Intelligent security robot operation and data center interaction data processing system based on AI large model

Abstract

The invention discloses an intelligent security robot operation and data center interaction data processing system based on an AI large model, and belongs to the technical field of interaction data processing. Three layers of key indexes of hardware, software and data interaction are integrated through an extended state vector, multi-dimensional state fusion is implemented, and one-sidedness of single-dimensional monitoring can be avoided; the historical data trend and the real-time state mutation are balanced by using the time-varying memory characteristic of the fractional differential operator; through cooperative calculation of the edge nodes and the data center, low-delay state perception under wide-area deployment is realized, and the real-time requirement of the security robot is met; through the differential design of the positive error direction convergence index and the negative error direction convergence index, the hardware fault convergence time can be effectively shortened, and the software vulnerability overshoot is reduced; by configuring an observer gain matrix, it can be ensured that observation errors converge within a controllable range, and full-period compensation before, during and after disturbance is achieved in combination with the fault prediction probability.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of interactive data processing, in particular to an intelligent security robot operation and data center interactive data processing system based on an AI large model. BACKGROUND

[0002] The intelligent security robot operation and data center interaction data processing refers to a complete data flow and processing mechanism in which a patrol robot, a monitoring robot and the like collect, transmit, analyze and feedback data during task execution in an intelligent security system, and efficiently cooperate with a background data center, and the core goal is to realize real-time sensing, intelligent analysis, rapid response and continuous optimization.

[0003] The existing technical solution has defects in system stability. The security system needs to be stably operated for a long time, but technical bottlenecks and design defects may cause system failure, such as software logic vulnerabilities or hardware reliability problems. A one-time system crash may cause serious consequences in the security field. The technical bottlenecks and design defects cannot be actively monitored and optimized from different dimensions, resulting in poor active monitoring and optimization effect of system stability. SUMMARY

[0004] The purpose of the present application is to provide an intelligent security robot operation and data center interactive data processing system based on an AI large model, which solves the technical problem that the existing solution cannot actively monitor and optimize the technical bottlenecks and design defects from different dimensions, resulting in poor active monitoring and optimization effect of system stability.

[0005] The purpose of the present application can be realized by the following technical solution: A system health and disturbance trend prediction processing module: a distributed collaborative state observation model is constructed, an extended state vector integrating hardware health degree, software logic node state and data interaction quality is defined, time-varying data characteristics are dynamically weighted through a fractional order differential operator, and system health degree index and disturbance trend prediction value are output in real time; A fault prediction and disturbance compensation analysis module: based on the defined extended state vector, a predictive disturbance observer driven by an asymmetric barrier function is designed, an observer gain matrix is configured, and the asymmetric convergence characteristics of hardware faults and software vulnerabilities are combined to output fault prediction probability and disturbance compensation amount; A system stability monitoring and optimization module: according to the fault prediction probability and disturbance compensation amount obtained by processing, a fractional order sliding mode adaptive interaction strategy is constructed, time-varying scaling factors and redundant node switching thresholds for data transmission are dynamically adjusted, high-frequency data jitter is suppressed through an integral sliding mode surface, and the communication bandwidth allocation of the security robot-data center is smoothly adjusted by using a nonlinear continuous operator, so as to realize the active monitoring and optimization of system stability.

[0006] Preferably, based on the collected hardware health data, software logic node status data, and data interaction quality data, the extended state vector is defined as follows: j, l, and v are all element indices in the corresponding vectors; T is the transpose operator; where, For hardware health vectors; This represents the state vector of a software logic node. This is the data interaction quality vector.

[0007] Preferably, a fractional differential operator is used to dynamically weight the time-varying data features, enhancing the influence of historical data on the current state. The relevant expression is as follows: ;in, For fractional derivatives, ; It is a fractional differential operator; For the Gamma function, t is the integral variable; t is the time variable; When performing dynamic weighted fusion, the expression involved is: ;in, It is a time-varying weight matrix; These are weights for hardware health, software logic, and data interaction, respectively.

[0008] Preferably, the extended state vector, fractional differential operator, and time-varying weight matrix are simultaneously calculated to obtain the system health index Health(t) and the disturbance trend prediction value Trend(t), and the relevant expressions are as follows: ; ;in, It is a fractional differential operator; The time derivative of the weight matrix; When Health(t) < 0.6 and Trend(t) < -0.1, the potential perturbation trend is determined to be significant, triggering the predictive observer to run.

[0009] Preferably, based on the defined extended state vector, the error dynamic equation of the predictive perturbation observer is constructed: ;in, Let A be the observation error vector; let A be the system matrix and B be the input matrix. The lumped disturbance vector; This is the estimated value of the disturbance.

[0010] Preferably, considering the asymmetric nature of hardware failures and software vulnerabilities, such as sensor drift, an asymmetric barrier function is introduced to reconstruct the observation error. The relevant expression is as follows: ;in, These are the convergence indices for the positive error direction and the negative error direction, respectively; sign() is the sign function. The errors of each component of the extended state vector are nonlinearly weighted. Both are observation error vectors The amount.

[0011] Preferably, based on the observation error vector and the disturbance estimate, the fault prediction probability is output through a nonlinear mapping function; when the fault prediction probability is greater than 0.7, the compensation intensity is dynamically adjusted.

[0012] Preferably, based on the constructed sliding surface, a control law containing integral and nonlinear damping terms is designed, and the time-varying scaling factor for data transmission is dynamically adjusted.

[0013] Preferably, a switching threshold is designed based on the sliding surface error and the disturbance compensation residual; when the communication link error is greater than the switching threshold, the redundant node switching is triggered.

[0014] Preferably, a nonlinear continuous operator is used. The expression involved in smoothly adjusting the communication bandwidth allocation between the security robot and the data center is as follows: ;in, Real-time communication bandwidth allocation for security robots and data centers; Configure the initial communication bandwidth. It is a time-varying smoothing factor; It is a sliding surface.

[0015] Compared to existing solutions, the beneficial effects achieved by this invention are: This invention integrates key indicators from three layers—hardware, software, and data interaction—by extending state vectors to implement multi-dimensional state fusion, thus avoiding the one-sidedness of single-dimensional monitoring. By utilizing the time-varying memory characteristics of fractional differential operators, it balances historical data trends with real-time state abrupt changes, effectively improving the lead time for disturbance prediction compared to traditional integer-order filtering. Through collaborative computing between edge nodes and data centers, it achieves low-latency state perception in wide-area deployments, meeting the real-time requirements of security robots.

[0016] This invention, through the differentiated design of convergence exponents in the positive and negative error directions, can effectively shorten the convergence time of hardware faults and reduce the overshoot of software vulnerabilities, thus solving the problem of unbalanced response of traditional symmetric observers to asymmetric faults. By configuring the observer gain matrix, it can ensure that the observation error converges within a controllable range, and combined with the fault prediction probability, it can achieve full-cycle compensation before, during, and after the disturbance occurs. Compared with passive fault tolerance mechanisms, it can effectively improve system reliability. By reusing the fractional-order differential state vector and the time-varying weight matrix, the disturbance estimation is deeply correlated with the system health, which can avoid the compensation lag caused by the disconnect between independent observers and state assessment.

[0017] The fractional sliding surface in this invention reduces high-frequency noise sensitivity through fractional differential operators and nonlinear continuous operators, which can suppress high-frequency jitter; the time-varying scaling factor and nonlinear operators work together to effectively improve communication bandwidth utilization and further improve packet loss rate control; the redundant node switching threshold is dynamically adjusted based on real-time error, which can shorten fault response time and improve system stability margin. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart illustrating the operation of the intelligent security robot based on an AI large model and the data processing system for interaction with the data center, as described in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this invention is a data processing system for intelligent security robot operation and data center interaction based on an AI large-scale model, comprising: The system health and disturbance trend prediction processing module constructs a distributed collaborative state observation model, defines an extended state vector that integrates hardware health, software logic node status, and data interaction quality, and dynamically weights time-varying data features using fractional-order differential operators to output system health indicators and disturbance trend prediction values ​​in real time. Specific steps include: In the distributed network of intelligent security robots and data centers, N sensing nodes are deployed, including robot-end embedded modules and data center edge computing units, where N is a positive integer, to collect hardware health data, software logic node status data and data interaction quality data in real time. Among them, hardware health data is collected by sensors, including the CPU temperature drift rate of the robot end, the memory ECC error count, the voltage fluctuation of the communication module, and the hard disk IO response latency of the data center server. Software logic node status data is collected through embedded logs, including the abnormal frequency of function call stack depth of robot-side path planning algorithm, data center database transaction rollback rate, and number of API timeout errors. Data interaction quality data is collected through network probes, including end-to-end transmission packet loss rate, TCP retransmission count, data frame verification failure rate, and transmission latency jitter. All data is synchronized to the distributed collaborative computing layer via 5G edge nodes, and the sampling period is adaptively adjusted according to the interaction intensity; the sampling period is 50ms during periods of high interaction intensity and 2s during periods of low activity. Based on the collected hardware health data, software logic node status data, and data interaction quality data, the extended state vector is defined as follows: j, l, and v are all element indices in the corresponding vectors; T is the transpose operator; in, For hardware health vectors, , These are CPU temperature drift rate, memory ECC error count, communication module voltage fluctuation, and hard disk I / O response latency; This is the state vector of the software logic node. , These are the frequency of abnormal function call stack depth, database transaction rollback rate, and number of API timeout errors, respectively. For data interaction quality vector, , These are packet loss rate, TCP retransmission count, data frame check failure rate, and transmission delay jitter, respectively. The time stamp alignment algorithm unifies asynchronously acquired multi-source data to the same time axis with an error of ≤1ms, and Z-score standardization is used to eliminate dimensional differences. The implemented time stamp alignment algorithm and Z-score standardization are both existing conventional technical solutions, and the specific implementation steps are not described here. Fractional differential operators are used to dynamically weight the features of time-varying data, enhancing the influence of historical data on the current state. The relevant expression is as follows: ;in, For fractional derivatives, This achieves short-term sensitivity and long-term smoothness through dynamic adjustment: when the system does not detect potential disturbances, The value is 0.3, focusing on historical trends; when the system detects a potential disturbance, The value is set to 0.8 to enhance real-time response; the detection of potential disturbances is a conventional technical solution, and the specific implementation steps will not be elaborated here. It is a fractional differential operator; For the Gamma function, is the integration variable, specifically a historical time point; t is the time variable. When performing dynamic weighted fusion, the expression involved is: ;in, It is a time-varying weight matrix; These are weights for hardware health, software logic, and data interaction, respectively. , is the first weighting coefficient, with a default value of 0.5; e is the natural constant; Let be the norm of the hardware health vector; , This is the second weighting coefficient, with a default value of 2; ; The diagonal matrix structure of W(t) enables the decoupling and weighting of hardware, software, and data interaction states, which can avoid excessive interference of single-dimensional anomalies on the overall health.

[0022] By simultaneously solving the extended state vector, the fractional-order differential operator, and the time-varying weight matrix, we obtain the system health index Health(t) and the predicted disturbance trend value Trend(t). The relevant expressions are as follows: ; ;in, It is a fractional differential operator; The time derivative of the weight matrix; When Health(t) < 0.6 and Trend(t) < -0.1, the potential perturbation trend is determined to be significant, triggering the predictive observer to run.

[0023] In this embodiment of the invention, by extending the state vector to integrate key indicators from three layers—hardware, software, and data interaction—multi-dimensional state fusion is implemented, which avoids the one-sidedness of single-dimensional monitoring. By utilizing the time-varying memory characteristics of fractional differential operators, the trends of historical data and real-time state changes are balanced, which can effectively improve the lead time for disturbance prediction compared to traditional integer-order filtering. Through collaborative computing between edge nodes and data centers, low-latency state perception under wide-area deployment is achieved, meeting the real-time requirements of security robots.

[0024] Fault prediction and disturbance compensation analysis module: Based on the defined extended state vector, an asymmetric barrier function-driven predictive disturbance observer is designed, the observer gain matrix is ​​configured, and the asymmetric convergence characteristics of hardware faults and software vulnerabilities are combined to output the fault prediction probability and disturbance compensation amount; the specific steps include: Based on the defined extended state vector, the error dynamic equation of the predictive perturbation observer is constructed: ;in, The time derivative of the error vector; For the observation error vector, , A is the estimated state vector output by the observer, which is calculated by a high-order finite-time homogeneous observer and contains real-time estimates of each component in X(t); A is the system state matrix, which is the state space matrix derived from the robot dynamics model and contains the coupling relationships of state variables such as position, velocity, and acceleration; B is the input matrix. This is a lumped disturbance vector, containing hardware faults such as sensor drift and software vulnerabilities such as data verification errors. This is the estimated value of the disturbance; Here, the observer order r=3 is set, corresponding to the three-dimensional components of the extended state vector, and the desired characteristic polynomial is constructed: ;in, Let be the expected characteristic polynomial; s be a complex variable; All are roots of the characteristic polynomial, with default values ​​of 50, 60, and 70 respectively; And, through the formula Calculate the gain matrix K. Configure functions for poles, The desired closed-loop pole vector is used to ensure the observer's dynamic performance meets the requirements. ; To adjust the duration; To address the asymmetric nature of hardware failures and software vulnerabilities—hardware failures such as sensor drift requiring rapid error convergence, and software vulnerabilities such as logic anomalies requiring smooth error convergence—an asymmetric barrier function is introduced to reconstruct the observation error. The relevant expression is as follows: ;in, These are the convergence indices for the positive and negative error directions, respectively, with default values ​​of 1.5 and 0.6; sign() is the sign function, ensuring that the error direction is consistent with the barrier function; The errors of each component of the extended state vector are nonlinearly weighted. Both are observation error vectors The amount; Combining the dynamic weighted results of fractional derivatives The expression involved in constructing the perturbation estimator is as follows: ; Based on observation error vector and disturbance estimates The fault prediction probability is output through a nonlinear mapping function. The expression involved is: ;in, This represents the steepness of the probability curve. The larger the value, the more significant the jump in failure probability. The default value is 0.3. The disturbance threshold is determined by the maximum allowable disturbance amplitude of the system. It is set to 0.5 for hardware failures and 0.7 for software vulnerabilities. The 2-norm of the perturbation estimate; When the probability of fault prediction When the value is greater than 0.7, the compensation intensity is dynamically adjusted, and the relevant expression is: ;in, M is the disturbance compensation amount; M is the system inertia matrix, referencing the hardware inertia parameters in the dynamic model, such as the transmission delay inertia of the robot communication module.

[0025] In this embodiment of the invention, by designing the convergence exponents in the positive and negative error directions differently, the convergence time of hardware faults can be effectively shortened and the overshoot of software vulnerabilities can be reduced, solving the problem of unbalanced response of traditional symmetric observers to asymmetric faults. By configuring the observer gain matrix, the observation error can be ensured to converge within a controllable range. Combined with the fault prediction probability, full-cycle compensation before, during, and after the disturbance occurs can be achieved, which can effectively improve system reliability compared to passive fault tolerance mechanisms. By reusing the fractional-order differential state vector and the time-varying weight matrix, the disturbance estimation is deeply correlated with the system health, avoiding the compensation lag caused by the disconnect between independent observers and state assessment.

[0026] The system stability monitoring and optimization module: Based on the obtained fault prediction probability and disturbance compensation amount, a fractional sliding mode adaptive interaction strategy is constructed to dynamically adjust the time-varying scaling factor of data transmission and the redundant node switching threshold. High-frequency data jitter is suppressed through an integral sliding surface, and the communication bandwidth allocation between the security robot and the data center is smoothly adjusted using nonlinear continuous operators, thereby achieving proactive monitoring and optimization of system stability. Specific steps include: Based on the fault prediction probability and disturbance compensation obtained from the processing, an integral fractional-order sliding surface is designed: ;in, is the multi-scale gain matrix; i is the component index, i=1,2,…,m; m is the total number of components; Let k be the i-th observation error component; k is the backstep recursion level. This is the integral gain matrix; Nonlinear weighting of the integral term error; For the power parameter of the integral term; For integration operators, it represents the integral operator with respect to the time variable. Integrate from 0 to t to accumulate historical error information; Based on sliding surface Design a control law containing integral and nonlinear damping terms, and dynamically adjust the time-varying scaling factor for data transmission. The expression involved is: Where u is the generalized control input, i.e. the actual control quantity executed by the system; For robustness gain; Nonlinear weighting of the integral term error; It is a nonlinear continuous operator; ;in, For the attenuation rate parameter, ; For power-order parameters, ; And, based on the slip surface error and disturbance compensation residual , Design a switching threshold to represent the total actual disturbance of the system. : ;in, This is the initial switching threshold; This is the threshold adjustment index, with a value range of (0,1); Let the sliding surface norm be denoted as . The residual norm is used to compensate for the disturbance. The initial sliding surface norm; The norm of the residual after initial disturbance compensation; When the communication link error is greater than the handover threshold When this occurs, redundant node switching is triggered; Using nonlinear continuous operators The expression involved in smoothly adjusting the communication bandwidth allocation between the security robot and the data center is as follows: ;in, Real-time communication bandwidth allocation for security robots and data centers; Configure the initial communication bandwidth. It is a time-varying smoothing factor. , This is the initial smoothing value; This is the attenuation rate parameter; The steady-state smoothing value is used; the time-varying smoothing factor decays dynamically, enabling the bandwidth to transition from initial high throughput to steady-state low jitter.

[0027] In this embodiment of the invention, the fractional sliding surface reduces the sensitivity to high-frequency noise through fractional differential operators and nonlinear continuous operators, thereby suppressing high-frequency jitter; the time-varying scaling factor and nonlinear operators work together to effectively improve the utilization of communication bandwidth, further enhancing the packet loss rate control effect; the redundant node switching threshold is dynamically adjusted based on real-time error, which can shorten the fault response time and improve the system stability margin.

[0028] In the several embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0029] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0030] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0031] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0032] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A data processing system for intelligent security robot operation and data center interaction based on AI large-scale model, characterized in that, include: System health and disturbance trend prediction processing module: Constructs a distributed collaborative state observation model, defines an extended state vector that integrates hardware health, software logic node status and data interaction quality, and dynamically weights time-varying data features through fractional differential operators to output system health indicators and disturbance trend prediction values ​​in real time; Fault prediction and disturbance compensation analysis module: Based on the defined extended state vector, an asymmetric barrier function-driven predictive disturbance observer is designed, the observer gain matrix is ​​configured, and the asymmetric convergence characteristics of hardware faults and software vulnerabilities are combined to output the fault prediction probability and disturbance compensation amount. System stability monitoring and optimization module: Based on the fault prediction probability and disturbance compensation amount obtained through processing, a fractional sliding mode adaptive interaction strategy is constructed to dynamically adjust the time-varying scaling factor of data transmission and the switching threshold of redundant nodes. High-frequency data jitter is suppressed by integral sliding surface, and the communication bandwidth allocation between the security robot and the data center is smoothly adjusted by nonlinear continuous operators to achieve proactive monitoring and optimization of system stability.

2. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 1, characterized in that, Based on the collected hardware health data, software logic node status data, and data interaction quality data, the extended state vector is defined as follows: j, l, and v are all element indices in the corresponding vectors; T is the transpose operator; where, For hardware health vectors; This represents the state vector of a software logic node. This is the data interaction quality vector.

3. The intelligent security robot operation and data center interaction data processing system based on AI large model according to claim 2, characterized in that, Fractional differential operators are used to dynamically weight the features of time-varying data, enhancing the influence of historical data on the current state. The relevant expression is as follows: ;in, For fractional derivatives, ; It is a fractional differential operator; For the Gamma function, t is the integral variable; t is the time variable; When performing dynamic weighted fusion, the expression involved is: ;in, It is a time-varying weight matrix; These are weights for hardware health, software logic, and data interaction, respectively.

4. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 3, characterized in that, By simultaneously solving the extended state vector, the fractional-order differential operator, and the time-varying weight matrix, we obtain the system health index Health(t) and the predicted disturbance trend value Trend(t). The relevant expressions are as follows: ; ;in, It is a fractional differential operator; The time derivative of the weight matrix; When Health(t) < 0.6 and Trend(t) < -0.1, the potential perturbation trend is determined to be significant, triggering the predictive observer to run.

5. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 4, characterized in that, Based on the defined extended state vector, the error dynamic equation of the predictive perturbation observer is constructed: ;in, Let A be the observation error vector; let A be the system matrix and B be the input matrix. The lumped disturbance vector; This is the estimated value of the disturbance.

6. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 5, characterized in that, To address the asymmetric nature of hardware failures and software vulnerabilities, an asymmetric barrier function is introduced to reconstruct the observation error. The relevant expression is as follows: ;in, These are the convergence indices for the positive error direction and the negative error direction, respectively; sign() is the sign function. The errors of each component of the extended state vector are nonlinearly weighted. Both are observation error vectors The amount.

7. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 6, characterized in that, Based on the observation error vector and the disturbance estimate, the fault prediction probability is output through a nonlinear mapping function; when the fault prediction probability is greater than 0.7, the compensation intensity is dynamically adjusted.

8. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 7, characterized in that, Based on the constructed sliding surface, a control law containing integral and nonlinear damping terms is designed, and the time-varying scaling factor for data transmission is dynamically adjusted.

9. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 8, characterized in that, Based on the sliding surface error and disturbance compensation residual, a switching threshold is designed; when the communication link error exceeds the switching threshold, redundant node switching is triggered.

10. The AI-based large-scale model-based intelligent security robot operation and data center interactive data processing system according to claim 9, characterized in that, Using nonlinear continuous operators The expression involved in smoothly adjusting the communication bandwidth allocation between the security robot and the data center is as follows: ;in, Real-time communication bandwidth allocation for security robots and data centers; Configure the initial communication bandwidth. It is a time-varying smoothing factor; It is a sliding surface.

Images