An automatic control method for a reverse-thrust garbage incineration mechanical grate feeder

Abstract

The present invention relates to an automatic control method for a reverse push type waste incineration mechanical grate feeder, belonging to the technical field of waste incineration. The method comprises the following steps: S1: obtaining historical data for more than half a year and classifying and dividing them; S2: screening the historical data, using the K-means clustering algorithm to automatically remove abnormal data of the start-stop furnace, equipment failure and measurement point maintenance process, and obtaining the normal operation data required for modeling; S3: selecting variables that affect the change of wind pressure as modeling feature variables; S4: applying the DNN deep learning algorithm, according to the selected feature variables and historical data, training the wind pressure model, and obtaining the wind pressure model; S5: connecting the wind pressure model to the real-time feature data, obtaining the real-time wind pressure theoretical output result, and comparing it with the actual wind pressure value, judging the garbage level in the furnace according to the deviation, and obtaining the indirect characterization of the garbage level.

Description

Technical Field

[0001] The invention belongs to the technical field of garbage incineration, and relates to an automatic control method for a reverse push type garbage incineration mechanical grate feeder. Background Art

[0002] The reverse push type domestic waste incineration mechanical grate is composed of fixed grates and movable grates installed alternately. The movable grate is opposite to the direction of waste movement. The grate is divided into drying section, combustion section, burning section and cooling section. Each section is equipped with a primary air door to provide oxygen for waste combustion. In the reverse push type mechanical grate waste incineration process, the stability of the amount of waste entering the furnace directly determines the quality of the combustion process. If there is too much waste entering the furnace, there will be risks of overheating, overload, and environmental protection parameters exceeding the standard. If there is too little waste entering the furnace, there will be risks of burning out, rapid drop in furnace temperature, or even lower than the environmental protection requirements of 850℃. However, due to the complex composition and changeable form of the waste itself, the waste in the furnace is very easy to be unevenly distributed, and the feeding amount of the waste cannot be simply calculated from the feeding speed; at the same time, due to factors such as the non-ignition and deflagration of the waste, it is easy to cause the operator to make mistakes in judgment, which leads to too much or too little artificial feeding, resulting in the situation where the waste level on the grate is too thick or too thin. How to accurately and effectively judge the waste level in the furnace has become the key to restricting whether the waste can be burned stably.

[0003] In addition, there is a complex environment of high temperature and high smoke in the furnace, and it is difficult to effectively measure the garbage level in the furnace. At present, the only characterization quantity that can indirectly reflect the material layer situation in the furnace is the wind pressure of each wind chamber of the grate, but the actual wind pressure is mostly affected by comprehensive factors such as the output of the primary fan, secondary fan, induced draft fan and the opening of the wind chamber damper. It is difficult for operators to effectively judge the garbage level in the furnace through the change of wind pressure. How to eliminate the external factors that affect the change of wind pressure, effectively judge the material layer situation in the furnace, and then apply it to the feeding and grate automatic control is particularly critical. Summary of the invention

[0004] In view of this, an object of the present invention is to provide an automatic control method for a reverse push type waste incineration mechanical grate feeder.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] An automatic control method for a reverse push type waste incineration mechanical grate feeder, the method comprising the following steps:

[0007] S1: Obtain historical data for more than half a year, including primary air frequency, primary air volume, secondary air frequency, secondary air volume, induced draft fan frequency, primary air pressure, air chamber damper opening, grate lower air chamber pressure, steam flow and furnace temperature data. The primary and secondary air and air chamber pressure data are directly related to the garbage level characterization and are greatly affected by such parameters. The furnace temperature data of the steam flow meter is used as a judgment of the working condition, and the data is classified based on this;

[0008] S2: Screen the historical data and use the K-means clustering algorithm to automatically remove abnormal data from the process of furnace startup and shutdown, equipment failure, and measurement point maintenance, and obtain the normal operation data required for modeling; among them, the data from the process of furnace startup and shutdown, equipment failure, and measurement point maintenance are abnormal data;

[0009] S3: According to the process principle, the variables that affect the wind pressure change are selected as the modeling characteristic variables;

[0010] S4: Apply the DNN deep learning algorithm to train the wind pressure model according to the selected characteristic variables and historical data to obtain the wind pressure model; wherein the configuration of the DNN algorithm is:

[0011] The input data are: primary air frequency, primary air volume, secondary air frequency, secondary air volume, induced draft fan frequency, primary air pressure, air chamber damper opening;

[0012] Output data: garbage level data;

[0013] The number of model layers is: 5 layers, each layer has 100 neurons;

[0014] The activation function is: RELU function;

[0015] The loss function is: mean square error loss function;

[0016] The training process is carried out using the SGD stochastic gradient descent method;

[0017] The number of training times is adjusted according to the data size. If the model accuracy is insufficient, more training data and training are performed;

[0018] S5: Connect the wind pressure model to the real-time characteristic data, obtain the real-time wind pressure theoretical output result, and compare it with the actual wind pressure value. According to the deviation, judge the garbage level in the furnace and obtain the indirect characterization of the garbage level.

[0019] Optionally, in S1, historical operation data of more than half a year is selected, abnormal operating conditions are eliminated through an unsupervised clustering learning method, and data of different normal operating conditions are obtained. Primary air frequency feedback, primary air volume, primary air inlet and outlet air pressure, secondary air frequency, secondary air volume, induced draft fan frequency and each section air door opening parameters are selected as model input features according to the process principle, and a DNN deep learning algorithm is applied to establish a wind pressure model;

[0020] After the modeling is completed, the model is connected to the distributed control system DCS and the characteristic variables to obtain the model output P predict With the actual material level P real Deviation: ΔP err =P predict -P real ;

[0021] By ΔP err The size of the garbage is used to determine the current amount of garbage entering the furnace, and then the feed speed of the feeder and the grate cycle are dynamically adjusted to ensure the uniformity of the feed amount.

[0022] Optionally, the K-means clustering algorithm uses Euclidean distance for calculation, and the distance calculation formula is:

[0023]

[0024] Where X i is the i-th data vector, C j is the center vector of each cluster; K=1…m is the dimension of the vector.

[0025] Optionally, in S5, the deviation between the theoretical value of the wind pressure data model and the actual wind pressure measurement value is connected to the feeder pushing control to guide the pusher to perform pushing control in real time; the deviation between the wind pressure model and the actual wind pressure value is connected to the feeder pushing control to guide the upper grate to perform grate control in real time.

[0026] Optionally, the S5 is specifically:

[0027] By ΔP err The size of the garbage is used to determine the current amount of garbage entering the furnace, and then the feed speed of the feeder and the grate cycle are dynamically adjusted to ensure the uniformity of the feed amount;

[0028] When the actual wind pressure value is higher than the model value by 0.1Kpa, it indicates that the garbage content in the drying section of the furnace is high, and the feed is reduced, and the feed rate is reduced from 2 to 1; when the actual wind pressure value is lower than the model value by 0.1Kpa, it indicates that the garbage content in the drying section of the furnace is low, and the feed is increased, and the feed rate is increased from 2 to 4, and the material level is raised;

[0029] The wind pressure data model is used to sense and judge the situation of the garbage material in the furnace in real time, and control the increase of the feed speed or the decrease of the feed amount.

[0030] The beneficial effects of the present invention are as follows: the DNN deep neural network is used to model the wind pressure in the drying section wind chamber, and a wind pressure model for the wind chamber under different primary air, secondary air, induced draft and different wind chamber damper openings is established. By comparing the actual wind pressure value with the model value, the factors affecting the wind pressure by the garbage in the furnace can be extracted, and the garbage level in the furnace can be characterized, which is then applied to the automatic control of the feeder and the grate to achieve the smooth and automatic operation of the waste incinerator.

[0031] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0033] Figure 1 Modeling architecture for wind pressure models;

[0034] Figure 2 This is the schematic diagram of the real-time feeder control based on the wind pressure model;

[0035] Figure 3 This is the wind pressure model and feed rate control curve of the drying section of a reverse waste incinerator. DETAILED DESCRIPTION

[0036] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0037] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0038] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0039] See also Figure 1 to Figure 3 , is an automatic control method for a reverse push type waste incineration mechanical grate feeder, comprising the following steps:

[0040] S1: Obtain historical data for more than half a year, including primary air frequency, primary air volume, secondary air frequency, secondary air volume, induced draft fan frequency, primary air pressure, air chamber damper opening, grate lower air chamber pressure, steam flow and furnace temperature data;

[0041] S2: Screen the historical data and use the K-means clustering algorithm to automatically remove abnormal data such as furnace startup and shutdown, equipment and measurement point maintenance, and obtain the normal operating data required for modeling;

[0042] S3: According to the process principle, the variables that affect the wind pressure change are selected as the modeling characteristic variables;

[0043] S4: Apply the DNN deep learning algorithm to train the wind pressure model based on the selected characteristic variables and historical data to obtain the wind pressure model. If the model accuracy is insufficient, add training data for training;

[0044] S5: Connect the wind pressure model to the real-time characteristic data, obtain the real-time wind pressure theoretical output result, and compare it with the actual wind pressure value. According to the deviation, judge the garbage level in the furnace and obtain the indirect characterization of the garbage level.

[0045] Furthermore, the deviation between the theoretical value of the wind pressure data model and the actual wind pressure measurement value is connected to the feeder push control to guide the feeder to perform push control in real time;

[0046] Furthermore, the deviation between the wind pressure model value and the actual wind pressure value is connected to the feeder push control to guide the upper grate to perform grate control in real time;

[0047] Furthermore, the K-means clustering analysis algorithm uses Euclidean distance for calculation, and its distance calculation formula is:

[0048]

[0049] Where X i is the i-th data vector, C j is the center vector of each cluster; K = 1…m is the dimension of the vector

[0050] Furthermore, the wind pressure model is advanced in that it applies big data modeling capabilities. Through the big data model, the effects of primary air, secondary air, induced draft and air chamber damper on wind pressure can be fully considered, so the changes in the material layer in the reactor can be more intuitive.

[0051] Furthermore, the garbage material level characterization method and application method apply the model deviation to the feeding and grate control, avoiding the risk of further accelerating the pushing of materials and causing material piles when there is material in the furnace but it is not ignited. It can also promptly sense when the material in the furnace is insufficient and quickly replenish the material, thereby ensuring the automatic long-term operation of the feeding and grate systems.

[0052] The historical operation data of more than half a year was selected, and the abnormal working conditions were eliminated through the unsupervised clustering learning method to obtain the data of different normal working conditions. According to the process principle, the primary air frequency feedback, primary air volume, primary air inlet and outlet air pressure, secondary air frequency, secondary air volume, induced draft fan frequency and each section of the air door opening were selected as the model input features, and the DNN deep learning algorithm was used to establish the wind pressure model. The whole modeling process is as follows: Figure 1 shown.

[0053] After the modeling is completed, the model is connected to the DCS real-time system and the characteristic variables to obtain the model output P predict Deviation from the actual wind pressure P: ΔP err =P predict -P real ;

[0054] By ΔP err The size of the garbage is used to determine the current amount of garbage entering the furnace, and then the feed speed of the feeder and the grate cycle are dynamically adjusted to ensure the uniformity of the feed amount. Figure 2 shown. Figure 3 This is the wind pressure model and feed speed control curve of the drying section of a reverse waste incinerator, and the horizontal axis is the time point. The actual wind pressure value is significantly higher than the model value of 0.1Kpa. At this time, it can be characterized as indicating that the garbage content in the drying section of the furnace is high, and the feed should be reduced. The feed speed is reduced from 2 to around 1, the feed speed is significantly reduced, and the feed amount is reduced; the actual wind pressure value is significantly lower than the model value of 0.1Kpa. At this time, it can be characterized as indicating that the garbage content in the drying section of the furnace is low, and the feed should be increased. The feed speed is increased from 2 to around 4, the feed speed is significantly increased, the feed amount is increased, and the subsequent material level is slowly raised. The wind pressure data model can be used to perceive and judge the situation of the garbage in the furnace in real time, and then guide the feed speed to increase or decrease the feed amount, solving the problem of difficult judgment of garbage in the furnace and uneven feeding.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. An automatic control method for a reverse-thrust waste incineration mechanical grate feeder, characterized in that: The method comprises the following steps: S1: Obtain historical data for more than half a year, including primary air frequency, primary air volume, secondary air frequency, secondary air volume, induced draft fan frequency, primary air pressure, air chamber damper opening, grate lower air chamber pressure, steam flow and furnace temperature data. The primary and secondary air and air chamber pressure data are directly related to the garbage level characterization and are greatly affected by such parameters. The furnace temperature data of the steam flow meter is used as a judgment of the working condition, and the data is classified based on this; S2: Screen the historical data and use the K-means clustering algorithm to automatically remove abnormal data from the process of furnace startup and shutdown, equipment failure, and measurement point maintenance, and obtain the normal operation data required for modeling; among them, the data from the process of furnace startup and shutdown, equipment failure, and measurement point maintenance are abnormal data; S3: According to the process principle, the variables that affect the wind pressure change are selected as the modeling characteristic variables; S4: Apply the DNN deep learning algorithm to train the wind pressure model according to the selected characteristic variables and historical data to obtain the wind pressure model; wherein the configuration of the DNN algorithm is: The input data are: primary air frequency, primary air volume, secondary air frequency, secondary air volume, induced draft fan frequency, primary air pressure, air chamber damper opening; Output data: garbage level data; The number of model layers is: 5 layers, each layer has 100 neurons; The activation function is: RELU function; The loss function is: mean square error loss function; The training process is carried out using the SGD stochastic gradient descent method; The number of training times is adjusted according to the data size. If the model accuracy is insufficient, more training data and training are performed; S5: Connect the wind pressure model to the real-time characteristic data, obtain the real-time wind pressure theoretical output result, and compare it with the actual wind pressure value. According to the deviation, judge the garbage level in the furnace and obtain the indirect characterization of the garbage level.

2. The automatic control method of a reverse push type waste incineration mechanical grate feeder according to claim 1 is characterized in that: In S1, historical operation data of more than half a year is selected, abnormal operating conditions are eliminated through unsupervised clustering learning methods, and data of different normal operating conditions are obtained. According to the process principle, primary air frequency feedback, primary air volume, primary air inlet and outlet air pressure, secondary air frequency, secondary air volume, induced draft fan frequency and each section air door opening parameters are selected as model input features, and the DNN deep learning algorithm is applied to establish the air pressure model; After the modeling is completed, the model is connected to the distributed control system DCS and the characteristic variables to obtain the model output P predict With the actual material level P real Deviation: ΔP err =P predict -P real ; By ΔP err The size of the garbage is used to determine the current amount of garbage entering the furnace, and then the feed speed of the feeder and the grate cycle are dynamically adjusted to ensure the uniformity of the feed amount.

3. The automatic control method of a reverse push type waste incineration mechanical grate feeder according to claim 1 is characterized in that: The K-means clustering algorithm uses Euclidean distance for calculation, and the distance calculation formula is: Where X i is the i-th data vector, C j is the center vector of each cluster; K=1…m is the dimension of the vector.

4. The automatic control method of a reverse push type waste incineration mechanical grate feeder according to claim 2 is characterized in that: In S5, the deviation between the theoretical value of the wind pressure data model and the actual wind pressure measurement value is connected to the feeder push control to guide the pusher to perform push control in real time; the deviation between the wind pressure model and the actual wind pressure value is connected to the feeder push control to guide the upper grate to perform grate control in real time.

Images