Long distance belt conveyor control system for TBM construction

Abstract

The application relates to a long-distance belt conveyor control system for TBM construction and belongs to the technical field of long-distance belt conveyor control in TBM construction. In view of the technical problem that the long-distance belt conveyor in TBM construction is easily affected by load fluctuation, the technical solution points of the application are as follows: a load detection unit and a speed detection unit are arranged; a central controller is connected with the load detection unit, the speed detection unit and a driving unit containing a frequency converter and a driving motor, a sampling and control cycle of 50ms-200ms is adopted, a model prediction control algorithm is built-in, target torque and speed are output by rolling optimization under the constraints of speed and acceleration absolute values, and the frequency converter adopts vector or direct torque control to drive the motor to operate. The system is mainly used for slag transportation control of the long-distance belt conveyor in TBM construction and realizes efficient and stable operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of long-distance belt control. More particularly, the present application relates to a long-distance belt conveyor control system for TBM construction. BACKGROUND

[0002] In the process of TBM construction, the long-distance belt conveyor is the core equipment for slag transportation, and its running state directly affects the tunneling efficiency and construction cost. Due to the complex TBM tunneling conditions, affected by factors such as rock quality changes and tunneling speed adjustment, the material load of the belt conveyor often presents frequent and large amplitude fluctuations, and the switching from empty load to full load can be completed in a short time, which puts high requirements on the dynamic adjustment capability of the belt conveyor.

[0003] In terms of running smoothness, the traditional control method has weak constraint ability on acceleration. Due to the large inertia of the long-distance belt conveyor and the long transmission link, if the adjustment amplitude is too large when the load fluctuates, it is easy to cause the belt acceleration to exceed the reasonable range, causing the slip between the belt and the drum, the relative sliding of the material on the belt, and even causing the belt tensioning device to act frequently, aggravating the equipment wear and shortening the service life of the belt and related parts. SUMMARY

[0004] It is another object of the present application to provide a long-distance belt conveyor control system for TBM construction.

[0005] In order to achieve these objects and other advantages according to the present application, a long-distance belt conveyor control system for TBM construction is provided, comprising:

[0006] A load detection unit is installed 5m-10m behind the material receiving point of the belt conveyor, for real-time detection of material load and generation of a load signal, and the range of the load detection unit is 0kg / m-500kg / m;

[0007] A speed detection unit is installed at the shaft end of the belt conveyor drive drum, for real-time detection of the belt running speed and generation of a speed signal, and the detection range of the speed detection unit is 0m / s-5m / s;

[0008] A drive unit comprising a frequency converter and a belt conveyor drive motor;

[0009] A central controller is connected with the load detection unit, the speed detection unit and the drive unit through industrial buses respectively, receives the load signal and the speed signal with a sampling period of 50-200 ms, and takes the sampling period as a control period; the central controller is built-in with a model predictive control algorithm, takes the load signal and the speed signal as inputs, takes the target torque and the target speed of the belt conveyor drive motor as outputs, establishes a prediction model containing the belt conveyor dynamics equation and the energy consumption equation, sets the prediction time domain as 3-10 s, and sets the control time domain as 1-3 s;

[0010] In each control period, the central controller performs the following calculation steps: according to the current load and speed state, the speed sequence and the torque sequence in the future time period are optimized within the prediction time domain to meet the speed range of 1.5-3.5 m / s required by the TBM discharging rate as the constraint condition, the absolute value of the belt conveyor acceleration is not more than 0.1 m / s 2 -0.3 m / s 2 for smoothness constraint, the total energy consumption of the drive unit is minimized as the optimization target; after the optimization calculation, the target torque and the target speed corresponding to the control period are taken as the control instructions output;

[0011] The frequency converter is configured to receive the target torque and the target speed instructions sent by the central controller, and adopts the vector control or the direct torque control mode to drive the belt conveyor drive motor to run according to the instructions to adjust the belt conveyor running speed and the torque.

[0012] Preferably, the model predictive control algorithm built-in in the central controller is further integrated with a dynamic disturbance observer, the dynamic disturbance observer takes the driving speed of the TBM as a feedforward signal, and estimates the equivalent material disturbance load introduced by the TBM driving mutation based on the difference between the actual value and the predicted value of the load signal and the speed signal;

[0013] In the rolling optimization step, the prediction model introduces the equivalent material disturbance load as a time-varying disturbance item into the belt conveyor dynamics equation, and corrects the optimization target to the total energy consumption minimization of the drive unit and the speed fluctuation variance minimization of the belt conveyor;

[0014] The speed fluctuation variance is calculated based on the prediction model and the equivalent torque disturbance in the predicted speed sequence within the prediction time domain.

[0015] Preferably, the dynamic disturbance observer is further configured to:

[0016] establish a moving disturbance source model based on the equivalent material disturbance load and the current running speed of the belt;

[0017] The moving disturbance source model characterizes the material disturbance introduced by the TBM tunneling mutation as an equivalent lumped mass block. The equivalent lumped mass block moves downstream from the receiving point at the belt running speed, and the current position of the equivalent lumped mass block on the belt conveyor is tracked in real time.

[0018] In the rolling optimization step of the model predictive control algorithm, the predictive model integrates the moving disturbance source model and introduces the current position and size of the equivalent lumped mass block as time-varying parameters into the segmented dynamic equation and energy consumption equation of the belt conveyor. The optimization objective is further modified to minimize the fluctuation variance of the dynamic tension in the high-tension zone along the path of the moving disturbance source in the prediction time domain. The high-tension zone includes at least the cutting zone of the drive roller.

[0019] Preferably, it also includes at least one position reference sensor, which is installed near the cutting side of the drive roller or redirecting roller along the belt conveyor, and is used to generate a reference trigger signal when the equivalent concentrated mass block passes by.

[0020] The central controller is further configured to perform the following calibration steps:

[0021] The reference trigger signal is used as the absolute position reference;

[0022] The cumulative position error is calculated based on the time deviation between the predicted arrival time of the equivalent lumped mass block and the actual generation time of the reference trigger signal.

[0023] By utilizing the position accumulation error, the current position of the equivalent lumped mass block tracked in the moving disturbance source model is corrected in real time, and the corrected position information is fed back to the prediction model for subsequent rolling optimization calculations.

[0024] Preferably, the central controller is also configured to execute an adaptive periodic adjustment method, which includes the following steps:

[0025] S1. In each control cycle, the central controller evaluates the severity of the dynamic response of the belt conveyor. The evaluation is based on the current disturbance intensity and the expected future fluctuations. The current disturbance intensity is judged based on the magnitude and rate of change of the load signal transmitted by the load detection unit in the near future. The expected future fluctuations are judged based on the stability of the future speed sequence output by the model predictive control algorithm in the current cycle in the prediction time domain.

[0026] S2. Based on the evaluation results of step S1, the central controller dynamically adjusts the duration of the next and subsequent control cycles. When the evaluation results indicate that the dynamic response of the belt conveyor tends to be more severe, the control cycle is shortened, and the lower limit of the cycle duration is not lower than the lower limit of the sampling cycle. When the evaluation results indicate that the dynamic response of the belt conveyor tends to be more stable, the control cycle is extended, and the upper limit of the cycle duration does not exceed a preset maximum value.

[0027] S3. When performing the control cycle adjustment in step S2, the central controller performs a gradual cycle switching with a fixed time step.

[0028] Preferably, the method for determining the current disturbance intensity is as follows:

[0029] Within a sliding evaluation time window, the central controller performs real-time analysis on the received load signal sequence and simultaneously calculates the change amplitude index and change rate index. Specifically, the difference between the maximum and minimum values ​​of the load signal within the evaluation time window is calculated to obtain the load fluctuation range as the change amplitude index; the absolute value of the first derivative of the load signal with respect to time within the evaluation time window is calculated, and the maximum value within the evaluation time window is taken as the change rate index.

[0030] The central controller compares the change amplitude indicator with a preset amplitude threshold and the change rate indicator with a preset rate threshold. When either the change amplitude indicator or the change rate indicator exceeds its corresponding threshold, the current disturbance intensity is determined to be high; when neither the change amplitude indicator nor the change rate indicator exceeds its corresponding threshold, the current disturbance intensity is determined to be low.

[0031] The preferred method is the one for judging future volatility expectations:

[0032] The central controller performs dynamic characteristic analysis on the future velocity sequence, including continuous fluctuation detection and constraint violation assessment. The central controller identifies all segments in the future velocity sequence that satisfy the condition that multiple consecutive sampling points maintain the same direction of change, and counts the number of segments as a continuous fluctuation index. The central controller calculates the acceleration sequence based on the future velocity sequence and counts the proportion of sampling points in the velocity sequence that violate smoothness constraints.

[0033] The central controller obtains a comprehensive evaluation value based on the continuous fluctuation index and the constraint violation ratio. When the comprehensive evaluation value exceeds a preset evaluation threshold, it is determined that the expected future fluctuation is high; when the comprehensive evaluation value does not exceed the preset evaluation threshold, it is determined that the expected future fluctuation is low.

[0034] Preferably, the method for judging the degree of dynamic response is as follows: when the current disturbance intensity is high or the expected future fluctuation is high, the dynamic response of the belt conveyor is judged to be drastic; when the current disturbance intensity is low and the expected future fluctuation is low, the dynamic response of the belt conveyor is judged to be stable.

[0035] Preferably, the model predictive control algorithm further includes an online parameter estimation module, which is used to estimate the dynamic parameters and energy consumption parameters of the belt conveyor in real time and update the prediction model. The online parameter estimation module takes the load signal and speed signal as input, performs parameter estimation at a preset sampling period, and feeds the updated parameters back to the model predictive control algorithm for rolling optimization calculation.

[0036] Preferably, the online parameter estimation module uses recursive least squares or extended Kalman filter for parameter estimation; wherein, recursive least squares is used to estimate the inertia coefficient and friction coefficient of the belt conveyor system, and extended Kalman filter is used to estimate the state variables and parameters of the belt conveyor in the presence of measurement noise, and improves the estimation accuracy through adaptive gain adjustment.

[0037] The present invention has at least the following beneficial effects:

[0038] First, this invention, through the precise layout and real-time monitoring of load detection units and speed detection units, combined with model predictive control executed by the central controller with a cycle of 50ms-200ms, can ensure the smooth operation of the belt conveyor under speed ranges of 1.5m / s-3.5m / s and acceleration constraints of 0.1m / s²-0.3m / s², reducing material spillage and mechanical damage to the belt caused by sudden speed changes. It can also optimize the energy consumption of the drive unit through energy consumption equations, adapting to the continuous operation requirements of long-distance, dynamic slag removal in TBM construction.

[0039] Secondly, this invention uses the TBM tunneling speed as a feedforward signal through a dynamic disturbance observer, and estimates the equivalent disturbance in real time by combining the deviation between the actual and predicted values ​​of load and speed, so that the prediction model can incorporate the influence of disturbance in advance; at the same time, the speed fluctuation variance is included in the optimization objective, which can reduce the belt speed oscillation caused by sudden changes in TBM tunneling, avoid the accumulation or disconnection caused by sudden changes in material conveying speed, and improve the continuity of slag discharge.

[0040] Third, this invention characterizes material disturbance as an equivalent concentrated mass block that moves with the belt by using a moving disturbance source model. Its position is tracked in real time and incorporated into the segmented dynamic equation. The dynamic tension fluctuation in high-tension areas (such as the drive roller cutting area) is optimized in a targeted manner. This can avoid the impact of sudden local tension rise on the belt body and joint during the disturbance movement, reduce the risk of belt breakage, and extend the equipment maintenance cycle.

[0041] Fourth, the present invention can correct the cumulative position error of the equivalent concentrated mass block by using the absolute position trigger signal provided by the position reference sensor, ensuring that the disturbance position tracking does not deviate from the actual path during long-distance belt conveyor operation, making the prediction model's tension optimization in the high-tension zone more in line with the actual working conditions, and reducing control lag or over-adjustment caused by position deviation.

[0042] Fifth, this invention dynamically adjusts the control cycle by assessing the severity of the dynamic response. When the disturbance is severe, the cycle is shortened to speed up the response, and when the operation is stable, the cycle is extended to reduce the computational load on the controller. At the same time, the cycle is switched gradually with a fixed step size, which can avoid control oscillations caused by abrupt changes in the cycle, thus balancing control accuracy and system operation economy.

[0043] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0044] Figure 1 This is a flowchart of a long-distance belt conveyor control system for TBM construction, as described in one of the technical solutions of the present invention. Detailed Implementation

[0045] The present invention will be further described in detail below with reference to embodiments, so that those skilled in the art can implement it based on the description.

[0046] like Figure 1 As shown, the present invention provides a long-distance belt conveyor control system for TBM construction, comprising:

[0047] The load detection unit is installed 5m-10m behind the material receiving point of the belt conveyor. It is used to detect the material load in real time and generate a load signal. The range of the load detection unit is 0kg / m-500kg / m.

[0048] The speed detection unit is installed on the shaft end of the belt conveyor drive drum. It is used to detect the belt running speed in real time and generate a speed signal. The detection range of the speed detection unit is 0m / s - 5m / s.

[0049] The drive unit includes a frequency converter and a belt conveyor drive motor;

[0050] The central controller is connected to the load detection unit, speed detection unit, and drive unit via an industrial bus. The central controller receives load and speed signals with a sampling period of 50ms-200ms and uses this sampling period as the control period. The central controller has a built-in model predictive control algorithm. The model predictive control algorithm takes the load and speed signals as inputs and the target torque and target speed of the belt conveyor drive motor as outputs to establish a predictive model that includes the belt conveyor dynamics equations and energy consumption equations. The prediction time domain is set to 3s-10s and the control time domain is set to 1s-3s.

[0051] Within each control cycle, the central controller performs the following calculation steps: Based on the current load and speed status, it continuously optimizes the speed and torque sequences for future time periods within the prediction time domain, using the speed range of 1.5 m / s-3.5 m / s required to meet the TBM slag discharge rate as a constraint, and ensuring that the absolute value of the belt conveyor acceleration does not exceed 0.1 m / s². 2 -0.3m / s 2 To ensure smoothness, the optimization objective is to minimize the total energy consumption of the drive unit. After the optimization calculation is completed, the target torque and target speed of the corresponding control cycle are output as control commands.

[0052] The frequency converter is configured to receive target torque and target speed commands from the central controller, and use vector control or direct torque control to drive the belt conveyor drive motor to operate according to the commands and regulate the belt conveyor speed and torque.

[0053] In the detection unit, the installation distance of the load detection unit can be selected as 5m, 7m, or 10m, with a fixed measuring range of 0-500kg / m. The detection range of the speed detection unit can be selected as 0.5m / s, 3m / s, or 5m / s. The load detection unit can be a belt scale or a load cell, while the speed detection unit can be a photoelectric encoder or a Hall sensor. The housing of the detection unit can be made of stainless steel or aluminum alloy, and the sensor probe can be made of wear-resistant alloy steel to ensure stable operation in the construction environment. The load detection unit must be installed at the selected distance behind the material receiving point of the belt conveyor, close to the belt bearing surface without affecting material conveying. The speed detection unit must be installed on the shaft end of the belt conveyor drive drum, maintaining coaxiality with the shaft to ensure detection accuracy. In the drive unit, a general-purpose low-voltage frequency converter can be used, and a three-phase asynchronous motor can be used. The frequency converter housing can be made of cold-rolled steel plate, and the motor housing can be made of cast iron. The frequency converter can be installed in the control cabinet of the belt conveyor room or on a bracket near the drive motor. The drive motor must be installed on one side of the drive drum and connected via a coupling. The central controller can be a PLC or an industrial control computer, with an aluminum alloy casing. It is installed in the TBM control room or a dedicated control box for the belt conveyor. The control box must be fixed in a dry and ventilated location, and the cables should be laid along the outside of the belt conveyor frame, avoiding areas with strong electromagnetic interference. The sampling and control cycle of the central controller can be selected as 50ms, 100ms, or 200ms. The prediction time domain of the model predictive control algorithm can be selected as 3s, 6s, or 10s, and the control time domain can be selected as 1s, 2s, or 3s. These parameters need to be adjusted according to the length of the belt conveyor and the frequency of material fluctuations. A shorter cycle should be selected for long distances and large fluctuations, and a longer cycle should be selected for short distances and small fluctuations.

[0054] The working process is as follows: the load detection unit and speed detection unit collect corresponding signals in real time and transmit them to the central controller via the industrial bus. The central controller receives the signals at a set cycle and calls the built-in model predictive control algorithm containing the belt conveyor dynamics equations and energy consumption equations. Under speed constraints of 1.5 m / s, 2.5 m / s, or 3.5 m / s, and acceleration constraints of 0.1 m / s², 0.2 m / s², or 0.3 m / s², it continuously optimizes the future speed and torque sequence with the goal of minimizing energy consumption. The target torque and speed of the current cycle are extracted as control commands and output to the frequency converter. The frequency converter drives the motor using vector control or direct torque control, thereby adjusting the belt conveyor's operating status. Functional testing uses a long-distance TBM slag discharge belt conveyor as the experimental object, simulating different load and speed conditions. Key parameters are recorded using a data acquisition instrument. The system runs continuously for 8 hours, recording data hourly, analyzing speed fluctuations and whether acceleration meets the constraints, and calculating the average energy consumption.

[0055] By adopting this technical solution, the present invention, through the precise layout and real-time monitoring of the load detection unit and speed detection unit, combined with model predictive control executed by the central controller with a cycle of 50ms-200ms, can ensure the smooth operation of the belt conveyor under the speed range of 1.5 m / s-3.5m / s and acceleration constraints of 0.1 m / s²-0.3m / s², reducing material spillage and mechanical damage to the belt caused by sudden speed changes. At the same time, it can optimize the energy consumption of the drive unit through the energy consumption equation, adapting to the continuous operation requirements of long-distance, dynamic slag removal in TBM construction.

[0056] In another technical solution, the model predictive control algorithm built into the central controller also integrates a dynamic disturbance observer. The dynamic disturbance observer uses the tunneling speed of the TBM as a feedforward signal and estimates the equivalent material disturbance load introduced by the sudden change in TBM tunneling in real time based on the difference between the actual and predicted values ​​of the load signal and the speed signal.

[0057] In the rolling optimization step, the prediction model introduces the equivalent material disturbance load as a time-varying disturbance term into the belt conveyor dynamic equation, and modifies the optimization objective to minimize the total energy consumption of the drive unit and the variance of the belt conveyor speed fluctuation.

[0058] Among them, the speed fluctuation variance is calculated based on the prediction model and equivalent torque disturbance, and is obtained from the predicted speed sequence in the prediction time domain;

[0059] The feedforward signal for the dynamic disturbance observer is the TBM tunneling speed, which can be selected as 3r / min, 5r / min or 8r / min. This signal can be acquired by a magnetoelectric speed sensor or a photoelectric speed sensor. The dynamic disturbance observer is a software module of the central controller, which does not require additional hardware. Its response time is set to 10ms, 20ms or 30ms to track disturbance changes in real time.

[0060] In real-time estimation of the equivalent material disturbance load introduced by TBM tunneling abrupt changes, the estimation process revolves around two types of signals: first, the abrupt change characteristics of the TBM tunneling speed (e.g., a "tunneling abrupt change" is determined when the speed change rate exceeds a preset threshold); and second, the difference between the actual load signal from the load detection unit and the predicted load signal output by the prediction model. The disturbance estimation process is first triggered by the speed change signal, then the degree of load deviation from expectations is quantified based on this difference. Combining the correlation between the speed change amplitude and load deviation (e.g., the larger the speed change, the higher the weighting coefficient of the corresponding load deviation can be), the equivalent material disturbance load is finally calculated. The calculation of the equivalent material disturbance load can adopt a "deviation quantification + feedforward correction" approach. First, the instantaneous difference between the actual load collected by the load detection unit and the predicted load output by the prediction model is calculated. Then, a correction coefficient is determined based on the abrupt change rate of the TBM tunneling speed. This correction coefficient is an adjustable value of 0.1-0.5, which can be calibrated according to actual working conditions. This correction coefficient is then used to weight and correct the aforementioned instantaneous load difference. Simultaneously, based on the structural parameters of the conveyor belt receiving point, a load conversion coefficient for speed change abruptly is preset, and the corresponding feedforward correction load is calculated by combining the abrupt change rate of the TBM tunneling speed. Finally, the corrected load deviation is added to the equivalent load corresponding to the speed change abruptly to obtain the equivalent material disturbance load;

[0061] During operation, the speed sensor collects the TBM's tunneling speed and transmits it to the dynamic disturbance observer. The observer simultaneously receives the actual load and speed signals, compares them with the predicted signals output by the prediction model, calculates the difference, and estimates the equivalent material disturbance load in real time using feedforward signals. In the rolling optimization step, the prediction model incorporates this disturbance load as a time-varying disturbance term into the conveyor belt dynamic equation. Simultaneously, it calculates the speed fluctuation variance based on the speed sequence within the predicted time domain. The weighting coefficient of the speed fluctuation variance can be adjusted according to the working conditions; the weight is increased when the disturbance is large. Minimizing the total energy consumption of the drive unit and minimizing the speed fluctuation variance are jointly used as optimization objectives for rolling optimization. Functional testing uses a TBM muck conveyor belt with dynamic disturbance as the experimental object, simulating material disturbance caused by sudden changes in TBM tunneling speed. The speed fluctuation curve is observed using an oscilloscope, and the disturbance scenario is repeated 10 times with the speed fluctuation amplitude recorded.

[0062] By adopting this technical solution, the present invention uses the TBM tunneling speed as a feedforward signal through a dynamic disturbance observer, and estimates the equivalent disturbance in real time by combining the deviation between the actual and predicted values ​​of load and speed, so that the prediction model can incorporate the influence of disturbance in advance; at the same time, the speed fluctuation variance is included in the optimization objective, which can reduce the belt speed oscillation caused by sudden changes in TBM tunneling, avoid the accumulation or discontinuity caused by sudden changes in material conveying speed, and improve the continuity of slag discharge.

[0063] In another technical solution, the dynamic disturbance observer is further configured as follows:

[0064] A model of the moving disturbance source is established based on the equivalent material disturbance load and the current running speed of the belt;

[0065] The moving disturbance source model characterizes the material disturbance introduced by the TBM tunneling mutation as an equivalent lumped mass block. The equivalent lumped mass block moves downstream from the receiving point at the belt running speed, and the current position of the equivalent lumped mass block on the belt conveyor is tracked in real time.

[0066] In the rolling optimization step of the model predictive control algorithm, the predictive model integrates the moving disturbance source model, and introduces the current position and size of the equivalent lumped mass block as time-varying parameters into the segmented dynamic equations and energy consumption equations of the belt conveyor. The optimization objective is further modified to minimize the fluctuation variance of the dynamic tension in the high-tension zone along the path of the moving disturbance source within the prediction time domain. The high-tension zone includes at least the cutting-in zone of the drive roller. The moving disturbance source model is a software algorithm module of the central controller, requiring no additional hardware. The mass of the equivalent lumped mass block is calculated based on the estimated disturbance load and can be selected as 50kg, 100kg, or 150kg. The initial position is set to the center of the receiving point, and the moving speed is consistent with the real-time belt speed.

[0067] During operation, a moving disturbance source model is established based on the equivalent material disturbance load and the current belt speed. The material disturbance is represented as an equivalent concentrated mass block, which moves downstream from the receiving point at the belt speed. The model calculates its movement distance in real time to track its current position. The high-tension zone includes the drive roller cutting area and the tension concentration section in the middle of the belt conveyor. The drive roller cutting area is defined as the belt section corresponding to the first 1 / 4 arc of the roller surface in contact with the belt. After integrating the moving disturbance source model, the prediction model uses the current position and mass of the equivalent concentrated mass block as time-varying parameters and substitutes them into the segmented dynamic equation and energy consumption equation of the belt conveyor. The optimization objective is modified to minimize the dynamic tension fluctuation variance of the high-tension zone along the path of the moving disturbance source in the prediction time domain. Functional testing was conducted using a long-distance TBM belt conveyor as the experimental object. Strain gauge or fiber optic tension sensors were installed along the line to detect tension changes in high-tension areas. The entire process of an equivalent concentrated mass block moving from the receiving point to the drive drum was simulated and tension fluctuation data were recorded. The mass block mass and belt speed were changed and the test was repeated 5 times to analyze the variance of tension fluctuation.

[0068] By adopting this technical solution, the present invention characterizes material disturbance as an equivalent concentrated mass block that moves with the belt through a moving disturbance source model, tracks its position in real time and incorporates it into a segmented dynamic equation, and optimizes the dynamic tension fluctuation in high-tension areas (such as the drive roller cutting area) in a targeted manner. This can avoid the impact of sudden local tension rise on the belt body and joint during the disturbance movement, reduce the risk of belt breakage, and extend the equipment maintenance cycle.

[0069] In another technical solution, at least one position reference sensor is also included. The position reference sensor is installed near the cutting side of the drive roller or redirecting roller along the belt conveyor and is used to generate a reference trigger signal when the equivalent concentrated mass block passes by.

[0070] The central controller is further configured to perform the following calibration steps:

[0071] The reference trigger signal is used as the absolute position reference;

[0072] The cumulative position error is calculated based on the time deviation between the predicted arrival time of the equivalent lumped mass block and the actual generation time of the reference trigger signal.

[0073] By utilizing the position accumulation error, the current position of the equivalent lumped mass block tracked in the moving disturbance source model is corrected in real time, and the corrected position information is fed back to the prediction model for subsequent rolling optimization calculations.

[0074] The position reference sensor can be a photoelectric, inductive, or Hall effect proximity switch. The housing can be made of aluminum alloy or engineering plastic, and the sensing probe can be made of ceramic or hard alloy. It is mounted near the cutting side of the drive or redirecting roller of the belt conveyor and connected to the frame by welding or bolting via a stainless steel bracket. The distance between the sensor probe and the belt surface is set to 5mm, 7mm, or 10mm. The sensor's trigger response time is ≤10ms, and the correction threshold for cumulative position error is set to ±0.5m.

[0075] During operation, when the equivalent lumped mass block moves with the belt to the vicinity of the sensor, the sensor generates a reference trigger signal and transmits it to the central controller. The central controller uses this signal as the absolute position reference, retrieves the predicted arrival time of the mass block, compares it with the actual trigger time, calculates the time deviation, and then combines it with the current belt speed to obtain the cumulative position error. This error is used to correct the current position of the equivalent lumped mass block in real time, and the corrected position information is fed back to the prediction model for subsequent rolling optimization. The correction cycle is consistent with the control cycle. Functional testing uses a TBM belt conveyor with a position reference sensor as the experimental object. A marker simulating the equivalent lumped mass block is placed on the belt, and the actual position of the marker is recorded using a laser rangefinder. This position is compared with the corrected position tracked by the model. The device is moved 10 times consecutively, and the position deviation is recorded each time.

[0076] By adopting this technical solution, the present invention can correct the cumulative position error of the equivalent lumped mass block through the absolute position trigger signal provided by the position reference sensor, ensuring that the disturbance position tracking does not deviate from the actual path during long-distance belt conveyor operation, making the prediction model's tension optimization in the high-tension zone more in line with the actual working conditions, and reducing control lag or over-adjustment caused by position deviation.

[0077] In another technical solution, the central controller is also configured to execute an adaptive periodic adjustment method, which includes the following steps:

[0078] S1. In each control cycle, the central controller evaluates the severity of the dynamic response of the belt conveyor. The evaluation is based on the current disturbance intensity and the expected future fluctuations. The current disturbance intensity is judged based on the magnitude and rate of change of the load signal transmitted by the load detection unit in the near future. The expected future fluctuations are judged based on the stability of the future speed sequence output by the model predictive control algorithm in the current cycle in the prediction time domain.

[0079] S2. Based on the evaluation results of step S1, the central controller dynamically adjusts the duration of the next and subsequent control cycles. When the evaluation results indicate that the dynamic response of the belt conveyor tends to be more severe, the control cycle is shortened, and the lower limit of the cycle duration is not lower than the lower limit of the sampling cycle (the sampling cycle of the central controller). When the evaluation results indicate that the dynamic response of the belt conveyor tends to be more stable, the control cycle is extended, and the upper limit of the cycle duration does not exceed a preset maximum value.

[0080] S3. When performing the control cycle adjustment in step S2, the central controller performs a gradual cycle switching with a fixed time step.

[0081] The lower limit of the control cycle is 50ms, and the preset maximum value is 500ms. The fixed time step can be selected as 20ms, 30ms, or 50ms, and the time step is set according to the control accuracy requirements. The preset maximum value of the control cycle is set according to the computing power of the central controller to avoid computing overload.

[0082] During operation, within each control cycle, the central controller assesses the severity of the conveyor belt's dynamic response based on the current disturbance intensity and expected future fluctuations. The current disturbance intensity is determined by analyzing the recent amplitude and rate of change in the load signal, while the expected future fluctuations are determined by analyzing the stationarity of the future speed sequence in the predicted time domain. If the assessment results indicate that the dynamic response tends to be severe, the next and subsequent control cycles are shortened to ensure a minimum of 50ms; if it tends to be stable, the control cycle is extended to ensure a minimum of 500ms. Cycle adjustments are made gradually in fixed time steps to avoid control fluctuations caused by abrupt changes. Functional testing uses a running long-distance TBM conveyor belt as the experimental object, simulating different dynamic response scenarios such as stable, moderately severe, and severe. Response changes are triggered by altering load fluctuations, and the adjustment process of the control cycle is recorded and analyzed for timeliness and smoothness.

[0083] By adopting this technical solution, the present invention dynamically adjusts the control cycle by assessing the severity of the dynamic response. When the disturbance is severe, the cycle is shortened to speed up the response, and when the operation is stable, the cycle is extended to reduce the computational load of the controller. At the same time, by gradually switching the cycle with a fixed step size, control oscillations caused by abrupt changes in the cycle can be avoided, thus balancing control accuracy and system operation economy.

[0084] In another technical solution, the method for determining the current disturbance intensity is as follows:

[0085] Within a sliding evaluation time window, the central controller performs real-time analysis on the received load signal sequence and simultaneously calculates the change amplitude index and change rate index. Specifically, the difference between the maximum and minimum values ​​of the load signal within the evaluation time window is calculated to obtain the load fluctuation range as the change amplitude index; the absolute value of the first derivative of the load signal with respect to time within the evaluation time window is calculated, and the maximum value within the evaluation time window is taken as the change rate index.

[0086] The central controller compares the change amplitude indicator with a preset amplitude threshold and the change rate indicator with a preset rate threshold. When either the change amplitude indicator or the change rate indicator exceeds its corresponding threshold, the current disturbance intensity is determined to be high; when neither the change amplitude indicator nor the change rate indicator exceeds its corresponding threshold, the current disturbance intensity is determined to be low.

[0087] The sliding evaluation time window can be selected as 1s, 2s or 3s, and the step size is consistent with the control cycle (50ms-200ms). The amplitude threshold can be selected as 100kg / m, 150kg / m or 200kg / m, and the speed threshold can be selected as 50kg / (m·s), 80kg / (m·s) or 100kg / (m·s). These thresholds are set according to the load-bearing capacity of the belt conveyor and determined based on historical operating data statistics.

[0088] During operation, the central controller sets a sliding evaluation time window, receiving and analyzing load signal sequences in real time within the window. It calculates the difference between the maximum and minimum load signal values ​​within the window to obtain the load fluctuation range as an indicator of the amplitude of change. It also calculates the absolute value of the first derivative of the load signal with respect to time within the window and takes the maximum value as the rate of change indicator. Both indicators are compared with their corresponding preset thresholds. If either indicator exceeds the threshold, the current disturbance intensity is determined to be high; otherwise, it is determined to be low. Functional testing uses the load detection system of a TBM belt conveyor as the experimental object. Simulated load signal sequences including stable, small fluctuations, and large sudden changes are input. The disturbance intensity judgment results of the central controller are recorded, and the consistency of the judgment results with the preset scenarios is compared. The test is repeated multiple times.

[0089] By adopting this technical solution, the present invention analyzes the fluctuation range and rate of change of the load signal through a sliding window, which can accurately distinguish the intensity of the current disturbance, provide an objective basis for the adjustment of the control cycle, avoid inappropriate control strategies caused by overreacting to minor disturbances or underreacting to strong disturbances, and enhance the system's adaptability to dynamic load changes.

[0090] In another technical approach, the method for judging expected future volatility is as follows:

[0091] The central controller performs dynamic characteristic analysis on the future velocity sequence, including continuous fluctuation detection and constraint violation assessment. The central controller identifies all segments in the future velocity sequence that satisfy the condition that multiple consecutive sampling points maintain the same direction of change, and counts the number of segments as a continuous fluctuation index. The central controller calculates the acceleration sequence based on the future velocity sequence and counts the proportion of sampling points in the velocity sequence that violate smoothness constraints.

[0092] The central controller obtains a comprehensive evaluation value based on the continuous fluctuation index and the constraint violation ratio. When the comprehensive evaluation value exceeds a preset evaluation threshold, it is determined that the expected future fluctuation is high; when the comprehensive evaluation value does not exceed the preset evaluation threshold, it is determined that the expected future fluctuation is low.

[0093] Multiple consecutive sampling points can be selected from 3, 5 or 8, and the preset evaluation threshold can be selected from 0.3, 0.5 or 0.7. This threshold is determined through multiple experimental calibrations. The number of consecutive sampling points is set according to the control period and the prediction time domain, and is usually 1 / 10 to 1 / 5 of the total number of sampling points in the prediction time domain.

[0094] During operation, the central controller acquires the future velocity sequence within the prediction time domain and performs dynamic characteristic analysis. It identifies segments in the sequence where multiple consecutive sampling points maintain the same direction of change and counts their number as a persistent fluctuation index. Based on the future velocity sequence, it calculates the acceleration sequence and counts the proportion of sampling points violating smoothness constraints. After normalizing the persistent fluctuation index and constraint violation ratio, it calculates a comprehensive evaluation value with each factor weighted 50%. The comprehensive evaluation value is compared with a preset evaluation threshold; if it exceeds the threshold, the expected future fluctuation is considered high; otherwise, it is considered low. Functional testing uses the central controller's model prediction output module as the experimental object. Future velocity sequences with different levels of stability are input, and the judgment results for future fluctuation expectations are recorded. The accuracy of the judgment results is calculated, and the judgment deviation under different sequence scenarios is analyzed.

[0095] By adopting this technical solution, the present invention can predict future fluctuation trends in advance by analyzing the continuous fluctuation range and constraint violation ratio of future speed sequences, making the control cycle adjustment more forward-looking, preparing for potential fluctuations before they occur, reducing the adjustment pressure when fluctuations actually occur, and improving the predictability and stability of belt conveyor operation.

[0096] In another technical solution, the method for judging the degree of dynamic response is as follows: when the current disturbance intensity is judged to be high, or the expected future fluctuation is judged to be high, the dynamic response degree of the belt conveyor is judged to be drastic; when the current disturbance intensity is judged to be low, and the expected future fluctuation is judged to be low, the dynamic response degree of the belt conveyor is judged to be stable.

[0097] During operation, the central controller receives the judgment results of the current disturbance intensity and the expected future fluctuation. If the current disturbance intensity is high, regardless of the expected future fluctuation, the dynamic response of the belt conveyor is judged to be drastic. If the expected future fluctuation is high and the current disturbance intensity is low, the dynamic response is also judged to be drastic. If the current disturbance intensity is low and the expected future fluctuation is low, the dynamic response is judged to be stable. Functional testing uses the response evaluation module of the central controller as the experimental object. Different combinations of disturbance intensity and expected fluctuation judgment results are input, the judgment results of the dynamic response are recorded, the consistency of the judgment results with the logical rules is verified, and the test is repeated multiple times.

[0098] By adopting this technical solution, the present invention can ensure that the judgment criteria for the operating status of the belt conveyor are consistent under different working conditions by clarifying the correlation rules between the degree of dynamic response and the intensity of disturbance and the expected fluctuation. This avoids the confusion of control strategy caused by the ambiguity of judgment logic, makes the triggering conditions for control cycle adjustment more reliable, and improves the overall stability of system control.

[0099] In another technical solution, the model predictive control algorithm also includes an online parameter estimation module. The online parameter estimation module is used to estimate the dynamic parameters and energy consumption parameters of the belt conveyor in real time and update the prediction model. The online parameter estimation module takes the load signal and speed signal as input, performs parameter estimation at a preset sampling period, and feeds the updated parameters back to the model predictive control algorithm for rolling optimization calculation.

[0100] The online parameter estimation module can be a software module integrated in the central controller, or a dedicated chip such as a DSP chip or FPGA chip can be selected; the sampling period for parameter estimation is 50ms, 100ms or 200ms, the convergence accuracy is set to ±5%, and the initial parameter values ​​are based on the design parameters and factory data of the belt conveyor.

[0101] During operation, the online parameter estimation module takes load signal and speed signal as input and estimates the dynamic parameters (such as inertia coefficient and friction coefficient) and energy consumption parameters (such as motor efficiency coefficient and belt drive loss coefficient) of the belt conveyor in real time according to a set period. The updated parameters are fed back to the model predictive control algorithm in real time. After receiving the updated parameters, the algorithm adjusts the correlation coefficient of the prediction model in a timely manner for the next round of rolling optimization calculation.

[0102] By adopting this technical solution, the present invention updates the dynamics and energy consumption parameters of the belt conveyor in real time through the online parameter estimation module. This can cope with parameter drift caused by wear and environmental changes (such as temperature and humidity) during long-term operation of the belt conveyor, so that the prediction model always matches the actual state of the equipment, ensuring the long-term effectiveness of the control algorithm and reducing the decrease in control accuracy caused by parameter mismatch.

[0103] In another technical solution, the online parameter estimation module uses recursive least squares or extended Kalman filter for parameter estimation; wherein, recursive least squares is used to estimate the inertia coefficient and friction coefficient of the belt conveyor system, and extended Kalman filter is used to estimate the state variables and parameters of the belt conveyor in the presence of measurement noise, and improves the estimation accuracy through adaptive gain adjustment;

[0104] The forgetting factor for the recursive least squares method can be selected as 0.95, 0.98, or 0.99. The process noise variance of the extended Kalman filter can be selected as 1e-4, 1e-3, or 5e-3, and the observation noise variance can be selected as 1e-5, 1e-4, or 5e-4. The forgetting factor is set according to the data update rate; a smaller value is selected when the data changes rapidly. The process noise variance and observation noise variance are set based on the sensor measurement accuracy and system disturbance; a smaller value is selected when the sensor accuracy is high. Both algorithms are software algorithms integrated into the online parameter estimation module. The forgetting factor of the recursive least squares method can be dynamically adjusted according to the load change rate. When the load change rate is >50 kg / (m·s), the forgetting factor is set to 0.95 to focus on the influence of new data; when the load change rate is <10 kg / (m·s), the forgetting factor is set to 0.99 to balance the new and old data, so that the parameter estimation can adapt to different load dynamic characteristics.

[0105] During operation, the online parameter estimation module selects the estimation method according to the working conditions. When the working conditions are stable and the measurement noise is low, the recursive least squares method is used to estimate the inertia coefficient and friction coefficient based on the linear characteristics of the belt conveyor system. The parameter estimates are corrected through iterative calculation, and the forgetting factor balances the weights of new and old data. When the working conditions are complex and there is significant measurement noise, the extended Kalman filter is used to simultaneously estimate the state variables and related parameters, and the noise impact is reduced by adaptively adjusting the gain coefficient.

[0106] By adopting this technical solution, the present invention can efficiently estimate the inertia coefficient and friction coefficient under stable operating conditions through the recursive least squares method, and accurately estimate the state variables and parameters in noisy environments through the extended Kalman filter. The two methods are adapted to different operating scenarios, which can ensure the accuracy of parameter estimation while taking into account the computational efficiency, providing reliable parameter support for the prediction model and improving the robustness of the control algorithm.

[0107] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A long distance belt conveyor control system for TBM construction, characterized by, The application relates to a belt conveyor load control system, which comprises the following parts: a load detection unit, which is installed 5-10 m behind the belt conveyor receiving point, is used for real-time detection of material load and generates a load signal, and the detection range of the load detection unit is 0 kg / m-500 kg / m; a speed detection unit, which is installed at the shaft end of the belt conveyor driving roller, is used for real-time detection of the belt running speed and generates a speed signal, and the detection range of the speed detection unit is 0 m / s-5 m / s; a driving unit, which comprises a frequency converter and a belt conveyor driving motor; a central controller, which is connected with the load detection unit, the speed detection unit and the driving unit through an industrial bus, receives the load signal and the speed signal at a sampling period of 50 ms-200 ms, and takes the sampling period as a control period; the central controller is provided with a model prediction control algorithm, the model prediction control algorithm takes the load signal and the speed signal as input, takes the target torque and the target rotating speed of the belt conveyor driving motor as output, establishes a prediction model containing a belt conveyor dynamics equation and an energy consumption equation, the prediction time domain is set as 3 s-10 s, and the control time domain is set as 1 s-3 s; In each control cycle, the central controller performs the following calculation steps: according to the current load and speed state, the speed sequence and torque sequence in the future period are optimized in the prediction time domain to meet the speed range 1.5m / s-3.5m / s of the TBM out-of-rag rate requirement as the constraint condition, and the absolute value of the belt conveyor acceleration is not more than 0.1m / s 2 -0.3m / s 2 The total energy consumption of the drive unit is minimized as the optimization objective; after the optimization calculation is completed, the target torque and target speed corresponding to the control cycle are output as the control command; the frequency converter is configured to receive the target torque and the target rotating speed instructions sent by the central controller, and adopts a vector control or direct torque control mode to drive the belt conveyor driving motor to run according to the instructions to adjust the belt running speed and the torque; the model prediction control algorithm of the central controller is further integrated with a dynamic disturbance observer, the dynamic disturbance observer takes the driving rotating speed of the TBM as a feedforward signal, and estimates the equivalent material disturbance load introduced by the TBM driving mutation based on the difference between the actual value and the predicted value of the load signal and the speed signal; in the rolling optimization step, the prediction model introduces the equivalent material disturbance load as a time-varying disturbance item into the belt conveyor dynamics equation, and corrects the optimization target to minimize the total energy consumption of the driving unit and minimize the speed fluctuation variance of the belt conveyor; wherein the speed fluctuation variance is calculated based on the prediction model and the equivalent torque disturbance, and is predicted in the prediction time domain; the dynamic disturbance observer is further configured to: establish a moving disturbance source model based on the equivalent material disturbance load and the current running speed of the belt; the moving disturbance source model characterizes the material disturbance introduced by the TBM driving mutation as an equivalent concentrated mass block, the equivalent concentrated mass block moves downstream from the receiving point at the running speed of the belt, and the current position of the equivalent concentrated mass block on the belt conveyor is tracked in real time; in the rolling optimization step of the model prediction control algorithm, the prediction model integrates the moving disturbance source model, introduces the current position and size of the equivalent concentrated mass block as time-varying parameters into the segmented dynamics equation and energy consumption equation of the belt conveyor; wherein the optimization target is further corrected to: in the prediction time domain, the fluctuation variance of the dynamic tension of the high-tension area on the path of the moving disturbance source is minimized, and the high-tension area at least includes the cutting-in area of the driving roller.

2. The long distance belt conveyor control system for TBM construction as claimed in claim 1 wherein, at least one position reference sensor is further included, the position reference sensor is installed near the cutting-in side of the driving roller or the turning roller of the belt conveyor, and a reference trigger signal is generated when the equivalent concentrated mass block passes. The central controller is further configured to perform the following correction steps: Taking the reference trigger signal as the absolute position reference; According to the time deviation between the predicted arrival time of the equivalent lumped mass and the actual generation time of the reference trigger signal, the position cumulative error is calculated; Using the position cumulative error, the current position of the equivalent lumped mass tracked in the moving disturbance source model is corrected in real time, and the corrected position information is fed back to the prediction model for subsequent rolling optimization calculation.

3. The long distance belt conveyor control system for TBM construction as claimed in claim 1 wherein, The central controller is also configured to perform an adaptive cycle adjustment method, which includes the following steps: S1. In each control cycle, the central controller evaluates the degree of dynamic response of the belt conveyor, which is based on the current disturbance intensity and the future fluctuation expectation. The current disturbance intensity is determined according to the change amplitude and rate of the load signal in the near future from the load detection unit. The future fluctuation expectation is determined according to the smoothness of the future speed sequence in the prediction time domain output by the model predictive control algorithm in the current cycle; S2. Based on the evaluation results of step S1, the central controller dynamically adjusts the length of the next and subsequent control cycles. When the evaluation results show that the degree of dynamic response of the belt conveyor tends to be intense, the control cycle is shortened, and the lower limit of the cycle length is not less than the lower limit of the sampling cycle. When the evaluation results show that the degree of dynamic response of the belt conveyor tends to be stable, the control cycle is lengthened, and the upper limit of the cycle length is not more than a preset maximum value; S3. When performing the control cycle adjustment of step S2, the central controller performs gradual cycle switching with a fixed time step.

4. The long distance belt conveyor control system for TBM construction as claimed in claim 3 wherein, The judgment method of the current disturbance intensity is: The central controller performs real-time analysis on the received load signal sequence in a sliding evaluation time window and simultaneously calculates the change amplitude indicator and the change rate indicator. The difference between the maximum value and the minimum value of the load signal in the evaluation time window is calculated to obtain the load fluctuation range as the change amplitude indicator. The absolute value of the first derivative of the load signal with respect to time in the evaluation time window is calculated, and the maximum value in the evaluation time window is taken as the change rate indicator; The central controller compares the change amplitude indicator with a preset amplitude threshold and compares the change rate indicator with a preset rate threshold. When either of the change amplitude indicator and the change rate indicator exceeds its corresponding threshold, it is determined that the current disturbance intensity is high. When neither of the change amplitude indicator and the change rate indicator exceeds its corresponding threshold, it is determined that the current disturbance intensity is low.

5. The long distance belt conveyor control system for TBM construction as claimed in claim 4 wherein, The judgment method of the future fluctuation expectation is: The central controller performs dynamic characteristic analysis on the future speed sequence, including continuous fluctuation detection and constraint violation evaluation. The central controller identifies all sections in the future speed sequence that satisfy the condition of maintaining the same change direction for a plurality of consecutive sampling points, and counts the number of sections as the continuous fluctuation indicator. The central controller calculates the acceleration sequence from the future speed sequence and counts the proportion of sampling points that violate the smoothness constraint in the speed sequence; The central controller obtains a comprehensive evaluation value according to the continuous fluctuation indicator and the constraint violation proportion. When the comprehensive evaluation value exceeds a preset evaluation threshold, it is determined that the future fluctuation expectation is high; When the comprehensive evaluation value does not exceed the preset evaluation threshold, it is determined that the future fluctuation expectation is low.

6. The long distance belt conveyor control system for TBM construction as claimed in claim 5 wherein, The judgment method of the dynamic response degree is that when it is determined that the current disturbance intensity is high or the future fluctuation expectation is high, it is determined that the dynamic response degree of the belt conveyor tends to be intense. When it is determined that the current disturbance intensity is low and the future fluctuation expectation is low, it is determined that the dynamic response degree of the belt conveyor tends to be stable.

7. The long distance belt conveyor control system for TBM construction as claimed in claim 1 wherein, The model predictive control algorithm further comprises an online parameter estimation module, which is used to estimate the dynamic parameters and energy consumption parameters of the belt conveyor in real time and update the prediction model. The online parameter estimation module takes the load signal and the speed signal as inputs, performs parameter estimation at a preset sampling period, and feeds back the updated parameters to the model predictive control algorithm for rolling optimization calculation.

8. The long distance belt conveyor control system for TBM construction as claimed in claim 7 wherein, The online parameter estimation module uses a recursive least square method or an extended Kalman filter to perform parameter estimation; wherein the recursive least square method is used to estimate the inertia coefficient and the friction coefficient of the belt conveyor system, and the extended Kalman filter is used to estimate the state variables and parameters of the belt conveyor when there is measurement noise, and to improve the estimation accuracy through adaptive gain adjustment.

Images