Numerical control machine tool energy-saving control method based on multi-objective optimization

Abstract

The invention discloses a numerical control machine tool energy-saving control method based on multi-objective optimization, and the method comprises the following steps: collecting the operation states of motor driving, main shaft operation and cooling execution in the machining process of a numerical control machine tool, and generating energy consumption disturbance characterization data; determining a bias absorbable sub-domain of a processing action space, and encoding to generate an action bias absorption set; determining a bias absorption relation and generating a bias absorption parameter; executing multi-objective optimization solution, and generating a bias absorption vector; executing absorption micro-actions to generate an absorption action sequence; forming a disturbance update quantity, updating a bias absorption parameter, and generating a parameter update sequence; and continuously executing convergence updating on the parameter updating sequence, and generating an energy-saving control result after processing is completed. According to the method, the machining action is adjusted and controlled through multi-target bias absorption optimization, energy consumption disturbance absorption and machining precision are synchronously kept, and the method has the advantages of being energy-saving, efficient and stable in machining.

Description

Technical Field

[0001] This invention relates to the field of energy-saving control of CNC machine tools, and in particular to an energy-saving control method for CNC machine tools based on multi-objective optimization. Background Technology

[0002] During CNC machine tool processing, multiple execution links such as motor drive, spindle operation, and cooling are affected by factors such as load changes, material properties, heat accumulation, and servo following errors. This results in energy consumption disturbances such as power fluctuations, speed deviations, and temperature rises exhibiting dynamic coupling characteristics. Existing energy-saving control methods are mostly based on fixed machining trajectories and stable operating parameters, reducing overall energy consumption through power limiting, constant speed control, and spindle start-stop optimization. However, energy consumption is usually treated as an independent indicator, lacking linkage adjustment for the dynamic response of the actuator, and failing to fully absorb disturbances during the continuous execution of machining tasks.

[0003] Against this backdrop, existing technologies cannot integrate energy consumption disturbances with the machining motion space, nor can they absorb disturbances in multiple channels such as path, attitude, and injection vector. As a result, energy consumption disturbances can only be directly superimposed on the execution process in the form of energy. This causes energy consumption fluctuations to be amplified into dimensional errors, servo deviations, and decreased machining stability, creating a contradiction between the cost of maintaining machining accuracy and the cost of energy consumption stability. At the same time, existing energy-saving control methods lack a closed-loop update mechanism between disturbance, motion, and parameters, making it impossible to continuously eliminate the impact of disturbances during the machining task execution cycle. This makes it difficult to achieve simultaneous maintenance of energy saving and accuracy in high energy consumption fluctuation environments. Summary of the Invention

[0004] One objective of this invention is to propose an energy-saving control method for CNC machine tools based on multi-objective optimization. This invention optimizes and controls machining actions through multi-objective bias absorption, thereby achieving simultaneous absorption of energy consumption disturbances and maintenance of machining accuracy, and possessing the advantages of energy saving, high efficiency and stable machining.

[0005] An energy-saving control method for CNC machine tools based on multi-objective optimization according to an embodiment of the present invention includes the following steps:

[0006] The system collects the operating status of motor drive, spindle operation and cooling execution during the CNC machine tool machining process, and extracts power fluctuation rate, servo current deviation, thermal excitation disturbance and spindle response overshoot to generate energy consumption disturbance characterization data.

[0007] Based on energy consumption disturbance characterization data, the bias absorbable subdomain of the processing motion space is determined, and the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are encoded to generate the motion bias absorption set.

[0008] Based on the energy consumption disturbance characterization data and the action bias absorption set, the bias absorption relationship is determined and the bias absorption parameters are generated according to the quantitative index set of energy consumption disturbance amplitude, processing size deviation and servo following error.

[0009] Multi-objective optimization is performed on real-time energy consumption disturbance characterization data using bias absorption parameters to generate bias absorption vectors;

[0010] Write the bias absorption vector into the corresponding bias absorbable subdomain, and perform absorption micro-actions in the attitude fine-tuning, path segment extension and jet vector offset channels to generate an absorption action sequence.

[0011] The energy consumption disturbance characterization data is updated based on the absorption action sequence to form the disturbance update amount, and the bias absorption parameter is updated accordingly to form the parameter update sequence.

[0012] Based on the online detection data of processing quality, the parameter update sequence is continuously updated to generate energy-saving control results after processing is completed.

[0013] Optionally, the generation of the energy consumption disturbance characterization data specifically includes:

[0014] During the processing task execution cycle, the motor drive operation status, spindle operation status and cooling execution operation status are collected at preset collection time intervals and arranged in the order of sampling time to form the original sequence of operation status;

[0015] Based on the original sequence of operating status, the amplitude deviation of the motor drive power change is identified according to the sampling time index, forming a power fluctuation rate record sequence;

[0016] Under the sampling time index corresponding to the original sequence of the running state, the change in servo phase current is reduced point by point according to the change in the reference current of the processing task, and the reduction result is recorded to form a servo phase current deviation recording sequence.

[0017] Based on the original sequence of operating status, the deviation between the spindle housing temperature change and the ambient temperature change is extracted item by item according to the sampling time index, forming a thermal excitation disturbance recording sequence.

[0018] Based on the original sequence of operating status, the actual spindle speed change and the machining task speed setting value are identified point by point according to the sampling time index, forming a spindle response overshoot record sequence;

[0019] The power fluctuation rate recording sequence, servo phase current deviation recording sequence, thermal excitation disturbance recording sequence, and spindle response overshoot recording sequence are combined according to the sampling time index to generate a combined recording sequence for the operation status characterization.

[0020] The combined sequence of operational status characterization records is stored continuously according to the sampling time index within the processing task execution cycle to form a time-ordered combination of operational status characterization data, generating energy consumption disturbance characterization data.

[0021] Optionally, the generation of the action bias absorption set specifically includes:

[0022] Within the processing task execution cycle, the energy consumption disturbance characterization data are mapped to each processing action segment according to the sampling time index to form an energy consumption disturbance characterization partition sequence;

[0023] Based on the energy consumption disturbance characterization sequence, the displacement range, attitude range, path range and injection direction range corresponding to each processing action segment are extracted to form the offset absorption range record corresponding to the offset tolerance, attitude micro-angle margin, micro-path flexible segment and injection vector adjustable range.

[0024] The bias absorption range records are filtered by the processing action segment index, and the processing action segments with bias absorption range are retained to form a set of bias absorbable subdomains.

[0025] In the processing action segments corresponding to the bias absorbable subdomain set, the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are extracted according to the processing action segment index to form a bias absorption range segment sequence.

[0026] The bias absorption range segment sequence is encoded using a unified encoding method, maintaining a consistent correspondence between the processing action segment index and the sampling time index, to generate an encoded sequence containing multiple processing action segment bias absorption ranges.

[0027] The encoded sequences are continuously aggregated to generate an action bias absorption set.

[0028] Optionally, the generation of the bias absorption parameter specifically includes:

[0029] Within the processing task execution cycle, the energy consumption disturbance characterization data is divided according to the processing action segment index to form an energy consumption disturbance characterization segment sequence;

[0030] Arrange the action bias absorption set according to the corresponding processing action segment index to form the action bias absorption segment sequence;

[0031] Based on the segmented sequence of energy consumption disturbance characterization, the amplitude changes of energy consumption disturbance characterization data within each processing action segment are identified to form an energy consumption disturbance amplitude record;

[0032] Based on the machining dimension detection data within the machining task execution cycle, the machining dimension deviation under the machining action segment index is extracted to form a machining dimension deviation record;

[0033] Based on the servo follow error data within the processing task execution cycle, extract the servo follow error under the processing action segment index to form a servo follow error record;

[0034] The energy consumption disturbance amplitude records, machining dimension deviation records, and servo follow error records are combined according to the corresponding machining action segment index to form a set of quantitative indicators;

[0035] Based on the set of quantitative indicators and the segmented sequence of action bias absorption, the corresponding bias absorption relationship is determined under the processing action segment index, forming a segmented record of bias absorption relationship;

[0036] The bias absorption relationship is segmented and continuously aggregated according to the processing action segment index to form the bias absorption parameter.

[0037] Optionally, the generation of the bias absorption vector specifically includes:

[0038] During the processing task execution cycle, the bias absorption parameters and real-time energy consumption disturbance characterization data are arranged according to the processing action segment index to form the bias absorption solution input sequence;

[0039] Based on the input sequence of bias absorption solution, the disturbance amplitude of real-time energy consumption disturbance characterization data and the corresponding bias absorption parameters are identified in each processing action segment, forming a bias absorption solution index record;

[0040] Within each processing action segment, based on the bias absorption solution index record, multi-objective optimization solution is performed on the real-time energy consumption disturbance characterization data to determine the disturbance bias amount within the bias absorbable subdomain and form a disturbance bias amount record sequence.

[0041] The disturbance offset recording sequence is arranged according to the processing action segment index and sampling time index. The disturbance offset is encoded under each sampling time index. Each disturbance offset is represented in a unified format as the offset component record corresponding to the displacement offset component, attitude offset component, path offset component and injection offset component.

[0042] Based on the offset component record, the displacement offset component, attitude offset component, path offset component and injection offset component are combined in a preset order under each processing action segment and sampling time index to form an offset absorption vector segment.

[0043] Each bias absorption vector segment is continuously aggregated according to the processing action segment index and sampling time index to generate a bias absorption vector.

[0044] Optionally, the generation of the absorption action sequence specifically includes:

[0045] Write the bias absorption vector into the corresponding bias absorbable subdomain according to the processing action segment index to form the bias absorption writing sequence;

[0046] Based on the offset absorption writing sequence, within the offset absorbable subdomain corresponding to each processing action segment, the offset absorption vector is decomposed into displacement offset component, attitude offset component, path offset component and injection offset component to form an offset absorption component sequence.

[0047] Based on the offset absorption component sequence, the displacement offset component, attitude offset component, path offset component and injection offset component are written into the position adjustment channel, attitude fine-tuning channel, path segment extension channel and injection vector offset channel respectively to form a channel offset sequence.

[0048] Within the channels corresponding to each channel bias sequence, absorb micro-actions are performed according to the sampling time index to form a sequence of absorb micro-action segments.

[0049] Based on the sequence of absorbed micro-action segments, each absorbed micro-action segment is continuously merged under the index of processed action segments to form a sequence of absorbed micro-action segments.

[0050] The micro-motion segment sequence is assembled into an absorption motion fragment sequence according to the sampling time index, while maintaining the correspondence between the processing motion segment indices;

[0051] The absorption action sequence is generated by continuously accumulating the sequence of absorption action segments within the processing task execution cycle.

[0052] Optionally, the generation of the parameter update sequence specifically includes:

[0053] Based on the absorption action sequence and energy consumption disturbance characterization data, a correspondence between absorption actions and energy consumption disturbance characterization data is established under the processing action segment index, forming a corresponding sequence of absorption actions;

[0054] Based on the sequence of absorption actions, the changes in energy consumption disturbance characterization data caused by absorption actions are identified in each processing action segment, and disturbance change segments are formed by indexing the sampling time.

[0055] The perturbation change segments are continuously aggregated under the processing action segment index to form the perturbation update quantity;

[0056] Based on the perturbation update amount and the bias absorption parameter, establish the correspondence between the perturbation update amount and the bias absorption parameter under the processing action segment index to form the corresponding sequence of bias absorption parameters;

[0057] Based on the corresponding sequence of bias absorption parameters, the parameters are updated item by item according to the sampling time index in each processing action segment to form a bias absorption parameter update segment;

[0058] The bias absorption parameter update fragments are continuously aggregated under the processing action segment index to generate a parameter update sequence.

[0059] Optionally, the generation of the energy-saving control result specifically includes:

[0060] Based on the parameter update sequence and online processing quality detection data, a quality detection comparison sequence corresponding to the parameter update sequence is formed under the processing action segment index and sampling time index;

[0061] Based on the quality inspection comparison sequence, the changes in online processing quality inspection data are recorded under each sampling time index to form a processing quality change segment sequence;

[0062] The sequence of processing quality change segments is arranged continuously under the index of processing action segments to form a processing quality change sequence;

[0063] The parameter update sequence is updated item by item according to the processing quality change sequence, and the parameter convergence update segment is formed according to the sampling time index.

[0064] The parameter convergence update segments are continuously aggregated under the processing action segment index to form a convergence parameter update sequence. The convergence parameter update sequence replaces the parameter update sequence to enter the next convergence update cycle.

[0065] When the preset convergence criteria are met, the convergence parameter update sequence is finally summarized according to the processing action segment index to form a stable parameter update sequence.

[0066] Based on the stable parameter update sequence and online processing quality detection data, a processing control adjustment sequence is formed within the processing task execution cycle. At the end of the processing task, the processing control adjustment sequence is integrated to form the energy-saving control result of the completed processing.

[0067] The beneficial effects of this invention are:

[0068] This invention transforms the passive acceptance of energy consumption disturbances during CNC machine tool machining into active absorption by constructing energy consumption disturbance characterization data, motion bias absorption set, bias absorption parameters, and convergence iterative update mechanism. This enables the machining execution action to have dynamic adjustment capability in response to disturbances. Compared with existing technologies that rely solely on fixed machining paths, constant parameters, or a single energy consumption suppression strategy, this invention uses a multi-objective optimization approach to map multi-dimensional disturbance information such as power fluctuation rate, servo current deviation, thermal excitation disturbance, and spindle response overshoot to the machining motion space. By utilizing displacement bias, attitude micro-angle margin, flexible micro-path segment, and adjustable range of injection vector to form a bias absorption subdomain, the disturbance no longer directly acts on the spindle and servo system in the form of energy fluctuations. Instead, it is decomposed and absorbed through multiple channels such as attitude, path, and injection, blocking the transmission link of disturbance to machining error at the source.

[0069] This invention further employs a multi-objective optimization solution mechanism driven by bias absorption parameters, which uniformly quantifies and evaluates the amplitude of energy consumption disturbance, machining size deviation, and servo following error. This ensures that the disturbance absorption process not only pursues energy consumption stability but also guarantees that machining accuracy and execution response stability are not compromised. In addition, this invention performs gradual convergence updates of the bias absorption parameters based on the absorption action sequence, constructing an adaptive control closed loop that couples disturbance, action, and parameter. This allows energy consumption disturbances generated during machining to be identified, absorbed, and continuously weakened in real time, thereby enabling the energy-saving control strategy to dynamically evolve and converge to stability as the machining task progresses.

[0070] The beneficial effects of this invention are that, in the dynamic environment of CNC machining, it achieves the elimination of energy consumption disturbances at the machining action level through multi-channel absorption, so that the cost of maintaining machining accuracy and the cost of stable energy consumption no longer conflict with each other; through continuous iterative updates, it achieves full-cycle energy-saving regulation, so that the machining process under high energy consumption fluctuation conditions has stability, reliability and efficiency; overall, it significantly reduces the adverse effects of energy consumption fluctuations on product quality and equipment performance, has higher energy efficiency and better machining consistency, and is suitable for CNC machining application scenarios with high precision, high stability and high energy efficiency requirements. Attached Figure Description

[0071] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0072] Figure 1 This is a flowchart of a CNC machine tool energy-saving control method based on multi-objective optimization proposed in this invention;

[0073] Figure 2 This is a schematic diagram illustrating the relationship between the generation of bias absorption parameters and the solution of bias absorption vector in a multi-objective optimization-based energy-saving control method for CNC machine tools proposed in this invention.

[0074] Figure 3 This is a schematic diagram illustrating the generation of energy-saving control results for a multi-objective optimization-based energy-saving control method for CNC machine tools proposed in this invention. Detailed Implementation

[0075] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0076] refer to Figures 1-3 A method for energy-saving control of CNC machine tools based on multi-objective optimization includes the following steps:

[0077] The system collects the operating status of motor drive, spindle operation and cooling execution during the CNC machine tool machining process, and extracts power fluctuation rate, servo current deviation, thermal excitation disturbance and spindle response overshoot to generate energy consumption disturbance characterization data.

[0078] Based on energy consumption disturbance characterization data, the bias absorbable subdomain of the processing motion space is determined, and the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are encoded to generate the motion bias absorption set.

[0079] Based on the energy consumption disturbance characterization data and the action bias absorption set, the bias absorption relationship is determined and the bias absorption parameters are generated according to the quantitative index set of energy consumption disturbance amplitude, processing size deviation and servo following error.

[0080] Multi-objective optimization is performed on real-time energy consumption disturbance characterization data using bias absorption parameters to generate bias absorption vectors;

[0081] Write the bias absorption vector into the corresponding bias absorbable subdomain, and perform absorption micro-actions in the attitude fine-tuning, path segment extension and jet vector offset channels to generate an absorption action sequence.

[0082] The energy consumption disturbance characterization data is updated based on the absorption action sequence to form the disturbance update amount, and the bias absorption parameter is updated accordingly to form the parameter update sequence.

[0083] Based on the online detection data of processing quality, the parameter update sequence is continuously updated to generate energy-saving control results after processing is completed.

[0084] In this embodiment, the generation of the energy consumption disturbance characterization data specifically includes:

[0085] During the processing task execution cycle, the motor drive operation status, spindle operation status and cooling execution operation status are collected at preset collection time intervals and arranged in the order of sampling time to form the original sequence of operation status;

[0086] Based on the original sequence of operating status, the amplitude deviation of the motor drive power change is identified according to the sampling time index, forming a power fluctuation rate record sequence;

[0087] Under the sampling time index corresponding to the original sequence of the running state, the change in servo phase current is reduced point by point according to the change in the reference current of the processing task, and the reduction result is recorded to form a servo phase current deviation recording sequence.

[0088] Based on the original sequence of operating status, the deviation between the spindle housing temperature change and the ambient temperature change is extracted item by item according to the sampling time index, forming a thermal excitation disturbance recording sequence.

[0089] Based on the original sequence of operating status, the actual spindle speed change and the machining task speed setting value are identified point by point according to the sampling time index, forming a spindle response overshoot record sequence;

[0090] The power fluctuation rate recording sequence, servo phase current deviation recording sequence, thermal excitation disturbance recording sequence, and spindle response overshoot recording sequence are combined according to the sampling time index to generate a combined recording sequence for the operation status characterization.

[0091] The combined sequence of operational status characterization records is stored continuously according to the sampling time index within the processing task execution cycle to form a time-ordered combination of operational status characterization data, generating energy consumption disturbance characterization data.

[0092] In this embodiment, the generation of the action bias absorption set specifically includes:

[0093] Within the processing task execution cycle, the energy consumption disturbance characterization data are mapped to each processing action segment according to the sampling time index to form an energy consumption disturbance characterization partition sequence;

[0094] Based on the energy consumption disturbance characterization sequence, the displacement range, attitude range, path range and injection direction range corresponding to each processing action segment are extracted to form the offset absorption range record corresponding to the offset tolerance, attitude micro-angle margin, micro-path flexible segment and injection vector adjustable range.

[0095] The bias absorption range records are filtered by the processing action segment index, and the processing action segments with bias absorption range are retained to form a set of bias absorbable subdomains.

[0096] The formation of the set of biased absorbing subdomains specifically includes:

[0097] Within the processing task execution cycle, the bias absorption range records are arranged according to the corresponding processing action segment index to form a bias absorption range corresponding sequence; based on the bias absorption range corresponding sequence, processing action segments that do not have a bias absorption range are removed, and processing action segments that have a bias absorption range are retained to form a preliminary selection set of processing action segments with a bias absorption range; the preliminary selection set of processing action segments with a bias absorption range is continuously merged under adjacent sampling time indices to form an absorbable segment sequence; the start and end sampling time indices of each absorbable segment are identified in the absorbable segment sequence to form subdomain boundary marker records; the subdomain boundary marker records are continuously arranged according to the processing action segment index to form bias absorbable subdomain segments; the bias absorbable subdomain segments are continuously aggregated and merged within the processing task execution cycle to form a bias absorbable subdomain set;

[0098] In the processing action segments corresponding to the bias absorbable subdomain set, the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are extracted according to the processing action segment index to form a bias absorption range segment sequence.

[0099] The bias absorption range segment sequence is encoded using a unified encoding method, maintaining a consistent correspondence between the processing action segment index and the sampling time index, to generate an encoded sequence containing multiple processing action segment bias absorption ranges.

[0100] The generation of the encoded sequence specifically includes:

[0101] Under the processing action segment index and sampling time index corresponding to the offset absorption range segment sequence, each offset absorption range segment is divided into fields according to the sampling time index in a preset field order to form an offset absorption range field sequence. In the offset absorption range field sequence, the offset tolerance field, attitude micro-angle margin field, micro-path flexible segment field, and injection vector adjustable range field are concatenated according to the processing action segment index to form an offset absorption range segment field combination sequence. Under the processing action segment index corresponding to the offset absorption range segment field combination sequence, a unique code identifier is assigned to each field combination according to a unified coding marking method to form an offset absorption range segment coding record sequence. The offset absorption range segment coding record sequence is concatenated in the execution order under the sampling time index to form an offset absorption range coding segment sequence. The offset absorption range coding segment sequence is continuously aggregated under the processing action segment index to generate a coding sequence containing offset absorption ranges of multiple processing action segments.

[0102] The encoded sequences are continuously aggregated to generate an action bias absorption set;

[0103] The generation of the action bias absorption set specifically includes:

[0104] Within the processing task execution cycle, the encoded sequences are arranged sequentially according to the processing action segment index to form an encoded sequence index aligned record; the encoded sequence index aligned record is continuously connected under adjacent sampling time indexes to form an encoded continuous segment sequence; the processing action segment indices corresponding to the encoded continuous segment sequence are continuously merged to form an encoded aggregate segment record; the encoded aggregate segment record is connected as a whole according to the processing action segment index to form an action bias absorption segment; the action bias absorption segments are continuously aggregated and merged according to the sampling time index within the processing task execution cycle to generate an action bias absorption set.

[0105] In this embodiment, the generation of the bias absorption parameter specifically includes:

[0106] Within the processing task execution cycle, the energy consumption disturbance characterization data is divided according to the processing action segment index to form an energy consumption disturbance characterization segment sequence;

[0107] Arrange the action bias absorption set according to the corresponding processing action segment index to form the action bias absorption segment sequence;

[0108] Based on the segmented sequence of energy consumption disturbance characterization, the amplitude changes of energy consumption disturbance characterization data within each processing action segment are identified to form an energy consumption disturbance amplitude record;

[0109] Based on the machining dimension detection data within the machining task execution cycle, the machining dimension deviation under the machining action segment index is extracted to form a machining dimension deviation record;

[0110] Based on the servo follow error data within the processing task execution cycle, extract the servo follow error under the processing action segment index to form a servo follow error record;

[0111] The energy consumption disturbance amplitude records, machining dimension deviation records, and servo follow error records are combined according to the corresponding machining action segment index to form a set of quantitative indicators;

[0112] Based on the set of quantitative indicators and the segmented sequence of action bias absorption, the corresponding bias absorption relationship is determined under the processing action segment index, forming a segmented record of bias absorption relationship;

[0113] The generation of the bias absorption relationship segmented record specifically includes:

[0114] Within the processing task execution cycle, the set of quantitative indicators is arranged according to the processing action segment index and the action offset absorption segment sequence to form a quantitative indicator correspondence sequence; in the quantitative indicator correspondence sequence, the corresponding indexes of energy consumption disturbance amplitude records, processing size deviation records, and servo follow error records and offset absorption range segments are identified according to the sampling time index to form quantitative indicator and offset absorption segment correspondence records; the quantitative indicator and offset absorption segment correspondence records are continuously connected according to the processing action segment index to form offset absorption relationship segments; the offset absorption relationship segments are continuously aggregated and merged according to the sampling time index within the processing task execution cycle to form offset absorption relationship segment records;

[0115] The bias absorption relationship is segmented and recorded, and then continuously aggregated according to the processing action segment index to form the bias absorption parameter.

[0116] The generation of the bias absorption parameter specifically includes:

[0117] Within the processing task execution cycle, the bias absorption relationship is segmented and recorded, and then arranged sequentially according to the processing action segment index to form a bias absorption relationship index alignment sequence. The bias absorption relationship index alignment sequence is then continuously connected under adjacent sampling time indices to form a continuous bias absorption relationship segment sequence. The continuous bias absorption relationship segment sequence is then continuously merged according to the processing action segment index to form a bias absorption relationship aggregated segment record. The bias absorption relationship aggregated segment record is then encoded under the condition that the processing action segment index and the sampling time index correspond to maintain index consistency to form a bias absorption relationship mapping segment. The bias absorption relationship mapping segment is then continuously aggregated and merged according to the sampling time index within the processing task execution cycle to form the bias absorption parameter.

[0118] In this embodiment, the generation of the bias absorption vector specifically includes:

[0119] During the processing task execution cycle, the bias absorption parameters and real-time energy consumption disturbance characterization data are arranged according to the processing action segment index to form the bias absorption solution input sequence;

[0120] Based on the input sequence of bias absorption solution, the disturbance amplitude of real-time energy consumption disturbance characterization data and the corresponding bias absorption parameters are identified in each processing action segment, forming a bias absorption solution index record;

[0121] Within each processing action segment, based on the bias absorption solution index record, multi-objective optimization solution is performed on the real-time energy consumption disturbance characterization data to determine the disturbance bias amount within the bias absorbable subdomain and form a disturbance bias amount record sequence.

[0122] The generation of the perturbation bias recording sequence specifically includes:

[0123] Within the processing action segment corresponding to the offset absorption solution index record, real-time energy consumption disturbance characterization data and offset absorption parameters are continuously read in the order of sampling time index to form a solution input sequence. Based on the solution input sequence, multi-objective optimization is performed on the energy consumption disturbance amplitude within the offset absorbable subdomain to generate disturbance offset amount fragments. The disturbance offset amount fragments are filtered according to the processing action segment index to maintain the disturbance offset amount within the offset absorbable subdomain to form a disturbance offset amount filtering fragment. The disturbance offset amount filtering fragments are continuously arranged according to the sampling time index to form a disturbance offset amount arrangement sequence. The disturbance offset amount arrangement sequence is processed under the processing action segment index to perform fragment aggregation to generate a disturbance offset amount sequence. The disturbance offset amount sequence is continuously stored under the sampling time index to maintain the original order to form a disturbance offset amount record sequence.

[0124] The disturbance offset recording sequence is arranged according to the processing action segment index and sampling time index. The disturbance offset is encoded under each sampling time index. Each disturbance offset is represented in a unified format as the offset component record corresponding to the displacement offset component, attitude offset component, path offset component and injection offset component.

[0125] The generation of the bias component record specifically includes:

[0126] Based on the disturbance bias recording sequence, the disturbance bias reading segments are formed by sequentially reading each item according to the index correspondence between the processing action segment index and the sampling time index. These segments are then split according to the bias dimensions corresponding to displacement bias, attitude bias, path bias, and injection bias to generate bias component segments. Under each sampling time index, the displacement bias, attitude bias, path bias, and injection bias are encoded using a uniform format based on the bias component segments to form encoded component segments. These encoded component segments are then continuously merged according to the processing action segment index to generate a bias component arrangement sequence. Finally, the bias component arrangement sequence is stored in the same order as the disturbance bias recording sequence under the sampling time index to form the bias component record.

[0127] Based on the offset component record, the displacement offset component, attitude offset component, path offset component and injection offset component are combined in a preset order under each processing action segment and sampling time index to form an offset absorption vector segment.

[0128] Each bias absorption vector segment is continuously aggregated according to the processing action segment index and sampling time index to generate a bias absorption vector.

[0129] In this embodiment, the generation of the absorption action sequence specifically includes:

[0130] Write the bias absorption vector into the corresponding bias absorbable subdomain according to the processing action segment index to form the bias absorption writing sequence;

[0131] Based on the offset absorption writing sequence, within the offset absorbable subdomain corresponding to each processing action segment, the offset absorption vector is decomposed into displacement offset component, attitude offset component, path offset component and injection offset component to form an offset absorption component sequence.

[0132] Based on the offset absorption component sequence, the displacement offset component, attitude offset component, path offset component and injection offset component are written into the position adjustment channel, attitude fine-tuning channel, path segment extension channel and injection vector offset channel respectively to form a channel offset sequence.

[0133] The generation of the channel bias sequence specifically includes:

[0134] The offset absorption component sequence is arranged according to the corresponding processing action segment index and sampling time index to form an offset absorption component alignment sequence. Based on the offset absorption component alignment sequence, under each processing action segment, according to the recording order of displacement offset component, attitude offset component, path offset component, and injection offset component, the displacement offset component is written into the position adjustment channel to form a position offset recording segment, the attitude offset component is written into the attitude fine-tuning channel to form an attitude offset recording segment, the path offset component is written into the path segment extension channel to form a path offset recording segment, and the injection offset component is written into the injection vector offset channel to form an injection offset recording segment. Under each processing action segment index and sampling time index, the position offset recording segment, attitude offset recording segment, path offset recording segment, and injection offset recording segment are continuously spliced ​​in the order of arrangement to form a channel offset segment sequence. The channel offset segment sequence is continuously aggregated within the processing task execution cycle while keeping the processing action segment index unchanged to generate a channel offset sequence.

[0135] Within the channels corresponding to each channel bias sequence, absorb micro-actions are performed according to the sampling time index to form a sequence of absorb micro-action segments.

[0136] The generation of the absorption micro-motion fragment sequence specifically includes:

[0137] Based on the channel offset sequence, the position offset record segments in the position adjustment channel, the attitude offset record segments in the attitude fine-tuning channel, the path offset record segments in the path segment extension channel, and the injection offset record segments in the injection vector offset channel are continuously read according to the index correspondence under the processing action segment index and the sampling time index to form a channel offset reading sequence. The channel offset reading sequence is then sequentially executed to absorb micro-actions and write them according to the arrangement order of position offset record segments, attitude offset record segments, path offset record segments, and injection offset record segments under each sampling time index to form absorb micro-action write segments. Each absorb micro-action write segment is then continuously merged under the processing action segment index and the sampling time index, maintaining the sampling time index correspondence consistent with the channel offset sequence, to generate an absorb micro-action segment sequence.

[0138] Based on the sequence of absorbed micro-action segments, each absorbed micro-action segment is continuously merged under the index of processed action segments to form a sequence of absorbed micro-action segments.

[0139] The generation of the micro-motion segment sequence specifically includes:

[0140] The micro-motion fragment sequence is arranged according to the processing action segment index to form an aligned sequence of micro-motion fragments; based on the aligned sequence of micro-motion fragments, micro-motion fragments are continuously extracted according to the sampling time index under each processing action segment index to form a continuous micro-motion fragment; the continuous micro-motion fragments are merged segment by segment while keeping the processing action segment index unchanged to form a segment of micro-motion fragments; the segment of micro-motion fragments is stored in the order of processing action segment index within the processing task execution cycle to form a segmented sequence of micro-motions.

[0141] The micro-motion segment sequence is assembled into an absorption motion fragment sequence according to the sampling time index, while maintaining the correspondence between the processing motion segment indices;

[0142] The absorption action sequence is generated by continuously accumulating the sequence of absorption action segments within the processing task execution cycle.

[0143] In this embodiment, the generation of the parameter update sequence specifically includes:

[0144] Based on the absorption action sequence and energy consumption disturbance characterization data, a correspondence between absorption actions and energy consumption disturbance characterization data is established under the processing action segment index, forming a corresponding sequence of absorption actions;

[0145] The generation of the sequence corresponding to the absorption action specifically includes:

[0146] The absorption action sequence and energy consumption disturbance characterization data are arranged item by item according to the index correspondence under the processing action segment index and the sampling time index to form an index-aligned sequence of absorption actions and energy consumption disturbance characterization data. Based on the index-aligned sequence, absorption action sequence segments and energy consumption disturbance characterization data segments are continuously extracted under the sampling time index of each processing action segment to form a sequence of corresponding segments of absorption actions and energy consumption disturbance characterization data. The sequence of corresponding segments of absorption actions and energy consumption disturbance characterization data is continuously aggregated within the processing task execution cycle while maintaining the correspondence between the processing action segment index and the sampling time index to form a sequence of absorption actions.

[0147] Based on the sequence of absorption actions, the changes in energy consumption disturbance characterization data caused by absorption actions are identified in each processing action segment, and disturbance change segments are formed by indexing the sampling time.

[0148] The generation of the disturbance change segment specifically includes:

[0149] The absorption action corresponding sequence is continuously arranged according to the index correspondence under the processing action segment index and the sampling time index to form the absorption action corresponding alignment sequence; based on the absorption action corresponding alignment sequence, the absorption action record segment and the energy consumption disturbance characterization data record segment are extracted sequentially according to the sampling time index in each processing action segment to form the action disturbance corresponding segment sequence; the action disturbance corresponding segment sequence is subjected to item-by-item differential extraction according to the sequential arrangement of the absorption action record segment and the energy consumption disturbance characterization data record segment under each sampling time index to form the disturbance difference segment; the disturbance difference segment is continuously aggregated under the processing action segment index and the sampling time index while maintaining the index correspondence consistent with the action disturbance corresponding segment sequence to form the disturbance change segment;

[0150] The perturbation change segments are continuously aggregated under the processing action segment index to form the perturbation update quantity;

[0151] Based on the perturbation update amount and the bias absorption parameter, establish the correspondence between the perturbation update amount and the bias absorption parameter under the processing action segment index to form the corresponding sequence of bias absorption parameters;

[0152] The generation of the sequence corresponding to the bias absorption parameters specifically includes:

[0153] The perturbation update quantities are continuously arranged under the processing action segment index and the sampling time index to form a perturbation update quantity alignment sequence; the bias absorption parameters are continuously arranged under the processing action segment index and the sampling time index to form a bias absorption parameter alignment sequence; based on the perturbation update quantity alignment sequence and the bias absorption parameter alignment sequence, item-by-item alignment is performed under the processing action segment index and the sampling time index according to the index correspondence to form a perturbation parameter corresponding segment sequence; the perturbation parameter corresponding segment sequence is continuously aggregated under the processing action segment index and the sampling time index while maintaining the index correspondence to form a bias absorption parameter corresponding sequence;

[0154] Based on the corresponding sequence of bias absorption parameters, the parameters are updated item by item according to the sampling time index in each processing action segment to form a bias absorption parameter update segment;

[0155] The bias absorption parameter update fragments are continuously aggregated under the processing action segment index to generate a parameter update sequence.

[0156] In this embodiment, the generation of the energy-saving control result specifically includes:

[0157] Based on the parameter update sequence and online processing quality detection data, a quality detection comparison sequence corresponding to the parameter update sequence is formed under the processing action segment index and sampling time index;

[0158] Based on the quality inspection comparison sequence, the changes in online processing quality inspection data are recorded under each sampling time index to form a processing quality change segment sequence;

[0159] The sequence of processing quality change segments is arranged continuously under the index of processing action segments to form a processing quality change sequence;

[0160] The parameter update sequence is updated item by item according to the processing quality change sequence, and the parameter convergence update segment is formed according to the sampling time index.

[0161] The generation of the parameter convergence update segment specifically includes:

[0162] Within the processing task execution cycle, the processing quality change sequence and parameter update sequence are arranged according to the sampling time index to form a corresponding record of quality change and parameter update. Based on the corresponding record of quality change and parameter update, under each sampling time index, the record position corresponding to the parameter component in the parameter update sequence is compared item by item with the change record in the processing quality change sequence to form a parameter and quality corresponding segment under the corresponding sampling time index. The parameter and quality corresponding segment is continuously updated under each sampling time index, keeping the corresponding relationship of the processing action segment index of each parameter component in the parameter update sequence unchanged, forming a parameter item update segment sequence arranged by sampling time index. The parameter item update segment sequence is continuously aggregated under the processing action segment index to form a parameter convergence update segment.

[0163] The parameter convergence update segments are continuously aggregated under the processing action segment index to form a convergence parameter update sequence. The convergence parameter update sequence replaces the parameter update sequence to enter the next convergence update cycle.

[0164] When the preset convergence criteria are met, the convergence parameter update sequence is finally summarized according to the processing action segment index to form a stable parameter update sequence.

[0165] The generation of the stable parameter update sequence specifically includes:

[0166] When the preset convergence criteria are met, the convergence parameter update sequence is marked with a termination period according to the processing action segment index. The recording order of the parameter update results under each sampling time index remains unchanged. Under the termination period mark, the final merging operation corresponding to the processing action segment index is performed on the convergence parameter update sequence to form a convergence parameter merge segment sequence with dual correspondence between the processing action segment index and the sampling time index. Under each processing action segment index, the convergence parameter merge segment sequence is arranged continuously according to the sampling time index to generate a convergence parameter aggregation sequence containing the final parameter update results within all processing task execution cycles. In the convergence parameter aggregation sequence, the correspondence between the processing action segment indexes remains unchanged. The parameter update results under each sampling time index are extracted as final parameter update component segments according to the termination period mark. The final parameter update component segments are continuously aggregated under the processing action segment index to form a stable parameter update sequence.

[0167] Based on the stable parameter update sequence and online processing quality detection data, a processing control adjustment sequence is formed within the processing task execution cycle. At the end of the processing task, the processing control adjustment sequence is integrated to form the energy-saving control result of the completed processing.

[0168] The generation of the energy-saving control results specifically includes:

[0169] The stable parameter update sequence and the online processing quality detection data are arranged correspondingly under the processing action segment index and sampling time index to form a stable parameter quality correspondence sequence. Based on the stable parameter quality correspondence sequence, the change records corresponding to the online processing quality detection data are identified under each sampling time index to form processing quality change segments. The processing quality change segments are arranged continuously according to the processing action segment index to form a processing quality change sequence. The stable parameter update sequence is corrected item by item with the processing quality change sequence to form stable parameter correction segments under the sampling time index. The stable parameter correction segments are continuously aggregated under the processing action segment index to form a stable parameter correction sequence. The stable parameter correction sequence is arranged in the order of the sampling time index within the processing task execution cycle to form a processing control adjustment sequence. At the end of the processing task, the processing control adjustment sequence is integrated to form the energy-saving control result of the completed processing.

[0170] Example 1:

[0171] To verify the feasibility of this invention in practice, it was applied to a CNC machining center that has long been engaged in precision machinery manufacturing. The machining center is located in an industrial park and is responsible for processing multiple batches of high-precision metal parts of various types. The parts being processed have complex structures, including curved contours, deep cavity structures, and high-gloss surfaces. The process cycle is long, and the motor load, ambient temperature, and spindle speed often fluctuate dynamically during the processing, resulting in frequent energy consumption disturbances. This leads to the superposition and amplification of spindle response deviation and servo following error, making it difficult to simultaneously ensure machining accuracy and control energy-saving costs.

[0172] On a high-precision component production line in this machining center, this invention is applied within the control system, working in conjunction with the machine tool servo drive module, spindle control module, and cooling control module. Before the machining task starts, this invention collects energy consumption disturbance information in real time, including changes in motor power, servo drive current, spindle speed response deviation, and cooling temperature changes. This information is then written into the control system's cache as energy consumption disturbance characterization data. During the continuous execution of machining actions, this energy consumption disturbance characterization data is divided into machining action segments and correlated with the action offset absorption set. If power fluctuations, load impacts, temperature rise accumulation, or cutting disturbances occur during machining, this invention prevents energy consumption fluctuations from being transmitted to machining accuracy and machine tool dynamic response. Instead, it identifies flexibly adjustable parts in the machining action space, including finely adjustable tool posture, slightly extended toolpath segments, and slightly offset injection directions. Within its achievable range, it introduces offset absorption actions, transforming energy consumption disturbances into micro-actions that can be eliminated by the execution system, thus actively absorbing the disturbances.

[0173] During continuous processing, this invention continuously monitors the weakening effect of the absorption action on energy consumption disturbances. If the recorded energy consumption disturbance data shows a decay trend, it indicates that the bias absorption parameter direction is correct, and the system will progressively converge and update the parameter to make subsequent bias absorption actions more accurate. If the bias action introduces new execution errors or the processing quality inspection system reports that the dimensional deviation exceeds the allowable range, the system will perform reverse correction on the bias parameter through quality feedback, thereby forming a closed-loop adjustment mechanism for disturbance absorption and accuracy maintenance. This real-time adaptive update process runs through the entire processing cycle, making the final energy-saving control result highly convergent.

[0174] In a multi-axis machining operation, due to the high-strength metal being processed, significant cutting heat is generated during the cutting process, causing the spindle housing temperature to gradually rise. This affects the spindle speed control, leading to deviations from the set value. Upon detecting this disturbance, this invention identifies micro-path flexible segments within the current machining path, allowing for a minimal offset in the tool feed direction. This alleviates the stress on the spindle, thereby reducing power fluctuations. Simultaneously, the cooling execution section guides a slight deviation in the jet direction, concentrating the cooling medium more effectively on the heated area and suppressing further temperature increases. The entire adjustment process maintains machining accuracy requirements during execution, preventing potential degradation in machining quality.

[0175] In another production batch, this invention still proved effective in addressing current fluctuations caused by frequent acceleration and deceleration in the servo system. By automatically fine-tuning the tool posture through absorption motion, the impact of inertia changes was dispersed and absorbed, thus preventing trajectory deviation caused by the continuous accumulation of servo following errors. The online machining quality detection module within the system continuously monitors dimensional accuracy. When the accuracy deviation caused by the absorption motion approaches the allowable boundary of machining quality, this invention immediately updates the offset absorption parameters through a convergence mechanism, reducing offset motion and ensuring stable accuracy.

[0176] Table 1. Comprehensive comparison data between the method of this invention and traditional energy-saving control methods.

[0177] Indicator Item Traditional energy-saving control methods The multi-target bias absorption method of the present invention Improvement range Standard deviation of power fluctuation (kW) 0.82 0.46 ↓43.9% Spindle response overshoot average value (rpm) 53 28 ↓47.2% Servo tracking error mean (µm) 11.4 7.1 ↓37.7% Processing quality inspection deviation (µm) 7.2 4.3 ↓40.3% Tool usage time before tool change (h) 38.6 44.2 ↑14.5% Average energy consumption per batch (kWh) 126.4 115.2 ↓8.9% Number of processing cycle interruptions (times / batch) 3 1 ↓66.7% Finished product consistency stability score (0-10) 7.8 9.1 ↑16.7%

[0178] As can be seen from Table 1, the present invention shows significant advantages in three indicators that directly reflect the degree of influence of energy consumption disturbance: standard deviation of power fluctuation, mean overshoot of spindle response, and servo following error. The reduction ranges from 37% to 48%. The fundamental reason for this phenomenon is that the present invention does not passively limit energy consumption or simply implement constant parameter control like traditional methods. Instead, it actively transfers the influence of disturbance by using an action bias absorption mechanism, so that energy consumption disturbance is absorbed and dispersed at the processing action level. This ultimately weakens the disturbance transmission chain and prevents the power fluctuation inside the execution system from being amplified into accuracy degradation.

[0179] In terms of processing quality detection deviation, the present invention reduces it by about 40% compared with traditional methods. This is closely related to the closed-loop control mechanism formed by the continuous updating of the bias absorption parameters in the present invention. Traditional energy-saving methods are prone to lag or insufficient compensation when facing disturbance changes in the processing process. However, the present invention, through real-time convergence updates, makes the bias absorption action gradually close to the disturbance characteristics, reduces the cumulative error in the processing process, and keeps the processing size deviation stable and shows a long-term stabilization characteristic.

[0180] The increased tool life before tool replacement is not due to improved tool material or coating performance, but rather because this invention effectively reduces load impact and thermal excitation disturbance during machining, making tool wear more uniform and controllable. Therefore, under the same materials and machining strategies, this invention achieves a natural extension of tool life, further reducing production costs and minimizing downtime caused by tool replacement.

[0181] The energy consumption per batch decreased by about 9%. Although the decrease is not extremely large, this result has practical value from the perspective of simultaneously saving energy and maintaining processing accuracy. Traditional methods often sacrifice processing stability or reduce speed to reduce energy consumption. However, this invention relies on disturbance absorption and bias adjustment to reduce energy consumption without significantly reducing processing speed and stability, thus achieving both economy and sustainability.

[0182] The decrease in the number of processing cycle interruptions and the improvement in the consistency score of finished products further illustrate that the present invention makes the processing process more "immune" to disturbances, and provides a buffer margin for the machine tool execution process by making the motion bias absorption space, thereby reducing the fluctuations and uncertainties caused by temporary corrections, error compensation or abnormal processing shutdowns.

[0183] In summary, this invention does not rely on improving equipment hardware performance or adding additional energy consumption compensation equipment. It achieves substantial results in stabilizing accuracy, extending tool life, suppressing energy consumption disturbances, and reducing machining anomalies simply by optimizing control strategies and motion bias absorption mechanisms. This demonstrates that the invention is suitable for energy saving and stability improvement in long-cycle, multi-batch, and high-precision machining tasks.

[0184] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A CNC machine tool energy-saving control method based on multi-objective optimization, characterized in that, Includes the following steps: The system collects the operating status of motor drive, spindle operation and cooling execution during the CNC machine tool machining process, and extracts power fluctuation rate, servo current deviation, thermal excitation disturbance and spindle response overshoot to generate energy consumption disturbance characterization data. Based on energy consumption disturbance characterization data, the bias absorbable subdomain of the processing motion space is determined, and the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are encoded to generate the motion bias absorption set. Based on the energy consumption disturbance characterization data and the action bias absorption set, the bias absorption relationship is determined and the bias absorption parameters are generated according to the quantitative index set of energy consumption disturbance amplitude, processing size deviation and servo following error. Multi-objective optimization is performed on real-time energy consumption disturbance characterization data using bias absorption parameters to generate bias absorption vectors; Write the bias absorption vector into the corresponding bias absorbable subdomain, and perform absorption micro-actions in the attitude fine-tuning, path segment extension and jet vector offset channels to generate an absorption action sequence. The energy consumption disturbance characterization data is updated based on the absorption action sequence to form the disturbance update amount, and the bias absorption parameter is updated accordingly to form the parameter update sequence. Based on the online detection data of processing quality, the parameter update sequence is continuously updated to generate energy-saving control results after processing is completed.

2. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the energy consumption disturbance characterization data specifically includes: During the processing task execution cycle, the motor drive operation status, spindle operation status and cooling execution operation status are collected at preset collection time intervals and arranged in the order of sampling time to form the original sequence of operation status; Based on the original sequence of operating status, the amplitude deviation of the motor drive power change is identified according to the sampling time index, forming a power fluctuation rate record sequence; Under the sampling time index corresponding to the original sequence of the running state, the change in servo phase current is reduced point by point according to the change in the reference current of the processing task, and the reduction result is recorded to form a servo phase current deviation recording sequence. Based on the original sequence of operating status, the deviation between the spindle housing temperature change and the ambient temperature change is extracted item by item according to the sampling time index, forming a thermal excitation disturbance recording sequence. Based on the original sequence of operating status, the actual spindle speed change and the machining task speed setting value are identified point by point according to the sampling time index, forming a spindle response overshoot record sequence; The power fluctuation rate recording sequence, servo phase current deviation recording sequence, thermal excitation disturbance recording sequence, and spindle response overshoot recording sequence are combined according to the sampling time index to generate a combined recording sequence for the operation status characterization. The combined sequence of operational status characterization records is stored continuously according to the sampling time index within the processing task execution cycle to form a time-ordered combination of operational status characterization data, generating energy consumption disturbance characterization data.

3. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the action bias absorption set specifically includes: Within the processing task execution cycle, the energy consumption disturbance characterization data are mapped to each processing action segment according to the sampling time index to form an energy consumption disturbance characterization partition sequence; Based on the energy consumption disturbance characterization sequence, the displacement range, attitude range, path range and injection direction range corresponding to each processing action segment are extracted to form the offset absorption range record corresponding to the offset tolerance, attitude micro-angle margin, micro-path flexible segment and injection vector adjustable range. The bias absorption range records are filtered by the processing action segment index, and the processing action segments with bias absorption range are retained to form a set of bias absorbable subdomains. In the processing action segments corresponding to the bias absorbable subdomain set, the bias tolerance, attitude micro-angle margin, micro-path flexible segment and jet vector adjustable range are extracted according to the processing action segment index to form a bias absorption range segment sequence. The bias absorption range segment sequence is encoded using a unified encoding method, maintaining a consistent correspondence between the processing action segment index and the sampling time index, to generate an encoded sequence containing multiple processing action segment bias absorption ranges. The encoded sequences are continuously aggregated to generate an action bias absorption set.

4. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the bias absorption parameter specifically includes: Within the processing task execution cycle, the energy consumption disturbance characterization data is divided according to the processing action segment index to form an energy consumption disturbance characterization segment sequence; Arrange the action bias absorption set according to the corresponding processing action segment index to form the action bias absorption segment sequence; Based on the segmented sequence of energy consumption disturbance characterization, the amplitude changes of energy consumption disturbance characterization data within each processing action segment are identified to form an energy consumption disturbance amplitude record; Based on the machining dimension detection data within the machining task execution cycle, the machining dimension deviation under the machining action segment index is extracted to form a machining dimension deviation record; Based on the servo follow error data within the processing task execution cycle, extract the servo follow error under the processing action segment index to form a servo follow error record; The energy consumption disturbance amplitude records, machining dimension deviation records, and servo follow error records are combined according to the corresponding machining action segment index to form a set of quantitative indicators; Based on the set of quantitative indicators and the segmented sequence of action bias absorption, the corresponding bias absorption relationship is determined under the processing action segment index, forming a segmented record of bias absorption relationship; The bias absorption relationship is segmented and continuously aggregated according to the processing action segment index to form the bias absorption parameter.

5. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the bias absorption vector specifically includes: During the processing task execution cycle, the bias absorption parameters and real-time energy consumption disturbance characterization data are arranged according to the processing action segment index to form the bias absorption solution input sequence; Based on the input sequence of bias absorption solution, the disturbance amplitude of real-time energy consumption disturbance characterization data and the corresponding bias absorption parameters are identified in each processing action segment, forming a bias absorption solution index record; Within each processing action segment, based on the bias absorption solution index record, multi-objective optimization solution is performed on the real-time energy consumption disturbance characterization data to determine the disturbance bias amount within the bias absorbable subdomain and form a disturbance bias amount record sequence. The disturbance offset recording sequence is arranged according to the processing action segment index and sampling time index. The disturbance offset is encoded under each sampling time index. Each disturbance offset is represented in a unified format as the offset component record corresponding to the displacement offset component, attitude offset component, path offset component and injection offset component. Based on the offset component record, the displacement offset component, attitude offset component, path offset component and injection offset component are combined in a preset order under each processing action segment and sampling time index to form an offset absorption vector segment. Each bias absorption vector segment is continuously aggregated according to the processing action segment index and sampling time index to generate a bias absorption vector.

6. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the absorption action sequence specifically includes: Write the bias absorption vector into the corresponding bias absorbable subdomain according to the processing action segment index to form the bias absorption writing sequence; Based on the offset absorption writing sequence, within the offset absorbable subdomain corresponding to each processing action segment, the offset absorption vector is decomposed into displacement offset component, attitude offset component, path offset component and injection offset component to form an offset absorption component sequence. Based on the offset absorption component sequence, the displacement offset component, attitude offset component, path offset component and injection offset component are written into the position adjustment channel, attitude fine-tuning channel, path segment extension channel and injection vector offset channel respectively to form a channel offset sequence. Within the channels corresponding to each channel bias sequence, absorb micro-actions are performed according to the sampling time index to form a sequence of absorb micro-action segments. Based on the sequence of absorbed micro-action segments, each absorbed micro-action segment is continuously merged under the index of processed action segments to form a sequence of absorbed micro-action segments. The micro-motion segment sequence is assembled into an absorption motion fragment sequence according to the sampling time index, while maintaining the correspondence between the processing motion segment indices; The absorption action sequence is generated by continuously accumulating the sequence of absorption action segments within the processing task execution cycle.

7. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the parameter update sequence specifically includes: Based on the absorption action sequence and energy consumption disturbance characterization data, a correspondence between absorption actions and energy consumption disturbance characterization data is established under the processing action segment index, forming a corresponding sequence of absorption actions; Based on the sequence of absorption actions, the changes in energy consumption disturbance characterization data caused by absorption actions are identified in each processing action segment, and disturbance change segments are formed by indexing the sampling time. The perturbation change segments are continuously aggregated under the processing action segment index to form the perturbation update quantity; Based on the perturbation update amount and the bias absorption parameter, establish the correspondence between the perturbation update amount and the bias absorption parameter under the processing action segment index to form the corresponding sequence of bias absorption parameters; Based on the corresponding sequence of bias absorption parameters, the parameters are updated item by item according to the sampling time index in each processing action segment to form a bias absorption parameter update segment; The bias absorption parameter update fragments are continuously aggregated under the processing action segment index to generate a parameter update sequence.

8. The energy-saving control method for CNC machine tools based on multi-objective optimization according to claim 1, characterized in that, The generation of the energy-saving control results specifically includes: Based on the parameter update sequence and online processing quality detection data, a quality detection comparison sequence corresponding to the parameter update sequence is formed under the processing action segment index and sampling time index; Based on the quality inspection comparison sequence, the changes in online processing quality inspection data are recorded under each sampling time index to form a processing quality change segment sequence; The sequence of processing quality change segments is arranged continuously under the index of processing action segments to form a processing quality change sequence; The parameter update sequence is updated item by item according to the processing quality change sequence, and the parameter convergence update segment is formed according to the sampling time index. The parameter convergence update segments are continuously aggregated under the processing action segment index to form a convergence parameter update sequence. The convergence parameter update sequence replaces the parameter update sequence to enter the next convergence update cycle. When the preset convergence criteria are met, the convergence parameter update sequence is finally summarized according to the processing action segment index to form a stable parameter update sequence. Based on the stable parameter update sequence and online processing quality detection data, a processing control adjustment sequence is formed within the processing task execution cycle. At the end of the processing task, the processing control adjustment sequence is integrated to form the energy-saving control result of the completed processing.

Images