South China Sea fishery intelligent forecasting system based on AI

Abstract

The invention, which relates to the technical field of the South China Sea fishery, discloses an AI-based intelligent forecasting system for a South China Sea fishery, comprising a ship identification and classification unit and a behavior supervision unit. The ship identification and classification unit is used for identifying ships entering the South China Sea fishery through a multi-source data fusion technology, verifying consistency of appearance characteristics, fishing license numbers and ship grade information of the ships, obtaining fishing legality of the ships, and obtaining fishing legality of the ships through integrated ship identification and classification, behavior supervision and forecasting function modules. According to the method, the ships entering the fishery can be quickly and accurately identified and classified, the fishery supervision efficiency is effectively improved, meanwhile, the sensors integrated on the ships and the fishery patrol boats are used for monitoring fishing activity data in real time, and the compliance of fishing behaviors is accurately evaluated through the fishing intensity evaluation model, so that the fishing efficiency is improved. And timely and accurate supervision information and law enforcement basis are provided for fishery administration departments.

Description

Technical Field

[0001] The present invention relates to the technical field of South China Sea fisheries, and in particular to an AI-based intelligent forecasting system for South China Sea fisheries. Background Art

[0002] The South China Sea fishing grounds are China's most important tropical fishing grounds. Located in the central continental shelf of the South China Sea, they are a key location for catching a variety of commercial marine fish, shrimp, and crabs. With high water temperature, high salinity, fertile water, and abundant food, they are an ideal place for the reproduction and growth of numerous marine organisms and a vital component of China's marine fishery resources.

[0003] In existing technologies, traditional fishery supervision methods mainly rely on manual patrols and observations. This method is not only inefficient, but also difficult to achieve comprehensive monitoring of fisheries. In addition, due to the lack of intelligent early warning systems, it is difficult for fishery administration departments to promptly detect and respond to illegal fishing activities, which not only undermines the sustainable use of fishery resources, but also affects the balance of the marine ecological environment. Although some fishery management systems have begun to try to introduce information technology, most of these systems have problems such as single functions and low intelligence. They can only achieve simple identification and tracking of ships, but cannot effectively assess the compliance of fishing activities. They also lack data analysis and early warning functions, making it difficult to provide timely and accurate supervision information for fishery administration departments.

[0004] In view of this, the present invention proposes an AI-based South China Sea fishery intelligent forecasting system to make up for and improve the shortcomings of the existing technology. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides an AI-based South China Sea fishing ground intelligent forecasting system to solve the corresponding technical problems raised in the above background technology.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is: an AI-based South China Sea fishing ground intelligent forecasting system, comprising: a vessel identification and classification unit and a behavior supervision unit;

[0007] The vessel identification and classification unit is used to identify vessels entering the South China Sea fishing grounds through multi-source data fusion technology, verify the consistency of their appearance characteristics, fishing license numbers, and vessel grade information, obtain the fishing legality of the vessel, and classify the vessel into a fishing vessel or a non-fishing vessel based on the fishing legality result, and send the classification results to the behavior supervision unit;

[0008] The behavior supervision unit includes a fishing vessel supervision module and a non-fishing vessel supervision module. The fishing vessel supervision module is used to monitor the fishing activity data of the fishing vessel in real time through sensors integrated on the vessel and the fishery patrol boat, establish a fishing intensity assessment model based on the fishing activity data, obtain the current fishing intensity F, and calculate the fishing intensity according to the preset fishing limit F. q , the current fishing intensity F and the fishing limit F q Compare and evaluate the compliance of current fishing activities. Based on the compliance assessment results and the preset grading standards, the fishing degree is divided into compliance, minor violation, moderate violation, and severe violation levels, and the fishing degree level is sent to the forecasting unit;

[0009] The non-fishing vessel supervision module is used to obtain image data through surveillance cameras and drone equipment installed around the fishing grounds, combine AI image recognition technology, and analyze the behavior patterns of non-fishing vessels. It combines historical data with real-time intelligence to establish an illegal fishing behavior prediction model, and obtains a preset behavior threshold p based on historical illegal fishing behavior data and law enforcement experience. th , evaluate the behavior of non-fishing vessels, and divide the evaluation results into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior and high-risk illegal behavior levels and send them to the forecast unit.

[0010] Preferably, a forecasting unit is also included, which is used to obtain the fishing degree level of fishing vessels and the behavior level of non-fishing vessels, generate corresponding early warning information according to the fishing degree level and the behavior level, push the early warning information and related vessel information to fishery administration departments and fishery enterprises, and automatically assign processing tasks to fishery law enforcement personnel and managers of fishery enterprises according to the urgency and type of the early warning information.

[0011] As a preferred method, the specific process of establishing a fishing intensity assessment model is as follows:

[0012] S101, obtaining fishing activity data, wherein the fishing activity data includes fishing time T, number of fishing nets N, area of ​​fishing area A, and catch per unit time M;

[0013] S102, cleaning, denoising and standardizing the acquired fishing activity data;

[0014] S103: Based on the fishing activity data, a fishing intensity assessment index system is constructed. The fishing intensity assessment index includes the fishing duration density D T , fishing net density D N and fishing efficiency E;

[0015] Among them, the fishing time density D Tis the fishing time per unit fishing area, and the calculation formula is:

[0016] Fishing gear density D N is the number of fishing gear per unit fishing area, and the calculation formula is:

[0017] Fishing efficiency E is the amount of fish caught per unit fishing gear per unit time, and the calculation formula is:

[0018] S104. Construct a judgment matrix, determine the relative importance of each indicator through expert scoring, calculate the maximum eigenvalue and corresponding eigenvector of the judgment matrix, perform consistency test, normalize the eigenvector, and obtain the weight w of each indicator. T 、w N and w E , and satisfy w T +w N +w E =1;

[0019] S105. Based on each indicator and its weight, a fishing intensity assessment model is established to calculate the current fishing intensity F. The calculation formula is: F = w T *D T +w N *D N +w E *E;

[0020] S106. Use historical fishing data to verify the model and optimize the model by calculating the error between the model prediction value and the actual value.

[0021] As a preferred method, the specific process of fishing degree division is as follows:

[0022] S201, based on the current fishing intensity F and the preset fishing limit F q , calculate the fishing intensity deviation rate R, the calculation formula is:

[0023] S202. Based on actual regulatory needs and experience, set thresholds for fishing severity classification, including the light violation threshold R1, the moderate violation threshold R2, and the severe violation threshold R3, and satisfy the requirement of 0% ≤ R1. <R2<R3≤100%;

[0024] S203: Based on the fishing intensity deviation rate R and the set fishing intensity classification threshold, the fishing intensity is classified into the following levels:

[0025] When R=0%, that is, F=F q , the fishing degree level is compliant;

[0026] When 0% < R ≤ R1, the fishing intensity level is a minor violation;

[0027] When R1 < R ≤ R2, the fishing intensity level is a moderate violation;

[0028] When R > R3, the fishing intensity level is a severe violation.

[0029] Preferably, the specific process of establishing the illegal fishing behavior prediction model is as follows:

[0030] S301. Obtain image data and form an image data set D img ={I1, I2,..., I n}, where n is the total number of images. Collect historical illegal fishing behavior data and real-time intelligence to form a non-image data set D non ={X1, X2,..., X m}, where m is the number of non-image data entries. Correlate and integrate the image data and non-image data, and match the corresponding non-image data X i for each image data I j to form a comprehensive data set D ={(I i , X j )};

[0031] S302. Use a pre-trained convolutional neural network to extract features from the image data I i to obtain an image feature vector α i . Use the one-hot encoding method to transform the categorical variables in the non-image data X j and normalize the numerical variables to obtain a non-image feature vector β j . Concatenate the image feature vector α i and the non-image feature vector β j to obtain a comprehensive feature vector Z ij =|α i , β j |;

[0032] S303. Select the random forest algorithm as the illegal fishing behavior prediction model, and input the comprehensive feature vector set Z ={(Z ij , Y ij )} into the model for training, where Y ij is the label, 0 represents non-fishing behavior, and 1 represents illegal fishing behavior. Use the classification accuracy L as the evaluation index for model training, and its calculation formula is:

[0033] where k is the total number of samples;

[0034] Y prThe label of the kth sample predicted by the model;

[0035] Y pr is the true label of the k-th sample;

[0036] I(·) is an indicator function, which takes the value 1 when the condition in the brackets is met and 0 otherwise.

[0037] As a preferred method, the specific process of classifying behavior levels is as follows:

[0038] According to the model prediction probability p and the behavior threshold p th Based on the relationship between the two, the behavior levels of non-fishing vessels are divided into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior and high-risk illegal behavior. The specific classification rules are as follows:

[0039] When p <p th -Δp sa , it is divided into the non-fishing behavior level, where Δp sa is the preset safety probability interval;

[0040] When p≤p th When the illegal behavior is classified as low risk;

[0041] When p th <p≤p th When +Δp, it is classified as a medium-risk illegal behavior level;

[0042] When p>p th +Δp, it is classified as a high-risk illegal behavior level, where Δp is the preset risk level interval.

[0043] Compared with the existing technology, the beneficial effects of the present invention are: by integrating functional modules of ship identification and classification, behavior supervision and forecasting, it is possible to quickly and accurately identify and classify ships entering the fishing grounds, effectively improving the efficiency of fishery supervision. At the same time, sensors integrated on ships and fishery patrol boats are used to monitor fishing activity data in real time, and the compliance of fishing behavior is accurately evaluated through the fishing intensity assessment model, providing timely and accurate supervision information and law enforcement basis for fishery departments, effectively avoiding overfishing and illegal fishing, and protecting the sustainable use of fishery resources. In addition, through AI image recognition technology and illegal fishing behavior prediction models, the behavior patterns of non-fishing vessels can be analyzed, and potential illegal fishing behaviors can be warned, further strengthening the supervision of fisheries. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the overall structure of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] Embodiments of the present invention:

[0047] Please refer to Figure 1 As shown, an AI-based South China Sea fishing ground intelligent forecasting system includes: a ship identification and classification unit and a behavior supervision unit;

[0048] The vessel identification and classification unit is used to identify vessels entering the South China Sea fishing grounds through multi-source data fusion technology, verify the consistency of their appearance characteristics, fishing license numbers, and vessel classification information, and determine the legality of the vessel's fishing. Based on the legality of the vessel's fishing, the vessel is classified as a fishing vessel or a non-fishing vessel, and the classification results are sent to the behavior supervision unit.

[0049] The behavior supervision unit includes a fishing vessel supervision module and a non-fishing vessel supervision module. The fishing vessel supervision module is used to monitor the fishing activity data of the fishing vessel in real time through sensors integrated on the vessel and the fishery patrol boat. Based on the fishing activity data, a fishing intensity assessment model is established to obtain the current fishing intensity F and the preset fishing limit F is used to evaluate the fishing intensity. q , the current fishing intensity F and the fishing limit F q Compare and evaluate the compliance of current fishing activities. Based on the compliance assessment results and the preset grading standards, the fishing degree is divided into compliance, minor violation, moderate violation, and severe violation levels, and the fishing degree level is sent to the forecasting unit;

[0050] The non-fishing vessel supervision module is used to obtain image data through surveillance cameras and drone equipment installed around the fishing grounds, combine AI image recognition technology, and analyze the behavior patterns of non-fishing vessels. It combines historical data with real-time intelligence to establish an illegal fishing behavior prediction model, and obtains a preset behavior threshold based on historical illegal fishing behavior data and law enforcement experience. th , evaluate the behavior of non-fishing vessels, and classify the evaluation results into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior, and high-risk illegal behavior levels and send them to the forecast unit;

[0051] The system also includes a forecasting unit for obtaining the fishing degree level of fishing vessels and the behavior level of non-fishing vessels, generating corresponding warning information based on the fishing degree level and behavior level, and pushing the warning information and related vessel information to fishery administration departments and fishery enterprises. Based on the urgency and type of the warning information, the system automatically assigns processing tasks to fishery law enforcement personnel and fishery enterprise managers;

[0052] The specific process of establishing a fishing intensity assessment model is as follows:

[0053] S101, obtaining fishing activity data, the fishing activity data including fishing time T, number of fishing nets N, fishing area A, and catch per unit time M;

[0054] S102, cleaning, denoising and standardizing the acquired fishing activity data;

[0055] S103. Based on the fishing activity data, a fishing intensity assessment index system is constructed. The fishing intensity assessment index includes fishing duration density D T , fishing net density D N and fishing efficiency E;

[0056] Among them, the fishing time density D T is the fishing time per unit fishing area, and the calculation formula is:

[0057] Fishing gear density D N is the number of fishing gear per unit fishing area, and the calculation formula is:

[0058] Fishing efficiency E is the amount of fish caught per unit fishing gear per unit time, and the calculation formula is:

[0059] S104. Construct a judgment matrix, determine the relative importance of each indicator through expert scoring, calculate the maximum eigenvalue and corresponding eigenvector of the judgment matrix, perform consistency test, normalize the eigenvector, and obtain the weight w of each indicator. T 、w N and w E , and satisfy w T +w N +w E =1;

[0060] S105. Based on each indicator and its weight, a fishing intensity assessment model is established to calculate the current fishing intensity F. The calculation formula is: F = w T *D T +w N *D N +wE *E;

[0061] S106. Use historical fishing data for model verification, and optimize the model by calculating the error between the model prediction value and the actual value.

[0062] The specific process of fishing degree classification is as follows:

[0063] S201. According to the current fishing intensity F and the preset fishing quota F q , calculate the fishing intensity deviation rate R, and the calculation formula is:

[0064] S202. According to the actual supervision requirements and experience, set the fishing degree level classification thresholds, including the mild violation threshold R1, the moderate violation threshold R2, and the severe violation threshold R3, and satisfy 0% ≤ R1 < R2 < R3 ≤ 100%;

[0065] S203. According to the fishing intensity deviation rate R and the set fishing degree level classification thresholds, classify the fishing degree into the following levels:

[0066] When R = 0%, that is, F = F[[ID=2​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​Perform feature extraction to obtain the image feature vector α i , use one-hot encoding to encode non-image data X j The categorical variables in are transformed and the numerical variables are normalized to obtain the non-image feature vector β j , the image feature vector α i and the non-image feature vector β j Splice and get the comprehensive feature vector Z ij =|α i ,β j |;

[0073] S303, select the random forest algorithm as the illegal fishing behavior prediction model, and transform the comprehensive feature vector set Z={(Z ij ,Y ij )} is input into the model for training, where Y ij is a label, 0 indicates non-fishing behavior, and 1 indicates illegal fishing behavior. The classification accuracy L is used as the evaluation indicator for model training, and its calculation formula is:

[0074] Where k is the total number of samples;

[0075] Y pr The label of the kth sample predicted by the model;

[0076] Y pr is the true label of the k-th sample;

[0077] I(·) is an indicator function, which takes the value 1 when the condition in the brackets is met and 0 otherwise;

[0078] The specific process of classifying behavior levels is as follows:

[0079] According to the model prediction probability p and the behavior threshold p th Based on the relationship between the two, the behavior levels of non-fishing vessels are divided into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior and high-risk illegal behavior. The specific classification rules are as follows:

[0080] When p <p th -Δp sa , it is divided into the non-fishing behavior level, where Δp sa is the preset safety probability interval;

[0081] When p≤p th When the illegal behavior is classified as low risk;

[0082] When p th <p≤p th When +Δp, it is classified as a medium-risk illegal behavior level;

[0083] When p>p th +Δp, it is classified as a high-risk illegal behavior level, where Δp is the preset risk level interval.

[0084] By integrating functional modules for ship identification and classification, behavior supervision and forecasting, it is possible to quickly and accurately identify and classify ships entering the fishing grounds, effectively improving the efficiency of fishery supervision. At the same time, sensors integrated on ships and fishery patrol boats are used to monitor fishing activity data in real time, and the compliance of fishing behavior is accurately assessed through the fishing intensity assessment model, providing timely and accurate supervision information and law enforcement basis for fishery departments, effectively avoiding overfishing and illegal fishing, and protecting the sustainable use of fishery resources. In addition, through AI image recognition technology and illegal fishing behavior prediction models, the behavior patterns of non-fishing vessels can be analyzed, and potential illegal fishing behavior can be warned, further strengthening the supervision of fisheries.

[0085] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technical personnel in this field for each set of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0086] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by those skilled in the art according to actual conditions.

[0087] In the two embodiments provided in this application, it should be understood that the disclosed devices and systems can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the module division is merely a logical function division. In actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not implemented. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interface, and the indirect coupling or communication connection of devices or modules may be electrical, mechanical or other forms.

[0088] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An AI-based intelligent forecasting system for South China Sea fishing grounds, characterized by: It includes: A ship identification and classification unit and a behavior supervision unit; The ship identification and classification unit is used to identify ships entering the South China Sea fishing ground through multi-source data fusion technology, verify the consistency of their appearance features, fishing license numbers and ship grade information, obtain the fishing legality of the ships, and classify the ships into fishing ships and non-fishing ships according to the fishing legality results of the ships, and send the classification results to the behavior supervision unit; The behavior supervision unit includes a fishing vessel supervision module and a non-fishing vessel supervision module. The fishing vessel supervision module is used to monitor the fishing activity data of the fishing vessel in real time through sensors integrated on the vessel and the fishery patrol boat, establish a fishing intensity assessment model based on the fishing activity data, obtain the current fishing intensity F, and calculate the fishing intensity according to the preset fishing limit F. q , the current fishing intensity F and the fishing limit F q Compare and evaluate the compliance of current fishing activities. Based on the compliance assessment results and the preset grading standards, the fishing degree is divided into compliance, minor violation, moderate violation, and severe violation levels, and the fishing degree level is sent to the forecasting unit; The non-fishing vessel supervision module is used to obtain image data through surveillance cameras and drone equipment installed around the fishing grounds, combine AI image recognition technology, and analyze the behavior patterns of non-fishing vessels. It combines historical data with real-time intelligence to establish an illegal fishing behavior prediction model, and obtains a preset behavior threshold p based on historical illegal fishing behavior data and law enforcement experience. th , evaluate the behavior of non-fishing vessels, and divide the evaluation results into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior and high-risk illegal behavior levels and send them to the forecast unit.

2. The AI-based South China Sea fishing ground intelligent forecasting system according to claim 1 is characterized in that: It also includes a forecasting unit, which is used to obtain the fishing degree level of fishing ships and the behavior level of non-fishing ships, generate corresponding early warning information according to the fishing degree level and the behavior level, push the early warning information and related ship information to the fishery administration department and fishing enterprises, and automatically allocate processing tasks to fishery law enforcement officers and managers of fishing enterprises according to the urgency and type of the early warning information.

3. The AI-based South China Sea fishery intelligent forecasting system according to claim 2 is characterized in that: The specific process of establishing a fishing intensity assessment model is as follows: S101. Obtain fishing activity data, where the fishing activity data includes fishing duration T, number of fishing gears N, fishing area A and catch per unit time M; S102. Clean, denoise and standardize the obtained fishing activity data; S103: Based on the fishing activity data, a fishing intensity assessment index system is constructed. The fishing intensity assessment index includes the fishing duration density D T , fishing net density D N and fishing efficiency E; Among them, the fishing time density D T is the fishing time per unit fishing area, and the calculation formula is: Fishing gear density D N is the number of fishing gear per unit fishing area, and the calculation formula is: Fishing efficiency E is the amount of fish caught per unit fishing gear per unit time, and the calculation formula is: S104. Construct a judgment matrix, determine the relative importance of each indicator through expert scoring, calculate the maximum eigenvalue and corresponding eigenvector of the judgment matrix, perform consistency test, normalize the eigenvector, and obtain the weight w of each indicator. T 、w N and w E , and satisfy w T +w N +w E =1; S105. Based on each indicator and its weight, a fishing intensity assessment model is established to calculate the current fishing intensity F. The calculation formula is: F = w T *D T +w N *D N +w E *E; S$.

06. Use historical fishing data for model verification, and optimize the model by calculating the error between the model prediction value and the actual value.

4. The AI-based South China Sea fishery intelligent forecasting system according to claim 3 is characterized in that: The specific process of dividing the fishing degree is as follows: S201, based on the current fishing intensity F and the preset fishing limit F q , calculate the fishing intensity deviation rate R, the calculation formula is: S202. According to the actual supervision requirements and experience, set the thresholds for dividing the fishing degree levels, including a mild violation threshold R1, a moderate violation threshold R2 and a severe violation threshold R3, and satisfy 0% ≤ R1 < R2 < R3 ≤ 100%; S203. According to the fishing intensity deviation rate R and the set thresholds for dividing the fishing degree levels, divide the fishing degree into the following levels: When R=0%, that is, F=F q , the fishing degree level is compliant; When 0% < R ≤ R1, the fishing degree level is mild violation; When R1 < R ≤ R2, the fishing degree level is moderate violation; When R > R3, the fishing degree level is severe violation.

5. The AI-based South China Sea fishery intelligent forecasting system according to claim 4 is characterized in that: The specific process of establishing an illegal fishing behavior prediction model is as follows: S301, obtain image data and form image data set D img ={I1,I2,...,I n }, where n is the total number of images, and historical illegal fishing behavior data and real-time intelligence are collected to form the non-image dataset D non ={X1,X2,...,X m }, where m is the number of non-image data entries. The image data and non-image data are associated and integrated to form a i Match the corresponding non-image data X j , forming a comprehensive data set D = {(I i ,X j )}; S302, using pre-trained convolutional neural network to image data I i Perform feature extraction to obtain the image feature vector α i , use one-hot encoding to encode non-image data X j The categorical variables in are transformed and the numerical variables are normalized to obtain the non-image feature vector β j , the image feature vector α i and the non-image feature vector β j Splice and get the comprehensive feature vector Z ij =|α i ,β j |; S303, select the random forest algorithm as the illegal fishing behavior prediction model, and transform the comprehensive feature vector set Z={(Z ij ,Y ij )} is input into the model for training, where Y ij is a label, 0 indicates non-fishing behavior, and 1 indicates illegal fishing behavior. The classification accuracy L is used as the evaluation indicator for model training, and its calculation formula is: Where k is the total number of samples; Y pr The label of the kth sample predicted by the model; Y pr is the true label of the k-th sample; I(·) is an indicator function, which takes 1 when the condition in the parentheses holds, and 0 otherwise.

6. The AI-based South China Sea fishery intelligent forecasting system according to claim 5 is characterized in that: The specific process of dividing the behavior level is as follows: According to the model prediction probability p and the behavior threshold p th Based on the relationship between the two, the behavior levels of non-fishing vessels are divided into non-fishing behavior, low-risk illegal behavior, medium-risk illegal behavior and high-risk illegal behavior. The specific classification rules are as follows: When p <p th -Δp sa , it is divided into the non-fishing behavior level, where Δp sa is the preset safety probability interval; When p≤p th When the illegal behavior is classified as low risk; When p th <p≤p th When +Δp, it is classified as a medium-risk illegal behavior level; When p>p th +Δp, it is classified as a high-risk illegal behavior level, where Δp is the preset risk level interval.

Images