Four-cart efficient control method for coke tank truck APS optimization

Abstract

The invention relates to the technical field of coke oven mechanical automation control, and discloses a four-cart efficient control method for coke tank truck APS optimization. The system collects a vehicle position, an equipment alignment signal and a step sequence reaching mark in real time; verifying four-vehicle linkage conditions and signal integrity based on fuzzy logic; door picking, coke pushing, door cleaning and frame cleaning, walking, coal charging and door closing process actions are sequentially controlled; adjusting the walking trigger logic according to the process parameters; and switching to a standby control mode when the control mode is abnormal. The system constructs a full-automatic closed-loop control framework from state perception, condition verification to process execution, realizes whole-course unmanned operation of cooperative operation of four carts, and solves the problems of discontinuous automatic process, extensive control granularity, insufficient system flexibility and the like of a traditional control method. And the production automation level, the safety and the operation efficiency are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of automated control technology for coke oven machinery, and in particular to a high-efficiency control method for four types of coke tank cars based on APS optimization. Background Technology

[0002] Current automated control technology for the four main coke oven cars typically centers on the Advanced Power System (APS) of the coke car, achieving vehicle movement and precise positioning by setting reference points for each furnace number. This system utilizes ready signals and positioning status transmitted between vehicles to construct basic interlocking logic; for example, it requires both the coke quenching car and the coke pushing car to be aligned before allowing coke pushing operations. This method focuses its automation on the movement and alignment stages, aiming to improve equipment movement efficiency and reduce human error, shortening the single-furnace operation cycle by optimizing coordination. However, its control scope is mainly limited to position coordination and basic interlocks. For the specific process execution stages after positioning, it still relies heavily on operator intervention, representing a high-efficiency operation mode centered on position management.

[0003] The aforementioned and existing related technologies often suffer from the following drawbacks: First, the automated process is not seamless; the system only enables vehicle positioning, and subsequent process steps still require manual operation, making it impossible to achieve fully unmanned production. Second, the control granularity is coarse, lacking fine verification of each process action, resulting in insufficient safety and reliability. Third, the system lacks flexibility, unable to dynamically adjust the process flow according to production needs, which restricts further improvement in equipment utilization and production efficiency. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing technology has the disadvantages of discontinuous automation process, coarse control granularity and poor system flexibility. To this end, we propose four high-efficiency control methods for coke tanker cars based on APS optimization.

[0005] To achieve the above objectives, this application adopts the following technical solution: a four-vehicle high-efficiency control method for coke tanker trucks with optimized APS, comprising the following steps: S1 receives the planned furnace number sequence and the door and frame clearing process selection parameters, establishes a control starting point based on the planned furnace number, initializes the vehicle controller and status monitor, and verifies the consistency between the initial position of the four main vehicles and the planned furnace number.

[0006] S2 collects the current furnace number data of the SCP integrated car, coke quenching car, and smoke guiding car through the vehicle bus network, obtains the equipment alignment signal and semi-automatic step sequence achievement mark, and monitors the status of the coke tank car and the interlocking conditions of the four cars.

[0007] S3 performs multiple preset condition checks based on the collected real-time status data. The conditions include the integrity of the equipment alignment signal and the semi-automatic step sequence achievement mark, the four-car interlocking condition, and the permission of the door and frame clearing process selection parameters. If all conditions are met, the vehicle controller will sequentially execute the processes of door removal, coke pushing, door and frame clearing, traveling, coal loading, and door closing.

[0008] S4 monitors the timing of process action completion signals during execution, dynamically adjusts the travel triggering logic based on the door and frame clearing process selection parameters, activates pre-alarm and delay mechanisms to coordinate step sequence switching, and updates the execution sequence of vehicle control commands.

[0009] S5: When an abnormal equipment alignment signal, an invalid semi-automatic step sequence achievement flag, a missing four-vehicle interlock condition, or an incorrect status of the coke tanker is detected, the current control process is interrupted, the system switches to standby control mode, and the interruption event data is recorded.

[0010] Preferably, step S3 includes the following steps: S31, verify the matching status between the current furnace number and the planned furnace number, compare the real-time collected data of the current furnace number of the vehicle with the planned furnace number for a second time, and generate a verification result of the furnace number consistency. S32, verify the integrity of the four-vehicle interlocking conditions and the equipment alignment signal, confirm that the interlocking conditions, alignment signals and semi-automatic step sequence achievement flags are all in a valid state, and generate a parallel permission signal; S33, confirm the execution conditions of the door and frame cleaning process selection parameters, and activate or skip the corresponding door and frame cleaning control sub-processes according to the received process parameters; S34 executes the door removal and coke pushing processes through the vehicle controller. When the parallel permission signal is valid, it sends door removal and coke pushing drive commands to the SCP integrated car and the coke blocking car. S35 sequentially executes the processes of clearing the door and frame, traveling, loading coal, and closing the door. Based on the process selection parameters and action completion signals, it triggers a series of subsequent automated operations according to the preset timing sequence.

[0011] Furthermore, step S32 specifically involves performing the following operations: A multi-source information verification system based on fuzzy logic is constructed. It collects signal strength sequences of four-vehicle interlocking conditions, deviation measurements of equipment alignment sensors, binary states of semi-automatic step sequence achievement markers, and environmental monitoring data (temperature, humidity, and electromagnetic interference) in real time via a high-speed data bus. A complete set of fuzzy membership functions is established, where each interlocking signal corresponds to a Gaussian membership function to characterize its signal quality confidence level, and each interference factor corresponds to an S-shaped membership function to quantify its destructive impact. Fuzzy logic aggregation operators are used to combine all these membership values ​​into a holistic verification index. The calculation of this index involves geometric mean of the signal membership and exponential decay adjustment of the interference impact index to balance signal reliability and environmental robustness. The system compares the verification index with a dynamically adjusted threshold. If it exceeds the threshold, a parallel permission signal is immediately generated and transmitted to the subsequent controller; if it falls below the threshold, a multi-verification loop is automatically triggered. Time series analysis is used to identify instantaneous anomalies and update the membership parameters. The verification index formula is as follows: , in, This indicates the total number of interlock signals and equipment alignment signals, including four-vehicle interlock condition signals, equipment alignment signals, and semi-automatic sequence achievement flags. This indicates the number of environmental disturbances, such as temperature fluctuations, electromagnetic noise, and mechanical vibration. This represents the real-time measured value of the i-th signal, which originates from the output of a sensor or control system. This represents the real-time measured value of the j-th interfering factor, obtained through environmental monitoring equipment. The fuzzy membership function of the i-th signal is defined as follows: ,in It is the ideal value of the signal. It is the allowable deviation range of the signal. The fuzzy influence function of the j-th interfering factor number, ,in It is the interference threshold. It is the interference sensitivity coefficient. surface It displays a global interference attenuation factor, which dynamically adjusts the degree of influence of environmental interference on the system verification results based on historical environmental data.

[0012] Furthermore, step S34 specifically involves performing the following operations: An adaptive control model based on Lyapunov stability is constructed. By real-time acquisition of displacement sensor data of the gate-removing mechanism, measured values ​​of the push rod torque, equipment vibration spectrum characteristics, and temperature drift parameters, a four-dimensional state vector is established, including position tracking error, torque fluctuation coefficient, spectral energy distribution, and temperature drift compensation. A nonlinear observer is used to estimate the unmodeled dynamics of the system in real time. A composite control strategy is designed, including proportional-integral-derivative control terms and unmodeled dynamic estimation terms. The proportional term is used to quickly respond to trajectory deviations, the integral term is used to eliminate steady-state errors, and the derivative term is used to predict the system's changing trend. When a sudden change in push rod resistance or gate-removing mechanism jamming is detected, a sliding mode compensation term based on the sliding surface function and switching gain is automatically activated. Based on the stability analysis results of the Lyapunov function, the control parameters are adjusted online by real-time calculation of the time derivative of the Lyapunov function to ensure the system's robustness in the presence of parameter uncertainties and external disturbances.

[0013] Furthermore, step S35 specifically involves performing the following operations: A state transition control model based on stochastic processes is established. By real-time monitoring of the four main vehicle positions, equipment alignment signal integrity, semi-automatic step sequence achievement marker states, and coke car body rotation angle, a six-tuple state vector is constructed to represent the system's operating state. An improved Markov decision process is used for action sequence optimization. A multi-dimensional state evaluation index is defined, including five dimensions: door clearing completion probability, frame clearing trolley displacement accuracy, travel positioning error, coal loading floor travel, and door closing alignment deviation. When the frame clearing trolley displacement exceeds the limit or the coke car rotation fails to reach the specified angle, 1-3 waiting cycles are automatically inserted and the optimal execution path is recalculated. The state transition probability matrix is ​​dynamically adjusted based on real-time operating data, and unreachable states are filtered out using a confidence threshold. The formula for the state transition control model is: , in, Let represent the five-dimensional state vector at time t, where This indicates the probability of completing the door clearing process. This indicates the displacement accuracy of the frame clearing trolley. This indicates the travel and positioning error. Indicates the stroke of the coal loading bottom plate. This indicates the door closing alignment deviation. Indicates in The process action chosen at each moment, Indicates the first The ideal target value of each state variable Indicates the first The permissible deviation range of each state variable. This represents the base incidence rate of abnormal events in the system per unit of time. This indicates the actual number of abnormal events detected. , These are the adjustment coefficients for the normal distribution term and the abnormal event term, respectively.

[0014] Preferably, step S4 includes the following steps: S41, monitors the timing of the generation of process action completion signals, and detects the limit signals or PLC completion flags of each process mechanism reaching the end point in real time. S42, adjust the travel trigger logic according to the parameters selected for the door and frame cleaning process. When the door or frame cleaning function is disabled, the travel preparation program is activated immediately after the corresponding process step is completed. S43, initiate the coordinated switching of the pre-alarm and delay mechanism, trigger the audible and visual alarm and start the timer before the vehicle moves, and issue the driving command after the delay ends; S44 dynamically updates the execution sequence of vehicle control commands, rearranging or refreshing the queue of control commands to be executed based on real-time process status and alarm feedback.

[0015] Furthermore, step S42 specifically involves performing the following operations: A multi-agent reinforcement learning-based travel triggering coordination control system is established. This system defines a state space containing the position states of the four main vehicles, equipment readiness flags, and process completion levels. An action space including travel trigger timing and speed curves is constructed. A multi-objective reward function considering energy efficiency, timing coordination, and safety margin is designed. A distributed Q-learning algorithm is employed, enabling each vehicle agent to collaboratively learn the optimal triggering strategy based on local observations. Distributed decision optimization is achieved by independently updating the action value function. The behavioral strategies of each agent are dynamically adjusted using a policy gradient method. When the door or frame clearing function is disabled, the corresponding equipment readiness flag is set to the immediately ready state in the state space, accelerating the travel triggering decision-making process. Travel triggering commands are generated in real-time based on the learned optimal strategy.

[0016] Furthermore, the characteristic is that step S44 specifically performs the following operations: A dynamic command scheduling system based on real-time data analysis is constructed. By continuously collecting the position coordinates of the four main vehicles, equipment operating status, process completion progress, and environmental parameters, a multi-dimensional evaluation system is established, including system service intensity, command queue length, and expected waiting time. State transition probability analysis technology is used to monitor the system operation. When the displacement deviation of the actuator and abnormal posture of the process equipment are detected, such as the displacement of the clearing trolley exceeding the allowable range or the rotation angle of the coke tank car not reaching the predetermined position, the command sequence reconstruction program is automatically started. The execution priority and timing arrangement of each control command are recalculated through optimization algorithms. This optimization process comprehensively considers multiple factors such as task weight, waiting time, and queue capacity limit. The control command queue is refreshed in real time based on the optimization results, and the feasibility of sequence changes is ensured through probabilistic verification methods.

[0017] Preferably, step S5 includes the following steps: S51 continuously compares the effectiveness of the equipment alignment signal with the four-vehicle interlocking conditions, and monitors the stability of the key interlocking signals in real time during the automatic cycle. S52 monitors abnormal changes in the status of coke tank cars, and detects faults such as out-of-tolerance tank position, loss of status signals, or mismatch with the coke pushing and coal loading cycle. S53 triggers a control flow interruption and switches to standby control mode. When an abnormal signal is detected, the automatic cycle is immediately paused, and the system switches to semi-automatic operation or safe shutdown state according to the preset switching logic.

[0018] Furthermore, step S53 specifically involves performing the following operations: A fuzzy Petri net-based intelligent fault diagnosis and decision-making system is constructed. This system collects multi-source state monitoring data from the four vehicle control systems in real time, including multi-dimensional parameters such as equipment alignment deviation, signal transmission delay, actuator response error, and environmental interference characteristics. A fault reasoning mechanism with self-learning capabilities is established, and a dynamic confidence propagation algorithm is used for fault level assessment. By analyzing the changing trends and correlations of state variables at each node, the system's comprehensive risk index is calculated. When a deviation from the preset safe operating boundary is detected, including persistent abnormal equipment alignment signals or the four-vehicle interlocking conditions not meeting the safety threshold, a multi-level fault confirmation process is initiated, including state monitoring, risk assessment, and decision execution. Based on real-time diagnostic results, a control mode switching decision is executed. The optimal backup control mode is selected through a preset rule base and case base to achieve a smooth transition of control modes. Simultaneously, the complete fault handling process is recorded for system self-optimization.

[0019] The technical effects and advantages of this invention are as follows: 1. In this invention, by constructing a fully automated closed-loop control architecture from status monitoring and condition verification to process execution, unmanned intelligent operation of the four vehicles working together is realized, which solves the problems of low automation level and low operating efficiency caused by process interruption and manual intervention in traditional control methods.

[0020] 2. In this invention, a collaborative control mechanism based on multiple signal verification and intelligent decision-making is adopted to achieve fine verification and safety interlocking of each process action, which solves the problem of insufficient safety and reliability caused by the coarse control granularity of the existing technology.

[0021] 3. In this invention, a flexible control strategy with configurable process parameters and dynamic reconfiguration of instruction sequences is adopted to achieve adaptive optimization scheduling for complex working conditions, thus solving the problems of insufficient equipment utilization and limited production efficiency caused by poor system flexibility. Attached Figure Description

[0022] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts: Figure 1 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0023] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0024] Reference Figure 1 As shown, this invention provides a technical solution: a four-vehicle high-efficiency control method for coke tanker trucks with optimized APS, comprising the following steps: S1 receives the planned furnace number sequence and the door and frame clearing process selection parameters, establishes a control starting point based on the planned furnace number, initializes the vehicle controller and status monitor, and verifies the consistency between the initial position of the four main vehicles and the planned furnace number.

[0025] S2 collects the current furnace number data of the SCP integrated car, coke quenching car, and smoke guiding car through the vehicle bus network, obtains the equipment alignment signal and semi-automatic step sequence achievement mark, and monitors the status of the coke tank car and the interlocking conditions of the four cars.

[0026] S3 performs multiple preset condition checks based on the collected real-time status data. The conditions include the integrity of the equipment alignment signal and the semi-automatic step sequence achievement mark, the four-car interlocking condition, and the permission of the door and frame clearing process selection parameters. If all conditions are met, the vehicle controller will sequentially execute the processes of door removal, coke pushing, door and frame clearing, traveling, coal loading, and door closing.

[0027] S4 monitors the timing of process action completion signals during execution, dynamically adjusts the travel triggering logic based on the door and frame clearing process selection parameters, activates pre-alarm and delay mechanisms to coordinate step sequence switching, and updates the execution sequence of vehicle control commands.

[0028] S5: When an abnormal equipment alignment signal, an invalid semi-automatic step sequence achievement flag, a missing four-vehicle interlock condition, or an incorrect status of the coke tanker is detected, the current control process is interrupted, the system switches to standby control mode, and the interruption event data is recorded.

[0029] In this embodiment, step S1 includes the following steps: S11 receives the planned furnace number sequence and the process selection parameters for clearing the gate and clearing the frame from the central dispatch system and transmits them to the vehicle controller through the system communication bus.

[0030] S12, establish a control starting point based on the planned furnace number, write the planned furnace number into the target register of each vehicle controller, and initialize the furnace number tracking and comparison program.

[0031] S13, initialize the vehicle controller and status monitor, load the pre-stored equipment parameters and control logic. The pre-stored equipment parameters include the mechanical dimensions, travel limits and positioning coordinates of each vehicle. The control logic includes the process sequence flow, safety interlock rules and abnormal handling strategies. Start the cyclic reading function of the four vehicle position feedback signals.

[0032] S14, verify the consistency between the initial position of the four cars and the planned furnace number, read the current furnace number sensor data of each car, and perform the first matching verification with the planned furnace number in the target register.

[0033] In this embodiment, step S2 includes the following steps: S21 collects the current furnace number data of the SCP integrated car, coke quenching car, and flue gas guiding car, and cyclically reads the position information output by the encoder or RFID tag reader of each car.

[0034] S22, acquire the equipment alignment signal and the semi-automatic step sequence completion flag, receive the fine alignment completion signal sent by the limit switches of each vehicle and the vision positioning system, and the step sequence preparation ready flag generated by the PLC program.

[0035] S23 reads the status of the coke tank and the position of the rotating mechanism in real time, and receives the cooperative operation permission signal sent by the four-car interlock device.

[0036] In this embodiment, step S3 includes the following steps: S31, verify the matching status between the current furnace number and the planned furnace number, compare the real-time collected data of the current furnace number of the vehicle with the planned furnace number for a second time, and generate a verification result of the furnace number consistency. S32, verify the integrity of the four-vehicle interlocking conditions and the equipment alignment signal, confirm that the interlocking conditions, alignment signals and semi-automatic step sequence achievement flags are all in a valid state, and generate a parallel permission signal; S33, confirm the execution conditions of the door and frame cleaning process selection parameters, and activate or skip the corresponding door and frame cleaning control sub-processes according to the received process parameters; S34 executes the door removal and coke pushing processes through the vehicle controller. When the parallel permission signal is valid, it sends door removal and coke pushing drive commands to the SCP integrated car and the coke blocking car. S35 sequentially executes the processes of clearing the door and frame, traveling, loading coal, and closing the door. Based on the process selection parameters and action completion signals, it triggers a series of subsequent automated operations according to the preset timing sequence.

[0037] Furthermore, the S32 verification process for the integrity of the four-vehicle interlocking conditions and equipment alignment signals involves establishing a fuzzy logic verification model as follows: Define the input set of the verification system, including One verification signal and One interfering factor The verification signals include four-vehicle interlocking condition signals, equipment alignment signals, and semi-automatic step sequence achievement flags. Interference factors include environmental parameters such as electromagnetic interference intensity and temperature fluctuations. Based on fuzzy logic theory, a Gaussian membership function is established for each verification signal. , in, Indicates the first The ideal reference value for a signal, This indicates the permissible deviation range of the signal.

[0038] Simultaneously, an S-shaped influence function is established for each interfering factor: , in, Indicates the first The activation threshold of each interfering factor This represents the sensitivity coefficient of the interference.

[0039] Considering that multiple verification signals need to meet the requirements simultaneously, the confidence scores of each signal are aggregated using a geometric mean, and the formula is as follows: , To suppress the cumulative effects of interfering factors, an exponential decay function is used, the formula of which is: , in It is a global interference attenuation factor that is dynamically adjusted based on historical operating data.

[0040] The final complete formula for calculating the verification index is obtained as follows: .

[0041] Specifically, during the formula processing, real-time data is collected every 100 milliseconds to calculate the membership value of each signal and the influence of each interference factor. The overall verification score is then obtained through the above formula. ,when If a parallel permission signal is generated, a re-verification process is initiated and the abnormal event is recorded.

[0042] Furthermore, step S34 specifically involves performing the following operations: An adaptive control model based on Lyapunov stability is constructed. By real-time acquisition of displacement sensor data of the gate-removing mechanism, measured values ​​of the push rod torque, equipment vibration spectrum characteristics, and temperature drift parameters, a four-dimensional state vector is established, including position tracking error, torque fluctuation coefficient, spectral energy distribution, and temperature drift compensation. A nonlinear observer is used to estimate the unmodeled dynamics of the system in real time. A composite control strategy is designed, including proportional-integral-derivative control terms and unmodeled dynamic estimation terms. The proportional term is used to quickly respond to trajectory deviations, the integral term is used to eliminate steady-state errors, and the derivative term is used to predict the system's changing trend. When a sudden change in push rod resistance or gate-removing mechanism jamming is detected, a sliding mode compensation term based on the sliding surface function and switching gain is automatically activated. Based on the stability analysis results of the Lyapunov function, the control parameters are adjusted online by real-time calculation of the time derivative of the Lyapunov function to ensure the system's robustness in the presence of parameter uncertainties and external disturbances.

[0043] Furthermore, the process of establishing the state transition control model in step S35 is as follows: First, define The state vector of the system at time t is ,in This indicates the probability of completing the door clearing process. This indicates the displacement accuracy of the frame clearing trolley. This indicates the travel and positioning error. Indicates the stroke of the coal loading bottom plate. This indicates the door closing alignment deviation. Based on the Markov property assumption, the system in action... The formula for the state transition probability under the action is: , in, This represents the environmental disturbance factors at time t.

[0044] Considering the independent Gaussian distribution characteristics of each state component, a basic probability model for state transition is established, and its formula is: , in, This represents the ideal target value of the i-th state variable. This represents the allowable deviation range for the i-th state variable.

[0045] Meanwhile, the occurrence of abnormal events in the system follows a Poisson distribution, with the distribution formula as follows: , in, This represents the base incidence rate of abnormal events per unit of time. This indicates the actual number of abnormal events detected.

[0046] By combining normal state transitions with the impact of abnormal events, we obtain the complete state transition probability formula: , in , These are the adjustment coefficients for the normal distribution term and the anomalous event term, respectively, and they satisfy... .

[0047] During formula processing, the state vector is collected every 2 seconds. Real-time data is used to calculate the relationship between each state component and the ideal target value. The degree of deviation, and the number of abnormal events within the current time window. The state transition probability is adjusted using a Bayesian update rule. When the deviation of any state component exceeds... When the time is right, the waiting period insertion mechanism is triggered, and the optimal action sequence is finally output to the vehicle controller for execution.

[0048] In this embodiment, step S4 includes the following steps: S41, monitors the timing of the generation of process action completion signals, and detects the limit signals or PLC completion flags of each process mechanism reaching the end point in real time. S42, adjust the travel trigger logic according to the parameters selected for the door and frame cleaning process. When the door or frame cleaning function is disabled, the travel preparation program is activated immediately after the corresponding process step is completed. S43, initiate the coordinated switching of the pre-alarm and delay mechanism, trigger the audible and visual alarm and start the timer before the vehicle moves, and issue the driving command after the delay ends; S44 dynamically updates the execution sequence of vehicle control commands, rearranging or refreshing the queue of control commands to be executed based on real-time process status and alarm feedback.

[0049] Furthermore, step S42 specifically involves performing the following operations: A multi-agent reinforcement learning-based travel triggering coordination control system is established. This system defines a state space containing the position states of the four main vehicles, equipment readiness flags, and process completion levels. An action space including travel trigger timing and speed curves is constructed. A multi-objective reward function considering energy efficiency, timing coordination, and safety margin is designed. A distributed Q-learning algorithm is employed, enabling each vehicle agent to collaboratively learn the optimal triggering strategy based on local observations. Distributed decision optimization is achieved by independently updating the action value function. The behavioral strategies of each agent are dynamically adjusted using a policy gradient method. When the door or frame clearing function is disabled, the corresponding equipment readiness flag is set to the immediately ready state in the state space, accelerating the travel triggering decision-making process. Travel triggering commands are generated in real-time based on the learned optimal strategy.

[0050] Furthermore, the characteristic is that step S44 specifically performs the following operations: A dynamic command scheduling system based on real-time data analysis is constructed. By continuously collecting the position coordinates of the four main vehicles, equipment operating status, process completion progress, and environmental parameters, a multi-dimensional evaluation system is established, including system service intensity, command queue length, and expected waiting time. State transition probability analysis technology is used to monitor the system operation. When the displacement deviation of the actuator and abnormal posture of the process equipment are detected, such as the displacement of the clearing trolley exceeding the allowable range or the rotation angle of the coke tank car not reaching the predetermined position, the command sequence reconstruction program is automatically started. The execution priority and timing arrangement of each control command are recalculated through optimization algorithms. This optimization process comprehensively considers multiple factors such as task weight, waiting time, and queue capacity limit. The control command queue is refreshed in real time based on the optimization results, and the feasibility of sequence changes is ensured through probabilistic verification methods.

[0051] Specifically, the dynamic instruction scheduling system is responsible for acquiring vehicle location, equipment status, and process parameters in real time. It analyzes system load and performance through a multi-dimensional indicator system, and re-plans the instruction execution order when an abnormal situation is detected, ensuring that the updated instruction sequence meets process requirements and safety standards.

[0052] In this embodiment, step S5 includes the following steps: S51 continuously compares the effectiveness of the equipment alignment signal with the four-vehicle interlocking conditions, and monitors the stability of the key interlocking signals in real time during the automatic cycle. S52 monitors abnormal changes in the status of coke tank cars, and detects faults such as out-of-tolerance tank position, loss of status signals, or mismatch with the coke pushing and coal loading cycle. S53 triggers a control flow interruption and switches to standby control mode. When an abnormal signal is detected, the automatic cycle is immediately paused, and the system switches to semi-automatic operation or safe shutdown state according to the preset switching logic.

[0053] Furthermore, step S53 specifically involves performing the following operations: A fuzzy Petri net-based intelligent fault diagnosis and decision-making system is constructed. This system collects multi-source state monitoring data from the four vehicle control systems in real time, including multi-dimensional parameters such as equipment alignment deviation, signal transmission delay, actuator response error, and environmental interference characteristics. A fault reasoning mechanism with self-learning capabilities is established, and a dynamic confidence propagation algorithm is used for fault level assessment. By analyzing the changing trends and correlations of state variables at each node, the system's comprehensive risk index is calculated. When a deviation from the preset safe operating boundary is detected, including persistent abnormal equipment alignment signals or the four-vehicle interlocking conditions not meeting the safety threshold, a multi-level fault confirmation process is initiated, including state monitoring, risk assessment, and decision execution. Based on real-time diagnostic results, a control mode switching decision is executed. The optimal backup control mode is selected through a preset rule base and case base to achieve a smooth transition of control modes. Simultaneously, the complete fault handling process is recorded for system self-optimization.

[0054] Specifically, the intelligent fault diagnosis and decision-making system comprises three core processing stages: the fault feature extraction stage identifies abnormal patterns through real-time data stream analysis and tracks parameter change trends using sliding time window technology; the risk assessment stage comprehensively considers the probability of fault occurrence, degree of impact, and propagation range to establish a dynamic risk map; and the decision execution stage selects the optimal solution from multiple backup control modes based on the risk level and equipment status, including degraded operation, partial function disabling, and system-wide safe shutdown.

[0055] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

[0056] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A coke drum car APS optimization four-car high efficiency control method, characterized in that, The method comprises the following steps: S1, receiving a planned furnace number sequence and a door clearing and frame clearing process selection parameter, establishing a control starting point based on the planned furnace number, initializing a vehicle controller and a state monitor, and verifying the consistency of the initial positions of the four vehicles and the planned furnace number; S2, collecting current furnace number data of the SCP integrated vehicle, the coke blocking vehicle, and the smoke guiding vehicle, obtaining a device alignment signal and a semi-automatic step sequence completion flag, and monitoring the state of the coke tank vehicle tank body and the four-vehicle interlocking condition; S3, performing multiple preset condition checks based on the collected real-time state data, and sequentially executing the door picking, coke pushing, door clearing and frame clearing, walking, coal loading, and door closing process actions if they are simultaneously satisfied; S4, monitoring the generation timing of the process action completion signal during the execution process, dynamically adjusting the walking trigger logic according to the door clearing and frame clearing process selection parameter, starting the pre-alarm and delay mechanism to coordinate the step sequence switching, and updating the execution sequence of the vehicle control instructions; S5, when a device state error is monitored, interrupting the current control process, switching to a backup control mode, and recording the interruption event data.

2. The coke drum car APS optimized four-car high efficiency control method according to claim 1, characterized in that: The step S3 comprises the following steps: S31, checking the matching state of the current furnace number and the planned furnace number; S32, checking the integrity of the four-vehicle interlocking condition, the device alignment signal, and the semi-automatic step sequence completion flag; S33, confirming the execution condition of the door clearing and frame clearing process selection parameter; S34, executing the door picking and coke pushing process actions through the vehicle controller; S35, sequentially executing the door clearing and frame clearing, walking, coal loading, and door closing process actions.

3. The coke drum car APS optimized four-car high efficiency control method according to claim 2, characterized in that: The step S32 specifically performs the following operations: A multi-source information verification system based on fuzzy logic is constructed, and a complete set of fuzzy membership functions is established, in which each interlocking signal corresponds to a Gaussian membership function to represent the confidence level of its signal quality, and each interference factor corresponds to an S-shaped membership function to quantify its destructive effect. A fuzzy logic aggregation operator is used to combine all these membership values into a comprehensive verification index. The system compares the verification index with a dynamically adjusted threshold value. If it is higher than the threshold value, a parallel permission signal is immediately generated and transmitted to the subsequent controller. If it is lower than the threshold value, a multiple verification loop is automatically triggered to identify transient abnormalities through time series analysis and update the membership parameters. The verification index formula is: ， wherein, denotes the total number of chained signals and device-to-device signals, denotes the number of environmental interference factors, denotes the real-time measurement value of the ith signal, denotes the real-time measurement value of the jth interference factor, denotes the fuzzy membership function of the ith signal, defined as wherein is the signal ideal value, is the signal allowed deviation range, denotes the fuzzy influence function of the jth interference factor, wherein is the interference threshold value, is the interference sensitivity coefficient, denotes the global interference attenuation factor.

4. The coke drum car APS optimized four-car high efficiency control method according to claim 2, characterized in that: The step S34 specifically performs the following operations: An adaptive control model is constructed, a state vector is established, unmodeled dynamics of the system are estimated in real time, and a composite control strategy is designed. When a coke pushing resistance mutation or a door picking mechanism jam is detected, a sliding mode compensation term based on a sliding mode function and a switching gain is automatically activated. The stability analysis result is used to adjust the control parameters online through real-time calculation of the time derivative.

5. The coke drum car APS optimized four car high efficient control method according to claim 2, characterized in that: The step S35 specifically performs the following operations: A state transition control model based on a random process is established, a six-tuple state vector is constructed to represent the system operation state by monitoring the position data of the four cars, the completeness of the equipment alignment signal, the semi-automatic step sequence achievement flag state and the rotation angle of the coke pot car tank body in real time, a Markov decision process is used for action sequence optimization, multi-dimensional state evaluation indexes are defined, when the action result deviates from the preset process standard, a waiting period is automatically inserted and the optimal execution path is recalculated, the state transition probability matrix is dynamically adjusted according to the real-time working condition data, and the unreachable state is filtered through a confidence threshold, wherein the formula of the state transition control model is: ， wherein, represents a multi-dimensional state vector at time t, is the vector length, represents a process action selected at time t, represents a process action selected at time t, represents an ideal target value of the jth state variable, represents an ideal target value of the jth state variable, represents an allowable deviation range of the jth state variable, represents an allowable deviation range of the jth state variable, represents a basic occurrence rate of system abnormal events per unit time, represents the actual number of abnormal events monitored, , are the adjustment coefficients of the normal distribution term and the abnormal event term, respectively.

6. The coke drum car APS optimized four car high efficient control method according to claim 1, characterized in that: The step S4 comprises the following steps: S41, monitoring the generation timing of the process action completion signal; S42, adjusting the walking trigger logic according to the door and frame cleaning process selection parameters; S43, starting the pre-alarm and delay mechanism coordination step sequence switching; S44, dynamically updating the execution sequence of the vehicle control instruction.

7. The coke drum car APS optimized four-car high efficiency control method of claim 6, wherein: The step S42 specifically performs the following operations: A walking trigger coordination control system based on multi-agent reinforcement learning is established, a state space is defined, an action space is constructed, a multi-objective reward function is designed, a distributed Q-learning algorithm is used, each vehicle agent learns the optimal trigger strategy based on local observation, distributed decision optimization is realized by updating the action value function independently, the behavior strategy of each agent is dynamically adjusted by combining the policy gradient method, when the door or frame cleaning function is disabled, the corresponding equipment readiness flag in the state space is set to an immediate readiness state, the walking trigger decision process is accelerated, and the walking trigger instruction is generated in real time based on the learned optimal strategy.

8. The coke drum car APS optimized four car high efficient control method according to claim 6, characterized in that: The step S44 specifically performs the following operations: A dynamic instruction scheduling system based on real-time data analysis is constructed, a multi-dimensional evaluation system is established, a state transition probability analysis technology is used to monitor the system operation condition, when the execution mechanism displacement deviation and the process equipment posture anomaly are detected, the instruction sequence reconstruction program is automatically started, the execution priority and timing arrangement of each control instruction are recalculated through the optimization algorithm, the control instruction queue is refreshed in real time based on the optimization result, and the feasibility of sequence change is ensured through the probability verification method.

9. The coke drum car APS optimized four car high efficient control method according to claim 1, characterized in that: The step S5 comprises the following steps: S51, continuously comparing the effectiveness of the equipment alignment signal and the four-car interlocking condition; S52, monitoring the abnormal change of the coke pot car tank body state; S53, triggering the control process interruption and switching to the standby control mode.

10. The coke drum car APS optimized four car high efficient control method according to claim 9, characterized in that: The step S53 specifically performs the following operations: An intelligent fault diagnosis and decision system is constructed, multi-source state monitoring data of the four-car control system are collected in real time, a fault reasoning mechanism with self-learning ability is established, a dynamic confidence degree propagation algorithm is used for fault level evaluation, the change trend and correlation of each node state quantity are analyzed, the system comprehensive risk index is calculated, when the deviation from the preset safe operation boundary is detected, a multi-level fault confirmation process is started, a control mode switching decision is executed based on the real-time diagnosis result, the optimal standby control mode is selected through the preset rule base and case base, the smooth transition of the control mode is realized, and the complete fault handling process is recorded.

Images