Intelligent prison correction system and method based on data and model middle platform

Abstract

The invention discloses a prison intelligent correction method based on data and a model middle platform, and relates to the technical field of prison management, and the method comprises the following steps: deploying a plurality of monitoring devices, and carrying out the real-time collection of the behavior, position and physiological data of a prison in combination with a standard protocol; a data management middle platform is used for uniformly storing multi-source heterogeneous data from different sources and in different formats, providing data cleaning, format standardization and structured conversion capabilities, supporting multi-dimensional correlation analysis, mining potential relationships among the data, setting a strict authority management and data auditing mechanism, and improving the data auditing efficiency. A plurality of AI models are integrated in a model management platform, and when the precision of the model is reduced or the environment is changed, the model is automatically or manually triggered to be retrained. Through real-time monitoring of multi-source data and intelligent identification of AI model scheduling, potential safety risks can be found as early as possible, and precautionary measures can be taken in time.

Description

Technical Field

[0001] The present invention relates to the technical field of prison management, and in particular to a prison intelligent correction system and method based on a data and model middle platform. Background Art

[0002] A prison is an organ for executing national penalties. It is a place for detaining criminals sentenced to fixed-term imprisonment, life imprisonment, and death penalty with a two-year suspension of execution by the people's court. In order to facilitate the management of criminals, some prisons have deployed basic digital management systems, such as access control systems, face recognition systems, and monitoring video storage and playback.

[0003] However, there is often a lack of data interconnection and interoperability between the above systems, and each module is relatively independent; there is a lack of a mechanism for unified management and comprehensive analysis of massive data, making it difficult to deeply mine the behavior characteristics and potential risks of detainees, unable to conduct refined analysis and tracking of the behavior and psychological conditions of each detainee, and it is difficult for prison managers to adjust correction measures in a timely manner, resulting in insufficient personalization and precision. Therefore, there is room for improvement. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a prison intelligent correction system and method based on a data and model middle platform. Its advantage lies in being able to detect potential security risks early and take preventive measures in a timely manner.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] A prison intelligent correction method based on a data and model middle platform, comprising the following steps:

[0007] Step 1: Deploy a variety of monitoring devices, and collect real-time data on the behavior, location, and physiological data of detainees in combination with standard protocols; use the data management middle platform to uniformly store multi-source heterogeneous data from different sources and in various formats, provide data cleaning, format standardization, and structured conversion capabilities, and at the same time support multi-dimensional correlation analysis, mine the potential relationships between data, and set strict permission management and data audit mechanisms;

[0008] Step 2: Integrate a variety of AI models in the model management middle platform, and support automatic or manual triggering of model retraining when the model accuracy drops or the environment changes;

[0009] Step 3: Perform multi-level processing and analysis on the collected real-time data. First, denoise, normalize, and detect outliers through the data preprocessing module, and then conduct in-depth analysis based on the behavior recognition algorithm. The system uses a rule engine and machine learning models to identify abnormal behaviors and calculate key metrics. When the score exceeds the set threshold or triggers a specific pattern matching rule, the system automatically generates an alarm message and links to relevant management personnel to take corresponding measures. In high-risk scenarios, the system pushes alarm messages to relevant staff and links to security measures for rapid intervention.

[0010] Step 4: Based on the comprehensive information of the detainees, including age, case background, type of crime, length of sentence, psychological assessment results, behavior performance, contraband records, social relations, and recidivism risk assessment, systematically analyze their correction needs. Combine artificial intelligence analysis and professional correction models to accurately evaluate the psychological state, educational needs, adaptability to vocational skills training, and social integration potential of each detainee, and accordingly develop personalized correction plans. The system continuously tracks the implementation of the correction plans and records the intervention effects and behavior changes of the detainees to provide a basis for subsequent model iteration and decision-making optimization.

[0011] Step 5: In the visual command platform, comprehensively present the overall situation of the prison, including the operation status of the security system, the situation of prison areas, and key information on the distribution of on-duty police forces. At the same time, dynamically display the real-time distribution, detention situation, personnel flow trajectories, and monitoring of key personnel of the detainees. The platform integrates a real-time alarm system to promptly push information on emergencies, abnormal behavior detection, and security warnings. In addition, based on data analysis and intelligent models, the system can also provide behavior pattern analysis, risk prediction, and security assessment to assist management personnel in making efficient decisions.

[0012] The present invention is further configured such that the monitoring devices include cameras, sensors, access control systems, smart bracelets, and ankle rings. The cameras are used for video monitoring and behavior recognition, the sensors monitor environmental parameters and individual states, the access control system records personnel entry and exit information, and the smart bracelets and ankle rings collect physiological movement data such as heart rate, blood oxygen, and activity trajectories.

[0013] The present invention is further configured such that the standard protocol is at least one of GB28181, Onvif, and Modbus.

[0014] The present invention is further configured such that the AI model includes a risk assessment model based on multi-feature synthesis and a psychological state prediction model. The psychological state prediction model dynamically updates the prediction results of the psychological state, and the formula is: Where H is the hypothesized state and E is the new evidence.

[0015] The present invention is further configured such that the correction plan includes hierarchical legal education, vocational skills training, psychological intervention measures, behavior correction courses, and personalized assistance plans, and the psychological intervention measures include psychological counseling and group counseling.

[0016] The present invention is further configured such that the visual command platform simultaneously supports data drilling, trend analysis, business indicator monitoring, and multi-level role permission management.

[0017] The present invention is further configured such that the behavior recognition algorithm is a spatio-temporal behavior detection algorithm based on deep learning for skeleton behavior recognition, and the formula of the spatio-temporal graph convolution algorithm is: where B(vi) is the domain of node vi, and Z ij is a normalization factor, and W is a weight function.

[0018] A prison intelligent correction system based on a data and model middle platform, the prison intelligent correction system applies the above-mentioned prison intelligent correction method, and the prison intelligent correction system includes a data collection and management module, a model management module, a real-time risk assessment and alarm module, a personalized correction and assessment module, and a visual decision support module. The data collection and management module and the model management module are communicatively connected to the real-time risk assessment and alarm module, and the real-time risk assessment and alarm module is communicatively connected to the real-time risk assessment and alarm module.

[0019] A computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor, when executing the computer program, implements the steps of the above method.

[0020] A computer-readable storage medium stores a computer program thereon. The computer program, when executed by a processor, implements the steps of the above method.

[0021] The beneficial effects of the present invention are as follows:

[0022] 1. Enhance the security prevention ability of the prison

[0023] Through real-time monitoring of multi-source data and intelligent identification of AI model scheduling, potential security risks can be detected early, and preventive measures can be taken in a timely manner.

[0024] 2. Improve the accuracy and efficiency of correction work

[0025] The system can formulate differentiated correction plans for different groups of people, avoiding the "one-size-fits-all" management method; and automatically track the correction effect and continuously optimize the intervention strategy.

[0026] 3. Realize the centralized management and application of data and models

[0027] By combining the data management middleware and the model management middleware, the data processing efficiency and model scheduling speed are greatly improved, realizing the in-depth mining and value transformation of massive prison business data.

[0028] 4. Provide visual decision-making and full-process tracking

[0029] Prison managers can view real-time status, risk warnings, and correction progress in a unified visual platform, and carry out command and dispatch and decision-making more efficiently.

[0030] 5. Improve the stability and scalability of the system

[0031] Adopting a distributed architecture design, it can be flexibly expanded according to the growth of prison scale and data volume, and through a perfect permission management and security mechanism, the security of sensitive data is guaranteed. Description of the Drawings

[0032] Figure 1 It is a schematic structural diagram of the workflow of a prison intelligent correction method based on data and model middleware proposed by the present invention. Detailed Embodiments

[0033] The technical solutions of this patent will be further described in detail below in conjunction with the specific embodiments.

[0034] The embodiments of this patent are described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below by referring to the drawings are exemplary and are only used to explain this patent and should not be construed as a limitation of this patent.

[0035] In the description of this patent, it should be understood that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing this patent and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this patent.

[0036] In the description of this patent, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "set" should be understood in a broad sense. For example, it can be fixedly connected and set, or detachably connected and set, or integrally connected and set. For those of ordinary skill in the art, the specific meanings of the above terms in this patent can be understood according to specific circumstances.

[0037] Refer to Figure 1, a prison intelligent correction method based on data and model middle platform, including the following steps:

[0038] Step 1: Deploy a variety of monitoring equipment and collect the behavior, location and physiological data of detainees in real time in combination with standard protocols (GB28181, Onvif or Modbus); use the data management platform to uniformly store multi-source heterogeneous data from different sources and formats, provide data cleaning, format standardization and structured conversion capabilities, support multi-dimensional correlation analysis, explore potential relationships between data, ensure data consistency, accuracy and availability, and provide a solid data foundation for business decision-making and intelligent analysis; and set up strict authority management and data audit mechanisms;

[0039] Monitoring equipment includes cameras, sensors, access control systems, smart bracelets and anklets. Cameras are used for video surveillance and behavior recognition, sensors monitor environmental parameters and individual status, access control systems record people's entry and exit information, and smart bracelets and anklets collect heart rate, blood oxygen and activity trajectory physiological motion data to ensure the comprehensiveness and accuracy of data acquisition, providing reliable support for security management and intelligent analysis.

[0040] Step 2: Integrate multiple AI models in the model management center. When the model accuracy decreases or the environment changes, it supports automatic or manual triggering of model retraining to maintain the effectiveness and reliability of the model.

[0041] Step 3: Multi-level processing and analysis of the collected real-time data. First, the data preprocessing module is used for denoising, normalization and outlier detection, and then in-depth analysis is performed based on the behavior recognition algorithm. The system uses the rule engine and machine learning model to identify abnormal behavior and calculate key indicators. When the score exceeds the set threshold or triggers a specific pattern matching rule, the system automatically generates an alarm message and links the management personnel to take corresponding measures. In high-risk scenarios, the system pushes alarm information to relevant staff and links security measures for rapid intervention.

[0042] Step 4: Based on the comprehensive information of the detainees, including age, case background, crime type, length of sentence, psychological assessment results, behavioral performance, violation records, social relations and recidivism risk assessment, systematically analyze their correction needs; combine artificial intelligence analysis and professional correction models to accurately assess the psychological state, educational needs, vocational skills training adaptability and social integration potential of each detainee, and formulate personalized correction plans accordingly; the system continuously tracks the implementation of the correction plan and records the intervention effect and behavioral changes of detainees, providing a basis for subsequent model iteration and decision optimization;

[0043] The correction plan includes hierarchical legal education, vocational skills training, psychological intervention measures, behavior correction courses, and personalized assistance programs. The psychological intervention measures include psychological counseling and group counseling.

[0044] Step 5: In the visual command platform, comprehensively present the overall situation of the prison, including the operation status of the security system, the situation of prison areas, and key information on the distribution of on-duty police forces; at the same time, dynamically display the real-time distribution, detention situation, personnel flow trajectories, and monitoring of key personnel of the detainees; the platform integrates a real-time alarm system to promptly push information on emergencies, abnormal behavior detection, and security warnings; in addition, based on data analysis and intelligent models, the system can also provide behavior pattern analysis, risk prediction, and security assessment to assist managers in making efficient decisions and enhancing the intelligent level of prison security management; the visual command platform also supports data drilling, trend analysis, business indicator monitoring, and multi-level role permission management to facilitate collaborative decision-making by different functional departments.

[0045] In this embodiment, the AI model includes a risk assessment model based on multi-feature integration and a psychological state prediction model. The psychological state prediction model dynamically updates the prediction results of the psychological state, and the formula is: Among them, H is the hypothesized state, and E is the new evidence.

[0046] The behavior recognition algorithm is a spatio-temporal behavior detection algorithm based on deep learning for skeleton behavior recognition. The formula of the spatio-temporal graph convolution algorithm is: Among them, B(vi) is the domain of node vi, Z ij is the normalization factor, and W is the weight function.

[0047] (1) Unified access and management of multi-source monitoring data

[0048] ① Based on the data management middle platform, integrate multi-source data such as video monitoring, access control systems, electronic fences, and wearable devices to achieve centralized storage, real-time analysis, and governance of the data.

[0049] ② Provide functions such as data cleaning, data lineage tracking, and data security control to meet the strict requirements of the judicial industry for data compliance and security.

[0050] (2) Generation and execution of personalized correction plans

[0051] ① Combining the individual portraits, behavior habits, and risk assessment results of the detainees, the system automatically generates personalized correction or education plans.

[0052] ② Through visual decision-making tools, display the risk levels, psychological conditions, and correction progress of the detainees to prison police officers and managers in the form of charts, dashboards, etc., providing a basis for rapid decision-making and intervention.

[0053] (3) Visual Decision-making and Omnidirectional Supervision

[0054] ① Integrate real-time data, model prediction results, and business processes organically in a unified prison management command platform and present them in a multi-dimensional visual manner.

[0055] ② Support both remote and local command and dispatch simultaneously to ensure real-time control and scientific decision-making of the overall prison situation.

[0056] A prison intelligent correction system based on a data and model middle platform, applying the above-mentioned prison intelligent correction method. The prison intelligent correction system includes a data collection and management module, a model management module, a real-time risk assessment and alarm module, a personalized correction and assessment module, and a visual decision-making support module. The data collection and management module and the model management module are communicatively connected to the real-time risk assessment and alarm module, and the real-time risk assessment and alarm module is communicatively connected to the real-time risk assessment and alarm module.

[0057] A computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. It is characterized in that when the processor executes the computer program, the steps of the above method are implemented.

[0058] A computer-readable storage medium, on which a computer program is stored. It is characterized in that when the computer program is executed by the processor, the steps of the above method are implemented.

[0059] As mentioned above, it is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A prison intelligent correction method based on data and model platform, characterized in that: The following steps are involved: Step 1: Deploy a variety of monitoring equipment and collect the behavior, location and physiological data of detainees in real time in combination with standard protocols; use the data management platform to uniformly store multi-source heterogeneous data from different sources and formats, provide data cleaning, format standardization and structured conversion capabilities, support multi-dimensional correlation analysis, explore potential relationships between data, and set up strict permission management and data audit mechanisms; Step 2: Integrate multiple AI models in the model management platform. When the model accuracy decreases or the environment changes, it supports automatic or manual triggering of model retraining. Step 3: Multi-level processing and analysis of the collected real-time data. First, the data preprocessing module is used for denoising, normalization and outlier detection, and then in-depth analysis is performed based on the behavior recognition algorithm. The system uses the rule engine and machine learning model to identify abnormal behavior and calculate key indicators. When the score exceeds the set threshold or triggers a specific pattern matching rule, the system automatically generates an alarm message and links the management personnel to take corresponding measures. In high-risk scenarios, the system pushes alarm information to relevant staff and links security measures for rapid intervention. Step 4: Based on the detainee’s comprehensive information, including age, case background, crime type, length of sentence, psychological assessment results, behavioral performance, violation record, social relationships and recidivism risk assessment, systematically analyze their correction needs; Combining artificial intelligence analysis and professional correction models, the system accurately assesses the psychological state, educational needs, adaptability to vocational skills training, and social integration potential of each detainee, and formulates a personalized correction plan based on this. The system continuously tracks the implementation of the correction plan and records the intervention effect and behavioral changes of detainees, providing a basis for subsequent model iteration and decision optimization. Step 5: The visual command platform comprehensively presents the overall situation of the prison, including the operation status of the security system, the situation of the prison area and the key information of the distribution of the police force on duty; at the same time, it dynamically displays the real-time distribution of detainees, detention conditions, personnel flow trajectories and key personnel monitoring conditions; the platform integrates a real-time alarm system to timely push emergency events, abnormal behavior detection and security warning information; In addition, based on data analysis and intelligent models, the system can also provide behavioral pattern analysis, risk prediction and safety assessment to assist managers in making efficient decisions.

2. According to claim 1, a prison intelligent correction method based on data and model platform is characterized in that: The monitoring equipment includes cameras, sensors, access control systems, smart bracelets and anklets. The cameras are used for video monitoring and behavior recognition, the sensors monitor environmental parameters and individual status, the access control system records personnel entry and exit information, and the smart bracelets and anklets collect heart rate, blood oxygen and activity trajectory physiological motion data.

3. According to claim 2, a prison intelligent correction method based on data and model platform is characterized in that: The standard protocol is at least one of GB28181, Onvif and Modbus.

4. According to claim 1, a prison intelligent correction method based on data and model platform is characterized in that: The AI ​​model includes a risk assessment model based on multi-feature integration and a psychological state prediction model. The psychological state prediction model dynamically updates the prediction results of the psychological state. The formula is: Among them, H is the hypothetical state and E is the newly added evidence.

5. According to claim 1, a prison intelligent correction method based on data and model platform is characterized in that: The correction program includes tiered legal education, vocational skills training, psychological intervention measures, behavior correction courses and personalized assistance plans, and the psychological intervention measures include psychological counseling and group counseling.

6. According to claim 1, a prison intelligent correction method based on data and model platform is characterized in that: The visual command platform also supports data drilling, trend analysis, business indicator monitoring, and multi-level role authority management.

7. According to claim 6, a prison intelligent correction method based on data and model platform is characterized in that: The behavior recognition algorithm is a spatiotemporal behavior detection algorithm based on deep learning, which is used for skeleton behavior recognition. The spatiotemporal graph convolution algorithm formula is: Among them, B(vi) is the domain of node vi, Z ij is the normalization factor, and W is the weight function.

8. A prison intelligent correction system based on data and model platform, characterized in that: The prison intelligent correction system applies the prison intelligent correction method described in any one of claims 1 to 5, and the prison intelligent correction system includes a data acquisition and management module, a model management module, a real-time risk assessment and alarm module, a personalized correction and assessment module and a visual decision support module. The data acquisition and management module and the model management module are communicatively connected to the real-time risk assessment and alarm module, and the real-time risk assessment and alarm module is communicatively connected to the real-time risk assessment and alarm module.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images