CN-121978102-A - High-precision precipitation particle phase state identification method and system

CN121978102ACN 121978102 ACN121978102 ACN 121978102ACN-121978102-A

Abstract

The invention discloses a high-precision precipitation particle phase state identification method and a high-precision precipitation particle phase state identification system, which relate to the technical field of atmospheric detection and solve the problems of difficult precipitation particle tracking, multiple misjudgment of phase state identification and inaccurate basis of misjudgment point restoration in the prior art; and constructing a special traceable sample set for the misjudgment point, and deducing and repairing the real phase state of the sample set according to the phase state association type of the sample set. The invention realizes the unique tracking of the whole particle flow, accurately screens and restores the misjudgment point, improves the integral accuracy of precipitation particle phase identification, and ensures the continuity of phase evolution.

Inventors

CHEN YICHEN
ZHANG XING
ZHANG LEI
SHI SHAOYING
HUANG MENGYU
WANG YONG
TANG JINPING
An shuangdeng
LI XIA
JING YINGYING
MA NINGKUN

Assignees

北京市人工影响天气中心
南京旗云中天科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (10)

1. The high-precision precipitation particle phase state identification method is characterized by comprising the following steps of: step 1, carrying out continuous frame synchronous sampling on precipitation particles in a detection area, distributing a unique particle identification for each collected precipitation particle, synchronously collecting three-dimensional space coordinates, phase state identification results and sampling time stamps of the particle in each sampling frame, constructing a space-time track of the particle according to the sequence of the sampling time stamps aiming at all continuous sampling points corresponding to the same particle identification, carrying out integrity check on the space-time track, and marking the space-time track passing through the integrity check with a corresponding phase state identification result sequence; Step 2, based on the space-time track passing through the integrity check, respectively carrying out time continuity check and space continuity check, marking the sampling points which do not pass through any one of the two types of check as suspicious misjudgment points, otherwise marking the sampling points as effective sampling points, and forming suspicious misjudgment point data sets and effective sampling point data sets of each particle; And 3, aiming at each suspicious misjudgment point, determining the time sequence position of the suspicious misjudgment point in the space-time track according to the unique identifier of the particle to which the suspicious misjudgment point belongs, constructing a tracing sample set of the suspicious misjudgment point, and repairing the real phase state of the suspicious misjudgment point based on the phase state change of the effective sampling point in the tracing sample set.
2. A method of high precision precipitation particle phase identification according to claim 1, wherein each collected precipitation particle is assigned a unique particle identification specific operation as: Extracting the contour complexity C and the particle size morphology integrated value F of each particle from each particle image obtained by synchronous sampling of continuous frames, wherein C, F is E [0,1]; Presetting a unified interval dividing number S, equally dividing [0,1] into S continuous and non-overlapping subintervals, namely [0,1/S ], [1/S,2/S ], [ S-1)/S, 1], and distributing fixed interval numbers 1,2 and 3. For the quantization characteristics (C, F) of the current particles, respectively matching the quantization characteristics with the sections to obtain section numbers Rc and Rf corresponding to the sections corresponding to the C and the F; A fixed integer N which is larger than the total number S of the sections is preset, a unique particle identification ID is generated, namely ID=Rc multiplied by N+Rf, the ID generated by the first frame of the particle is used as a unique life identification, and subsequent frames are not recalculated and are directly used.
3. The method for identifying the phase state of the high-precision precipitation particles according to claim 2, wherein the specific rule for extracting the contour complexity C of each particle is to traverse the edge pixels of the particle image, calculate the ratio of the total number of the edge pixels of the particle to the total number of the circumscribed rectangular pixels of the particle, and record the ratio as C.
4. The method for identifying phase states of precipitation particles with high precision according to claim 2, wherein the specific rule for extracting the particle size morphology integrated value F of each particle is as follows: calculating the ratio of the length of the long axis to the length of the short axis of the particle by scanning the three-dimensional profile of the particle, and marking the ratio as M, and simultaneously calculating the ratio of the volume of the particle to the volume of the sphere with the same diameter, and marking the ratio as K; taking the reciprocal of M to obtain the normalized elongation M ' =1/M, and carrying out characteristic coupling operation on M ' and K to obtain the particle size morphology integrated value F=M ' ×K.
5. The method for identifying the phase state of the high-precision precipitation particles according to claim 1, wherein the specific steps of carrying out the integrity check on the time-space trajectory are as follows: calculating the three-dimensional space linear distance of every two groups of adjacent sampling points in the space-time track, and marking as Di, i E [1, n-1], wherein n is the total number of sampling points of the space-time track; Calculating an arithmetic mean value Davg and a variance SD of all the three-dimensional space straight-line distances based on the three-dimensional space straight-line distances; Calibrating a smoothness judgment threshold value Ts=k×Davg based on the characteristics of the space-time track, judging as a smoothness qualified track if the variance SD of the space-time track is less than or equal to Ts, otherwise, judging as a smoothness unqualified track, marking the smoothness unqualified track as a candidate incomplete track, and temporarily entering the next step of verification, wherein k is a smoothness proportionality coefficient; Setting a Z-axis coordinate of an entrance boundary of a detection area as Zin and a Z-axis coordinate of an exit boundary as Zout according to the smoothness qualified track, and setting a boundary quantization adaptive difference value delta Z at the same time, wherein Zin > Zout and delta Z are adjustable positive real numbers; Extracting initial sampling point coordinates (X1, Y1, Z1) and end sampling point coordinates (Xn, yn, zn) of a smoothness qualified track, calculating a difference absolute value |Z1-zin| of the Z-axis coordinates of the initial sampling point and the Z-axis coordinates of an inlet boundary, judging that the inlet adaptation is qualified if the absolute value |Z1-zin| is less than or equal to delta Z, calculating a difference absolute value |Zn-zout| of the Z-axis coordinates of the end sampling point and the Z-axis coordinates of an outlet boundary, and judging that the outlet adaptation is qualified if the absolute value |Zn-zout| is less than or equal to delta Z; If the space-time track with qualified smoothness, qualified inlet adaptation and qualified outlet adaptation is simultaneously satisfied, the space-time track is judged to pass the integrity check and is reserved, otherwise, the space-time track is directly removed.
6. The method for identifying the phase state of the precipitation particles with high precision according to claim 1, wherein the specific steps of performing the time consistency check are as follows: Extracting a corresponding phase state identification result sequence aiming at the particle space-time track passing through the integrity verification, extracting a phase state identification result P of a front adjacent sampling point and a phase state identification result Q of a rear adjacent sampling point of each sampling point to be verified in the sequence, and marking the phase state identification result of the current sampling point to be verified as R; Comparing the phase state identification result R of the current sampling point to be checked with the preamble phase state P and the follow-up phase state Q, if R is the physical gradual transition intermediate state of P and Q or is consistent with P, Q, judging that the time continuity check is passed, otherwise, judging that the time continuity check is not passed; The physical gradual-change intermediate state refers to that the same precipitation particle is positioned between two different phase states in a continuous and single physical evolution process, and has core physical characteristics of an initial phase state and a final phase state.
7. The method for identifying the phase state of high-precision precipitation particles according to claim 1, wherein the specific steps of performing the spatial continuity check are as follows: Setting a fixed radius A by taking the three-dimensional space coordinates of the current sampling point to be checked as the center, defining a three-dimensional spherical space neighborhood, traversing all particles in the neighborhood, and incorporating the space-time track which passes the integrity check to form a neighborhood effective track set; Extracting corresponding initial three-dimensional coordinates and corresponding termination three-dimensional coordinates of each complete space-time track in the neighborhood effective track set, calculating initial coordinate spacing and termination coordinate spacing of any two space-time tracks in the set, classifying the initial coordinate spacing and the termination coordinate spacing into the same cluster set if the initial coordinate spacing and the termination coordinate spacing are both in a spacing judgment interval, and repeating the operation until all tracks in the neighborhood effective track set are classified into corresponding cluster sets, wherein the spacing judgment interval is an inherent coordinate error range generated in the equipment sampling process; Counting the number of space-time tracks contained in each cluster set, marking the cluster set with the largest number of space-time tracks as a dominant cluster set, counting the number of space-time tracks corresponding to each phase state in the dominant cluster set, marking the phase state with the largest number of tracks as a neighborhood track dominant phase state, comparing the phase state of the current sampling point to be checked with the neighborhood track dominant phase state, if the two phases are consistent, judging that the space continuity check is passed, otherwise, judging that the space continuity check is not passed.
8. The method for identifying the phase state of the high-precision precipitation particles according to claim 1, wherein the specific operation of constructing the traceable sample set of suspicious misjudgment points is as follows: Based on particle identification associated with suspicious misjudgment points, invoking a space-time track which corresponds to the identification and passes through the integrity verification, and carrying out ascending sort on all sampling points in the space-time track according to sampling time stamps, and simultaneously, distributing a unique time sequence number for each sampling point; taking the time sequence serial numbers corresponding to the suspicious misjudgment points as the center, extending the B time sequence serial numbers forwards and extending the B time sequence serial numbers backwards, dividing the time sequence serial numbers into exclusive time sequence windows of the suspicious misjudgment points, screening all sampling points in the exclusive time sequence windows, and reserving effective sampling points which pass through time continuity check and space continuity check; And integrating all the screened effective sampling points according to the one-way increasing sequence of the corresponding time sequence numbers to construct a exclusive traceability sample set of the suspicious misjudgment point, and marking the corresponding time sequence numbers and phase recognition results for each effective sampling point in the exclusive traceability sample set.
9. The method for identifying the phase state of the precipitation particles with high precision according to claim 1, wherein the repairing of the true phase state of the suspected misjudgment point comprises the following steps: The exclusive tracing sample set corresponding to the suspicious misjudgment point is called, the time sequence number and the phase state identification result of all the effective sampling points in the sample set are extracted, the time sequence number and the phase state identification result are sorted according to the one-way increasing sequence of the time sequence number, a time sequence number-phase state ordered sequence is formed, meanwhile, the phase state association type of the sample set is judged based on the ordered sequence, and the specific rule is that if the phase states of all the effective sampling points in the ordered sequence are completely the same, the phase state consistency association is judged; The real phase state of the suspicious misjudgment point is determined according to the determined phase state association type, and the specific rule is that if the correlation is phase state consistency, the real phase state of the suspicious misjudgment point is completely consistent with the phase state of all effective sampling points in the exclusive tracing sample set; binding the determined real phase state with the suspicious misjudgment point, and incorporating the repaired suspicious misjudgment point into an effective sampling point data set of the particle.
10. A high-precision precipitation particle phase recognition system for performing a high-precision precipitation particle phase recognition method according to any of claims 1-9, comprising: The track construction module is used for carrying out continuous frame synchronous sampling on precipitation particles in the detection area, distributing a unique particle identification for each collected precipitation particle, synchronously collecting three-dimensional space coordinates, phase state identification results and sampling time stamps of the particle in each sampling frame, constructing a space-time track of the particle according to the sequence of the sampling time stamps aiming at all continuous sampling points corresponding to the same particle identification, carrying out integrity check on the space-time track, and marking the space-time track passing through the integrity check with a corresponding phase state identification result sequence; The track checking module is used for respectively carrying out time continuity check and space continuity check based on the space-time track passing through the integrity check, marking the sampling points which do not pass through any one of the two types of check as suspicious misjudgment points, otherwise marking the sampling points as effective sampling points, and forming a suspicious misjudgment point data set and an effective sampling point data set of each particle; The phase state restoration module is used for defining the time sequence position of the suspicious misjudgment point in the space-time track according to the unique identifier of the particle to which each suspicious misjudgment point belongs, constructing a tracing sample set of the suspicious misjudgment point, and restoring the real phase state of the suspicious misjudgment point based on the phase state change of the effective sampling point in the tracing sample set.

Description

High-precision precipitation particle phase state identification method and system Technical Field The invention relates to the technical field of atmospheric detection, in particular to a high-precision precipitation particle phase state identification method and system. Background The accurate identification of the precipitation particle phase state is a core technology of weather monitoring and precipitation type prejudging, has important significance for works such as weather forecast, disaster prevention and reduction, and the like, however, the existing precipitation particle phase state identification technology still has the following defects in practical application: Firstly, particle identification distribution mostly depends on single morphological characteristics, so that the problem of identification repetition is easy to occur, the unique tracking of the particle full sampling process can not be realized, and dislocation and confusion occur in space-time track construction; Secondly, phase state identification is usually completed based on single frame sampling data, a complete space-time track is not constructed aiming at a particle motion process, and consistency verification of phase state change is lacked; thirdly, the processing of suspicious misjudgment points comprises direct elimination or blind replacement based on an external preset rule, so that not only is effective data lost, but also the natural evolution continuity of the particle phase state is easily damaged, and the deviation between a repair result and the real phase state is larger; Fourthly, although some technologies try track verification, collaborative verification is not carried out by combining time and space double dimensions, and abnormal sampling points of phase identification cannot be effectively screened; therefore, a high-precision precipitation particle phase identification method and system are needed. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a high-precision precipitation particle phase state identification method and a high-precision precipitation particle phase state identification system, and solves the problems that the existing precipitation particles are difficult to track, the number of misjudgment points for phase state identification is large, and the repairing of misjudgment points is not accurate. In order to achieve the purpose, the invention is realized by the following technical scheme that the high-precision precipitation particle phase state identification method comprises the following steps: step 1, carrying out continuous frame synchronous sampling on precipitation particles in a detection area, distributing a unique particle identification for each collected precipitation particle, synchronously collecting three-dimensional space coordinates, phase state identification results and sampling time stamps of the particle in each sampling frame, constructing a space-time track of the particle according to the sequence of the sampling time stamps aiming at all continuous sampling points corresponding to the same particle identification, carrying out integrity check on the space-time track, and marking the space-time track passing through the integrity check with a corresponding phase state identification result sequence; Step 2, based on the space-time track passing through the integrity check, respectively carrying out time continuity check and space continuity check, marking the sampling points which do not pass through any one of the two types of check as suspicious misjudgment points, otherwise marking the sampling points as effective sampling points, and forming suspicious misjudgment point data sets and effective sampling point data sets of each particle; And 3, aiming at each suspicious misjudgment point, determining the time sequence position of the suspicious misjudgment point in the space-time track according to the unique identifier of the particle to which the suspicious misjudgment point belongs, constructing a tracing sample set of the suspicious misjudgment point, and repairing the real phase state of the suspicious misjudgment point based on the phase state change of the effective sampling point in the tracing sample set. As a further aspect of the invention, each collected precipitation particle is assigned a unique particle identification specific operation as: Extracting the contour complexity C and the particle size morphology integrated value F of each particle from each particle image obtained by synchronous sampling of continuous frames, wherein C, F is E [0,1]; Presetting a unified interval dividing number S, equally dividing [0,1] into S continuous and non-overlapping subintervals, namely [0,1/S ], [1/S,2/S ], [ S-1)/S, 1], and distributing fixed interval numbers 1,2 and 3. For the quantization characteristics (C, F) of the current particles, respectively matching the quantization characteristics with the sections to obtain