption by 32%, extends driving range by 28%, improves operating efficiency by 36%, reduces failure rate by 25%, and has no exhaust emissions compared to existing single lithium battery electric drive systems. The operating noise is controlled below 60 decibels, fully meeting the actual needs of green and intelligent handling in large-scale warehousing and logistics centers.

[0041] In summary, the system described in this invention adopts a multi-energy collaborative power supply architecture of photovoltaics, lithium batteries, and supercapacitors, achieving energy self-sufficiency under low-power conditions. This significantly reduces reliance on grid power and lithium battery energy storage, thereby reducing carbon emissions. Simultaneously, the combination of tiered charging and discharging and battery health management technologies effectively extends the cycle life of lithium batteries, significantly reducing waste battery pollution and fully aligning with the green development concept. Furthermore, based on reinforcement learning-based energy consumption optimization algorithms and high-precision scene perception technology, it achieves dynamic power allocation and optimal path planning under multiple operating conditions, flexibly adapting to the operational needs of different loads and storage scenarios. Finally, through the integrated application of distributed dual-motor drive, triple braking protection, liquid cooling, and fault prediction technologies, it not only achieves early warning and accurately predicts various potential faults, effectively improving the system's operational stability under heavy-load and continuous operation scenarios and reducing the failure rate, but also ensures that the operation process and fault handling are traceable and controllable through blockchain log recording and dual-link communication mechanisms, fully meeting industrial-grade safety standards.

[0042] The technical solution provided by this invention has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make several improvements and modifications to this invention without departing from the principles of this invention, and these improvements and modifications also fall within the protection scope of the claims of this invention.

Claims

1. A new energy electric drive system for a novel green intelligent handling rack, characterized in that, It includes a multi-source energy supply module, an intelligent drive module, an energy efficiency management module, an energy recovery module, and a communication interaction module, among which: The multi-source energy supply module is used to provide energy for the transport rack under different operating scenarios, and to provide working power for the intelligent drive module, energy efficiency management module, energy recovery module and communication interaction module. The intelligent drive module is used to achieve stepless speed change and precise torque output by using a vector control algorithm to drive the walking mechanism located on both sides of the transport frame. The energy efficiency management module is used to collect data on the load weight and lifting torque of the transport rack, and, in conjunction with the operation scenario map of the transport rack, uses an energy consumption optimization algorithm to generate dynamic energy consumption adjustment decisions and send them to the multi-source energy supply module and the intelligent drive module. The energy recovery module is used to monitor the operating status of the transport rack in real time to start the energy recovery process and output electrical energy to the multi-source energy supply module. The communication interaction module is used to establish an information transmission channel between the intelligent drive module, the energy efficiency management module, and the warehouse intelligent scheduling system.

2. The new energy electric drive system of the novel green intelligent handling rack according to claim 1, characterized in that: The multi-source energy supply module includes: A flexible photovoltaic array is installed on top of the transport frame to provide power for low-power operation scenarios of the transport frame; Lithium-ion battery packs are used to provide the main driving force for the transport rack; An energy coordination controller is used to dynamically allocate the output ratio of each energy unit using a fuzzy control algorithm.

3. The new energy electric drive system of the novel green intelligent handling rack according to claim 2, characterized in that: The steps of the energy co-controller dynamically allocating the output ratio of each energy unit using a fuzzy control algorithm include: Step A1: Parameter Acquisition; Step A2: Fuzzification processing, converting the collected parameters into fuzzy variables; Step A3: Establish core fuzzy rules, infer fuzzy variables based on Mamdani inference method, and output fuzzy sets of power supply ratios of each energy unit; Step A4: The fuzzy set is converted into a precise power supply ratio using the center of gravity method. The output power of each energy unit is adjusted by the PWM controller, and the adjustment is fed back in real time to ensure that the power supply ratio is dynamically matched with the operating conditions.

4. The new energy electric drive system of the novel green intelligent handling rack according to claim 1, characterized in that: The intelligent drive module includes a permanent magnet synchronous motor, a power control unit, a distributed drive controller, and a transmission mechanism; wherein: The permanent magnet synchronous motor unit adopts a dual-motor distributed layout, which drives the walking mechanisms on both sides of the transport frame through the transmission mechanism. The distributed drive controller is used to dynamically adjust the control strategy of the permanent magnet synchronous motor set. The power control unit uses a vector control algorithm based on the control strategy to achieve stepless speed change and precise torque output of the permanent magnet synchronous motor.

5. The new energy electric drive system of the novel green intelligent handling rack according to claim 4, characterized in that: The power control unit uses a vector control algorithm to achieve stepless speed regulation and precise torque output of the permanent magnet synchronous motor, including the following steps: Step B1: Collect the three-phase stator current, rotor position and speed of the permanent magnet synchronous motor using a high-precision Hall sensor, and filter the collected current signal. Step B2: Coordinate transformation decoupling, transforming the complex three-phase AC control into independent DC control on the dq axis; Step B3: Receive the speed command and torque command output by the energy efficiency management module, and construct speed closed-loop and torque closed-loop control respectively; Step B4: Voltage limiting and SVPWM modulation to achieve stepless speed change and precise torque output.

6. The new energy electric drive system of the novel green intelligent handling rack according to claim 1, characterized in that: The energy efficiency management module includes a core control unit, a load detection unit, and a scene perception unit, wherein: The load detection unit is used to collect data on the load weight and lifting torque of the transport frame through the coordinated use of pressure and torque sensors. The scene perception unit is used to collect the positioning data of the transport frame in collaboration with the lidar and the inertial measurement unit, and to build a high-precision operation scene map in combination with the SLAM algorithm. The core control unit is used to dynamically adjust decisions based on changes in the work scenario, load, and energy status according to the data collected by the load detection unit and the scene perception unit, and optimize the unit energy consumption efficiency under different working conditions.

7. The new energy electric drive system of the novel green intelligent handling rack according to claim 6, characterized in that: The core control unit dynamically adjusts its decisions based on changes in the work scenario, load, and energy status using an energy consumption optimization algorithm. The steps for optimizing the unit energy consumption efficiency under different working conditions include: Step C1: Initialize the parameters of the improved DQN neural network model and simultaneously collect multi-dimensional state parameters; Step C2: The improved DQN neural network outputs three types of decision instructions by fitting the improved action value function: operation mode, motor power adjustment coefficient, and path planning correction amount, while calling the built-in working condition rule library to correct the decision. Step C3: The core control unit sends decision commands to the energy co-controller and intelligent drive module to perform corresponding power adjustment, mode switching and path correction, while collecting energy consumption data after execution in real time; Step 4: Calculate the reward value based on the unit energy consumption operation efficiency. If the efficiency meets the standard, a positive reward is given; if the equipment is abnormal, a negative reward is given. Backpropagate the reward value and the state vector of the next time step to the neural network to update the weight parameters. Step C5: Continuously cycle through steps 1-4, dynamically adjusting decisions based on changes in the work scenario, load, and energy status.

8. The new energy electric drive system of the novel green intelligent handling rack according to claim 7, characterized in that: The formula for calculating the reward value is: Reward value = k1 × (carrying weight × distance / power consumption) - k2 × (motor overheating time + battery degradation rate) Where k1 and k2 are weighting coefficients.

9. The new energy electric drive system of the novel green intelligent handling rack according to any one of claims 6-8, characterized in that: The energy efficiency management module also includes a fault prediction unit, which is used to predict potential faults of the transport rack in advance.

10. The new energy electric drive system of the novel green intelligent handling rack according to claim 1, characterized in that: The energy recovery module includes a full-condition energy recovery controller, a braking recovery unit, a lifting recovery unit, and a charging / discharging regulation unit, wherein: The full-condition energy recovery controller is used to monitor the operating status of the transport frame in real time and automatically start the energy recovery process under conditions such as braking, downhill, lifting and lowering. The braking recovery unit is used to convert the kinetic energy recovered during braking into electrical energy through a motor reverse drag power generation method; The lifting and recovery unit is used to convert the gravitational potential energy released during the lowering of the heavy object into electrical energy through a hydraulic-electrical energy conversion mechanism. The charge / discharge regulation unit is used to preferentially store the recovered electrical energy into the supercapacitor energy storage unit, and the excess electrical energy is used to replenish the lithium battery pack.

Images