Intelligent autonomous control method of coal-fired machine for novel electric power system

Abstract

The invention relates to the technical field of coal-fired machine intelligent control, in particular to a coal-fired machine intelligent autonomous control method for a novel electric power system, and the method comprises the steps: obtaining a flame gray-scale map during the combustion of flame in a coal-fired machine at each collection moment; determining flame combustion deviation at each acquisition moment; and according to the average distribution condition of gray values of all pixel points in a flame region in the flame gray-scale map at each collection moment and the size of the flame region, determining a combustion state evaluation coefficient at each collection moment by combining the flame combustion deviation so as to control the flame combustion state in the coal-fired machine. By analyzing the interference of the impurities contained in the pulverized coal on the flame combustion state, the stable state of flame combustion is obtained by using the neural network, the combustion parameters of the coal-fired machine are regulated and controlled in combination with the control algorithm, and the control precision of the coal-fired machine is improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent control of coal-fired machines, and in particular to an intelligent autonomous control method for coal-fired machines used in new power systems. Background Art

[0002] As my country takes energy conservation and emission reduction as its energy strategy, developing renewable clean energy is an effective means to achieve this strategy. When a high proportion of renewable energy is connected to the power grid, due to the limited peak-shaving capacity and poor stability of renewable energy, thermal power generation, especially coal-fired units, is required to have deep flexibility adjustment capabilities to adapt to changes in renewable energy generation load. However, due to the minimum stability and pollution emission limits of coal-fired units, the deep adjustment and control of the coal-fired unit power grid is improper, which easily leads to energy waste of "abandoning wind and solar power".

[0003] Coal-fired power plants are devices that use coal as fuel to convert fossil energy into electrical energy. Therefore, the process of converting electrical energy is affected by the demand of the power grid and the quality of coal. Flame stability is the primary issue that coal-fired power plants pay attention to during peak load regulation. Flame stability is directly related to the reliability of coal-fired power plant operation. Under deep peak load regulation, the fuel supply needs to be regulated, which affects the stability of the flame during combustion.

[0004] The traditional detection method is mainly based on grayscale variance to test the stability of flame combustion, mainly by analyzing the correlation between the grayscale variance of the flame image in the furnace and the flame stability. However, due to the variable coal quality and coal types, mixed coal burning, ash accumulation and other serious phenomena in my country, the combustion inside the coal-fired machine will be affected by coal powder and other impurities, resulting in short-term pulses. As a result, the traditional grayscale variance detection method has a large deviation in the combustion stability in the furnace, which leads to poor autonomous control ability of the coal-fired machine. Summary of the invention

[0005] In order to solve the above technical problems, the present application provides an intelligent autonomous control method for a coal-fired machine in a new power system to solve the existing problems.

[0006] The present invention discloses an intelligent autonomous control method for a coal-fired machine used in a new power system, which adopts the following technical solutions:

[0007] An embodiment of the present application provides an intelligent autonomous control method for a coal-fired machine in a new power system, the method comprising the following steps:

[0008] Obtain the flame grayscale image of the flame burning inside the coal-fired machine at each acquisition time;

[0009] Segment the flame grayscale image at each acquisition time to obtain the flame area and extract the backbone line of the flame area;

[0010] Determine the flame combustion deviation at each acquisition moment according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition moment, and the distance from all pixels on the backbone line to the edge of the flame area;

[0011] According to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time and the size of the flame area, the combustion state evaluation coefficient at each acquisition time is determined in combination with the flame combustion deviation to control the state of flame combustion in the coal-fired machine.

[0012] Preferably, the method for determining the flame combustion deviation at each acquisition moment is:

[0013] Determine the brightness gradient deviation at each acquisition time according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition time;

[0014] Determine the distance deviation at each acquisition moment according to the distance from all pixel points on the backbone line of the flame region in the flame grayscale image at each acquisition moment to the edge of the flame region;

[0015] The flame combustion deviation at each acquisition time is determined based on the brightness gradient deviation and the distance deviation.

[0016] Preferably, the process of determining the brightness gradient deviation at each acquisition moment is:

[0017] The backbone line is divided into two sections using the central pixel point of the backbone line of the flame area in the flame grayscale image at each acquisition time as the dividing point, and the grayscale values ​​of all the pixels between the two sections and the central pixel point are respectively used to form the first grayscale sequence and the second grayscale sequence at each acquisition time;

[0018] The brightness gradient deviation at each acquisition moment is the distance between the first grayscale sequence and the second grayscale sequence at each acquisition moment. Preferably, the distance deviation at each acquisition moment is the cumulative difference in distances from all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition moment to the upper and lower edges of the flame area.

[0019] Preferably, the expression of the flame combustion deviation at each acquisition moment is: i =B i +C i Where A i represents the flame combustion deviation at the i-th acquisition moment; B i represents the brightness gradient deviation at the i-th acquisition moment; C i Represents the distance deviation at the i-th acquisition moment.

[0020] Preferably, the method for determining the combustion state evaluation coefficient at each acquisition moment is:

[0021] Determine the evaluation value at each acquisition time according to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time and the size of the flame area;

[0022] The evaluation values ​​of a preset number of collection moments before each collection moment are combined into a combustion evaluation sequence at each collection moment;

[0023] Determine the combustion fluctuation value at each acquisition moment based on the discrete degree and change trend of all evaluation values ​​in the combustion evaluation sequence at each acquisition moment;

[0024] A combustion state evaluation coefficient at each acquisition time is determined based on the combustion fluctuation value and the flame combustion deviation.

[0025] Preferably, the evaluation value at each acquisition moment is the product of the average grayscale value of all pixels in the flame area in the flame grayscale image at each acquisition moment and the proportion of the flame area in the flame grayscale image.

[0026] Preferably, the combustion fluctuation value at each acquisition moment is the ratio of the variance of the combustion evaluation sequence at each acquisition moment to the absolute value of the trend strength of the combustion evaluation sequence.

[0027] Preferably, the expression of the combustion state evaluation coefficient at each acquisition moment is: E i =D i ×A i Where, E i represents the combustion state evaluation coefficient at the i-th acquisition moment; D i A represents the combustion fluctuation value at the i-th acquisition moment; i Represents the flame combustion deviation at the i-th acquisition moment.

[0028] Preferably, the controlling the state of the flame burning in the coal-fired machine includes:

[0029] The mean and variance of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time, the circularity and moment of inertia of the flame area, and the combustion state evaluation coefficient are used to form a feature vector at each acquisition time;

[0030] The flame grayscale image and feature vector at each acquisition time are used as the input of the neural network model to obtain the combustion stability at each acquisition time.

[0031] If the combustion stability is less than the second preset value, the combustion parameters of the coal-fired machine are adjusted, and the deviation between the combustion stability at each collection moment and the second preset value and the rate of change of the deviation are used as the input of the fuzzy control algorithm to obtain the parameters of the fuzzy control algorithm to regulate the combustion parameters of the coal-fired machine; otherwise, the combustion parameters of the coal-fired machine will not be adjusted.

[0032] This application has at least the following beneficial effects:

[0033] The present application collects flame image data through a camera, determines the flame combustion deviation at each collection moment according to the change of the grayscale value of all pixels on the backbone line of the flame area in the flame grayscale image at each collection moment, and the distance from all pixels on the backbone line to the edge of the flame area, reflecting the interference of coal powder impurities in the coal-fired machine at the current collection moment; determines the combustion state evaluation coefficient at each collection moment according to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each collection moment and the size of the flame area, combined with the flame combustion deviation, reflecting the overall short-time pulse situation of the coal-fired machine within the local time range; based on the combustion state evaluation coefficient and combined with the combustion stability output by the neural network, autonomous control of the coal-fired machine is achieved through a fuzzy PID algorithm.

[0034] This application analyzes the short-term pulse characteristics that occur when coal powder particles are disturbed by impurities, and analyzes them in conjunction with normal stable combustion and combustion fluctuation characteristics. It uses a neural network to obtain the stable state of flame combustion, and combines the fuzzy PID control algorithm to achieve intelligent and autonomous control of the coal-fired machine, thereby maximizing energy combustion and reducing production costs and control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 A flowchart of the steps of an intelligent autonomous control method for a coal-fired machine used in a new power system;

[0037] Figure 2 A flow chart of the steps for extracting flame combustion deviation;

[0038] Figure 3 This is a schematic diagram of the combustion state evaluation coefficient extraction process;

[0039] Figure 4 Schematic diagram of the autonomous control process of the coal-fired machine. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of an intelligent autonomous control method for a coal-fired machine for a new power system proposed in the present application, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0041] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0042] The specific scheme of the intelligent autonomous control method of a coal-fired machine for a new power system provided by the present application is described in detail below with reference to the accompanying drawings.

[0043] An embodiment of the present application provides an intelligent autonomous control method for a coal-fired machine for a new power system. Specifically, the following intelligent autonomous control method for a coal-fired machine for a new power system is provided. Figure 1 , the method comprises the following steps:

[0044] Step S1: Obtain the flame grayscale image of the internal flame of the coal-fired machine at each acquisition time and pre-process it.

[0045] S101: Collecting flame images of the flame burning inside the coal-fired machine.

[0046] Since the light inside the coal-fired machine is strong and the temperature is high, the requirements for the image acquisition device are high. Therefore, this embodiment uses a high-temperature resistant CCD camera, and the requirements for the camera are as follows: first, it is assumed that the flame inside the coal-fired machine has a virtual object surface, and the image output by the CCD camera is the result of the accumulation of longitudinal spatial light radiation of the flame burning in the boiler. Secondly, the number of lines of the CCD camera is required to be sufficient, at least greater than or equal to 1080 lines, and it is assumed that the temperature in a small area is the same. Therefore, each pixel point on the target surface of the CCD camera can only receive the light radiation stimulus corresponding to this pixel point, and each point on the flame in the coal-fired machine furnace has a unique corresponding pixel point in the imaging image.

[0047] A CCD camera is used to obtain the flame image of the flame burning inside the coal-fired machine, and the sampling interval of the CCD camera is set to continuously sample the flame image of the flame burning inside the coal-fired machine, and the flame image at each sampling moment is grayed to obtain the flame grayscale map at each sampling moment.

[0048] It should be noted that the value of the sampling interval is manually set. In this embodiment, the value of the sampling interval is 1s. The implementer can also set it according to the specific situation. This embodiment does not impose any special restrictions.

[0049] S102: Preprocessing the flame grayscale image at each acquisition time.

[0050] When the CCD camera collects the flame grayscale image inside the coal-fired machine, it is inevitably disturbed by environmental factors, resulting in certain noise and blur in the image. Therefore, the image denoising algorithm is used to denoise the flame grayscale image at each acquisition moment to obtain the preprocessed flame grayscale image.

[0051] It should be noted that there are many commonly used image denoising algorithms. In this embodiment, the flame grayscale image is denoised and smoothed by using histogram weighted averaging and mean filtering. The implementer can also use image denoising algorithms such as Gaussian filtering, bilateral filtering or BM3D algorithm to denoise the flame grayscale image. This embodiment does not impose any special restrictions.

[0052] Among them, histogram weighted averaging and mean filtering are both well-known technologies, and the specific processes of image enhancement and denoising are not described in detail.

[0053] Step S2: determining the flame combustion deviation at each acquisition moment according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition moment and the distance from all pixels on the backbone line to the edge of the flame area.

[0054] When deep peak-shaving is carried out on the coal-fired units in thermal power plants, the main adjustment is made between the minimum and maximum power generation of the coal-fired units. Often during peak-shaving, the coal-fired units operate at low load. At this time, the operating characteristics of the coal-fired units deviate far from full-load operation, resulting in poor combustion stability of the coal-fired units during peak-shaving and large emissions of pollutant gases.

[0055] When the coal-fired machine is in a low-load operation state, the flame combustion stability is easily affected by the impurities in the coal powder, causing the flame to change in brightness or shape in a short period of time. This phenomenon is called a short-time pulse. Therefore, when the traditional grayscale variance method is used to detect the flame combustion stability, a large deviation occurs, which leads to a deviation in the low-load intelligent control of the coal-fired machine, and further analysis is required in combination with the collected flame grayscale map.

[0056] The flame image at each acquisition time often includes the flame area and smoke area of ​​the coal powder burning without impurities. In the flame area, the closer to the center of the flame, the higher the combustion temperature is, and the combustion temperature at the edge of the flame is lower than that at the center of the flame. Therefore, the flame image basically shows that the center is bright, and the brightness gradually decreases as it approaches the edge. The coal powder particles gradually become smaller from the center of the flame to the edge of the flame, and the combustion characteristics of the flame area are symmetrical, that is, from the center of the flame to the two ends of the flame edge, the change trend of the flame brightness is the same, and the change trend of the coal powder particle size is also the same. For the smoke area, due to the full combustion of the coal powder, the proportion of the smoke area is often smaller.

[0057] However, if the coal powder contains impurities, the interference of the impurities will cause the flame to burn incompletely, causing the shape of the flame area to change, resulting in fluctuations and distortions in the flame area. At the same time, depending on the impurities in the coal powder, flame color reactions will occur during combustion, causing the brightness of the flame to change.

[0058] This embodiment uses a semantic segmentation neural network to extract the flame grayscale image at each acquisition moment and perform binary classification extraction to obtain the flame area and the smoke area in the flame grayscale image at each acquisition moment.

[0059] It should be noted that there are many commonly used semantic segmentation neural network models. In this embodiment, the SCTNet model is used to perform binary classification extraction on the flame images at each acquisition moment. The implementer can also use the PSPNet model or the Mask R-CNN model to perform binary classification extraction on the image according to the specific situation. This embodiment does not impose any special restrictions.

[0060] Among them, the SCTNet model is a well-known technology, and the specific process of implementing the segmentation is not repeated here.

[0061] S201: Determine the brightness gradient deviation at each acquisition moment according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition moment.

[0062] In the flame area, the closer to the center, the higher the combustion temperature. Therefore, the flame image basically shows that the center is bright, and the brightness gradually decreases as it approaches the edge. Therefore, by analyzing the changes in the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time, the brightness gradient deviation at each acquisition time is determined to reflect the full degree of flame combustion in the coal-fired machine. Specifically:

[0063] Firstly, a backbone extraction algorithm is used to obtain the backbone line of the flame area of ​​the flame grayscale image at each acquisition moment. The backbone line is divided into two segments with the central pixel point of the flame area as the dividing point. The grayscale values ​​of all pixels between the two segments and their central pixel points are respectively combined into the first grayscale sequence and the second grayscale sequence at each acquisition moment.

[0064] It should be noted that there are many commonly used backbone extraction algorithms. In this embodiment, the Zhang-Suen backbone extraction algorithm is used to obtain the backbone line of the flame area. The implementer may also use other algorithms to extract the backbone line, and this embodiment does not impose any special restrictions.

[0065] Among them, the Zhang-Suen backbone extraction algorithm is a well-known technology, and the specific process of extracting the backbone line is not repeated here.

[0066] Furthermore, according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition time, the brightness gradient deviation at each acquisition time is determined, specifically:

[0067] The brightness gradient deviation at each acquisition moment is the distance between the first grayscale sequence and the second grayscale sequence of the flame area in the flame grayscale image at each acquisition moment.

[0068] It should be noted that there are many methods for measuring the distance between sequences. In this embodiment, the similarity between the two sequences is measured by calculating the DTW distance between the first sequence and the second sequence. The implementer may also use other methods such as Manhattan distance or Chebyshev distance to measure the similarity between the two sequences according to the specific situation. This embodiment does not impose any special restrictions.

[0069] The DTW distance is a well-known technology, and its specific calculation process will not be described in detail.

[0070] If the current flame region is a flame region where pulverized coal without impurities is burned, according to the trend characteristics of flame combustion, the closer to the center of the flame, the higher the brightness, and the closer to the edge of the flame, the lower the brightness. Therefore, the change trends of the first grayscale sequence and the second grayscale sequence are basically the same, and the brightness gradient deviation of the current flame region is smaller; if the current flame region is a flame region where pulverized coal containing impurities is burned, the greater the difference in the change trends of the first grayscale sequence and the second grayscale sequence, the greater the brightness gradient deviation of the current flame region.

[0071] S202: Determine the distance deviation at each acquisition moment according to the distance from all pixel points on the backbone line of the flame region in the flame grayscale image at each acquisition moment to the edge of the flame region.

[0072] In the flame area, along the direction from the center of the flame area to the upper and lower edges of the flame area, the change trend of the flame brightness and the change trend of the size of the coal powder particles are symmetrical about the center of the flame area. Therefore, according to the distance from all pixel points on the backbone line of the flame area in the flame grayscale image at each acquisition time to the edge of the flame area, the distance deviation at each acquisition time is determined to determine whether the current flame area is a flame area where coal powder without impurities is burned, specifically:

[0073] The distance deviation at each acquisition time is the result of the accumulation of the differences in the distances from all pixel points on the backbone line of the flame area in the flame grayscale image at each acquisition time to the upper and lower edges of the flame area.

[0074] It should be noted that there are many methods for measuring differences. In this embodiment, the absolute value of the difference between the distances of each pixel point on the backbone line of the flame area in the flame grayscale image at each acquisition time to the upper and lower edges of the flame area is calculated as a method for measuring the difference. The implementer may also use other methods for measuring the difference such as ratio according to the specific situation, and this embodiment does not impose any special restrictions.

[0075] If the current flame area is a flame area of ​​pulverized coal burning without impurities, according to the characteristics of pulverized coal particle combustion, the current flame area often presents a certain symmetrical area. At this time, the distances from the pixels on the backbone line to the upper and lower edges of the flame area are basically the same, and the distance deviation of the flame area in the flame grayscale image at the acquisition moment is small; conversely, if the current flame area is a flame area of ​​pulverized coal burning with impurities, the difference in the distances from the pixels on the backbone line to the upper and lower edges of the flame area is large, and the distance deviation of the flame area in the flame grayscale image at the acquisition moment is large.

[0076] S203: Determine the flame combustion deviation at each acquisition time based on the flame combustion deviation and the distance deviation.

[0077] If the coal powder contains impurities, the interference of the impurities will cause the flame to burn incompletely, causing the shape of the flame area to change, resulting in fluctuations and distortions in the flame area. At the same time, depending on the impurities in the coal powder, flame color reactions will occur during combustion, causing the brightness of the flame to change.

[0078] Therefore, the flame combustion deviation at each acquisition moment is determined based on the brightness gradient deviation and the distance deviation to eliminate the interference of impurities in the coal powder on the flame brightness and shape, thereby more accurately controlling the adjustment of combustion parameters in the coal-fired machine.

[0079] The expression of flame combustion deviation at each acquisition time is: A i =B i +C i Where A i represents the flame combustion deviation at the i-th acquisition moment; B i represents the brightness gradient deviation at the i-th acquisition moment; C i Represents the distance deviation at the i-th acquisition moment.

[0080] If the current acquisition moment is the interference fluctuation of coal powder impurity combustion, the flame color reaction and the fluctuation and distortion of the flame shape will occur due to the combustion of impurities, and the flame change trend characteristics of the flame area of ​​the coal powder that originally did not contain impurities will be broken. The larger the value of the brightness gradient deviation, the greater the symmetry deviation of the flame area shape due to the fluctuation and distortion of the flame shape, and finally the distance deviation value at the current acquisition moment increases, and finally the flame combustion deviation value is larger; on the contrary, if the flame area at the current acquisition moment is the area of ​​coal powder combustion that does not contain impurities, the flame change trends on both sides of the flame center are consistent, and there will be no fluctuation and distortion of the flame shape, and the two sides of the flame center are highly symmetrical, and the flame combustion deviation is smaller.

[0081] Preferably, as an embodiment of the present application, the flame combustion deviation extraction step flow chart is as follows: Figure 2 shown.

[0082] Step S3: determining the combustion state evaluation coefficient at each acquisition moment according to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition moment and the size of the flame area in combination with the flame combustion deviation.

[0083] During the peak load regulation of the power grid, the coal-fired unit will adjust the combustion parameters to meet the needs of the power grid, so the coal-fired unit needs to maintain stable operation under a certain load. Ideally, the pulverized coal in the coal-fired unit can be fully burned and remain relatively stable under the current load, that is, the flame state remains basically stable. However, due to impurities in the coal, the flame state may fluctuate, that is, the flame is larger for a while, and the output power of the coal-fired unit is greater than the load demand; the flame is smaller for a while, and the output power of the coal-fired unit is less than the load demand. At this time, the coal combustion state needs to be regulated. However, due to the interference of impurities in the coal powder, the size of the flame will mutate in a short time, that is, at this time, the flame combustion will produce short-term pulses, which will interfere with the detection of flame combustion stability.

[0084] Therefore, in order to further measure the stability of flame combustion in the coal-fired machine, each acquisition moment is taken as the time end point, and the flame grayscale images of a preset number of acquisition moments are obtained forward. The product of the mean grayscale value of all pixels in the flame area in the flame grayscale image at each acquisition moment and the proportion of the flame area in the flame grayscale image is used as the evaluation value of each acquisition moment, and the combustion evaluation sequence of each acquisition moment is composed in chronological order.

[0085] It should be noted that the value of the preset number is artificially set. In this embodiment, the value of the preset number is 30. The implementer can set it according to the specific situation. This embodiment does not impose any special restrictions.

[0086] Wherein, for each collection moment, when there are no preset number of collection moments before it, the collection moment is not processed.

[0087] S301: Determine the combustion fluctuation value at each collection moment based on the discrete degree and change trend of all evaluation values ​​in the combustion evaluation sequence at each collection moment.

[0088] Under normal circumstances, coal powder combustion has a certain time correlation, that is, there is a time correlation between the ignition of coal powder and its full combustion. However, if the coal powder contains trace impurities, the impurities will cause the flame size to change suddenly during the combustion process, forming a short-time pulse, which will affect the stability detection of flame combustion.

[0089] Therefore, based on the discrete degree and change trend of all evaluation values ​​in the combustion evaluation sequence at each acquisition moment, the combustion fluctuation value at each acquisition moment is determined to exclude the short-term pulse caused by impurity combustion from the detection of flame combustion stability, specifically:

[0090] Analyze the variance of the combustion assessment sequence at each acquisition moment;

[0091] It should be noted that in this embodiment, the variance is used to measure the degree of dispersion of all evaluation values ​​in the combustion evaluation sequence. In other embodiments, the variance can be replaced by the standard deviation; in other embodiments, the variance can also be replaced by the mean deviation to measure the degree of dispersion of all evaluation values.

[0092] Furthermore, the trend strength of the combustion assessment sequence at each acquisition moment is analyzed;

[0093] It should be noted that there are many methods for calculating the trend strength of the combustion assessment sequence. In this embodiment, the moving average method is used to obtain the trend strength of the combustion assessment sequence at each acquisition moment. The implementer may also use the differential method or other methods to obtain the trend strength of the combustion assessment sequence according to the specific situation. This embodiment does not impose any special restrictions.

[0094] Among them, the moving average method is a well-known technology, and the specific process of calculating the trend strength will not be repeated here.

[0095] It can be obtained that the combustion fluctuation value at each acquisition time is the ratio of the variance of the combustion evaluation sequence at each acquisition time to the absolute value of the trend intensity.

[0096] If a short-time pulse is generated in the pulverized coal combustion process at the current acquisition moment, the average grayscale value of the flame area at the corresponding pulverized coal impurity interference acquisition moment is large, and the area of ​​the flame area accounts for a small proportion, so the evaluation value is large. Therefore, the greater the fluctuation of the combustion evaluation sequence, the greater the variance. At the same time, due to the brevity and randomness of the short-time pulse, it does not have a trend. Therefore, the smaller the value of the trend intensity of the combustion evaluation sequence, the greater the combustion fluctuation value obtained in the end. On the contrary, if no short-time pulse is generated in the pulverized coal combustion process at the current acquisition moment, the average grayscale value of the flame area is small, and the area of ​​the flame area accounts for a large proportion, so the evaluation value is small. Therefore, the smaller the fluctuation of the combustion evaluation sequence, the smaller the variance obtained, the larger the value of the trend intensity of the combustion evaluation sequence, and the smaller the combustion fluctuation value obtained in the end.

[0097] S302: Determine a combustion state evaluation coefficient at each acquisition time based on the combustion fluctuation value and the flame combustion deviation.

[0098] In order to more precisely monitor and analyze the dynamic characteristics of the flame and ensure the combustion efficiency, the combustion state evaluation coefficient at each acquisition time is determined based on the flame combustion deviation and the combustion fluctuation value. The expression of the combustion state evaluation coefficient at each acquisition time is: E i =B i ×A i Where, E i represents the combustion state evaluation coefficient at the i-th acquisition moment; D i A represents the combustion fluctuation value at the i-th acquisition moment; i Represents the flame combustion deviation at the i-th acquisition moment.

[0099] The combustion state evaluation coefficient at each acquisition time is the product of the flame combustion deviation and the combustion fluctuation value at each acquisition time.

[0100] If a short-time pulse occurs in the flame combustion process at the current acquisition moment, the fluctuation mutation in the combustion evaluation sequence is large, and the combustion fluctuation value is large; at the same time, the flame combustion deviation of the flame image at the current acquisition moment is large, and finally the combustion state evaluation coefficient at the current acquisition moment is large; conversely, if no short-time pulse occurs in the flame combustion process at the current acquisition moment, the combustion fluctuation value is small; at the same time, the flame combustion deviation of the flame image at the current acquisition moment is small, and finally the combustion state evaluation coefficient at the current acquisition moment is small.

[0101] Preferably, as an embodiment of the present application, the combustion state evaluation coefficient extraction process diagram is as follows: Figure 3 shown.

[0102] Step S4: Based on the combustion state evaluation coefficient at each acquisition moment, the state of flame combustion in the coal-fired machine is controlled.

[0103] The combustion state evaluation coefficient at each acquisition moment is obtained by step S3, which reflects the situation of short-time pulses occurring in the coal-fired machine at the current acquisition moment, and selects other image feature parameters directly related to flame stability, which are specifically manifested as: the mean and variance of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition moment, the circularity and moment of inertia of the flame area, and the combustion state evaluation coefficient together to construct the feature vector at each acquisition moment.

[0104] It should be noted that the calculation method of circularity and moment of inertia is a well-known technology, and the specific calculation process will not be repeated here.

[0105] The flame grayscale image and feature vector at each acquisition moment are used as the input of the BP back propagation neural network, the cross entropy loss function is adopted, the optimizer is Adam, the number of training rounds epoch is the first preset value, and the model output is the combustion stability at each acquisition moment.

[0106] It should be noted that the value of the first preset value is manually set. In this embodiment, the value of the first preset value is 20. The implementer can also set it according to the specific situation. This embodiment does not impose any special restrictions.

[0107] Among them, the BP back propagation neural network is a well-known technology, and its specific operation process will not be repeated here.

[0108] Based on the combustion stability at each collection moment, if the combustion stability is greater than or equal to the second preset value, it indicates that the combustion state of the coal-fired machine at the current collection moment is relatively stable, and there is no need to adjust the combustion parameters of the coal-fired machine. If the combustion stability is less than the second preset value, it indicates that the combustion state of the coal-fired machine at the current collection moment is unstable, and the combustion parameters of the coal-fired machine need to be adjusted.

[0109] It should be noted that the value of the second preset value is manually set. In this embodiment, the value of the second preset value is 0.8. The implementer can set it according to the specific situation, and this embodiment does not impose any special restrictions.

[0110] The specific control process is: calculate the deviation between the combustion stability at each collection moment and the second value; calculate the change rate of the deviation between each collection moment and the previous collection moment; use the deviation at each collection moment and the change rate of the deviation as the input of the fuzzy PID control algorithm, and the algorithm outputs the parameters of the fuzzy PID control algorithm, namely the proportional gain Kp, integral gain Ki, and differential gain Kd, and adjusts the combustion parameters of the coal-fired machine based on this, monitors the combustion state of the coal-fired machine and adjusts the combustion parameters of the coal-fired machine in real time, realizes autonomous control of the coal-fired machine, and improves the anti-interference ability of the coal-fired machine and the autonomous control ability of the algorithm.

[0111] It should be noted that the deviation is calculated by taking the absolute value of the difference between the combustion stability at each acquisition moment and the second preset value.

[0112] Among them, the fuzzy PID control algorithm is a well-known technology, and its specific process will not be repeated here.

[0113] At this point, based on the combustion state evaluation coefficient, the real-time supervision and regulation of the combustion state of the coal-fired machine is achieved in combination with the neural network model and the fuzzy PID control algorithm.

[0114] Preferably, as an embodiment of the present application, the schematic diagram of the autonomous control process of the coal-fired machine is as follows: Figure 4 shown.

[0115] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0116] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0117] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An intelligent autonomous control method for a coal-fired machine in a new power system, characterized in that: The method comprises the following steps: Obtain the flame grayscale image of the flame burning inside the coal-fired machine at each acquisition time; Segment the flame grayscale image at each acquisition time to obtain the flame area and extract the backbone line of the flame area; Determine the flame combustion deviation at each acquisition moment according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition moment, and the distance from all pixels on the backbone line to the edge of the flame area; According to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time and the size of the flame area, the combustion state evaluation coefficient at each acquisition time is determined in combination with the flame combustion deviation to control the state of flame combustion in the coal-fired machine.

2. The intelligent autonomous control method for coal-fired machines used in new power systems according to claim 1, characterized in that: The method for determining the flame combustion deviation at each acquisition moment is: Determine the brightness gradient deviation at each acquisition time according to the change of the grayscale values ​​of all pixels on the backbone line of the flame area in the flame grayscale image at each acquisition time; Determine the distance deviation at each acquisition moment according to the distance from all pixel points on the backbone line of the flame region in the flame grayscale image at each acquisition moment to the edge of the flame region; The flame combustion deviation at each acquisition time is determined based on the brightness gradient deviation and the distance deviation.

3. The intelligent autonomous control method of a coal-fired machine for a new power system according to claim 2, characterized in that: The process of determining the brightness gradient deviation at each acquisition moment is as follows: The backbone line is divided into two sections using the central pixel point of the backbone line of the flame area in the flame grayscale image at each acquisition time as the dividing point, and the grayscale values ​​of all the pixels between the two sections and the central pixel point are respectively used to form the first grayscale sequence and the second grayscale sequence at each acquisition time; The brightness gradient deviation at each acquisition moment is the distance between the first grayscale sequence and the second grayscale sequence at each acquisition moment.

4. The intelligent autonomous control method of a coal-fired machine for a new power system according to claim 2, characterized in that: The distance deviation at each acquisition moment is the result of accumulating the difference in distances from all pixel points on the backbone line of the flame area in the flame grayscale image at each acquisition moment to the upper and lower edges of the flame area.

5. The intelligent autonomous control method of a coal-fired machine for a new power system according to claim 2, characterized in that: The expression of the flame combustion deviation at each acquisition time is: i =B i +C i Where A i represents the flame combustion deviation at the i-th acquisition moment; B i represents the brightness gradient deviation at the i-th acquisition moment; C i Represents the distance deviation at the i-th acquisition moment.

6. The intelligent autonomous control method of a coal-fired machine for a new power system according to claim 1, characterized in that: The method for determining the combustion state evaluation coefficient at each acquisition moment is: Determine the evaluation value at each acquisition time according to the average distribution of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time and the size of the flame area; The evaluation values ​​of a preset number of collection moments before each collection moment are combined into a combustion evaluation sequence at each collection moment; Determine the combustion fluctuation value at each acquisition moment based on the discrete degree and change trend of all evaluation values ​​in the combustion evaluation sequence at each acquisition moment; A combustion state evaluation coefficient at each acquisition time is determined based on the combustion fluctuation value and the flame combustion deviation.

7. The method for intelligent autonomous control of coal-fired machines for a new power system according to claim 6, characterized in that: The evaluation value at each acquisition moment is the product of the mean grayscale value of all pixels in the flame area of ​​the flame grayscale image at each acquisition moment and the proportion of the flame area in the flame grayscale image.

8. The method for intelligent autonomous control of coal-fired machines for a new power system according to claim 6, characterized in that: The combustion fluctuation value at each acquisition moment is the ratio of the variance of the combustion evaluation sequence at each acquisition moment to the absolute value of the trend strength of the combustion evaluation sequence.

9. The intelligent autonomous control method of a coal-fired machine for a new power system according to claim 6, characterized in that: The expression of the combustion state evaluation coefficient at each acquisition time is: E i =D i ×A i Where, E i represents the combustion state evaluation coefficient at the i-th acquisition moment; D i A represents the combustion fluctuation value at the i-th acquisition moment; i Represents the flame combustion deviation at the i-th acquisition moment.

10. The method for intelligent autonomous control of coal-fired machines for a new power system according to claim 1, characterized in that: The method of controlling the state of flame combustion in the coal-fired machine includes: The mean and variance of the grayscale values ​​of all pixels in the flame area in the flame grayscale image at each acquisition time, the circularity and moment of inertia of the flame area, and the combustion state evaluation coefficient are used to form a feature vector at each acquisition time; The flame grayscale image and feature vector at each acquisition time are used as the input of the neural network model to obtain the combustion stability at each acquisition time. If the combustion stability is less than the second preset value, the combustion parameters of the coal-fired machine are adjusted, and the deviation between the combustion stability at each collection moment and the second preset value and the rate of change of the deviation are used as the input of the fuzzy control algorithm to obtain the parameters of the fuzzy control algorithm to regulate the combustion parameters of the coal-fired machine; otherwise, the combustion parameters of the coal-fired machine will not be adjusted.

Images