A remote vehicle power-up control method for a construction machine

Abstract

The present application belongs to the technical field of vehicle control, and particularly relates to a remote whole vehicle power-on control method for engineering machinery, which comprises the following steps: constructing an electrical-thermal-mechanical coupling digital twin of the engineering machinery, and completing the power supply loop configuration of the main battery, the backup battery and the relay control module; a remote control module establishes a secure communication channel with a remote control system through a quantum key distribution protocol and generates a quantum key; the present application solves the problem of blindness in starting caused by the lack of state prediction in the traditional "hard start" mode by constructing an electrical-thermal-mechanical coupling digital twin and combining the non-steady-state heat conduction equation and the stress balance equation for state prediction. The optimal starting vector is solved through multi-objective Pareto optimization, and multi-physical field collaborative preheating is carried out in cooperation with federated learning edge computing, so that the present application can avoid the low-temperature large-current impact on the battery and mechanical thermal stress damage.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, specifically relating to a remote power-on control method for engineering machinery. Background Technology

[0002] With the development of the Internet of Things and remote control technology, remote power-on and operation of construction machinery has become an industry trend. However, existing remote control technologies have significant shortcomings in energy efficiency management and system security under complex working conditions (such as low temperature, electromagnetic interference, and network instability), making it difficult to meet the needs of intelligent high-end equipment.

[0003] Traditional remote start control for construction machinery typically employs a simple "hard start" mode or logic control based on fixed timing. These solutions lack the ability to finely sense and predict the multi-physics (electro-thermal-mechanical) coupling states of critical components (such as power batteries and mechanical structures). In low-temperature environments, battery internal resistance increases and activity decreases; blindly starting with high current not only leads to excessive energy consumption but may also cause irreversible damage to the battery cells and the risk of thermal runaway. Simultaneously, the non-uniform thermal stress generated by drastic temperature changes in mechanical components accelerates material fatigue and shortens equipment lifespan. Furthermore, in open-air, multi-equipment construction scenarios, communication links are susceptible to interference from complex electromagnetic environments, posing security risks such as eavesdropping, tampering, or denial-of-service attacks, challenging traditional encryption and authentication methods. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides a remote vehicle power-on control method for engineering machinery, comprising: S1. Construct an electro-thermal-mechanical coupled digital twin for engineering machinery and complete the power supply circuit configuration for the main battery, backup battery, and relay control module; S2. The remote control module establishes a secure communication channel with the remote control system and generates a quantum key through a quantum key distribution protocol. S3. Receive and verify remote power-on commands stored on the blockchain based on the communication channel, complete command hash value and digital signature verification, perform multi-physics field state prediction based on digital twin, solve the optimal start-up vector through multi-objective Pareto optimization and complete feasibility verification. S4. Based on the optimal start-up vector output in step S3, combined with the battery temperature field and stress field output in real time by the digital twin in step S1, the heating device is controlled to perform multi-physics field collaborative preheating; and federated learning edge computing is used to optimize the heating parameters and calculate the cumulative energy consumption and equipment life loss index in real time. S5. Based on the preheating completion signal in step S4 and the optimal start-up timing within the optimal start-up vector in step S3, output a pull-in signal to drive the dual redundant relays to close, so that the main battery is connected to the vehicle; establish a virtual-real synchronization mechanism based on the digital twin in S1, and use Kalman filtering to fuse the state and compensate for the system response delay; S6. Based on the synchronization state after compensation in step S5, control the power system to complete the vehicle start-up, and make the hydraulic and working devices ready; based on the local optimization parameters and optimization contribution output by federated learning edge computing in step S4, calculate the local contribution of this round of operation, and use hierarchical compression coding to upload the state data through the quantum encryption link in step S2. S7. After receiving the stop command, retrieve the full-process data that has been stored on the blockchain in step S6, and automatically calculate the global energy efficiency index and carbon footprint audit report for this operation.

[0005] Further improvements to this technical solution include step S1, which includes: S101. Construct the electro-thermal-mechanical multi-physics coupled control equations for the engineering machinery to be controlled. The control equations include the unsteady heat conduction equation for simulating the evolution of the battery temperature field and the stress balance equation for calculating the stress field distribution of the engineering machinery. The control equations are spatially discretized and time-integrated using the finite element method. S102. Connect the positive and negative terminals of the main battery to the DC bus of the vehicle circuit through the contacts of the main relay and the backup relay in the relay control module. At the same time, connect the positive and negative terminals of the backup battery directly to the power management chip of the remote control module to provide the remote control module with a continuous power supply independent of the vehicle power supply. S103. Initialize key system parameters, including generating an initial sequence of entangled photon pairs for quantum key distribution to build a quantum key pool, creating a copy of the blockchain network's genesis block in local storage to establish a trusted ledger baseline, and downloading the initial global model weights from the federated learning server. The remote control module serves as prior knowledge for the local federated learning model, and the weight vector for multi-objective optimization is set. Used to quantify the relative importance of energy consumption, time, and lifetime loss in the optimization step S3.

[0006] Further improvements to this technical solution include the control equations in step S101, which serve as the core of digital twin calculation, specifically including: S1011. Establish the coupled unsteady-state heat conduction control equations: ; And the stress balance governing equations: ; in, The density of the material; is the specific heat capacity at constant pressure; T is the temperature field; t is time; k(T) is the thermal conductivity related to temperature; I represents the electrochemical reaction heat generation rate; I represents the current; SOC represents the battery state of charge. This refers to the battery's internal resistance. This refers to the volume of a single battery cell; denoted as Cauchy stress tensor; b represents volume force. For displacement field; S1012. Discretize the governing equations using the finite element method to construct the global heat capacity matrix. Heat conduction matrix Mass matrix M and stiffness matrix And assemble them to obtain the discrete system equations:

[0007] and ; Where T and U are the nodal temperature and displacement vectors, respectively; These are the nodal load vectors for chemical reaction heat and Joule heat, respectively; This is the equivalent nodal load vector for thermal stress. , e is the unit number, Let B be the element domain, B be the strain-displacement matrix, and D be the elasticity matrix. The coefficient of thermal expansion is The temperature shape function matrix; S1013. The implicit Newmark-β method is used to perform time integration on the discretized system equations. Within each simulation step Δt, the nonlinear coupled equations are solved through Newton-Raphson iteration, and the three-dimensional temperature field distribution T(x,y,z,t) inside the battery pack is output in real time. Mises equivalent force field distribution This data is then used as input for step S4.

[0008] Further improvements to this technical solution include step S2, which includes: S201. The remote control module uses the entangled photon pairs in the quantum key pool initialized in step S1 to perform a quantum key distribution process based on the BB84 protocol with the remote control system, and negotiates to generate a dynamic quantum key specific to this session. And utilize dynamic quantum key distribution Establish high-level secure communication channels; S202. In standby mode, the remote control module monitors the Received Signal Strength Indicator (RSSI) and packet loss rate of the wireless communication link in real time. And calculate the overall communication channel quality index: ; Where RSSI is the sliding window average signal strength; Sensitivity coefficient; Reference signal strength; S203, Electromagnetic interference intensity spectral density of the frequency band where the radio frequency front-end of the synchronous monitoring remote control module is located. And calculate the equivalent electromagnetic interference intensity: ; in, , These are the minimum and maximum operating frequencies for wireless communication of the remote control module, respectively. Based on the comprehensive communication channel quality index With equivalent electromagnetic interference intensity Dynamically adjust the heartbeat packet sending cycle:

[0009] With radio frequency transmit power ; in, This is the weighting coefficient for adjusting the heartbeat cycle based on communication channel quality. This is the adjustment weighting coefficient of electromagnetic interference intensity on the heartbeat cycle; This is a preset reference value for electromagnetic interference intensity; The minimum permissible radio frequency transmit power for the remote control module; This represents the maximum permissible radio frequency transmit power for the remote control module. This is the exponential attenuation coefficient of electromagnetic interference on the transmitted power.

[0010] Further improvements to this technical solution include step S3, which includes: S301. The remote control module receives the remote power-on command ciphertext from the remote control system through the quantum-encrypted secure channel established in step S2, and uses the dynamic quantum key generated in step S2. Decrypt to obtain the instruction hash value. With digital signatures The original instructions; S302. Call the local blockchain ledger copy established in step S1 to verify the instruction hash value. Has it been stored in the blockchain network consensus and verified using the public key corresponding to the remote control system? Verify digital signature The effectiveness; S303. After the instruction verification is successful, drive the digital twin constructed in step S1 to perform multiphysics coupling simulation using the current physical state read by the sensor as the initial condition, and predict the battery temperature field sequence within the preset time window [t, t+ΔT]. and stress field sequence ; S304, Constructing a system based on total energy consumption Total startup time and equipment lifespan loss For a multi-objective Pareto optimization problem where the objective is to be optimized: ; Among them, decision variables Includes start-up time, heating power curve, and controller parameters; constraints include channel quality calculated in step S2. and temperature predicted by digital twins ; S305. A non-dominated sorting genetic algorithm with an elitist strategy is used to solve the Pareto optimization problem, obtaining a Pareto optimal solution set. Based on the weight vector λ set in step S1, the optimal compromise solution is selected from the solution set, and the output is the optimal starting vector. The optimal start vector will then be passed to steps S4 and S5.

[0011] Further improvements to this technical solution include step S4, which includes: S401, Analyze the optimal start vector output in step S3. Extract the optimal heating power curve. and optimal controller parameters And set it as the benchmark target for preheating control; S402, at startup time Previously, the digital twin constructed in step S1 was used to obtain the three-dimensional temperature field inside the battery pack in real time. and stress field And calculate the error between the current state and the target state:

[0012] and ; S403, State error , and ambient temperature Input a federated learning local model deployed on the remote control module. This federated learning local model is a neural network containing long short-term memory network layers and fully connected layers. Its output is the real-time power adjustment amount for each zone heating device. This makes the actual heating power ; S404, based on actual heating power Real-time integral calculation from the start of warm-up to the current moment Cumulative energy consumption: ; Based on the stress field time history, the equipment life loss index is estimated using a fatigue damage model: ; in, The fatigue damage coefficient of the battery pack material; The fatigue index of the battery pack material; is the fatigue strength limit of the battery pack material; is the maximum von Mises equivalent stress inside the battery pack at time t. S405, when the core temperature Reach the target temperature And the maximum von Mises equivalent force Below safety limit When the preheating is complete, a preheating completion signal is sent to step S5, and the final local model parameters obtained in this step are simultaneously sent to the system. Cumulative energy consumption and life loss index The "local optimization parameters and optimization contribution" are passed to step S6.

[0013] Further improvements to this technical solution include the optimization of heating control using a federated learning local model deployed in step S403, specifically including: S4031. The federated learning local model adopts an encoder-decoder structure, with the encoder being a bidirectional LSTM layer used to extract the input state sequence. The temporal characteristics are mapped to a two-layer fully connected network by the decoder, which is used to map the power adjustment to the time domain. ; S4032, Federated learning local model calculates output via forward propagation: ; ; in, This is the current input feature vector; It is in a hidden state; These are the model parameters, and their initial values ​​are... , Inherited from the federated learning global model weights downloaded in step S1 ; S4033. During the warm-up process, the federated learning global model uses real-time collected sensor data and digital twin prediction data to calculate the local loss: ; Where N is the total number of temperature sampling points used in a single forward inference by the local model of federated learning; This refers to the measured internal temperature of the battery pack or the simulated temperature output by the digital twin. The weighting coefficient for the heating power regularization term; And calculate the parameter gradient through backpropagation. This gradient is related to the final Together they constitute the "local optimization contribution" required for step S6.

[0014] Further improvements to this technical solution include step S5, which includes: S501. When the preheating completion signal from step S4 is received, and the system clock reaches the optimal start time in the optimal start vector of step S3. At that time, the remote control module generates the dynamic quantum key through step S2. Encrypted suction command; S502, the relay control module receives and decrypts the activation command, and simultaneously drives its internal main relay coil and backup relay coil, so that the contacts of the dual redundant relays are in a preset, minimal time difference. The internal circuits close sequentially, and the main battery is safely connected to the vehicle's DC bus. S503. Establish a digital twin-physical entity synchronization loop: using the battery voltage predicted by the digital twin based on the coupling equation in step S1. Bus current As a priori state estimate The voltage measured by the sensor Current As observation vector ; S504. State fusion using the Kalman filter algorithm: Calculate the Kalman gain. ; And update the posterior state estimate: ; Where H is the observation matrix; P is the error covariance matrix; and R is the observation noise covariance matrix. S505, Based on the trace of the posterior error covariance matrix Assess system uncertainty and dynamically calculate the compensated response delay: ; in, As a reference delay; This is the proportional coefficient, which is advanced in subsequent control timing. Issue commands to dynamically compensate for system response delays and estimate the compensated synchronization state. Proceed to step S6.

[0015] Further improvements to this technical solution include step S6, which includes: S601. Based on the compensated synchronization state output from step S5, estimate... Including main battery voltage With bus current The remote control module sends a start command to the engine control unit to control the engine ignition or drive motor start, and sequentially activates the hydraulic main pump solenoid valve and the working device control valve, so that the whole vehicle working system enters the standby state. S602. Based on the local optimization parameters and optimization contributions passed in step S4, i.e. the final weights of the local federated learning model. and training gradient Calculate the local contribution of this device in this round of operation: ; in, The cosine similarity function; The global average gradient is obtained from the federated learning server; For reference, energy consumption and lifespan loss values; Weighting coefficients that contribute to the optimization of the local model in federated learning; For energy consumption control effect; Weighting coefficients for the effectiveness of lifespan loss control; S603, For matters requiring reporting, including The system state data is hierarchically compressed and encoded: a lightweight neural network classifier categorizes it into high, medium, and low priorities based on data sensitivity. ; S604. Channel quality evaluated in real time according to step S2. With data priority Adaptive coding strategy selection: high-priority data is encrypted with quantum key distribution using one-time pad and forward error correction code is added; medium-priority data is encrypted with AES-256 and lossless compression is used; low-priority data is encrypted with lightweight encryption and lossy compression. S605. Apply the processed data packet to the dynamic quantum key generated in step S2. The data is encrypted and uploaded to the remote control system via a secure channel. At the same time, the hash value of the uploaded data packet is submitted to the blockchain network for evidence storage, forming a traceable full-process operation record.

[0016] Further improvements to this technical solution include step S7, which includes: S701. The remote control module receives a stop command from the remote control system, triggers the operation termination process, and queries and retrieves the hash sequence of the entire operation data packet stored in step S6 from the blockchain network based on the device identifier and timestamp. S702. Based on the full-process data, calculate the global energy efficiency index of this operation: ; in, To estimate the effective mechanical work through the fusion of digital twins and sensors; These are the energy consumption for operation and communication, respectively. , The mean and standard deviation of the total energy consumption of all equipment participating in this group operation; S703, Simultaneous carbon footprint audit report, calculating total carbon emissions: ; And generate a carbon emission intensity index: ; in, These are the carbon emission factors for the power grid and diesel, respectively. For the corresponding power; Total operation time; This refers to the total effective mechanical work generated by all the engineering machinery involved in the collaborative control during this group remote power-on operation; S704. The federated learning server collects the local contribution calculated by each client in step S6. and the corresponding model update gradient Perform weighted aggregation: ; To update the global model, where The global learning rate; The weights of the new global model obtained by federated learning are the weighted aggregates of all construction machinery clients in this group operation. This is the sum of the local contributions of all engineering machinery clients participating in this collaborative control; The weights of the old global model stored in the federated learning server before this global model update; S705, the calculated results and the hash value of the new global model after aggregation A new blockchain transaction is jointly written, and after consensus is reached, it is stored on the blockchain to ensure the immutability of this operation. The session state is then cleared, and the system switches to a deep security standby mode based on a quantum key pool.

[0017] The beneficial effects of this invention are as follows: This invention overcomes the blindness of traditional "hard start" modes by constructing an electro-thermal-mechanical coupled digital twin in step S1 and performing state prediction by combining unsteady-state heat conduction equations and stress balance equations in steps S3 and S4. By solving for the optimal start vector through multi-objective Pareto optimization and coordinating multi-physics preheating with federated learning edge computing, this solution avoids battery low-temperature high-current impact and mechanical thermal stress damage. This not only significantly reduces the peak energy consumption at the moment of vehicle start-up but also effectively reduces fatigue losses of electrical components and mechanical structures, significantly improving the reliability and service life of engineering machinery under extreme conditions.

[0018] To address the vulnerabilities of traditional remote control communication to attacks and the ease with which commands can be tampered with, this invention establishes a secure channel using a quantum key distribution protocol in step S2. Communication parameters are adaptively adjusted based on channel quality and electromagnetic interference intensity to ensure link stability under complex operating conditions. Simultaneously, by introducing blockchain-based verification commands in step S3, and then storing the entire process operation data and carbon footprint audit report on the blockchain in step S7, the invention achieves immutability of commands and operation records, and full traceability of the entire process. This mechanism, combining anti-interference communication, quantum encrypted transmission, and blockchain-based verification, significantly enhances the security and attack resistance of the remote control system for construction machinery.

[0019] In steps S4 and S6, this invention introduces a federated learning edge computing architecture and a local contribution evaluation model. At the edge, multiphysics data is used to optimize heating and startup strategies. In step S7, the global model is updated based on contribution weights, achieving collaborative optimization through collective intelligence while protecting node data privacy. Simultaneously, based on the local contribution calculated in step S6, data is subjected to hierarchical compression and prioritized transmission. This not only optimizes the utilization efficiency of communication bandwidth but also ensures the real-time and reliable transmission of critical control data, enabling the entire system to continuously optimize its energy efficiency and response performance with increasing number of runs.

[0020] Unlike traditional technologies that only focus on whether the equipment can start, this invention constructs a complete global energy efficiency index calculation and carbon footprint audit model in step S7. By retrieving the entire process data stored on the blockchain, it automatically calculates the total energy consumption, including the energy consumption of auxiliary systems, and the corresponding carbon emissions, generating a standardized carbon footprint audit report. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the specific embodiments. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0025] Figure 1 This is a schematic flowchart illustrating a remote power-on control method for engineering machinery provided by the present invention. The order of steps in this flowchart can be changed, and some steps can be omitted, depending on different requirements.

[0026] like Figure 1 As shown, the method includes: S1. Construct an electro-thermal-mechanical coupled digital twin for engineering machinery and complete the power supply circuit configuration for the main battery, backup battery, and relay control module; S2. The remote control module establishes a secure communication channel with the remote control system and generates a quantum key through a quantum key distribution protocol. S3. Receive and verify remote power-on commands stored on the blockchain based on the communication channel, complete command hash value and digital signature verification, perform multi-physics field state prediction based on digital twin, solve the optimal start-up vector through multi-objective Pareto optimization and complete feasibility verification. S4. Based on the optimal start-up vector output in step S3, combined with the battery temperature field and stress field output in real time by the digital twin in step S1, the heating device is controlled to perform multi-physics field collaborative preheating; and federated learning edge computing is used to optimize the heating parameters and calculate the cumulative energy consumption and equipment life loss index in real time. S5. Based on the preheating completion signal in step S4 and the optimal start-up timing within the optimal start-up vector in step S3, output a pull-in signal to drive the dual redundant relays to close, so that the main battery is connected to the vehicle; establish a virtual-real synchronization mechanism based on the digital twin in S1, and use Kalman filtering to fuse the state and compensate for the system response delay; S6. Based on the synchronization state after compensation in step S5, control the power system to complete the vehicle start-up, and make the hydraulic and working devices ready; based on the local optimization parameters and optimization contribution output by federated learning edge computing in step S4, calculate the local contribution of this round of operation, and use hierarchical compression coding to upload the state data through the quantum encryption link in step S2. S7. After receiving the stop command, retrieve the full-process data that has been stored on the blockchain in step S6, and automatically calculate the global energy efficiency index and carbon footprint audit report for this operation.

[0027] To facilitate understanding of the present invention, the following description further illustrates the remote vehicle power-on control method for construction machinery provided by the present invention, based on the principle of the method and the process of remotely powering on construction machinery in the embodiments.

[0028] First, step S1 includes: S101. Construct the electro-thermal-mechanical multi-physics coupled control equations for the engineering machinery to be controlled. The control equations include the unsteady heat conduction equation for simulating the evolution of the battery temperature field and the stress balance equation for calculating the stress field distribution of the engineering machinery. The control equations are spatially discretized and time-integrated using the finite element method. S102. Connect the positive and negative terminals of the main battery to the DC bus of the vehicle circuit through the contacts of the main relay and the backup relay in the relay control module. At the same time, connect the positive and negative terminals of the backup battery directly to the power management chip of the remote control module to provide the remote control module with a continuous power supply independent of the vehicle power supply. S103. Initialize key system parameters, including generating an initial sequence of entangled photon pairs for quantum key distribution to build a quantum key pool, creating a copy of the blockchain network's genesis block in local storage to establish a trusted ledger baseline, and downloading the initial global model weights from the federated learning server. The remote control module serves as prior knowledge for the local federated learning model, and the weight vector for multi-objective optimization is set. Used to quantify the relative importance of energy consumption, time, and lifetime loss in the optimization step S3.

[0029] Furthermore, the governing equations that form the core of the digital twin computation in step S101 specifically include: S1011. Establish the coupled unsteady-state heat conduction control equations: ; And the stress balance governing equations: ; in, The density of the material; is the specific heat capacity at constant pressure; T is the temperature field; t is time; k(T) is the thermal conductivity related to temperature; I represents the electrochemical reaction heat generation rate; I represents the current; SOC represents the battery state of charge. This refers to the battery's internal resistance. This refers to the volume of a single battery cell; denoted as Cauchy stress tensor; b represents volume force. For displacement field; S1012. Discretize the governing equations using the finite element method to construct the global heat capacity matrix. Heat conduction matrix Mass matrix M and stiffness matrix And assemble them to obtain the discrete system equations:

[0030] and ; Where T and U are the nodal temperature and displacement vectors, respectively; These are the nodal load vectors for chemical reaction heat and Joule heat, respectively; This is the equivalent nodal load vector for thermal stress. , e is the unit number, Let B be the element domain, B be the strain-displacement matrix, and D be the elasticity matrix. The coefficient of thermal expansion is The temperature shape function matrix; S1013. An implicit Newmark-β method (usually referring to the generalized α method used in structural dynamics or the backward Euler method in thermal analysis, which is a generalization of the Newmark-β method to thermal problems) is used to perform time integration on the discretized system equations. Within each simulation step Δt, the nonlinear coupled equations are solved through Newton-Raphson iteration, and the three-dimensional temperature field distribution T(x,y,z,t) inside the battery pack is output in real time. Mises equivalent force field distribution This data is then used as input for step S4.

[0031] Solution strategy: Sequential coupling analysis is employed. Within each time step Δt: Solve the heat conduction equation and update the temperature field. .

[0032] Substitute Tn+1 into the calculation of thermal load .

[0033] Solve the mechanical equilibrium equations and update the displacement field. and stress field .

[0034] Nonlinear treatment: Since k(T) and Rint(T,SOC) are functions of temperature or state, the system of equations is nonlinear. The Newton-Raphson iterative method is used to solve the problem: In each iteration, the material properties are updated based on the current temperature estimate, the matrix is ​​reassembled, and the linear equations are solved until the change in the temperature solution is less than the preset tolerance.

[0035] Within each time step, due to material nonlinearity (temperature dependence of thermal conductivity, variation of internal resistance) and geometric nonlinearity (large deformation effect), the Newton-Raphson iterative method is required to solve the nonlinear coupled equations. Taking the heat conduction equation as an example, the iterative format is: ; ; in, Represents the Jacobian matrix of heat conduction; Represents the heat conduction residual vector; the iterative convergence criterion is... or .

[0036] Through the above solution process, the three-dimensional temperature field distribution T(x,y,z,t) and the von Mises equivalent force field distribution inside the battery pack are output in real time.

[0037] Battery packs in construction machinery typically consist of multiple lithium-ion battery cells connected in series and parallel, encapsulated within a metal casing. During low-temperature startup, the battery generates significant ohmic heat and electrochemical reaction heat due to high-current discharge, leading to uneven temperature rise. This, in turn, causes internal thermal stress due to differences in the thermal expansion coefficients of different materials (electrodes, separators, electrolytes, and casing). Furthermore, the intervention of external heaters exacerbates this unevenness in temperature and stress.

[0038] The positive terminal of the main battery is connected to the first contact of the main relay in the relay control module via a first copper busbar, and the negative terminal of the main battery is connected to the second contact of the main relay via a second copper busbar. The first contact of the main relay's output terminal is connected to the positive terminal of the vehicle's DC bus via a third copper busbar, and the second contact of the output terminal is connected to the negative terminal of the DC bus via a fourth copper busbar. The contacts of the standby relay are connected in parallel across the main relay contacts, forming a dual redundancy structure. When the main relay fails, the system automatically switches to the standby relay.

[0039] The main battery uses a lithium iron phosphate power battery with a rated voltage of Rated capacity Maximum continuous discharge current Operating temperature range .

[0040] The positive terminal of the backup battery is directly connected to the positive input terminal of the power management chip of the remote control module via the fifth copper busbar, and the negative terminal of the backup battery is directly connected to the negative input terminal of the power management chip via the sixth copper busbar. The power management chip integrates a low dropout linear regulator and a switching power converter to convert the backup battery voltage into the operating voltages (3.3V, 5V, 12V) required by the various functional units of the remote control module.

[0041] The backup battery is a lithium titanate battery with a rated voltage of [missing information]. Rated capacity Operating temperature range With a self-discharge rate of less than 2% per month, it can support the remote control module to work continuously for more than 180 days in deep sleep mode.

[0042] Quantum Key Pool Construction: An initial sequence of entangled photon pairs is generated for quantum key distribution. Entangled photon pairs are generated using a spontaneous parametric down-conversion process. The pump source is a 405nm continuous-wave laser, which, through a nonlinear crystal (barium borate crystal), generates signal photons and idle photons with wavelengths of 810nm. The entangled photon pairs are in Bell states. ; in, and These represent the horizontal and vertical polarization states, respectively; the subscripts s and i represent the signal photon and the idle photon, respectively.

[0043] The generated sequence of entangled photon pairs is stored in a quantum key pool with a capacity of [value missing]. Yes, each pair of photons is assigned a unique identifier and a quantum state fidelity record. The quantum key pool is integrated into the quantum communication subsystem of the remote control module, and maintains an entangled association with the quantum key pool of the remote control system through a quantum channel.

[0044] Geometric modeling and mesh generation: Based on the 3D computer-aided design (CAD) model of the engineering machinery battery pack, the model is discretized into a finite number of elements (e.g., tetrahedral or hexahedral elements) using pre-processing software (e.g., ANSYS, COMSOL, or a lightweight mesh generation library integrated into the remote control module). The mesh density is balanced between simulation accuracy and computational resources, with localized refinement in areas with large temperature gradients and stress concentrations (e.g., tab connections, surfaces near heaters).

[0045] System matrix assembly: For each element, calculate its heat capacity matrix, heat conduction matrix, mass matrix, and stiffness matrix based on its shape function and material properties. Assemble all element matrices according to node numbers to obtain the global system matrix.

[0046] Blockchain genesis block creation: A copy of the blockchain network's genesis block is created in local storage (non-volatile flash memory, 32GB capacity). Genesis block. The data structure is as follows: ; Wherein, Version represents the protocol version number; PrevHash represents the hash of the previous block, with the genesis block set to a 256-bit zero vector; MerkleRoot represents the Merkle root hash, initially the hash value of an empty transaction tree; and Timestamp represents the genesis timestamp. Take the UTC time of the system's first startup; Nonce represents a random number. The proof-of-work algorithm is used to calculate the satisfaction. Target value; This represents the SHA-256 double hash function.

[0047] After the genesis block is established, subsequent blocks in the blockchain network expand upon it to form an immutable and trusted ledger benchmark, which is used to store hashes of remote power-on instructions, operation records, and global model update hashes.

[0048] The remote control module establishes a secure connection with the federated learning server via an Ethernet interface and downloads the initial global model weights. These weights are parameters for a multi-layer neural network, and the model architecture includes an input layer (input dimension). (corresponding to state features such as temperature, stress, current, and voltage), two hidden layers (128 neurons per layer, ReLU activation function) and an output layer (output dimension) (corresponding to the heating power adjustment). The total weighted parameters are... .

[0049] Downloaded initial global model weights As prior knowledge for the local federated learning model, it is fine-tuned and optimized based on locally collected data during the warm-up process in subsequent step S4. The optimized local model weights are: .

[0050] Secondly, step S2 includes: S201. The remote control module uses the entangled photon pairs in the quantum key pool initialized in step S1 to perform a quantum key distribution process based on the BB84 protocol with the remote control system, and negotiates to generate a dynamic quantum key specific to this session. And utilize dynamic quantum key distribution Establish high-level secure communication channels; S202. In standby mode, the remote control module monitors the Received Signal Strength Indicator (RSSI) and packet loss rate of the wireless communication link in real time. And calculate the overall communication channel quality index: ; Where RSSI is the sliding window average signal strength; Sensitivity coefficient; Reference signal strength; S203, Electromagnetic interference intensity spectral density of the frequency band where the radio frequency front-end of the synchronous monitoring remote control module is located. And calculate the equivalent electromagnetic interference intensity: ; in, , These are the minimum and maximum operating frequencies for wireless communication of the remote control module, respectively. Based on the comprehensive communication channel quality index With equivalent electromagnetic interference intensity Dynamically adjust the heartbeat packet sending cycle:

[0051] With radio frequency transmit power ; in, This is the weighting coefficient for adjusting the heartbeat cycle based on communication channel quality. This is the adjustment weighting coefficient of electromagnetic interference intensity on the heartbeat cycle; This is a preset reference value for electromagnetic interference intensity; The minimum permissible radio frequency transmit power for the remote control module; This represents the maximum permissible radio frequency transmit power for the remote control module. This is the exponential attenuation coefficient of electromagnetic interference on the transmitted power.

[0052] The remote control module integrates a quantum communication subsystem, which includes a quantum signal transmitting unit, a quantum signal receiving unit, a quantum random number generator, and a classical communication interface. The quantum signal transmitting unit uses a 785nm pulsed laser diode as the light source, with a pulse width of 1ns and a repetition frequency of 10MHz, attenuated to a single-photon level with an average photon number μ=0.1 by an attenuator. The quantum signal receiving unit uses a silicon-based single-photon avalanche photodiode detector with a quantum efficiency of [missing information]. The dark count rate is less than 100Hz and the time jitter is less than 300ps.

[0053] The remote control system is equipped with a quantum communication subsystem of the same specifications. The two sides are connected through a polarization-maintaining single-mode fiber quantum channel with a channel length not exceeding 50km and a channel loss of less than 0.2dB / km. The classic communication interface uses gigabit Ethernet for transmitting basis vector alignment information and error correction verification data.

[0054] The remote control module and remote control system execute a quantum key distribution process based on the BB84 protocol, which includes six stages: quantum state preparation, quantum state transmission, basis vector selection and measurement, key screening, information coordination, and privacy amplification.

[0055] Quantum state preparation stage: The quantum random number generator of the remote control module generates a random bit sequence a∈{0,1} and a random basis vector sequence b∈{0,1}. When b=0, the basis {|0,1} is selected for computation. ,|1 When b=1, choose the diagonal basis {|+ ,| },in Prepare the corresponding quantum state based on the values ​​of a and b: ; Quantum state transmission stage: The prepared quantum state is transmitted to the remote control system via a polarization-maintaining single-mode fiber quantum channel. During transmission, the quantum state may be affected by channel loss, decoherence, and eavesdropping attacks.

[0056] Basis selection and measurement phase: The quantum random number generator of the remote control system generates a random basis vector sequence b′∈{0,1}, which is used to select the measurement basis vector. When b′=0, the measurement is performed under the computational basis; when b′=1, the measurement is performed under the diagonal basis. The measurement result is denoted as a′∈{0,1}.

[0057] Key filtering phase: Through the classic communication interface, the remote control module and remote control system publicly compare the basis vector sequences b and b′, ​​retaining measurement results with identical basis vectors. The length of the filtered key is approximately 50% of the original transmission volume.

[0058] Information Coordination Phase: A cascaded error correction algorithm is used to correct errors in the filtered key, eliminating bit errors caused by channel noise and detector dark counting. The bit error rate of the corrected qubits. Less than 11%, phase error rate Less than 11%.

[0059] Privacy amplification stage: A universal hash function is used to amplify the privacy of the corrected key, eliminating information that could be obtained by eavesdroppers. The formula for calculating the length of the privacy-amplified key is: ; in, H2 is the length of the filtered key, and H2 is the binary Shannon entropy function, H2(x) = xlog2x (1 x)log2(1 x).

[0060] The dynamic quantum key generated by final negotiation The key is 256 bits long and is used for Advanced Encryption Standard (AES) symmetric encryption for this session. Key generation rate. satisfy: ; in, For detector efficiency; The average number of photons; This is the pulse repetition frequency.

[0061] Using dynamic quantum key distribution Establish a high-level secure communication channel. Employ a one-time pad encryption method using quantum key distribution to encrypt subsequent transmitted remote power-on commands and status data. The encryption process is as follows: ; Where C represents ciphertext and M represents plaintext; For dynamic quantum keys, ⊕ represents bitwise XOR operation.

[0062] The security provided by quantum key distribution is based on the fundamental principles of quantum mechanics. Any eavesdropping will introduce a detectable disturbance, thereby improving the security of communication.

[0063] In standby mode, the RF transceiver unit of the remote control module monitors the Received Signal Strength Indicator (RSSI) and packet loss rate of the wireless communication link in real time. .

[0064] Measurement of Received Signal Strength Indication (RSSI): The RF transceiver unit acquires the instantaneous received power through the received channel power indicator. The sampling frequency is 1kHz, and the duration of each sampling point is 1ms. A sliding window averaging method is used to smooth the instantaneous received power. The sliding window length W=100. The formula for calculating the RSSI after sliding window averaging is as follows: ; in, is the instantaneous received power at the i-th sampling point, in dBm; k is the current sampling point number.

[0065] Packet loss rate Measurement: The remote control module sends a heartbeat packet every 60 seconds, and records the total number of heartbeat packets sent. and the total number of acknowledgment packets received The formula for calculating packet loss rate is: ; Received Signal Strength Indicator (RSSI) and Packet Loss Rate after Sliding Window Averaging Calculate the overall communication channel quality index This index takes into account both signal strength and transmission reliability, and its value ranges from 0 to 1. The closer it is to 1, the better the channel quality.

[0066] When RSSI= 80dBm and When =0, =0.5; when RSSI is better than 80dBm and When =0, >0.5; when RSSI is worse than 80dBm or When >0, <0.5.

[0067] The remote control module integrates a broadband spectrum monitoring unit to monitor the electromagnetic interference intensity spectral density in the frequency band where the radio frequency front-end is located in real time. The spectrum monitoring unit adopts a superheterodyne receiver architecture, with a frequency sweep range of 30MHz to 6GHz, a resolution bandwidth of 100kHz, and a peak detection method.

[0068] Electromagnetic interference intensity spectral density at each frequency point f The measured value is: ; in, Here is the interference power at frequency point f, in W; This refers to the resolution bandwidth, measured in Hz. The unit is .

[0069] Then, step S3 includes: S301. The remote control module receives the remote power-on command ciphertext from the remote control system through the quantum-encrypted secure channel established in step S2, and uses the dynamic quantum key generated in step S2. Decrypt to obtain the instruction hash value. With digital signatures The original instructions; S302. Call the local blockchain ledger copy established in step S1 to verify the instruction hash value. Has it been stored in the blockchain network consensus and verified using the public key corresponding to the remote control system? Verify digital signature The validity of the instructions is ensured, guaranteeing their integrity and non-repudiation. S303. After the instruction verification is successful, drive the digital twin constructed in step S1 to perform multiphysics coupling simulation using the current physical state read by the sensor as the initial condition, and predict the battery temperature field sequence within the preset time window [t, t+ΔT]. and stress field sequence ; S304, Constructing a system based on total energy consumption Total startup time and equipment lifespan loss For a multi-objective Pareto optimization problem where the objective is to be optimized: ; Among them, decision variables Includes start-up time, heating power curve, and controller parameters; constraints include channel quality calculated in step S2. and temperature predicted by digital twins ; S305. A non-dominated sorting genetic algorithm with an elitist strategy is used to solve the Pareto optimization problem, obtaining a Pareto optimal solution set. Based on the weight vector λ set in step S1, the optimal compromise solution is selected from the solution set, and the output is the optimal starting vector. The optimal start vector will then be passed to steps S4 and S5.

[0070] Receiving and decrypting quantum encryption instructions: Channel and Key: The dynamic quantum key established by the remote control module in step S2 and negotiated based on the BB84 protocol. The encrypted secure channel is being monitored. This channel uses authentication and encryption modes such as AES-GCM to ensure confidentiality and integrity.

[0071] Command Format: The remote power-on command sent by the remote control system is a structured data packet with the format: Command Ciphertext = AES_GCM_Encrypt(K_{QKD}, Original Command). The original command must include at least: a command action code (e.g., power-on), the target device ID, and a timestamp. Random numbers and the instruction hash value used for subsequent verification. Elliptic curve digital signatures .

[0072] Decryption process: After receiving the encrypted command, the remote control module uses the same stored session key. Perform the AES_GCM_Decrypt operation. After successful decryption and verification of the Message Authentication Code (MAC), readable raw instruction data is obtained, from which the following can be extracted: and .

[0073] Local blockchain ledger: When the remote control module is initialized in step S1, it has synchronized the genesis block and subsequent block headers of the blockchain network and maintained a lightweight SPV (Simplified Payment Verification) node, which can verify the existence of transactions.

[0074] Hash-based verification: The remote control module calculates the hash value of the received original instruction (excluding the signature): =SHA256(Instruction ActionCode | Target Device ID |) || It then queries the blockchain network (or checks the locally cached event log) to confirm the hash value. The process verifies whether the transaction has been packaged into a block that has passed Proof-of-Work (PoW) or Proof-of-Stake (PoS) consensus and received a sufficient number of subsequent block confirmations (e.g., 6 confirmations). This process verifies the "existence" and "immutability" of the instruction.

[0075] Digital signature verification: using a pre-set public key or one obtained from the blockchain that is bound to the identity of the remote control system. For instruction hash value and digital signatures Execute a verification algorithm (such as the ECDSA verify function). If the verification passes, it proves the authenticity and non-repudiation of the instruction's origin. Instruction verification is considered successful only if both conditions are met: "the hash is stored on the blockchain" and "the digital signature is valid."

[0076] Initialize the digital twin: Drive the electro-thermal-mechanical coupled digital twin constructed in step S1. Read the current physical state in real time from the engineering machinery's battery management system (BMS), temperature sensor network, tilt sensor, etc., including: main battery state of charge. Temperature at each sampling point Ambient temperature The static attitude of the equipment, etc., are then mapped to the node initial conditions (e.g., initial temperature field T0) and boundary conditions of the digital twin finite element model.

[0077] Perform coupled simulation prediction: Input settings: Set a future time window [ , +ΔT], where ΔT is the preset prediction duration, such as 600 seconds. This window needs to cover possible warm-up and startup processes.

[0078] Simulation Execution: The digital twin, based on its built-in control equations (see step S1), performs transient coupled simulations driven by a variety of preset candidate control strategies (i.e., sampling points of the decision variable X generated by the subsequent optimization algorithm). The key driving input for the simulation is the assumed "heating power curve," which originates from the exploration of the optimization algorithm.

[0079] Output sequence: After the simulation is completed, for each candidate strategy, the digital twin outputs the battery pack core temperature sequence {T^core(k)} and the maximum von Mises equivalent stress sequence {σ^vM,max(k)} sampled at fixed intervals (e.g., 1 second) within the future time window under that strategy, where k=1,2,...,N, and N=ΔT / sampling interval.

[0080] This step transforms the engineering objective into a mathematical optimization problem, the input of which is closely dependent on the predicted output of the digital twin.

[0081] Definition of decision variables: Define the decision variable vector X = ( , , ).

[0082] The startup time is defined at... A continuous or discrete variable within.

[0083] The heating power curve can be parameterized as a spline curve defined by m control points, for example... B i (t) is a basis function (e.g., a B-spline), w i Let be the weight coefficients to be optimized. The decision variable can then be expressed as: .

[0084] The parameters of the proportional-integral-derivative controller are used for closed-loop fine-tuning in step S4.

[0085] Objective function construction: Construct three conflicting optimization objectives: Total energy consumption target : Calculate from the current time to the completion of startup ( The total energy consumption (over a fixed period of time) mainly includes preheating energy consumption and peak start-up energy consumption. Preheating energy consumption is obtained through the integral heating power curve. Startup energy consumption can be estimated based on startup current and voltage predictions using digital twins. That is... .

[0086] Total Startup Time Target Defined as the time from the current moment until the device is ready ( The total time (after system stabilization) can be simplified to: Where C is the fixed time of the startup process, or a more precise calculation can be made by combining the "time to reach the startable temperature" predicted by the digital twin.

[0087] Equipment life loss target Based on the stress sequence {σ^vM,max(k)} predicted by a digital twin, the equivalent life loss caused by the preheating process is calculated using a fatigue damage accumulation model (e.g., Miner's linear accumulation rule). For example, =∑k(σ^vM,max(k) / σ f )β Δt k ,in β and β are material fatigue parameters. Then... .

[0088] Constraint settings: Communication quality constraints: g1(X): This is an X-independent constraint measured in real time by step S2, ensuring that decisions are made under conditions of reliable communication.

[0089] Temperature safety constraint: g2(X):max({T^core(k)})≤ This constraint relies directly on the predictive output of the digital twin to ensure that the maximum battery temperature does not exceed the safety limit under any strategy.

[0090] Stress safety constraint: g3(X):max({σ^vM,max(k)})≤ It also relies on digital twin predictions to prevent thermal stress from exceeding limits.

[0091] Battery power constraint: g4(X): Ensure sufficient power during startup.

[0092] Power and time boundaries: , .

[0093] In summary, the multi-objective Pareto optimization problem can be formally described as follows: ; Where Ω represents the feasible region of the decision variable.

[0094] Non-dominated sorting genetic algorithm with elitist strategy (NSGA-II): Initialization: Randomly generate an initial population P0 of size N, where each individual represents the encoding of a decision variable X.

[0095] Fitness assessment: For each individual X in the population j Call the digital twin constructed in step S303 and input the heating power curve represented by that individual. and startup time Perform a multiphysics coupling simulation prediction.

[0096] Using the simulation output sequences {T^core(k)} and {σ^vM,max(k)}, calculate the three objective function values ​​corresponding to this individual. And determine whether all constraints are satisfied. Individuals that do not meet the constraints are given a large penalty value (or dealt with through constraint domination).

[0097] Non-dominated ranking and crowding calculation: All individuals in the population are compared pairwise, and non-dominated ranking is performed according to Pareto dominance, dividing the population into multiple fronts (F1, F2, ..., where F1 is the Pareto optimal front). For individuals within the same non-dominated front, their crowding distance is calculated to measure the distribution density of individuals in the target space.

[0098] Selection, crossover, and mutation: A binary tournament selection method is used, prioritizing individuals with high non-dominated rank (smaller numbers). If ranks are the same, individuals with greater crowding distance are selected (to maintain population diversity). Simulated binary crossover (SBX) and polynomial mutation are performed on the selected parent individuals to generate the offspring population. .

[0099] Elite Preservation: Preserving the parent population's P t and offspring population Merge the populations, perform non-dominated sorting and crowding calculations on the merged populations, and then select the top N individuals to form a new parent population P. t+1 This strategy ensures the preservation of elite individuals.

[0100] Iteration Termination: Repeat the above process of evaluation, sorting, selection, crossover, mutation, and merging until the preset maximum number of generations is reached. .

[0101] Selection of the optimal compromise solution: After the algorithm terminates, the final Pareto optimal frontier (F1) contains the set of all optimal compromise solutions that cannot be mutually improved. From this frontier, the optimal solution is selected based on the weight vector set in step S1. Choose the solution that best matches your current preference. For example, use the weighted Chebyshev method: ; in, and These are the minimum and maximum values ​​of the i-th objective on the Pareto front, used for normalization. The final output is... This is the "optimal initiation vector".

[0102] Next, step S4 includes: S401, Analyze the optimal start vector output in step S3. Extract the optimal heating power curve. and optimal controller parameters And set it as the benchmark target for preheating control; S402, at startup time Previously, the digital twin constructed in step S1 was used to obtain the three-dimensional temperature field inside the battery pack in real time. and stress field And calculate the error between the current state and the target state:

[0103] and ; S403, State error , and ambient temperature Input a federated learning local model deployed on the remote control module. This federated learning local model is a neural network containing long short-term memory network layers and fully connected layers. Its output is the real-time power adjustment amount for each zone heating device. This makes the actual heating power ; S404, based on actual heating power Real-time integral calculation from the start of warm-up to the current moment Cumulative energy consumption: ; Based on the stress field time history, the equipment life loss index is estimated using a fatigue damage model: ; in, The fatigue damage coefficient of the battery pack material; The fatigue index of the battery pack material; is the fatigue strength limit of the battery pack material; is the maximum von Mises equivalent stress inside the battery pack at time t. S405, when the core temperature Reach the target temperature And the maximum von Mises equivalent force Below safety limit When the preheating is complete, a preheating completion signal is sent to step S5, and the final local model parameters obtained in this step are simultaneously sent to the system. Cumulative energy consumption and life loss index The "local optimization parameters and optimization contribution" are passed to step S6.

[0104] Furthermore, the federated learning local model deployed in step S403 optimizes the heating control, specifically including: S4031. The federated learning local model adopts an encoder-decoder structure, with the encoder being a bidirectional LSTM layer used to extract the input state sequence. The temporal characteristics are mapped to a two-layer fully connected network by the decoder, which is used to map the power adjustment to the time domain. ; S4032, Federated learning local model calculates output via forward propagation: ; ; in, This is the current input feature vector; It is in a hidden state; These are the model parameters, and their initial values ​​are... , Inherited from the federated learning global model weights downloaded in step S1 ; S4033. During the warm-up process, the federated learning global model uses real-time collected sensor data and digital twin prediction data to calculate the local loss: ; Where N is the total number of temperature sampling points used in a single forward inference by the local model of federated learning; This refers to the measured internal temperature of the battery pack or the simulated temperature output by the digital twin. The weighting coefficient for the heating power regularization term; And calculate the parameter gradient through backpropagation. This gradient is related to the final Together they constitute the "local optimization contribution" required for step S6.

[0105] In addition, step S5 includes: S501. When the preheating completion signal from step S4 is received, and the system clock reaches the optimal start time in the optimal start vector of step S3. At that time, the remote control module generates the dynamic quantum key through step S2. Encrypted suction command; S502, the relay control module receives and decrypts the activation command, and simultaneously drives its internal main relay coil and backup relay coil, so that the contacts of the dual redundant relays are in a preset, minimal time difference. The internal circuits close sequentially, and the main battery is safely connected to the vehicle's DC bus. S503. Establish a digital twin-physical entity synchronization loop: using the battery voltage predicted by the digital twin based on the coupling equation in step S1. Bus current As a priori state estimate The voltage measured by the sensor Current As observation vector ; S504. State fusion using the Kalman filter algorithm: Calculate the Kalman gain. ; And update the posterior state estimate: ; Where H is the observation matrix; P is the error covariance matrix; and R is the observation noise covariance matrix. S505, Based on the trace of the posterior error covariance matrix Assess system uncertainty and dynamically calculate the compensated response delay: ; in, As a reference delay; This is the proportional coefficient, which is advanced in subsequent control timing. Issue commands to dynamically compensate for system response delays and estimate the compensated synchronization state. Proceed to step S6.

[0106] Conditional Judgment: The remote control module continuously monitors two signals: one is the "preheating complete" flag signal (a Boolean value) from step S4; the other is the system real-time clock. The module will determine whether to proceed if and only if the "preheating complete" signal is true and the system clock reaches or exceeds the optimal start vector output in step S3. The optimal start time included in When the triggering condition is met.

[0107] Command Encryption: The remote control module immediately generates a structured "activation command". This command includes at least: command code (e.g., RELAY_CLOSE), target relay identifier (primary / backup), and timestamp. And the serial number. Then, using the currently valid dynamic quantum key generated from the quantum key distribution session established in step S2. The instruction is encrypted using an authentication encryption algorithm (such as AES-256-GCM) to generate instruction ciphertext. The encryption process also generates a Message Authentication Code (MAC) to ensure the integrity and immutability of the instructions.

[0108] Additionally, step S6 includes: S601. Based on the compensated synchronization state output from step S5, estimate... Including main battery voltage With bus current The remote control module sends a start command to the engine control unit to control the engine ignition or drive motor start, and sequentially activates the hydraulic main pump solenoid valve and the working device control valve, so that the whole vehicle working system enters the standby state. S602. Based on the local optimization parameters and optimization contributions passed in step S4, i.e. the final weights of the local federated learning model. and training gradient Calculate the local contribution of this device in this round of operation: ; in, The cosine similarity function; The global average gradient is obtained from the federated learning server; For reference, energy consumption and lifespan loss values; Weighting coefficients that contribute to the optimization of the local model in federated learning; For energy consumption control effect; Weighting coefficients for the effectiveness of lifespan loss control; S603, For matters requiring reporting, including The system state data is hierarchically compressed and encoded: a lightweight neural network classifier categorizes it into high, medium, and low priorities based on data sensitivity. ; S604. Channel quality evaluated in real time according to step S2. With data priority Adaptive selection of encoding strategy: for high-priority data ( Quantum key encryption is used for one-time pad encryption with forward error correction codes added; for middle-priority data ( AES-256 encryption and lossless compression are used; for low-priority data ( It employs lightweight encryption and lossy compression; S605. Apply the processed data packet to the dynamic quantum key generated in step S2. The data is encrypted and uploaded to the remote control system via a secure channel. At the same time, the hash value of the uploaded data packet is submitted to the blockchain network for evidence storage, forming a traceable full-process operation record.

[0109] Hierarchical startup sequence execution: After successful verification, the remote control module sends a start command to the engine control unit to execute a tiered start sequence: Pre-start phase (duration) ): Activates the fuel pump to build fuel pressure, preheats the glow plugs to increase combustion chamber temperature, and the battery management system performs a self-test. Control Input ,in For fuel pump power, This refers to the power of the glow plug.

[0110] Start-up motor drive phase (duration) When the starter relay is closed, the starter motor drives the engine crankshaft to rotate, and the speed increases from 0 to... Control input ,in The starting motor power is typically 5kW.

[0111] Ignition and Combustion Phase (Duration) When the crankshaft speed reaches, At this time, the engine control unit controls the fuel injectors to inject fuel and ignite it. The combustion pressure pushes the piston to do work, and the engine speed increases rapidly. Control input ,in This refers to the fuel injection mass flow rate.

[0112] Starter motor disengagement phase (duration) When the engine speed exceeds When the starter relay disconnects, the starter motor stops working, and the engine starts running automatically. Control input .

[0113] Idle speed stabilization phase (duration) The engine speed stabilizes at the idle speed setting. Oil pressure builds up, coolant circulates. Control input. Idle speed is regulated using a proportional-integral-derivative controller.

[0114] The complete control input function for the hierarchical start sequence is: .

[0115] After the engine idles to a stable speed, the remote control module sequentially activates the hydraulic system and the working device: Hydraulic main pump solenoid valve activation: Sends a proportional control signal to the hydraulic main pump solenoid valve to set the initial displacement. Establish system pressure to Monitor the slope of the pressure rise; if... If the hydraulic oil viscosity is too high, extend the preheating time.

[0116] Working device control valve activation: Sequentially activate the control valves of the boom, stick, and bucket hydraulic cylinders to perform no-load cycle operations and verify the hydraulic system response. The control valve drive signal is a pulse width modulation signal with a duty cycle gradually changing from 0 to 50% to avoid hydraulic shock.

[0117] The vehicle's operating system enters standby mode, awaiting operational instructions from the remote control system.

[0118] Finally, step S7 includes: S701. The remote control module receives a stop command from the remote control system, triggers the operation termination process, and queries and retrieves the hash sequence of the entire operation data packet stored in step S6 from the blockchain network based on the device identifier and timestamp. S702. Based on the full-process data, calculate the global energy efficiency index of this operation: ; in, To estimate the effective mechanical work through the fusion of digital twins and sensors; These are the energy consumption for operation and communication, respectively. , The mean and standard deviation of the total energy consumption of all equipment participating in this group operation; S703, Simultaneous carbon footprint audit report, calculating total carbon emissions: ; And generate a carbon emission intensity index: ; in, These are the carbon emission factors for the power grid and diesel, respectively. For the corresponding power; Total operation time; This refers to the total effective mechanical work generated by all the engineering machinery involved in the collaborative control during this group remote power-on operation; S704. The federated learning server collects the local contribution calculated in step S6 from each client (i.e., the remote control module of each piece of construction machinery). and the corresponding model update gradient Perform weighted aggregation: ; To update the global model, where The global learning rate; The weights of the new global model obtained by federated learning are the weighted aggregates of all construction machinery clients in this group operation. This is the sum of the local contributions of all engineering machinery clients participating in this collaborative control; The weights of the old global model stored in the federated learning server before this global model update; S705, the calculated results and the hash value of the new global model after aggregation A new blockchain transaction is jointly written, and after consensus is reached, it is stored on the blockchain to ensure the immutability of this operation. The session state is then cleared, and the system switches to a deep security standby mode based on a quantum key pool.

[0119] Unlike traditional technologies that only focus on whether the equipment can start, this invention constructs a complete global energy efficiency index calculation and carbon footprint audit model in step S7. By retrieving the entire process data stored on the blockchain, it automatically calculates the total energy consumption, including the energy consumption of auxiliary systems, and the corresponding carbon emissions, generating a standardized carbon footprint audit report.

[0120] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.

Claims

1. A remote power-on control method for engineering machinery, characterized in that, include: S1. Construct an electro-thermal-mechanical coupled digital twin for engineering machinery and complete the power supply circuit configuration for the main battery, backup battery, and relay control module; S2. The remote control module establishes a secure communication channel with the remote control system and generates a quantum key through a quantum key distribution protocol. S3. Receive and verify remote power-on commands stored on the blockchain based on the communication channel, complete command hash value and digital signature verification, perform multi-physics field state prediction based on digital twin, solve the optimal start-up vector through multi-objective Pareto optimization and complete feasibility verification. S4. Based on the optimal start-up vector output in step S3, combined with the battery temperature field and stress field output in real time by the digital twin in step S1, the heating device is controlled to perform multi-physics field collaborative preheating; and federated learning edge computing is used to optimize the heating parameters and calculate the cumulative energy consumption and equipment life loss index in real time. S5. Based on the preheating completion signal in step S4 and the optimal start-up timing within the optimal start-up vector in step S3, output a pull-in signal to drive the dual redundant relays to close, so that the main battery is connected to the vehicle; establish a virtual-real synchronization mechanism based on the digital twin in S1, and use Kalman filtering to fuse the state and compensate for the system response delay; S6. Based on the synchronization state after compensation in step S5, control the power system to complete the vehicle start-up, and make the hydraulic system and working device ready. Based on the local optimization parameters and optimization contribution output by federated learning edge computing in step S4, the local contribution of this round of operation is calculated, and the state data is uploaded via the quantum encryption link in step S2 using hierarchical compression coding. S7. After receiving the stop command, retrieve the full-process data that has been stored on the blockchain in step S6, and automatically calculate the global energy efficiency index and carbon footprint audit report for this operation.

2. The remote vehicle power-on control method for engineering machinery according to claim 1, characterized in that, Step S1 includes: S101. Construct the electro-thermal-mechanical multi-physics coupled control equations for the engineering machinery to be controlled. The control equations include the unsteady heat conduction equation for simulating the evolution of the battery temperature field and the stress balance equation for calculating the stress field distribution of the engineering machinery. The control equations are spatially discretized and time-integrated using the finite element method. S102. Connect the positive and negative terminals of the main battery to the DC bus of the vehicle circuit through the contacts of the main relay and the backup relay in the relay control module. At the same time, connect the positive and negative terminals of the backup battery directly to the power management chip of the remote control module to provide the remote control module with a continuous power supply independent of the vehicle power supply. S103. Initialize key system parameters, including generating an initial sequence of entangled photon pairs for quantum key distribution to build a quantum key pool, creating a copy of the blockchain network's genesis block in local storage to establish a trusted ledger baseline, and downloading the initial global model weights from the federated learning server. The remote control module serves as prior knowledge for the local federated learning model, and the weight vector for multi-objective optimization is set. Used to quantify the relative importance of energy consumption, time, and lifetime loss in the optimization step S3.

3. The remote vehicle power-on control method for engineering machinery according to claim 2, characterized in that, The governing equations, which form the core of digital twin computation, in step S101 specifically include: S1011. Establish the coupled unsteady-state heat conduction control equations: ； And the stress balance governing equations: ； in, The density of the material; is the specific heat capacity at constant pressure; T is the temperature field; t is time; k(T) is the thermal conductivity related to temperature; I represents the electrochemical reaction heat generation rate; I represents the current; SOC represents the battery state of charge. This refers to the battery's internal resistance. This refers to the volume of a single battery cell; denoted as Cauchy stress tensor; b represents volume force. For displacement field; S1012. Discretize the governing equations using the finite element method to construct the global heat capacity matrix. Heat conduction matrix Mass matrix M and stiffness matrix And assemble them to obtain the discrete system equations: and ； Where T and U are the nodal temperature and displacement vectors, respectively; These are the nodal load vectors for chemical reaction heat and Joule heat, respectively; This is the equivalent nodal load vector for thermal stress. , e is the unit number, Let B be the element domain, B be the strain-displacement matrix, and D be the elasticity matrix. The coefficient of thermal expansion is The temperature shape function matrix; S1013. The implicit Newmark-β method is used to perform time integration on the discretized system equations. Within each simulation step Δt, the nonlinear coupled equations are solved through Newton-Raphson iteration, and the three-dimensional temperature field distribution T(x,y,z,t) inside the battery pack is output in real time. Mises equivalent force field distribution This data is then used as input for step S4.

4. The remote vehicle power-on control method for engineering machinery according to claim 2, characterized in that, Step S2 includes: S201. The remote control module uses the entangled photon pairs in the quantum key pool initialized in step S1 to perform a quantum key distribution process based on the BB84 protocol with the remote control system, and negotiates to generate a dynamic quantum key specific to this session. And utilize dynamic quantum key distribution Establish high-level secure communication channels; S202. In standby mode, the remote control module monitors the Received Signal Strength Indicator (RSSI) and packet loss rate of the wireless communication link in real time. And calculate the overall communication channel quality index: ； Where RSSI is the sliding window average signal strength; Sensitivity coefficient; Reference signal strength; S203, Electromagnetic interference intensity spectral density of the frequency band where the radio frequency front-end of the synchronous monitoring remote control module is located. And calculate the equivalent electromagnetic interference intensity: ； in, , These are the minimum and maximum operating frequencies for wireless communication of the remote control module, respectively. Based on the comprehensive communication channel quality index With equivalent electromagnetic interference intensity Dynamically adjust the heartbeat packet sending cycle: ； With radio frequency transmit power: ； in, This is the weighting coefficient for adjusting the heartbeat cycle based on communication channel quality. This is the adjustment weighting coefficient of electromagnetic interference intensity on the heartbeat cycle; This is a preset reference value for electromagnetic interference intensity; The minimum permissible radio frequency transmit power for the remote control module; This represents the maximum permissible radio frequency transmit power for the remote control module. This is the exponential attenuation coefficient of electromagnetic interference on the transmitted power.

5. The remote vehicle power-on control method for engineering machinery according to claim 4, characterized in that, Step S3 includes: S301. The remote control module receives the remote power-on command ciphertext from the remote control system through the quantum-encrypted secure channel established in step S2, and uses the dynamic quantum key generated in step S2. Decrypt to obtain the instruction hash value. With digital signatures The original instructions; S302. Call the local blockchain ledger copy established in step S1 to verify the instruction hash value. Has it been stored in the blockchain network consensus and verified using the public key corresponding to the remote control system? Verify digital signature The effectiveness; S303. After the instruction verification is successful, drive the digital twin constructed in step S1 to perform multiphysics coupling simulation using the current physical state read by the sensor as the initial condition, and predict the battery temperature field sequence within the preset time window [t, t+ΔT]. and stress field sequence ; S304, Constructing a system based on total energy consumption Total startup time and equipment lifespan loss For a multi-objective Pareto optimization problem where the objective is to be optimized: ； Among them, decision variables Includes start-up time, heating power curve, and controller parameters; constraints include channel quality calculated in step S2. and temperature predicted by digital twins ; S305. A non-dominated sorting genetic algorithm with an elitist strategy is used to solve the Pareto optimization problem, obtaining a Pareto optimal solution set. Based on the weight vector λ set in step S1, the optimal compromise solution is selected from the solution set, and the output is the optimal starting vector. The optimal start vector will then be passed to steps S4 and S5.

6. The remote vehicle power-on control method for engineering machinery according to claim 5, characterized in that, Step S4 includes: S401, Analyze the optimal start vector output in step S3. Extract the optimal heating power curve. and optimal controller parameters And set it as the benchmark target for preheating control; S402, at startup time Previously, the digital twin constructed in step S1 was used to obtain the three-dimensional temperature field inside the battery pack in real time. and stress field And calculate the error between the current state and the target state: and ； S403, State error , and ambient temperature Input a federated learning local model deployed on the remote control module. This federated learning local model is a neural network containing long short-term memory network layers and fully connected layers. Its output is the real-time power adjustment amount for each zone heating device. This makes the actual heating power ； S404, based on actual heating power Real-time integral calculation from the start of warm-up to the current moment Cumulative energy consumption: ； Based on the stress field time history, the equipment life loss index is estimated using a fatigue damage model: ； in, The fatigue damage coefficient of the battery pack material; The fatigue index of the battery pack material; is the fatigue strength limit of the battery pack material; is the maximum von Mises equivalent stress inside the battery pack at time t. S405, when the core temperature Reach the target temperature And the maximum von Mises equivalent force Below safety limit When the preheating is complete, a preheating completion signal is sent to step S5, and the final local model parameters obtained in this step are simultaneously sent to the system. Cumulative energy consumption and life loss index The "local optimization parameters and optimization contribution" are passed to step S6.

7. The remote vehicle power-on control method for engineering machinery according to claim 6, characterized in that, The federated learning local model deployed in step S403 optimizes the heating control, specifically including: S4031. The federated learning local model adopts an encoder-decoder structure, with the encoder being a bidirectional LSTM layer used to extract the input state sequence. The temporal characteristics are mapped to a two-layer fully connected network by the decoder, which is used to map the power adjustment to the time domain. ; S4032, Federated learning local model calculates output via forward propagation: ； ； in, This is the current input feature vector; It is in a hidden state; These are the model parameters, and their initial values ​​are... , Inherited from the federated learning global model weights downloaded in step S1 ; S4033. During the warm-up process, the federated learning global model uses real-time collected sensor data and digital twin prediction data to calculate the local loss: ； Where N is the total number of temperature sampling points used in a single forward inference by the local model of federated learning; This refers to the measured internal temperature of the battery pack or the simulated temperature output by the digital twin. The weighting coefficient for the heating power regularization term; And calculate the parameter gradient through backpropagation. This gradient is related to the final Together they constitute the "local optimization contribution" required for step S6.

8. The remote vehicle power-on control method for engineering machinery according to claim 6, characterized in that, Step S5 includes: S501. When the preheating completion signal from step S4 is received, and the system clock reaches the optimal start time in the optimal start vector of step S3. At that time, the remote control module generates the dynamic quantum key through step S2. Encrypted suction command; S502, the relay control module receives and decrypts the activation command, and simultaneously drives its internal main relay coil and backup relay coil, so that the contacts of the dual redundant relays are in a preset, minimal time difference. The internal circuits close sequentially, and the main battery is safely connected to the vehicle's DC bus. S503. Establish a digital twin-physical entity synchronization loop: using the battery voltage predicted by the digital twin based on the coupling equation in step S1. Bus current As a priori state estimate The voltage measured by the sensor Current As observation vector ; S504. State fusion using the Kalman filter algorithm: Calculate the Kalman gain. ； And update the posterior state estimate: ； Where H is the observation matrix; P is the error covariance matrix; and R is the observation noise covariance matrix. S505, Based on the trace of the posterior error covariance matrix Assess system uncertainty and dynamically calculate the compensated response delay: ； in, As a reference delay; This is the proportional coefficient, which is advanced in subsequent control timing. Issue commands to dynamically compensate for system response delays and estimate the compensated synchronization state. Proceed to step S6.

9. The remote vehicle power-on control method for engineering machinery according to claim 8, characterized in that, Step S6 includes: S601. Based on the compensated synchronization state output from step S5, estimate... Including main battery voltage With bus current The remote control module sends a start command to the engine control unit to control the engine ignition or drive motor start, and sequentially activates the hydraulic main pump solenoid valve and the working device control valve, so that the whole vehicle working system enters the standby state. S602. Based on the local optimization parameters and optimization contributions passed in step S4, i.e. the final weights of the local federated learning model. and training gradient Calculate the local contribution of this device in this round of operation: ； in, The cosine similarity function; The global average gradient is obtained from the federated learning server; For reference, energy consumption and lifespan loss values; Weighting coefficients that contribute to the optimization of the local model in federated learning; For energy consumption control effect; Weighting coefficients for the effectiveness of lifespan loss control; S603, For matters requiring reporting, including The system state data is hierarchically compressed and encoded: a lightweight neural network classifier categorizes it into high, medium, and low priorities based on data sensitivity. ; S604. Channel quality evaluated in real time according to step S2. With data priority Adaptive coding strategy selection: high-priority data is encrypted with quantum key distribution using one-time pad and forward error correction code is added; medium-priority data is encrypted with AES-256 and lossless compression is used; low-priority data is encrypted with lightweight encryption and lossy compression. S605. Apply the processed data packet to the dynamic quantum key generated in step S2. The data is encrypted and uploaded to the remote control system via a secure channel. At the same time, the hash value of the uploaded data packet is submitted to the blockchain network for evidence storage, forming a traceable full-process operation record.

10. The remote vehicle power-on control method for engineering machinery according to claim 6, characterized in that, Step S7 includes: S701. The remote control module receives a stop command from the remote control system, triggers the operation termination process, and queries and retrieves the hash sequence of the entire operation data packet stored in step S6 from the blockchain network based on the device identifier and timestamp. S702. Based on the full-process data, calculate the global energy efficiency index of this operation: ； in, To estimate the effective mechanical work through the fusion of digital twins and sensors; These are the energy consumption for operation and communication, respectively. , The mean and standard deviation of the total energy consumption of all equipment participating in this group operation; S703, Simultaneous carbon footprint audit report, calculating total carbon emissions: ； And generate a carbon emission intensity index: ； in, These are the carbon emission factors for the power grid and diesel, respectively. For the corresponding power; Total operation time; This refers to the total effective mechanical work generated by all the engineering machinery involved in the collaborative control during this group remote power-on operation; S704. The federated learning server collects the local contribution calculated by each client in step S6. and the corresponding model update gradient Perform weighted aggregation: ； To update the global model, where The global learning rate; The weights of the new global model obtained by federated learning are the weighted aggregates of all construction machinery clients in this group operation. This is the sum of the local contributions of all engineering machinery clients participating in this collaborative control; The weights of the old global model stored in the federated learning server before this global model update; S705, the calculated results and the hash value of the new global model after aggregation A new blockchain transaction is jointly written, and after consensus is reached, it is stored on the blockchain to ensure the immutability of this operation. The session state is then cleared, and the system switches to a deep security standby mode based on a quantum key pool.

Images