Fly ash low-temperature pyrolysis optimization control method and system based on multi-modal data
By identifying the latent heat trap state through a multimodal data processing unit and switching to asynchronous pulse control mode, the problem of control integral saturation and oscillation caused by the nonlinear phase change of materials during pyrolysis is solved, and the stability and energy consumption optimization of the fly ash low-temperature pyrolysis system are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG HUIHEYUAN ENVIRONMENTAL TECH CO LTD
- Filing Date
- 2026-06-12
- Publication Date
- 2026-07-14
AI Technical Summary
Existing technologies cannot effectively identify the nonlinear phase transitions of materials during pyrolysis when treating special solid waste containing high concentrations of inorganic salts and time-varying moisture. This leads to excessive stacking of integral terms in the control system, causing temperature overshoot and valve oscillation. Furthermore, the technology cannot respond promptly to changes in material composition, resulting in increased energy consumption and excessive flue gas emissions.
The latent heat trap state is determined by the multimodal data processing unit, the continuous feedback control loop is forcibly suspended, and the asynchronous pulse control mode is switched to generate discrete timing pulse opening commands. Combined with a multidimensional safety constraint closed loop, the heat source flow rate is precisely regulated to avoid integral saturation and command oscillation.
It effectively eliminates integral saturation and control command overshoot, reduces energy consumption, ensures stable operation and environmental performance of the system under extreme conditions, and avoids excessive flue gas emissions.
Smart Images
Figure CN122386638A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of industrial control system technology, and in particular relates to an optimized control method and system for low-temperature pyrolysis of fly ash based on multimodal data. Background Technology
[0002] Current multivariable composite control systems for large hysteresis temperature fields typically employ an architecture that integrates feedback control loops with multi-path feedforward compensation strategies. Temperature measurement components deployed inside the reactor collect real-time apparent temperatures, compare them with set benchmark parameters to generate deviation signals, and a proportional-integral-derivative controller calculates the opening command of the heat source feed valve to smooth out temperature disturbances introduced into the pyrolysis center reaction zone by fluctuations in the feed of multiple materials.
[0003] However, when the control system deals with special solid waste containing high concentrations of inorganic salts and time-varying moisture, the material undergoes a violent phase change and absorbs a large amount of latent heat in a specific process section. This causes local stagnation in the apparent temperature data fed back by the temperature sensing components. Because conventional continuous feedback calculation loops cannot detect the state jumps caused by sudden changes in the material, they continue to accumulate residual deviations from temperature targets based on fixed periodic sampling data. This leads to excessive stacking of internal integral terms and continuous blind amplification of valve opening commands. To suppress this control overshoot caused by large time delays and nonlinear coupling of phase changes, conventional improvement approaches usually focus on establishing a static correction network by introducing additional offline detection data of material composition, or slowing down the system's command response speed by reducing the gain coefficient of the proportional-integral-derivative controller. However, under continuous large-scale pressure fluctuations, simply reducing the controller gain cannot eliminate the static deviations in conventional process sections in a timely manner, and parameter compensation based on a fixed time constant cannot adapt to the dynamic time delay evolution caused by changes in material flow rate and composition. Optimizing the pyrolysis physical configuration or changing the shape of the mechanical rollers are also necessary to address this issue. The improved control of material distribution loses its dynamic adjustment flexibility once the hardware is manufactured and formed, making it unable to counteract the heat absorption jumps in the physical field with varying compositions. At the same time, the existing supporting software algorithm control also has shortcomings. For example, Chinese invention patent CN119692181B discloses a multimodal data-driven low-temperature pyrolysis optimization control method and system for fly ash dioxins. It constructs a neural network with shared multimodal features to estimate the pyrolysis rate and uses the gradient descent algorithm to solve for the overall optimal pyrolysis temperature and heat treatment time. This purely data-driven control method relies on a global continuous mapping network constructed from historical static samples. When facing fly ash materials with drastic fluctuations in composition in industrial settings, the local transient phase change heat absorption of the material causes the physical field to exhibit an extremely nonlinear state fracture. This instantaneous local apparent temperature stagnation is sudden and discrete. The neural network and gradient optimization architecture are based on the continuous gradient and the preset overall steady-state characteristics. They have a blind spot for perceiving the transient physical changes across seconds and cannot respond to the sudden latent heat traps and switch control laws in real time at the surface time sequence, causing control time lag and phase interference.
[0004] Therefore, the technical problem to be solved by this invention is how to adaptively identify the nonlinear phase transition control dead zone of the pyrolysis physical field through multi-source heterogeneous parameter fusion, and to realize the conditional suspension of the continuous loop, the phase advance injection of the open-loop pulse, and the precise reset of the internal historical error integral register at the microprocessor bottom layer, so as to eliminate integral saturation and instruction oscillation under the large hysteresis heat treatment condition. Summary of the Invention
[0005] This invention aims to solve the problems of integral saturation and control command overshoot oscillation caused by nonlinear phase transitions of materials in the continuous calculation loop of the control system.
[0006] In this technical solution, a method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data includes the following steps: Step S1: The processing unit obtains the real-time furnace temperature parameters under controlled conditions through the temperature sensing external element, and simultaneously reads the real-time valve opening command value sent to the controlled heat source flow regulation mechanism. Step S2: The processing unit calculates the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period. When the real-time valve opening command value is greater than the basic load threshold stored in the memory connected to the processing unit, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the memory in three consecutive sampling periods, the processing unit determines that the controlled operating condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of the internal material phase change. Step S3: During the transient control cycle in which the latent heat trap state is determined to be established, the processing unit forcibly suspends the analog output channel of the currently calculated conventional feedback adjustment command, terminates the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switches to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the memory, it generates a discontinuous pulse opening command with discrete timing characteristics and directly uses the pulse opening command as the control law input to act on the controlled heat source flow regulation mechanism until the absolute value of the real-time temperature rise slope parameter in the subsequent continuous sampling cycle is greater than the stagnation threshold. Then, it is determined that the controlled operating condition has moved out of the phase change stagnation section and exited the latent heat trap state.
[0007] Preferably, in step S1, flue gas monitoring data is acquired; when it triggers the environmental protection threshold, the conventional feedback adjustment command is blocked and the opening of the auxiliary air volume valve drive mechanism is adjusted so that the controlled operating condition is stabilized within the process safety range.
[0008] Preferably, step S1 includes: calculating the real-time communication delay value using timestamps within the communication channel, and disconnecting the external channel and switching to the safety backup feedback loop when the delay exceeds a safe time threshold.
[0009] Preferably, step S3 includes: calling the dead zone compensation operator to calculate the reverse compensation amount before outputting the command and superimposing it into the conventional feedback regulation command to generate an actual control signal and send it to the controlled heat source flow regulation mechanism.
[0010] Preferably, step S3 includes: accumulating the current valve reference opening by a fixed step size during the high-level segment to generate a pulse opening command, and resetting the pulse opening command to the current valve reference opening during the low-level segment.
[0011] Preferably, step S2 includes: recording the cumulative duration of latent heat trap states to generate a condition deviation index, and outputting a warning signal and adjusting the subsequent base load threshold when the deviation exceeds the abnormal alarm threshold.
[0012] Preferably, the basic load threshold stored in the memory is 60% to 75% of the maximum opening of the controlled heat source flow regulation mechanism, the stagnation threshold stored in the memory is 0.1℃ to 0.3℃, and the continuous sampling period is 5s to 10s.
[0013] Preferably, step S1 includes: performing median filtering on the original measurement signal to generate real-time furnace temperature parameters, and performing full-range normalization transformation on the real-time valve opening command value.
[0014] Preferably, step S3 includes: when the absolute value of the real-time temperature rise slope parameter is greater than the stagnation threshold for two consecutive sampling cycles, the phase transition section is determined to end, and the output of the normal feedback adjustment command is restored.
[0015] A low-temperature pyrolysis optimization control system for fly ash based on multimodal data is used to implement a low-temperature pyrolysis optimization control method for fly ash based on multimodal data. The system includes a temperature sensing module, a controlled heat source flow rate regulation module, a data storage module, and a logic processing module. The logic processing module is communicatively connected to the temperature sensing module, the controlled heat source flow regulation module, and the data storage module, respectively. The logic processing module is configured to obtain the real-time furnace temperature parameters under controlled conditions through the temperature sensing module and simultaneously read the real-time valve opening command value of the controlled heat source flow regulation module. The logic processing module is also configured to calculate the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period, and to determine that the controlled condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of internal material phase change when the real-time valve opening command value is greater than the basic load threshold delay range stored in the data storage module, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the data storage module in three consecutive sampling periods. The logic processing module is also configured to forcibly suspend the analog output channel of the currently calculated conventional feedback adjustment command in the latent heat trap state, terminate the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switch to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the data storage module, it generates intermittent pulse opening commands with discrete timing characteristics, and uses the pulse opening commands as the control law input to the controlled heat source flow regulation module until the absolute value of the real-time temperature rise slope parameter in subsequent continuous sampling cycles is greater than the stagnation threshold. At this point, it is determined that the controlled condition has moved out of the phase change stagnation section and exited the latent heat trap state.
[0016] Compared with existing technologies, the fly ash low-temperature pyrolysis optimization control method based on multimodal data of the present invention has the following advantages: 1. In the optimized control of fly ash low-temperature pyrolysis using multimodal data, the controlled pyrolysis space is determined to enter a latent heat stagnation state caused by material phase change by using a dual conditional allocation of the temperature rise slope parameter and the main heat source valve opening command value. Subsequently, the underlying control state machine is called to forcibly suspend the conventional proportional-integral-derivative continuous calculation loop and freeze the output port, blocking the integral saturation phenomenon caused by the temperature stagnation deviation inside the controller. Simultaneously, the pre-component data is introduced to calculate the compensation enthalpy value required to cross the latent heat stagnation state and convert it into an open-loop pulse control command of a specific width. The interrupt channel bypasses the periodic communication queue and directly writes it into the analog output module, driving the lower valve to maintain a full-load heating state within the corresponding pulse width time. After the pulse width time ends, the historical error integral register in the continuous calculation loop is cleared and the closed-loop control process is restored, so that the timing of the control command issuance is phase-aligned with the material phase change energy flow, avoiding temperature overshoot and continuous valve oscillation caused by excessive accumulation of integrals at the end of the phase change due to continuous calculation.
[0017] 2. The processing unit introduces energy consumption unit price characteristic parameters and heat conversion efficiency parameters into the dynamic energy supply system. In each calculation cycle, it compares the real-time unit prices of multiple heat sources, integrates the heat conversion efficiency parameters to calculate the current marginal utility cost ratio of each energy supply medium, and dynamically updates the scheduling allocation coefficient in the basic energy scheduling strategy according to the marginal utility cost ratio. It assigns the heat source with the current cost advantage to share the basic heat load. This adaptive energy supply allocation logic and open-loop pulse control command achieve nonlinear scheduling coupling at the logic layer, so that the economic boundary constraint of the current cycle directly restricts the allocation of system energy supply commands, avoiding the increase in energy consumption caused by the fixed ratio of multiple energy sources in traditional control and the deterioration of operating economy caused by fluctuations in the price of energy supply medium.
[0018] 3. The processing unit establishes a multi-dimensional safety constraint closed loop. It collects oxygen content parameters and pollutant concentration data in parallel through a sensor network. An emergency-level interlocking protection topology is constructed outside the conventional adjustment path of the closed-loop control command. When the real-time pollutant concentration data triggers the preset environmental protection threshold, the control logic directly blocks the analog output channel of the microprocessor to prevent the continued issuance of conventional closed-loop control commands. Simultaneously, it activates the pre-stored automated emergency program interlocking logic in the local memory, allocates secondary air adjustment compensation signals, and adjusts the corresponding valve actions to bring the pyrolysis physical conditions back to the safe range limited by the process baseline constraint parameters. This safety control mechanism does not rely on an external dedicated independent safety system. It opens up the response path from data acquisition to control arbitration at the control algorithm level, avoiding the risk of excessive flue gas emissions caused by the delay or overshoot of the heat replenishment command in extreme conditions due to the large lag physical system. Attached Figure Description
[0019] Figure 1 This is a flowchart of the optimized control method for low-temperature pyrolysis of fly ash according to the present invention.
[0020] Figure 2 This is a state structure diagram of the fly ash low-temperature pyrolysis optimization control system of the present invention. Detailed Implementation
[0021] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.
[0022] A method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data includes the following steps: Step S1: The processing unit obtains the real-time furnace temperature parameters under controlled conditions through the temperature sensing external element, and simultaneously reads the real-time valve opening command value sent to the controlled heat source flow regulation mechanism. Step S2: The processing unit calculates the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period. When the real-time valve opening command value is greater than the basic load threshold stored in the memory connected to the processing unit, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the memory in three consecutive sampling periods, the processing unit determines that the controlled operating condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of the internal material phase change. Step S3: During the transient control cycle in which the latent heat trap state is determined to be established, the processing unit forcibly suspends the analog output channel of the currently calculated conventional feedback adjustment command, terminates the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switches to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the memory, it generates a discontinuous pulse opening command with discrete timing characteristics and directly uses the pulse opening command as the control law input to act on the controlled heat source flow regulation mechanism until the absolute value of the real-time temperature rise slope parameter in the subsequent continuous sampling cycle is greater than the stagnation threshold. Then, it is determined that the controlled operating condition has moved out of the phase change stagnation section and exited the latent heat trap state.
[0023] Preferably, in step S1, flue gas monitoring data is acquired; when it triggers the environmental protection threshold, the conventional feedback adjustment command is blocked and the opening of the auxiliary air volume valve drive mechanism is adjusted so that the controlled operating condition is stabilized within the process safety range.
[0024] Preferably, step S1 includes: calculating the real-time communication delay value using timestamps within the communication channel, and disconnecting the external channel and switching to the safety backup feedback loop when the delay exceeds a safe time threshold.
[0025] Preferably, step S3 includes: calling the dead zone compensation operator to calculate the reverse compensation amount before outputting the command and superimposing it into the conventional feedback regulation command to generate an actual control signal and send it to the controlled heat source flow regulation mechanism.
[0026] Preferably, step S3 includes: accumulating the current valve reference opening by a fixed step size during the high-level segment to generate a pulse opening command, and resetting the pulse opening command to the current valve reference opening during the low-level segment.
[0027] Preferably, step S2 includes: recording the cumulative duration of latent heat trap states to generate a condition deviation index, and outputting a warning signal and adjusting the subsequent base load threshold when the deviation exceeds the abnormal alarm threshold.
[0028] Preferably, the basic load threshold stored in the memory is 60% to 75% of the maximum opening of the controlled heat source flow regulation mechanism, the stagnation threshold stored in the memory is 0.1℃ to 0.3℃, and the continuous sampling period is 5s to 10s.
[0029] Preferably, step S1 includes: performing median filtering on the original measurement signal to generate real-time furnace temperature parameters, and performing full-range normalization transformation on the real-time valve opening command value.
[0030] Preferably, step S3 includes: when the absolute value of the real-time temperature rise slope parameter is greater than the stagnation threshold for two consecutive sampling cycles, the phase transition section is determined to end, and the output of the normal feedback adjustment command is restored.
[0031] Preferably, a fly ash low-temperature pyrolysis optimization control system based on multimodal data includes a temperature sensing module, a controlled heat source flow regulation module, a data storage module, and a logic processing module. The logic processing module is communicatively connected to the temperature sensing module, the controlled heat source flow regulation module, and the data storage module, respectively. The logic processing module is configured to obtain the real-time furnace temperature parameters under controlled conditions through the temperature sensing module and simultaneously read the real-time valve opening command value of the controlled heat source flow regulation module. The logic processing module is also configured to calculate the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period, and to determine that the controlled condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of internal material phase change when the real-time valve opening command value is greater than the basic load threshold delay range stored in the data storage module, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the data storage module in three consecutive sampling periods. The logic processing module is also configured to forcibly suspend the analog output channel of the currently calculated conventional feedback adjustment command in the latent heat trap state, terminate the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switch to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the data storage module, it generates intermittent pulse opening commands with discrete timing characteristics, and uses the pulse opening commands as the control law input to the controlled heat source flow regulation module until the absolute value of the real-time temperature rise slope parameter in subsequent continuous sampling cycles is greater than the stagnation threshold. At this point, it is determined that the controlled condition has moved out of the phase change stagnation section and exited the latent heat trap state.
[0032] Example 1: In a low-temperature pyrolysis control environment of fly ash at a continuously operating industrial solid waste disposal site, where the control system employs the method claimed in this invention, when the system faces the continuous injection of fly ash material containing high concentrations of inorganic salts and time-varying moisture, the crystal water and volatile chlorides inside the fly ash material undergo an endothermic phase change when the pyrolysis temperature rises to a specific critical threshold. This forms a latent heat trap section in the physical thermal field, causing high-frequency random fluctuations in the apparent temperature. If the conventional continuous closed-loop control architecture relying solely on furnace temperature feedback is used, the control system often implicitly assumes that the thermal response curve of the controlled object evolves linearly. This leads to the temperature sensing component reading a continuous deviation signal indicating that the temperature has not reached the target within the latent heat trap section. Consequently, the integral term of the proportional-integral-derivative (PID) regulator inside the control system is excessively stacked, continuously amplifying the opening command of the energy supply valve. When the material phase change ends instantaneously and the latent heat trap closes, the excessively accumulated control quantity will transfer excessive energy to subsequent process stages, inducing an overshoot spike in the actual pyrolysis temperature and wasted energy in the energy supply medium.
[0033] The processing unit acquires real-time furnace temperature parameters under controlled operating conditions through a temperature sensing external element and simultaneously reads the real-time valve opening command value sent to the controlled heat source flow regulation mechanism. To eliminate high-frequency random thermocouple measurement noise interference in the industrial environment, after acquiring the initial continuous measurement signal, the processing unit calls a sliding time window averaging filtering algorithm to extract the original measurement values from the ten consecutive discrete sampling periods preceding the current moment, performing arithmetic mean integration to generate smoothed real-time furnace temperature parameters. Based on the fundamental physical theorems of energy conservation and overall heat conduction, it calculates the difference in smoothed real-time furnace temperature parameter values between two adjacent sampling periods, divides it by the corresponding time interval to generate the variation, and calculates the real-time temperature rise slope parameter based on the variation of the real-time furnace temperature parameters within the continuous sampling period. When the real-time valve opening command value is greater than the value stored in the memory connected to the processing unit... When the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the memory for three consecutive sampling periods, the controlled condition is determined to be in a latent heat trap state due to the heat absorption of internal material phase change. During the transient control period in which the latent heat trap state is determined, the processing unit forcibly suspends the analog output channel of the currently calculated conventional feedback adjustment command, terminates the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switches to asynchronous pulse control mode. It calls the pre-loaded multimodal component data and generates a discontinuous pulse opening command with discrete timing characteristics based on the fixed step size parameter pre-stored in the memory. The open-loop pulse control command containing the pulse width parameter is directly written to the analog output module through the interrupt channel, bypassing the periodic communication queue. The calculation formula of the pulse width parameter is as follows: ,in, To calculate the generated pulse width parameter, To obtain the physical quality parameters of the current batch of fly ash, The characteristic value of the extracted volatile matter mass concentration, The theoretical latent heat constant is set for this volatile compound. The reference heat release power of the main heat source valve at its maximum physical opening. The processing unit drives the controlled heat source flow regulation mechanism at pulse width parameters, which serves as the system heat transfer efficiency conversion factor. Within a limited time frame, a full-load start-up physical action is performed to pre-inject counter-heat energy into the pyrolysis space before the actual physical temperature drops, and the duration of the open-loop pulse control command reaches the pulse width parameter. At this time, a truncation instruction is written to the analog output module, and the historical error integral register in the pending proportional-integral-differential continuous calculation loop is cleared, restoring the normal periodic closed-loop instruction calculation process. During this energy alignment process, considering the time scale difference between the large time delay of the overall valve heating and the transient heat absorption of the surface phase change, the pre-injected counteracting thermal energy does not directly intervene in the surface phase change interface. Instead, it stores the feedforward thermal energy in advance in the physical inner wall of the reactor and the high-temperature gaseous medium through full-load strong drive, forming an overall thermal capacity buffer zone. When the fly ash material undergoes a violent surface latent heat phase change due to the temperature reaching the critical point, this thermal capacity buffer zone uses the previously predicted local high temperature gradient as the physical medium for energy transfer, passively and instantaneously releasing the latent heat dissipation required in the form of radiation and forced convection. The energy is thus indirectly coordinated and aligned with the energy compensation based on the overall physical heat storage prediction and the surface phase change dissipation time sequence in essence, in terms of thermodynamic transfer. The cumulative occurrence time of a single occurrence of latent heat trap state is compared with the abnormal alarm time threshold stored in the memory. According to the negative deviation compensation principle of industrial control system, if the cumulative occurrence time of a single occurrence is greater than the abnormal alarm time threshold, the difference between the cumulative occurrence time of a single occurrence and the abnormal alarm time threshold is extracted. The ratio of the difference to the abnormal alarm time threshold is calculated to generate the duration deviation rate. The basic correction step size is called and multiplied by the duration deviation rate to generate the threshold correction increment. The current basic load threshold is added to the threshold correction increment to generate the updated basic load threshold and overwrite the original storage unit data. The basic correction step size is preset to 2% of the maximum physical opening of the controlled heat source flow regulation mechanism.
[0034] After receiving the open-loop pulse control command directly issued by the processing unit, bypassing the communication queuing cycle, the controlled heat source flow regulation mechanism completes a determined high-level full-load heat replenishment action and a low-level restoration of the normal reference opening within a specific pulse width time. This avoids the valve reciprocating oscillation and subsequent temperature overshoot defects caused by the large physical time delay and phase change nonlinearity in the traditional control architecture. The processing unit introduces energy consumption unit price characteristic parameters and heat conversion efficiency parameters into the dynamic energy supply system. In each calculation cycle, it compares and calculates the real-time unit price and heat conversion efficiency parameters of multiple heat sources to generate a dynamic update mechanism for the basic energy dispatch strategy. The current marginal utility cost ratio of the degree allocation coefficient is used to assign the heat source with the best current cost to share the basic heat load. This adaptive energy allocation logic and the above-mentioned open-loop pulse control command are coupled nonlinearly at the logic layer, so that the dynamic variance of the furnace temperature parameter fed back by the external temperature sensing element converges when facing the aforementioned compositional fluctuation conditions. It also completes the constraint alignment of the multivariable system of the energy supply medium, so that the entire industrial control system can compensate for the spatiotemporal misalignment at the physical level by the state jump of the control logic and the reprogramming of the command timing under material disturbance conditions. This allows the phase change resistance of the physical world to be directly transformed into a steady-state control output that meets the requirements of safe, environmentally friendly and stable operation within the control domain.
[0035] Example 2: This example verifies the actual performance of the method and system claimed in this invention in resolving runaway phase transition nonlinear control. The test was conducted on a fly ash low-temperature pyrolysis industrial programmable test platform with multi-source data sensing capabilities. It includes a temperature sensing external element for monitoring the internal state of the controlled operating conditions. The measurement resolution of the temperature sensing external element is set to 0.1℃, the measurement accuracy to 0.2℃, and the data sampling frequency to 10Hz. The fly ash low-temperature pyrolysis industrial programmable test platform is connected to a physical transformation unit for weighing the continuously injected material. The weighing resolution of the physical transformation unit is set to 0.5kg. To simulate electromagnetic interference and operating condition disturbances in an industrial setting, a 50Hz power frequency interference harmonic and standard deviation are introduced into the data acquisition bus. To measure the noise of a random thermocouple at 1.2℃, in this controlled thermal environment, the processing unit determines the first preset sampling period. The specific actions for determining the first preset sampling period include: the microprocessor reads the mass of material weighed by the physical transformation unit and calculates the mass injection change rate; simultaneously, it collects the transient temperature drop change rate through the temperature sensing external element; when the mass injection change rate is greater than 10% / min and the transient temperature drop change rate is greater than 1.5℃ / s, the microprocessor shortens the sampling window, switching the value of the first preset sampling period from the initial 2.0s to the lower limit of the working window, 0.5s. Under the condition that the fly ash continuous injection rate is 50kg / h and the moisture content is 15%, the microprocessor performs the above actions to determine the specific value of the first preset sampling period as 0.5s.
[0036] The experiment used the volatile components and water of crystallization content in fly ash as variable parameters to characterize the intensity of disturbances under controlled conditions, and divided them into three levels: low concentration, medium concentration, and high concentration to complete gradient tests. A multidimensional control system was also established. Control group one used a continuous closed-loop control loop with temperature feedback and no pulse control mode. Control group two, while maintaining the control architecture of this invention, fixed the pulse width parameter to a constant and closed the multimodal data input channel. The sample group of this invention set the stagnation threshold to 0.2℃ / s and initiated the complete control action of the method of this invention. In the initial test section where the material was at a low concentration, the initial temperature parameters collected by the temperature sensing external element fluctuated frequently between 280.4℃ and 285.6℃ due to the aforementioned noise. After the material entered the pyrolysis zone, the control group 1, lacking a latent heat trap identification mechanism, experienced a continuous accumulation of deviations in its continuous feedback calculation loop regarding the temperature stagnation signal caused by the endothermic phase change. This resulted in the real-time valve opening command value being amplified from the initial 35.2% to 98.5% of the maximum limit. Consequently, after the material phase change ended and the latent heat trap closed, the deviation was transmitted to the controlled thermal field. The energy input caused the final measured temperature parameter to rise to 354.2℃. In control group two, because the pulse width parameter was fixed as a constant, the energy supply could not be dynamically adjusted according to the time-varying material mass. The duration of its discrete-time pulse commands was too long, also causing localized overheating in the later stages of the reaction and pushing the final measured temperature parameter up to 331.6℃. In contrast, in the control flow of the present invention with a first preset sampling period of 0.5s, the real-time temperature rise slope parameter calculated by the microprocessor was 0.04℃ / s, 0.03℃ / s, and 0.03℃ / s respectively in three consecutive sampling periods. Both s and 0.02℃ / s are below the stagnation threshold of 0.2℃ / s, causing the control state to switch to asynchronous pulse control mode and suspend the output of conventional feedback adjustment commands. The physical mass parameters of the current batch of fly ash are extracted as 25.4kg, the characteristic value of volatile matter mass concentration is 0.12, the reference heat release power of the main heat source valve at maximum physical opening is 15000W, the system heat transfer efficiency conversion factor is 0.85, and the theoretical latent heat constant set for this volatile matter is 245000J / kg. The microprocessor calls the following formula to calculate the pulse width parameter: ,in, The calculated pulse width parameter, The physical quality parameters of the current batch of fly ash. The characteristic value of the extracted volatile matter mass concentration, The theoretical latent heat constant, As the reference heat release power, As the system heat transfer efficiency conversion factor, the above known values are substituted into the formula to calculate the pulse width parameter as 62.3s. Based on this, the microprocessor drives the controlled heat source flow regulation mechanism to start the physical action of full-load heat supplementation within a time range of 62.3s. When the heat supplementation time reaches 62.3s, a cutoff command is issued and the historical error integration register is cleared, so that the measured furnace temperature parameter is smoothly converged to 298.6℃.
[0037] As the intensity of the controlled disturbance in the experiment increased from medium to high concentration, the nonlinear latent heat phase change resistance within the fly ash material increased. This exacerbated the integral saturation phenomenon of the continuous feedback control loop within the control group, lengthening the reciprocating oscillation period of its lower-level valve to over 320 seconds, increasing the maximum temperature exceedance to 386.4℃, and increasing the ineffective energy loss ratio of the power supply medium to 42.6%. However, under this high-disturbance condition, the sample group of this invention, by establishing a causal relationship between the latent heat of physical phase change and pulse compensation within the control domain, and by inputting dynamic data showing that the physical mass parameter of the current batch of fly ash increased to 45.8 kg and the characteristic value of volatile matter concentration increased to 0.28, calculated the... The pulse width parameter is adaptively extended to 156.4s. The microprocessor calls the open-loop pulse control instruction containing the pulse width parameter to directly act on the lower-level actuator. This counteracts the energy dissipation of the material's endothermic phase change during continuous phase transitions, controlling the variance of the final output furnace temperature parameter to within 2.4℃. This confirms a quantifiable, deterministic, monotonic correlation between the technical effect and the intensity of the controlled operating condition disturbance. To verify the reasonable performance of the parameter boundaries defined in this invention, control group three set the stagnation threshold to exceed the upper limit of 2.5℃ / s and fall below the lower limit of 0.01℃ / s. When using a stagnation threshold exceeding the upper limit of 2.5℃ / s, due to the judgment working window... If the threshold is too wide, the microprocessor misjudges the transient temperature drop caused by conventional high-frequency electromagnetic noise disturbances as a latent heat trap state, leading to frequent misalignment of the asynchronous pulse control mode. This causes overload of the control system command and accelerates wear and tear on the valve actuator due to frequent switching. Furthermore, when a stagnation threshold of 0.01℃ / s (below the lower limit) is used, the control system becomes sluggish in responding to the actual endothermic phase change when the temperature rise slope approaches zero. This results in the inability to suspend the integral term in time at the critical point of the phase change, causing the system to revert to integral saturation under the continuous feedback control architecture. The data from the aforementioned out-of-range control test show that when the process judgment parameter deviates from the operating window defined by this invention, both the system's control stability and hardware lifespan are affected. A sharp deterioration occurs, thus confirming the rationality of the numerical limit of the stagnation threshold range with the fact of the physical inflection point. Based on the physical measurement data collected in multiple gradient experiments, the processing unit can identify the latent heat trap state of the apparent temperature by relying on the dual state judgment of the temperature rise slope and the opening command. Under the coordinated operation of discrete timing pulse command and continuous calculation loop forced suspension mechanism, the timing of the dynamic energy supply command is aligned with the phase change energy flow in the controlled pyrolysis space. The steady-state convergence time of the measured furnace temperature parameters is shortened, the valve reciprocating oscillation of the lower-level actuator is eliminated, and under the energy consumption constraint of multiple energy supply media, the entire fly ash low-temperature pyrolysis industrial control system maintains a stable and disturbance-resistant operating state.
[0038] Example 3: This example combines Figures 1 to 2 A description of the optimized control method and system for low-temperature pyrolysis of fly ash based on multimodal data, such as... Figure 1As shown, step S1: Read the real-time parameters of the controlled operating condition. The temperature sensing external element obtains the furnace temperature parameters and simultaneously reads the valve opening command value of the controlled heat source flow regulation mechanism. After completing step S1, the system proceeds to step S2: Calculate the temperature rise slope parameter. The processing unit completes the numerical integration based on the real-time furnace temperature parameter changes within the continuous sampling period. The system enters the decision branch to determine whether it is in a latent heat trap state, verifying that the opening command value is greater than the basic load threshold and the absolute value of the slope for three consecutive cycles is less than the stagnation threshold. If the above condition determination result is negative, the system proceeds to maintain the normal feedback regulation mode, controls the internal continuous feedback control loop to continue to accumulate and output the integral term, and returns to before step S1. If the above condition determination result is positive, the system proceeds to step S3: Suspend and reset the analog channel. The system forcibly suspends the output channel of the conventional feedback regulation command and terminates the accumulation of the integral term in the continuous closed loop to prevent saturation overshoot. After completing step S3, the system enters the asynchronous pulse control mode and generates discrete timing commands based on the fixed step size parameters pre-stored in the memory. The pulse opening command is directly input to the controlled heat source flow regulation mechanism. During this period, the system continuously determines whether to move out of the phase change stagnation section and exit the latent heat trap state, and detects that the absolute value of the real-time temperature rise slope parameter is greater than the stagnation threshold in the subsequent continuous sampling period. If the determination result is negative, the system cycles to the asynchronous pulse control mode. If the determination result is positive, the system flows to the recovery closed loop control process, performs the operation of terminating the asynchronous pulse control mode, clearing the historical error integral register, and restoring the conventional command output, and finally returns to the conventional feedback regulation mode.
[0039] like Figure 2As shown, the fly ash low-temperature pyrolysis optimization control system includes four control states: a safety baseline feedback loop, a conventional feedback adjustment mode, an asynchronous pulse control mode, and an automated emergency procedure interlocking logic. In the conventional feedback adjustment mode, the internal continuous feedback control loop accumulates integral terms. When the system is in the conventional feedback adjustment mode, if the real-time communication delay exceeds the safety time threshold, the system switches to the safety baseline feedback loop, disconnecting the external channel and enabling the local infrared temperature measurement parameter. If the system is determined to have entered a latent heat trap state in the conventional feedback adjustment mode, i.e., the real-time valve opening command value is greater than the basic load threshold and the absolute value of the real-time temperature rise slope parameter is greater than the basic load threshold within three consecutive sampling periods... If all values are less than the stagnation threshold, the system switches to asynchronous pulse control mode, suspends the conventional feedback adjustment command, and generates intermittent pulse opening commands. Subsequently, when it is determined in asynchronous pulse control mode that the system has moved out of the phase change stagnation section and exited the latent heat trap state, and the absolute value of the real-time temperature rise slope parameter within the continuous sampling period is greater than the stagnation threshold, the system returns to conventional feedback adjustment mode. If the flue gas monitoring data triggers the environmental protection threshold in conventional feedback adjustment mode, the system switches to the automated emergency procedure interlock logic, executes the operation of blocking the conventional feedback adjustment command and adjusting the opening of the auxiliary air volume valve drive mechanism until it stabilizes within the process safety range, and then returns to conventional feedback adjustment mode through the dotted line path.
[0040] Example 4: When the control system operates continuously for 720 hours in a low-temperature pyrolysis reactor environment with continuously injected fly ash material, the deposition of fly ash particles on the reactor wall surface forms a thermal resistance layer of fixed thickness, which physically increases the multidimensional resistance to heat conduction. This causes a time lag between the static heat conduction parameters stored in the memory and the controlled pyrolysis space where physical losses occur. Consequently, the calculated pulse width parameter cannot fully compensate for the latent heat energy dissipated by the phase transition of the material. To correct the deviation of the system heat conduction efficiency conversion coefficient caused by the thermal resistance layer, the processing unit corrects it through a sliding time window data damping loop.
[0041] Within each calibration cycle, the furnace temperature parameter sequence is read when the main heat source valve is at a fixed opening. The current wall thermal resistance variable is calculated by measuring the temperature rise slope of the external temperature sensing element over 200 consecutive sampling points. The wall thermal resistance variable is used as a dynamic negative feedback compensation quantity by utilizing the historical database stored in the memory to automatically adjust the value of the corresponding system heat transfer efficiency conversion factor in the mathematical formula. The updated formula is shown below: ,in, This is the dynamically updated conversion factor for the system's heat transfer efficiency. The initial heat transfer efficiency baseline value is dimensionless and is set to 0.85; It is the thermal resistance decay constant, and is stored in memory as 0.035; The cumulative operating hours of the control system for injecting fly ash material. The preset baseline value for the maintenance cycle of the controlled heat source is set at 1000 hours; the microprocessor will update the calculated system heat transfer efficiency conversion factor each time. The operation register is written into the memory in real time, so that the open-loop pulse control command of the subsequent calculation cycle can output the pulse width parameter based on the current physical transmission characteristics. When the cumulative running time reaches the maintenance cycle benchmark value, a dust cleaning prompt signal is generated and sent to the terminal actuator, thereby completing the dynamic adjustment and correction of the heat transfer attenuation fluctuation of the physical equipment.
[0042] Example 5: When the system faces the initial access condition of the controlled operating condition, the processing unit calls the pre-stored initialization control program in the memory before opening the control loop to determine the temperature rise response characteristics of the controlled pyrolysis space. The processing unit drives the main heat source flow regulating mechanism to adjust the valve opening, controlling the valve opening to increase from 10% to 80%. The temperature sensing external element collects the temperature measurement response data in the pyrolysis space. The microprocessor calculates the average slope value of the pyrolysis space temperature rise segment when the valve opening is 50%. 20% of this average slope value is written into the memory as the stagnation threshold. The opening command value required to maintain the pyrolysis space temperature at 250℃ is stored in the corresponding register unit as the basic load threshold, thus determining the basic control boundary of the controlled operating condition.
[0043] After storing the basic load threshold and stagnation threshold in the memory, to overcome the inherent uneven spatial composition distribution of heterogeneous solid waste and before the spectroscopic analyzer acquires optical data, a mechanically rotating physical disturbance fluidization component installed on the side wall of the main feed pipeline is activated. This component performs a periodic, cross-sectional, multi-point staggered mixing action on the freely falling solid fly ash particle flow, forcibly homogenizing local concentration differences to form a surface of material with a uniform apparent composition. To ensure that the material flow subjected to strong mechanical disturbance meets the measurement conditions of subsequent precision instruments, an additional layer is added to the inner cavity of the pipeline below the fluidization component. The system features a multi-layered, stepped mechanical flow-restricting baffle structure. High-speed, freely falling fly ash particles, after being homogenized and mixed, impact these baffles, causing a rapid dissipation of their physical kinetic energy and the formation of a localized, slow-flowing stagnation zone above the baffles. This forces the overall falling motion to slow down, converting it into a stationary material surface that meets the stringent requirements of the microsecond-level excitation and detection window of the spectrometer. This avoids spectral blurring and signal distortion caused by the high transient displacement of particles capturing the optical focus. The processing unit reads the raw elemental mass concentration parameters obtained from the spectrometer to determine the chlorine mass percentage. Percentage of sulfur by mass The characteristic value of volatile matter mass concentration is calculated based on the pre-stored functional relationship in the memory. The calculation formula is as follows: ,in, The characteristic value of volatile matter mass concentration. This represents the mass percentage of chlorine. The formula represents the mass percentage of sulfur. The weighting coefficients 0.01 and 0.015 in this mathematical formula are derived from the standardized latent heat of vaporization ratio of specific inorganic salts in fly ash at low-temperature pyrolysis frequencies, determined through engineering calibration. Specifically, 0.01 corresponds to the normalized energy conversion coefficient of the standard molar enthalpy change during the decomposition of chloride salts, represented by sodium chloride, and 0.015 corresponds to the relative endothermic reference weight of the gaseous molecules released during the pyrolysis desulfurization of sulfates, represented by calcium sulfate. By invoking these two fixed constants, the system mathematically maps the mass percentage of isomeric elements into a dimensionless comprehensive physical quantity representing the endothermic intensity per unit mass of the material. It is 8.5% and When the concentration is 2.1%, the microprocessor calculates the characteristic value of the volatile matter mass concentration. The value is 0.1165. The processing unit switches between a proportional-integral-derivative continuous calculation loop and an asynchronous pulse control mode. The industrial control system uses the characteristic value of volatile matter mass concentration. Calculate control parameters and adjust the main heat source flow regulation mechanism to stabilize the furnace temperature parameters.
[0044] Example 6: When the system faces the condition of multiple low-temperature pyrolysis reactors deployed in parallel clusters and the total injected material flow rate fluctuates by more than 50% within the sampling period, the main control unit periodically obtains the real-time temperature deviation variation value collected by the temperature measuring external element corresponding to each low-temperature pyrolysis reactor through the network communication interface, and inputs the real-time heat load of each low-temperature pyrolysis reactor as characteristic data into the pre-stored power distribution attenuation matrix in the memory. Based on the power distribution attenuation matrix, the dynamic feed rate correction coefficient of each low-temperature pyrolysis reactor is calculated, and the feedforward adjustment command of the feed frequency conversion speed regulation actuator of a single low-temperature pyrolysis reactor is corrected, so as to control the furnace temperature parameter deviation of each low-temperature pyrolysis reactor in the parallel cluster to converge under transient material impact.
[0045] The main control unit, in response to transient changes caused by localized material accumulation within a single low-temperature pyrolysis reactor, collects real-time furnace temperature parameters that deviate from the process target value, and controls the microprocessor to calculate and adaptively adjust the step size factor according to the following formula: ,in, To adaptively adjust the step size factor, The feedback sensitivity coefficient, which is pre-stored in the memory, is set to 0.015. The microprocessor will calculate the generated adaptive adjustment step size factor as the absolute value of the temperature tracking error. Multiplying the basic control step size, the adjustment gain is calculated and directly written into the physical register of the analog output module. This reduces the action time constant of the controlled heat source flow regulation mechanism from 15.2s to 11.3s after receiving the real-time valve opening command value. It also causes the response dead zone of the internal thermal field of the low-temperature pyrolysis reactor to converge, allowing the real-time furnace temperature parameters fed back from the external temperature sensing element to return to the steady-state control range.
[0046] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application 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 this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.
Claims
1. A method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data, characterized in that, Includes the following steps: Step S1: The processing unit obtains the real-time furnace temperature parameters under controlled conditions through the temperature sensing external element, and simultaneously reads the real-time valve opening command value sent to the controlled heat source flow regulation mechanism. Step S2: The processing unit calculates the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period. When the real-time valve opening command value is greater than the basic load threshold stored in the memory connected to the processing unit, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the memory in three consecutive sampling periods, the processing unit determines that the controlled operating condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of the internal material phase change. Step S3: During the transient control cycle in which the latent heat trap state is determined to be established, the processing unit forcibly suspends the analog output channel of the currently calculated conventional feedback adjustment command, terminates the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switches to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the memory, it generates a discontinuous pulse opening command with discrete timing characteristics and directly uses the pulse opening command as the control law input to act on the controlled heat source flow regulation mechanism until the absolute value of the real-time temperature rise slope parameter in the subsequent continuous sampling cycle is greater than the stagnation threshold. Then, it is determined that the controlled operating condition has moved out of the phase change stagnation section and exited the latent heat trap state.
2. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, In step S1, flue gas monitoring data is acquired; when it triggers the environmental protection threshold, the conventional feedback adjustment command is blocked and the opening of the auxiliary air volume valve drive mechanism is adjusted to keep the controlled operating conditions stable within the process safety range.
3. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, Step S1 includes: calculating the real-time communication delay value using timestamps within the communication channel, and disconnecting the external channel and switching to the safety backup feedback loop when the delay exceeds the safe time threshold.
4. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, Step S3 includes: calling the dead zone compensation operator to calculate the reverse compensation amount before outputting the command and superimposing it into the conventional feedback regulation command to generate an actual control signal and send it to the controlled heat source flow regulation mechanism.
5. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, Step S3 includes: during the high-level segment, accumulating the current valve reference opening by a fixed step size to generate a pulse opening command, and during the low-level segment, resetting the pulse opening command to the current valve reference opening.
6. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, Step S2 includes: recording the cumulative duration of latent heat trap states to generate a condition deviation index, and outputting a warning signal and adjusting the subsequent base load threshold when the deviation exceeds the abnormal alarm threshold.
7. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, The basic load threshold stored in the memory is 60% to 75% of the maximum opening of the controlled heat source flow regulation mechanism, the stagnation threshold stored in the memory is 0.1℃ to 0.3℃, and the continuous sampling period is 5s to 10s.
8. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 1, characterized in that, Step S1 includes: performing median filtering on the original measurement signal to generate real-time furnace temperature parameters, and performing full-range normalization transformation on the real-time valve opening command value.
9. The method for optimizing and controlling the low-temperature pyrolysis of fly ash based on multimodal data according to claim 5, characterized in that, Step S3 includes: when the absolute value of the real-time temperature rise slope parameter is greater than the stagnation threshold for two consecutive sampling cycles, the phase transition section is determined to end, and the output of the normal feedback adjustment command is restored.
10. A fly ash low-temperature pyrolysis optimization control system based on multimodal data, used to implement the fly ash low-temperature pyrolysis optimization control method based on multimodal data as described in claim 1, characterized in that, It includes a temperature sensing module, a controlled heat source flow regulation module, a data storage module, and a logic processing module. The logic processing module is communicatively connected to the temperature sensing module, the controlled heat source flow regulation module, and the data storage module, respectively. The logic processing module is configured to obtain the real-time furnace temperature parameters under controlled conditions through the temperature sensing module and simultaneously read the real-time valve opening command value of the controlled heat source flow regulation module. The logic processing module is also configured to calculate the real-time temperature rise slope parameter based on the change of the real-time furnace temperature parameter in the continuous sampling period, and to determine that the controlled condition is in a latent heat trap state with apparent temperature stagnation due to the heat absorption of internal material phase change when the real-time valve opening command value is greater than the basic load threshold delay range stored in the data storage module, and the absolute value of the real-time temperature rise slope parameter is less than the stagnation threshold stored in the data storage module in three consecutive sampling periods. The logic processing module is also configured to forcibly suspend the analog output channel of the currently calculated conventional feedback adjustment command in the latent heat trap state, terminate the accumulation of the integral term of the internal continuous feedback control loop to prevent saturation overshoot, and automatically switch to asynchronous pulse control mode. Based on the fixed step size parameters pre-stored in the data storage module, it generates intermittent pulse opening commands with discrete timing characteristics, and uses the pulse opening commands as the control law input to the controlled heat source flow regulation module until the absolute value of the real-time temperature rise slope parameter in subsequent continuous sampling cycles is greater than the stagnation threshold. At this point, it is determined that the controlled condition has moved out of the phase change stagnation section and exited the latent heat trap state.