A security control system optimization method and scheduling system based on standardized strategy table

Abstract

The present application discloses a security control system optimization method and scheduling system based on a standardized policy table, which relates to the field of industrial automation. The method includes: in response to receiving real-time monitoring data of a security control system, determining the trigger conditions in the standardized policy table, and obtaining the policy rules corresponding to the trigger conditions; when the number of trigger conditions is greater than 1, determining the importance value of each policy rule corresponding to the trigger condition according to a pre-trained reinforcement learning model; obtaining the execution time and resource requirement of each policy rule corresponding to the trigger condition; determining the first priority of each policy rule according to the importance value, execution time and resource requirement of each policy rule; executing loop steps to obtain the effect of each policy rule; sending the policy rule with the best effect to the security control system, which is used to execute the policy rule. The present application can be used to optimize the security control system and effectively improve the performance of the security control system.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of industrial automation, and in particular to a security control system optimization method and a scheduling system based on a standardized strategy table. Background Art

[0002] At present, with the rapid development of industrial automation, energy management, intelligent transportation and other fields, safety control systems (security control systems) play a vital role in ensuring system stability and security. In the field of industrial automation, security control systems usually use programmable logic controllers to collect field data through sensors and actuators, and operate according to preset control logic to achieve monitoring and control of the production process. Among them, the control logic is a fixed control strategy designed by engineers based on the expected operating environment and known operating conditions of the security control system. However, the industrial environment is usually dynamic and changeable. For example, the aging of equipment on the production line and changes in the external environment (such as temperature, humidity, etc.) will affect the operation of the security control system, which will cause the fixed control logic to be unable to identify and adjust changes, and ultimately lead to a decline in the performance of the security control system. Therefore, a security control system optimization method is urgently needed to effectively solve the above problems. Summary of the invention

[0003] The embodiments of the present application provide a security control system optimization method and a scheduling system based on a standardized strategy table, which are used to optimize the security control system and effectively improve the performance of the security control system.

[0004] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0005] In a first aspect, a security control system optimization method based on a standardized policy table is provided, which is applied to a central control device, wherein the central control device stores a standardized policy table, wherein the standardized policy table includes a trigger condition and a policy rule, wherein the trigger condition corresponds to the policy rule one by one, and the method includes:

[0006] In response to receiving the real-time monitoring data of the security control system, determining the trigger condition in the standardized policy table, and obtaining the policy rule corresponding to the trigger condition;

[0007] When the number of the trigger conditions is greater than 1, determining the importance value of each of the policy rules corresponding to the trigger condition according to the pre-trained reinforcement learning model;

[0008] Obtaining the execution time and resource requirement of each of the policy rules corresponding to the trigger condition;

[0009] Determine a first priority of each of the policy rules according to the importance value of each of the policy rules, the execution time, and the resource requirement;

[0010] Execute the loop step until the effect of each of the policy rules is obtained;

[0011] Sending the best policy rule to the security control system, and the security control system is used to execute the policy rule;

[0012] Wherein, the circulation step comprises:

[0013] Taking the policy rule with the highest current first priority as the target policy rule;

[0014] Obtaining the execution steps of the target policy rule;

[0015] In a preset simulation model, under the influence of different change factors, the execution steps are executed, and simulation feedback data is received in real time, wherein the change factors include equipment change factors and environment change factors;

[0016] Determining and storing the effect of the target policy rule according to the simulation feedback data, and filtering out the target policy rule, wherein the effect of the target policy rule is characterized by an environmental adaptability index and an economic benefit index;

[0017] According to the first priority of each of the remaining policy rules, the step of taking the policy rule with the highest current first priority as the target policy rule is performed.

[0018] In a possible implementation manner of the first aspect, the central control device includes a plurality of execution units, and determining the first priority of each of the policy rules according to the importance value, the execution time, and the resource requirement of each of the policy rules includes:

[0019] Constructing a matrix R of the resource requirements of each of the policy rules, wherein the element Ri j in the matrix R is used to represent the requirement of the i-th policy rule for the j-th execution unit;

[0020] For each of the strategy rules, according to the importance value, the execution time and the matrix R, a fitness value calculation formula is used to calculate the fitness value of each of the strategy rules;

[0021] According to the fitness value of each of the policy rules, the order of the first priority of each of the policy rules is determined in descending order.

[0022] In another possible implementation manner of the first aspect, before performing the executing step, the method includes:

[0023] The execution order of the execution steps is determined according to the device change factor and the environment change factor.

[0024] In another possible implementation manner of the first aspect, determining the target execution order of the execution steps according to the device change factor and the environment change factor includes:

[0025] Extracting feature vectors of the device change factor and the environment change factor to obtain a feature vector set, and inputting the feature vector set into a pre-trained machine learning model to obtain a second priority of each of the execution steps;

[0026] The execution steps are arranged in descending order according to the second priorities to obtain the execution order.

[0027] In another possible implementation manner of the first aspect, the machine learning model is a gradient boosting tree model, and the training step of the gradient boosting tree model includes:

[0028] Divide the acquired historical data feature vectors into a training set and a validation set;

[0029] Inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps;

[0030] The execution steps are arranged according to the priority score of each execution step to obtain the execution order of the execution steps.

[0031] In another possible implementation manner of the first aspect, inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps includes:

[0032] Inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps, wherein the training set includes a device change feature vector and an environment change feature vector;

[0033] The gradient boosting tree model is used to cyclically execute the prediction step until the change in the verification error when the gradient boosting tree model is verified by using the verification set within a preset round is less than a preset change threshold; the gradient boosting tree model is also used to calculate the priority score of each of the execution steps according to the device change feature vector and the environment change feature vector using a priority score calculation formula when the verification error is the lowest;

[0034] The prediction step includes:

[0035] Initializing the gradient boosting tree model using an initialization formula;

[0036] The prediction error of the current decision tree of the gradient boosting tree model is calculated using a prediction error calculation formula;

[0037] Add a new decision tree to obtain an updated model, wherein the new decision tree is used to fit the current decision tree;

[0038] Using a validation error calculation formula, calculate the validation error of the updated model on the validation set;

[0039] When the change of the verification error within the preset rounds is less than a preset change threshold, terminating the prediction step;

[0040] When the change of the verification error within the preset rounds is greater than or equal to the preset change threshold, the step of initializing the gradient boosting tree model using the initialization formula is performed.

[0041] In another possible implementation manner of the first aspect, the central control device further includes a storage medium, the simulation feedback data includes failure time, normal operation time, energy cost, maintenance cost, and economic benefit, and determining and storing the effect of the target policy rule according to the simulation feedback data includes:

[0042] Calculate the environmental adaptability index of the target policy rule according to the failure time and the normal operation time;

[0043] Calculating the economic benefit index of the target strategy rule according to the energy cost, maintenance cost and economic benefit;

[0044] Performing a weighted summation on the environmental adaptability index and the economic benefit index to obtain a first result, where the first result is used to characterize the effect of the target policy rule;

[0045] The first result is stored in the storage medium.

[0046] In another possible implementation manner of the first aspect, calculating the economic benefit index of the target strategy rule according to the energy cost, the maintenance cost, and the economic benefit includes:

[0047] Using an economic benefit indicator calculation formula, the economic benefit indicator of the target strategy rule is calculated according to the energy cost, maintenance cost and economic benefit;

[0048] The economic benefit index calculation formula includes:

[0049]

[0050] In the formula, ROI is the economic benefit index, A is the economic benefit, Y n is the energy cost, Y w is the maintenance cost.

[0051] In a second aspect, the present application provides a central control device, including:

[0052] a memory configured to store instructions; and

[0053] The processor includes a plurality of execution units, and is configured to call the instructions from the memory and implement the above-mentioned security control system optimization method based on the standardized strategy table when executing the instructions.

[0054] In a third aspect, the present application provides a scheduling system, comprising:

[0055] A plurality of central control devices, wherein each of the central control devices is configured to execute the above-mentioned security control system optimization method based on the standardized strategy table;

[0056] A shared server, the shared server includes a storage disk, the storage disk includes a plurality of round-robin algorithm storage nodes, wherein each of the round-robin algorithm storage nodes encapsulates the above-mentioned round-robin steps, each of the round-robin algorithm storage nodes corresponds to a priority order of a policy rule, and a plurality of the central control devices are connected to the shared server;

[0057] Wherein, the central control device is used to determine the trigger condition in the standardized policy table in response to receiving the real-time monitoring data of the security control system, and obtain the policy rule corresponding to the trigger condition; when the number of the trigger conditions is greater than 1, determine the importance value of each policy rule corresponding to the trigger condition according to the pre-trained reinforcement learning model; obtain the execution time and resource requirement of each policy rule corresponding to the trigger condition; determine the first priority of each policy rule according to the importance value, the execution time and the resource requirement of each policy rule; and

[0058] The central control device is used to determine the first priority order of the policy rules according to the first priority of each of the policy rules, and send the first priority order to the shared server;

[0059] The shared server is used to compare the first priority order with the priority order corresponding to each of the round-robin algorithm storage nodes, and when there is a round-robin algorithm storage node corresponding to a priority order that is consistent with the first priority order, send the round-robin steps encapsulated by the round-robin algorithm node to the central control device, so that the central control device executes the encapsulated round-robin steps.

[0060] Through the above technical solutions, through real-time monitoring of data, dynamic trigger conditions and policy rules, combined with pre-trained reinforcement learning models, the importance, execution time and resource requirements of each policy rule are evaluated, the priority is scientifically determined, and the policy effect is verified through simulation models, ensuring that the system can respond to changes in the environment and equipment status in a timely manner, dynamically adjust the control strategy, and improve the adaptability and effectiveness of the security control system. Finally, through cyclic execution and effect evaluation, the best policy rules are selected and applied in practice, ensuring that the security control system can maintain efficient and safe operation in various complex environments, significantly improving overall performance and reliability, and meeting the needs of modern industrial environments.

[0061] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of a process flow of a security control system optimization method based on a standardized strategy table provided in an embodiment of the present application;

[0063] Figure 2 A schematic diagram of a flow chart of a cycle step provided in an embodiment of the present application;

[0064] Figure 3 A schematic diagram of the structure of a central control device provided in an embodiment of the present application;

[0065] Figure 4 A schematic diagram of a scheduling system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0067] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0068] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0069] Figure 1 The flowchart of a security control system optimization method based on a standardized strategy table according to an embodiment of the present application is schematically shown. Figure 1 As shown, an embodiment of the present application provides a security control system optimization method based on a standardized policy table, which is applied to a central control device. The central control device stores a standardized policy table. The standardized policy table includes trigger conditions and policy rules. The trigger conditions correspond one-to-one to the policy rules. The method may include the following steps.

[0070] S110, in response to receiving the real-time monitoring data of the security control system, determining the triggering condition in the standardized policy table, and obtaining the policy rule corresponding to the triggering condition;

[0071] S120, when the number of trigger conditions is greater than 1, determining the importance value of each policy rule corresponding to the trigger condition according to the pre-trained reinforcement learning model;

[0072] S130, obtaining the execution time and resource requirement of each policy rule corresponding to the trigger condition;

[0073] S140, determining the first priority of each policy rule according to the importance value, execution time and resource requirement of each policy rule;

[0074] S150, executing the loop step until the effect of each policy rule is obtained;

[0075] S160, sending the best policy rule to the security control system, which is used to execute the policy rule;

[0076] The cycle steps include:

[0077] S101, taking the current policy rule with the highest first priority as the target policy rule;

[0078] S102, obtaining execution steps of target policy rules;

[0079] S103, in a preset simulation model, executing the execution steps under the influence of different change factors, and receiving simulation feedback data in real time, the change factors including equipment change factors and environment change factors;

[0080] S104, determining and storing the effect of the target policy rule according to the simulation feedback data, and filtering out the target policy rule, wherein the effect of the target policy rule is characterized by an environmental adaptability index and an economic benefit index;

[0081] S105 . According to the first priority of each remaining policy rule, execute the step of taking the policy rule with the highest current first priority as the target policy rule.

[0082] When the security control system receives real-time monitoring data, it first needs to parse and process the monitoring data. The monitoring data can include a variety of information such as device status, environmental parameters, system load, etc. Through the pre-set parsing algorithm, the key parameters in the data are identified and compared with the trigger conditions in the standardized strategy table. The standardized strategy table stores various possible trigger conditions, which are set based on common and abnormal conditions in the operation of the security control system. For example, if the temperature of a device detected in the real-time monitoring data exceeds the preset safety threshold, the condition will be triggered.

[0083] The policy rules corresponding to the trigger condition are extracted from the policy table. The policy rules predefine the response measures to be taken under specific conditions, such as adjusting equipment operating parameters, starting backup equipment or sounding an alarm, etc., which can effectively ensure that the security control system can respond to various emergencies in a timely manner and ensure the safe and stable operation of the system.

[0084] When multiple trigger conditions are detected, the importance of each trigger condition needs to be evaluated to determine the order of priority. Specifically, the reinforcement learning model can be trained with a large amount of historical data to output the importance value of each policy rule corresponding to the trigger condition. The input of the reinforcement learning model includes the multiple trigger conditions currently detected and their related parameters, such as device status, environmental conditions, etc. The output of the reinforcement learning model is the importance value of the policy rule corresponding to each trigger condition. The importance value reflects the degree of impact of executing the policy rule on the overall safety and performance of the system under the current circumstances. For example, if the two trigger conditions are device overheating and voltage abnormality, the device overheating is judged to be more urgent based on historical data and the current environment, thereby giving it a higher importance value, which can effectively ensure that the most urgent and important issues are handled first, and optimize the allocation and use of resources.

[0085] After determining the importance value of each policy rule, it is necessary to further evaluate the execution time and resource requirements of these rules. The execution time refers to the length of time from the start of executing the policy rule to the completion of the required operation, and the resource requirements include the required computing resources, storage resources and other execution units. This embodiment not only considers the importance, but also comprehensively considers the resource and time constraints, so as to achieve more efficient system optimization.

[0086] After obtaining the importance value, execution time and resource requirement of each policy rule, the first priority of each policy rule can be determined. The priority can be determined by weighted scoring or a preset algorithm. Specifically, for the weighted scoring method, a comprehensive scoring formula can be set, for example: priority = α × importance value - β × execution time - γ × resource requirement, where α, β and γ are weight coefficients used to balance the influence of various factors.

[0087] The higher the priority score, the more priority the policy rule should be executed in the current situation. For example, a policy rule with a high importance value but a large execution time and resource requirements may not have a high comprehensive score as another policy rule with a slightly lower importance value but a smaller execution time and resource requirements. This can ensure the scientificity and rationality of the decision, so that the system can maximize the overall performance within limited resources and time.

[0088] After determining the priority of each policy rule, it is necessary to verify the effects of these policy rules through actual execution. This embodiment uses a cyclic step method to execute each policy rule one by one and evaluate its effect.

[0089] Specifically, first select the policy rule with the highest current priority as the target policy rule and obtain the execution steps. The execution steps may include specific operation instructions, parameter adjustment, device control and other processes. Then, simulate and execute the above execution steps in the preset simulation model. The simulation model can simulate various changing factors in the actual environment, such as changes in device status, changes in ambient temperature, etc., and evaluate the effect of the policy rules in real time through simulation feedback data. The effect evaluation includes environmental adaptability indicators and economic benefit indicators. The environmental adaptability indicators are used to measure the adaptability of policy rules under different environmental conditions, and the economic benefit indicators are used to evaluate the economic benefits they bring. Through cyclic execution and evaluation, the policy rules with the best effect are gradually screened out to ensure that the final implemented policy can achieve the best effect in actual applications.

[0090] After determining the most effective policy rule in the loop step, it is sent to the security control system for actual execution. The sending process includes transmitting the detailed execution steps, parameters and other information of the policy rule to the security control system to ensure that the system can correctly understand and execute these rules. After receiving the policy rule, the security control system operates according to the predetermined steps and parameters. The operation may include adjusting the operating parameters of the equipment, starting or shutting down certain equipment, and sounding an alarm. Through the above-mentioned optimization method, this embodiment enables the security control system to respond quickly and accurately to various emergencies in a complex and changing environment, thereby improving the overall safety and reliability of the system.

[0091] In the loop step, firstly, the current policy rule with the highest first priority is used as the target policy rule, and the execution steps of the target policy rule are obtained.

[0092] Each policy rule contains specific operation instructions and parameter settings, which can describe in detail the operations that need to be performed under specific conditions. For example, if the target policy rule is to adjust the operating parameters of a device, the execution steps may include reading the current status of the device, calculating the parameter value that needs to be adjusted, and sending adjustment instructions. The above execution steps need to be recorded in detail so that the operations can be performed accurately during actual execution. By obtaining and recording these execution steps, it can be ensured that the predetermined process can be followed in different situations, reducing human operation errors and improving the stability and reliability of the system.

[0093] In the preset simulation model, the above execution steps are executed under the influence of different changing factors, and the simulation feedback data is received in real time. The preset simulation model is a virtual environment that can simulate various changing factors in the actual industrial environment, such as changes in equipment status, ambient temperature and humidity, etc. During the simulation process, operations are performed according to the execution steps of the target policy rules, and simulation feedback data is received in real time. The simulation feedback data includes failure time, normal operation time, energy cost, maintenance cost and economic benefits. The use of the simulation model can fully test the policy rules before actual execution, identify potential problems and make adjustments, thereby improving the effectiveness and reliability of the policy rules.

[0094] The effect of the target policy rules is determined and stored based on the simulation feedback data, and the target policy rules are screened out. The effect of the target policy rules is characterized by environmental adaptability indicators and economic benefit indicators. The environmental adaptability indicator measures the adaptability of the policy rules under different environmental conditions, such as high temperature, low temperature, high humidity and other conditions. The economic benefit indicator is used to evaluate the economic benefits of the policy rules in actual applications. By analyzing the simulation feedback data, the effect of the target policy rules is determined and stored for subsequent reference.

[0095] According to the first priority of each remaining policy rule, execute the above step of taking the policy rule with the highest current first priority as the target policy rule. That is, after filtering out the target policy rule with unsatisfactory effect, reselect the policy rule with the highest current priority as the new target policy rule and repeat the above steps. This ensures that all policy rules can be fully tested and evaluated, and the policy rules with the best effect can be selected for practical application. Through the method of cyclic execution and evaluation, the policy rules can be continuously optimized to improve the overall efficiency and reliability of the system.

[0096] In summary, through the above-mentioned cyclic steps, the comprehensive optimization of the security control system policy rules can be achieved. First, by prioritizing and selecting, ensure that the most important policy rules are executed first. Secondly, by recording and obtaining the execution steps in detail, ensure the accuracy and reliability of the operation. Then, through the preset simulation model, test under the influence of different changing factors to comprehensively evaluate the effect of the policy rules. Finally, determine the effect of the policy rules through environmental adaptability indicators and economic benefit indicators, and select the policy rules with the best effect for practical application. Finally, ensure that the security control system can operate efficiently and safely in a complex and changeable industrial environment, and significantly improve the overall performance and reliability.

[0097] This embodiment uses real-time monitoring data, dynamic trigger conditions and policy rules, combined with a pre-trained reinforcement learning model, to evaluate the importance, execution time and resource requirements of each policy rule, scientifically determine the priority, and verify the policy effect through a simulation model. This process ensures that the system can respond to changes in the environment and equipment status in a timely manner, dynamically adjust the control strategy, and improve the adaptability and effectiveness of the system. Ultimately, through cyclic execution and effect evaluation, the best policy rules are screened and applied in practice, ensuring that the security control system can maintain efficient and safe operation in various complex environments, significantly improving overall performance and reliability, and meeting the needs of modern industrial environments.

[0098] In one implementation of this embodiment, the central control device includes multiple execution units, and determines the first priority of each policy rule according to the importance value, execution time and resource requirement of each policy rule, including the following steps:

[0099] S210, constructing a matrix R of resource requirements for each policy rule, where the element Rij in the matrix R is used to represent the requirement of the i-th policy rule for the j-th execution unit;

[0100] S220, for each policy rule, using a fitness value calculation formula according to the importance value, execution time and matrix R, calculate the fitness value of each policy rule;

[0101] S230 . Determine the order of the first priority of each policy rule in descending order according to the fitness value of each policy rule.

[0102] In one of the embodiments, in the genetic algorithm, it is first necessary to construct a resource requirement matrix R for each policy rule. The element Rij in the matrix R is used to characterize the requirement of the i-th policy rule for the j-th execution unit. Assuming there are m policy rules and n execution units, the matrix R is an m×n matrix. Each element Rij can be obtained through actual measurement, representing the requirement of policy rule i for resource unit j during execution. For example, policy rule A may require 2 computing resources and 1 storage resource, while policy rule B may only require 2 storage resources. By constructing the matrix, the requirements of each policy rule for different resources can be understood.

[0103] For each policy rule, the fitness value of each policy rule is calculated using the fitness value calculation formula based on its importance value, execution time and resource requirement matrix R. The fitness value calculation formula can comprehensively consider the importance, execution time and resource requirement of the policy rule, including:

[0104]

[0105] Where F(i) is the fitness value, Ii is the importance value of the i-th strategy rule, Ti is the execution time of the i-th strategy rule, and Rij is the demand for the j-th resource unit when the i-th strategy rule is executed.

[0106] According to the fitness value of each policy rule, the first priority order of each policy rule is determined in descending order. After the sorting is completed, the policy rules with high fitness values ​​will be executed first, and the policy rules with low fitness values ​​will be ranked behind. In this way, through sorting, the execution order of each policy rule can be clarified, thereby optimizing the overall performance of the system.

[0107] Among them, the higher the importance value, the more important the policy rule is, so the higher the fitness value is, the shorter the execution time, the faster the policy rule is executed, so the higher the fitness value is, and the smaller the resource requirement, the lower the resource requirement of the policy rule is, so the higher the fitness value is. Therefore, the fitness function can balance the importance, execution time and resource requirement of the policy rule to evaluate the pros and cons of each policy rule.

[0108] In another embodiment, a genetic algorithm may be used to further process the fitness value of each policy rule to obtain the best fitness of each policy rule, and the order of the best fitness is used as the first priority order of each policy rule. In a genetic algorithm, the higher the fitness value of a policy rule, the better the performance of the policy rule in the current problem, that is, the fitness value of a policy rule represents the importance and execution effect of the policy rule under given resource and time constraints.

[0109] Specifically, in a genetic algorithm, the fitness value can be used for selection, crossover and mutation operations. For ease of understanding, the selection, crossover and mutation operations of the genetic algorithm in this embodiment are explained as follows: selection means that individuals (strategy rules) with high fitness values ​​have a greater probability of being selected as parents; crossover means that the selected parents perform crossover operations to generate new offspring; mutation means that the offspring perform mutation operations to introduce diversity. Through multiple iterations of selection, crossover and mutation operations, the genetic algorithm can gradually optimize the individuals in the population, so that the fitness value gradually increases until the optimal fitness value is obtained.

[0110] In summary, in order to further optimize the fitness value of each strategy rule, a genetic algorithm can be used for processing. The genetic algorithm can gradually optimize the fitness value of the strategy rule by simulating the process of natural selection and genetic variation. If the number of strategy rules is m, according to the fitness value of each strategy rule, the steps of using the genetic algorithm to determine the first priority of each strategy rule are as follows:

[0111] The genetic steps are executed cyclically until the fitness value is no longer significantly improved, that is, the change in the fitness value is less than a first preset threshold value, which may be determined according to specific implementations and is not limited in this application;

[0112] The genetic steps include:

[0113] S1. Take the first n strategy rules with the largest fitness value as the parent generation, and cross the parent generation to obtain n new strategy rules, where n new strategy rules are used to characterize the new offspring generated after the parent generation crossover, n is less than or equal to m, and n is a positive integer;

[0114] S2, mutate the n new strategy rules to obtain n mutated offspring (mutated strategy rules) and the fitness value of each mutated offspring, and calculate the change between the fitness value of each mutated offspring and the initial fitness value;

[0115] When the change amount is less than a second preset threshold, stopping the genetic step;

[0116] When the change amount is greater than or equal to the second preset threshold, S1 is executed.

[0117] In the above genetic steps, the first n strategy rules with the largest fitness values ​​are used as parents, and the parents are crossovered to obtain n new strategy rules. The crossover operation can be performed in the form of single-point crossover. Single-point crossover is to make a cut at a certain position of the parent, exchange the first half and the second half, and thus generate a new offspring. Through the crossover operation, the excellent characteristics of the parent can be combined to generate a better quality offspring. After the crossover operation, the new strategy rules obtained will be used for subsequent mutation operations.

[0118] The mutation operation for mutating n new strategy rules can adopt the method of randomly exchanging the values ​​of two gene positions. Among them, the gene position refers to a position or element in the strategy rule. The purpose of the mutation operation is to introduce diversity and prevent the algorithm from falling into the local optimal solution. The fitness value of the mutated offspring will be recalculated and compared with the initial fitness value to calculate the change in the fitness value. If the change is less than the second preset threshold, the genetic step is stopped; if the change is greater than or equal to the second preset threshold, the genetic step is continued until the fitness value is no longer significantly improved.

[0119] Through multiple iterations of selection, crossover and mutation operations, the genetic algorithm can gradually optimize the individuals in the population, so that the fitness value gradually increases until the optimal fitness value is obtained. Through this optimization method, it can be ensured that the strategy rules are executed in the best order under given resource and time constraints, thereby improving the efficiency and effectiveness of the system.

[0120] In general, genetic algorithms gradually optimize the fitness value of strategy rules by simulating the process of natural selection and genetic variation, ensuring that the system executes strategy rules in the best order under resource and time constraints, thereby ensuring that the execution effect of strategy rules reaches the best state.

[0121] This implementation method can effectively determine the first priority order of each strategy rule by constructing a resource demand matrix, calculating fitness values, sorting strategy rules, and further optimizing the fitness values ​​using a genetic algorithm. The genetic algorithm gradually optimizes the fitness values ​​of the strategy rules by simulating the process of natural selection and genetic variation, ensuring that the system executes the strategy rules in the best order under resource and time constraints, thereby improving the efficiency and effectiveness of the system and effectively ensuring that the execution effect of the strategy rules reaches the best state.

[0122] In one implementation of this embodiment, before executing the execution step, the following steps are included:

[0123] S310: Determine the execution order of the execution steps according to the equipment change factors and the environment change factors.

[0124] Before executing the execution steps, the execution order of the execution steps needs to be determined based on the equipment change factors and environmental change factors. First, the data of equipment change factors and environmental change factors need to be collected and analyzed. Equipment change factors include the operating status, aging degree, fault records, etc. of the equipment; environmental change factors include external environmental conditions such as temperature, humidity, and air pressure. These data can be obtained in real time through sensors installed in the equipment and environment, and stored in the database for analysis.

[0125] This implementation can effectively improve the resilience and reliability of the security control system by determining the execution order of the execution steps according to equipment change factors and environmental change factors.

[0126] Specifically, determining the target execution order of the execution steps according to the equipment change factors and the environment change factors includes the following steps:

[0127] S311, extracting feature vectors of equipment change factors and environment change factors to obtain a feature vector set, and inputting the feature vector set into a pre-trained machine learning model to obtain a second priority of each execution step;

[0128] S312. Arrange the execution steps in descending order according to the second priority level to obtain an execution order.

[0129] This embodiment extracts the feature vector of the change factor from the equipment and the environment. Equipment change factors include the operating status of the equipment (such as temperature, vibration, power consumption), aging degree (such as usage time, maintenance record), fault record (such as fault frequency, fault type), etc. Environmental change factors include external environmental conditions such as temperature, humidity, air pressure, air quality, etc. These data can be obtained in real time through various sensors installed in the equipment and the environment. For example, a temperature sensor can measure the temperature of the equipment and the environment, a vibration sensor can monitor the vibration of the equipment, and a humidity sensor can measure the humidity in the air.

[0130] Specifically, feature engineering can be used to extract feature vectors of device change factors and environmental change factors. That is, the device change factors and environmental change factors can first be cleaned, normalized, and feature extracted to obtain feature vectors of the device change factors and environmental change factors. The set of feature vectors of the device change factors and environmental change factors is a feature vector set. The feature vector set represents the current state of the device and environment. Assuming there are n device change factors and m environmental change factors, the feature vector set can be represented as an (n+m)-dimensional vector X=[x1,x2,…,x n+m ].

[0131] The feature vector set is input into a pre-trained machine learning model to obtain the second priority of each execution step. The pre-trained machine learning model can be a regression model or a classification model based on supervised learning. The machine learning model can predict the second priority of each execution step under the current feature vector set.

[0132] After obtaining the second priority of each execution step, the execution steps are sorted according to the second priority. First, all execution steps and their corresponding second priority scores are sorted into a list. Assume that there are k execution steps S1, S2, …, S k , the second priority scores of each execution step are P1, P2, ..., P k . These steps and their priority scores can be represented as a two-dimensional array:

[0133]

[0134] Next, this two-dimensional array is scored according to the second priority P i Sort from largest to smallest. This can be achieved using efficient sorting algorithms such as quick sort and merge sort. The result of the sorting is a new two-dimensional array, in which the execution steps are arranged from high to low according to the priority score:

[0135]

[0136] Among them, P i1 ≥P i2 ≥P ik The sorted array represents the target execution order of the execution steps, with steps with higher priority scores being executed first and steps with lower priority scores being executed later.

[0137] The sorting method of this embodiment ensures that in actual operation, the most important and urgent steps can be executed first, thereby minimizing the adverse effects caused by equipment changes and environmental changes. For example, if the operating temperature of a device is too high and the ambient temperature is also high, then the execution steps related to cooling will be executed first to prevent the device from overheating and being damaged. In this way, the response speed and adaptability of the system can be improved, ensuring that it can still operate efficiently and safely in a complex and changing environment.

[0138] This implementation method can significantly improve the resilience and reliability of the security control system by determining the target execution order of the execution steps according to the equipment change factors and the environmental change factors. First, by extracting the feature vector and inputting the pre-trained machine learning model, the priority of each execution step under the current equipment and environmental status can be accurately evaluated. Secondly, by sorting the execution steps, it is ensured that the most important and urgent steps are executed first, reducing the adverse effects caused by the changing factors. Finally, through optimization methods, the stability and adaptability of the system can be improved to ensure that it can still operate efficiently and safely in a complex and changing environment to meet the needs of the modern industrial environment. In this way, not only can the operating efficiency of the system be improved, but also equipment failures and downtime can be effectively reduced, maintenance costs can be reduced, and overall economic benefits can be improved.

[0139] In one implementation of this embodiment, the machine learning model is a gradient boosting tree model, and the training steps of the gradient boosting tree model include the following steps:

[0140] S410, dividing the acquired historical data feature vectors into a training set and a validation set;

[0141] S420, inputting a training set into the gradient boosting tree model to obtain a priority score for each execution step;

[0142] S430. Arrange the execution steps according to the priority score of each execution step to obtain the execution order of the execution steps.

[0143] In this embodiment, the historical data includes the feature vectors of the equipment change factors and the environment change factors, as well as the corresponding execution steps and their priority scores. Assume that there are N pieces of historical data, each piece of data consists of a feature vector X i and the corresponding execution step priority score y i Composition, where i = 1, 2,…, N.

[0144] Divide the historical data into a training set and a validation set. The training set is used to train the gradient boosting tree model, and the validation set is used to evaluate the performance of the model. Usually, the data is randomly divided into a training set and a validation set according to a certain ratio (such as 80% and 20%). Assuming that 80% of the data is used as the training set and 20% of the data is used as the validation set, the training set contains 0.8N data and the validation set contains 0.2N data.

[0145] After the data is divided, the training set is used to train the gradient boosting tree model. Gradient Boosting Decision Tree (GBDT) is an ensemble learning method that gradually builds multiple decision tree models and combines the prediction results of these models to improve the overall prediction performance. Specifically, GBDT iteratively trains a series of weak learners (usually decision trees), and adjusts the model parameters according to the residual (prediction error) of the previous model at each iteration, thereby gradually approaching the true value. By inputting the training set into the gradient boosting tree model, the priority score of each execution step can be obtained.

[0146] After training is completed, the execution steps are sorted according to the prediction results of the gradient boosting tree model (the priority score of each execution step). Steps with higher priority scores will be executed first, and steps with lower priority scores will be executed later. This sorting method ensures that in actual operation, the most important and urgent steps can be executed first, thereby minimizing the adverse effects caused by equipment changes and environmental changes.

[0147] This implementation method can significantly improve the accuracy and effectiveness of determining the priority of execution steps by using the gradient boosting tree model to train and predict the feature vectors of equipment change factors and environmental change factors. First, by dividing the historical data into a training set and a validation set, it is ensured that the model can fully learn the characteristics of the data during the training process and effectively evaluate the generalization ability of the model during the validation process. Secondly, through the iterative training process of the gradient boosting tree model, the true value is gradually approached to improve the prediction performance of the model. Finally, by sorting the execution steps, it is ensured that the most important and urgent steps are executed first, reducing the adverse effects caused by the changing factors, which not only improves the operating efficiency of the system, but also effectively reduces equipment failures and downtime, reduces maintenance costs, and improves the overall economic benefits. In addition, it can improve the stability and adaptability of the security control system, ensuring that it can still operate efficiently and safely in a complex and changing environment to meet the needs of a modern industrial environment.

[0148] In one implementation of this embodiment, inputting a training set into a gradient boosting tree model to obtain a priority score for each execution step includes the following steps:

[0149] S510, inputting a training set into the gradient boosting tree model to obtain a priority score for each execution step, wherein the training set includes a device change feature vector and an environment change feature vector;

[0150] The gradient boosting tree model is used to cyclically execute the prediction steps until the change in the verification error of the gradient boosting tree model when the verification set is used to verify the gradient boosting tree model within the preset rounds is less than the preset change threshold; the gradient boosting tree model is also used to calculate the priority score of each execution step according to the device change feature vector and the environment change feature vector using the priority score calculation formula when the verification error is the lowest;

[0151] The prediction steps include:

[0152] S511, using an initialization formula to initialize the gradient boosting tree model;

[0153] S512, using a prediction error calculation formula to calculate the prediction error of the current decision tree of the gradient boosting tree model;

[0154] S513, adding a new decision tree to obtain an updated model, the new decision tree is used to fit the current decision tree;

[0155] S514, using a verification error calculation formula to calculate the verification error of the updated model on the verification set;

[0156] S515, when the change amount of the verification error within the preset round is less than the preset change amount threshold, terminating the prediction step;

[0157] S516: When the change in the verification error within the preset rounds is greater than or equal to the preset change threshold, perform the step of initializing the gradient boosting tree model using an initialization formula.

[0158] This embodiment first obtains historical data containing device change feature vectors and environment change feature vectors, and divides them into a training set and a validation set. The training set is used to train the gradient boosting tree model, and the validation set is used to evaluate the model performance. The gradient boosting tree model is used to cyclically execute the prediction step until the change in the verification error within a preset round is less than a preset change threshold. Specifically, the gradient boosting tree model trains a new decision tree in each round of iteration to reduce the prediction error of the current model. After each iteration, the verification error on the validation set is calculated, and it is determined whether its change is less than the preset change threshold. If so, the training is terminated; otherwise, the iteration continues. Finally, when the verification error is the lowest, the model calculates the priority score of each execution step based on the device change feature vector and the environment change feature vector.

[0159] Specifically, before training the gradient boosting tree model, the model needs to be initialized first. The purpose of the initialization step is to provide an initial prediction value for the model so that it can be gradually improved in subsequent iterations. The initialization formula usually uses the average value of the target variable in the training set as the initial prediction value. Specifically, assuming that the training set contains 0.8N data, and the priority score of each data is y i , then the initial prediction value is the average of all priority scores:

[0160]

[0161] Through initialization, the gradient boosting tree model predicts the same value for all data in the initial state, providing a benchmark for subsequent iterative training. The effect of initialization is to provide a starting point for the model, enabling it to gradually learn and improve the prediction results in subsequent iterations.

[0162] After the model is initialized, the prediction error of the current model is calculated. The prediction error is used to measure the prediction ability of the model on the training set. Assume that the prediction value of the current model is The actual priority score is y i , then the prediction error E calculation formula is:

[0163]

[0164] By calculating the prediction error, the performance of the current model can be evaluated and guidance can be provided for subsequent iterative training. The larger the error, the worse the prediction ability of the model, and the model parameters need to be further adjusted to reduce the error.

[0165] After calculating the prediction error, add a new decision tree to reduce the prediction error of the current model. The new decision tree is used to fit the residual of the current model, that is, the difference between the actual priority score and the value predicted by the current model. The specific steps are as follows:

[0166] S10, calculate the residual of the current model

[0167] S20, using the residual ri as the target value, train a new decision tree model. The new decision tree reduces the prediction error of the current model by learning the characteristics of the residual;

[0168] S30: Add the prediction results of the new decision tree to the prediction value of the current model to obtain the updated model prediction value. Where α is the learning rate, which controls the step size of each update.

[0169] Through S10-S30, the model gradually improves the prediction results and reduces the prediction error in each round of iteration. The addition of new decision trees enables the model to better fit the training data and improve the prediction performance.

[0170] After each iteration, the validation error of the updated model needs to be calculated on the validation set to evaluate the generalization ability of the model. The validation error is used to measure the prediction ability of the model on unseen data. The validation error calculation formula is:

[0171]

[0172] By calculating the validation error, we can evaluate the prediction performance of the model on the validation set and determine whether the model is overfitting or underfitting. The smaller the validation error, the stronger the generalization ability of the model is, and it can maintain good prediction performance on unseen data.

[0173] After each round of iteration, it is necessary to determine whether the change in the verification error is less than the preset change threshold. If the change in the verification error within the preset rounds is less than the preset change threshold, it means that the performance of the model has stabilized, and further iterations may not significantly improve the model performance. At this time, the prediction step can be terminated. Specifically, set a change threshold ∈ and a preset round T. If the change in the verification error |ΔEval| is less than ∈ in T consecutive rounds of iterations, the training is terminated. In this way, over-training of the model can be avoided, training efficiency can be improved, and the performance of the model on the validation set can be optimized.

[0174] If the change in the validation error within the preset rounds is greater than or equal to the preset change threshold, it means that the performance of the model has not yet reached the optimal level and iterative training is required. In this case, re-execute the initialization step to initialize the gradient boosting tree model and continue iterative training. In this way, the model can be continuously improved during the training process to improve the prediction performance until the validation error reaches the optimal level.

[0175] The gradient boosting tree model of this embodiment can effectively train and predict the device change feature vector and the environment change feature vector, and calculate the priority score of each execution step. First, through initialization and iterative training, the model gradually improves the prediction results and reduces the prediction error. Secondly, by calculating the verification error on the verification set, the generalization ability of the model is evaluated to ensure that the model maintains good prediction performance on unseen data. Finally, by judging the change in the verification error, it is decided whether to terminate the training or continue to iterate to ensure that the model achieves the best balance between training efficiency and prediction performance. It not only improves the prediction accuracy of the model, but also effectively avoids overfitting and underfitting problems and improves the generalization ability of the model. Through this optimization method, the operating efficiency and stability of the system can be significantly improved, ensuring that it can still operate efficiently and safely in a complex and changing environment to meet the needs of a modern industrial environment.

[0176] In one implementation of this embodiment, the central control device further includes a storage medium, the simulation feedback data includes failure time, normal operation time, energy cost, maintenance cost and economic benefit, and determining and storing the effect of the target policy rule according to the simulation feedback data includes the following steps:

[0177] S610, calculating the environmental adaptability index of the target policy rule according to the failure time and the normal operation time;

[0178] S620, calculating the economic benefit index of the target strategy rule according to the energy cost, maintenance cost and economic benefit;

[0179] S630, performing weighted summation on the pipe diameter adaptability index and the economic benefit index to obtain a first result, where the first result is used to characterize the effect of the target strategy rule;

[0180] S640: Store the first result in a storage medium.

[0181] First, the environmental adaptability index of the target policy rule is calculated based on the failure time and normal operation time in the simulation feedback data. Failure time refers to the cumulative time when the equipment fails during operation, and normal operation time refers to the cumulative operation time of the equipment in a fault-free state. The environmental adaptability index reflects the reliability and stability of the equipment under different environmental conditions. The specific calculation method can use the failure rate formula, that is, the failure rate is equal to the failure time divided by the total operation time (failure time plus normal operation time). For example, if the failure time of the equipment is 100 hours and the normal operation time is 900 hours, the total operation time is 1000 hours, and the failure rate is 100 / 1000=0.1, that is, the environmental adaptability index is 0.9. In this way, the adaptability of the equipment in a specific environment can be quantified to provide data support for subsequent policy optimization.

[0182] Next, the economic benefit index of the target strategy rule is calculated based on the energy cost, maintenance cost and economic benefit in the simulation feedback data. Energy cost refers to the energy cost consumed during the operation of the equipment, maintenance cost refers to the maintenance cost incurred during the operation of the equipment, and economic benefit refers to the economic income brought by the normal operation of the equipment. The economic benefit index is used to evaluate the net income generated by the equipment during operation.

[0183] The economic benefit index can adopt the net benefit formula, that is, the net benefit is equal to the economic benefit minus the sum of energy cost and maintenance cost. For example, if the economic benefit of the equipment is 10,000 yuan, the energy cost is 2,000 yuan, and the maintenance cost is 1,000 yuan, then the net benefit is 10,000-2,000-1,000=7,000 yuan. This embodiment can quantify the economic benefit of the equipment during operation and provide data support for subsequent strategy optimization.

[0184] Then, the environmental adaptability index and the economic benefit index are weighted and summed to obtain the first result. Weighted summation means that different weights are assigned according to the importance of different indicators, and the weighted values ​​of each indicator are added to obtain a comprehensive evaluation result. The specific calculation method can adopt the weighted average formula, that is, the first result is equal to the environmental adaptability index multiplied by its weight plus the economic benefit index multiplied by its weight. For example, if the environmental adaptability index is 0.9, the economic benefit index is 7,000 yuan, the weight of the environmental adaptability index is 0.4, and the weight of the economic benefit index is 0.6, then the first result is equal to 0.9×0.4+7000×0.6=4200.36. In this way, the effect of the target policy rule can be comprehensively evaluated.

[0185] Finally, the first result is stored in a storage medium. The storage medium may be a part of a central control device, which is used to store and manage various data. Storing the first result in a storage medium may facilitate subsequent query and analysis. A specific implementation method may adopt database storage, and insert the first result as a record into a database table. For example, a SQL statement is used to insert the first result into a database table named "Strategy Effectiveness Evaluation", and the table structure includes fields such as "Strategy ID", "Environmental Adaptability Index", "Economic Benefit Index" and "Comprehensive Evaluation Results". In this way, subsequent query and analysis can be facilitated, providing data support for strategy optimization.

[0186] This implementation method can systematically evaluate the effect of the target policy rules and store the evaluation results. First, by calculating the environmental adaptability index and the economic benefit index, the reliability and economic benefit of the equipment under different environmental conditions are quantified. Then, by weighted summation, the effect of the target policy rules is comprehensively evaluated. Finally, the comprehensive evaluation results are stored in a storage medium for subsequent query and analysis. It can not only improve the scientificity and accuracy of the policy evaluation, but also provide data support for policy optimization, and ultimately improve the operating efficiency and economic benefits of the equipment.

[0187] In one implementation of this embodiment, calculating the economic benefit index of the target strategy rule according to the energy cost, maintenance cost and economic benefit includes the following steps:

[0188] S710, using an economic benefit index calculation formula to calculate the economic benefit index of the target strategy rule according to energy cost, maintenance cost and economic benefit;

[0189] The calculation formula for economic benefit indicators includes:

[0190]

[0191] In the formula, ROI is the economic benefit index, A is the economic benefit, and Y is n is the energy cost, Y w For maintenance costs.

[0192] Please refer to the description of this embodiment in steps S610-S640, which will not be repeated here.

[0193] Reference Figure 3 , the embodiment of the present application further provides a central control device, including:

[0194] A memory 10 configured to store instructions; and

[0195] The processor 20 includes a plurality of execution units, and is configured to call instructions from the memory 10 and implement the above-mentioned security control system optimization method based on the standardized policy table when executing the instructions.

[0196] Reference Figure 4 , the embodiment of the present application also provides a scheduling system, including:

[0197] A plurality of central control devices, wherein each central control device is configured to execute the above-mentioned security control system optimization method based on the standardized strategy table;

[0198] A shared server, the shared server includes a storage disk, the storage disk includes a plurality of round-robin algorithm storage nodes, wherein each round-robin algorithm storage node encapsulates the above-mentioned round-robin steps, each round-robin algorithm storage node corresponds to a priority order of a policy rule, and a plurality of central control devices are connected to the shared server;

[0199] The central control device is used to determine the trigger conditions in the standardized policy table in response to receiving the real-time monitoring data of the security control system, and obtain the policy rules corresponding to the trigger conditions; when the number of trigger conditions is greater than 1, determine the importance value of each policy rule corresponding to the trigger condition according to the pre-trained reinforcement learning model; obtain the execution time and resource requirement of each policy rule corresponding to the trigger condition; determine the first priority of each policy rule according to the importance value, execution time and resource requirement of each policy rule; and

[0200] The central control device is used to determine a first priority order of the policy rules according to the first priority of each policy rule, and send the first priority order to the shared server;

[0201] The shared server is used to compare the first priority order with the priority order corresponding to each round-robin algorithm storage node, and when there is a round-robin algorithm storage node whose priority order is consistent with the first priority order, send the round-robin algorithm node encapsulated round-robin steps to the central control device, so that the central control device executes the encapsulated round-robin steps.

[0202] In this embodiment, the shared server and storage disk can enable multiple central control devices to share computing resources and data storage, which reduces the storage requirements for duplicate data and the waste of computing resources. And because each central control device can run independently and communicate with the shared server, it has good scalability and can increase control devices as needed.

[0203] Specifically, encapsulating the loop steps in independent storage nodes makes the system design more modular. Each node can be independently developed, tested and maintained, which improves the flexibility and maintainability of the scheduling system. In addition, multiple loop algorithm storage nodes can run in parallel, which significantly improves the system processing capacity and efficiency. Different central control devices can access and execute loop algorithm storage nodes at the same time, thereby reducing waiting time and processing delays. In addition, since the loop steps are distributed on multiple nodes, when new loop steps need to be added or existing steps need to be modified, only the corresponding storage nodes need to be updated or added, without large-scale changes to the entire algorithm, thus improving the flexibility and scalability of the scheduling system. Encapsulating complex loop steps in the storage nodes of shared servers can effectively reduce the computing and storage burden of the central control device, allowing it to focus more on the processing and preliminary decision-making of real-time monitoring data.

[0204] It should be noted that when the central control device executes the above-mentioned security control system optimization method based on the standardized policy table, if the loop algorithm storage node in the shared server is not called, and the loop step is executed autonomously, the central control device will encapsulate the loop step after executing the loop step and send the loop step to the shared server for storage. In addition, each loop algorithm storage node in the shared server is an independent node, and each loop algorithm storage node only stores one loop step. The loop algorithm storage node can be expanded based on actual needs.

[0205] An embodiment of the present application also provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the above-mentioned security control system optimization method based on the standardized policy table.

[0206] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0207] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0208] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0209] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0210] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0211] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0212] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media 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 memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0213] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0214] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A security control system optimization method based on a standardized strategy table, characterized in that: Applied to a central control device, the central control device stores a standardized policy table, the standardized policy table includes trigger conditions and policy rules, the trigger conditions correspond to the policy rules one by one, the method includes: In response to receiving the real-time monitoring data of the security control system, determining the trigger condition in the standardized policy table, and obtaining the policy rule corresponding to the trigger condition; When the number of the trigger conditions is greater than 1, determining the importance value of each of the policy rules corresponding to the trigger condition according to the pre-trained reinforcement learning model; Obtaining the execution time and resource requirement of each of the policy rules corresponding to the trigger condition; Determine a first priority of each of the policy rules according to the importance value of each of the policy rules, the execution time, and the resource requirement; Execute the loop step until the effect of each of the policy rules is obtained; Sending the best policy rule to the security control system, and the security control system is used to execute the policy rule; Wherein, the circulation step comprises: Taking the policy rule with the highest current first priority as the target policy rule; Obtaining the execution steps of the target policy rule; In a preset simulation model, under the influence of different change factors, the execution steps are executed, and simulation feedback data is received in real time, wherein the change factors include equipment change factors and environment change factors; Determining and storing the effect of the target policy rule according to the simulation feedback data, and filtering out the target policy rule, wherein the effect of the target policy rule is characterized by an environmental adaptability index and an economic benefit index; According to the first priority of each remaining policy rule, executing the step of taking the policy rule with the highest current first priority as the target policy rule; Before executing the execution step, the method includes: Determining the execution order of the execution steps according to the equipment change factor and the environment change factor; The determining the target execution order of the execution steps according to the device change factor and the environment change factor includes: Extracting feature vectors of the device change factor and the environment change factor to obtain a feature vector set, and inputting the feature vector set into a pre-trained machine learning model to obtain a second priority of each of the execution steps; Arrange the execution steps in descending order according to the second priorities to obtain the execution order; The machine learning model is a gradient boosting tree model, and the training steps of the gradient boosting tree model include: Divide the acquired historical data feature vectors into a training set and a validation set; Inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps; Arranging the execution steps according to the priority score of each of the execution steps to obtain an execution order of the execution steps; Inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps includes: Inputting the training set into the gradient boosting tree model to obtain a priority score for each of the execution steps, wherein the training set includes a device change feature vector and an environment change feature vector; The gradient boosting tree model is used to cyclically execute the prediction step until the change in the verification error when the gradient boosting tree model is verified by using the verification set within a preset round is less than a preset change threshold; the gradient boosting tree model is also used to calculate the priority score of each of the execution steps according to the device change feature vector and the environment change feature vector using a priority score calculation formula when the verification error is the lowest; The prediction step includes: Initializing the gradient boosting tree model using an initialization formula; The prediction error of the current decision tree of the gradient boosting tree model is calculated using a prediction error calculation formula; Add a new decision tree to obtain an updated model, wherein the new decision tree is used to fit the current decision tree; Using a validation error calculation formula, calculate the validation error of the updated model on the validation set; When the change of the verification error within the preset rounds is less than a preset change threshold, terminating the prediction step; When the change of the verification error within the preset rounds is greater than or equal to the preset change threshold, the step of initializing the gradient boosting tree model using the initialization formula is performed.

2. The method according to claim 1, characterized in that The central control device includes a plurality of execution units, and the determining of the first priority of each of the policy rules according to the importance value, the execution time and the resource requirement of each of the policy rules includes: Constructing a matrix R of the resource requirements of each of the policy rules, wherein the element Rij in the matrix R is used to represent the requirement of the i-th policy rule for the j-th execution unit; For each of the strategy rules, according to the importance value, the execution time and the matrix R, a fitness value calculation formula is used to calculate the fitness value of each of the strategy rules; According to the fitness value of each of the policy rules, the order of the first priority of each of the policy rules is determined in descending order.

3. The method according to claim 1, characterized in that The central control device further includes a storage medium, the simulation feedback data includes failure time, normal operation time, energy cost, maintenance cost and economic benefit, and the determining and storing the effect of the target policy rule according to the simulation feedback data includes: Calculate the environmental adaptability index of the target policy rule according to the failure time and the normal operation time; Calculating the economic benefit index of the target strategy rule according to the energy cost, maintenance cost and economic benefit; Performing a weighted summation on the environmental adaptability index and the economic benefit index to obtain a first result, where the first result is used to characterize the effect of the target policy rule; The first result is stored in the storage medium.

4. The method according to claim 1, characterized in that: According to energy cost, maintenance cost and economic benefit, the economic benefit index of the target strategy rule is calculated, including: Using an economic benefit indicator calculation formula, the economic benefit indicator of the target strategy rule is calculated according to the energy cost, maintenance cost and economic benefit; The economic benefit index calculation formula includes: ； Wherein, ROI is the economic benefit index, A is the economic benefit, Yn is the energy cost, and Yw is the maintenance cost.

5. A central control device, characterized in that: include: a memory configured to store instructions; as well as The processor comprises a plurality of execution units, and is configured to call the instructions from the memory and to implement the security control system optimization method based on the standardized strategy table according to any one of claims 1 to 4 when executing the instructions.

6. A scheduling system, characterized in that: include: A plurality of central control devices, wherein each of the central control devices is configured to execute the security control system optimization method based on the standardized strategy table according to any one of claims 1 to 4; A shared server, the shared server comprising a storage disk, the storage disk comprising a plurality of round-robin algorithm storage nodes, wherein each of the round-robin algorithm storage nodes encapsulates the round-robin step according to claim 1, each of the round-robin algorithm storage nodes corresponds to a priority order of a policy rule, and a plurality of the central control devices are connected to the shared server; Wherein, the central control device is used to determine the trigger condition in the standardized policy table in response to receiving the real-time monitoring data of the security control system, and obtain the policy rule corresponding to the trigger condition; when the number of the trigger conditions is greater than 1, determine the importance value of each policy rule corresponding to the trigger condition according to the pre-trained reinforcement learning model; obtain the execution time and resource requirement of each policy rule corresponding to the trigger condition; determine the first priority of each policy rule according to the importance value, the execution time and the resource requirement of each policy rule; and The central control device is used to determine the first priority order of the policy rules according to the first priority of each of the policy rules, and send the first priority order to the shared server; the shared server is used to compare the first priority order with the priority order corresponding to each of the round-robin algorithm storage nodes, and if there is a round-robin algorithm storage node whose corresponding priority order is consistent with the first priority order, send the round-robin steps encapsulated by the round-robin algorithm node to the central control device, so that the central control device executes the encapsulated round-robin steps.

Images