CN-121834715-B - MICP strength prediction method and system based on machine learning and multi-feature fusion

CN121834715BCN 121834715 BCN121834715 BCN 121834715BCN-121834715-B

Abstract

The invention discloses a MICP strength prediction method and a MICP strength prediction system based on machine learning and multi-feature fusion, and relates to the field of solidification data analysis, wherein an accurate quality prediction mapping model and an anomaly compensation parameter mapping model are constructed based on historical data, so that the system has the capability of predicting solidification quality from process parameters and intelligently recommending an optimal treatment scheme from an anomaly state; by setting a dual-threshold quality control mechanism and a real-time abnormality monitoring and early warning system, various abnormal conditions in the curing process can be actively identified and responded in time, the complete closed loop processing from parameter deviation detection to automatic compensation measure execution to hardware standby switching is effectively ensured, the stability of the curing process and the consistency of engineering quality are effectively ensured, the requirement of manual intervention and the misjudgment risk are greatly reduced, a powerful technical support is provided for the standardized, normalized and industrialized development of the sand curing technology, and the engineering efficiency and economic benefit are remarkably improved.

Inventors

PENG JIE
Si Shaomin
JIANG ZHAO
WEI RENJIE
Shang Zhiyang
Gu Xuanming
SHEN SHICHENG
FENG ZEWEI
Si Jixu

Assignees

河海大学

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (8)

1. The MICP strength prediction method based on machine learning and multi-feature fusion is characterized by comprising the following steps of: s1, setting an average unconfined compressive strength and a calcium carbonate distribution difference coefficient quality index, and setting a plurality of influence factor parameters to acquire historical sand column solidification data; s2, constructing a final sand column solidification quality index mapping model based on the data acquired in the S1; s3, identifying a plurality of easy-to-abnormality parameters including bacterial liquid activity, grouting pressure and environmental temperature, setting corresponding countermeasures and parameters thereof, and acquiring historical data based on the countermeasures and parameters thereof to construct a final sand column solidification abnormality measure parameter mapping model; the method specifically comprises the following steps: S31, setting parameter types which are easy to be abnormal in the sand column solidification process according to operation steps of carrying out sand column solidification for a plurality of times historically to obtain an easy-abnormal solidification parameter type set, wherein the easy-abnormal solidification parameter type set comprises bacterial liquid activity, grouting pipeline pressure and ambient temperature, setting corresponding countermeasures and measure parameter types of each easy-abnormal solidification parameter according to the easy-abnormal solidification parameter type set to obtain a solidification parameter abnormal countermeasure set and a solidification parameter abnormal measure parameter set, wherein the solidification parameter abnormal countermeasure set comprises bacterial liquid supplementing, a back flushing program executing and a cooling device starting, and the solidification parameter abnormal measure parameter set comprises bacterial liquid supplementing flow rate, bacterial liquid supplementing time, bacterial liquid supplementing total quantity, back flushing direction, ‌ flushing medium type, ‌ flushing pressure, single ‌ flushing time, ‌ flushing cycle number, ‌ flushing flow, cooling power, cooling target temperature and cooling duration; S32, acquiring the numerical values of various easy-to-abnormal curing parameters before and after abnormality and the parameter data of corresponding countermeasures in the operation steps of carrying out sand column curing for a plurality of times in the history in S31 according to the easy-to-abnormal curing parameter type set, the curing parameter abnormality countermeasures set and the curing parameter abnormality measure parameter set, and obtaining a curing parameter data set before abnormality, a curing parameter data set after abnormality and a curing abnormality measure parameter data set; s33, constructing a mapping model which is input into curing parameter data before and after abnormality and output into abnormal measure parameter data according to the curing parameter data set before and after abnormality, the curing parameter data set after abnormality and the abnormal measure parameter data set after abnormality, and obtaining a final sand column curing abnormal measure parameter mapping model; s4, acquiring current sand column solidification parameter data in real time and inputting the current sand column solidification parameter data into a mapping model in the S2 for mapping; S5, setting a compressive strength and distribution difference coefficient double threshold, and repeatedly adjusting the curing parameters and re-mapping the curing parameters until the result meets the requirement if the mapping result in S4 meets the double threshold; S6, inputting the real-time parameters and the preset parameters which are acquired in the S5 and easy to be cured abnormally into the mapping model in the S3 for mapping, performing the compensation operation according to the mapping result, and then performing feedback verification on the recovery condition of the easy-to-be-abnormal parameters, if the recovery condition is not expected, starting the hardware standby switching, otherwise, finishing the compensation.
2. The MICP intensity prediction method based on machine learning and multi-feature fusion according to claim 1, wherein S1 comprises the steps of: s11, setting a plurality of index types for measuring the quality of the sand column after being solidified by adopting an MICP mode to obtain a sand column solidification quality index type set, wherein the sand column solidification quality index type set comprises the average unconfined compressive strength of the sand column after solidification and the calcium carbonate distribution difference coefficient; Setting a plurality of parameter types which have influence on the quality after curing in the process of curing the sand column by adopting an MICP mode to obtain a sand column curing parameter type set, wherein the sand column curing parameter type set comprises fungus liquid characteristics, cementing liquid characteristics, environmental parameters, grouting modes, grouting interval time, grouting cycle times, soil quality and geometric parameters before curing the sand column; And S12, acquiring sand column curing quality index data and sand column curing parameter data corresponding to the MICP sand column curing operation process for a plurality of times in history according to the sand column curing quality index type set and the sand column curing parameter type set, and obtaining a history sand column curing quality index data set and a history sand column curing parameter data set.
3. The MICP intensity prediction method based on machine learning and multi-feature fusion according to claim 2, wherein S2 comprises the steps of: S21, constructing a mapping model which is input into various sand column curing parameter data and output into various sand column curing quality index data according to the historical sand column curing quality index data set and the historical sand column curing parameter data set, and obtaining a final sand column curing quality index mapping model.
4. The MICP intensity prediction method based on machine learning and multi-feature fusion of claim 3, wherein S4 comprises the steps of: s41, acquiring corresponding sand column curing parameter data in the current curing process of the sand column in real time according to the sand column curing parameter type set to obtain a current sand column curing parameter data set; S42, inputting the current sand column curing parameter data set into a final sand column curing quality index mapping model for mapping to obtain the current sand column curing quality index data set.
5. The MICP intensity prediction method based on machine learning and multi-feature fusion of claim 4, wherein S5 comprises the steps of: S51, setting a corresponding unconfined compressive strength threshold and a calcium carbonate distribution difference coefficient threshold according to the current sand column curing requirement; s52, repeatedly adjusting the current sand column curing parameter data set if the unconfined compressive strength data in the current sand column curing quality index data set is smaller than the unconfined compressive strength threshold or the calcium carbonate distribution difference coefficient data in the current sand column curing quality index data set is larger than or equal to the calcium carbonate distribution difference coefficient threshold, and inputting the adjusted current sand column curing parameter data set into the final sand column curing quality index mapping model again for mapping in each repetition until the unconfined compressive strength data in the mapping result is larger than or equal to the unconfined compressive strength threshold and the calcium carbonate distribution difference coefficient data in the mapping result is smaller than the calcium carbonate distribution difference coefficient threshold, so as to obtain the current final sand column curing parameter data set; S53, setting deviation threshold values of each type of easy-to-abnormal curing parameters and corresponding preset values according to the current actual sand column curing requirements, and obtaining a current easy-to-abnormal curing parameter deviation threshold value set.
6. The machine learning and multi-feature fusion based MICP intensity prediction method of claim 5, wherein S5 further comprises: S54, setting parameters for carrying out sand column curing currently according to the current final sand column curing parameter data set, starting to carry out current sand column curing operation after setting is completed, acquiring corresponding easy-abnormal curing parameter data in real time according to the easy-abnormal curing parameter type set in the operation process to obtain a current real-time easy-abnormal curing parameter set, and acquiring preset values corresponding to various current easy-abnormal curing parameters according to the current final sand column curing parameter data set to obtain a current initial easy-abnormal curing parameter set; and S6, executing the corresponding deviation threshold value set of the current easy-to-abnormal curing parameters, if the absolute value of the difference value between the corresponding parameters between the current real-time easy-to-abnormal curing parameter set and the current initial easy-to-abnormal curing parameter set is larger than or equal to the corresponding deviation threshold value, otherwise, taking no measures.
7. The MICP intensity prediction method based on machine learning and multi-feature fusion of claim 6, wherein S6 comprises the steps of: s61, inputting the current initial easy-to-abnormal curing parameter set and the current real-time easy-to-abnormal curing parameter set into a final sand column curing abnormal measure parameter mapping model for mapping to obtain a current abnormal measure parameter data set; S62, setting an execution feedback period, setting and executing corresponding parameters in the processes of replenishing bacterial liquid, executing a back flushing program and starting a cooling device according to the current abnormal measure parameter data set, and acquiring corresponding abnormal curing parameter data again in real time after the execution feedback period to obtain a current compensated abnormal curing parameter set; if the absolute value of the difference value between the corresponding parameters between the current compensated easy-to-abnormal curing parameter set and the current initial easy-to-abnormal curing parameter set is smaller than the corresponding deviation threshold value, the abnormal compensation is completed, otherwise, S63 is executed; S63, switching from the currently used bacterial liquid tank to the standby bacterial liquid tank and switching from the currently used grouting pipeline to the standby pipeline.
8. A system for realizing the MICP strength prediction method based on machine learning and multi-feature fusion according to any one of claims 1-7, which is characterized by comprising a historical sand column curing process data acquisition module, a sand column curing quality index mapping model construction module, a sand column curing abnormal measure parameter mapping model construction module, a current sand column curing parameter mapping module, a current curing parameter adjustment and easy abnormal curing parameter judgment module and a current abnormal measure parameter setting and measure execution feedback module; The historical sand column solidification process data acquisition module sets average unconfined compressive strength and calcium carbonate distribution difference coefficients and a plurality of influence factor parameters so as to acquire historical sand column solidification data; The sand column solidification quality index mapping model construction module constructs a final sand column solidification quality index mapping model based on the data acquired by the historical sand column solidification process data acquisition module; The sand column curing abnormal measure parameter mapping model construction module identifies a plurality of easy abnormal parameters including bacterial liquid activity, grouting pressure and environmental temperature, then sets corresponding counter measures and parameters thereof, and acquires historical data based on the counter measures and parameters thereof to construct a final sand column curing abnormal measure parameter mapping model; The current sand column curing parameter mapping module acquires current sand column curing parameter data in real time and inputs the current sand column curing parameter data into a mapping model in the sand column curing quality index mapping model construction module for mapping; Setting a compression strength and distribution difference coefficient double threshold value by the current curing parameter adjustment and easy-to-abnormality curing parameter judgment module, repeatedly adjusting the curing parameters and re-mapping until the result meets the requirement if the mapping result in the current sand column curing parameter mapping module meets the double threshold value; The current abnormal measure parameter setting and measure executing feedback module inputs the current curing parameter adjustment and the real-time parameter easy to be abnormally cured and the preset parameter acquired in the easy to be abnormally cured parameter judging module into a mapping model in the sand column curing abnormal measure parameter mapping model building module for mapping, executes the compensation operation according to the mapping result, and then feeds back and verifies the recovery condition of the easy to be abnormally parameter, if the recovery condition does not reach the expected condition, the hardware standby switching is started, otherwise, the compensation is completed.

Description

MICP strength prediction method and system based on machine learning and multi-feature fusion Technical Field The invention belongs to the field of solidification data analysis, and particularly relates to a MICP strength prediction method and system based on machine learning and multi-feature fusion. Background The prior art has insufficient recognition and response capability to abnormal states in the curing process, is difficult to discover and process burst process problems such as reduced bacterial liquid activity, pipeline blockage, abnormal environment temperature and the like in time, often causes unqualified curing quality or process interruption, and parameter adjustment under abnormal conditions mainly depends on manual experience, has slow response speed and unstable processing effect, lacks effective fault tolerance and redundancy design, and cannot guarantee the continuity and engineering reliability of the curing process. Disclosure of Invention Aiming at the problems in the related art, the invention provides a MICP strength prediction method and a MICP strength prediction system based on machine learning and multi-feature fusion, so as to overcome the technical problems in the prior art. In order to solve the technical problems, the invention is realized by the following technical scheme: the invention discloses a MICP strength prediction method based on machine learning and multi-feature fusion, which comprises the following steps: s1, setting an average unconfined compressive strength and a calcium carbonate distribution difference coefficient quality index, and setting a plurality of influence factor parameters to acquire historical sand column solidification data; s2, constructing a final sand column solidification quality index mapping model based on the data acquired in the S1; s3, identifying a plurality of easy-to-abnormality parameters including bacterial liquid activity, grouting pressure and environmental temperature, setting corresponding countermeasures and parameters thereof, and acquiring historical data based on the countermeasures and parameters thereof to construct a final sand column solidification abnormality measure parameter mapping model; s4, acquiring current sand column solidification parameter data in real time and inputting the current sand column solidification parameter data into a mapping model in the S2 for mapping; S5, setting a compressive strength and distribution difference coefficient double threshold, and repeatedly adjusting the curing parameters and re-mapping the curing parameters until the result meets the requirement if the mapping result in S4 meets the double threshold; S6, inputting the real-time parameters and the preset parameters which are acquired in the S5 and easy to be cured abnormally into the mapping model in the S3 for mapping, performing the compensation operation according to the mapping result, and then performing feedback verification on the recovery condition of the easy-to-be-abnormal parameters, if the recovery condition is not expected, starting the hardware standby switching, otherwise, finishing the compensation. Preferably, the step S1 includes the steps of: s11, setting a plurality of index types for measuring the quality of the sand column after being solidified by adopting an MICP mode to obtain a sand column solidification quality index type set, wherein the sand column solidification quality index type set comprises the average unconfined compressive strength of the sand column after solidification and the calcium carbonate distribution difference coefficient; Setting a plurality of parameter types which have influence on the quality after curing in the process of curing the sand column by adopting an MICP mode to obtain a sand column curing parameter type set, wherein the sand column curing parameter type set comprises fungus liquid characteristics, cementing liquid characteristics, environmental parameters, grouting modes, grouting interval time, grouting cycle times, soil quality and geometric parameters before curing the sand column; S12, acquiring sand column curing quality index data and sand column curing parameter data corresponding to the MICP sand column curing operation process for a plurality of times in history according to the sand column curing quality index type set and the sand column curing parameter type set, and obtaining a history sand column curing quality index data set and a history sand column curing parameter data set; By introducing key quality indexes such as average unconfined compressive strength, calcium carbonate distribution difference coefficient and the like, the curing effect can be comprehensively evaluated from two dimensions of mechanical property and curing uniformity, so that scientific basis is provided for subsequent process parameter adjustment. Preferably, the step S2 includes the steps of: S21, constructing a mapping model which is input into various sand column curing parame