Gas turbine high-capacity historical data analysis method

Abstract

The invention discloses a gas turbine high-capacity historical data analysis method which comprises the following steps: configuring storage mode parameters through a data acquisition module, including a change rate threshold value, a dead zone threshold value and a storage period parameter; the data preprocessing module is used for preliminarily filtering the gas turbine operation data collected in real time based on the parameters; storing the filtered data in a classified manner according to the measuring points, performing secondary compression on the time sequence data by adopting a second-order compression algorithm based on temporal correlation and spatial correlation, and circularly covering and storing the compressed data in a time sequence database; in response to a query condition input by a user, obtaining historical data of the target measurement point from the time sequence database through a multi-node concurrent reading mechanism, and performing data granularity optimization based on a downsampling algorithm; the processed data is subjected to normalization conversion, a multi-axis dynamic visualization chart with a time synchronization coordinate system is generated, and the chart supports superposition comparative analysis and interactive operation of multi-measuring-point data; the method is high in data compression rate and small in storage space occupation.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbine operation and maintenance software, and particularly to a method for analyzing large-capacity historical data of a gas turbine. Background Art

[0002] As a device widely used in the fields of energy, transportation, and industry, the historical data generated during the operation of a gas turbine is of great significance for performance evaluation, fault prediction, and maintenance strategy formulation. In fields such as gas turbine power generation, the control system is usually required to operate stably for a long time, and the planned shutdown and maintenance cycle is usually relatively long, generally 1 to 3 months, which will bring a series of problems such as the storage and management of a large amount of data and the playback and analysis of a large time span.

[0003] Due to the complexity and enormity of gas turbine operation data, traditional data analysis methods often have difficulty effectively processing these large-capacity historical data and cannot deeply mine the valuable information therein. Summary of the Invention

[0004] Object of the Invention: The object of the present invention is to provide a method for analyzing large-capacity historical data of a gas turbine to solve the problem that traditional data analysis methods often have difficulty effectively processing these large-capacity historical data.

[0005] Technical Solution: A method for analyzing large-capacity historical data of a gas turbine according to the present invention includes the following steps: S1: Configure storage mode parameters through a data acquisition module, including a change rate threshold, a dead zone threshold, and a storage period parameter; the data preprocessing module preliminarily filters the real-time collected gas turbine operation data based on the parameters. S2: Classify and store the filtered data by measurement points, perform secondary compression on the time series data using a second-order compression algorithm based on time correlation and space correlation, and circularly overwrite and store the compressed data in a time series database. S3: In response to the query conditions input by the user, obtain the historical data of the target measurement point from the time series database through a multi-node concurrent reading mechanism, and perform data granularity optimization processing based on a downsampling algorithm. S4: Perform normalization conversion on the processed data to generate a multi-axis dynamic visualization chart with a time-synchronized coordinate system, and the chart supports the overlay comparison analysis and interactive operation of multi-measurement point data.

[0006] Further, in step S1, the change rate threshold is the critical value for triggering storage by percentage change, and the dead zone threshold is the critical value for triggering storage by absolute value change. When the change amount of the collected point value exceeds both the change rate threshold and the dead zone threshold, a storage operation is performed.

[0007] Further, in step S2, the second-order compression algorithm includes time dimension compression and space dimension compression; the time dimension compression adopts piecewise linear fitting. According to the local linear characteristics of the time series data, the data corresponding to continuous timestamps is divided into multiple linear segments, and only the starting point, slope, and ending point of each segment are stored; the space dimension compression adopts vectorized preprocessing. After converting the data into binary encoding, redundant bits are eliminated through XOR operation, and then the result is compressed again.

[0008] Further, in step S3, the downsampling algorithm is the LTTB algorithm.

[0009] Further, in step S4, the normalization process includes aligning the timestamps of the data on the x-axis. Based on the measured point data with clear timestamps, the measured point data of other points is supplemented by interpolation; for the y-axis, the full-scale display of the data is planned, and the minimum and maximum values of the measured point data curve in this segment are used as the minimum and maximum values of the y-axis coordinate system of this measured point data.

[0010] Further, the storage strategy of the data storage module includes storing by measured point classification, compressing and storing using the second-order compression algorithm, and circular overwriting storage.

[0011] Further, the data retrieval module reads data in a multi-node concurrent manner.

[0012] Further, the visualization analysis component module intuitively compares and analyzes the data of different measured points in the form of a multi-axis curve graph.

[0013] Further, the multi-axis curve graph supports functions such as zooming, panning, and color change.

[0014] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: multiple storage strategies, high data compression rate, and small storage space occupation; combined with multi-node concurrent reading and data downsampling algorithm, the data loading speed is fast, and the data trend details are not lost; using a normalized multi-axis curve, intuitive comparison and analysis of data can be carried out under the same time section, and the data analysis efficiency is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is the flow chart of the present invention; Figure 2 is the schematic diagram of the storage strategy of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0016] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0017] As Figure 1 shown, the embodiment of the present invention provides a CFG reconstruction method based on a PPC binary file, including the following steps: Step S1: The data acquisition module configures the storage mode, including the change rate, dead zone, and storage period; the data preprocessing module performs preliminary filtering in real time according to the configuration information; Among them, the change rate is used to judge the percentage threshold range of the change of the acquisition point value. If it exceeds the threshold, it is stored; otherwise, it is ignored. The dead zone is used to judge a threshold range of a value set for whether the acquisition point value changes. If it exceeds the threshold, it is stored; otherwise, it is ignored.

[0018] Step S2: The data storage module stores the data classified by measurement points. After compressing the data through the second-order compression algorithm, it is stored in the time series database; when the data exceeds the storage period, cyclic overwrite storage is performed; Among them, the second-order compression algorithm refers to the secondary compression performed by using different compression algorithms for the time correlation and space correlation of time series data respectively.

[0019] Step S3: The data retrieval module obtains the manually input time period information and measurement point selection information, and obtains the data of the selected measurement points within this time period from the time series database; the data retrieval module reads the data in a multi-node concurrent manner, and after calculation by the downsampling algorithm, it transfers the data to the visualization analysis component; Among them, the downsampling algorithm is the LTTB algorithm, which ensures that the amount of the sampled data set is further reduced without losing the trend, reduces the bandwidth consumption during the data transmission process, and at the same time, the small sample data volume can improve the efficiency of data alignment and result display.

[0020] Step S4: After the visualization analysis component module normalizes the data, it plots the measurement point data on the same multi-axis curve graph with curves of different colors (the x-axis is time, and the y-axis is multi-axis); the multi-axis curve graph supports zooming, panning, and color change, which is convenient for detailed data comparison and analysis; Among them, the normalization process includes two aspects: aligning the timestamps of the data on the x-axis, using the measurement point data with clear timestamps as the benchmark, and filling in the other measurement point data through interpolation; planning the full-scale display of the data on the y-axis, using the minimum and maximum values of the measurement point data curve in this section as the minimum and maximum values of the coordinate system of the y-axis of this measurement point data.

Claims

1. A method for analyzing large-capacity historical data of a gas turbine, characterized in that, It includes the following steps: S1: Configure the storage mode parameters through the data acquisition module, including the change rate threshold, dead zone threshold, and storage period parameters; The data preprocessing module performs preliminary filtering on the real-time collected gas turbine operation data based on the above parameters; S2: Classify and store the filtered data by measurement point, perform secondary compression on the time series data using a second-order compression algorithm based on time correlation and space correlation, and circularly overwrite and store the compressed data in the time series database; S3: In response to the query conditions input by the user, obtain the historical data of the target measurement point from the time series database through a multi-node concurrent reading mechanism, and perform data granularity optimization processing based on the downsampling algorithm; S4: Perform normalization conversion on the processed data to generate a multi-axis dynamic visualization chart with a time synchronization coordinate system. The chart supports the overlay comparison analysis and interactive operation of multi-measurement point data.

2. The method for analyzing large-capacity historical data of a gas turbine according to claim 1, wherein In step S1, the change rate threshold is the critical value for triggering storage by percentage change, and the dead zone threshold is the critical value for triggering storage by absolute value change. The storage operation is executed when the change amount of the collected point value exceeds both the change rate threshold and the dead zone threshold.

3. A method for analyzing large-capacity historical data of a gas turbine according to claim 1, characterized in that, In step S2, the second-order compression algorithm includes time dimension compression and space dimension compression; the time dimension compression uses piecewise linear fitting. According to the local linear characteristics of the time series data, the data corresponding to consecutive timestamps is divided into multiple linear segments, and only the starting point, slope, and ending point of each segment are stored; the space dimension compression uses vectorized preprocessing. After converting the data into binary encoding, redundant bits are eliminated through exclusive OR operations, and then the result is further compressed.

4. A method for analyzing large-capacity historical data of a gas turbine according to claim 1, characterized in that In step S3, the downsampling algorithm is the LTTB algorithm.

5. A method for analyzing large-capacity historical data of a gas turbine according to claim 1, characterized in that, In step S4, the normalization process includes aligning the timestamps of the x-axis data. Based on the measurement point data with clear timestamps, the data of other measurement points are filled in by interpolation; for the y-axis, the full-scale display of the data is planned, and the minimum and maximum values of the measurement point data curve in this segment are used as the minimum and maximum values of the y-axis coordinate system of this measurement point data.

6. The method for analyzing large-capacity historical data of a gas turbine according to claim 1, wherein The storage strategy of the data storage module includes classification storage by measurement point, compression storage using the second-order compression algorithm, and circular overwrite storage.

7. A large-capacity historical data analysis method for a gas turbine according to claim 1, characterized in that The data retrieval module reads data in a multi-node concurrent manner.

8. A method for analyzing large-capacity historical data of a gas turbine according to claim 1, characterized in that The visualization analysis component module visually compares the data of different measurement points in the form of a multi-axis curve graph.

9. A method for analyzing large-capacity historical data of a gas turbine according to claim 8, characterized in that, The multi-axis curve graph supports functions such as zooming, panning, and color change.

Images