Solid waste incineration working condition intelligent sensing and self-adaptive adjustment method

Abstract

The present application relates to solid waste incineration working condition adjustment technical field, especially, it relates to a kind of solid waste incineration working condition intelligent sensing and self-adapting adjustment method, realize the direct quantitative sensing of combustion state in furnace, based on Darcy law calculation layer equivalent permeability, based on Stefan-Boltzmann law calculation local effective heat release flux, can reflect the air permeability and combustion intensity of layer in real time, fundamentally solve the problem of traditional parameter lag.Equivalent heat-pneumatic impedance factor is proposed as comprehensive working condition evaluation index, the coupling relationship of heat release and gas flow is uniformly represented, a scientific quantitative basis is provided for control decision, and the control precision is significantly improved.A double-layer control architecture of adaptive fuzzy neural network combined model predictive control is constructed, the former realizes the multivariate coordinated regulation of primary air, secondary air, feeding and speed of grate, and the latter realizes the precise predictive control of auxiliary burner, and the working condition stability and adjustment speed are greatly improved by the cooperation of the two.

Description

Technical Field

[0001] This invention relates to the field of solid waste incineration operating condition adjustment technology, and in particular to a method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions. Background Technology

[0002] Municipal solid waste incineration is the mainstream technological route for achieving waste reduction, harmlessness, and resource recovery. Mechanical grate incinerators are widely used due to their large processing capacity and stable operation. However, my country's municipal solid waste is characterized by complex composition, high moisture content, and large fluctuations in calorific value, making the combustion conditions within the incinerator furnace prone to instability. This leads to problems such as incomplete combustion, excessive flue gas pollutants, and coking and corrosion of the furnace body. Traditional incineration control methods mostly rely on manual experience or single-parameter PID regulation, which can only passively control based on lagging parameters such as furnace outlet temperature and flue gas oxygen concentration. They cannot perceive the actual combustion state of the material bed in the furnace in real time and are unable to cope with the complex conditions of strong coupling between heat release and gas flow. There is an urgent need to develop an intelligent control technology that can accurately perceive the furnace conditions in real time and achieve multi-variable collaborative adaptive regulation.

[0003] Existing solid waste incineration operating condition control technologies have the following significant shortcomings: First, the sensing layer suffers from severe lag, only detecting downstream parameters such as the furnace outlet and flue, failing to directly acquire core combustion state information such as the permeability of the material layer above the grate and the intensity of local heat release. This inherent delay in control adjustment easily leads to significant fluctuations in operating conditions. Second, the control logic is fragmented, failing to consider the strong coupling relationship between heat release and gas flow. Single-parameter or simple multi-parameter PID control is prone to overshoot or oscillation when there are sudden changes in material layer thickness or calorific value, even causing material layer burn-through or flameout accidents. Third, the intelligent control methods have poor generalization ability. Most solutions use a single neural network or fuzzy control, relying on training data under specific operating conditions, making it difficult to cover complex scenarios with different seasons, different waste compositions, and different processing loads. Fourth, the main and auxiliary combustion systems are decoupled. The auxiliary burner only passively starts when the furnace temperature is below the threshold, unable to coordinate with the adjustment actions of primary air, feeding, etc., resulting in high auxiliary fuel consumption and slow operating condition recovery. Fifth, there is a lack of unified comprehensive evaluation indicators for operating conditions. Control decisions are mostly based on experience thresholds and lack scientific quantitative basis, which can easily lead to erroneous adjustments. Summary of the Invention

[0004] The main objective of this invention is to provide an intelligent sensing and adaptive adjustment method for solid waste incineration operating conditions. Firstly, it achieves direct quantitative sensing of the combustion state within the furnace. Based on Darcy's law, it calculates the equivalent permeability of the feed bed, and based on the Stefan-Boltzmann law, it calculates the local effective heat release flux, enabling real-time reflection of feed bed permeability and combustion intensity, fundamentally solving the problem of lag in traditional parameters. Secondly, it proposes an equivalent thermal-aerodynamic impedance factor as a comprehensive operating condition evaluation index, uniformly characterizing the coupling relationship between heat release and gas flow, providing a scientific quantitative basis for control decisions, and significantly improving control accuracy. Thirdly, it constructs a two-layer control architecture combining an adaptive fuzzy neural network and model predictive control. The former achieves multi-variable coordinated adjustment of primary air, secondary air, feed, and grate speed, while the latter achieves precise predictive control of the auxiliary burner. The synergy between the two significantly improves operating condition stability and adjustment speed.

[0005] The technical solution of the present invention is as follows:

[0006] A method for intelligent sensing and adaptive adjustment of solid waste incineration conditions is proposed, which includes the following steps:

[0007] S1. Synchronously sample the operating status data inside the incinerator furnace, including the material layer thickness, primary air chamber static pressure, primary air volume flow rate, local surface radiation temperature, and flue gas carbon monoxide concentration.

[0008] S2. Substitute the material layer thickness, primary air chamber static pressure, and primary air volume flow rate into the permeability calculation formula based on Darcy's law to obtain the equivalent permeability of the material layer.

[0009] S3. Based on the local surface radiation temperature, flue gas carbon monoxide concentration, and primary air volume flow rate, the local effective heat release flux is obtained.

[0010] S4. Based on the equivalent permeability of the material layer and the local effective heat release flux, combined with the static pressure of the primary air chamber, the coupled state modulus is synthesized and the trend is extracted. The equivalent thermal-aerodynamic resistance factor and the rate of change of operating conditions are output, and the benign cooperative steady-state range of the incinerator is preset to obtain the operating condition deviation.

[0011] S5. Input the equivalent thermal-aerodynamic resistance factor and the operating condition change rate into the adaptive fuzzy neural network controller, and output the primary air volume adjustment increment, secondary air volume adjustment increment, feeding rate adjustment increment and grate movement speed adjustment increment.

[0012] S6. Input the equivalent thermo-aerodynamic impedance factor, operating condition change rate and the current operating power of the auxiliary burner into the model prediction and control module to generate the auxiliary burner power adjustment increment.

[0013] S7. The primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate movement speed adjustment increment, and auxiliary burner power adjustment increment are combined into a control command vector and sent to the corresponding adjustment layer of the incinerator to complete the adaptive adjustment of the solid waste incineration conditions.

[0014] A further improvement of this invention is that the expression for the equivalent permeability of the material layer in S2 is:

[0015] ;

[0016] in, The equivalent permeability of the material layer, Aerodynamic viscosity, For material layer thickness, The primary air volumetric flow rate, The effective cross-sectional area of ​​the grate, This refers to the static pressure of the primary air chamber.

[0017] A further improvement of this invention is that the expression for the local effective heat release flux in S3 is:

[0018] ;

[0019] in, For local effective heat release flux, For the surface emissivity of the material layer, Let be the blackbody radiation constant. The local surface radiation temperature, is the enthalpy conversion coefficient for carbon monoxide combustion.

[0020] A further improvement of the present invention is that step S4 includes the following specific steps:

[0021] S41, Based on equivalent permeability of the material layer With local effective heat release flux Combined with the static pressure of the primary air chamber Coupled state modulus synthesis is performed to obtain the equivalent thermal-aerodynamic resistance factor. , ;

[0022] S42, Equivalent thermo-aerodynamic resistance factor The rate of change of the operating condition is obtained by taking the first time derivative. , The first-order time derivative is calculated by difference between the values ​​at two adjacent sampling times, and the formula is as follows: ;in, The sampling period is t, and the index of the sampling time is t.

[0023] S43, Pre-set benign and coordinated steady-state range for the incinerator ,in, This represents the lower limit of the benign, coordinated, steady-state range for the incinerator. The upper limit of the benign and coordinated steady-state range of the incinerator is used to obtain the operating condition deviation. , , This represents the median value of the benign, coordinated, steady-state range for the incinerator.

[0024] A further improvement of the present invention is that the specific content of S5 is: to determine the working condition deviation. and operating condition change rate An adaptive fuzzy neural network controller is input, which adopts a five-layer feedforward network structure, including an input layer, a fuzzification layer, a rule layer, a normalization layer, and an output layer. The input layer has two input nodes, corresponding to the operating condition deviations. and operating condition change rate The fuzzification layer sets 5 fuzzification nodes for each input variable, for a total of 10 nodes. Each node corresponds to a fuzzy subset. The fuzzy subsets for both input variables are set as {NB, NS, ZO, PS, PB}, where NB corresponds to a large negative deviation, NS corresponds to a small negative deviation, ZO corresponds to no deviation in the steady-state range, PS corresponds to a small positive deviation, and PB corresponds to a large positive deviation. Each fuzzification node uses a Gaussian membership function to perform the fuzzification calculation of the input variables. The calculation formula is as follows: Where i is the index of the input variable, and i=1 corresponds to the operating condition deviation. The rate of change of the operating condition corresponding to i=2 j is the index of the fuzzy subset, j=1 corresponds to NB, j=2 corresponds to NS, j=3 corresponds to ZO, j=4 corresponds to PS, and j=5 corresponds to PB. Let the membership function center value be the value of the fuzzy subset of the ith input variable. For the corresponding width parameter, the output is the membership value of each fuzzy subset corresponding to each input variable. The rule layer has 25 rule nodes, each corresponding to a fuzzy rule. The number of rule nodes matches the number of fuzzy subset combinations of the two input variables. Each rule node receives the corresponding membership value from the fuzzification layer and uses a product operation to calculate the fit of the fuzzy rule. The output of each node is the product of the two corresponding membership values, calculated as follows: Where k is the index of the rule node, ranging from 1 to 25; m and n are the fuzzy subset indices corresponding to the two input variables, respectively; the fourth layer is the normalization layer, with 25 normalization nodes, each corresponding one-to-one with the rule layer nodes. The normalization fit weight is calculated using the following formula: The output layer is configured with four output nodes, each corresponding to a primary airflow adjustment increment. Secondary air volume adjustment increment Feed rate adjustment increment Incremental adjustment of grate movement speed The calculation formula is: ;in, This represents the p-th adjustment increment output by the adaptive fuzzy neural network controller, where p is the index of the output node, and p=1 corresponds to... p=2 corresponds to p=3 corresponds to p=4 corresponds to , , , These are the coefficients of the linear function corresponding to the k-th rule of the p-th output node.

[0025] A further improvement of the present invention is that the model predictive control module in S6 takes the equivalent thermo-aerodynamic impedance factor, the rate of change of operating conditions, and the current operating power of the auxiliary burner as inputs, and uses the power adjustment increment of the auxiliary burner as input. As the output, a controlled autoregressive integral moving average model is used as the prediction model, and its expression is: ;in, For the shift operator, For difference operators, For the model, white noise, and The model polynomial has the following expressions: ; ; It is a second-order shift operator. The coefficients are polynomials, and the order of the model is set to... , The pure time delay is set to one sampling period, and the model predictive control module uses a rolling optimization method to obtain the power adjustment increment of the auxiliary burner. The objective function is optimized as follows: ;in, This is the predicted value of the equivalent thermo-aerodynamic resistance factor at the k-th time in the future. The weighting coefficient is used to control the increment.

[0026] A further improvement of the present invention is that the specific content of S7 is as follows: the primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate movement speed adjustment increment, and auxiliary burner power adjustment increment are combined into a control command vector and sent to the corresponding adjustment layer of the incinerator. When the equivalent thermal-aerodynamic resistance factor is greater than the upper limit of the benign cooperative steady-state range of the incinerator, it is determined that the permeability of the material layer in the furnace is poor and there is local oxygen deficiency. The adjustment actions of inhibiting the increase of feed, increasing the secondary air volume, and increasing the grate movement speed are executed. When the equivalent thermal-aerodynamic resistance factor is less than the lower limit of the benign cooperative steady-state range of the incinerator, it is determined that the material layer is burned through and forms an airflow short circuit. The adjustment actions of reducing the primary air volume in the corresponding area and accelerating the feed rate are executed.

[0027] The technical effects of this invention are as follows:

[0028] A method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions was developed. First, it achieves direct quantitative sensing of the combustion state within the furnace. Based on Darcy's law, the equivalent permeability of the feed bed is calculated, and based on the Stefan-Boltzmann law, the local effective heat release flux is calculated. This allows for real-time reflection of feed bed permeability and combustion intensity, fundamentally solving the problem of lag in traditional parameters. Second, an equivalent thermal-aerodynamic impedance factor is proposed as a comprehensive operating condition evaluation index, uniformly characterizing the coupling relationship between heat release and gas flow, providing a scientific quantitative basis for control decisions and significantly improving control accuracy. Third, a two-layer control architecture combining adaptive fuzzy neural networks and model predictive control is constructed. The former enables multi-variable coordinated adjustment of primary air, secondary air, feed, and grate speed, while the latter enables precise predictive control of the auxiliary burner. The synergy between the two significantly improves operating condition stability and adjustment speed. Attached Figure Description

[0029] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0030] Figure 1 This is a flowchart illustrating an intelligent sensing and adaptive adjustment method for solid waste incineration operating conditions according to Embodiment 1 of the present invention. Detailed Implementation

[0031] Example 1: This example proposes an intelligent sensing and adaptive adjustment method for solid waste incineration operating conditions. Firstly, it achieves direct quantitative sensing of the combustion state within the furnace. Based on Darcy's law, it calculates the equivalent permeability of the feed bed, and based on the Stefan-Boltzmann law, it calculates the local effective heat release flux, reflecting the feed bed permeability and combustion intensity in real time, fundamentally solving the problem of lag in traditional parameters. Secondly, it proposes an equivalent thermal-aerodynamic impedance factor as a comprehensive operating condition evaluation index, uniformly characterizing the coupling relationship between heat release and gas flow, providing a scientific quantitative basis for control decisions, and significantly improving control accuracy. Thirdly, it constructs a two-layer control architecture combining an adaptive fuzzy neural network and model predictive control. The former achieves multi-variable coordinated adjustment of primary air, secondary air, feed, and grate speed, while the latter achieves precise predictive control of the auxiliary burner. The synergy of both significantly improves operating condition stability and adjustment speed. Specifically, as shown... Figure 1 As shown in the figure, the intelligent sensing and adaptive adjustment method for solid waste incineration operating conditions proposed in this embodiment includes the following specific steps:

[0032] S1. Synchronously sample the operating status data inside the incinerator furnace, including the material layer thickness, primary air chamber static pressure, primary air volume flow rate, local surface radiation temperature, and flue gas carbon monoxide concentration.

[0033] In this embodiment, the operating status data within the incinerator furnace is synchronously sampled. This data includes bed thickness, primary air chamber static pressure, primary air volumetric flow rate, local surface radiation temperature, and flue gas carbon monoxide concentration. Bed thickness is measured using laser rangefinders installed above the grate. These sensors are evenly spaced along the grate width, with a sampling frequency of 1Hz for each measurement point. The overall bed thickness is obtained by arithmetic averaging of the measurements from multiple points. Primary air chamber static pressure is measured using pressure transmitters installed at the outlet of each primary air chamber. The pressure transmitters have a measurement range of 0 to 5000 Pa and a sampling frequency of 10Hz. Primary air volumetric flow rate is measured using vortex flow meters installed on the primary air main pipe. The vortex flow meters have a measurement range of 0 to 50000 Pa. The sampling frequency was set to 10Hz. Local surface radiation temperature was measured using an infrared thermal imager installed on the furnace sidewall. The infrared thermal imager's measurement range was 0 to 1500℃, and the sampling frequency was set to 5Hz. By dividing the images acquired by the infrared thermal imager into regions and performing temperature statistics, the local surface radiation temperature of different areas above the grate was obtained. The carbon monoxide concentration in the flue gas was measured using an infrared gas analyzer installed at the furnace outlet. The infrared gas analyzer's measurement range was 0 to 5000ppm, and the sampling frequency was set to 1Hz. All sensor sampling data was transmitted to the central controller via industrial Ethernet. The central controller synchronously processed data from different sampling frequencies, with a control cycle of 1 second. The average value of the sensor measurements within each control cycle was used as the input data for subsequent calculations.

[0034] S2. Substitute the material layer thickness, primary air chamber static pressure, and primary air volume flow rate into the permeability calculation formula based on Darcy's law to obtain the equivalent permeability of the material layer.

[0035] In this embodiment, the expression for the equivalent permeability of the material layer in S2 is:

[0036] ;

[0037] in, The equivalent permeability of the material layer, Aerodynamic viscosity, For material layer thickness, The primary air volumetric flow rate, The effective cross-sectional area of ​​the grate, This refers to the static pressure of the primary air chamber.

[0038] In this embodiment, the bed thickness, primary air chamber static pressure, and primary air volumetric flow rate are substituted into the permeability calculation formula based on Darcy's law to obtain the equivalent permeability of the bed. Darcy's law describes the relationship between pressure drop and flow velocity when a fluid flows in a porous medium. For gas flow in a fixed bed, when the flow is in a laminar state, Darcy's law can be expressed as follows: ;in, The pressure drop of the gas passing through the bed. For bed thickness, For fluid dynamic viscosity, The apparent velocity of the fluid in the bed. The permeability of the bed. In a solid waste incinerator, primary air passes upward through the material bed from the primary air chamber, and the static pressure of the primary air chamber... Approximately equal to the pressure drop of the gas passing through the material layer The apparent velocity of primary air in the material layer It can be expressed as primary air volumetric flow rate. With the effective cross-sectional area of ​​the grate The ratio, i.e. Substituting the above relationships into Darcy's law expression and rearranging, we obtain the formula for calculating the equivalent permeability of the substrate layer: ;in, The equivalent permeability of the material layer is expressed in units of 1000 m³ / s. , Here, aerodynamic viscosity is expressed in Pa·s. Within the operating temperature range of the incinerator, aerodynamic viscosity can be approximated as [value missing]. Pa・s, This refers to the thickness of the material layer, in meters (m). The primary air volumetric flow rate is expressed in units of... s, The effective cross-sectional area of ​​the grate is given in units of... For a specific incinerator, the effective cross-sectional area of ​​the grate is a fixed value; The static pressure of the primary air chamber is expressed in Pa. This formula converts the directly measured bed thickness, static pressure of the primary air chamber, and primary air volumetric flow rate into quantitative indicators that directly reflect the gas flow capacity of the bed. A higher equivalent permeability indicates better permeability, making it easier for gas to pass through; a lower equivalent permeability indicates poorer permeability, resulting in greater resistance to gas flow.

[0039] S3. Based on the local surface radiation temperature, flue gas carbon monoxide concentration, and primary air volume flow rate, the local effective heat release flux is obtained.

[0040] In this embodiment, the expression for the local effective heat release flux in S3 is:

[0041] ;

[0042] in, For local effective heat release flux, For the surface emissivity of the material layer, Let be the blackbody radiation constant. The local surface radiation temperature, is the enthalpy conversion coefficient for carbon monoxide combustion.

[0043] In this embodiment, the local effective heat release flux is obtained based on the local surface radiation temperature, flue gas carbon monoxide concentration, and primary air volumetric flow rate. The local effective heat release flux represents the actual heat released per unit area of ​​the fuel layer on the grate per unit time; it is equal to the radiative heat flux on the fuel layer surface minus the heat flux lost due to incomplete combustion. The radiative heat flux on the fuel layer surface can be calculated using the Stefan-Boltzmann law, which describes the relationship between the radiative heat flux and surface temperature of a blackbody surface. For non-blackbody surfaces, the radiative heat flux can be expressed as… ,in, For surface radiative heat flux, For the surface emissivity of the material layer, Let be the blackbody radiation constant. The surface temperature is [value missing]. The heat flux loss due to incomplete combustion is mainly caused by the incomplete combustion of carbon monoxide in the flue gas. The enthalpy of combustion of carbon monoxide is [value missing]. The heat released by the complete combustion of a unit mass of carbon monoxide is The mass flow rate of carbon monoxide in flue gas can be expressed as: ,in, This refers to the volume concentration of carbon monoxide in the flue gas. Let be the air density. Therefore, the heat flux due to incomplete combustion per unit area of ​​grate can be expressed as... The expression for the heat flux loss due to incomplete combustion is rearranged, and the enthalpy conversion coefficient of carbon monoxide combustion is introduced. ,in The heat flux lost due to incomplete combustion can be expressed as: The local effective heat release flux equals the surface radiative heat flux of the bed minus the heat flux lost due to incomplete combustion. Therefore, the formula for calculating the local effective heat release flux is: ;in, Local effective heat release flux, in units of , The surface emissivity of the feed bed is approximately taken as 0.85 for solid waste incineration feed beds. Let be the blackbody radiation constant, and its value is . , This refers to the local surface radiation temperature, expressed in Kelvin (K). This is the enthalpy conversion coefficient for carbon monoxide combustion, in units of... Under standard conditions, the enthalpy of combustion of carbon monoxide is approximately 12640 kJ / m³. Therefore, the enthalpy conversion coefficient of carbon monoxide combustion can be approximated as . , This refers to the carbon monoxide concentration in flue gas, expressed in ppm, dimensionless (ppm is a unit of volume fraction). ), The primary air volumetric flow rate is expressed in units of... , The effective cross-sectional area of ​​the grate is given in units of... This calculation formula can transform the directly measured local surface radiation temperature, flue gas carbon monoxide concentration, and primary air volume flow rate into quantitative indicators that directly reflect the actual heat release intensity of the bed. The larger the local effective heat release flux, the more intense the combustion of the bed and the more heat is released per unit time; the smaller the local effective heat release flux, the slower the combustion of the bed and the less heat is released per unit time.

[0044] S4. Based on the equivalent permeability of the material layer and the local effective heat release flux, combined with the static pressure of the primary air chamber, the coupled state modulus is synthesized and the trend is extracted. The equivalent thermal-aerodynamic resistance factor and the rate of change of operating conditions are output, and the benign cooperative steady-state range of the incinerator is preset to obtain the operating condition deviation.

[0045] In this embodiment, step S4 includes the following specific steps:

[0046] S41, Based on equivalent permeability of the material layer With local effective heat release flux Combined with the static pressure of the primary air chamber Coupled state modulus synthesis is performed to obtain the equivalent thermal-aerodynamic resistance factor. , ;

[0047] S42, Equivalent thermo-aerodynamic resistance factor The rate of change of the operating condition is obtained by taking the first time derivative. , The first-order time derivative is calculated by difference between the values ​​at two adjacent sampling times, and the formula is as follows: ;in, The sampling period is t, and the index of the sampling time is t.

[0048] S43, Pre-set benign and coordinated steady-state range for the incinerator ,in, This represents the lower limit of the benign, coordinated, steady-state range for the incinerator. The upper limit of the benign and coordinated steady-state range of the incinerator is used to obtain the operating condition deviation. , , This represents the median value of the benign, coordinated, steady-state range for the incinerator.

[0049] In this embodiment, based on the equivalent permeability of the material layer With local effective heat release flux Combined with the static pressure of the primary air chamber Coupled state modulus synthesis is performed to obtain units of The equivalent thermal-aerodynamic impedance factor comprehensively reflects the coupling relationship between the heat release intensity and gas flow resistance in the furnace. When the equivalent thermal-aerodynamic impedance factor is within a reasonable range, it indicates that the heat release and gas flow in the furnace are in a coordinated state, and the combustion conditions are stable. When the equivalent thermal-aerodynamic impedance factor is too large, it indicates that the heat release intensity in the furnace is too high or the gas flow resistance is too large, which can easily lead to local oxygen deficiency and incomplete combustion. When the equivalent thermal-aerodynamic impedance factor is too small, it indicates that the heat release intensity in the furnace is too low or the gas flow resistance is too small, which can easily lead to bed burn-through and airflow short circuit. The equivalent thermal-aerodynamic impedance factor... The rate of change of the operating condition is obtained by taking the first time derivative. The first-order time derivative is obtained by calculating the difference between the values ​​at two adjacent sampling times. The sampling period is 1 second (s), the same as the control period. t is the index of the sampling time (s). The rate of change of operating conditions reflects the rate of change of the equivalent thermo-aerodynamic impedance factor over time. The larger the absolute value of the rate of change, the faster the incineration conditions change, and the more unstable the system. The smaller the absolute value of the rate of change, the slower the incineration conditions change, and the more stable the system. Next, proceed to step S43, presetting the benign cooperative steady-state range for the incinerator. For a mechanical grate incinerator with a capacity of 500 t / d, the benign cooperative steady-state range can be preset as follows: This leads to the operating condition deviation. For the aforementioned preset benign cooperative steady-state range, the median value of the benign cooperative steady-state range for the incinerator is... Operating condition deviation reflects the degree of deviation between the current equivalent thermal-aerodynamic resistance factor and the ideal steady-state value. The larger the absolute value of the operating condition deviation, the greater the deviation of the current incineration operating condition from the ideal steady state, and the greater the adjustment range required. The smaller the absolute value of the operating condition deviation, the closer the current incineration operating condition is to the ideal steady state, and the smaller the adjustment range required.

[0050] S5. Input the equivalent thermal-aerodynamic resistance factor and the operating condition change rate into the adaptive fuzzy neural network controller, and output the primary air volume adjustment increment, secondary air volume adjustment increment, feeding rate adjustment increment and grate movement speed adjustment increment.

[0051] In this embodiment, the specific content of S5 is: to determine the operating condition deviation. and operating condition change rate An adaptive fuzzy neural network controller is input, which adopts a five-layer feedforward network structure, including an input layer, a fuzzification layer, a rule layer, a normalization layer, and an output layer. The input layer has two input nodes, corresponding to the operating condition deviations. and operating condition change rate The fuzzification layer sets 5 fuzzification nodes for each input variable, for a total of 10 nodes. Each node corresponds to a fuzzy subset. The fuzzy subsets for both input variables are set as {NB, NS, ZO, PS, PB}, where NB corresponds to a large negative deviation, NS corresponds to a small negative deviation, ZO corresponds to no deviation in the steady-state range, PS corresponds to a small positive deviation, and PB corresponds to a large positive deviation. Each fuzzification node uses a Gaussian membership function to perform the fuzzification calculation of the input variables. The calculation formula is as follows: Where i is the index of the input variable, and i=1 corresponds to the operating condition deviation. The rate of change of the operating condition corresponding to i=2 j is the index of the fuzzy subset, j=1 corresponds to NB, j=2 corresponds to NS, j=3 corresponds to ZO, j=4 corresponds to PS, and j=5 corresponds to PB. Let be the center value of the membership function of the j-th fuzzy subset of the i-th input variable, where the unit is when i=1. When i=2, the unit is , For the corresponding width parameter, and the corresponding The units are the same. Regarding operating condition deviations... The membership function center values ​​of its fuzzy subsets are respectively set as follows: , , , , The corresponding width parameter is set to For the rate of change of operating conditions The membership function center values ​​of its fuzzy subsets are respectively set as follows: , , , , The corresponding width parameters are set to The output of the fuzzification layer is the membership value of each input variable for each fuzzy subset. The membership value ranges from [0,1], representing the degree to which the input variable belongs to that fuzzy subset. The rule layer has 25 rule nodes, each corresponding to a fuzzy rule. The number of rule nodes matches the number of fuzzy subset combinations of the two input variables. Each rule node receives the corresponding membership value from the fuzzification layer and uses a product operation to calculate the fit of the fuzzy rule. The output of each node is the product of the two corresponding membership values, calculated as follows: Where k is the index of the rule node, ranging from 1 to 25; m and n are the fuzzy subset indices corresponding to the two input variables, each ranging from 1 to 5; the correspondence between k and m, n is k = (m-1) × 5 + n; the output of the rule layer is the fitness value of each fuzzy rule; the larger the fitness value, the higher the degree of activation of the fuzzy rule. The fourth layer is the normalization layer, with 25 normalization nodes, corresponding one-to-one with the rule layer nodes, to calculate the normalization fitness weights, using the following formula: The output of the normalization layer is the normalized fit weight of each fuzzy rule, and the sum of all normalized fit weights is 1. The output layer has four output nodes, each corresponding to a primary airflow adjustment increment. Secondary air volume adjustment increment Feed rate adjustment increment Incremental adjustment of grate movement speed The calculation formula is: ;in, This represents the p-th adjustment increment output by the adaptive fuzzy neural network controller, where p is the index of the output node, and p=1 corresponds to... p=2 corresponds to p=3 corresponds to p=4 corresponds to , , , The coefficients of the linear function corresponding to the k-th rule of the p-th output node are... Dimensions and same, Let be the operating condition deviation coefficient corresponding to the k-th rule of the p-th output node, with dimensions . Divide the dimensions by , Let be the coefficient of the rate of change of operating conditions corresponding to the k-th rule of the p-th output node, with dimensions . Divide the dimensions by .

[0052] In this embodiment, the adaptive fuzzy neural network controller is trained using a backpropagation algorithm combined with gradient descent. During training, a suitable training dataset needs to be constructed first. The construction of the training dataset includes four parts: data source selection, data preprocessing, input / output sample pair construction, and dataset partitioning. The data source is selected from actual operating data of the incinerator under different solid waste compositions, different processing loads, and different seasons. The solid waste compositions include various combinations of 40% to 60% kitchen waste, 10% to 20% plastic, 10% to 15% paper, and 15% to 30% other waste. The processing load covers 60% to 110% of the rated processing capacity. The data collection period is no less than 3 months, and the collected data volume is no less than 1 million sets. Data preprocessing includes two steps: outlier removal and data normalization. Outlier removal uses… In principle, for each sensor's measurement data sequence, its mean is calculated. and standard deviation , will satisfy Data points were identified as outliers and removed. After outlier removal, linear interpolation was used to fill in the missing data. Data normalization employed a min-max normalization method, mapping the range of all input and output variables to the [0,1] interval. The normalization formula is as follows: , For the normalized data, The minimum value of this variable across all data; This represents the maximum value of the variable across all data. Input-output sample pairs are constructed using operating condition deviation and rate of change as inputs, and the actual optimal control quantity as the output. The actual optimal control quantity is obtained through statistical analysis of operator manual adjustment records or historical automatic control data from historical operation data where the incineration operating conditions are within a benign coordinated steady-state range. For each set of input data, the control quantity that causes the equivalent thermo-aerodynamic impedance factor to return to the benign coordinated steady-state range fastest and with the smallest fluctuation within the next 10 seconds is selected as the corresponding output data. The dataset is randomly partitioned, with all input-output sample pairs divided into training, validation, and test sets at a ratio of 70%, 15%, and 15%, respectively. The training set is used to adjust the controller parameters, the validation set is used to monitor for overfitting during training, and the test set is used to evaluate the generalization performance of the controller after training. During training, the mean square error between the controller output and the actual optimal control quantity is used as the loss function, calculated as follows: , where N is the number of training set samples; s is the sample number; This represents the actual output of the p-th output node for the s-th sample. This represents the expected output of the p-th output node for the s-th sample. The training process uses batch gradient descent to update parameters, with the initial learning rate set to 0.01. During training, the learning rate is dynamically adjusted based on changes in the validation set loss function. If the validation set loss function does not decrease for 5 consecutive training cycles, the learning rate is halved. If the validation set loss function does not decrease for 10 consecutive training cycles, training is terminated early, and the model parameters with the minimum validation set loss function are saved. After training, the adaptive fuzzy neural network controller can automatically output appropriate increments for primary air volume, secondary air volume, feed rate, and grate speed based on input operating condition deviations and rates of change, achieving adaptive adjustment of the incineration conditions.

[0053] S6. Input the equivalent thermo-aerodynamic impedance factor, operating condition change rate and the current operating power of the auxiliary burner into the model prediction and control module to generate the auxiliary burner power adjustment increment.

[0054] In this embodiment, the model prediction control module in S6 uses the equivalent thermo-aerodynamic impedance factor, the rate of change of operating conditions, and the current operating power of the auxiliary burner as inputs, and uses the power adjustment increment of the auxiliary burner as input. As the output, a controlled autoregressive integral moving average model is used as the prediction model, and its expression is: ;in, For the shift operator, satisfying , For difference operators, For the model, white noise, and The model polynomial has the following expressions: ; ; It is a second-order shift operator. The coefficients are polynomials, and the order of the model is set to... , The pure time delay is set to one sampling period, consistent with the control period of 1 second. This pure time delay corresponds to the minimum response time for a measurable change in the furnace thermal state after the auxiliary burner power adjustment. The model predictive control module uses a rolling optimization method to obtain the auxiliary burner power adjustment increment. The objective function is optimized as follows: ;in, This is the predicted value of the equivalent thermo-aerodynamic resistance factor at the k-th time in the future. The weighting coefficient is used to control the increment.

[0055] In this embodiment, to ensure the consistency of dimensions on both sides of the prediction model equation, the input and output variables need to be dimensionless before being substituted into the model for calculation. The dimensionless calculation formula is as follows: ,in The variable is dimensionless; x is the original variable. The nominal value of the variable; This is a scaling factor for the variable. For the equivalent thermo-aerodynamic resistance factor, the nominal value is taken as the median value within the benign cooperative steady-state range of the incinerator. Scaling factor For the auxiliary burner power adjustment increment, the nominal value is taken as 0, the scaling factor is taken as 50kW, and the dimensionless equivalent thermo-aerodynamic resistance factor is... and auxiliary burner power adjustment increment Substitute the values ​​into the above prediction model for calculation.

[0056] In this embodiment, the model predictive control module uses a rolling optimization method to obtain the auxiliary burner power adjustment increment. The objective function is optimized as follows: ; ; represents the predicted value of the equivalent thermo-aerodynamic resistance factor at the k-th time in the future. To control the weighting coefficient of the increment, For the prediction time domain, a sampling period of 10 seconds (10 sampling periods) is used, covering the entire time range during which auxiliary burner power regulation significantly affects the furnace thermal state. For the control time domain, M is set to 3 sampling periods (3 seconds), effectively reducing online computation while maintaining control accuracy. The weighting coefficients for the control increments are also specified. Setting it to 0.1 is used to balance control performance with the magnitude of changes in the control quantity. The higher the value, the more stable the auxiliary burner power regulation, but the slower the operating condition recovery speed; The smaller the value, the faster the operating condition recovers, but the greater the fluctuations during the adjustment process.

[0057] In this embodiment, the coefficients of the model polynomial , , , , The identification was obtained through systematic analysis of actual incinerator operating data, using a recursive least squares method. The identification data originated from step response test data of the incinerator within the rated load range of 70% to 110%. During the test, primary air volume, secondary air volume, feed rate, and grate speed were kept constant; only the auxiliary burner power was changed. The incremental adjustment of the auxiliary burner power and the corresponding changes in the equivalent thermal-aerodynamic impedance factor were recorded. The data acquisition length was no less than 10,000 sets, with a sampling period of 1 second. The data was preprocessed before identification using... Outliers were removed as a rule, and then high-frequency noise was filtered out using a first-order low-pass filter with a cutoff frequency of 0.1Hz. The preprocessed data was then fitted using a recursive least squares method to obtain the typical coefficients of the model polynomial. , , , , The specific coefficient values ​​for different incinerators can be obtained using the same identification method.

[0058] In the rolling optimization calculation process, the dimensionless equivalent thermo-aerodynamic impedance factor at the current time t and historical times is first used as the basis. , ...and dimensionless auxiliary burner power regulation increment , ...Substitute the values ​​into the prediction model to calculate the dimensionless equivalent thermo-aerodynamic resistance factor predictions for the next P time points. The value of k ranges from 1 to P, and then the predicted value is converted into a dimensional value. Substitute these values ​​into the optimization objective function. To ensure that the control quantity varies within the allowable range of the equipment and avoids impacting the equipment, two sets of constraints are introduced during the optimization process. The first set is the control increment constraint, which limits the maximum change in auxiliary burner power within each control cycle, expressed as: Where k takes values ​​from 0 to M-1, The lower limit for the auxiliary burner power regulation increment is set to -50kW. The upper limit for the auxiliary burner power adjustment increment is set to 50kW. The second group consists of control constraints, limiting the operating power of the auxiliary burner within its rated operating range; the expression is: The value of k ranges from 1 to P. The predicted operating power of the auxiliary burner at the k-th time in the future is calculated using the following recursive formula: ; To ensure the minimum stable operating power of the auxiliary burner, For a mechanical grate incinerator with a capacity of 500 t / d, the rated power of the auxiliary burner is set to 2000 kW. After introducing constraints, the original optimization problem is transformed into a quadratic programming problem with linear constraints. The effective set method is used to solve this quadratic programming problem, yielding the optimal dimensionless auxiliary burner power adjustment increment at the current moment. Then, the actual auxiliary burner power regulation is obtained through inverse normalization calculation. And use it as the output of the model predictive control module.

[0059] S7. The primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate movement speed adjustment increment, and auxiliary burner power adjustment increment are combined into a control command vector and sent to the corresponding adjustment layer of the incinerator to complete the adaptive adjustment of the solid waste incineration conditions.

[0060] In this embodiment, S7 specifically involves: composing a control command vector from the primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate speed adjustment increment, and auxiliary burner power adjustment increment, and sending it to the corresponding adjustment layer of the incinerator. When the equivalent thermal-aerodynamic resistance factor is greater than the upper limit of the benign cooperative steady-state range of the incinerator, it is determined that the permeability of the material layer inside the furnace is poor and there is local oxygen deficiency. Adjustments are then made to suppress the increase in feed, increase the secondary air volume, and increase the grate speed. When the equivalent thermal-aerodynamic resistance factor is less than the lower limit of the benign cooperative steady-state range of the incinerator, it is determined that the material layer has burned through, forming an airflow short circuit. Adjustments are then made to reduce the primary air volume in the corresponding area and accelerate the feed rate.

[0061] Example 2: This example provides an electronic device, including a processor and a memory, wherein the memory stores a computer program that can be called by the processor; the processor executes the above-mentioned method for intelligent sensing and adaptive adjustment of solid waste incineration conditions by calling the computer program stored in the memory.

[0062] The electronic device can vary considerably depending on its configuration or performance. It may include one or more Central Processing Units (CPUs) and one or more memories, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement the intelligent sensing and adaptive adjustment method for solid waste incineration operations provided in the above-described embodiment. The electronic device may also include other components for implementing its functions; for example, it may have wired or wireless network interfaces and input / output interfaces for data input and output. Details will not be elaborated upon in this embodiment.

[0063] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0064] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0065] This invention is described with reference to flowchart illustrations and block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and block diagrams, as well as combinations of blocks in the flowchart illustrations and block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions, characterized in that: The specific steps include the following: S1. Synchronously sample the operating status data inside the incinerator furnace, including the material layer thickness, primary air chamber static pressure, primary air volume flow rate, local surface radiation temperature, and flue gas carbon monoxide concentration. S2. Substitute the material layer thickness, primary air chamber static pressure, and primary air volume flow rate into the permeability calculation formula based on Darcy's law to obtain the equivalent permeability of the material layer. S3. Based on the local surface radiation temperature, flue gas carbon monoxide concentration, and primary air volume flow rate, the local effective heat release flux is obtained. S4. Based on the equivalent permeability of the material layer and the local effective heat release flux, combined with the static pressure of the primary air chamber, the coupled state modulus is synthesized and the trend is extracted. The equivalent thermal-aerodynamic resistance factor and the rate of change of operating conditions are output, and the benign cooperative steady-state range of the incinerator is preset to obtain the operating condition deviation. S5. Input the equivalent thermal-aerodynamic resistance factor and the operating condition change rate into the adaptive fuzzy neural network controller, and output the primary air volume adjustment increment, secondary air volume adjustment increment, feeding rate adjustment increment and grate movement speed adjustment increment. S6. Input the equivalent thermo-aerodynamic impedance factor, operating condition change rate and the current operating power of the auxiliary burner into the model prediction and control module to generate the auxiliary burner power adjustment increment. S7. The primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate movement speed adjustment increment, and auxiliary burner power adjustment increment are combined into a control command vector and sent to the corresponding adjustment layer of the incinerator to complete the adaptive adjustment of the solid waste incineration conditions.

2. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 1, characterized in that: The expression for the equivalent permeability of the material layer in S2 is: ； in, The equivalent permeability of the material layer, Aerodynamic viscosity, For material layer thickness, The primary air volumetric flow rate, The effective cross-sectional area of ​​the grate, This refers to the static pressure of the primary air chamber.

3. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 2, characterized in that: The expression for the local effective heat release flux in S3 is: ； in, For local effective heat release flux, For the surface emissivity of the material layer, Let be the blackbody radiation constant. The local surface radiation temperature, is the enthalpy conversion coefficient for carbon monoxide combustion.

4. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 3, characterized in that: S4 includes the following specific steps: S41, Based on equivalent permeability of the material layer With local effective heat release flux Combined with the static pressure of the primary air chamber Coupled state modulus synthesis is performed to obtain the equivalent thermal-aerodynamic resistance factor. , ; S42, Equivalent thermo-aerodynamic resistance factor The rate of change of the operating condition is obtained by taking the first time derivative. , The first-order time derivative is calculated by difference between the values ​​at two adjacent sampling times, and the formula is as follows: ;in, The sampling period is t, and the index of the sampling time is t. S43, Pre-set benign and coordinated steady-state range for the incinerator ,in, This represents the lower limit of the benign, coordinated, steady-state range for the incinerator. The upper limit of the benign and coordinated steady-state range of the incinerator is used to obtain the operating condition deviation. , , This represents the median value of the benign, coordinated, steady-state range for the incinerator.

5. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 4, characterized in that: The specific content of S5 is: to determine the operating condition deviation. and operating condition change rate An adaptive fuzzy neural network controller is input, which adopts a five-layer feedforward network structure, including an input layer, a fuzzification layer, a rule layer, a normalization layer, and an output layer. The input layer has two input nodes, corresponding to the operating condition deviations. and operating condition change rate The fuzzification layer sets 5 fuzzification nodes for each input variable, for a total of 10 nodes. Each node corresponds to a fuzzy subset. The fuzzy subsets for both input variables are set as {NB, NS, ZO, PS, PB}, where NB corresponds to a large negative deviation, NS corresponds to a small negative deviation, ZO corresponds to no deviation in the steady state range, PS corresponds to a small positive deviation, and PB corresponds to a large positive deviation. Each fuzzification node uses a Gaussian membership function to perform the fuzzification calculation of the input variables. The calculation formula is as follows: Where i is the index of the input variable, and i=1 corresponds to the operating condition deviation. The rate of change of the operating condition corresponding to i=2 j is the index of the fuzzy subset, j=1 corresponds to NB, j=2 corresponds to NS, j=3 corresponds to ZO, j=4 corresponds to PS, and j=5 corresponds to PB. Let the membership function center value be the value of the fuzzy subset of the ith input variable. For the corresponding width parameter, the output is the membership value of each fuzzy subset corresponding to each input variable. The rule layer has 25 rule nodes, each corresponding to a fuzzy rule. The number of rule nodes matches the number of fuzzy subset combinations of the two input variables. Each rule node receives the corresponding membership value from the fuzzification layer and uses a product operation to calculate the fit of the fuzzy rule. The output of each node is the product of the two corresponding membership values, calculated as follows: Where k is the index of the rule node, ranging from 1 to 25; m and n are the fuzzy subset indices corresponding to the two input variables, respectively; the fourth layer is the normalization layer, with 25 normalization nodes, each corresponding one-to-one with the rule layer nodes. The normalization fitness weight is calculated using the following formula: The output layer is configured with four output nodes, each corresponding to a primary airflow adjustment increment. Secondary air volume adjustment increment Feed rate adjustment increment Incremental adjustment of grate movement speed The calculation formula is: ;in, This represents the p-th adjustment increment output by the adaptive fuzzy neural network controller, where p is the index of the output node, and p=1 corresponds to... p=2 corresponds to p=3 corresponds to p=4 corresponds to , , , These are the coefficients of the linear function corresponding to the k-th rule of the p-th output node.

6. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 5, characterized in that: The model predictive control module in S6 takes the equivalent thermo-aerodynamic impedance factor, the rate of change of operating conditions, and the current operating power of the auxiliary burner as inputs, and the power adjustment increment of the auxiliary burner as inputs. As the output, a controlled autoregressive integral moving average model is used as the prediction model, and its expression is: ;in, For the shift operator, For difference operators, For the model, white noise, and The model polynomial has the following expressions: ; ; It is a second-order shift operator. The coefficients are polynomials, and the order of the model is set to... , The pure time delay is set to one sampling period, and the model predictive control module uses a rolling optimization method to obtain the power adjustment increment of the auxiliary burner. The objective function is optimized as follows: ;in, This is the predicted value of the equivalent thermo-aerodynamic resistance factor at the k-th time in the future. The weighting coefficient is used to control the increment.

7. The method for intelligent sensing and adaptive adjustment of solid waste incineration operating conditions according to claim 6, characterized in that: The specific content of S7 is as follows: the primary air volume adjustment increment, secondary air volume adjustment increment, feed rate adjustment increment, grate movement speed adjustment increment, and auxiliary burner power adjustment increment are combined into a control command vector and sent to the corresponding adjustment layer of the incinerator. When the equivalent thermal-aerodynamic resistance factor is greater than the upper limit of the benign cooperative steady-state range of the incinerator, it is determined that the permeability of the material layer in the furnace is poor and there is local oxygen deficiency. The adjustment actions of inhibiting the increase of feed, increasing the secondary air volume, and increasing the grate movement speed are executed. When the equivalent thermal-aerodynamic resistance factor is less than the lower limit of the benign cooperative steady-state range of the incinerator, it is determined that the material layer is burned through and forms an airflow short circuit. The adjustment actions of reducing the primary air volume in the corresponding area and accelerating the feed rate are executed.

Images