Industrial process steady-state detection method, system, medium, equipment and terminal

Abstract

The present invention belongs to the technical field of industrial steady-state detection, and discloses an industrial process steady-state detection method, system, medium, equipment and terminal. A Gaussian filter is used to filter the target sequence; the first-order difference of the filtered target sequence is calculated, and the direction and degree of data change are reflected by the differential value; a Gaussian mixture model is used to model the differential value sequence to obtain an estimated value of the fluctuation standard deviation of the target parameter in the steady state; and a CUSUM control chart is used to perform steady-state discrimination detection of industrial process time series data of coal-fired power units. Based on a large amount of historical process data, the present invention estimates the fluctuation standard deviation of the target operating condition parameter in the steady state from a global perspective, and the larger the amount of historical data, the stronger the robustness of the method. The comparison results show that the adaptive CUSUM control chart industrial process steady-state detection method of the present invention has better effect, can detect steady-state sequences of continuous periods, and is more suitable for scenarios with frequent switching of operating conditions.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial steady-state detection, and in particular relates to an industrial process steady-state detection method, system, medium, equipment and terminal. Background Art

[0002] At present, coal-fired power units undertake the main peak-shaving tasks in my country's power production. Frequent operating condition switching occurs during the operation of the units, causing the unit process data to exhibit a "steady-state-unsteady-state" mode. When the operating conditions are adjusted, there are different degrees of time delays between different state parameters, resulting in differences in the information representation of steady-state process data and unsteady-state process data. Therefore, steady-state discrimination is of great significance to the modeling process that is sensitive to operating condition changes.

[0003] There is widespread measurement noise interference and sudden system fluctuations in industrial process data measurement, which results in some data points exceeding the allowable fluctuation level even at the specified operating level, causing the steady state of judgment to be continuously destroyed by these noise points and affecting the integrity.

[0004] Through the above analysis, the problems and defects of the existing technology are as follows: the measurement noise interference and sudden system fluctuations in the existing industrial process data measurement methods cause the operating parameters to have data points that exceed the allowable fluctuation level even at the specified operating level, resulting in the destruction of the steady state judgment and affecting the integrity. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a method, system, medium, equipment and terminal for steady-state detection of an industrial process, and in particular, relates to a method, system, medium, equipment and terminal for steady-state detection of industrial process data of a coal-fired power unit based on an adaptive cumulative sum control chart (CUSUM).

[0006] The present invention is implemented as follows: an industrial process steady-state detection method, the industrial process steady-state detection method comprising: using a Gaussian filter to perform Gaussian filtering on a historical working process data set of a coal-fired power unit to obtain a filtered historical working process data set of the coal-fired power unit; performing differential processing on the filtered historical working process data set of the coal-fired power unit to obtain a first-order differential data set; using a Gaussian mixture model (GMM) to model the first-order differential data set to obtain an estimated value of the standard deviation of the sequence fluctuation under steady state; using an adaptive CUSUM control chart to process the continuous working process data of the coal-fired power unit to be processed to obtain a steady-state judgment result.

[0007] Furthermore, the industrial process steady-state detection method also includes:

[0008] A Gaussian filter is used to filter the target sequence; the first-order difference of the filtered target sequence is calculated, and the difference value reflects the direction and degree of data change; the Gaussian mixture model is used to model the difference value sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state sequence, and then the estimated value of the fluctuation standard deviation of the target parameter in the steady state is obtained; finally, the CUSUM control chart is used to perform steady-state discrimination detection on the industrial process time series data of coal-fired power units to obtain the steady-state discrimination result.

[0009] Further, the industrial process steady-state detection method comprises the following steps:

[0010] Step 1, data filtering: Gaussian filtering is performed on the historical process data set using a Gaussian filter;

[0011] Step 2: Obtain the volatility distribution of the steady-state sequence in the target data, and based on the sequence difference value, select a mixed Gaussian model to describe the volatility distribution of the sequence in the steady state;

[0012] Step three, based on the mean and fluctuation standard deviation of the detection subsequence, the steady-state judgment of the industrial process time series data of the coal-fired power unit is carried out based on the cumulative sum control chart.

[0013] Furthermore, in step 2, the Gaussian mixture model is used to estimate the standard deviation of the target sequence in steady state. In coal-fired power units, the change trend of operating parameters is divided into three types: rising, steady state and falling, and the change trend is the cumulative result of the first-order difference of the sequence; the Gaussian mixture model is used to model the distribution of the first-order difference value of the target sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state subsequence in the sequence, and then the estimated value of the standard deviation of the operating parameters in steady state is obtained.

[0014] Furthermore, in step three, the steady-state subsequences in the target sequence are identified by traversal, and the concept of "growing steady-state subsequences" is proposed to estimate the mean of the current steady-state subsequence. Each steady-state subsequence is generated by a point, and the next point in the sequence is identified by the mean of the current subsequence; if it belongs to the current steady-state subsequence, the current subsequence grows, the current identification point is added to the current steady-state subsequence, and the mean is updated, and the subsequent data points are identified; if it does not belong to the current steady-state subsequence, the current steady-state subsequence stops growing, and the current identification point is used as the starting point of the next steady-state subsequence to identify the subsequent data points; until all the data points in the target sequence are traversed, the steady-state identification process ends.

[0015] Furthermore, during the entire traversal process, the mean of the current steady-state subsequence is updated as the current steady-state subsequence grows.

[0016] Another object of the present invention is to provide an industrial process steady-state detection system using the industrial process steady-state detection method, the industrial process steady-state detection system comprising:

[0017] A data filtering module, used for performing Gaussian filtering on the historical process data set using a Gaussian filter;

[0018] The sequence difference module is used to obtain the fluctuation distribution of the steady-state sequence in the target data. Based on the sequence difference value, a mixed Gaussian model is preferably used to describe the fluctuation distribution of the sequence in the steady state.

[0019] The steady-state discrimination module is used to perform steady-state discrimination of industrial process time series data of coal-fired power units based on the mean and fluctuation standard deviation of the detection subsequence and the cumulative control chart.

[0020] Another object of the present invention is to provide a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the industrial process steady-state detection method.

[0021] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the industrial process steady-state detection method.

[0022] Another object of the present invention is to provide an information data processing terminal, which is used to implement the industrial process steady-state detection system.

[0023] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0024] The present invention provides a steady-state discrimination method for process data of a coal-fired power unit. First, a Gaussian filter is used to filter a target sequence. Then, a first-order difference is calculated for the filtered target sequence, and the difference value can reflect the direction and degree of data change. A Gaussian mixture model is used to model the difference value sequence, and a sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state sequence, thereby obtaining an estimated value of the fluctuation standard deviation of the target parameter in the steady state. Finally, a CUSUM control chart is used to perform steady-state discrimination of time series data.

[0025] The steady-state detection method based on the adaptive CUSUM control chart proposed in the present invention estimates the fluctuation standard deviation of the target operating condition parameter in the steady state from a global perspective based on a large amount of historical process data, and the larger the amount of historical data, the more robust the method is.

[0026] On the premise of obtaining the estimated value of the steady-state fluctuation standard deviation in the load time series data of the unit, the adaptive CUSUM control chart proposed in the present invention is used to detect the process data of the target parameter, and a certain segment of time series data is taken to demonstrate the detection effect, and compared with the R detection method. The comparison results show that the adaptive CUSUM control chart industrial process steady-state detection method provided by the present invention has a better effect and can detect the steady-state sequence of a continuous period, so it is more suitable for the scene where the operating conditions are frequently switched. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 is a flow chart of an industrial process steady-state detection method provided by an embodiment of the present invention;

[0029] Figure 2 is a schematic diagram of an industrial process steady-state detection method provided by an embodiment of the present invention;

[0030] Figure 3 It is a schematic diagram of the distribution of the first-order difference value of the load time series data of a certain unit provided by an embodiment of the present invention;

[0031] Figure 4 It is a schematic diagram of the steady-state detection effect of a certain section of unit load time series data provided by an embodiment of the present invention;

[0032] Among them, Figure (a) is the effect diagram of the R detection method, and the filter coefficient is (0 . 2 , 0 . 1 , 0 . 01), the detection threshold at 0.05 confidence level is 1.53, and the light gray area is the detected steady-state sequence; Figure (b) is the effect diagram of the adaptive CUSUM control chart detection method, and the light gray area is the detected steady-state sequence. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0034] In view of the problems existing in the prior art, the present invention provides an industrial process steady-state detection method, system, medium, equipment and terminal. The present invention is described in detail below with reference to the accompanying drawings.

[0035] In order to enable those skilled in the art to fully understand how to implement the present invention in detail, this section is an explanatory embodiment that expands and describes the technical solution of the claims.

[0036] The industrial process steady-state detection method provided by the embodiment of the present invention first uses a Gaussian filter to filter the target sequence; then, the first-order difference of the filtered target sequence is calculated, and the difference value can reflect the direction and degree of data change. The Gaussian mixture model is used to model the difference value sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state sequence, thereby obtaining an estimated value of the fluctuation standard deviation of the target parameter in the steady state; finally, the CUSUM control chart is used to distinguish the steady state of time series data.

[0037] like Figure 1 As shown, the industrial process steady-state detection method provided by the embodiment of the present invention includes the following steps:

[0038] S101, using a Gaussian filter to perform Gaussian filtering on a historical working process data set of a coal-fired power unit to obtain a filtered historical working process data set of the coal-fired power unit;

[0039] S102, performing differential processing on the filtered historical working process data set of the coal-fired power unit to obtain a first-order differential data set;

[0040] S103, using Gaussian mixture model to model the first-order difference data set, and obtain the estimated value of the standard deviation of the series fluctuation in the steady state;

[0041] S104, using an adaptive CUSUM control chart to process the continuous working process data of the coal-fired power unit to obtain a steady-state judgment result.

[0042] As a preference, Figure 2 As shown, the industrial process steady-state detection method provided by the embodiment of the present invention is divided into three steps: the first is data filtering, preferably Gaussian filtering; the second is to obtain the fluctuation distribution of the steady-state sequence in the target data, based on the sequence difference value, preferably a mixed Gaussian model to describe the fluctuation distribution of the sequence under steady-state; the third is to perform steady-state judgment based on the mean and fluctuation standard deviation of the detection subsequence and the cumulative and control chart.

[0043] The industrial process steady-state detection method provided by the embodiment of the present invention uses a Gaussian mixture model to estimate the standard deviation of the target sequence in steady state. In coal-fired power units, the change trend of operating parameters can be divided into three types: rising, steady state and falling, and the change trend is the cumulative result of the first-order difference of the sequence. The Gaussian mixture model is used to model the distribution of the first-order difference value of the target sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state subsequence in the sequence, and the estimated value of the standard deviation of the operating parameters in steady state is obtained.

[0044] The present invention uses a traversal method to discriminate the steady-state subsequence in the target sequence, and proposes the concept of "growing steady-state subsequence" to estimate the mean of the current steady-state subsequence. Each steady-state subsequence is generated by a point, and the value of the point is used as the mean to discriminate the next point in the sequence. If it belongs to the current steady-state subsequence, the current subsequence grows, the point is added to the current steady-state subsequence, and the mean is updated, and the subsequent data points are continued to be discriminated; if it does not belong to the current steady-state subsequence, the current steady-state subsequence stops growing, and the current discrimination point is used as the starting point of the next steady-state subsequence to discriminate the subsequent data points; until all the data points in the target sequence are traversed, that is, the steady-state discrimination process ends. During the entire traversal process, the mean of the current steady-state subsequence is updated as the current steady-state subsequence grows.

[0045] The industrial process steady-state detection system provided by the embodiment of the present invention includes:

[0046] A data filtering module, used for performing Gaussian filtering on the historical process data set using a Gaussian filter;

[0047] The sequence difference module is used to obtain the fluctuation distribution of the steady-state sequence in the target data. Based on the sequence difference value, a mixed Gaussian model is preferably used to describe the fluctuation distribution of the sequence in the steady state.

[0048] The steady-state discrimination module is used to perform steady-state discrimination of industrial process time series data of coal-fired power units based on the mean and fluctuation standard deviation of the detection subsequence and the cumulative control chart.

[0049] The embodiments of the present invention have achieved some positive effects during the development or use process, and indeed have great advantages over the prior art. The following content is described in conjunction with data, charts, etc. of the test process.

[0050] The steady-state discrimination method proposed in the present invention is compared with the most widely used R test method. The basic idea of ​​the R test method is to use two different methods to obtain an unbiased estimate of the variance of the filtered data, and to establish an F test statistic with their ratio. If the corresponding test conditions are met, it means that the process is in a steady state.

[0051] The steady-state detection method based on the adaptive CUSUM control chart proposed in the present invention estimates the fluctuation standard deviation of the target operating condition parameter in the steady state from a global perspective based on a large amount of historical process data, and the larger the amount of historical data, the more robust the method is.

[0052] First, the unit load data is filtered using a Gaussian filter with a template length of 5; then, the distribution of the first-order difference value of the filtered time series data is fitted using a Gaussian mixture model with a sub-model number of 3. In the embodiment of the present invention, the empirical density distribution of the first-order difference value of the unit load time series data and the GMM fitting distribution are as follows: Figure 3 and as shown in Table 1. The Gaussian mixture model obtained by solving is shown in formula (1). The coefficient of determination of the model is as high as 0.993, indicating that the model can well fit the distribution of the first-order difference value of the unit load time series data. From the composition of the model, it can be seen that in the change trend of the unit load process data, the average speed of the downward trend is -4.7MW / min, and the standard deviation of fluctuation is 6.883; the average value of fluctuation under the steady trend is 0.029MW / min, and the standard deviation of fluctuation is 0.834; the average speed under the rising trend is 4.449MW / min, and the standard deviation of fluctuation is 6.49. The average value of fluctuation under the steady trend is very close to 0, so it is reasonable to use the corresponding standard deviation of fluctuation 0.834 as the estimated value of the standard deviation in the steady state of the unit load time series data.

[0053] P(x)=0.265f(x|-4.769,6.883)+0.454f(x|0.029,0.834)+0.281f(x|4.449,6.49)(1)

[0054] Table 1 Empirical density distribution and GMM fitting distribution of the first-order difference value of unit load time series data

[0055]

[0056] On the premise of obtaining the estimated value of the steady-state fluctuation standard deviation in the load time series data of the unit, the adaptive CUSUM control chart proposed in this paper is used to detect the process data of the target parameter. A certain fragment of time series data is taken to show the detection effect and compared with the R detection method. The results are as follows: Figure 4 As shown. It can be seen that the adaptive CUSUM control chart detection method has a better effect and can detect steady-state sequences of a continuous period, so it is more suitable for scenarios with frequent switching of operating conditions. In addition, the R detection method has an obvious "boundary effect" when detecting steady-state sequences, that is, some points at the beginning of the new steady state will be missed, and some points after the steady state will be misjudged. This is due to the nature of linear filtering in the R detection method.

[0057] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0058] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for detecting steady state of an industrial process, characterized in that: include: A Gaussian filter is used to perform Gaussian filtering on the historical working process data set of the coal-fired power unit to obtain a filtered historical working process data set of the coal-fired power unit; The filtered coal-fired power unit historical working process data set is differentially processed to obtain a first-order difference data set; the first-order difference data set is modeled using a Gaussian mixture model to obtain an estimate of the standard deviation of the sequence fluctuation under steady state; the continuous working process data of the coal-fired power unit to be processed is processed using an adaptive CUSUM control chart to obtain a steady-state discrimination result; The industrial process steady-state detection method includes the following steps: Step 1, data filtering: Gaussian filtering is performed on the historical process data set using a Gaussian filter; Step 2: Obtain the volatility distribution of the steady-state sequence in the target data, and select a mixed Gaussian model to describe the volatility distribution of the sequence under steady state based on the sequence difference value; Step 3: Based on the mean and fluctuation standard deviation of the detection subsequence, the steady-state discrimination of the industrial process time series data of the coal-fired power unit is carried out based on the cumulative sum control chart; In step 3, the steady-state subsequence in the target sequence is identified by traversal, and the concept of "growing steady-state subsequence" is proposed to estimate the mean of the current steady-state subsequence; each steady-state subsequence is generated by a point, and the point value is used as the mean to identify the next point in the sequence; if it belongs to the current steady-state subsequence, the current subsequence grows, the point is added to the current steady-state subsequence, and the mean is updated, and the subsequent data points are continuously identified; if it does not belong to the current steady-state subsequence, the current steady-state subsequence stops growing, and the current identification point is used as the starting point of the next steady-state subsequence to identify the subsequent data points; until all the data points in the target sequence are traversed, the steady-state identification process ends; During the entire traversal process, the mean of the current steady-state subsequence is updated as the current steady-state subsequence grows.

2. The method for detecting steady state of an industrial process according to claim 1, characterized in that: Industrial process steady-state detection methods also include: A Gaussian filter is used to filter the target sequence; the first-order difference of the filtered target sequence is calculated, and the difference value reflects the direction and degree of data change; the Gaussian mixture model is used to model the difference value sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state sequence, and then the estimated value of the fluctuation standard deviation of the target parameter in the steady state is obtained; finally, the CUSUM control chart is used to perform steady-state discrimination detection on the industrial process time series data of coal-fired power units to obtain the steady-state discrimination result.

3. The method for detecting steady state of an industrial process according to claim 1, characterized in that: In step 2, the Gaussian mixture model is used to estimate the standard deviation of the target sequence in steady state; in coal-fired power units, the change trend of operating parameters is divided into three types: rising, steady state and falling, and the change trend is the cumulative result of the first-order difference of the sequence; The Gaussian mixture model is used to model the distribution of the first-order difference values ​​of the target sequence, and the sub-model with a mean approximately equal to 0 is considered to be the fluctuation distribution of the steady-state subsequence in the sequence, thereby obtaining the estimated value of the standard deviation of the operating parameters in the steady state.

4. An industrial process steady-state detection system using the industrial process steady-state detection method according to any one of claims 1 to 3, characterized in that: Industrial process steady-state detection system includes: A data filtering module, used for performing Gaussian filtering on the historical process data set using a Gaussian filter; The sequence difference module is used to obtain the volatility distribution of the steady-state sequence in the target data. Based on the sequence difference value, a mixed Gaussian model is selected to describe the volatility distribution of the sequence under steady state; The steady-state discrimination module is used to perform steady-state discrimination of industrial process time series data of coal-fired power units based on the mean and fluctuation standard deviation of the detection subsequence and the cumulative control chart.

5. A computer device, characterized in that: The computer device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the industrial process steady-state detection method according to any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the industrial process steady-state detection method according to any one of claims 1 to 3.

7. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the industrial process steady-state detection system as described in claim 4.

Images