Information system integration service optimization method based on artificial intelligence

Abstract

The invention discloses an information system integration service optimization method based on artificial intelligence, and relates to the field of integration service digital optimization transformation, and the method comprises the steps: obtaining the past operation data and historical business data of an information system integration platform, carrying out the preprocessing of the past operation data and historical business data, and obtaining a standard data set; carrying out feature extraction on the standard data set, respectively carrying out long-term and short-term time sequence capture and spatial association to obtain a long-term prediction parameter and a short-term prediction parameter, obtaining current operation data, and predicting the information system integration service based on the long-term prediction parameter and the short-term prediction parameter to obtain a predicted operation condition and a predicted service condition; and based on the predicted operation condition and the predicted business condition, performing information system integration service optimization by adopting an improved simulated annealing algorithm to obtain a target scheduling optimization strategy. The method focuses on resource scheduling and process optimization, and gives consideration to dynamic adaptation and optimization precision.

Description

Technical Field

[0001] This invention relates to the field of digital optimization and transformation of integrated services, and in particular to an information system integrated service optimization method based on artificial intelligence. Background Technology

[0002] Information system integration service integration optimization refers to the process of continuously improving the overall system operating efficiency, reducing resource consumption, and ensuring business continuity through technical means, based on the integration of computer software, hardware, network communication and other technologies and products to meet the specific business needs of users, and ultimately achieving synergistic improvement of system performance and business value. Its core objective is to break down barriers between systems, optimize resource allocation, and make the integrated system more in line with the dynamic development needs of business.

[0003] In the current field of information system integration services, existing technical solutions have many shortcomings, which have become key bottlenecks restricting the optimization effect.

[0004] In the prediction and optimization phase, traditional solutions often use single-dimensional time series analysis, ignoring the spatial correlation between data (such as the linkage relationship of resource consumption in different subsystems and the dependency relationship between business modules), and failing to distinguish between long-term (such as quarterly, annual) and short-term (such as hourly, daily) business fluctuation patterns, resulting in insufficient accuracy of prediction parameters and difficulty in effectively predicting the trend of system operation and business development.

[0005] In addition, existing optimization algorithms (such as traditional genetic algorithms and greedy algorithms) are prone to getting stuck in local optima when dealing with multi-objective scheduling problems. They cannot take into account multiple dimensions of requirements such as system stability, resource utilization and business response speed, and it is difficult to generate efficient scheduling strategies.

[0006] To address these issues, there is an urgent need for AI-based optimization methods for information system integration services. Summary of the Invention

[0007] To address the aforementioned issues, this application proposes an information system integration service optimization method based on artificial intelligence, comprising the following steps: S1. Obtain past operating data and historical business data from the information system integration platform, and preprocess the past operating data and historical business data to obtain a standard dataset; S2. Extract features from the standard dataset and perform long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters. S3. Obtain current operating data: Based on long-term and short-term forecast parameters, predict the information system integration service to obtain the predicted operating status and predicted business status. S4. Based on the predicted operational and business conditions, an improved simulated annealing algorithm is used to optimize the information system integration service and obtain the target scheduling optimization strategy.

[0008] Preferably, the past operational data includes server CPU utilization, server memory utilization, node liveness status, microservice call volume, response time, interface call log, step dependency relationship, and process trigger volume; The historical business data includes past resource call scenarios and corresponding past performance metrics, past process optimization schemes and past time consumption data, and past error rates.

[0009] Preferably, the specific content of extracting features from the standard dataset and performing long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters includes: The standard dataset includes standard operational data and standard business data; Establish an operational data curve with time as the horizontal axis and standard operational data as the vertical axis; By combining the resource call scenarios in standard business data, the runtime data curve is divided into several time stages of business processing scenarios, and scenario labels are established for the corresponding time stages. And the time period for establishing scene tags is defined as a normal time period; An experience pool is established on the scene label. The experience pool includes past resource allocation schemes and corresponding past performance indicators, past process optimization schemes and past time consumption data, and past error rates.

[0010] Preferably, the specific details of extracting features from the standard dataset and performing long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters also include: Peak features are obtained by extracting features from the running data curves; Based on the Prophet model, time-series feature extraction is performed on peak features to obtain business periodicity features and holiday effect features; The Prophet model is trained based on the cyclical characteristics of business operations and the holiday effect to obtain long-term prediction parameters. The running data curves are divided into short-term time groups according to time stages; The Transformer model was used to train short-term time groups to obtain short-term prediction parameters for different time groups.

[0011] Preferably, the specific content of the target scheduling optimization strategy obtained by using an improved simulated annealing algorithm to optimize information system integration services based on predicted operational and business conditions is as follows: An initial solution is generated using a greedy algorithm by combining predicted operational and business conditions. Set the initial temperature T0, the termination temperature, and the decay coefficient; A target temperature is preset, and a fitness function is constructed by integrating multiple objectives and constraint penalties, which is continuously updated and iterated until the target temperature is reached. The hill-climbing algorithm is used to fine-tune the traversal of neighborhood solutions to obtain the target scheduling optimization strategy.

[0012] Preferably, the specific content of generating the initial solution using a greedy algorithm by combining the predicted operational situation and the predicted business situation is as follows: The predicted operational status and predicted business status are analyzed to obtain node load information and service call information; By combining the stability of the information system with the node load information, we can determine the available nodes and the dangerous nodes; Extract the call type and corresponding call volume from the service call information, and sort the call types from high to low according to the call volume; Define the peak order range according to business needs, and mark the peak service label for the call type within the peak order range; Greedy rules and their priorities are established based on the core rules of the information system. By combining service call information and node load information, a corresponding list of services to be allocated and a list of nodes are generated, and a blank resource allocation framework is generated. Select a service from the list of services to be assigned and filter it in the node list to obtain available nodes; Calculate the fit score of the available nodes and select the node with the highest fit score as the target node; Adaptation score = Node remaining resources × Service priority weight; Resource allocation is performed based on the target node, and the resource allocation results are saved to a blank resource allocation framework. The process continues until all services in the service list to be assigned have completed node allocation and resource quota settings, resulting in an initial solution.

[0013] Preferably, the specific content of constructing the fitness function by fusing constraint penalties includes: The fitness function includes a scheduling fitness function established for resource scheduling and a process fitness function established for process optimization. The expression for the scheduling fitness function is: ; in, For scheduling fitness, , , , These are dynamic coefficients for response time, resource utilization, cost, and load overload penalty, respectively. For response time, For resource utilization, For cost, Penalty for exceeding load limits; The expression for the process fitness function is: in, For scheduling fitness, , , These are dynamic coefficients for process time, resource utilization, error rate, and penalties for violations of step sequence. The process takes a long time. For error rate, Penalties will be imposed for violations of the correct procedure sequence.

[0014] A system for optimizing information system integration services based on artificial intelligence, comprising: Data acquisition unit: acquires past operational data and historical business data of the information system integration platform, and preprocesses the past operational data and historical business data to obtain a standard dataset; Real-time prediction unit: extracts features from the standard dataset and performs long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters respectively; Strategy optimization generation unit: Based on long-term and short-term prediction parameters, the scheduling optimization prediction of information system integration services is performed to obtain an initial scheduling optimization strategy. The initial scheduling optimization strategy is then optimized based on the improved simulated annealing algorithm to obtain the target scheduling optimization strategy.

[0015] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor invokes the computer program in the memory to implement the content of an information system integration service optimization method based on artificial intelligence.

[0016] A storage medium storing computer-executable instructions, wherein when the computer-executable instructions are loaded and executed by a processor, the content of an information system integration service optimization method based on artificial intelligence is disclosed.

[0017] In summary, the information system integration service optimization method based on artificial intelligence of the present invention has the following advantages compared with traditional technologies: 1. This application integrates multi-source heterogeneous historical data, combines time series modeling of long-term trends and short-term fluctuations with spatial correlation analysis, deeply mines the behavioral patterns of the system, provides highly reliable input for prediction, and breaks through the decision-making blind spots caused by data silos; 2. This application dynamically generates multi-dimensional prediction parameters to map future operating trends in real time; the improved simulated annealing algorithm breaks through local optima, achieves a global balance between resource scheduling and business needs, and ensures controllable SLA and costs.

[0018] The technical method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of the information system integration service optimization method based on artificial intelligence of the present invention. Figure 2 This is a unit diagram of the information system integration service optimization system based on artificial intelligence according to the present invention. Detailed Implementation

[0020] The technical method of the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0021] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0022] Techniques, systems, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the instruction manual.

[0023] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0024] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0025] This invention provides an information system integration service optimization method based on artificial intelligence, such as... Figure 1 As shown, it includes the following steps: S1. Obtain past operational data and historical business data from the information system integration platform, and preprocess the past operational data and historical business data to obtain a standard dataset.

[0026] Furthermore, the past operational data includes server CPU utilization, server memory utilization, node liveness status, microservice call volume, response time, interface call chain logs, step dependencies, and process trigger volume. The historical business data includes past resource call scenarios and corresponding past performance indicators, past process optimization schemes and past time consumption data, and past error rates.

[0027] Understandably, during the preprocessing stage, isolated forests can be used to filter abnormal load data caused by server failures, LSTM can be used to filter abnormal process timeouts caused by interface timeouts, and missing values ​​(such as missing offline node data) can be filled in using a Transformer model.

[0028] S2. Extract features from the standard dataset and perform long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters. Furthermore, the specific details of extracting features from the standard dataset and performing long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters include: The standard dataset includes standard operational data and standard business data; Establish an operational data curve with time as the horizontal axis and standard operational data as the vertical axis; By combining the resource call scenarios in standard business data, the runtime data curve is divided into several time stages of business processing scenarios, and scenario labels are established for the corresponding time stages. And the time period for establishing scene tags is defined as a normal time period; An experience pool is established on the scene label. The experience pool includes past resource allocation schemes and corresponding past performance indicators, past process optimization schemes and past time consumption data, and past error rates.

[0029] Furthermore, the specific details of extracting features from the standard dataset and performing long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters also include: Peak features are obtained by extracting features from the running data curves; Based on the Prophet model, time-series feature extraction is performed on peak features to obtain business periodicity features and holiday effect features; The Prophet model is trained based on the cyclical characteristics of business operations and the holiday effect to obtain long-term prediction parameters. The running data curves are divided into short-term time groups according to time stages; The Transformer model was used to train short-term time groups to obtain short-term prediction parameters for different time groups.

[0030] S3. Obtain current operating data: Based on long-term and short-term forecast parameters, predict the information system integration service to obtain the predicted operating status and predicted business status. S4. Based on the predicted operational and business conditions, an improved simulated annealing algorithm is used to optimize the information system integration service and obtain the target scheduling optimization strategy.

[0031] Furthermore, based on predicted operational and business conditions, an improved simulated annealing algorithm is used to optimize information system integration services, resulting in the following specific content of the target scheduling optimization strategy: The initial solution is generated by a greedy algorithm by combining the predicted operational status and the predicted business status. For example, the predicted operational status and the predicted business status can include the peak service call volume in the next 1 hour (short) / 6 hours (short), the predicted load of each node, the process trigger volume in the next 1 hour, and the predicted response time of key interfaces.

[0032] Furthermore, the specific details of using a greedy algorithm to generate the initial solution, based on the predicted operational and business conditions, are as follows: The predicted operational status and predicted business status are analyzed to obtain node load information and service call information; By combining the stability of the information system with the node load information, we can determine the available nodes and the dangerous nodes; Extract the call type and corresponding call volume from the service call information, and sort the call types from high to low according to the call volume; Define the peak order range according to business needs, and mark the peak service label for the call type within the peak order range; Greedy rules and their priorities are established based on the core rules of the information system. For example, core rules for information system integration may include: core services (payment, login) should be deployed on dual nodes (master-slave architecture) to avoid single point of failure; resource allocation for peak services should reserve 20% redundancy (to cope with prediction deviations); non-core services (logs, reports) should be deployed on low-configuration nodes first to save high-configuration resources; and services in the same business domain (such as orders, inventory) should be deployed on the same node first to reduce cross-node communication latency.

[0033] Greedy selection rules (ranked by priority) may include prioritizing the allocation of available redundant node resources for peak services, prioritizing core services to occupy high-configuration nodes (CPU / memory ≥ 8 cores 16GB), and allocating resources according to the "predicted call volume percentage" (e.g., if the order service is predicted to account for 60% of the total traffic, allocate 60% of the core node resources).

[0034] By combining service call information and node load information, a corresponding list of services to be allocated and a list of nodes are generated, and a blank resource allocation framework is generated. Select a service from the list of services to be assigned and filter it in the node list to obtain available nodes; Calculate the fit score of the available nodes and select the node with the highest fit score as the target node; Adaptation score = Node remaining resources × Service priority weight; Resource allocation is performed based on the target node, and the resource allocation results are saved to a blank resource allocation framework. The process continues until all services in the service list to be assigned have completed node allocation and resource quota settings, resulting in an initial solution.

[0035] Understandably, this adjustment is particularly applicable to scenarios with large business fluctuations (such as e-commerce promotions and peak processing times in government systems), upgrading SA's optimization solution from "passively adapting to the present" to "proactively adapting to the future."

[0036] Among them, the initial solution is more forward-looking, avoiding the solution that is "currently optimal but will fail in the future" (such as insufficient resources due to failure to predict traffic peaks). The fitness evaluation is more comprehensive: it not only focuses on current performance, but also evaluates the adaptability of the solution to future states, improving the robustness of optimization. The dynamic response is more accurate, and the re-optimization is triggered by updating the prediction model, which is more timely than the response based solely on the current state (such as expanding capacity 30 minutes in advance for predicted peaks).

[0037] Set the initial temperature T0, the termination temperature, and the decay coefficient; It is understandable that the initial temperature T0 can be dynamically output by the reinforcement learning (RL) agent according to the complexity of the scenario. For example, in the resource scheduling process, T0=150 (enhancing global exploration) in high-load scenarios (such as e-commerce promotions), T0=80 (focusing on local optimization) in low-load scenarios, T0=120 (exploring more paths) in long-process scenarios and T0=60 (fast convergence) in short-process scenarios.

[0038] The termination temperature ensures that the algorithm converges to a stable solution, while the decay coefficient can be dynamically adjusted according to the "improvement of the solution". For example, in resource scheduling, if the fitness improvement is >5% for three consecutive generations (search effective), the decay coefficient = 0.95 (slow cooling); if the improvement is <1% (search stalls), the decay coefficient = 0.85 (fast cooling); in process optimization, if the process time optimization is >10%, the decay coefficient = 0.93; if the optimization is <2%, the decay coefficient = 0.80.

[0039] A target temperature is preset, and a fitness function is constructed by integrating multiple objectives and constraint penalties, which is continuously updated and iterated until the target temperature is reached. Furthermore, the specific details of constructing the fitness function by fusing constraints and penalties include: The fitness function includes a scheduling fitness function established for resource scheduling and a process fitness function established for process optimization. The expression for the scheduling fitness function is: ; in, For scheduling fitness, , , , These are dynamic coefficients for response time, resource utilization, cost, and load overload penalty, respectively. For response time, For resource utilization, For cost, Penalty for exceeding load limits; The expression for the process fitness function is: in, For scheduling fitness, , , These are dynamic coefficients for process time, resource utilization, error rate, and penalties for violations of step sequence. The process takes a long time. For error rate, Penalties will be imposed for violations of the correct procedure sequence.

[0040] The hill-climbing algorithm is used to fine-tune the traversal of neighborhood solutions to obtain the target scheduling optimization strategy. For example, during resource scheduling, the resource allocation of the current optimal solution is fine-tuned by ±1%, the neighborhood solutions are traversed, and local optima are retained (such as CPU allocation from 60% to 62%, further reducing the response time by 5ms). During process optimization, the order of steps is fine-tuned (such as postponing "log reporting"), and the neighborhood solutions are traversed to find the combination with shorter execution time.

[0041] A system for optimizing information system integration services based on artificial intelligence, such as Figure 2 As shown, it includes: Data acquisition unit: acquires past operational data and historical business data of the information system integration platform, and preprocesses the past operational data and historical business data to obtain a standard dataset; Real-time prediction unit: extracts features from the standard dataset and performs long-term and short-term temporal capture and spatial correlation to obtain long-term and short-term prediction parameters respectively; Strategy optimization generation unit: Based on long-term and short-term prediction parameters, the scheduling optimization prediction of information system integration services is performed to obtain an initial scheduling optimization strategy. The initial scheduling optimization strategy is then optimized based on the improved simulated annealing algorithm to obtain the target scheduling optimization strategy.

[0042] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor invokes the computer program in the memory to implement the content of an information system integration service optimization method based on artificial intelligence.

[0043] A storage medium storing computer-executable instructions, wherein when the computer-executable instructions are loaded and executed by a processor, the content of an information system integration service optimization method based on artificial intelligence is disclosed.

[0044] Finally, it should be noted that the above embodiments are only used to illustrate the technical methods of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical methods of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical methods to deviate from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for optimizing services of an information system integration based on artificial intelligence, characterized in that, The method comprises the following steps: S1, obtaining past running data and historical business data of an information system integration platform, and preprocessing the past running data and the historical business data to obtain a standard data set; S2, performing feature extraction on the standard data set and performing long-term and short-term time series capture and spatial correlation to obtain long-term prediction parameters and short-term prediction parameters; S3, obtaining current running data, and predicting the information system integration service based on the long-term prediction parameters and the short-term prediction parameters to obtain a predicted running condition and a predicted business condition; S4, performing information system integration service optimization based on the predicted running condition and the predicted business condition by using an improved simulated annealing algorithm to obtain a target scheduling optimization strategy.

2. The artificial intelligence-based information system integration service optimization method of claim 1, wherein, The past running data comprises server CPU usage, server memory usage, node survival state, micro-service call volume, response time, interface call link log, step dependency relationship, and process trigger volume. The historical business data comprises past resource call scenarios and corresponding past performance indicators, past process optimization schemes and past time consumption data, and past error rates.

3. The artificial intelligence-based information system integration service optimization method of claim 2, wherein, The specific content of performing feature extraction on the standard data set and performing long-term and short-term time series capture and spatial correlation to obtain long-term prediction parameters and short-term prediction parameters comprises: The standard data set comprises standard running data and standard business data; a running data curve is established with time as the horizontal axis and the standard running data as the vertical axis; a plurality of time stages of business processing scenarios are divided from the running data curve in combination with resource call scenarios in the standard business data, and scenario labels are established in the corresponding time stages; and the time stages in which the scenario labels are established are defined as ordinary time periods; an experience pool is established on the scenario labels, and the experience pool comprises past resource allocation schemes and corresponding past performance indicators, past process optimization schemes and past time consumption data, and past error rates.

4. The artificial intelligence-based information system integration service optimization method of claim 3, wherein, The specific content of performing feature extraction on the standard data set and performing long-term and short-term time series capture and spatial correlation to obtain long-term prediction parameters and short-term prediction parameters further comprises: peak features are extracted from the running data curve; time series features are extracted from the peak features based on a Prophet model to obtain business periodicity features and holiday effect features; the Prophet model is trained based on the business periodicity features and the holiday effect features to obtain long-term prediction parameters; the running data curve is divided into short-term time groups according to the time stages; different short-term prediction parameters under different time groups are obtained by training the short-term time groups by using a Transformer model.

5. The artificial intelligence-based information system integration service optimization method of claim 4, wherein, The specific content of performing information system integration service optimization based on the predicted running condition and the predicted business condition by using an improved simulated annealing algorithm to obtain a target scheduling optimization strategy is: an initial solution is generated by using a greedy algorithm in combination with the predicted running condition and the predicted business condition; an initial temperature T0, a termination temperature, and a decay coefficient are set; a target temperature is preset, a fitness function is constructed by fusing a multi-objective and a constraint penalty, and iteration is continuously updated until the target temperature is reached; a target scheduling optimization strategy is obtained by fine-tuning the neighborhood solution by using a hill climbing algorithm.

6. The artificial intelligence-based information system integration service optimization method of claim 5, wherein, The specific content of generating an initial solution by using a greedy algorithm in combination with predicted operation and predicted business conditions is: The predicted operation and predicted business conditions are analyzed to obtain node load information and service call information; The node load information is judged in combination with information system stability to obtain available nodes and dangerous nodes; The call types and corresponding call amounts of the service call information are extracted, and the call types are sorted from high to low according to the call amounts; According to the peak order range defined by the business requirements, the call types in the peak order range are marked with a peak service label; The greedy rules and their priorities are established based on the core rules of the information system; The corresponding service list to be allocated and node list are generated in combination with the service call information and node load information, and a blank resource allocation framework is generated; The services in the service list to be allocated are selected to obtain available nodes in the node list; The highest node is selected as the target node by calculating the adaptation score of the available nodes; The adaptation score = node remaining resources x service priority weight; The resource allocation is performed in combination with the target node, and the resource allocation result is saved to the blank resource allocation framework; Until the services in the service list to be allocated are all allocated to nodes and the resource quotas are set, an initial solution is obtained.

7. The artificial intelligence-based information system integration service optimization method of claim 6, wherein, The specific content of constructing a fitness function by fusing constraint punishment includes: The fitness function includes a scheduling fitness function established for resource scheduling and a process fitness function established for process optimization; The expression of the scheduling fitness function is: ； wherein, is a scheduling fitness, , , , are dynamic coefficients for response time, resource utilization, cost, load overrun penalty, respectively, is a response time, is a resource utilization, is a cost, is a load overrun penalty; The expression of the process fitness function is: wherein, is the scheduling fitness, , , are the dynamic coefficients for the process duration, resource utilization, error rate, and step order violation penalty, respectively, is the process duration, is the error rate, is the step order violation penalty.

8. A system for implementing the method for optimizing the service of integrating information systems based on artificial intelligence according to claim 1, characterized in that, It includes: A data acquisition unit acquires past operation data and historical business data of an information system integration platform, and pre-processes the past operation data and historical business data to obtain a standard data set; A real-time prediction unit extracts features from the standard data set and performs long-term and short-term time series capture and spatial correlation to obtain long-term prediction parameters and short-term prediction parameters; A strategy optimization generation unit predicts the scheduling optimization of the information system integration service based on the long-term prediction parameters and short-term prediction parameters to obtain an initial scheduling optimization strategy, and optimizes the initial scheduling optimization strategy based on an improved simulated annealing algorithm to obtain a target scheduling optimization strategy.

9. An electronic device, comprising: It includes a memory and a processor, the memory stores a computer program, and the processor calls the computer program in the memory to realize the content of the information system integration service optimization method based on artificial intelligence in any one of claims 1 to 7.

10. A storage medium, characterized by The storage medium stores computer executable instructions, and the computer executable instructions are loaded and executed by the processor to realize the content of the information system integration service optimization method based on artificial intelligence in any one of claims 1 to 7.

Images