Software system performance continuous optimization operation and maintenance method and system

Abstract

The invention belongs to the technical field of industrial system data processing operation and maintenance, and provides a software system performance continuous optimization operation and maintenance method and system.The method comprises the following steps that in the process of industrial data transmission, transmission parameter data in the current transmission time period are collected, and state analysis is conducted on the transmission parameter data; acquiring a transmission parameter with an abnormal state in the current transmission time period as a determined optimization target of the next transmission time period; dividing the duration of the next transmission time period into an acquisition time period and an analysis time period according to a proportion, obtaining transmission parameters in an abnormal state in the acquisition time period, optimizing the determined optimization target, predicting the transmission parameter data of the analysis time period based on the transmission parameter data of the acquisition time period, and outputting the predicted transmission parameter data of the analysis time period. Obtaining transmission parameter prediction data in an analysis period; and according to the transmission parameter prediction data of the analysis time period, performing similarity matching analysis with the transmission data of the historical transmission time period.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial system data processing and operation and maintenance, and specifically relates to a method and system for continuous optimization and operation and maintenance of software system performance. Background Art

[0002] In the process of industrial intelligent transformation, the industrial local area network (ILAN) serves as the core infrastructure connecting production equipment, control systems, and data centers. Its stable operation directly determines the real-time and reliability of industrial data transmission and the continuity of industrial production. As the scale of industrial equipment networking expands and the amount of data transmitted (such as real-time control instructions, sensor monitoring data, and production execution data) surges, the maintenance services and industrial data processing of the ILAN face multiple challenges.

[0003] During data processing, handling abnormal parameters is a key step in ensuring transmission stability and reliability. As networks expand in size and complexity, traditional maintenance methods are no longer able to meet the demands of modern industrial production. For example, manual inspections and troubleshooting are not only inefficient but also lack real-time and accuracy, often leading to delays in problem discovery and resolution, which in turn impacts production progress. Furthermore, the continuous optimization process lacks a process for identifying recurring abnormal parameters and a mechanism for reusing historical experience, requiring the formulation of new strategies for each processing step, increasing processing costs and time.

[0004] To this end, the present invention provides a method and system for continuously optimizing the performance of a software system. Summary of the Invention

[0005] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0006] The technical solution adopted by the present invention to solve the technical problem is: a method for continuously optimizing the performance of a software system, comprising the following steps:

[0007] During the industrial data transmission process, the transmission parameter data of the current transmission period is collected, and the transmission parameter data status is analyzed to obtain the transmission parameters with abnormal status in the current transmission period as the determination optimization target for the next transmission period;

[0008] The duration of the next transmission period is divided into a collection period and an analysis period according to a ratio, and the transmission parameters that are abnormal during the collection period are obtained. The determined optimization target is optimized, and the transmission parameter data of the analysis period is predicted based on the transmission parameter data of the collection period to obtain the transmission parameter prediction data of the analysis period.

[0009] Based on the transmission parameter prediction data of the analysis period, similarity matching analysis is performed with the transmission data of the historical transmission period to identify the historical similar transmission period, and the transmission parameters of the historical similar transmission period with abnormal status are obtained as the historical abnormal transmission parameters;

[0010] The historical abnormal transmission parameters are compared with the abnormal transmission parameters that appeared during the collection period, and repeated transmission parameters and non-repeated transmission parameters are extracted, and the repeated transmission parameters and non-repeated transmission parameters are optimized separately.

[0011] The beneficial effects of the present invention are as follows:

[0012] This invention collects industrial data transmission parameters in real time, quantifies the proportion of abnormal collection time points to identify abnormal transmission parameters, provides clear targets for subsequent optimization, reduces blind optimization, ensures that the optimization direction fits the actual transmission shortcomings, and improves the targeted optimization from the source.

[0013] The present invention divides the transmission period into collection and analysis periods to reduce resource conflicts. Real-time optimization of the collection period provides accurate data for analysis period prediction. The LSTM neural network model is used to predict the analysis period parameters, capture long-term and short-term dependencies, improve prediction accuracy, and provide advance time for potential problem handling.

[0014] The present invention matches historically similar transmission periods, reuses fine-tuned historical optimization strategies to process repeated abnormal parameters, and formulates new strategies for non-repeated abnormal parameters, thereby combining experience reuse with new problem solving, thereby continuously and comprehensively optimizing software system performance and ensuring stable and efficient industrial data transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further described below with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart of the steps of a method for continuous optimization and operation and maintenance of software system performance according to the present invention;

[0017] Figure 2 This is a logic judgment flow chart of step 1 in a method for continuous optimization and operation of software system performance according to the present invention;

[0018] Figure 3 This is an architecture diagram of an operation and maintenance system for continuous optimization of software system performance according to the present invention. DETAILED DESCRIPTION

[0019] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0020] Example 1

[0021] See also Figure 1 As shown, a method for continuously optimizing and maintaining software system performance according to an embodiment of the present invention includes the following steps:

[0022] Step 1: During the industrial data transmission process, the transmission parameter data of the current transmission period is collected, and the transmission parameter data status is analyzed to obtain the transmission parameters with abnormal status in the current transmission period as the optimization target for the next transmission period;

[0023] Step 1 mainly involves real-time monitoring and analysis of the current industrial data transmission process to identify transmission problems, provide targets for optimization of the next transmission period, and provide stage targets for continuous optimization;

[0024] In some embodiments, the duration of the transmission period is set by those skilled in the art based on experience;

[0025] Collect all nodes, transmission links and transmission protocols involved in industrial data transmission during the current transmission period;

[0026] Real-time collection of transmission parameter data for the current transmission period, wherein the transmission parameters include at least basic performance parameters, data quality parameters, and environment-related parameters;

[0027] Specifically, the basic performance parameters include at least: transmission rate, transmission delay, data packet loss rate, and number of retransmissions;

[0028] Data quality parameters: data integrity, data accuracy, and data temporal consistency;

[0029] Environmental parameters: CPU, memory usage, network bandwidth utilization, link signal strength, and device temperature of the transmission node;

[0030] Regarding the collection method of transmission data: the above transmission data is collected in real time through embedded collection modules, network sniffing tools or industrial data middle platform interfaces;

[0031] The acquisition frequency is set according to the needs of the industrial scenario. For example, if the industrial scenario requires real-time control, high-frequency acquisition is required and the acquisition frequency is set to 10ms / time; if the industrial scenario requires non-real-time monitoring, low-frequency acquisition is required and the acquisition frequency is set to 1s / time;

[0032] Preprocessing the transmission parameter data collected in real time during the current transmission period, wherein the preprocessing includes: data cleaning and storing the transmission data in a temporary database;

[0033] Exemplarily, the temporary database is the time series database InfluxDB;

[0034] Comparing the collected transmission parameter data with corresponding preset thresholds, which are set by those skilled in the art based on the service characteristics, technical specifications, and industry experience of the transmission parameter data;

[0035] For example, the industrial Ethernet in automobile manufacturing workshops requires a packet loss rate of ≤0.001% to reduce errors in welding, assembly and other processes caused by data loss;

[0036] For each transmission parameter data, according to the comparison result, the collection time point corresponding to the abnormal state is extracted as the abnormal collection time point;

[0037] For example, if the transmission rate is lower than the preset minimum rate, it is marked as insufficient rate abnormal state, and the corresponding collection time point is the abnormal collection time point; if the packet loss rate is higher than the preset maximum packet loss rate, it is marked as abnormal packet loss abnormal state, and the corresponding collection time point is the abnormal collection time point;

[0038] Count the number of abnormal collection time points and calculate the ratio with the total number of collection time points in the transmission period to obtain the proportion of abnormal collection time points;

[0039] Extract the transmission parameters whose proportion of abnormal collection time points exceeds the preset proportion limit as the abnormal transmission parameters of the current transmission period;

[0040] For example, in the case of monitoring data transmission in an industrial Ethernet network of an automobile manufacturing plant, the transmission period is set to 1 hour and the acquisition frequency is 1 time / second (i.e., the total number of acquisition time points is 3600);

[0041] The data packet loss rate has a preset threshold of 0.0001%. If the data packet loss rate exceeds the preset threshold, the corresponding collection time point is marked as an abnormal collection time point. The abnormal collection time points are counted for 200 time points, and the proportion of abnormal collection time points is 0.056, which exceeds the preset limit (5%). Therefore, the data packet loss rate is determined to be an abnormal transmission parameter.

[0042] Obtain all abnormal transmission parameters of the current transmission period and the parameter types corresponding to the abnormal transmission parameters to determine the optimization target for the next transmission period;

[0043] Since this embodiment continuously optimizes and maintains the performance of the software system, the parameter types corresponding to all abnormal transmission parameters that appear in the current transmission period are the targets that need to be optimized in the next transmission period, because abnormal transmission parameters directly reflect the shortcomings of data transmission in the current transmission period.

[0044] This step has at least the following effects: by monitoring transmission parameters in real time and quantifying the proportion of abnormalities, abnormal transmission parameters in the current period can be identified, providing clear targets for optimization in the next period, reducing blind optimization, and ensuring that the optimization direction is aligned with actual transmission shortcomings;

[0045] Step 2: Divide the duration of the next transmission period into a collection period and an analysis period in proportion, obtain the transmission parameters that are abnormal during the collection period, optimize the determined optimization target, and then predict the transmission parameter data for the analysis period based on the transmission parameter data during the collection period;

[0046] In some embodiments, the duration of the next transmission period is divided proportionally into a collection period and an analysis period;

[0047] The collection period is used to collect transmission parameter data and optimize the optimization target during the collection period, providing a basis for predicting transmission parameter data during the analysis period, and obtaining the parameter type of abnormal transmission parameter status during the collection period;

[0048] This can increase the reliability and accuracy of analysis and prediction, reduce prediction deviations caused by insufficient problems, and through division, the collection and analysis environments can be focused separately, reducing resource usage conflicts and improving the efficiency of data processing and optimization within the corresponding transmission period;

[0049] The ratio of the collection period to the analysis period is set to 3:7, that is, 30% of the time is divided into the collection period and 70% of the time is divided into the analysis period;

[0050] During the collection period, transmission parameter data is collected to obtain abnormal transmission parameters during the collection period;

[0051] It should be noted that the method for obtaining the abnormal transmission parameters of the collection period is the same as the method for obtaining the abnormal transmission parameters of the current transmission period described in the aforementioned step 1;

[0052] During the acquisition period, the optimization target is optimized. The specific optimization strategy is set based on the transmission parameter type of the optimization target.

[0053] Optimization strategies for abnormal transmission parameters are well understood by those skilled in the art, for example:

[0054] If the optimization target is determined to be the packet loss rate among the basic performance parameters, the optimization strategies include, but are not limited to: dynamically adjusting the packet size, splitting the packet into 50% of its original size when the network link signal strength is weak, reducing the pressure of a single transmission; enabling the link redundancy mechanism, automatically switching to the backup link for data transmission when the packet loss rate of the primary link exceeds 0.0005%;

[0055] If the optimization target is determined to be the CPU usage of the transmission node in the environment-related parameters, the optimization strategy includes but is not limited to: prioritizing the processes of the transmission node, raising the priority of data transmission-related processes to the highest, and limiting the CPU usage of non-critical processes;

[0056] Based on any transmission parameter data and the data sequence of the acquisition period, the data sequence of the analysis period is predicted. The specific process is as follows:

[0057] The transmission parameter data of the analysis period is predicted using the LSTM neural network model;

[0058] It will be understood by those skilled in the art that based on the transmission parameter data of the acquisition period, the time series features are extracted and used as the input of the LSTM neural network model, and the output is the transmission parameter data of the analysis period;

[0059] Time series feature extraction: Determine the time step of the input sequence based on the acquisition frequency, that is, the length of historical data input to the model at one time;

[0060] For example, if the acquisition frequency is 10ms / time, the first 100 time steps are taken as input to capture short-term fluctuation characteristics; if the acquisition frequency is 1s / time, the first 60 time steps are taken as input;

[0061] Taking multivariate time series data as input, each time step contains all transmission parameters, forming an input matrix with the dimension of "time step × number of parameters";

[0062] For example, if the acquisition frequency is 1s / time and contains 5 transmission parameters, the input matrix is ​​a 60×5 two-dimensional array;

[0063] Construction and training of LSTM neural network model, the structure includes input layer, LSTM hidden layer, fully connected layer and output layer;

[0064] The input layer is a time series feature matrix with the same dimensions as the input sequence. Two to three hidden layers are set, with the number of neurons in each layer adjusted based on parameter complexity (e.g., 64 neurons in the first layer and 32 neurons in the second layer). Gating mechanisms (input gate, forget gate, and output gate) are used to capture long-term and short-term dependencies (e.g., the lagged correlation between network bandwidth fluctuations and packet loss rate). The fully connected layer maps the feature vectors output by the LSTM to the predicted dimensions for the analysis period. The output layer uses a linear activation function to directly output the normalized values ​​of the predicted transmission parameters.

[0065] The data from the historical collection period and the actual data from the corresponding analysis period are used as sample pairs (for example, over the past 30 days, the data from the daily collection period is used as input, and the data from the current analysis period is used as the label). The mean squared error (MSE) is used to measure the difference between the predicted value and the actual value. The Adam optimizer is used with a learning rate set to 0.001, and the model parameters are updated iteratively through backpropagation until the loss function converges.

[0066] The last complete time series sequence after preprocessing during the acquisition period (e.g., the parameter data of the last 60 time steps) is input into the trained LSTM model. The model generates a transmission parameter prediction data sequence for the analysis period based on the learned time series patterns.

[0067] The normalized prediction value output by the model is converted into actual physical quantities using the following formula to obtain the predicted value of the transmission parameters at each time point in the analysis period (e.g., the transmission delay in the analysis period is predicted to be 20ms±5ms, the packet loss rate is 0.0005%, etc.);

[0068] This step has at least the following effects: by dividing the next transmission period into a collection period and an analysis period in proportion, the collection and analysis processes are each focused, resource usage conflicts are reduced, and the efficiency of data processing and optimization is improved;

[0069] Real-time optimization of the determined optimization target during the collection period provides more accurate basic data for parameter prediction during the analysis period, reducing prediction errors caused by data deviations;

[0070] The LSTM neural network model is used to predict the transmission parameters of the analysis period based on the time series data of the collection period. This can capture the long-term and short-term dependencies of the parameters, improve the accuracy and foresight of the parameter prediction during the analysis period, and provide a reliable prediction basis for subsequent historical similarity matching.

[0071] Step 3: Based on the predicted transmission parameter data of the analysis period, perform similarity matching analysis with the transmission data of the historical transmission period, identify the historical similar transmission period, and obtain the transmission parameters of the historical similar transmission period with abnormal status as the historical abnormal transmission parameters;

[0072] In some embodiments, obtaining a transmission parameter prediction data sequence for an analysis period;

[0073] Perform sequence alignment and normalization on the transmission parameter prediction data sequence of the analysis period and the transmission data of any historical transmission period;

[0074] For example, if the analysis period is 70 minutes (with a collection frequency of 1 time / second), the prediction sequence contains 4200 time steps of transmission parameter data. The historical sequence needs to select historical period data with a duration of 70 minutes and a consistent collection frequency to ensure that the sequence length is 4200 time steps.

[0075] Based on the combination of the transmission parameter prediction data sequence and the transmission data of any historical transmission period, similarity analysis is performed based on the overall trend. The process is as follows:

[0076] Extracting statistical features of the transmission parameter prediction data sequence and the transmission data of the historical transmission period respectively, wherein the statistical features include at least: the mean, variance, peak value, valley value, and trend slope of the transmission parameter;

[0077] Convert statistical features into feature vectors, with each sequence corresponding to a feature vector;

[0078] The distance between two feature vectors is calculated using the Euclidean distance calculation formula, and the distance value is used to represent the similarity;

[0079] The smaller the distance value, the higher the similarity; the distance threshold is set by those skilled in the art based on experience;

[0080] Extracting historical transmission periods whose distance values ​​are less than a distance threshold as historical similar transmission periods;

[0081] Extract all abnormal transmission parameters corresponding to similar historical transmission periods, perform deduplication processing, and then construct a set of historical abnormal transmission parameters;

[0082] For example, assuming that the analysis period is set to 70 minutes and the acquisition frequency is 1 time / second, the transmission parameter prediction data series during the analysis period contains 4200 time steps (covering parameters such as transmission rate, packet loss rate, and CPU usage);

[0083] Filter out historical transmission period data with a duration of 70 minutes and a collection frequency of 1 time per second from the historical database (e.g., transmission data from 9:00 to 10:10 on July 10, 2025), form a historical sequence of 4200 time steps, and align it with the predicted sequence.

[0084] After normalizing both, statistical features are extracted:

[0085] In the prediction sequence, the transmission rate has a mean of 50 Mbps, a variance of 5, a peak of 60 Mbps, a valley of 40 Mbps, and a trend slope of 0.02 (slowly increasing). The packet loss rate has a mean of 0.0003% and a variance of 0.0001%.

[0086] In the historical series, the transmission rate has a mean of 48 Mbps, a variance of 6, a peak of 58 Mbps, a valley of 39 Mbps, and a trend slope of 0.015; the packet loss rate has a mean of 0.0004% and a variance of 0.00012%.

[0087] Convert the above statistical features into feature vectors:

[0088] The predicted sequence vector is [50, 5, 60, 40, 0.02, 0.0003%, 0.0001%];

[0089] The historical sequence vector is [48, 6, 58, 39, 0.015, 0.0004%, 0.00012%];

[0090] The distance value calculated by the Euclidean distance formula is 3.2;

[0091] If the preset distance threshold is 5, the historical period is determined to be a historical similar transmission period. Assuming that the abnormal transmission parameters of this period are transmission rate and packet loss rate, and the abnormal parameters of another similar period are transmission rate and CPU occupancy, the historical abnormal transmission parameter set constructed after deduplication is: {transmission rate, packet loss rate, CPU occupancy};

[0092] This step has at least the following effects: by similarly matching the predicted parameter sequence of the analysis period with the historical transmission period data, it is possible to identify historically similar transmission periods and extract abnormal transmission parameters in these periods. This fully utilizes historical operation and maintenance experience, provides historical data support for the formulation of subsequent continuous optimization strategies, and enhances the predictability and effectiveness of optimization.

[0093] Step 4: Compare historical abnormal transmission parameters with abnormal transmission parameters that appeared during the collection period, extract repeated transmission parameters and non-repeated transmission parameters, use the non-repeated transmission parameters as the optimization target for the analysis period, and optimize the repeated transmission parameters based on the historical optimization strategy;

[0094] In some embodiments, abnormal transmission parameters identified during the acquisition period are obtained and compared with a set of historical abnormal transmission parameters; repeated transmission parameters and non-repeated transmission parameters are extracted respectively;

[0095] It should be noted that both repeated transmission parameters and non-repeated transmission parameters refer to abnormal transmission parameters identified during the collection period;

[0096] Exemplarily, the abnormal transmission parameters during the collection period are: transmission delay, memory usage, CPU usage;

[0097] The abnormal transmission parameters of historical similar transmission period 1 are: transmission delay, packet loss rate;

[0098] The abnormal transmission parameters of historical similar transmission period 2 are: transmission delay, CPU usage;

[0099] The abnormal transmission parameters of historical similar transmission period 3 are: packet loss rate, network bandwidth;

[0100] After deduplication, the historical abnormal transmission parameter set includes: transmission delay, packet loss rate, CPU usage, and network bandwidth;

[0101] The parameters for repeated transmission are: transmission delay, CPU occupancy; the parameters for non-repeated transmission are: memory occupancy;

[0102] For non-repeated transmission parameters, which are the optimization targets during the analysis period, the specific optimization strategy is set based on the type of transmission parameters that determine the optimization target;

[0103] The optimization strategy for abnormal transmission parameters is well understood by those skilled in the art, for example,

[0104] For parameter memory usage that only appears during the collection period,

[0105] Analyze the root cause of the anomaly, such as memory leaks in a process during the collection period, and formulate immediate optimization strategies, such as forcibly releasing redundant memory and limiting the memory quota of the process;

[0106] For repeated transmission parameters, call the corresponding historical optimization strategy, fine-tune it in combination with the current scenario, and then execute it.

[0107] For example, for repeated parameter transmission delays, the optimization strategy for similar historical periods is to dynamically adjust the packet size to 60% of the original size and enable link load balancing. Based on the environmental parameters of the current collection period (e.g., link signal strength is 10% lower than in historical scenarios), the adjustment strategy is to compress the packet size to 50% of the original size and prioritize switching to backup links to enhance strategy adaptability.

[0108] This step has at least the following effects: For recurring abnormal transmission parameters, by retrieving historical optimization strategies and fine-tuning them based on the current scenario, we reuse historical experience, improve strategy adaptability, reduce the cost of repeated decision-making, and improve optimization efficiency.

[0109] For non-repeated abnormal parameters that only appear during the acquisition period, use them as new optimization targets during the analysis period and formulate targeted strategies to promptly address emerging problems and ensure that all shortcomings are covered by the optimization.

[0110] By classifying and processing repeated and non-repeated parameters, the combination of experience reuse and new problem solving is achieved, so that the performance of the software system can be comprehensively and continuously optimized.

[0111] Example 2

[0112] Based on the same inventive concept as the method for continuous optimization and operation of software system performance in the aforementioned embodiment, Figure 2 As shown, the present application provides a software system performance continuous optimization operation and maintenance system, wherein the system specifically includes:

[0113] Abnormal parameter identification module: During the process of industrial data transmission, the transmission parameter data of the current transmission period is collected and the transmission parameter data status is analyzed. The transmission parameters with abnormal status in the current transmission period are obtained as the optimization targets for the next transmission period.

[0114] The execution process is as follows: Industrial data transmission parameters within the current transmission period are collected in real time, pre-processed, and compared with preset thresholds. Abnormal collection time points are marked and their proportions are calculated. Abnormal transmission parameters with a proportion exceeding the threshold are screened out and determined as optimization targets for the next transmission period, providing a clear direction for subsequent optimization.

[0115] Time period prediction and optimization module: This module divides the duration of the next transmission period into a collection period and an analysis period according to the proportion, obtains the transmission parameters that are abnormal during the collection period, optimizes the determined optimization target, and then predicts the transmission parameter data for the analysis period based on the transmission parameter data during the collection period.

[0116] The execution process is as follows: the next transmission period is divided into a collection period and an analysis period in a ratio of 3:7. During the collection period, transmission parameters are collected and abnormal parameters are identified. At the same time, targeted optimization is implemented for the optimization target determined in step 1. Based on the parameter data sequence during the collection period, the LSTM neural network model is used to extract time series features and train, and the transmission parameter data sequence during the analysis period is predicted to provide a basis for subsequent similarity matching.

[0117] Historical similarity matching module: Based on the predicted transmission parameter data of the analysis period, similarity matching analysis is performed with the transmission data of the historical transmission period to identify the historical similar transmission period and obtain the transmission parameters with abnormal status in the historical similar transmission period as the historical abnormal transmission parameters;

[0118] The execution process is as follows: Using the transmission parameter prediction data sequence of the analysis period as a benchmark, the sequence is aligned and normalized with the historical transmission period data. Feature vectors are constructed by extracting statistical features such as mean and variance. The Euclidean distance is calculated to measure similarity. Historically similar transmission periods with distance values ​​less than a threshold are screened out. Abnormal transmission parameters in these periods are aggregated and deduplicated to form a set of historical abnormal transmission parameters.

[0119] Abnormal parameter comparison and optimization module: compares historical abnormal transmission parameters with abnormal transmission parameters that appeared during the collection period, extracts repeated transmission parameters and non-repeated transmission parameters, uses non-repeated transmission parameters as the optimization target during the analysis period, and optimizes repeated transmission parameters based on historical optimization strategies;

[0120] The execution process is as follows: the abnormal transmission parameters identified during the collection period are compared with the historical abnormal transmission parameter set, and repeated parameters (abnormal parameters common to both) and non-repeated parameters (abnormal parameters that only appear during the collection period) are extracted. The historical optimization strategy is reused and fine-tuned for the repeated parameters, and the non-repeated parameters are used as the new optimization target for the analysis period. Targeted new strategies are formulated to achieve continuous performance optimization.

[0121] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for continuous optimization and maintenance of software system performance, characterized by: The following steps are involved: During the industrial data transmission process, the transmission parameter data of the current transmission period is collected, and the transmission parameter data status is analyzed to obtain the transmission parameters with abnormal status in the current transmission period as the determination optimization target for the next transmission period; The duration of the next transmission period is divided into a collection period and an analysis period according to a ratio, and the transmission parameters that are abnormal during the collection period are obtained. The determined optimization target is optimized, and the transmission parameter data of the analysis period is predicted based on the transmission parameter data of the collection period to obtain the transmission parameter prediction data of the analysis period. Based on the transmission parameter prediction data of the analysis period, similarity matching analysis is performed with the transmission data of the historical transmission period to identify the historical similar transmission period, and the transmission parameters of the historical similar transmission period with abnormal status are obtained as the historical abnormal transmission parameters; The historical abnormal transmission parameters are compared with the abnormal transmission parameters that appeared during the collection period, and repeated transmission parameters and non-repeated transmission parameters are extracted, and the repeated transmission parameters and non-repeated transmission parameters are optimized separately.

2. A method for continuous optimization and operation of software system performance according to claim 1, characterized in that: The process of obtaining the transmission parameters whose status is abnormal during the current transmission period is as follows: Set the duration of the transmission period, collect the transmission parameter data of the current transmission period in real time, and perform comparative analysis to obtain abnormal collection time points; Count the number of abnormal collection time points and calculate the ratio with the total number of collection time points in the transmission period to obtain the proportion of abnormal collection time points; The transmission parameters with an excessive proportion of abnormal collection time points are extracted as the abnormal transmission parameters of the current transmission period.

3. The method for continuous optimization and operation of software system performance according to claim 2, characterized in that: The process of obtaining the abnormal collection time point is as follows: Compare the collected transmission parameter data with the corresponding preset thresholds respectively, and for each transmission parameter data, extract the collection time point corresponding to the abnormal state according to the comparison result as the abnormal collection time point; 4. The method for continuous optimization and operation of software system performance according to claim 1, characterized in that: The acquisition process of the transmission parameters of the abnormal state during the acquisition period is as follows: The duration of the next transmission period is divided into a collection period and an analysis period in a ratio of 3:7; During the collection period, transmission parameter data is collected to obtain abnormal transmission parameters during the collection period.

5. A method for continuous optimization and operation of software system performance according to claim 4, characterized in that: The acquisition process of the abnormal transmission parameters during the acquisition period is as follows: Real-time acquisition of transmission parameter data during the collection period; Compare the transmission parameter data of the collection period with the corresponding preset thresholds respectively, and for each transmission parameter data, extract the collection time point corresponding to the abnormal state according to the comparison result as the abnormal collection time point; Count the number of abnormal collection time points and calculate the ratio with the total number of collection time points in the collection period to obtain the proportion of abnormal collection time points; The transmission parameters with an excessive proportion of abnormal collection time points are extracted as the abnormal transmission parameters of the collection period.

6. The method for continuous optimization and operation of software system performance according to claim 1, characterized in that: The acquisition process of the transmission parameter prediction data during the analysis period is as follows: The transmission parameter data of the analysis period is predicted through the LSTM neural network model. The transmission parameter data of the collection period is used to extract the time series features as the input of the LSTM neural network model, and the transmission parameter prediction data of the analysis period is output.

7. The method for continuous optimization and operation of software system performance according to claim 1, characterized in that: The process of identifying historically similar transmission periods is as follows: Based on a combination of a transmission parameter prediction data sequence and transmission data of any historical transmission period, statistical features of the transmission parameter prediction data sequence and the transmission data of the historical transmission period are extracted respectively; Convert statistical features into feature vectors, with each sequence corresponding to a feature vector; The distance between two feature vectors is calculated using the Euclidean distance calculation formula, and the distance value is used to represent the similarity; The historical transmission periods whose distance values ​​are less than the distance threshold are extracted as the historical similar transmission periods.

8. The method for continuous optimization and operation of software system performance according to claim 1, characterized in that: The extraction process of the repeated transmission parameters and the non-repeated transmission parameters is as follows: The abnormal transmission parameters identified during the collection period are obtained and compared with the historical abnormal transmission parameter set; repeated transmission parameters and non-repeated transmission parameters are extracted respectively.

9. The method for continuous optimization and operation of software system performance according to claim 1, characterized in that: described The non-repeated transmission parameters are used as the optimization targets during the analysis period, and the repeated transmission parameters are optimized based on the historical optimization strategy.

10. A software system performance continuous optimization operation and maintenance system, characterized by: The system is used to execute the method according to any one of claims 1 to 9, and the system comprises: Abnormal parameter identification module: During the process of industrial data transmission, the transmission parameter data of the current transmission period is collected and the transmission parameter data status is analyzed. The transmission parameters with abnormal status in the current transmission period are obtained as the optimization targets for the next transmission period. Time period prediction and optimization module: This module divides the duration of the next transmission period into a collection period and an analysis period according to the proportion. It obtains the transmission parameters that are abnormal during the collection period, optimizes the determined optimization target, and then predicts the transmission parameter data for the analysis period based on the transmission parameter data during the collection period to obtain the transmission parameter prediction data for the analysis period. Historical similarity matching module: Based on the transmission parameter prediction data of the analysis period, similarity matching analysis is performed with the transmission data of the historical transmission period to identify the historical similar transmission period and obtain the transmission parameters of the historical similar transmission period with abnormal status as the historical abnormal transmission parameters; Abnormal parameter comparison and optimization module: compares historical abnormal transmission parameters with abnormal transmission parameters that appeared during the collection period, extracts repeated transmission parameters and non-repeated transmission parameters, and optimizes them separately based on the repeated transmission parameters and non-repeated transmission parameters.

Images