CN-121746745-B - Risk early warning method and system for contraband detection equipment by utilizing cloud network end

CN121746745BCN 121746745 BCN121746745 BCN 121746745BCN-121746745-B

Abstract

The invention provides a contraband detection equipment risk early warning method and system by utilizing a cloud network side, which are characterized in that a feature spectrum to be matched is generated by acquiring Raman spectrum data of a sample to be detected and carrying out wavelet transformation, a cloud server gathers non-alarm spectrum data uploaded by multiple terminals, an environmental background feature spectrum set is generated by clustering, the feature spectrum to be matched is matched with a standard spectrum library and the background spectrum set, comprehensive matching scores are obtained based on spectral cross-correlation, peak position deviation and peak shape correlation weighted operation, when the comprehensive matching scores of the feature spectrum to be matched and any background spectrum are higher than a background suppression threshold, an adjusted comprehensive matching score of the contraband is generated, a matching ambiguity coefficient is determined based on the ratio of the highest score to the next highest score, and a preset threat level and a matching ambiguity coefficient of the corresponding contraband are combined, and a risk early warning level is output.

Inventors

REN HAIJIE
CHEN YU
ZHENG LU
YANG ZHEN
HAO PENGFEI
CHANG YU
Shao Jialiang
GAO LINQIANG
YAN SHUO

Assignees

雷神光电技术(天津)有限公司
河南雷神光电技术有限公司
郑州雷神光电技术研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260212

Claims (10)

1. A contraband detection equipment risk early warning method by utilizing a cloud network end is characterized by comprising the following steps: the cloud server gathers the non-alarm spectrum data uploaded by the multi-terminal in the early warning area in a variable time period, and clusters to generate an environmental background characteristic spectrum set; matching the characteristic spectrum to be matched with each standard spectrum in a cloud contraband standard spectrum library and each background spectrum in the environmental background characteristic spectrum set, and carrying out weighting operation based on spectrum cross correlation, peak position offset and peak shape correlation to obtain comprehensive matching scores corresponding to each standard spectrum and each background spectrum; when the comprehensive matching score of the characteristic spectrum to be matched and any background spectrum is higher than a background suppression threshold, nonlinear suppression is carried out on the comprehensive matching score of the characteristic spectrum to be matched and each standard spectrum, and a group of adjusted contraband comprehensive matching scores are generated; And calculating the ratio of the highest score to the next highest score based on the adjusted comprehensive matching score of the contraband to determine a matching ambiguity coefficient, and outputting a risk early warning grade by combining the highest score, the preset threat grade of the contraband corresponding to the highest score and the matching ambiguity coefficient.
2. The method according to claim 1, wherein the acquiring raman spectrum data of the sample to be tested and performing wavelet transform to generate a characteristic spectrum to be matched comprises: and decomposing the Raman spectrum data by adopting a wavelet base of a preset type, and extracting a low-frequency approximation coefficient obtained by decomposing a designated layer as the characteristic spectrum to be matched.
3. The method of claim 1, wherein the cloud server aggregates non-alarm spectrum data uploaded by multiple terminals in the early warning area in a variable time period, and clusters to generate an environmental background feature spectrum set, comprising: And carrying out cluster analysis on the converged non-alarm spectrum data by adopting a density-based spatial clustering algorithm, wherein the neighborhood radius of the algorithm is a preset value Eps, the minimum sample number of a core object is a preset value MinPts, and the arithmetic average value of all spectrum data in each cluster is used as the environmental background characteristic spectrum of the cluster.
4. The method of claim 1, wherein the duration of the time period By the formula Calculation of wherein As a reference time period of time, The average value of the risk level values of all early warning events in the previous period is given, k is the risk sensitivity coefficient, Is a positive constant.
5. The method of claim 1, wherein the integrated match score S is determined by the formula Calculating; Wherein the method comprises the steps of For normalized cross-correlation coefficient values of the characteristic spectrum to be matched with the standard spectrum or the background spectrum, For the normalized peak position matching score calculated based on the peak position deviation of the two main characteristic peaks, For the correlation coefficient values of both peak-shaped profiles, 、 And Is a corresponding preset weight coefficient.
6. The method of claim 1, wherein the background suppression threshold By the formula Calculation of wherein As a basis for the suppression threshold value, In the cluster to which the background spectrum belongs number of non-alarm spectrum samples contained.
7. The method of claim 1, wherein said non-linearly suppressing the combined match scores of the to-be-matched characteristic spectrum and each of the standard spectra to generate a set of adjusted contraband combined match scores comprises: Adjusted contraband comprehensive matching score By the formula Calculating; Wherein the method comprises the steps of To match the original composite score with the ith standard spectrum, The highest comprehensive matching score obtained by matching the characteristic spectrum to be matched with each background spectrum in the environmental background characteristic spectrum set, For the background suppression threshold for the background spectrum corresponding to the highest overall match score, Is a suppression coefficient.
8. The method of claim 1, wherein the outputting the risk pre-warning level in combination with the highest score, the preset threat level of the contraband corresponding to the highest score, and the matching ambiguity factor comprises: By the formula Calculating a Risk score Risk; dividing the risk score into a plurality of early warning levels according to a multi-level preset threshold; Wherein the method comprises the steps of For the adjusted highest score value, For a preset threat level where the highest score corresponds to contraband, In order to match the coefficient of ambiguity, 、 And Is a preset weight coefficient and 。
9. A contraband detection equipment risk early warning system utilizing a cloud network end is characterized by comprising the following modules: The cloud server gathers the non-alarm spectrum data uploaded by the multi-terminal in the early warning area in a variable time period and clusters the non-alarm spectrum data to generate an environmental background characteristic spectrum set; The operation module is used for matching the characteristic spectrum to be matched with each standard spectrum in the cloud contraband standard spectrum library and each background spectrum in the environment background characteristic spectrum set, and carrying out weighted operation based on spectrum cross correlation, peak position offset and peak shape correlation to obtain a comprehensive matching score corresponding to each standard spectrum and each background spectrum; The second generation module is used for carrying out nonlinear inhibition on the comprehensive matching scores of the characteristic spectrum to be matched and each standard spectrum when the comprehensive matching score of the characteristic spectrum to be matched and any background spectrum is higher than a background inhibition threshold value, so as to generate a group of adjusted comprehensive matching scores of contraband; And the adjusting module is used for calculating the ratio of the highest score to the next highest score based on the adjusted comprehensive matching score of the contraband to determine a matching ambiguity coefficient, and outputting a risk early warning grade by combining the highest score, the preset threat grade of the contraband corresponding to the highest score and the matching ambiguity coefficient.
10. The system of claim 9, wherein the acquiring raman spectrum data of the sample to be tested and performing wavelet transform to generate a characteristic spectrum to be matched comprises: and decomposing the Raman spectrum data by adopting a wavelet base of a preset type, and extracting a low-frequency approximation coefficient obtained by decomposing a designated layer as the characteristic spectrum to be matched.

Description

Risk early warning method and system for contraband detection equipment by utilizing cloud network end Technical Field The application belongs to the field of equipment early warning, and particularly relates to a contraband detection equipment risk early warning method and system by utilizing a cloud network end. Background The Raman spectrum technology is used as a molecular spectrum analysis technology for rapid, nondestructive and fingerprint identification. The molecular structure information can be rapidly obtained by carrying out laser irradiation on the substances and analyzing the scattering spectrum, so that the accurate identification of prohibited articles such as explosives, drugs, dangerous chemicals and the like is realized. However, the actual detection environment is complex, the sample to be detected often exists in various packaging materials, containers or mixtures, and the coexisting background substances, such as plastics, textiles, beverages, cosmetics and the like, can cause serious interference to the characteristic peak of the target contraband due to the raman signal of the sample to be detected, and even completely mask the target signal. Background interference is usually subtracted by establishing a static background spectrum library in the prior art, but the method is difficult to cope with the change of the field environment. For example, factors such as packaging materials of different batches, temperature and humidity changes of field environment, detergent residues and the like can input new and unknown background signals, so that a static spectrum library is invalid, a large number of false positives and false negatives are caused, and the accuracy and reliability of detection are seriously affected. When the spectrum matching is carried out on most systems, the single similarity calculation method such as the spectrum cross-correlation or the Euclidean distance is relied on, and the calculated score is compared with a fixed threshold value to judge whether contraband exists or not. When the spectrum to be measured has a certain degree of similarity with various standard substances, the simple principle of highest score matching cannot reliably process the ambiguity in the identification. More importantly, the existing early warning mechanism is mostly binary, namely only outputs an alarm or safety result, and the reliability of the identification result and the inherent danger level difference of different contraband products are not comprehensively considered. For example, the risks of high-concentration explosives and low-toxicity chemicals are obviously different, but the conventional early warning system cannot distinguish the risks, lacks multi-level risk assessment and early warning capability, particularly when the recognition result has ambiguity such as spectrum similarity with various substances, cannot quantify and incorporate the uncertainty into the risk assessment, and is difficult to meet the requirement of rapid and grading of the risks in the modern advanced security scene. Disclosure of Invention The invention provides a risk early warning method for contraband detection equipment by utilizing a cloud network end, which is used for solving the problems that the existing method is difficult to deal with the change of the field environment, and the reliability of the identification result and the inherent dangerous grade difference of different contraband are not comprehensively considered, and comprises the following steps: the cloud server gathers the non-alarm spectrum data uploaded by the multi-terminal in the early warning area in a variable time period, and clusters to generate an environmental background characteristic spectrum set; matching the characteristic spectrum to be matched with each standard spectrum in a cloud contraband standard spectrum library and each background spectrum in the environmental background characteristic spectrum set, and carrying out weighting operation based on spectrum cross correlation, peak position offset and peak shape correlation to obtain comprehensive matching scores corresponding to each standard spectrum and each background spectrum; when the comprehensive matching score of the characteristic spectrum to be matched and any background spectrum is higher than a background suppression threshold, nonlinear suppression is carried out on the comprehensive matching score of the characteristic spectrum to be matched and each standard spectrum, and a group of adjusted contraband comprehensive matching scores are generated; And calculating the ratio of the highest score to the next highest score based on the adjusted comprehensive matching score of the contraband to determine a matching ambiguity coefficient, and outputting a risk early warning grade by combining the highest score, the preset threat grade of the contraband corresponding to the highest score and the matching ambiguity coefficient. Optionally, the acquir