CN-122026920-A - Process time sequence data self-adaptive multi-mode compression method and system in material adding process

CN122026920ACN 122026920 ACN122026920 ACN 122026920ACN-122026920-A

Abstract

The invention relates to the technical field of additive manufacturing data compression, and discloses a process time sequence data self-adaptive multi-mode compression method and a system in an additive process, wherein the method comprises the steps of obtaining original process time sequence data; extracting window level feature vectors, generating corresponding mode identification information, generating a target compression strategy, executing dynamic switching control of the compression strategy, and carrying out segmented compression processing on process time sequence data. Compared with the scheme adopting a fixed compression strategy or a single time sequence compression mode in the prior art, the method has the technical problem that the high compression ratio and the compression stability are difficult to be simultaneously considered under the condition that the data change modes of different process stages are frequently switched in the additive manufacturing process. According to the invention, by introducing a window-level feature-driven multi-mode identification and segmented compression switching strategy, the self-adaptive segmented compression aiming at different process data change modes is realized, and the overall compression ratio and compression stability of process time sequence data are improved.

Inventors

Hou Huyang
LIU XIN

Assignees

南京衍构科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (10)

1. A process time sequence data self-adaptive multi-mode compression method in an additive process, which is characterized by comprising the following steps: Step S10, acquiring original process time sequence data in the additive manufacturing process, executing a feature extraction task by adopting a sectional feature construction mechanism based on a sliding time window based on the original process time sequence data, and outputting a window level feature vector ; Step S20 based on the window level feature vector Executing a process data change mode identification task and outputting mode identification information M; step S30, generating a target compression strategy C based on the mode identification information M and a preset mode compression strategy mapping table; Step S40, executing a compression strategy dynamic switching task by adopting a state driving control mechanism based on a mode switching identifier according to the mode identification information M and the target compression strategy C, and outputting compression control state information S; and S50, finally, executing the adaptive multi-mode compression task in a segmented compression mode according to the compression control state information S.
2. The adaptive multi-mode compression method of process time series data in an additive manufacturing process as set forth in claim 1, wherein in step S10, original process time series data in the additive manufacturing process is obtained, a feature extraction task is performed by adopting a segment feature construction mechanism based on a sliding time window based on the original process time series data, and a window level feature vector is output Specifically comprises the following steps: Step S101, acquiring original process time sequence data in the additive manufacturing process, wherein the original process time sequence data comprises current parameters, voltage parameters and wire feeding speed parameters; Step S102, segmenting original process time sequence data according to sliding time windows with preset lengths, outputting sliding time window data subsequences k, and extracting window-level feature vectors for each sliding time window data subsequence k Window-level feature vector Comprising a process time sequence data average value Variance of process time sequence data First order differential mean value Linear fit residual energy And periodic correlation coefficient ; Wherein, the first order difference means For describing the overall trend of technological parameters over time, linear fitting residual energy For measuring the deviation of the current data, periodic correlation coefficient For characterizing whether a stable repeated periodic variation characteristic exists in a current time window of a process parameter.
3. The process timing data adaptive multi-mode compression method of claim 2, wherein in step S20, the process timing data adaptive multi-mode compression method is based on window-level feature vectors Executing a process data change mode identification task and outputting mode identification information M, wherein the process data change mode identification task specifically comprises the following steps of: Window level feature vector Inputting the data into a preset mode judgment logic to carry out classification judgment: Process timing data variance in sliding time window data subsequence k Less than a preset variance stability judgment threshold and process time sequence data mean value When the preset steady-state mean constraint condition is met, judging that the sliding time window data subsequence k belongs to a stable mode ; When linear fitting residual energy When the energy judgment threshold value of the linear fitting residual error is smaller than a preset threshold value, judging that the sliding time window data subsequence k belongs to a linear change mode ; When the periodic correlation coefficient When the data sub-sequence k is larger than a preset period correlation judgment threshold value, judging that the sliding time window data sub-sequence k belongs to a period change mode ; When an abnormal change point exceeding a preset mutation threshold exists in the sliding time window data subsequence k, judging that the sliding time window data subsequence k belongs to a mutation mode ; Judging that the sliding time window data subsequence k belongs to a random fluctuation mode ; Finally based on stationary mode Linear change pattern Periodic variation pattern Mutation pattern Random wave pattern The determination result of (2) outputs the mode identification information M.
4. The process time series data adaptive multi-mode compression method in an additive process according to claim 1, wherein in step S30, a target compression policy C is generated based on mode identification information M in combination with a preset mode compression policy mapping table, and specifically includes: Acquiring pattern identification information M including stationary pattern Linear change pattern Periodic variation pattern Mutation pattern Random wave pattern ; For stationary mode A run-length code or an approximate run-length code is selected as a target compression strategy C; For linear variation mode Selecting a compression strategy combining linear fitting parameter storage and residual error coding as a target compression strategy C; For periodic variation patterns Selecting a compression strategy which retains a main frequency component after frequency domain transformation as a target compression strategy C; For mutation patterns Selecting a compression strategy of high-precision coding of the mutation point position mark and the original value as a target compression strategy C; For random fluctuation mode And selecting a compression strategy combining differential coding and entropy coding as a target compression strategy C.
5. The adaptive multi-mode compression method of process time sequence data in an additive process according to claim 1, wherein in step S40, a state driving control mechanism based on a mode switching identifier is adopted to execute a dynamic switching task of a compression strategy according to mode identification information M and a target compression strategy C, and the step of outputting compression control state information S specifically includes: step S401, obtaining mode identification information of the current time window t And a previous time window Pattern identification information of (2) ; Step S402, based on the mode identification information And pattern identification information When the mode characteristic difference value exceeds a preset mode characteristic difference threshold value, judging that the data segment corresponding to the current sliding window where the current time window t is positioned is subjected to mode conversion; step S403, finally, the compression control state information S is output based on the mode switching flag.
6. The adaptive multi-mode compression method of process time series data in an additive process according to claim 1, wherein in step S50, the segmented compression mode is to perform segmented compression on original process time series data in real time by adopting different compression strategies based on compression control state information S.
7. The method of adaptive multi-mode compression of process timing data in an additive process of claim 1, wherein before the step of performing the adaptive multi-mode compression task in a segmented compression manner based on the compression control state information S, the method further comprises identifying a set of hybrid modes using a wavelet transform or a piecewise linear fitting method based on the compression control state information S, and assigning a compression priority to each segmented mode in the set of hybrid modes in step S50.
8. A process time series data self-adaptive multi-mode compression system in an additive process, which is applied to the process time series data self-adaptive multi-mode compression method in the additive process according to any one of claims 1 to 7, and is characterized in that the process time series data self-adaptive multi-mode compression system in the additive process comprises: The process time sequence feature construction module is used for acquiring original process time sequence data in the additive manufacturing process, executing a feature extraction task based on the original process time sequence data by adopting a sectional feature construction mechanism based on a sliding time window, and outputting window level feature vectors ; Process data change pattern recognition module for window-level feature vector based Executing a process data change mode identification task and outputting mode identification information M; the mode driving compression strategy generation module is used for generating a target compression strategy C based on the mode identification information M and combining a preset mode compression strategy mapping table; the compression strategy state switching control module is used for executing a compression strategy dynamic switching task by adopting a state driving control mechanism based on a mode switching identifier according to the mode identification information M and the target compression strategy C and outputting compression control state information S; And the self-adaptive multi-mode segmented compression module is used for executing the self-adaptive multi-mode compression task in a segmented compression mode according to the compression control state information S.
9. Process timing data adaptive multi-mode compression device in an additive process, comprising a memory, a processor and a process timing data adaptive multi-mode compression program in an additive process stored on the memory and executable on the processor, wherein the process timing data adaptive multi-mode compression program in an additive process is executed by the processor to implement the process timing data adaptive multi-mode compression method in an additive process according to any one of claims 1 to 7.
10. A computer program product comprising a process timing data adaptive multi-mode compression program in an additive process that when executed by a processor implements the process timing data adaptive multi-mode compression method in an additive process of any one of claims 1 to 7.

Description

Process time sequence data self-adaptive multi-mode compression method and system in material adding process Technical Field The invention relates to the technical field of additive manufacturing data compression, in particular to a process time sequence data self-adaptive multi-mode compression method and system in an additive process. Background At present, in the additive manufacturing process, in order to realize quality tracing, process optimization analysis and abnormal working condition tracing of the process, various process parameters such as current, voltage, wire feeding speed and the like are required to be continuously collected, and large-scale process time sequence data are formed for long-term storage. However, the existing technical data compression and storage schemes mostly adopt a fixed compression strategy or a general time sequence data compression method, and the characteristics of frequent process stage change and various data change modes in the additive manufacturing process are not fully considered, so that certain defects exist. For example, in the actual additive manufacturing process, there are significant differences in the variation characteristics of the process parameters corresponding to different process stages (such as a start stage, a stable forming stage and an end stage), and part of the stages have gentle data variation, and part of the stages may exhibit linear variation, periodic fluctuation or abrupt change characteristics. In the prior art, a single compression strategy is generally adopted to process all process time sequence data, and targeted optimization is difficult to carry out according to different data change modes, so that the compression efficiency is low in part of process stages, the overall compression effect is not ideal, and the data storage cost is difficult to fully reduce. In addition, the prior art often lacks an effective recognition and control mechanism for switching the process data change modes in adjacent time periods in the compression process, and when the process parameter change modes are switched, the original compression strategy is still used, so that the compression performance fluctuation is easily caused, and even the stability of the subsequent data analysis is influenced. Therefore, the prior art cannot fully meet the requirements of high compression ratio and stable compression on process time sequence data in the additive manufacturing application scene of frequent switching at the process stage and complex data change mode. Therefore, there is a need for a method that can adaptively adjust the compression mode according to the process time series data variation characteristics without depending on a fixed compression strategy, so as to improve the overall compression efficiency and long-term storage applicability of the additive manufacturing process time series data. Disclosure of Invention Aiming at the technical defects, the invention aims to provide a process time sequence data self-adaptive multi-mode compression method in an additive manufacturing process, and aims to solve the technical problems that a scheme of adopting a fixed compression strategy or a single time sequence compression mode in the prior art is difficult to simultaneously consider high compression ratio and compression stability under the condition that data change modes in different process stages are frequently switched in the additive manufacturing process. In order to solve the technical problems, the invention adopts the following technical scheme that the invention provides a process time sequence data self-adaptive multi-mode compression method in the material adding process. The process time sequence data self-adaptive multi-mode compression method in the material adding process comprises the following steps: Step S10, acquiring original process time sequence data in the additive manufacturing process, executing a feature extraction task by adopting a sectional feature construction mechanism based on a sliding time window based on the original process time sequence data, and outputting a window level feature vector ; Step S20 based on the window level feature vectorExecuting a process data change mode identification task and outputting mode identification information M; step S30, generating a target compression strategy C based on the mode identification information M and a preset mode compression strategy mapping table; Step S40, executing a compression strategy dynamic switching task by adopting a state driving control mechanism based on a mode switching identifier according to the mode identification information M and the target compression strategy C, and outputting compression control state information S; and S50, finally, executing the adaptive multi-mode compression task in a segmented compression mode according to the compression control state information S. Preferably, in step S10, original process time sequence data in the additive