High-precision machining method and system for cylindrical new energy battery shell

Abstract

The invention discloses a high-precision machining method and system for a cylindrical new energy battery shell, and relates to the technical field of machining regulation and control. The method comprises the steps that along with the machining process of a numerical control machine tool on a target battery shell, a sensing array group is synchronously driven for monitoring, and machining state data are determined; the method comprises the following steps: defining a machining mode and a machining error type, and constructing and generating a machining regulation and control module based on an intelligent agent cluster of grinding force, thermal deformation, vibration and wear; and if the error of the processing state data is out of limit, the processing regulation and control module is triggered to execute dynamic empowerment and agent cluster parallel decision and balanced game regulation, a processing regulation and control strategy is generated, and processing regulation and control driving is executed. The technical problem that in the prior art, due to the fact that a cylindrical new energy battery shell is affected by multi-factor coupling in the machining process, the machining precision fluctuates, and the machining quality is unstable is solved, and the technical effect of improving the machining precision consistency and the machining quality stability is achieved through real-time monitoring and dynamic regulation and control over the machining process.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of processing regulation, and in particular to a high-precision processing method and system for a cylindrical new energy battery shell. BACKGROUND

[0002] The cylindrical new energy battery shell is an important structural component of power batteries and energy storage batteries, and the processing quality thereof directly affects the assembly precision, thermal management performance and overall safety and reliability of the battery cell. The existing battery shell is mostly made of aluminum alloy or stainless steel material and is completed through multi-process processing such as numerical control turning, grinding and polishing. Due to the thin wall and poor rigidity of the battery shell, the processing precision fluctuates, the roundness deviation and surface defects frequently occur due to the influence of multi-factor coupling such as grinding force fluctuation, tool wear, thermal deformation and vibration during the processing. In addition, the traditional numerical control processing mainly relies on fixed process parameters and manual experience adjustment, lacks real-time sensing and self-adaptive regulation mechanism for the processing state, and is difficult to cope with the dynamically changing processing conditions, resulting in insufficient product consistency and low yield. SUMMARY

[0003] The application provides a high-precision processing method and system for a cylindrical new energy battery shell, which solves the technical problem that the processing precision fluctuates due to the influence of multi-factor coupling during the processing of the cylindrical new energy battery shell in the prior art, resulting in unstable processing quality.

[0004] In a first aspect, the application provides a high-precision processing method for a cylindrical new energy battery shell, which comprises: With the processing progress of the target battery shell by the numerical control machine tool, the sensing array group is synchronously driven for monitoring to determine the processing state data; a processing regulation module is generated through definition of the processing mode and the processing error type and construction of an intelligent agent cluster based on the grinding force, thermal deformation, vibration and wear; if the processing state data has error overrun, the processing regulation module is triggered to perform dynamic weighting based on processing mode matching and parallel decision and balanced game adjustment of the intelligent agent cluster to generate a processing regulation strategy and perform processing regulation driving.

[0005] In a second aspect, the application provides a high-precision processing system for a cylindrical new energy battery shell, which comprises: A monitoring module: with the processing progress of the target battery shell by the numerical control machine tool, the sensing array group is synchronously driven for monitoring to determine the processing state data; a module generation module: a processing regulation module is generated through definition of the processing mode and the processing error type and construction of an intelligent agent cluster based on the grinding force, thermal deformation, vibration and wear; a strategy execution module: if the processing state data has error overrun, the processing regulation module is triggered to perform dynamic weighting based on processing mode matching and parallel decision and balanced game adjustment of the intelligent agent cluster to generate a processing regulation strategy and perform processing regulation driving.

[0006] The one or more technical solutions provided in the application have at least the following technical effects or advantages: With the machining process of the numerical control machine tool on the target battery shell, the sensing array group is synchronously driven for monitoring to determine machining state data. By defining machining modes and machining error types, a machining regulation module is generated based on the intelligent agent cluster construction of grinding force, thermal deformation, vibration and wear. If the machining state data has an error out of limit, the machining regulation module triggers the dynamic weighting based on machining mode matching, and the intelligent agent cluster parallel decision and balanced game adjustment, to generate a machining regulation strategy and execute machining regulation driving. The technical problem of fluctuation of machining precision and instability of machining quality caused by the influence of multi-factor coupling on the cylindrical new energy battery shell in the machining process is solved. Through real-time monitoring and dynamic regulation of the machining process, the technical effects of improving machining precision consistency and machining quality stability are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0008] Figure 1 A high-precision machining method flowchart for a cylindrical new energy battery shell provided by the embodiment of the present application is shown. Figure 2 A high-precision machining system structure schematic diagram for a cylindrical new energy battery shell provided by the embodiment of the present application is shown.

[0009] Legend of the drawings: monitoring module 11, module generation module 12, strategy execution module 13. DETAILED DESCRIPTION

[0010] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects according to the present application are described in detail as follows.

[0011] Embodiment one, as shown in the present application, a high-precision machining method for a cylindrical new energy battery shell is provided, wherein the method comprises: Figure 1 With the machining process of the numerical control machine tool on the target battery shell, the sensing array group is synchronously driven for monitoring to determine machining state data.

[0012] ​In the embodiment of the present application, a multi-channel sensor array composed of force sensors, temperature sensors, vibration acceleration sensors and acoustic emission sensors is installed on the tool holder or end effector of the numerical control machine tool; when the machine tool performs turning, grinding or precision polishing processing, the system activates the acquisition channels of the sensor array in real time to synchronously acquire the grinding force signals, temperature rise signals, vibration signals and acoustic characteristic signals in the processing process at a fixed sampling frequency; the original signals output by each sensor are transmitted to the edge processing unit, the signals are filtered in time domain and frequency domain, abnormal noise is removed and features are extracted, forming a multi-dimensional processing state vector containing force mean value, force fluctuation amplitude, temperature rise rate, vibration acceleration peak value and acoustic emission energy and other indicators; through normalization processing and time series reconstruction of the processing state vector, the processing state data of the target battery shell at the current processing time is obtained.

[0013] By defining the processing mode and the processing error type, the intelligent agent cluster based on the grinding force, thermal deformation, vibration and wear is constructed to generate the processing control module.

[0014] Further, by defining the processing mode and the processing error type, the intelligent agent cluster based on the grinding force, thermal deformation, vibration and wear is constructed to generate the processing control module, including: Classify the processing state and define the processing mode, wherein the processing mode at least contains normal stable mode, slight chatter mode, thermal dominant mode and wear end stage mode; define the processing error type, wherein the processing error type contains the main processing error based on roundness and coaxiality, and the secondary processing error based on surface roughness and residual stress; construct the intelligent agent in multiple dimensions from the grinding force, thermal deformation, vibration and wear, and through the first-order training based on the dynamic weighting of the intelligent agent in the processing mode and the second-order training based on the balanced game of the intelligent agent of the processing error, the processing control module is constituted.

[0015] The collected processing state data is classified and processed, and according to the grinding force fluctuation characteristics, temperature change rate, vibration frequency spectrum distribution and tool wear degree and other multi-source characteristic parameters, a clustering algorithm such as K-means is used to classify the processing state, and a processing mode set is defined. The processing mode at least includes a normal stable mode, a slight chatter mode, a heat dominant mode and a wear end mode. Among them, the force signal in the normal stable mode is smooth, the vibration amplitude is low, and the temperature gradient is gentle; the slight chatter mode shows that the characteristic peak in the vibration frequency spectrum shifts; the heat dominant mode corresponds to a significant increase in the temperature rise rate and is accompanied by micro-plastic deformation; and the wear end mode reflects the sudden increase of tool wear rate and abnormal force fluctuation. The processing error type is defined, and a multi-level error index system is established, wherein the main processing error includes the shape and position error based on roundness and coaxiality, which is used to represent the macro geometric accuracy; the secondary processing error includes the micro quality error based on surface roughness and residual stress, which is used to reflect the surface integrity and fatigue performance. Through the hierarchical definition of error type, multi-dimensional evaluation and hierarchical control of processing quality can be realized. Four core physical dimensions of grinding force, thermal deformation, vibration and wear are taken as the core, and corresponding agent models are constructed. Each agent is composed of a state space, an action space and a reward function, wherein the state space is composed of real-time monitoring characteristic parameters of the corresponding dimension, the action space is related to the adjustable processing control quantity (such as feed speed, spindle speed, cooling flow, tool compensation quantity, etc.) of the dimension, and the reward function takes the minimization of error convergence rate or the optimization of energy consumption as the optimization goal.

[0016] In the training stage, first, the agent dynamic weighting first-order training based on the processing mode is performed, that is, the weight coefficients of the agents are adjusted according to the working condition characteristics in different processing modes, so that they can automatically focus on the dominant influencing factors according to the real-time mode; then, the agent balanced game second-order training based on the processing error constraint is performed, and the regulation and control strategies of the agents in different dimensions are coordinated through the multi-agent game learning mechanism, so that the global optimal distribution and coupled coordination of the processing parameters are realized. After the two-stage training, the processing regulation and control module generated has the self-learning, self-adaptation and dynamic game ability, and can automatically select the appropriate regulation and control strategy according to different modes in the processing process, so as to realize the intelligent stable control of the processing precision and quality of the battery shell.

[0017] Further, the multi-dimensional agent construction is performed from the grinding force, thermal deformation, vibration and wear, including: The first state space and the first parameter space are defined to deploy the grinding force agent with the greedy goal of minimizing the force fluctuation; the second state space and the second parameter space are defined to deploy the thermal deformation agent with the greedy goal of minimizing the temperature rise; the third state space and the third parameter space are defined to deploy the vibration agent with the greedy goal of minimizing the amplitude; and the fourth state space and the fourth parameter space are defined to deploy the wear agent with the greedy goal of minimizing the tool wear rate.

[0018] The first state space is composed of the average value of grinding force, fluctuation amplitude, contact time and tool feed depth, and the first parameter space corresponds to control variables such as spindle speed, feed speed and cutting depth; the grinding force agent performs dynamic fitting based on real-time force signals to achieve mechanical stability optimization by minimizing force fluctuation.

[0019] The second state space is composed of temperature gradient, temperature rise rate and heat dissipation efficiency, and the second parameter space includes variables such as coolant flow, tool contact time and environmental temperature; the thermal deformation agent responds to the change law of temperature field for response control, taking temperature stability as the evaluation function to reduce thermal-induced dimensional error.

[0020] The third state space is composed of vibration acceleration amplitude, frequency spectrum characteristics and chatter frequency, and the third parameter space includes variables such as tool feed rate, clamping stiffness and path smoothness; the vibration agent adjusts the machining process by real-time analysis of vibration energy distribution to achieve vibration energy minimization and machining trajectory stabilization.

[0021] The fourth state space is composed of tool edge wear, wear rate and tool contact stress, and the fourth parameter space includes variables such as tool compensation, feed angle and cooling method; the wear agent aims to prolong tool life and uniformize wear, and adjusts tool usage strategy through multi-round game learning to prevent sudden drop in machining performance.

[0022] If the machining state data has error out-of-limit, the machining control module is triggered to perform dynamic weighting based on machining mode matching, and the agent cluster parallel decision and balanced game adjustment to generate machining control strategy and execute machining control driving.

[0023] When the real-time collected machining state data has error out-of-limit, the machining control module is triggered. The machining control module first performs machining mode matching based on the current machining state data and historical mode database, calculates the similarity between the machining state vector and the mode feature vector to determine the current machining mode. Then, according to the matched machining mode, the grinding force agent, thermal deformation agent, vibration agent and wear agent are executed for dynamic weighting calculation. The dominant influence factors have different weight distributions in different modes: for example, the vibration agent weight increases in the slight chatter mode, and the thermal deformation agent weight increases in the thermal dominant mode. The system updates the agent weight matrix in real time according to the mode matching result to realize dynamic focusing of machining characteristics on the dominant agent.

[0024] After the dynamic empowerment is completed, the agents execute decision reasoning in parallel to generate an initial set of regulation strategies for their target dimensions, including spindle speed correction, feed rate adjustment, cooling flow compensation, tool posture correction, etc. Subsequently, the machining regulation module processes the strategy sets output by the four types of agents through balanced game processing, solves the conflict and coordination relationship between the regulation strategies through the establishment of a joint optimization model based on Nash equilibrium, and obtains a globally optimal machining regulation strategy with multi-dimensional balance. Finally, the system decouples the machining regulation strategy into multiple executable sub-strategies, converts it into corresponding control instruction parameters of the numerical control machine tool, including speed correction instructions, feed path compensation instructions, cooling control instructions, and tool posture adjustment instructions, and issues them to the corresponding drive units for execution, realizing real-time closed-loop regulation of the machining process.

[0025] Further, the drive sensor array is used to monitor and determine the machining state data, including: A sensor array is installed on the end effector of the numerical control machine tool. As the target battery shell is being processed, the machining state data is collected in synchronization. Error overrun judgment is performed on the machining state data. If there is an error overrun, a machining regulation instruction is generated.

[0026] A multi-dimensional sensor array composed of force sensors, temperature sensors, vibration acceleration sensors, and acoustic emission sensors is installed on the end effector or tool holder of the numerical control machine tool. The sensor array is electrically connected to the machine tool control system through a high-speed sampling module. When the target battery shell enters the machining stage, the sensor array collects signals such as grinding force, temperature change, vibration amplitude, and acoustic emission energy in real time as the tool path moves synchronously, and transmits the signals to the data acquisition unit after analog-to-digital conversion. Through band-pass filtering, denoising, and feature extraction of the original signals, machining state data containing force fluctuation mean, temperature rise rate, vibration frequency spectrum characteristics, and tool wear characteristics is formed.

[0027] Error overrun judgment is performed on the machining state data, including: comparing the real-time state data with the reference threshold interval in the machining mode database. When the roundness deviation, coaxiality, or surface roughness index exceeds the set threshold range, or abnormal features such as rapid temperature gradient rise and vibration peak shift occur, it is determined that there is an error overrun event. After the system determines that there is an error overrun, it automatically generates a machining regulation instruction and transmits it to the machining regulation module to trigger the subsequent machining mode matching, agent dynamic empowerment, and regulation strategy generation process, realizing closed-loop control from abnormal detection to regulation response.

[0028] Further, triggering the machining regulation module includes: Upon receipt of the machining regulation instruction, the mechanical drive parameters of the numerical control machine tool are read. Based on the machining state data and the mechanical drive parameters, the machining regulation module is triggered to make industrial control adjustment decisions.

[0029] After receiving the machining control command, the system automatically reads the mechanical drive parameters of the CNC machine tool, including spindle speed, feed rate, tool position, coolant flow rate, drive current, and servo response delay. This data is collected in real time by the machine tool control bus and synchronized to the machining control module to characterize the current dynamic execution status of the machine tool.

[0030] The processing control module jointly analyzes the mechanical drive parameters and the processing status data collected by the sensor array, and compares and analyzes the degree of deviation between the current execution state and the target processing mode; if the deviation exceeds the preset threshold, the industrial control adjustment decision process is triggered.

[0031] During the adjustment decision-making process, the machining control module, based on the agent states in four dimensions—grinding force, thermal deformation, vibration, and wear—calls a pre-trained pattern matcher and policy inferencer to comprehensively calculate the response priority and control weight of the agents in each dimension. A fast decision-making algorithm determines the optimal adjustment path, including control commands such as spindle speed correction, feed rate adjustment, tool compensation step size, and cooling system adjustment coefficient. The adjustment decision result is then output as a machining control strategy, providing input for subsequent machining parameter updates and machine tool execution unit driving.

[0032] Furthermore, performing dynamic weighting based on processing pattern matching includes: Using the processing status data, a target processing mode is determined through matching; based on the target processing mode and the processing status data, multi-dimensional dynamic weighting is performed to determine the agent weight distribution.

[0033] Using the aforementioned machining status data as input, the machining mode matching module is invoked. A feature similarity algorithm compares the real-time status with mode templates in the machining mode database. The machining mode database contains four preset typical status templates: normal stable mode, slight chatter mode, heat-dominated mode, and late-wear mode. Each mode template consists of multi-dimensional feature vectors, including grinding force fluctuation rate, temperature rise rate, vibration energy spectrum characteristics, and tool wear rate. During the matching process, the system calculates the cosine similarity or Euclidean distance between the real-time machining status vector and each mode template vector to determine the closest mode category, thereby identifying the target machining mode.

[0034] After determining the target processing mode, the system performs multi-dimensional dynamic weighting calculation according to the dominant influencing factors of the mode and real-time processing state data. The dynamic weighting process includes: establishing a four-dimensional intelligent agent weight vector; the system calculates the weight distribution of each dimension according to the dominant physical characteristics under the target mode, for example, in the heat dominant mode, the weight of the thermal deformation intelligent agent increases, and in the vibration mode, the weight of the vibration intelligent agent increases. The weight update adopts an adaptive adjustment strategy, which dynamically corrects the weight coefficient by real-time calculation of the variance and sensitivity index of each dimension state variable, so that the main influencing factor obtains a higher priority for regulation and control, and the secondary factor is automatically de-weighted; At the same time, the time sequence sliding window mechanism is introduced to smooth the filtering of historical processing state data, avoiding the instability of control caused by frequent fluctuations of weight. The final output of the intelligent agent weight distribution is used as the input parameter of the subsequent parallel decision and balanced game, realizing the adaptive focusing of the processing control module on multi-factor interference and the optimization of the processing strategy.

[0035] Further, the intelligent agent cluster parallel decision and balanced game adjustment includes: The parallel multi-dimensional directional control decision is executed to determine the control strategy set; the intelligent agent weight distribution is used to fine-tune the relative strategy of the control strategy set under balanced game to determine the processing control strategy; wherein the control process of the grinding force dimension includes: initializing the first state space according to the processing state data, and the grinding force intelligent agent executes directional decision based on the greedy target to determine the first control strategy.

[0036] When the agent weight distribution is obtained, the processing and regulation module starts the multi-agent parallel decision mechanism. Each agent independently executes directional regulation decision according to its own objective function and state space, and generates initial regulation strategies in four dimensions of grinding force, thermal deformation, vibration and wear. The parallel decision is realized through an asynchronous parallel computing framework, which can simultaneously process different dimensional regulation tasks and quickly output a regulation strategy set containing spindle speed adjustment, feed rate correction, cooling flow compensation coefficient and tool compensation parameters. With the agent weight distribution as the constraint condition, the regulation strategy set is subjected to balanced game analysis, and a joint optimization model under multi-agent interaction is constructed. By establishing a strategy coordination equation based on Nash equilibrium, the income difference and constraint conflict between strategies in each dimension are evaluated, and relative strategy adjustment is performed to make each regulation dimension tend to the energy optimal and error minimized state in the global range, and finally the balanced and modified processing and regulation strategy is output. Among them, the regulation process in the grinding force dimension includes: initializing the first state space according to the processing state data, the grinding force agent takes minimizing force fluctuation as the greedy target, executes directional decision algorithm based on reinforcement learning, calculates the optimal adjustment amount of spindle speed and feed depth in real time, and determines the first regulation strategy; the first regulation strategy dynamically modifies the parameter output through the convergence difference of the predicted force fluctuation response curve and the energy consumption model, realizing the mechanical stability control of the grinding process.

[0037] Further, the processing and regulation driving is executed, including: According to the minimum control unit of the numerical control machine tool, the processing and regulation strategy is decoupled to determine a plurality of sub-regulation strategies; a numerical control instruction based on the plurality of sub-regulation strategies is generated and issued to the corresponding control unit for feedback regulation of the target battery case processing process.

[0038] When the processing and regulation module outputs the processing and regulation strategy optimized by the balanced game, the system decouples and analyzes the processing and regulation strategy according to the minimum control unit of the numerical control machine tool. The minimum control unit of the numerical control machine tool includes spindle drive unit, feed servo unit, cooling control unit, tool compensation unit and clamping positioning unit, etc., each unit corresponds to an independent control interface and parameter set. The regulation strategy decoupling process is realized through a parameter mapping matrix, which decomposes the multi-dimensional control variables (such as spindle speed adjustment, feed rate correction, cooling flow compensation coefficient, tool posture adjustment angle, etc.) in the global processing and regulation strategy into sub-regulation strategies executable by the corresponding control unit.

[0039] The system generates numerical control instruction sequences according to each sub-control strategy, the instructions are in the form of standard G code or M code, and are accompanied by processing time stamps and execution priority labels, ensuring time sequence synchronization and conflict resolution in multi-unit control process. The controller issues the generated numerical control instructions to the corresponding execution unit through the machine tool bus, and each control unit executes the corresponding adjustment action immediately after receiving the instructions, including real-time correction of spindle speed, feed servo acceleration and deceleration control, cooling flow regulation, and tool position compensation, etc. During the execution process, the sensor array continuously collects processing feedback signals, and the system performs closed-loop correction according to the deviation of the feedback data and the target control parameters, so that the control result converges to the target interval in real time. When the feedback deviation is less than the set tolerance, the system determines that the processing control has reached a steady state and continues to execute the subsequent processing steps.

[0040] Further, the processing state data is updated with the processing time sequence after tracking monitoring according to the sensor array; error determination and adjustment decision based on the processing state data are continuously performed until the processing of the target battery shell is completed.

[0041] While executing the processing control driving, the sensor array remains in a working state to continuously collect grinding force, temperature, vibration and acoustic emission signals at a frequency synchronized with the spindle speed and feed beat of the numerical control machine tool, and to generate a feedback data set after processing control in real time; the system compares the feedback data set with the processing state data at the previous time according to the collection time sequence, forms an updated processing state vector, and continuously updates the processing state database in a sliding time window mechanism, realizing time sequence dynamic recording of the processing process. On this basis, the system continuously performs error determination and adjustment decision based on the updated processing state data, including: comparing the current processing state data with the standard processing threshold set, if any error index (such as roundness, coaxiality, surface roughness or residual stress) still exceeds the allowed range, the processing control module is automatically retriggered to execute new processing mode matching, dynamic weighting and game adjustment; if the error gradually converges to the set tolerance interval, the system enters a stable maintenance mode and only performs light monitoring. The cycle continues until the numerical control machine tool detects the target battery shell processing task completion identifier, the system stops sampling and stores the final processing state data.

[0042] In summary, the embodiments of the present application have at least the following technical effects: With the processing progress of the numerical control machine tool on the target battery shell, the sensing array group is synchronously driven to monitor and determine the processing state data. Through defining the processing mode and the processing error type, a processing control module is generated based on the agent cluster construction of grinding force, thermal deformation, vibration and wear. If the processing state data has error out of limit, the processing control module is triggered to execute dynamic weighting based on processing mode matching, and the agent cluster parallel decision and balanced game adjustment, to generate a processing control strategy and execute processing control driving. The technical problem of fluctuation of processing precision and instability of processing quality caused by multi-factor coupling in the processing of the cylindrical new energy battery shell is solved. Through real-time monitoring and dynamic control of the processing process, the technical effects of improving the consistency of processing precision and the stability of processing quality are achieved.

[0043] In the embodiment two, based on the same inventive concept as the high-precision processing method for the cylindrical new energy battery shell in the foregoing embodiments, as shown in the following table, the present application provides a high-precision processing system for the cylindrical new energy battery shell, wherein the system comprises: Figure 2 A monitoring module 11: with the processing progress of the numerical control machine tool on the target battery shell, the sensing array group is synchronously driven to monitor and determine the processing state data; a module generation module 12: through defining the processing mode and the processing error type, a processing control module is generated based on the agent cluster construction of grinding force, thermal deformation, vibration and wear; a strategy execution module 13: if the processing state data has error out of limit, the processing control module is triggered to execute dynamic weighting based on processing mode matching, and the agent cluster parallel decision and balanced game adjustment, to generate a processing control strategy and execute processing control driving.

[0044] Further, the module generation module 12 is used to execute the following method: Classify the processing state and define the processing mode, wherein the processing mode at least includes normal stable mode, slight chatter mode, thermal dominant mode and wear end mode; define the processing error type, wherein the processing error type includes main processing error based on roundness and coaxiality, and secondary processing error based on surface roughness and residual stress; perform multi-dimensional agent construction from grinding force, thermal deformation, vibration and wear, through executing first-order training based on dynamic weighting of agents under processing mode, and second-order training based on balanced game of agents based on processing error, to constitute a processing control module.

[0045] Further, the module generation module 12 is used to execute the following method: ​A first state space and a first parameter space are defined with the greedy goal of minimizing force fluctuations, and a grinding force agent is deployed; a second state space and a second parameter space are defined with the greedy goal of minimizing temperature rise, and a thermal deformation agent is deployed; a third state space and a third parameter space are defined with the greedy goal of minimizing amplitude, and a vibration agent is deployed; a fourth state space and a fourth parameter space are defined with the greedy goal of minimizing tool wear rate, and a wear agent is deployed.

[0046] Further, the monitoring module 11 is configured to execute the following method: The end effector of the numerical control machine tool is equipped with a sensor array group, and the processing state data is synchronously collected along with the processing progress of the target battery shell; error overrun determination is performed on the processing state data, and if there is error overrun, a processing control instruction is generated.

[0047] Further, the strategy execution module 13 is configured to execute the following method: Upon receiving the processing control instruction, the mechanical drive parameters of the numerical control machine tool are read; and according to the processing state data and the mechanical drive parameters, the processing control module is triggered to make an industrial control adjustment decision.

[0048] Further, the strategy execution module 13 is configured to execute the following method: The target processing mode is determined by matching the processing state data; and according to the target processing mode and the processing state data, multi-dimensional dynamic weighting is performed to determine the agent weight distribution.

[0049] Further, the strategy execution module 13 is configured to execute the following method: The multi-dimensional directional control decision is executed in parallel to determine the control strategy set; the control strategy set is fine-tuned under the relative strategy of balanced game according to the agent weight distribution to determine the processing control strategy; wherein the control process of the grinding force dimension includes: initializing the first state space according to the processing state data, and the grinding force agent executes directional decision based on the greedy goal to determine the first control strategy.

[0050] Further, the strategy execution module 13 is configured to execute the following method: According to the minimum control unit of the numerical control machine tool, the processing control strategy is decoupled to determine a plurality of sub-control strategies; numerical control instructions based on the plurality of sub-control strategies are generated and issued to the corresponding control units for feedback control of the target battery shell processing progress.

[0051] Further, the strategy execution module 13 is configured to execute the following method: According to the tracking monitoring after the processing regulation based on the sensor array, the processing state data is updated with the processing time sequence; the error determination and adjustment decision based on the processing state data are continuously performed until the processing of the target battery shell is completed.

[0052] The above is only the preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as the preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any modification, equivalent change and modification of the above embodiments based on the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A high-precision machining method for cylindrical new energy battery casings, characterized in that, The method includes: As the CNC machine tool processes the target battery casing, the sensor array is synchronously driven to monitor and determine the processing status data. By defining processing modes and processing error types, a processing control module is generated by constructing an intelligent agent cluster based on grinding force, thermal deformation, vibration and wear. If the processing status data has an error exceeding the limit, the processing control module is triggered to perform dynamic weighting based on processing mode matching and parallel decision-making and equilibrium game adjustment of intelligent agent clusters, generate processing control strategies, and execute processing control drive.

2. The high-precision machining method for cylindrical new energy battery casings as described in claim 1, characterized in that, By defining processing modes and processing error types, a processing control module is generated by constructing an intelligent agent cluster based on grinding force, thermal deformation, vibration, and wear. This module includes: The processing conditions are classified and the processing modes are defined, wherein the processing modes include at least the normal stable mode, the slight chatter mode, the heat-dominated mode, and the end-of-wear mode. Define machining error types, wherein the machining error types include primary machining errors based on roundness and coaxiality, and secondary machining errors based on surface roughness and residual stress; A multi-dimensional intelligent agent is constructed based on grinding force, thermal deformation, vibration and wear. A processing control module is formed by performing first-order training based on dynamic weighting of the intelligent agent under the processing mode and second-order training based on the equilibrium game of the intelligent agent based on processing error.

3. The high-precision machining method for cylindrical new energy battery casings as described in claim 2, characterized in that, The construction of intelligent agents in multiple dimensions, including grinding force, thermal deformation, vibration, and wear, includes: With minimizing force fluctuations as the greedy objective, a first state space and a first parameter space are defined, and a grinding force agent is deployed. With minimizing temperature rise as the greedy objective, a second state space and a second parameter space are defined, and a thermal deformation agent is deployed. With minimizing amplitude as the greedy objective, a third state space and a third parameter space are defined, and a vibration agent is deployed. With the greedy objective of minimizing tool wear rate, a fourth state space and a fourth parameter space are defined, and a wear agent is deployed.

4. The high-precision machining method for cylindrical new energy battery casings as described in claim 3, characterized in that, The drive sensor array monitors and determines processing status data, including: A sensor array is installed on the end effector of a CNC machine tool to synchronously collect processing status data as the target battery casing is processed. The processing status data is used to determine if there is an error exceeding the limit. If there is an error exceeding the limit, a processing control command is generated.

5. The high-precision machining method for cylindrical new energy battery casings as described in claim 4, characterized in that, Triggering the processing control module includes: Upon receiving the processing control command, the mechanical drive parameters of the CNC machine tool are read; Based on the processing status data and the mechanical drive parameters, the processing control module is triggered to make industrial control adjustment decisions.

6. The high-precision machining method for cylindrical new energy battery casings as described in claim 5, characterized in that, Perform dynamic weighting based on processing pattern matching, including: The target processing mode is determined by matching the processing status data. Based on the target processing mode and the processing status data, multi-dimensional dynamic weighting is performed to determine the agent weight distribution.

7. The high-precision machining method for cylindrical new energy battery casings as described in claim 6, characterized in that, Performing parallel decision-making and equilibrium game regulation for intelligent agent clusters includes: Parallel execution of multi-dimensional targeted control decisions to determine a set of control strategies; Based on the agent weight distribution, the relative strategy fine-tuning under equilibrium game is performed on the control strategy set to determine the processing control strategy; The process of regulating the grinding force dimension includes: initializing the first state space based on the processing state data, and the grinding force agent performing directional decision-making based on a greedy objective to determine the first regulation strategy.

8. The high-precision machining method for cylindrical new energy battery casings as described in claim 1, characterized in that, Execution of processing control drivers, including: Based on the minimum control unit of the CNC machine tool, the machining control strategy is decoupled to determine multiple sub-control strategies; Numerical control instructions based on multiple sub-control strategies are generated and sent to the corresponding control units to provide feedback control over the processing of the target battery casing.

9. The high-precision machining method for cylindrical new energy battery casings as described in claim 1, characterized in that, The processing status data is updated according to the processing time sequence based on the tracking and monitoring after processing control by the sensor array. Error judgment and adjustment decisions are continuously made based on processing status data until the processing of the target battery casing is completed.

10. A high-precision machining system for cylindrical new energy battery casings, characterized in that, The system for implementing the high-precision machining method for cylindrical new energy battery casings according to any one of claims 1-9 includes: Monitoring module: Synchronously drives the sensor array to monitor and determine the processing status data as the CNC machine tool processes the target battery case; Module generation module: By defining the processing mode and processing error type, a processing control module is generated by constructing an intelligent agent cluster based on grinding force, thermal deformation, vibration and wear. Strategy execution module: If the processing status data has an error exceeding the limit, the processing control module is triggered to execute dynamic weighting based on processing mode matching and parallel decision-making and equilibrium game adjustment of intelligent agent cluster, generate processing control strategy, and execute processing control drive.

Images