Intelligent operation and maintenance management method for security and protection video monitoring system

Abstract

The invention relates to the technical field of security and protection operation and maintenance, in particular to an intelligent operation and maintenance management method for a security and protection video monitoring system, and the method comprises the following steps: collecting three heterogeneous parameters of monitoring unit density, environment complexity and risk entropy, extracting corresponding hierarchical division parameters, and carrying out the normalization processing of the three heterogeneous parameters; constructing a comprehensive scoring model based on the normalized heterogeneous parameters, and outputting hierarchical scores through the scoring model; constructing a strategy priority model based on a hierarchical division result, and dynamically generating an optimal strategy combination through reinforcement learning; constructing a double-layer loop optimization mechanism, and reversely correcting hierarchical division parameters through a strategy execution effect; and verifying the feasibility of hierarchical division and optimal strategy combination. The intelligent operation and maintenance management method effectively solves the problems that in the prior art, heterogeneous data processing capacity is limited, static thresholds and rules lack adaptability, strategy generation and optimization means are single, and a closed-loop feedback mechanism is incomplete.

Description

Technical Field

[0001] The present invention relates to the technical field of security operation and maintenance, and particularly relates to an intelligent operation and maintenance management method for a security video monitoring system. Background Art

[0002] At present, security video monitoring systems are widely used in fields such as urban security, traffic management, and public place monitoring. Traditional operation and maintenance management methods mainly rely on manual inspections, preset static rules, and simple threshold alarms, and have the following deficiencies:

[0003] Limited heterogeneous data processing ability: Traditional methods are difficult to effectively collect and fuse multiple heterogeneous parameters such as monitoring unit density, environmental complexity, and risk entropy value at the same time, resulting in insufficient characterization of complex environments and dynamic risks, and unable to comprehensively reflect potential safety hazards in system operation.

[0004] Lack of adaptability of static thresholds and rules: The monitoring and alarm mechanisms based on fixed thresholds or preset rules cannot respond to the dynamic changes of the environment and risks in real time, and are prone to false alarms or missed alarms, reducing the reliability and practicality of the system.

[0005] Single means of strategy generation and optimization: Traditional operation and maintenance strategies often rely on expert experience or fixed processes, lack intelligent decision support, have inaccurate identification of the risk levels of monitored areas, and cannot achieve dynamic adjustment and continuous optimization of strategies.

[0006] Incomplete closed-loop feedback mechanism: In the prior art, the feedback of the operation and maintenance management system on the execution results often only stays at the monitoring and alarm level, lacking a mechanism for real-time adaptive correction of parameters and strategy models, and it is difficult to achieve long-term efficient and stable operation and maintenance management. Summary of the Invention

[0007] In order to overcome the disadvantages and deficiencies existing in the prior art, the purpose of the present invention is to provide an intelligent operation and maintenance management method for a security video monitoring system, which effectively solves the problems of limited heterogeneous data processing ability, lack of adaptability of static thresholds and rules, single means of strategy generation and optimization, and incomplete closed-loop feedback mechanism in the prior art.

[0008] The present invention is achieved through the following technical solutions:

[0009] In a first aspect, the present invention discloses an intelligent operation and maintenance management method for a security video monitoring system, which includes the following steps:

[0010] By collecting three heterogeneous parameters of monitoring unit density, environmental complexity, and risk entropy value, extracting corresponding hierarchical division parameters, and normalizing the three heterogeneous parameters;

[0011] Construct a comprehensive scoring model based on normalized heterogeneous parameters, and output hierarchical scores through the scoring model;

[0012] Construct a strategy priority model based on the hierarchical division results, and dynamically generate an optimal strategy combination through reinforcement learning;

[0013] Construct a double-loop optimization mechanism to reversely correct the hierarchical division parameters through the policy execution effect;

[0014] Verify the feasibility of the hierarchical division and the optimal strategy combination.

[0015] Combined with the first aspect, further, the modeling formula for monitoring unit density is:

[0016]

[0017] Among them, N cam is the number of effective monitoring devices, A is the monitoring area, ρ k is the spatial coverage rate of the k-th type of device, λ k is the failure coefficient of the k-th type of device, T active is the effective working duration of the device, T total is the total working duration of the device, and ω1, ω2, and ω3 are weight factors respectively;

[0018] The modeling formula for environmental complexity is:

[0019]

[0020] Among them, Entropy(E) is the Shannon entropy of the device type distribution, N type is the number of heterogeneous device types, N total is the total number of devices, ClusteringCoeff(G) is the network topology clustering coefficient, and x and y are complexity adjustment factors respectively;

[0021] The modeling formula for risk entropy value is:

[0022]

[0023] Among them, p i is the occurrence probability of the i-th type of security event, N alert is the number of effective alarms per unit time, Severity(E) is the event threat level coefficient, and z is the risk sensitivity factor.

[0024] Combined with the first aspect, further, after normalizing the heterogeneous parameters using the fuzzy membership function, construct a radial basis neural network scoring model. The mathematical expression formula of the scoring model is:

[0025]

[0026] where \(x = [D ＇ , C ′ , R ′ \) T is the normalized input vector, \(c k is the standardized feature vector of the \(k\)-th monitoring scenario, is the Gaussian kernel function, \(w k is the connection weight from the hidden layer to the output layer, and \(S\) is the output score.

[0027] Combined with the first aspect, further, the generation steps of the optimal policy combination are as follows:

[0028] Input the comprehensive score \(S t , the monitoring density change rate \(\Delta D t , the risk entropy difference \(\Delta R t , and the environmental complexity \(C hist , and output the state space mapping result \(S t :

[0029] s t =(S t , \(\Delta D t , \(\Delta R t , C hist )\in R 4 ;

[0030] Input the reward coefficients \(\alpha\), \(\beta\), \(\gamma\), \(\delta\), the mean time between failures \(MTBF\), the false alarm rate \(FRB\), the accident detection rate \(DR\), and the energy consumption of policy execution \(Energy\), and output the scalar reward value \(r t :

[0031] r t =\alpha\cdot MTBF - \beta\cdot FRB+\gamma\cdot DR-\delta\cdot Energy;

[0032] The Actor network outputs the original action for policy exploration and adds exploration noise:

[0033]

[0034] \sigma t =\sigma_0\cdot e -kt

[0035] where \(\pi θ is the policy function parameterized by the Actor network, \(\sigma_0\) is the initial noise intensity, and \(k\) is the noise decay coefficient;

[0036] Determine the gradient of the objective function of the actor network and the target value for updating the Critic network, input the original action of the Actor network, and output the discretized policy instruction.

[0037] In combination with the first aspect, further, a double-loop correction is adopted to divide the hierarchical parameters:

[0038] A parameter correction loop and a policy execution loop are set. The parameter correction loop and the policy execution loop are executed by backpropagation and policy effect feedback, and a second-order derivative optimization is performed using an optimization method based on Hessian-Free.

[0039] In combination with the first aspect, further, the steps for verifying the hierarchical division and the optimal policy combination include:

[0040] Use the division of time and space blocks for holdout verification and calculate the Kappa consistency coefficient;

[0041] Generate adversarial samples to verify the robustness of the model and introduce the ROI index;

[0042] Deploy a lightweight inference engine at the monitoring front end and design a sliding window model update.

[0043] In a third aspect, the present invention also discloses an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor runs the computer program to implement the intelligent operation and maintenance management method for the security video monitoring system as described above.

[0044] In a fourth aspect, the present invention also discloses a computer-readable storage medium, on which a computer program is stored. The program is executed by a processor to implement the intelligent operation and maintenance management method for the security video monitoring system as described above.

[0045] Advantages of the present invention:

[0046] 1. Multi-dimensional data fusion and fine risk assessment

[0047] By normalizing multi-dimensional heterogeneous parameters such as the density of monitoring units, environmental complexity, and risk entropy values, and using a comprehensive scoring model including interaction terms, a fine division of the security risks in the monitoring area is achieved. This method can effectively capture the non-linear relationships between different parameters and provide a more accurate risk assessment;

[0048] 2. Dynamic adaptive policy generation

[0049] Introduce reinforcement learning technology to build a policy priority model, enabling the system to dynamically generate an optimal operation and maintenance policy combination according to real-time monitoring data and historical operation and maintenance effects, overcoming the deficiency of traditional static policies in adapting to complex environments and greatly improving the efficiency of emergency response and fault repair;

[0050] 3. Double-layer closed-loop optimization mechanism

[0051] Through a double-loop feedback mechanism that optimizes the inner-layer real-time strategy and corrects the outer-layer parameters, the system can continuously adjust the normalization and scoring model parameters according to the policy execution effect, so as to achieve continuous adaptive learning and improvement;

[0052] 4. Having a negative feedback effect

[0053] By verifying the hierarchical division and the feasibility of the optimal policy combination, the stability and robustness of the system are significantly enhanced, ensuring efficient operation in a changing environment. Brief Description of the Drawings

[0054] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to the following drawings without creative efforts.

[0055] Figure 1 It is a flowchart of the steps of the intelligent operation and maintenance management method provided by the embodiment of the present invention. Detailed Embodiments

[0056] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the following will describe the detailed embodiments of the present invention with reference to the accompanying drawings. Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can be implemented in many other ways different from those described herein. Those skilled in the art can make similar improvements without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0057] At present, security video surveillance systems are widely used in fields such as urban security, traffic management, and public place monitoring. Traditional operation and maintenance management methods mainly rely on manual inspections, preset static rules, and simple threshold alarms, and have the following deficiencies:

[0058] Limited heterogeneous data processing ability: Traditional methods are difficult to effectively collect and integrate various heterogeneous parameters such as monitoring unit density, environmental complexity, and risk entropy value at the same time, resulting in insufficient characterization of complex environments and dynamic risks, and unable to comprehensively reflect the security hazards in system operation.

[0059] Lack of self-adaptability of static thresholds and rules: The monitoring and alarm mechanism based on fixed thresholds or preset rules cannot respond to the dynamic changes of the environment and risks in real time, and is prone to false alarms or missed alarms, reducing the reliability and practicality of the system.

[0060] Single means of policy generation and optimization: Traditional operation and maintenance policies often rely on expert experience or fixed processes, lack intelligent decision support, and are not precise enough in identifying the risk levels of monitored areas, and cannot achieve dynamic adjustment and continuous optimization of policies.

[0061] The closed-loop feedback mechanism is imperfect: In the existing technology, the feedback of the operation and maintenance management system on the execution result often only stays at the monitoring and alarm level, lacking a mechanism for real-time adaptive correction of parameters and policy models, and it is difficult to achieve long-term, efficient, and stable operation and maintenance management.

[0062] To solve the above problems, this embodiment discloses an intelligent operation and maintenance management method for a security video monitoring system, which includes the following steps:

[0063] Collect three heterogeneous parameters: monitoring unit density, environmental complexity, and risk entropy value, extract the corresponding hierarchical division parameters, and perform normalization processing on the three heterogeneous parameters;

[0064] Construct a comprehensive scoring model based on the normalized heterogeneous parameters, and output hierarchical scores through the scoring model;

[0065] Construct a policy priority model based on the hierarchical division results, and dynamically generate an optimal policy combination through reinforcement learning;

[0066] Construct a double-loop optimization mechanism, and reversely correct the hierarchical division parameters through the policy execution effect;

[0067] Verify the feasibility of the hierarchical division and the optimal policy combination.

[0068] Furthermore, the modeling formula for the monitoring unit density is:

[0069]

[0070] where N cam is the number of effective monitoring devices, A is the area of the monitoring area, ρ k is the spatial coverage rate of the k-th type of device, λ k is the failure coefficient of the k-th type of device, T active is the effective working duration of the device, T total is the total working duration of the device, and ω1, ω2, and ω3 are weight factors respectively;

[0071] The modeling formula for the environmental complexity is:

[0072]

[0073] where Entropy(E) is the Shannon entropy of the device type distribution, N type is the number of heterogeneous device types, N total is the total number of devices, ClusteringCoeff(G) is the network topology clustering coefficient, and x and y are complexity adjustment factors respectively;

[0074] The modeling formula for the risk entropy value is:

[0075]

[0076] Among them, p i is the occurrence probability of the i-th type of security event, N alert is the number of effective alarms per unit time, Severity(E) is the event threat level coefficient, and z is the risk sensitivity factor; the risk entropy value R characterizes the uncertainty of the regional risk, and the higher the entropy value, the more complex the risk type.

[0077] In this embodiment, the formula for the monitoring unit density introduces the product term of the spatial coverage rate ρ k and the device failure coefficient λ k of the i-th device, which can dynamically reflect the combined impact of device failures on the coverage blind area; the risk entropy value formula performs a power-law coupling on the threat level coefficient and the number of effective alarms, and this structure realizes the non-linear superposition of the event probability and the alarm frequency for the first time.

[0078] In the specific implementation, the normalization process uses a triangular fuzzy membership function, and the domain ranges of the input parameters are defined as follows:

[0079] The domain of the monitoring unit density D is [0, D max , where D max = 5.0, corresponding to the ultra-high density monitoring scenario;

[0080] The domain of the environmental complexity C is [0, 10], which is divided into three fuzzy intervals: low (0 - 3), medium (4 - 6), and high (7 - 10);

[0081] In addition, when the environmental complexity C > 6, the output of the scoring model shows a step change, which never appears in the traditional linear model.

[0082] The domain of the risk entropy value R is [0, log2M], where M is the total number of security event categories.

[0083] In the specific implementation, the parameters of the triangular fuzzy membership function are set as follows:

[0084] The monitoring unit density uses a semi-trapezoidal membership function, and the truncation threshold is set to [0, P max , where O max is determined according to the historical deployment peak;

[0085] The environmental complexity uses a symmetric triangular function, and the vertex position matches the statistical distribution of the Shannon entropy of the device type;

[0086] The risk entropy value uses a semi-Cauchy membership function, and the scale parameter γ = 1.5 to enhance the sensitivity of the high-risk area.

[0087] In this embodiment, normalization, the scoring model, and reinforcement learning constitute a one-way feature extraction channel, and the Gaussian kernel width of the radial basis neural network is strongly correlated with the output range of the fuzzy membership degree.

[0088] Furthermore, after normalizing heterogeneous parameters using the fuzzy membership function, a radial basis neural network scoring model is constructed. The mathematical expression formula of the scoring model is:

[0089]

[0090] where \(x = [D\) ′ , C ′ , R ′ \) T is the normalized input vector, \(c\) k is the standardized feature vector of the \(k\)th monitoring scenario, is the Gaussian kernel function, \(w\) k is the connection weight from the hidden layer to the output layer, and \(S\) is the output score. In this embodiment, the output \(S\in[0, 10]\) is the comprehensive score, and four monitoring levels are divided according to the quartile method.

[0091] Specifically, in the network training, the improved k-means++ algorithm is used to initialize the center of the basis function, and the number of hidden layer nodes is dynamically determined by the Bayesian information criterion. The model performs online incremental learning every 24 hours, and the update formula is:

[0092] \(\Delta w\) j =\(\eta\cdot(y\) ture - y\) pred )\cdot\varphi\) j(x)

[0093] where \(\eta\) is the adaptive learning rate, \(\varphi\) j(x) is the activation value of the \(j\)th hidden layer node, and the timeliness of the feature vector is maintained by the moving average method.

[0094] Furthermore, the generation steps of the optimal policy combination are as follows:

[0095] Input the comprehensive score \(S\) t , the monitoring density change rate \(\Delta D\) t , the risk entropy difference \(\Delta R\) t , and the environmental complexity \(C\) hist , and output the state space mapping result \(S\) t :

[0096] \(s\) t =(S\) t , \(\Delta D\) t , \(\Delta R\) t , \(C\) hist )\in R\) 4 ;

[0097] Input the reward coefficients α, β, γ, δ, the mean time between failures MTBF, the false alarm rate FRB, the accident discovery rate DR, and the energy consumption of policy execution Energy, and output the scalar reward value r t :

[0098] r t = α·MTBF - β·FRB + γ·DR - δ·Energy;

[0099] The Actor network outputs the original actions for policy exploration and adds exploration noise:

[0100]

[0101] σ t = σ0·e -kt

[0102] where π θ is the policy function parameterized by the Actor network, σ0 is the initial noise intensity, and k is the noise decay coefficient;

[0103] Determine the gradient of the objective function of the actor network and the target value for the update of the Critic network, input the original actions of the Actor network, and output the discretized policy instructions.

[0104] The action space dimension of the Actor network is compressed to (k is the number of levels), and this design significantly reduces the policy search complexity.

[0105] Furthermore, a double-layer loop is used to correct the level division parameters:

[0106] Set the parameter correction loop and the policy execution loop. The parameter correction loop and the policy execution loop are executed using backpropagation and policy effect feedback, and a Hessian-Free based optimization method is used for second-order derivative optimization.

[0107] Furthermore, the steps to verify the level division and the optimal policy combination include:

[0108] Use the method of dividing space-time blocks for holdout verification and calculate the Kappa consistency coefficient;

[0109] Generate adversarial samples, verify the robustness of the model, and introduce the ROI index;

[0110] Deploy a lightweight inference engine at the monitoring front end and design a sliding window model update.

[0111] In specific implementation, the generation of adversarial samples uses the FGSM (Fast Gradient Sign Method) attack, and the robustness of the model is verified by monitoring that the decrease in the Kappa coefficient does not exceed 15%.

[0112] Some embodiments of the present application also provide an electronic device, which includes: a processor, a memory, a bus, and a communication interface. The processor, the communication interface, and the memory are connected through the bus; a computer program that can run on the processor is stored in the memory, and when the processor runs the computer program, it executes the method provided in any of the foregoing embodiments of the present application.

[0113] The electronic device provided in the above embodiments of the present application and the intelligent operation and maintenance management method for the security video monitoring system provided in the embodiments of the present application are based on the same application concept and have the same beneficial effects in terms of the methods adopted, run, or implemented.

[0114] Among them, the memory may include a high-speed random access memory (RAM: Random Access Memory), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used.

[0115] The bus can be an ISA bus, a PCI bus, an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. Among them, the memory is used to store the program, and after receiving the execution instruction, the processor executes the program. The intelligent operation and maintenance management method for the security video monitoring system disclosed in any of the foregoing embodiments of the present application can be applied to the processor or implemented by the processor.

[0116] A processor may be an integrated circuit chip with the ability to process signals. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor or the instructions in the form of software. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by the hardware decoding processor, or executed and completed by the combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0117] The electronic device provided in the embodiments of the present application and the intelligent control method for heavy object handling provided in the embodiments of the present application are based on the same application concept and have the same beneficial effects as the method adopted, run, or implemented by it.

[0118] The embodiments of the present application also provide a computer-readable storage medium corresponding to the intelligent control method for heavy object handling provided in the foregoing embodiments. A computer program is stored thereon. The computer-readable storage medium is an optical disc, and a computer program (i.e., a program product) is stored thereon. When the computer program is run by a processor, it will execute the intelligent control method for heavy object handling provided in any of the foregoing embodiments.

[0119] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other optical or magnetic storage media, which will not be elaborated here one by one.

[0120] The computer-readable storage medium provided in the above embodiments of the present application and the intelligent operation and maintenance management method for the security video monitoring system provided in the embodiments of the present application are based on the same application concept and have the same beneficial effects as the method adopted, run, or implemented by the application program stored thereon.

[0121] It should be noted that in the above text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising such element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, but may also include performing functions in a substantially simultaneous manner or in a reverse order according to the functions involved. For example, the described methods may be performed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0122] From the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of the present application.

[0123] The embodiments of the present application have been described above in conjunction with the accompanying drawings, which are only specific embodiments of the present application. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Those of ordinary skill in the art, under the inspiration of the present application and without departing from the spirit and scope protected by the claims of the present application, can still make many forms, all of which fall within the protection scope of the present application.

[0124] In summary, an intelligent operation and maintenance management method, device, and storage medium for a security video monitoring system of the present invention have the following beneficial effects:

[0125] 1. Multi-dimensional data fusion and fine-grained risk assessment

[0126] By normalizing multi-dimensional heterogeneous parameters such as the density of monitoring units, environmental complexity, and risk entropy values, and adopting a comprehensive scoring model that includes interaction terms, a fine-grained division of the security risks in the monitored area is achieved. This method can effectively capture the non-linear relationships between different parameters and provide a more accurate risk assessment;

[0127] 2. Dynamic Adaptive Policy Generation

[0128] Introduce reinforcement learning technology to construct a policy priority model, enabling the system to dynamically generate an optimal combination of operation and maintenance policies based on real-time monitoring data and historical operation and maintenance effects, overcoming the deficiency of traditional static policies in adapting to complex environments and significantly improving the efficiency of emergency response and fault repair;

[0129] 3. Double-Layer Closed-Loop Optimization Mechanism

[0130] Through a double-layer loop feedback mechanism of inner-layer real-time policy optimization and outer-layer parameter correction, the system can continuously adjust the normalization and scoring model parameters according to the policy execution effects, thereby achieving continuous adaptive learning and improvement;

[0131] 4. Having a Negative Feedback Effect

[0132] By verifying the feasibility of hierarchical division and the optimal policy combination, the stability and robustness of the system are significantly enhanced, ensuring efficient operation in a changing environment.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. An intelligent operation and maintenance management method for a security video monitoring system, characterized in that It includes the following steps: By collecting three heterogeneous parameters of monitoring unit density, environmental complexity, and risk entropy value, extracting corresponding hierarchical division parameters, and normalizing the three heterogeneous parameters; Based on the normalized heterogeneous parameters, construct a comprehensive scoring model, and output hierarchical scores through the scoring model; Based on the hierarchical division results, construct a policy priority model, and dynamically generate an optimal policy combination through reinforcement learning; Construct a double-loop optimization mechanism, and reversely correct the hierarchical division parameters through the policy execution effect; Verify the feasibility of the hierarchical division and the optimal policy combination.

2. The intelligent operation and maintenance management method for a security video monitoring system according to claim 1, characterized in that The modeling formula of the monitoring unit density is: Among them, N cam is the number of effective monitoring devices, A is the area of the monitoring region, ρ k is the space coverage rate of the k-th type of device, λ k is the failure coefficient of the k-th type of device, T active is the effective working duration of the device, T total is the total working duration of the device, and ω1, ω2, and ω3 are weight factors respectively; The modeling formula of the environmental complexity is: Among them, Entropy(E) is the Shannon entropy of the device type distribution, N type is the number of heterogeneous device types, N total is the total number of devices, ClusteringCoeff(G) is the network topology clustering coefficient, and x and y are complexity adjustment factors respectively; The modeling formula of the risk entropy value is: where p i is the occurrence probability of the i-th type of security event, N alert is the number of effective alarms per unit time, Severity(E) is the event threat level coefficient, and z is the risk sensitivity factor.

3. An intelligent operation and maintenance management method for a security video monitoring system according to claim 1, characterized in that, After normalizing the heterogeneous parameters using the fuzzy membership function, construct a radial basis neural network scoring model, and the mathematical expression formula of the scoring model is: where x = [D′, C ＇ , R′] T is a normalized input vector, c k is the standardized feature vector of the k-th monitoring scenario, is the Gaussian kernel function, w l is the connection weight from the hidden layer to the output layer, and S is the output score.

4. An intelligent operation and maintenance management method for a security video monitoring system according to claim 1, characterized in that, The generation steps of the optimal policy combination are as follows: Input comprehensive score S t , monitoring density change rate ΔD t , risk entropy difference ΔR t , environmental complexity C hist , and output the state space mapping result S t : s t =(S t , ΔD t , ΔR t , C hist ) ∈ R 4 ; Input the reward coefficients α, β, γ, δ, the mean time between failures MTBF, the false alarm rate FRB, the accident discovery rate DR, and the energy consumption of policy execution Energy, and output the scalar reward value r t : r t = α·MTBF - β·FRB + γ·DR - δ·Energy; The Actor network outputs the original action for policy exploration and adds exploration noise: σ t = σ0·e -kt where, π θ is the policy function parameterized by the Actor network, σ0 is the initial noise intensity, and k is the noise attenuation coefficient; Determine the target function gradient of the actor network and the target value for the update of the Critic network, input the original action of the Actor network, and output the discretized policy instruction.

5. An intelligent operation and maintenance management method for a security video monitoring system according to claim 1, characterized in that, Adopt a double-loop to correct the hierarchical division parameters: Set the parameter correction loop and the policy execution loop. The parameter correction loop and the policy execution loop are executed using backpropagation and policy effect feedback, and the second-order derivative optimization is performed using the Hessian-Free based optimization method.

6. The intelligent operation and maintenance management method for a security video monitoring system according to claim 1, characterized in that The steps for verifying the hierarchical division and the optimal policy combination include: Adopt the method of dividing space-time blocks for holdout verification, and calculate the Kappa consistency coefficient; Generate adversarial samples, verify the robustness of the model, and introduce the ROI index; Deploy a lightweight inference engine at the monitoring front end, and design a sliding window model update.

7. An electronic device, characterized in that it includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor runs the computer program to implement the steps executed in the intelligent operation and maintenance management method for the security video monitoring system according to any one of claims 1-6.

8. A computer-readable storage medium, on which a computer program is stored, characterized in that the program is executed by the processor to implement the steps executed in the intelligent operation and maintenance management method for the security video monitoring system according to any one of claims 1-6.

Images