Coal-based solid waste fiber forming process technological parameter self-adaptive control system and control method

Abstract

The invention relates to a coal-based solid waste fiber forming process technological parameter self-adaptive control method and system, and the method comprises the following steps: collecting a raw material component electric signal, a process parameter electric signal and a product performance electric signal in real time through a distributed detection module, and transmitting the electric signals to an electric control unit after standardization processing; constructing an operation model based on a heat-force-flow coupling mechanism, substituting raw material components and process parameter electric signals, calculating a process parameter initial optimization interval, and outputting the process parameter initial optimization interval to a machine learning optimization layer; the machine learning optimization layer calculates the initial interval through a preset hybrid algorithm, generates an optimal process parameter instruction and transmits the optimal process parameter instruction to the adaptive control layer; and the adaptive control layer converts the optimal parameter instruction into an electric control signal which can be identified by the execution mechanism, and issues the electric control signal to each execution mechanism through a standardized bus to drive parameter adjustment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal-based solid waste into fiber, in particular, especially relates to a coal-based solid waste into fiber process parameter adaptive control system and control method. BACKGROUND

[0002] The annual output of industrial solid waste in China exceeds 2 billion tons, of which coal-based solid waste (power plant slag, fly ash, etc.) accounts for more than 50%. Although the existing coal-based solid waste into fiber technology has realized solid waste resource utilization (such as CN118324405B, a method for producing aluminum silicate fiber cotton by synergistic resource utilization of solid waste, CN109809700A, production of inorganic fiber from aluminum ash), there are still the following technical bottlenecks: First, the raw material adaptability is poor, the source of coal-based solid waste is complex, and the composition fluctuates greatly. The existing technology adopts fixed process parameters, which cannot be dynamically adapted, resulting in fiber forming defects. Second, it relies on manual experience or simple logic debugging. When the melt temperature, centrifugal speed and other parameters deviate from the optimal interval, the response is slow, resulting in waste of energy. Third, although the existing technology can control a single parameter, it cannot cooperatively control multiple parameters, resulting in uneven fiber diameter, which in turn causes large fluctuations in mechanical strength during subsequent use, making it difficult to meet the needs of high-end scenarios such as building insulation and composite reinforcement. SUMMARY

[0003] According to the above technical problems, a coal-based solid waste into fiber process parameter adaptive control system and control method are provided.

[0004] The technical means adopted by the present application are as follows: A coal-based solid waste into fiber process parameter adaptive control method, comprising the following steps: S1: Real-time acquisition of raw material composition electrical signal, process parameter electrical signal and product performance electrical signal by distributed detection module, and transmission to electric control unit after standardization processing; S2: Based on the heat-power-flow coupling mechanism, an operation model is constructed, the raw material composition and process parameter electrical signals are substituted, the initial optimization interval of process parameters is calculated, and output to the machine learning optimization layer; S3: The machine learning optimization layer generates optimal process parameter instructions by presetting a hybrid algorithm to operate on the initial interval and transmits them to the adaptive control layer; S4: The adaptive control layer converts the optimal parameter instructions into electric control signals recognizable by the actuator, and issues them to each actuator through the standardized bus to drive parameter adjustment; S5: The detection module real-time acquires the process parameter and product performance electrical signals after regulation and control, and feeds back to the correction layer and machine learning optimization layer to dynamically correct the model weight and process parameters, realizing closed-loop control; S6: when the raw material component electric signal fluctuates beyond the preset threshold, triggering the emergency instruction, quickly adjusting the actuator parameter, synchronously accelerating the model iteration, and outputting the optimal parameter adapted to the new raw material; In step S1, the raw material component electric signal is generated by an X-ray fluorescence detector, the process parameter electric signal includes temperature, viscosity, and pressure parameters, and the product performance electric signal is generated by a laser particle size analyzer.

[0005] Further, in step S2, the operation model is constructed based on a coupling mechanism equation and realized through a Python environment, including a melt viscosity equation and a fiber diameter equation.

[0006] Further, in step S3, the hybrid algorithm is a combination of random forest and LSTM algorithm, which is trained based on at least 5000 sets of training data, and the optimal parameter is output in XML format, including parameter target value, adjustment step, and actuator address information.

[0007] Further, in step S4, the electric control signal includes 0-10V analog signal, 4-20mA analog signal, and digital switching signal.

[0008] Further, the process parameters include electric melting furnace power, centrifugal speed, and air flow pressure, and the corresponding electric control signal and parameter are linearly related, and the control precision is ±1%.

[0009] The application also discloses a coal-based solid waste fiber forming process parameter self-adaptive control system adapted to the above method, which comprises a detection unit, a transmission unit, an electric control unit and an execution unit, and each unit cooperates through a standardized interface, and specifically comprises the following components: The detection unit is composed of an X-ray fluorescence component detector, a thermocouple, a melt viscometer and a laser particle size analyzer, and is used for converting physical parameters into standardized electric signals; The transmission unit is equipped with an industrial switch and supports signal hot standby transmission; The electric control unit is installed in the control cabinet in the control room; The execution unit includes an electric melting furnace transformer, a centrifuge frequency converter, an air flow fan pressure valve and a batching bin electric valve, and receives the electric control signal to execute parameter adjustment and feedback operation state electric signal.

[0010] Compared with the prior art, the application has the following advantages: through high-speed transmission of electric signals and fast response of the execution mechanism, the application solves the problem of parameter control hysteresis, and the raw material fluctuation is responded more timely, the whole process does not need manual intervention, the system automatically completes collection, operation, control and feedback, the manual intervention time is less, through accurate control, the energy consumption of the electric melting furnace is reduced by 12%-15%, the unit product power consumption is reduced by 8-10g / kWh, and the energy waste problem of traditional technology is solved. DETAILED DESCRIPTION

[0011] It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of this invention can be combined with each other. The technical solutions in the embodiments of this invention are clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0012] This invention discloses an adaptive control method for process parameters in a coal-based solid waste fiber forming process, comprising the following steps: S1: Real-time acquisition of raw material composition electrical signals, process parameter electrical signals, and product performance electrical signals through distributed detection modules, followed by standardized processing and transmission to the electrical control unit; S2: Based on the thermo-mechanical-fluid coupling mechanism, a computational model is constructed, and the raw material composition and process parameter electrical signals are substituted to calculate the initial optimization range of process parameters and output to the machine learning optimization layer; S3: The machine learning optimization layer uses a preset hybrid algorithm to calculate the initial interval, generate the optimal process parameter instructions, and transmit them to the adaptive control layer; S4: The adaptive control layer converts the optimal parameter instructions into electronic control signals that the actuators can recognize, and sends them to each actuator through a standardized bus to drive parameter adjustment; S5: The detection module collects process parameters and product performance electrical signals after regulation in real time, and feeds them back to the correction layer and machine learning optimization layer to dynamically correct model weights and process parameters to achieve closed-loop control. S6: When the fluctuation of the electrical signal of the raw material component exceeds the preset threshold, an emergency command is triggered to quickly adjust the parameters of the actuator, synchronously accelerate the model iteration, and output the optimal parameters adapted to the new raw material; the preset threshold is when the content of any major component (SiO2, Al2O3, CaO, Fe2O3) fluctuates by more than ±5% compared with the average value of the previous 10 minutes.

[0013] In step S1, the electrical signals of the raw material components are generated by an X-ray fluorescence detector, the electrical signals of the process parameters include temperature, viscosity, and pressure parameters, and the electrical signals of the product performance are generated by a laser particle size analyzer.

[0014] Furthermore, the computational model described in step S2 is constructed based on the coupling mechanism equation and is implemented through the Python environment, including the melt viscosity equation and the fiber diameter equation.

[0015] Furthermore, the hybrid algorithm described in step S3 is a combination of random forest and LSTM, which is generated based on at least 5000 sets of training data. The optimal parameters are output in XML format, including the target parameter values, adjustment step size and actuator address information.

[0016] Furthermore, the electronic control signals mentioned in step S4 include 0-10V analog signals, 4-20mA analog signals, and digital switch signals.

[0017] Furthermore, the process parameters include the power of the electric furnace, the centrifugal speed, and the airflow pressure. The corresponding electrical control signals are linearly related to the parameters, with an adjustment accuracy of ±1%.

[0018] This invention also discloses an adaptive control system for process parameters of coal-based solid waste fiber forming process adapted to the above method, including a detection unit, a transmission unit, an electrical control unit, and an execution unit. Each unit works collaboratively through a standardized interface, and its specific composition is as follows: The detection unit consists of an X-ray fluorescence spectrometer, thermocouples, a melt viscometer, and a laser particle size analyzer, and is used to convert physical parameters into standardized electrical signals. Transmission unit: Equipped with an industrial switch, supporting hot-standby signal transmission; Electrical control unit: Installed in the control cabinet in the control room; includes PLC controller, data acquisition card and industrial computer, with a built-in five-layer software architecture, specifically including data acquisition layer, model layer, machine learning layer, adaptive control layer and feedback correction layer, responsible for electrical signal calculation, control command generation and model iterative optimization, and supports interaction with TensorFlow framework and MySQL database.

[0019] Execution unit: includes electric furnace transformer, centrifuge frequency converter, airflow fan pressure valve, and batching silo electric valve. After receiving electrical control signals, it performs parameter adjustments and feeds back operating status electrical signals.

[0020] In this embodiment, the coal-based solid waste fiber conversion mentioned is a systematic process that transforms coal-based solid waste such as power plant slag and fly ash into high-quality mineral wool fibers. The core process consists of four main stages: raw material pretreatment, melt fiber conversion, fiber collection and forming, and finished product testing. In the raw material pretreatment stage, solid waste is stored separately, including a cold slag storage and a hot slag temporary storage. The cold slag has a moisture content of ≤10%, and the hot slag has a temperature of ≥800℃. The second step in the pretreatment stage is crushing and screening. The cold slag and hot slag are crushed into raw materials with a particle size of ≤50mm using a jaw crusher. The undersize material accounts for ≥95%, and the oversize material is returned to the crusher. After that, the components are blended and batched. The batching step can use existing technology. After the raw material pretreatment is completed, the process enters the melt fiber forming stage. This stage involves the sequential processes of raw material conveying, electric furnace melting, melt conveying, centrifugal fiber forming, and airflow drawing. During the raw material conveying process, cold slag is transferred to the cold material bin by a loader, and then conveyed to the buffer bin by a weighing belt and bucket elevator. Hot slag is introduced into the ladle through a diversion trough and then transferred to the electric furnace by a forklift and elevator. It is necessary to control the cold slag conveying speed to match the electric furnace feeding rate, maintaining it at 10 to 15 tons per hour. The hot slag transfer time should not exceed 30 minutes to prevent the hot slag from agglomerating due to heat dissipation. After the raw materials are transported to the electric melting furnace, they enter the melting stage. The electric melting furnace is equipped with a 2000KVA electric furnace transformer with a power adjustment range of 1200-2000 kilowatts. It uses graphite electrodes with a diameter of 500 mm for heating, and the melting temperature is controlled at 1450-1600 degrees Celsius to ensure that the melt viscosity is maintained in the optimal range of 800-1500 Pa·s. The melt stays in the furnace for no less than 30 minutes to ensure uniform composition and avoid temperature fluctuations that could lead to poor melt flowability or excessive electrode wear.

[0021] The molten material is transported to the centrifugal fiber forming stage through a flow channel and a chute. The flow channel needs to be equipped with a natural gas heating device to maintain the temperature at 1500-1550 degrees Celsius with a fluctuation of no more than 50 degrees Celsius. The chute has a preset inclination angle and the melt flow rate is stable at 460-780 kg per hour to prevent flow channel blockage from causing flow interruption or flow fluctuations that affect the fiber forming quality.

[0022] After the melt is conveyed to a four-roll centrifuge, it enters the centrifugal fiber-forming stage. The centrifuge speed is controlled at 1500-2500 rpm, and the roller surface temperature is maintained at 1200-1300 degrees Celsius to prevent the melt from sticking to the rollers. Centrifugal force spins the melt into primary fibers, ensuring that the fiber diameter is adjusted reasonably with the speed to prevent uneven fiber diameter caused by speed fluctuations. The primary fibers after centrifugal fiber-forming enter the airflow drawing stage. The airflow speed is controlled at 30-50 meters per second, and the channel temperature is maintained at 800-1000 degrees Celsius to prevent the fibers from solidifying prematurely. Airflow drawing extends the fiber length to 10-30 centimeters, ensuring the integrity of the fiber morphology and preventing fiber breakage due to insufficient air pressure or dust accumulation in the channel, which would affect the drawing effect.

[0023] After being melted and fiberized, the fibers enter the fiber collection and forming stage, which includes cotton collection, pendulum forming, curing and drying, and cutting and packaging. The cotton collection stage uses a negative pressure cotton collector. The collected fibers then enter the pendulum forming stage, where they are processed by a pendulum machine and forming conveyor belt. The pendulum machine's oscillation frequency is controlled at 10-15 times per minute, and the fibers are arranged in 5-8 layers to ensure tight interlayer bonding. After pendulum forming, the fibers enter the curing and drying stage using a curing oven at a temperature of 200℃±20℃. Alternatively, a segmented heating method can be used, sequentially performing pre-drying at 150℃, curing at 250℃, and cooling at 180℃, with a curing time controlled at 30-60 minutes to ensure complete adhesive curing. The cured and dried finished products then enter the cutting and packaging stage.

[0024] After the fiber is formed, it enters the finished product testing stage.

[0025] In this embodiment, in the actual production process, the electrical signal acquisition of S1 runs through the key nodes from raw material pretreatment to fiber forming, ensuring that the physical parameters of each production link are converted into controllable electrical signals in real time. Specifically, X-ray fluorescence detectors are deployed at the cold slag silo outlet conveyor and the hot slag temporary silo discharge chute to collect electrical signals of components such as silica and alumina in the cold and hot slag in real time. Thermocouples are installed in the upper, middle, and lower areas of the electric melting furnace cavity to collect melt temperature electrical signals. A melt viscometer is deployed in the melt flow channel at the bottom of the electric melting furnace to collect melt viscosity electrical signals. A pressure sensor is installed at the end of the flow channel to collect melt conveying pressure electrical signals. A laser particle size analyzer is installed at the end of the centrifuge outlet drawing channel to collect product performance electrical signals such as fiber diameter and slag ball content. A negative pressure sensor is deployed on the top of the cotton collecting machine housing to collect negative pressure electrical signals in the cotton collecting chamber, thereby monitoring the fiber forming quality and collection effect in real time.

[0026] In the component blending step of raw material pretreatment, the computational model, based on the thermo-mechanical-fluid coupling mechanism, substitutes the electrical signals of the raw material components collected by the X-ray fluorescence detector (in this embodiment, 45% silica and 18% alumina can be selected), and runs the melt viscosity equation in the Python environment to calculate the initial temperature range of the electric melting furnace suitable for this raw material composition (in this embodiment, it is 1500-1550℃) and the proportion of bauxite slag added in the batching bin. This provides parameter basis for component blending and batching, avoiding the experience error of manual batching; wherein, the formula is: η=k1T -3 exp(k2 / C), where k1 and k2 are fitting constants, specifically model constants obtained by fitting experimental data, η is the melt viscosity (Pa·s), T is the melt temperature, and C is the composition, specifically the mass percentage of CaO.

[0027] Before the melting step in the electric melting furnace for fiber formation, the model is synchronously fed with historical process parameter electrical signals (in this embodiment, the optimal viscosity of raw materials with similar composition is 800-1500 Pa). The initial speed range of the centrifuge and the initial air pressure of the airflow fan are calculated to ensure that the melting and fiber-forming parameters are initially adapted to the raw material characteristics, reducing the parameter debugging time during the production start-up phase. The calculated initial optimization range of process parameters is output to the machine learning optimization layer in CSV format.

[0028] The machine learning optimization layer uses a combination of random forest and LSTM algorithms. Based on 5000 sets of historical production data covering the correspondence between different raw material compositions, process parameters and product performance, it calculates the initial temperature range (1500-1550℃) and speed range (2000-2200r / min) output by S2, and outputs precise parameter instructions. Finally, it generates precise parameter control data for the electric furnace temperature of 1520℃, the centrifuge speed of 2100r / min, and the airflow pressure of 0.5MPa. The instructions are stored in XML format, including the parameter target value, the adjustment step size (temperature adjustment step size 5℃, speed adjustment step size 50r / min), and the addresses of the actuators including the electric furnace transformer and the centrifuge frequency converter, ensuring that the parameters can be accurately sent to the corresponding equipment. The algorithm combines the real-time signals from the cotton collector negative pressure sensor and the laser particle size analyzer to optimize the conveyor belt speed and the pendulum swing frequency, avoiding cotton layer thickness deviation or density unevenness.

[0029] The combination of random forest and LSTM adopts a cascaded structure: first, the random forest model makes preliminary predictions of the process parameter range based on the raw material composition and operating conditions; then, the LSTM model receives the time-series process parameters within this range, performs dynamic optimization, and outputs the final parameter instructions. In this embodiment, the LSTM network structure has two layers, with 64 neurons in each layer, using the Adam optimizer, and the loss function is the mean squared error.

[0030] The adaptive control layer converts the 1520℃ electric furnace temperature command into an analog signal, which is then transmitted to the electric furnace transformer via the bus. This drives the transformer to adjust the voltage, ensuring that the melt viscosity remains stable between 800-1500 Pa. The centrifuge speed command is converted into an analog signal and sent to the centrifuge frequency converter to drive the centrifuge roller speed adjustment, preventing fiber diameter dispersion caused by speed fluctuations. The batching ratio command is converted into a digital switch signal and transmitted to the electric valve of the batching hopper to control the valve opening, ensuring the silica to alumina ratio is maintained between 2.0 and 3.5. The cotton collection negative pressure command is converted into an analog signal and transmitted to the negative pressure controller to drive the cotton collector fan pressure adjustment. The response time of all electrical control signals is ≤5 seconds, synchronized with the continuous operation rhythm of the production process.

[0031] The feedback mechanism runs through the entire production process, correcting parameter deviations in real time and ensuring stable product quality: thermocouples collect electrical signals after the temperature of the electric melting furnace is adjusted in real time and feed them back to the correction layer; if the signal shows that the temperature exceeds the preset range, the correction layer generates a correction command to reduce the transformer power and prevent the melt from overheating; similarly, the laser particle size analyzer collects electrical signals after the fiber diameter is adjusted and feeds them back to the machine learning optimization layer, etc., to ensure that the finished product meets the standards.

[0032] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive control method for process parameters in a coal-based solid waste fiber forming process, characterized in that, Includes the following steps: S1: Real-time acquisition of raw material composition electrical signals, process parameter electrical signals, and product performance electrical signals through distributed detection modules, followed by standardized processing and transmission to the electrical control unit; S2: Based on the thermo-mechanical-fluid coupling mechanism, a computational model is constructed, and the raw material composition and process parameter electrical signals are substituted to calculate the initial optimization range of process parameters and output to the machine learning optimization layer. S3: The machine learning optimization layer uses a preset hybrid algorithm to calculate the initial interval, generate the optimal process parameter instructions, and transmit them to the adaptive control layer; S4: The adaptive control layer converts the optimal parameter instructions into electronic control signals that the actuators can recognize, and sends them to each actuator through a standardized bus to drive parameter adjustment; S5: The detection module collects process parameters and product performance electrical signals after regulation in real time, and feeds them back to the correction layer and machine learning optimization layer to dynamically correct model weights and process parameters to achieve closed-loop control. S6: When the fluctuation of the electrical signal of the raw material composition exceeds the preset threshold, an emergency command is triggered to quickly adjust the parameters of the actuator, synchronously accelerate the model iteration, and output the optimal parameters adapted to the new raw material. In step S1, the electrical signals of the raw material components are generated by an X-ray fluorescence detector, the electrical signals of the process parameters include temperature, viscosity, and pressure parameters, and the electrical signals of the product performance are generated by a laser particle size analyzer.

2. The method according to claim 1, characterized in that, The computational model described in step S2 is constructed based on the coupling mechanism equation and is implemented through the Python environment, including the melt viscosity equation and the fiber diameter equation.

3. The method according to claim 1, characterized in that, The hybrid algorithm described in step S3 is a combination of random forest and LSTM, which is generated based on at least 5000 sets of training data. The optimal parameters are output in XML format, including the target parameter values, adjustment step size and actuator address information.

4. The method according to claim 1, characterized in that, The electrical control signals mentioned in step S4 include 0-10V analog signals, 4-20mA analog signals, and digital switch signals.

5. The method according to claim 1, characterized in that, The process parameters include electric furnace power, centrifugal speed, and airflow pressure. The corresponding electrical control signals are linearly related to the parameters, with an adjustment accuracy of ±1%.

6. An adaptive control system for process parameters of a coal-based solid waste fiber forming process adapted to the method of any one of claims 1 to 5, characterized in that, It includes a detection unit, a transmission unit, an electronic control unit, and an execution unit. Each unit works collaboratively through a standardized interface. The specific composition is as follows: The detection unit consists of an X-ray fluorescence spectrometer, thermocouples, a melt viscometer, and a laser particle size analyzer, and is used to convert physical parameters into standardized electrical signals. Transmission unit: Equipped with an industrial switch, supporting hot-standby signal transmission; Electrical control unit: installed inside the control cabinet in the control room; Execution unit: includes electric furnace transformer, centrifuge frequency converter, airflow fan pressure valve, and batching silo electric valve. After receiving electrical control signals, it performs parameter adjustments and feeds back operating status electrical signals.