Intelligent belt conveyor detection and protection system and method based on multi-sensor fusion and self-adaptive control

Abstract

The invention relates to a belt conveyor intelligent detection protection system and method based on multi-sensor fusion and self-adaptive control, and belongs to the field of safety control of industrial conveying equipment. According to the technical scheme, running state information of the belt conveyor is collected in real time, wherein the running state information comprises a belt edge image obtained through an industrial camera, the belt speed obtained through an encoder and the weight of scattered materials obtained through a weighing sensor; based on the operation state information, a control module carries out intelligent decision-making to generate a control instruction; and the control instruction is sent to an actuator so as to adjust or protect the belt conveyor. According to the invention, intelligent perception, prediction control, adaptive decision and ordered protection are integrated, and by introducing core algorithms such as image entropy weighted control, a hierarchical state machine, a weight-time integral decision, priority arbitration and acceleration feedforward, technical leap from passive protection to active prevention and from single control to system collaboration is realized. And the accuracy and reliability of belt conveyor protection early warning are improved.

Description

TECHNICAL FIELD

[0001] The application relates to a belt conveyor intelligent detection protection system and method based on multi-sensor fusion and adaptive control, in particular to a belt conveyor intelligent detection protection system and method based on multi-sensor information fusion, adaptive control algorithm and priority dynamic arbitration, and belongs to the technical field of safety control of industrial conveying equipment. BACKGROUND

[0002] The belt conveyor is widely used in industrial production industry as an important material conveying equipment, and its efficient and stable operation plays a crucial role in guaranteeing production continuity and improving production efficiency. Once a fault occurs, not only will it cause production stagnation and economic losses, but also may cause safety accidents. Therefore, in the current era of industrial automation and intelligentization, higher requirements are put forward for the protection and early warning technology of the belt conveyor.

[0003] The prior art method often relies on simple sensors and threshold judgment, and has the following technical problems in actual application: 1. The control strategy is rigid; the traditional deviation and slip protection adopts fixed threshold switch control, which cannot adapt to the complex working conditions of the belt conveyor and is prone to false operation or missed report. 2. The sensing ability is single; the existing system has independent protection functions (such as deviation, tearing and slip), lacks information interaction and cooperative decision-making, and cannot cope with compound faults. 3. Lack of predictability and adaptability; the deviation adjustment relies on feedback, and the response is lagging. The speed control is based on the current speed deviation, and cannot predict the slip risk caused by load mutation. The image recognition has a high misjudgment rate under the interference of dust and water vapor. 4. Fault handling is disordered; when multiple faults occur simultaneously or successively, the system lacks priority management, which may delay the response to key faults. SUMMARY

[0004] The purpose of the present application is to provide a belt conveyor intelligent detection protection system and method based on multi-sensor fusion and adaptive control, which integrates intelligent sensing, predictive control, adaptive decision-making and orderly protection. By introducing image entropy weighted control, hierarchical state machine, weight-time integral decision-making, priority arbitration and acceleration feedforward, the application realizes a technical leap from passive protection to active prevention and from single control to system cooperation, improves the accuracy and reliability of belt conveyor protection and early warning, and solves the above problems in the background art.

[0005] The technical solution of the present application is: A belt conveyor intelligent detection protection method based on multi-sensor fusion and adaptive control, comprising the following steps: ① Real-time acquisition of the running state information of the belt conveyor, wherein the running state information at least includes a belt edge image acquired by an industrial camera, a belt speed acquired by an encoder, and a spilled material weight acquired by a weighing sensor; ② Based on the running state information, intelligent decision-making is performed by at least one control module to generate a control instruction: A. Run-off control module: calculating an entropy value of the belt edge image, and inputting the entropy value as a dynamic weight coefficient to a fuzzy PID controller to adaptively adjust the proportional, integral, and differential parameters of the fuzzy PID controller, and outputting a cylinder displacement amount for adjusting the belt run-off; B. Speed control module: calculating a real-time acceleration according to the belt speed acquired by the encoder, when the absolute value of the real-time acceleration exceeds a first acceleration threshold and continuously increases, generating a feedforward compensation amount, and superimposing the feedforward compensation amount and the output of a speed PID controller to jointly serve as a speed control instruction for the belt motor; C. Tearing protection module: based on the spilled material weight, running a light fault integrator and a heavy fault integrator in parallel; when the output value of the light fault integrator exceeds a first integral threshold, triggering a material cleaning instruction; when the output value of the heavy fault integrator exceeds a second integral threshold, triggering an emergency stop instruction; and when the heavy fault integrator is activated, freezing the operation of the light fault integrator; ③ Sending the control instruction to the corresponding actuator to adjust or protect the belt conveyor.

[0006] The execution steps of the run-off control module further include: defining three states of run-off protection: a normal state N, a first protection state P, and a second protection state S; when the run-off amount exceeds a first run-off threshold S1, migrating from the normal state N to the first protection state P, and driving the cylinder to perform a first-level adjustment action D1; when being in the first protection state P and the duration exceeding a first timing threshold T1, and the run-off amount exceeding a second run-off threshold S2, migrating to the second protection state S, and driving the cylinder to perform a second-level adjustment action D2; when being in the second protection state S and the duration exceeding a second timing threshold T2, triggering an emergency stop; wherein, when being in the second protection state S, if the run-off amount falls below the first run-off threshold S1, migrating to the first protection state P.

[0007] The method further includes a multi-fault priority arbitration mechanism: presetting priority codes for different types of fault signals; when multiple fault signals are received simultaneously, preferentially responding to a control instruction corresponding to a fault signal with a higher priority code, and allowing a high-priority instruction to preempt the control right of an executing low-priority instruction.

[0008] The priority codes are from high to low as follows: Mechanical switch signal of foreign matter penetration; Emergency shutdown signal triggered by heavy fault integrator; Emergency shutdown signal triggered by run-off control module in secondary protection state; Clearing signal triggered by light fault integrator; Adjustment signal triggered by run-off control module in primary protection state.

[0009] In the speed control module, when the absolute value of real-time acceleration is in a preset dead zone range , the output of feedforward compensation is shielded.

[0010] An intelligent detection and protection system of a belt conveyor based on multi-sensor fusion and adaptive control is used to realize the above method, comprising: ① a sensor group, including an industrial camera for collecting a belt edge image, an encoder for detecting a belt speed, and a weighing sensor for detecting the weight of spilled material; ② a controller, in communication connection with the sensor group, the controller being configured to execute a computer program stored therein to realize the functions of the run-off control module, the speed control module and the tearing protection module; ③ an actuator group, in communication connection with the controller, the actuator group at least including a cylinder for adjusting the run-off of the belt, a motor for driving the belt and a frequency converter for controlling the speed of the motor.

[0011] The industrial camera is built-in with an embedded processor for locally calculating the entropy value and the run-off amount of the belt edge image.

[0012] The system further comprises a human-computer interaction interface, in communication connection with the controller, for displaying the system state and fault information in real time, and for modifying the control parameters online, the control parameters including a first run-off threshold S1, a second run-off threshold S2, a first timing threshold T1, a second timing threshold T2, a first integral threshold , a second integral threshold , a first acceleration threshold and a feedforward compensation coefficient k.

[0013] The sensor group of the present application comprises an industrial camera, an encoder, a weighing sensor and a mechanical switch, the controller comprises an embedded processor and a PLC, and the actuator group comprises a servo driver, a frequency converter and a cylinder.

[0014] The core innovation point and technical features of the present application are as follows: 1. Run-off fuzzy PID self-tuning control of dynamic weight of image entropy; Technical features: use industrial camera to collect the image of the belt edge, calculate the entropy value of the image (representing the image definition and information content) in real time at the embedded end; the entropy value is used as the dynamic weight coefficient of the fuzzy PID controller, and the correction rate of the proportional, integral and differential parameters is adjusted in real time. Technical effect: when the image is clear, the adjustment response is accelerated, and when the image is blurred (disturbed), the system robustness is enhanced, forming a "image quality-control parameter" double closed loop adaptive adjustment, which fundamentally reduces the risk of miscontrol of visual detection.

[0015] 2. Protected hierarchical state machine migration model; Technical features: a state machine including normal state N, first protection state P and second protection state S is defined. The state migration condition not only depends on the run-off amount (S1, S2), but also introduces time threshold (T1, T2). A self-recovery path (S→P→N) with hysteresis loop is designed to avoid state chattering. Technical effect: by introducing time dimension and hysteresis logic, false shutdown caused by transient disturbance is avoided, while reliable judgment and processing of serious faults are ensured. All parameters (S1, S2, D1, D2, T1, T2) support online modification, with high flexibility.

[0016] 3. "Weight-time" hybrid integral decision algorithm for faults; Technical features: two parallel virtual integrators are constructed to process the weighing sensor signal. The light fault integrator reaches to trigger the material cleaning; the heavy fault integrator reaches to stop immediately. Set the interlocking logic, activated to freeze . Technical effect: through the integral algorithm, the transient impact is smoothed, the cumulative severity of the fault is accurately quantified, and the precise and decoupled response of light fault (material cleaning) and heavy fault (shutdown) is realized.

[0017] 4. Priority dynamic arbitration mechanism for multi-sensor fault signals; Technical features: all fault signals (foreign object penetration, heavy tear, secondary run-off, etc.) are assigned fixed interrupt priority codes (Level 0 highest). High-priority signals can preempt the control of low-priority tasks; signals of the same level use a combination of first-come-first-served and configurable time slice rotation; the fault state register has a latch function. Technical effect: an orderly, manageable and predictable fault response queue is established, ensuring that the most critical safety events are always handled most promptly, solving the control conflict problem when multiple faults occur.

[0018] 5. Encoder acceleration predictive speed regulation feedforward control; Technical features: through the encoder signal not only calculates the speed v(t), but also calculates the acceleration a(t) in real time. When |a(t)| exceeds the threshold and the trend continues, the feedforward compensation amount Δu = k·a(t) is generated and directly superimposed on the frequency converter speed given value. A dead zone is set to avoid steady-state oscillation. Technical effects: the transition from "feedback correction" to "feedforward prediction" is realized, the compensation is made in advance before the speed deviation occurs, the response speed of the system to load mutation is significantly improved, and slipping is effectively prevented. It constitutes a high-performance cascade control with PID feedback.

[0019] The beneficial effects of the present application are: All-round safety improvement: through multi-dimensional perception and intelligent decision-making, the incidence and harm of accidents such as deviation, tearing and slipping are significantly reduced. Control performance optimization: deviation adjustment response is faster and more stable; speed control lag is reduced, and dynamic quality is improved by more than 40%. Environmental adaptability is enhanced: image entropy weight and integral algorithm make the system have strong fault tolerance to dust, vibration and transient impact. Maintenance and operation efficiency is improved: online adjustable parameters, fault classification processing and state remote monitoring greatly facilitate debugging, optimization and fault diagnosis work, and reduce maintenance cost.

[0020] The present application integrates intelligent perception, predictive control, adaptive decision-making and orderly protection, and through the introduction of image entropy weighted control, hierarchical state machine, weight-time integral decision-making, priority arbitration and acceleration feedforward, realizes the technical leap from passive protection to active prevention and from single control to system cooperation, and improves the accuracy and reliability of the belt conveyor protection and early warning. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The present application is an embodiment system diagram. DETAILED DESCRIPTION

[0022] The present application is further described below through examples.

[0023] A belt conveyor intelligent detection and protection method based on multi-sensor fusion and adaptive control, comprising the following steps: ① Real-time acquisition of the running state information of the belt conveyor, the running state information at least including the belt edge image obtained through an industrial camera, the belt speed obtained through an encoder and the spilled material weight obtained through a weighing sensor; ② Based on the running state information, intelligent decision-making is made through at least one of the following control modules to generate a control instruction: A. Run-off control module: calculate the entropy value of the belt edge image, and input the entropy value as a dynamic weight coefficient to the fuzzy PID controller to adaptively adjust the proportional, integral, and differential parameters of the fuzzy PID controller, and output the cylinder displacement amount for adjusting the belt run-off; B. Speed control module: calculate the real-time acceleration according to the belt speed obtained by the encoder, when the absolute value of the real-time acceleration exceeds the first acceleration threshold and continues to increase, generate a feedforward compensation, and superimpose the feedforward compensation and the output of the speed PID controller to jointly serve as the speed control instruction of the belt motor; C. Tear protection module: based on the weight of the spilled material, run the light fault integrator and the heavy fault integrator in parallel; when the output value of the light fault integrator exceeds the first integral threshold, trigger the material cleaning instruction; when the output value of the heavy fault integrator exceeds the second integral threshold, trigger the emergency stop instruction; and when the heavy fault integrator is activated, freeze the operation of the light fault integrator; ③Send the control instruction to the corresponding actuator to adjust or protect the belt machine.

[0024] The execution steps of the run-off control module further include: Define three states of run-off protection: normal state (N), first protection state (P), and second protection state (S); when the run-off amount exceeds the first run-off threshold (S1), migrate from the normal state (N) to the first protection state (P), and drive the cylinder to perform the first level adjustment action (D1); When in the first protection state (P) and the duration exceeds the first timing threshold (T1), and the run-off amount exceeds the second run-off threshold (S2), migrate to the second protection state (S), and drive the cylinder to perform the second level adjustment action (D2); When in the second protection state (S) and the duration exceeds the second timing threshold (T2), trigger the emergency stop; Wherein, when in the second protection state (S), if the run-off amount falls below the first run-off threshold (S1), migrate to the first protection state (P).

[0025] The method further includes a multi-fault priority arbitration mechanism: preset priority codes for different types of fault signals; when multiple fault signals are received simultaneously, preferentially respond to the control instruction corresponding to the fault signal with a higher priority code, and allow the high-priority instruction to preempt the control of the low-priority instruction being executed.

[0026] The priority codes are in descending order as follows: Foreign object penetration mechanical switch signal; Emergency stop signal triggered by the heavy fault integrator; Emergency stop signal triggered by the run-off control module in the second protection state; a clear material signal triggered by the light fault integrator; an adjustment signal triggered by the deviation control module in the first protection state.

[0027] In the speed control module, when the absolute value of the real-time acceleration is in a preset dead zone range , the output of the feedforward compensation is shielded.

[0028] An intelligent detection and protection system of a belt conveyor based on multi-sensor fusion and adaptive control is used to realize the above method, comprising: ① a sensor group, including an industrial camera for collecting a belt edge image, an encoder for detecting a belt speed, and a weighing sensor for detecting the weight of spilled material; ② a controller, in communication connection with the sensor group, the controller being configured to execute a computer program stored therein to realize the functions of the deviation control module, the speed control module and the tearing protection module; ③ an actuator group, in communication connection with the controller, the actuator group at least including a cylinder for adjusting the deviation of the belt, a motor for driving the belt and a frequency converter for controlling the speed of the motor.

[0029] The industrial camera is built-in with an embedded processor for locally calculating the entropy value and the deviation amount of the belt edge image.

[0030] The system further comprises a human-computer interaction interface, in communication connection with the controller, for displaying the system state and fault information in real time, and for modifying the control parameters online, the control parameters including a first deviation threshold (S1), a second deviation threshold (S2), a first timing threshold (T1), a second timing threshold (T2), a first integral threshold , a second integral threshold , a first acceleration threshold and a feedforward compensation coefficient (k).

[0031] The sensor group of the present application comprises an industrial camera, an encoder, a weighing sensor, a mechanical switch, etc., the controller comprises an embedded processor, a PLC, etc., and the actuator group comprises a servo driver, a frequency converter, a cylinder, etc.

[0032] The core innovation points and technical features of the present application are as follows: 1. Deviation fuzzy PID self-tuning control of dynamic weight of image entropy value; Technical features: use industrial camera to collect the image of the edge of the belt, and calculate the entropy value (representing the image definition and information amount) of the image in real time at the embedded end; the entropy value is used as the dynamic weight coefficient of the fuzzy PID controller to adaptively adjust the correction rate of the proportional, integral and differential parameters in real time. Technical effect: when the image is clear, the adjustment response is accelerated, and when the image is blurred (disturbed), the system robustness is enhanced, forming a "image quality-control parameter" double closed loop adaptive adjustment, which fundamentally reduces the risk of miscontrol of visual detection.

[0033] 2, the protected hierarchical state machine migration model; Technical features: a state machine including normal state (N), first protection state (P) and second protection state (S) is defined. The state migration condition depends not only on the run-off amount (S1, S2), but also on the time threshold (T1, T2). A self-recovery path (S→P→N) with hysteresis loop is designed to avoid state chattering. Technical effect: by introducing time dimension and hysteresis logic, false shutdown caused by transient disturbance is avoided, while reliable judgment and processing of serious faults are ensured. All parameters (S1, S2, D1, D2, T1, T2) support online modification, with high flexibility.

[0034] 3, "weight-time" hybrid integral decision algorithm of fault; Technical features: two parallel virtual integrators are constructed to process the weighing sensor signal. The light fault integrator reaches to trigger the material cleaning; the heavy fault integrator reaches to stop immediately. Set the interlocking logic, activated . Technical effect: through the integral algorithm, the transient impact is smoothed, the cumulative severity of the fault is accurately quantified, and the precise and decoupled response of light fault (material cleaning) and heavy fault (shutdown) is realized.

[0035] 4, priority dynamic arbitration mechanism of sensor fault signal; Technical features: all fault signals (foreign object penetration, heavy tear, secondary run-off, etc.) are assigned fixed interrupt priority codes (Level 0 highest). High priority signals can preempt the control of low priority tasks; signals of the same level use a combination of first-come-first-served and configurable time slice rotation; the fault state register has a latch function. Technical effect: an orderly, manageable and predictable fault response queue is established, ensuring that the most critical safety events can always be handled most promptly, solving the control conflict problem when multiple faults occur.

[0036] 5, encoder acceleration predictive speed regulation feedforward control; Technical features: through the encoder signal not only calculates the speed v(t), but also calculates the acceleration a(t) in real time. When |a(t)| exceeds the threshold and the trend continues, the feedforward compensation amount Δu = k·a(t) is generated and directly superimposed on the frequency converter speed given value. Set the dead zone to avoid steady-state oscillation. Technical effects: the transition from "feedback correction" to "feedforward prediction" is realized, the compensation is made in advance before the speed deviation occurs, the response speed of the system to load mutation is significantly improved, and slipping is effectively prevented. It constitutes a high-performance cascade control with PID feedback.

[0037] The specific steps in the embodiments are as follows: 1. Run-off control implementation; The industrial camera periodically takes pictures of the belt edge, and the embedded processor calculates the image entropy H and the run-off amount ΔX. The normalized entropy H is used as the weight factor ω (0<ω<1). In the fuzzy PID controller of the PLC, the parameter adaptive law can be simplified as: Kp' = Kp0 + ω * ΔKp Ki' = Ki0 + ω * ΔKi Kd' = Kd0 + ω * ΔKd Where Kp0, Ki0, Kd0 are the reference parameters, and ΔKp, ΔKi, ΔKd are the adjustment amounts based on the fuzzy rule table. When ω is high (the image is clear), the parameter adjustment amplitude is large and the response is fast; when ω is low (the image is blurred), the adjustment amplitude is small and the system tends to be conservative and stable.

[0038] At the same time, the PLC maintains a state machine that defines the run-off protection. For example, set S1=20mm, S2=40mm, T1=5s, T2=3s. When ΔX>20mm, enter the P state, and the cylinder acts D1=10mm. If it is not restored within 5 seconds and ΔX>40mm, enter the S state, and the cylinder acts D2=20mm. If it is still not restored in the S state for 3 seconds, emergency stop.

[0039] 2. Speed control implementation; The encoder pulses are calculated by the PLC high-speed counter to obtain the speed v(t), and then the acceleration a(t) is calculated by the difference method. Set =0.5 m / s², =0.1 m / s², k=0.8.

[0040] If |a(t)|>0.5 and da / dt>0 (the trend continues to increase), then the feedforward compensation amount Δu = 0.8 * a(t).

[0041] If |a(t)|<0.1, then Δu=0.

[0042] Total speed control command U = Upid + Δu is sent to the frequency converter.

[0043] 3. Tearing protection implementation; The weighing sensor signal w(t) is sent to the PLC. Set the light fault threshold = 500g, the heavy fault threshold = 2000g, the light fault integral threshold = 1000*g*s, the heavy fault integral threshold = 5000*g*s.

[0044] The PLC calculates in each scan cycle: If w(t) > S1, and the light fault is not activated) If w(t) > S2, and the heavy fault is not activated) When > 1000, the material cleaning command is triggered, and the electric push rod is started. When > 5000, the immediate stop is triggered, and the integral operation of is frozen.

[0045] 4. Priority arbitration implementation; An interrupt priority register is set up in the PLC. When the foreign matter penetration switch (Level 0) signal comes, no matter whether the system is executing the speed regulation or the material cleaning, the current action is immediately stopped, and the emergency stop is executed.

[0046] 5. Parameter setting and monitoring; All the above parameters (S1, S2, T1, T2, k, etc.) can be set on the touch screen and written into the PLC through MODBUS. All running states, real-time data, fault codes and historical records can be viewed on the touch screen and uploaded to the DCS system.

[0047] This embodiment takes a comprehensive scene as an example to illustrate the running process of the system: System initialization: all initial parameters (such as S1, S2, T1, T2, k, etc.) are set through the touch screen.

[0048] Normal operation: the industrial camera continuously acquires images, calculates the entropy value and the running deviation. If the running deviation does not exceed S1, the system is in N state. The encoder continuously measures the speed and acceleration, and the feedforward control works dynamically according to a(t).

[0049] ​Runout fault handling: When runout > S1, state machine migrates from N to P, PLC drives cylinder to move D1 distance.

[0050] If runout recovers to < S1 within T1, state returns to N. If runout > S2 within T1, state migrates to S, PLC drives cylinder to move D2 distance. If runout does not recover within T2 in state S, Level 1 emergency stop is triggered.

[0051] Tear fault handling: Weighing sensor detects spillage, PLC initiates and integrator. If it is only a minor spillage ( reaches ), push rod is initiated to clean up the spillage, system alarms. If it is a serious tear ( reaches ), Level 1 emergency stop is immediately triggered, and current low-priority runout action is suspended through arbitration mechanism.

[0052] Concurrent fault arbitration: If mechanical foreign object penetration switch is triggered (Level 0) during the cleaning process, system will immediately interrupt all other tasks and execute the highest-priority emergency stop.

[0053] Communication and monitoring: Throughout the process, all key data, states and fault codes are uploaded to the touch screen and remote DCS system through MODBUS protocol, achieving full-process visual monitoring and recording.

Claims

1. A method for intelligent detection and protection of belt conveyors based on multi-sensor fusion and adaptive control, characterized in that... It includes the following steps: ① Real-time acquisition of the operating status information of the belt conveyor, the operating status information including at least the belt edge image obtained by an industrial camera, the belt speed obtained by an encoder, and the weight of spilled material obtained by a weighing sensor; ② Based on operational status information, intelligent decision-making is performed through at least one of the following control modules to generate control commands: A. Belt misalignment control module: Calculates the entropy value of the belt edge image and inputs the entropy value as a dynamic weighting coefficient to the fuzzy PID controller to adaptively adjust the proportional, integral, and derivative parameters of the fuzzy PID controller, and outputs the cylinder displacement amount used to adjust belt misalignment; B. Speed ​​control module: Calculates real-time acceleration based on the belt speed obtained by the encoder. When the absolute value of the real-time acceleration exceeds the first acceleration threshold and continues to increase, a feedforward compensation amount is generated. This feedforward compensation amount is then superimposed with the output of the speed PID controller and used together as the speed control command for the belt motor. C. Tear Protection Module: Based on the weight of spilled material, the minor fault integrator and the major fault integrator run in parallel; when the output value of the minor fault integrator exceeds the first integration threshold, a material clearing command is triggered; when the output value of the major fault integrator exceeds the second integration threshold, an emergency stop command is triggered; and when the major fault integrator is activated, the operation of the minor fault integrator is frozen. ③ Send control commands to the corresponding actuators to adjust or protect the belt conveyor.

2. The intelligent detection and protection method for belt conveyors based on multi-sensor fusion and adaptive control according to claim 1, characterized in that... The execution steps of the deviation control module also include: The three states of deviation protection are defined: normal state N, primary protection state P, and secondary protection state S. When the deviation exceeds the first deviation threshold S1, it transitions from normal state N to primary protection state P and drives the cylinder to execute the first-stage adjustment action D1. When the system is in the first-level protection state P and the duration exceeds the first timing threshold T1, and the deviation exceeds the second deviation threshold S2, it migrates to the second-level protection state S and drives the cylinder to perform the second-level adjustment action D2. When the system is in the secondary protection state S for a duration exceeding the second timing threshold T2, an emergency shutdown is triggered. When in the secondary protection state S, if the deviation amount falls back to less than the first deviation threshold S1, then it will migrate to the primary protection state P.

3. The intelligent detection and protection method for belt conveyors based on multi-sensor fusion and adaptive control according to claim 1 or 2, characterized in that... The method also includes a multi-fault priority arbitration mechanism: pre-setting priority codes for different types of fault signals; When multiple fault signals are received simultaneously, the control command corresponding to the fault signal with the higher priority code is responded to first, and the higher priority command is allowed to preempt the control of the lower priority command that is being executed.

4. The intelligent detection and protection method for belt conveyors based on multi-sensor fusion and adaptive control according to claim 3, characterized in that... The priority codes, from highest to lowest, are as follows: Mechanical switch signal for foreign object penetration; Emergency stop signal triggered by the major fault integrator; An emergency stop signal triggered by the deviation control module in the secondary protection state; The material clearing signal is triggered by the minor fault integrator; The adjustment signal is triggered by the deviation control module in the first-level protection state.

5. The intelligent detection and protection method for belt conveyors based on multi-sensor fusion and adaptive control according to claim 2, characterized in that: In the speed control module, when the absolute value of the real-time acceleration is within a preset dead zone range... When inside, the output of the feedforward compensation is shielded.

6. A belt conveyor intelligent detection and protection system based on multi-sensor fusion and adaptive control, used to implement the method described in any one of claims 1-5, characterized in that... include: ① Sensor group, including an industrial camera for acquiring images of the belt edge, an encoder for detecting belt speed, and a weighing sensor for detecting the weight of spilled material; ② Controller, which is communicatively connected to the sensor group, is configured to execute a computer program stored therein to realize the functions of the above-mentioned deviation control module, speed control module and tear protection module; ③ Actuator group, which is connected in communication with the controller, includes at least a cylinder for adjusting belt misalignment, a motor for driving the belt, and a frequency converter for controlling the speed of the motor.

7. The intelligent detection and protection system for belt conveyors based on multi-sensor fusion and adaptive control according to claim 6, characterized in that: The industrial camera has a built-in embedded processor for locally calculating the entropy and deviation of the belt edge image.

8. The intelligent detection and protection system for belt conveyors based on multi-sensor fusion and adaptive control according to claim 6, characterized in that: The system also includes a human-machine interface, which is communicatively connected to the controller. This interface is used to display system status and fault information in real time, and to modify control parameters online. The control parameters include a first deviation threshold S1, a second deviation threshold S2, a first timing threshold T1, a second timing threshold T2, and a first integral threshold. Second integration threshold First acceleration threshold And the feedforward compensation coefficient k.

Images