Search

CN-122018317-A - Multisource error fusion compensation control method in logistics line butt joint process

CN122018317ACN 122018317 ACN122018317 ACN 122018317ACN-122018317-A

Abstract

The invention relates to the technical field of logistics management, and particularly discloses a multisource error fusion compensation control method in a logistics line butt joint process, which comprises the steps of collecting environmental parameters and equipment operation data in real time through a deployment sensor network, constructing a proxy model by combining a Gaussian process regression method, and predicting energy consumption and environmental quality performance under different control strategies; the intelligent control system for the logistics line butt joint error has the advantages that the intelligent control system for the logistics line butt joint error is intelligent, self-adaptive and high in robustness, the intelligent control system for the logistics line butt joint error is capable of dynamically adapting to the change of the use mode of the logistics line and improving the optimization accuracy through a closed loop feedback mechanism, under abnormal conditions, the system can identify the sensor or equipment fault based on the prediction residual error and automatically switch to a standby control strategy, and the stable operation of the system is ensured.

Inventors

  • QIU XUEFEI
  • LIU YI

Assignees

  • 广东久盈精密技术有限公司

Dates

Publication Date
20260512
Application Date
20260204

Claims (10)

  1. 1. A multisource error fusion compensation control method in a logistics line butt joint process is characterized by comprising the following steps: acquiring environmental parameters, equipment running states and error data in real time through a sensor network deployed inside a logistics line, wherein the environmental parameters comprise temperature, humidity and the tension of a belt in the logistics line; Cleaning and normalizing the collected data, and constructing a proxy model for predicting control strategy errors; adopting a particle swarm optimization algorithm to cooperatively optimize the operation parameters of the motor in the object flow line in combination with the agent model; performing incremental training on the proxy model by periodically utilizing the newly-added operation data to enable the proxy model to adapt to the change of the use mode of the object flow line; The optimized control parameters are issued to a logistics line automation system, and the running state of the system is monitored in real time to form closed-loop optimization control; and identifying sensor abnormality or equipment fault based on the agent model prediction residual error, and switching to a standby control strategy when abnormality occurs.
  2. 2. The method for multi-source error fusion compensation control in a logistics line butt joint process according to claim 1, wherein the constructing a proxy model for predicting control strategy errors specifically comprises: Performing feature extraction on the cleaned and normalized multi-source historical data to extract input feature vectors related to energy consumption, current and motor speed; establishing a nonlinear mapping relation based on a Gaussian process regression method, taking an input characteristic vector as input, and outputting corresponding docking coordinate errors, docking placement form errors and docking completion time errors; the Bayesian optimization algorithm is utilized to automatically adjust and optimize the kernel function parameters of the Gaussian process, so that the generalization capability of the agent model under different working conditions is improved; And packaging the trained agent model into a callable interface for the particle swarm optimization algorithm to call in real time so as to evaluate the multi-objective performance of the candidate solution.
  3. 3. The method for multi-source error fusion compensation control in a process of logistics line docking according to claim 2, wherein the proxy model further comprises: After each particle swarm optimization iteration is completed, adding a newly generated control strategy and corresponding actual operation feedback data into a training sample set; a sliding window mechanism is adopted to retain effective data in the last N days, and an over-time or abnormal sample is removed to prevent model drift; the incremental learning strategy is used for online updating of the Gaussian process regression model, so that the agent model can dynamically adapt to the change of the use mode of the object streamline; And combining confidence intervals of model prediction, and introducing a search and utilization balance mechanism in the optimization process.
  4. 4. The method for multi-source error fusion compensation control in a logistics line butt joint process according to claim 3, wherein the method for collaborative optimization of the operation parameters of the motors in the logistics line by adopting a particle swarm optimization algorithm in combination with a proxy model is characterized by comprising the following steps: constructing a multidimensional decision variable space containing motor set current and start-stop time as a search space of a particle swarm optimization algorithm; Based on the agent model, rapidly predicting a docking coordinate error, a docking placement form error and a docking completion time error of each particle under a corresponding control strategy; designing a multi-target fitness function, comprehensively considering a docking coordinate error, a docking placement form error and a docking completion time error, and dynamically adjusting each target weight through a fuzzy logic; And iteratively searching the optimal control parameter combination under the guidance of the multi-objective fitness function by utilizing a particle swarm optimization algorithm, and outputting the optimal control parameter combination to a control execution module.
  5. 5. The method for multi-source error fusion compensation control in a logistics line docking process according to claim 4, wherein the collaborative optimization of the operation parameters of the motor in the logistics line comprises: introducing a local exploration mechanism into each optimization iteration, carrying out refined search on the control strategy in the neighborhood of the current optimal solution, and improving the solution precision; Establishing a memory library by combining the history optimization result, and recording logistics line load and environment response modes in different time periods, wherein the memory library is used for assisting rapid convergence in future similar scenes; And comparing the optimization result with the actual feedback error, and reversely correcting the prediction deviation of the proxy model.
  6. 6. The method for multi-source error fusion compensation control in a logistics line docking process according to claim 5, wherein the periodically performing incremental training on the proxy model by using the newly added operation data comprises the following steps: Periodically collecting and storing newly generated operation data, wherein the operation data comprises real-time energy consumption information, an environmental parameter acquisition result and an equipment state monitoring value; Performing outlier detection and elimination operation on the newly added data, and performing standardization processing to improve data consistency; Gradually fusing the newly added data into the existing agent model by adopting an incremental learning algorithm, and updating model parameters on the premise of not losing historical knowledge; And evaluating the prediction performance of the updated agent model by a cross verification method, and putting the model into the next optimization period for use after confirming that the model meets the set precision requirement.
  7. 7. The method for multi-source error fusion compensation control in a logistics line docking process of claim 6, wherein the incremental training of the proxy model comprises: identifying trend characteristics of the use mode of the object streamline along with the change of seasons or time by utilizing a time sequence analysis technology, and dynamically adjusting the learning strategy of the agent model according to the trend characteristics; Setting a sliding window mechanism, only reserving operation data within a preset time range as a model training sample, and eliminating outdated data to improve the response speed of the model; according to the feedback information of the user, the correction weight of the output result of the agent model is adjusted, so that the model prediction is closer to subjective evaluation indexes in the actual application scene; And establishing a model version management mechanism, and recording time nodes, training data sources and performance evaluation results of each model update.
  8. 8. The method for multi-source error fusion compensation control in a logistics line docking process according to claim 7, wherein the step of issuing the optimized control parameters to a logistics line automation system and monitoring the running state of the system in real time comprises the following steps: converting the optimal control parameters output by the multi-objective optimal regulation and control module into an instruction format conforming to the standard of the logistics line automation system; Transmitting the instruction to a corresponding logistics line equipment control system by using a communication protocol; acquiring operation data of all subsystems in the logistics line automation system in real time, wherein the operation data comprise a butt joint coordinate error, environmental parameters and equipment working states; And evaluating the effect of the current control strategy according to the real-time operation data, and feeding back the evaluation result to realize closed-loop control.
  9. 9. The method for multi-source error fusion compensation control in a process of interfacing a logistics line according to claim 8, wherein the forming a closed-loop optimization control comprises: Before issuing a control instruction, previewing a control strategy to be implemented through a simulation tool, predicting possible influence of the control strategy, and adjusting control parameters according to a previewing result; Introducing a fault-tolerant mechanism, and automatically switching to a standby control strategy when a communication fault or equipment abnormality is detected, so as to ensure the continuity and stability of the system; Comparing the difference between the actual running effect and the expected target, analyzing the deviation reason by adopting a machine learning algorithm, and dynamically adjusting the agent model and the optimization algorithm parameters according to the deviation reason; And periodically generating a detailed operation report, and recording indexes of control strategy execution conditions, equipment response time and energy consumption saving amount for subsequent performance audit and optimization decision.
  10. 10. The method for multi-source error fusion compensation control in a process of logistics line docking according to claim 9, wherein the identifying of sensor abnormality or equipment failure based on proxy model prediction residual error and switching to a standby control strategy when abnormality occurs comprises the steps of: acquiring actual operation data of each sensor and equipment in a commodity flow line in real time, and inputting the operation data into a trained agent model to generate a predicted value; Comparing the predicted value output by the proxy model with the corresponding actual measured value, calculating residual error between the predicted value and the corresponding actual measured value, and judging whether significant deviation exists or not through setting a threshold value; if the residual error is detected to continuously exceed the set threshold value, a multidimensional data analysis mechanism is started, the abnormal type is identified by combining historical data and a statistical analysis method, and sensor drift, data transmission errors or equipment faults are distinguished; after confirming that the abnormality exists, a preset standby control strategy is automatically activated to maintain the stable operation of the system, an alarm signal is synchronously triggered to inform operation and maintenance personnel to intervene, and meanwhile, an abnormal event is recorded for subsequent diagnosis and model correction.

Description

Multisource error fusion compensation control method in logistics line butt joint process Technical Field The invention relates to the technical field of logistics management, in particular to a multisource error fusion compensation control method in a logistics line butt joint process. Background With the acceleration of the urban process and the continuous increase of the energy consumption of logistics lines, the logistics line energy management system plays an increasingly important role in realizing the energy conservation and emission reduction targets. In recent years, with the development of internet of things (IoT), artificial intelligence and big data analysis technologies, intelligent logistics control systems gradually transition from traditional rule-based static control to data-driven dynamic optimization regulation. At present, most logistics line automation systems have the capability of monitoring the docking errors of environmental parameters and equipment running states in real time, and can execute a preset strategy through a logistics line controller to adjust the docking errors. The prior art has the following defects: The existing logistics line docking error optimization based on Particle Swarm Optimization (PSO) algorithm has the problems that the target weight setting is unreasonable and a dynamic weighing mechanism is lacked when the logistics line docking error optimization faces multi-target cooperative control. Specifically, multiple optimization targets are generally considered comprehensively, however, the conventional PSO algorithm adopts a static weight configuration mode, and it is difficult to dynamically adjust the priority of each target according to environmental changes and user requirements. Therefore, it is needed to provide an optimized control method capable of achieving multi-objective dynamic trade-off so as to improve the intelligent management level of the logistics line. Disclosure of Invention In order to solve the technical problems, the technical scheme solves the problems in the background technology. In order to achieve the above purpose, the invention adopts the following technical scheme: a multisource error fusion compensation control method in a logistics line butt joint process comprises the following steps: acquiring environmental parameters, equipment running states and error data in real time through a sensor network deployed inside a logistics line, wherein the environmental parameters comprise temperature, humidity and the tension of a belt in the logistics line; Cleaning and normalizing the collected data, and constructing a proxy model for predicting control strategy errors; adopting a particle swarm optimization algorithm to cooperatively optimize the operation parameters of the motor in the object flow line in combination with the agent model; performing incremental training on the proxy model by periodically utilizing the newly-added operation data to enable the proxy model to adapt to the change of the use mode of the object flow line; The optimized control parameters are issued to a logistics line automation system, and the running state of the system is monitored in real time to form closed-loop optimization control; and identifying sensor abnormality or equipment fault based on the agent model prediction residual error, and switching to a standby control strategy when abnormality occurs. The invention further provides a further scheme, wherein the construction of the proxy model for predicting the control strategy error specifically comprises the following steps: Performing feature extraction on the cleaned and normalized multi-source historical data to extract input feature vectors related to energy consumption, current and motor speed; establishing a nonlinear mapping relation based on a Gaussian process regression method, taking an input characteristic vector as input, and outputting corresponding docking coordinate errors, docking placement form errors and docking completion time errors; the Bayesian optimization algorithm is utilized to automatically adjust and optimize the kernel function parameters of the Gaussian process, so that the generalization capability of the agent model under different working conditions is improved; And packaging the trained agent model into a callable interface for the particle swarm optimization algorithm to call in real time so as to evaluate the multi-objective performance of the candidate solution. As a further scheme of the invention: the proxy model further includes: After each particle swarm optimization iteration is completed, adding a newly generated control strategy and corresponding actual operation feedback data into a training sample set; a sliding window mechanism is adopted to retain effective data in the last N days, and an over-time or abnormal sample is removed to prevent model drift; the incremental learning strategy is used for online updating of the Gaussian process regression mod