Adaptive control system and method for cutting transmission system of coal cutter under complex working condition

Abstract

Disclosed in the present invention are an adaptive control system and method for a cutting transmission system of a coal cutter under complex working conditions. An extended state observer in an adaptive torque compensation controller tracks state variables of a cutting transmission system, and estimates in real time disturbances of the system such as nonlinearity and time-varying parameters, so as to compensate a real-time action amount of a total disturbance into an actual control amount for feedforward compensation on disturbance terms; a BP network adaptively adjusts parameters of a second-order active disturbance rejection torque controller; an adaptive convolutional neural network fuzzy module realizes an adaptive adjustment for an output compensation of the active disturbance rejection torque compensation controller, so as to obtain an expert control rule that meets complex impact conditions. Thus, a nonlinear state error feedback and the adjustment amount of the adaptive convolutional neural network fuzzy module are combined to calculate an initial electromagnetic torque; and after the feedforward compensation, a compensated electromagnetic torque is generated and is superimposed with a torque compensation amount generated by a PI controller, thereby achieving an adaptive compensation control for dynamic load suppression of the system and effective suppression for a dynamic load impact energy of the cutting transmission system.

Description

Self-adaptive control system and method of shearer cutting transmission system under complex working conditions TECHNICAL FIELD

[0001] The present application belongs to the technical field of shearer control, and particularly relates to a self-adaptive control system and method of shearer cutting transmission system under complex working conditions. BACKGROUND

[0002] Coal is the basic energy of China, and coal industry is an important basic industry related to the economic lifeline and energy security of the country. In the structure of primary energy in China, coal will still be the main energy and important industrial raw material for a long time. In the development goals of the "Thirteenth Five-Year Plan" for the development of coal industry released by the National Development and Reform Commission on December 22, 2016, it is pointed out that by 2020, the long-term mechanism for safety production of coal industry will be further improved, the number of deaths in coal mine accidents will be reduced by more than 15% compared with 2015, and the death rate per million tons will be reduced by more than 15% compared with 2015; in addition, the degree of mechanization of coal mining in coal mines will reach 85%, the degree of mechanization of tunneling will reach 65%, the contribution rate of technological innovation to the development of the industry will be further improved, new progress will be made in the construction of information and intelligence in coal mines, and a number of advanced and efficient intelligent coal mines will be built. Therefore, higher requirements are put forward for the mechanization, automation and working condition adaptability of coal mining equipment.

[0003] Due to the complexity of underground coal seams, the coal seams contain high-strength rock partings, hard inclusions and rock faults, etc., so that the external load acting on the drum has the characteristics of randomness and strong impact, which easily leads to an increase in the dynamic load of the cutting transmission system, and the cutting part of the shearer becomes one of the weak parts of the whole machine. Due to the complexity of the cutting environment, it is an inevitable requirement for the automation and unmanned operation of the shearer to realize reliable operation and efficient production of the shearer. When the properties of the coal seam change, the adjustable motion parameters are the drum height and the traction speed, but the existing technology does not realize the speed regulation of the cutting drum; under different cutting conditions, when the traction speed changes in a large range, it is difficult to ensure that the shearer is in the best working state in terms of comprehensive performance such as strong cutting ability, small cutting specific energy consumption and high production rate. Therefore, under the condition of unmanned operation of the shearer, it is necessary to realize the optimal comprehensive performance of the shearer under different cutting conditions by means of variable speed cutting of the shearer drum and coordinated control of the cutting-traction motion parameters.

[0004] At present, scholars in the field of wind power, rolling mill and other fields have reduced the fatigue load of electromechanical transmission system caused by external load mutation through active control, that is, by designing a damping controller to superimpose a compensation torque on the basis of the original generator / motor torque to increase the electrical damping of the transmission chain. Some scholars have established a torsional vibration damping controller based on a band-pass filter (BPF) to suppress the impact load of the wind power gear transmission system caused by external load mutation and avoid resonance of the transmission chain caused by torsional vibration. A system fatigue load active control strategy based on BPF is proposed to suppress multiple resonance frequencies of the system with generator speed as the feedback signal. However, when the controlled object has model parameter disturbance and uncertainty factors, the stability of the damping controller based on BPF and its suppression performance on the fatigue load of the transmission system will be affected. In view of the shortcomings of the above controller, some scholars have established a damping controller based on the model of the controlled object with generator speed as the feedback signal, and a Kalman filter is used to estimate the aerodynamic torque load of the system. However, this method relies on relatively accurate transmission system parameters, and in fact the electromechanical transmission system is complex and the parameters are time-varying, so it is difficult to obtain accurate system parameters. In order to solve the above problems, some scholars have applied active disturbance rejection control technology, which does not rely on the accurate mathematical model of the controlled object and can automatically compensate for system parameter disturbances, to the field of torsional vibration suppression of rolling mill and wind power transmission system. However, these studies simplify the complex transmission system as an inertia mass model when modeling, only consider the torsional vibration of the shaft system, and do not establish a dynamic model including gears, which cannot consider the time-varying meshing stiffness of gears and the dynamic load of gear transmission system, and cannot analyze the influence of time-varying parameters on the performance of active disturbance rejection controller.

[0005] SUMMARY

[0006] The present application provides a coal mining machine cutting transmission system adaptive control system and method under complex working conditions, which can effectively suppress the dynamic load impact energy under complex cutting working conditions and realize the control method and control system of system parameter adaptive adjustment, so as to improve the safety and production efficiency of the coal mining machine under complex cutting working conditions.

[0007] To solve the above technical problems, the present application provides the following technical scheme: a coal mining machine cutting transmission system adaptive control system under complex working conditions, comprising: a sample construction module, an adaptive torque compensation controller, a superposition module, and a cutting motor speed adjustment module.

[0008] The sample construction module is used to calculate the difference between the actual speed of the cutting motor and the speed of the high-speed gear of the cutting transmission system, and construct a training sample data set based on the speed difference and the corresponding controller output compensation torque.

[0009] The adaptive torque compensation controller is configured to output a controller output compensation torque, and the adaptive torque compensation control neural network is trained using a sample data set, and a dynamic adjustment factor is introduced in the training process to dynamically adjust the controller output.

[0010] The second-order nonlinear active disturbance rejection torque module includes a tracking differentiator, an extended state observer, and a nonlinear error feedback law module; the second-order nonlinear active disturbance rejection torque module introduces a dynamic adjustment factor to enhance the adaptability and robustness of the active disturbance rejection controller, and the dynamic adjustment factor dynamically adjusts the control output according to the running state of the system and external disturbances; the adaptive convolutional neural network fuzzy module is configured to improve the working condition adaptability of the second-order nonlinear active disturbance rejection torque controller, and obtain expert control rules of the cutting transmission system under different coal rock impacts, and generate an adaptive adjustment amount ΔT according to changes in the working condition e ; the BP neural network optimizes the parameters of the NLSEF and the ESO in the second-order nonlinear active disturbance rejection torque controller;

[0011] The superposition module superimposes the controller output compensation torque and the PI controller output torque to obtain a torque sum.

[0012] The cutting motor speed adjustment module takes the torque sum as the motor controller input to obtain a motor controller output, and adjusts the cutting motor speed in real time according to the output.

[0013] Further, the tracking differentiator is configured to perform the following actions: arranging a transition process for an input signal:

[0014] In the formula, T is a sampling step; r is a speed factor; d = rh0; h0 is a filter factor, ω 11 is the input signal after transition, ω 22 is the differential of ω 11 , and fst is the fastest control synthesis function.

[0015] The extended state observer is configured to perform the following actions:

[0016] In the formula, y(k) is the system output; e(k) is the difference between the extended state observer observation output and the system output; β1, β2, and β3 are constants; α1, α2, and α3 are nonlinear parameters, all of which are positive

[0017] The nonlinear error feedback control law module is configured to perform the following actions:

[0018] In the formula, β4 and β5 are proportional factor and differential factor, respectively.

[0019] Further, the aforementioned dynamic adjustment factor includes a change rate factor a, a disturbance factor β, and an error factor γ; wherein γ = |e|, e is a control error, the change rate factor a, the disturbance factor β, and the error factor γ are weighted and fused to obtain a comprehensive adjustment factor λ, λ = s1a + s2β + s3γ, wherein s1, s2, and s3 are weighting coefficients.

[0020] Further, the aforementioned second-order nonlinear active-disturbance-rejection torque module compensates the output torque as follows: e2 = (1 + λ) [β4fal (e1 (t), a4, δ) + β5fal (e2 (t), a5, δ) - z3 / b0].

[0021] Further, the aforementioned BP neural network includes an input layer, a hidden layer, and an output layer; the input layer includes four neural nodes to track the differentiator output signals e1 and e2, the control system total output y, and 1 as neuron inputs; the hidden layer is determined to be six in combination with the cutting transmission system; and the output layer nodes are six, corresponding to the second-order nonlinear active-disturbance-rejection torque controller parameters β1, β2, β3, β4, and β5, respectively.

[0022] Further, the aforementioned adaptive convolutional neural network fuzzy module adopts an adaptive neuro-fuzzy control system ANFIS combined with a multi-layer convolutional neural network structure, specifically, a multi-layer convolutional neural network CNN is introduced in front of the input layer of the adaptive neuro-fuzzy control system ANFIS, the adaptive convolutional neural network fuzzy module is used to improve the working condition adaptive ability of the second-order nonlinear active-disturbance-rejection torque controller, and expert control rules of the cutting transmission system under different coal rock impacts are obtained, and an adaptive adjustment amount ΔT e is generated according to the working condition change.

[0023] The adaptive convolutional neural network fuzzy module includes a data input module, which converts the output signals of the tracking differentiator into a format suitable for CNN processing and performs normalization and standardization processing to form a CNN time slice;

[0024] The multi-layer convolutional neural network CNN includes multiple convolutional layers, activation function layers, and pooling layers, extracts high-level features of the input data after the output signals of the tracking differentiator are processed by multiple layers, and finally outputs a feature vector through a fully connected layer; and a feature fusion module fuses the features extracted by the multi-layer convolutional neural network CNN with other input features of the adaptive neuro-fuzzy control system ANFIS, and inputs them to the adaptive neuro-fuzzy control system ANFIS for adaptive fuzzy reasoning and decision-making.

[0025] Further, the aforementioned adaptive neuro-fuzzy control system ANFIS comprises an input layer, a fuzzification layer, a fuzzy rule layer, a fuzzy inference layer, a defuzzification layer;

[0026] The input layer accepts the fused feature vector, and defines the fused feature vector as input of the ANFIS as x i = [e1(k), e2(k)] T ;

[0027] The fuzzification layer assigns a linguistic variable value to each node through fuzzy inference rules, and calculates the membership function of each input component belonging to the fuzzy set of each linguistic variable value The membership function adopts a Gaussian function, and its formula is In the formula, i = 1, 2; j = 1, 2, …, m j , m j is the fuzzy partition number of x i ; c ij and δ ij are the center and width of the membership function, respectively;

[0028] The fuzzy rule layer is used to define fuzzy rules, that is In the formula, i1∈{1, 2, …, m1}, i1∈{1, 2, …, m1}, …, i1∈{1, 2, …, m1}, j = 1, 2, …, m, fuzzy inference is performed;

[0029] The fuzzy inference layer combines the fuzzy rules, calculates the applicability of each rule, and performs normalization calculation, that is

[0030] The defuzzification layer converts the fuzzy inference result into an explicit control output In the formula, w ij is the connection weight value between the fourth layer node and the output layer node.

[0031] Further, when the aforementioned adaptive neuro-fuzzy control system ANFIS performs adaptive fuzzy inference and decision-making, a performance index function is defined as In the formula, t i is the target output; the gradient descent method is used to calculate Then, a first-order gradient optimization algorithm is used to adjust w ij , c ij , δ ij , and the first-order gradient to be solved is:

[0032] The adaptive fuzzy inference parameter learning algorithm is:

[0033] 9. The self-adaptive control method of the shearer cutting transmission system under complex conditions according to claim 8, characterized in that the self-adaptive convolutional neural network fuzzy module is configured to perform the following actions: tra = -1 x (T me - T ste ), wherein T me is the motor output electromagnetic torque, T ste is the torque opposite to the direction of the motor output electromagnetic torque, and T tra is the motor electromagnetic torque steady-state target value; the adaptive convolutional neural network can be obtained by continuously training the input feature vector of the controller, and the adaptive convolutional neural network is used to compensate the electromagnetic torque adjustment amount ΔT e .

[0034] The final output torque of the adaptive torque compensation controller is: T e2 = (1 + λ) [β4fal(e1(t), a4, δ) + β5fal(e2(t), a5, δ) + ΔT e ]-z3 / b0

[0035] The superposition module superimposes the PI control output torque Te1;

[0036] The cutting motor speed adjustment module outputs the motor control total output compensation torque of the shearer self-adaptive control system: T eM = Te1+ (1 + λ) [β4fal(e1(t), a4, δ) + β5fal(e2(t), a5, δ) + ΔT e ]-z3 / b0.

[0037] The application also provides a control method based on the self-adaptive control system of the shearer cutting transmission system under complex conditions, which comprises the following steps:

[0038] S1, determining whether the shearer drum cutting impedance changes, if yes, determining that it is a full coal condition, executing step S2, otherwise, continuing to determine whether it is a parting condition, if yes, executing step S2, otherwise, determining that it is a fault condition, and executing step S2;

[0039] S2, determining the motor speed and impedance, inputting the signals to the adaptive torque compensation controller, and compensating the electromagnetic torque in real time according to the signals, outputting the electromagnetic torque to adjust the cutting motor speed, the traction speed and the drum speed;

[0040] S3, determining whether the cutting impedance changes, if yes, executing step S2, otherwise, ending.

[0041] Compared with the prior art, the application has the following beneficial technical effects by adopting the above technical scheme:

[0042] 1. The application realizes the application of active disturbance rejection technology in the cutting transmission system of a coal mining machine, overcomes the problem that it is difficult for a traditional direct torque control to quickly and effectively suppress the dynamic load of the cutting transmission system of the coal mining machine caused by impact load under the poor working conditions of the coal mining machine in deep coal seams. A dynamic adjustment factor is introduced into the second-order nonlinear active disturbance rejection torque compensation controller to improve the controller, so that the compensation torque can more accurately adapt to the changes in actual working conditions, the dynamic adjustment factor can automatically adjust the compensation strength according to real-time working conditions, the response speed and control accuracy of the system are improved, and the impact energy of the dynamic load is more effectively suppressed. The control method does not depend on the mathematical model of the controlled object, compared with the traditional direct torque control method, the control method not only has good tracking and estimation performance, but also is not sensitive to time-varying parameters of the system, and can reduce the influence of time-varying parameters on the second-order nonlinear active disturbance rejection torque controller.

[0043] 2. The application introduces a multi-layer convolutional neural network (CNN) before an adaptive neural-fuzzy control system (ANFIS) to perform hierarchical feature extraction on input data. CNN is good at processing high-dimensional data and can extract multi-level features and more accurately capture complex nonlinear relationships. By introducing a multi-layer convolutional neural network (CNN) after the input layer of the ANFIS, the learning ability and generalization ability of the system can be improved, the feature extraction ability of the system can be enhanced, and the processing effect of complex input data can be improved.

[0044] 3. The application adopts a control method combining an adaptive convolutional neural network fuzzy controller and a second-order nonlinear active disturbance rejection torque compensation controller, which solves the problem of adaptive adjustment of the output torque compensation of the second-order nonlinear active disturbance rejection torque compensation controller under complex working conditions. By combining the advantages of multi-layer convolutional neural network control and fuzzy control, adaptive adjustment of the output compensation of the active disturbance rejection torque compensation controller is realized, and expert control rules that meet complex impact working conditions are obtained. The adaptive adjustment amount can be generated according to the change of the working condition, so as to realize adaptive compensation control of the system dynamic load suppression. BRIEF DESCRIPTION OF DRAWINGS

[0045] Fig. 1 is a schematic diagram of an adaptive control system of a coal mining machine under complex working conditions.

[0046] Fig. 2 is a structure diagram of an adaptive torque compensation controller.

[0047] Fig. 3 is a structure diagram of a second-order nonlinear active disturbance rejection torque compensation controller.

[0048] Fig. 4 is a structure diagram of a BP neural network.

[0049] Fig. 5 is a structure diagram of an adaptive convolutional neural network fuzzy controller.

[0050] Fig. 6 is a structure diagram of an adaptive neural network-fuzzy control module.

[0051] Figure 7 is a flow chart of the self-adaptive control system of the coal mining machine under complex working conditions. DETAILED DESCRIPTION

[0052] In order to better understand the technical content of the present application, specific embodiments are described below with reference to the accompanying drawings.

[0053] Aspects of the present application are described in this patent document in relation to the accompanying drawings that illustrate a number of illustrative embodiments. The embodiments of the present application are not limited to the drawings described. It should be understood that the present application is realized by any one of the above-described various concepts and embodiments, and the concepts and embodiments disclosed below in detail, and the disclosed concepts and embodiments are not limited to any implementation. In addition, some aspects disclosed by the present application can be used alone or in any appropriate combination with other aspects disclosed by the present application.

[0054] As shown in Figure 1, it is a schematic diagram of the self-adaptive control system of the coal mining machine under complex working conditions. This control method collects the rotational speed ω1 of the flexible shaft end of the cutting motor m and the rotational speed ω1 of the gear transmission system flexible shaft end, and the difference Δω1 is used as the input quantity of the adaptive torque compensation controller. The output quantity Te2 of the adaptive torque compensation controller is superimposed with the output quantity Te1 of the PI controller to form the total compensation torque Te of the motor control. m The rotational speed of the cutting motor is adjusted in real time. Compared with the traditional direct torque control method, this control method not only has good tracking and estimation performance, but also is not sensitive to time-varying parameters of the system, and can more effectively suppress the impact energy of external dynamic load under complex cutting environment.

[0055] The present application provides a self-adaptive control system of the cutting transmission system of the coal mining machine under complex working conditions, comprising: a sample construction module, an adaptive torque compensation controller, a superposition module, and a cutting motor speed adjustment module.

[0056] The sample construction module is used to calculate the difference between the actual rotational speed of the cutting motor and the rotational speed of the high-speed gear of the cutting transmission system. Based on the rotational speed difference and the corresponding controller output compensation torque, a training sample data set is constructed.

[0057] As shown in Figure 2, it is a structure diagram of the adaptive torque compensation controller. The adaptive torque compensation controller comprises a second-order nonlinear self-disturbance torque module, an adaptive convolutional neural network fuzzy module, and a BP neural network. The controller is used to output the compensation torque, and the adaptive torque compensation control neural network is trained by using the sample data set. A dynamic adjustment factor is introduced in the training process to dynamically adjust the output of the controller.

[0058] The second-order nonlinear active-disturbance-rejection torque module includes a tracking differentiator, an extended state observer, and a nonlinear error feedback law module; the second-order nonlinear active-disturbance-rejection torque module introduces a dynamic adjustment factor to enhance the adaptability and robustness of the active-disturbance-rejection controller, and the dynamic adjustment factor dynamically adjusts the control output according to the running state of the system and external disturbance; an adaptive convolutional neural network fuzzy module is used to improve the working condition adaptability of the second-order nonlinear active-disturbance-rejection torque controller, and obtain expert control rules of the cutting transmission system under different coal rock impacts, and generate an adaptive adjustment amount ΔT according to the change of the working condition e ; a BP neural network is used to optimize the parameters of the NLSEF and the ESO in the second-order nonlinear active-disturbance-rejection torque controller;

[0059] The superposition module superimposes the controller output compensation torque on the PI controller output torque to obtain the sum of the torques.

[0060] The cutting motor speed adjustment module takes the sum of the torques as the motor controller input, obtains the motor controller output, and adjusts the cutting motor speed in real time according to the output.

[0061] As shown in FIG. 3, the structure diagram of the second-order nonlinear active-disturbance-rejection torque compensation controller, wherein the tracking differential controller (TD) is arranged to perform the following actions:

[0062] In the formula, T is the sampling step; r is the speed factor; d = rh0; h0 is the filter factor. ω 11 is the input signal after transition, and ω 22 is the differential of ω 11 . fst is the fastest control synthesis function.

[0063] The extended state observer (ESO) is configured to perform the following actions:

[0064] In the formula, y(k) is the system output; e(k) is the difference between the observation output of the extended state observer and the system output; β1, β2, and β3 are constants; α1, α2, and α3 are nonlinear parameters, all of which are positive values; the extended state observer is established based on the measurement output and the control input of the cutting transmission system, estimates the system and external disturbance, and compensates the torque in real time through the disturbance estimation value.

[0065] The nonlinear error feedback law module (NLSEF) is configured to perform the following actions:

[0066] In the formula, β4 and β5 are proportional factor and differential factor, respectively.

[0067] The introduced dynamic adjustment factor includes a change rate factor α, a disturbance factor β, and an error factor γ; wherein γ = |e|, e is control error. The above factors are weighted and fused to form a comprehensive adjustment factor λ, λ = s1α + s2β + s3γ, wherein s1, s2, s3 are weighting coefficients.

[0068] The second-order nonlinear active-disturbance-rejection torque controller adopts a disturbance compensation method, and compensates the real-time action amount of the total disturbance to the actual control amount. The compensation output torque of the second-order nonlinear active-disturbance-rejection torque controller after disturbance compensation is: e2 = (1 + λ) [β4fal (e1 (t), a4, δ) + β5fal (e2 (t), a5, δ) - z3 / b0]

[0069] As shown in FIG. 4, it is a BP neural network structure diagram. Since the parameters in the second-order nonlinear active-disturbance-rejection torque controller are too many and fixed, the torque compensation output in different cutting conditions is limited. The purpose of introducing the BP neural network is to optimize the parameters in the NLSEF and ESO of the active-disturbance-rejection controller. This BP neural network adopts a 3-layer BP neural network, including an input layer, a hidden layer and an output layer. The input layer adopts 4 neural nodes to track the differentiator output signals e1 and its differential e2, the total output y of the control system and 1 as the neuron input. The hidden layer is determined to be 6 in combination with the cutting transmission system. The output layer nodes are 5, respectively corresponding to the parameters β1, β2, β3, β4 and β5 of the second-order nonlinear active-disturbance-rejection torque controller. The BP neural network controller adjusts the parameters of the NLSEF and ESO according to the changes of the controlled object and external disturbances, which helps to improve the state estimation accuracy, improve the system robustness and improve the control effect.

[0070] As shown in Figure 5, it is a structure diagram of adaptive convolutional neural network fuzzy controller. In order to further improve the working condition adaptive ability of the second-order nonlinear active-disturbance-rejection torque controller, this adaptive convolutional neural network fuzzy module is designed. This controller adopts the structure of adaptive neural-fuzzy control system (ANFIS) combined with multi-layer convolutional neural network (CNN). CNN is good at processing high-dimensional data and can extract multi-level features and more accurately capture complex nonlinear relationships. By introducing multi-layer convolutional neural network (CNN) after the input layer of ANFIS, the learning ability and generalization ability of the system can be improved. In combination with Figure 5, the specific structure of the adaptive convolutional neural network fuzzy controller includes: a data input module, which converts the output signal of the tracking differentiator into a format suitable for CNN processing and performs normalization and standardization processing to form a CNN time slice; a multi-layer convolutional neural network (CNN) module, which includes multiple convolutional layers, activation function layers and pooling layers. After multi-layer processing, high-level features of the input data are extracted, and finally a feature vector is output through a fully connected layer. A feature fusion module fuses the features extracted by CNN with other input features of ANFIS and inputs them to ANFIS for fuzzy reasoning and decision-making. An adaptive neural-fuzzy reasoning module.

[0071] As shown in Figure 6, it is a structure diagram of adaptive neural network-fuzzy control module, which includes an input layer that accepts the fused feature vector, defines the fused feature vector x i = [e1(k), e2(k)] T of ANFIS input, a fuzzification layer that gives each node a linguistic variable value through fuzzy reasoning rules and calculates the membership function of each input component belonging to the fuzzy set of each linguistic variable value The membership function is represented by a Gaussian function, and its formula is where i = 1, 2; j = 1, 2,..., m j , m j is the fuzzy partition number of x i ; c ij and δ ij are the center and width of the membership function, respectively. A fuzzy rule layer defines fuzzy rules, i.e. where i1 ∈ {1, 2,..., m1}, i1 ∈ {1, 2,..., m1}, …, i1 ∈ {1, 2,..., m1}, j = 1, 2, …, m, performs fuzzy reasoning; a fuzzy reasoning layer calculates the applicability of each rule in combination with the fuzzy rules and performs normalization calculation, i.e. (j = 1, 2,..., m). A defuzzification layer converts the fuzzy reasoning result into a clear control output where w ij is the connection weight value between the fourth layer node and the output layer node.

[0072] In the adaptive neural-fuzzy inference module, a performance index function is defined. In the formula, t i Output the target value. Calculate using gradient descent. Then, the first-order gradient optimization algorithm is used to adjust w. ij c ij δ ij The first-order gradient is:

[0073] The adaptive neural-fuzzy inference module parameter learning algorithm can be obtained as follows:

[0074] To obtain expert control rules under different operating conditions, the adaptive convolutional neural network fuzzy controller training algorithm is defined as: T tra =-1×(T) me -T ste ), where T me T is the electromagnetic torque output by the motor itself. ste T is the torque that is opposite in direction to the electromagnetic torque output by the motor. tra Let ΔT be the steady-state target value of the motor's electromagnetic torque. By continuously training with the controller's input feature vector, an adaptive convolutional neural network fuzzy controller can be obtained to compensate for the electromagnetic torque adjustment ΔT in the control system. e .

[0075] The final output torque of the adaptive torque compensation controller is: T e2 =(1+λ)[β4fal(e1(t),a4,δ)+β5fal(e2(t),a5,δ)+ΔT e ]-z3 / b0

[0076] The total output compensation torque of the coal mining machine adaptive control system is: T, which is superimposed with the PI control output torque Te1. eM =Te1+(1+λ)[β4fal(e1(t),a4,δ)+β5fal(e2(t),a5,δ)+ΔT e ]-z3 / b0

[0077] Referring to Figure 7. On the other hand, the present invention also provides a control method based on an adaptive control system for a coal mining machine cutting transmission system under complex working conditions, comprising the following steps:

[0078] S1. Determine whether the cutting resistance of the coal mining machine drum changes. If yes, it is determined to be a full coal working condition and proceeds to step S2. Otherwise, continue to determine whether it is a rock interbedded working condition. If yes, proceed to step S2. Otherwise, it is determined to be a fault working condition and proceeds to step S2.

[0079] S2, determine the motor speed and impedance, input the signal to the adaptive torque compensation controller, the controller compensates the electromagnetic torque in real time according to the signal, outputs the electromagnetic torque to adjust the cutting motor speed, the traction speed and the roller speed;

[0080] S3, judge whether the cutting impedance changes, if yes, execute step S2, otherwise, end.

[0081] Although the present application has been described in connection with the preferred embodiment thereof with reference to the drawings, it is not intended to be limited to the embodiment but it will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and scope of the application. Accordingly, the scope of the present application should be governed by the appended claims rather than by the description.

Claims

1. A self-adaptive control system for a shearer cutting transmission system under complex working conditions, characterized in that, The application relates to a self-adaptive torque compensation controller for a coal winning machine. The sample construction module is used for calculating the difference between the actual rotating speed of a cutting motor and the rotating speed of a high-speed gear of a cutting transmission system; and constructing a training sample data set based on the rotating speed difference and a corresponding controller output compensation torque. The self-adaptive torque compensation controller is used for outputting a controller output compensation torque, training a self-adaptive torque compensation control neural network by using the sample data set, and introducing a dynamic adjustment factor in the training process to dynamically adjust the controller output. The self-adaptive torque compensation controller comprises a second-order nonlinear self-disturbance resisting torque module, an adaptive convolutional neural network fuzzy module and a BP neural network. The superposition module superimposes the controller output compensation torque and a PI controller output torque to obtain a torque sum. The second-order nonlinear active-disturbance-rejection torque module comprises a tracking differentiator, an extended state observer and a nonlinear error feedback law module; the second-order nonlinear active-disturbance-rejection torque module introduces a dynamic adjustment factor to enhance the adaptability and robustness of the active-disturbance-rejection controller, and the dynamic adjustment factor dynamically adjusts the control output according to the running state of the system and external disturbance; the adaptive convolutional neural network fuzzy module is used to improve the working condition adaptability of the second-order nonlinear active-disturbance-rejection torque controller, and obtain expert control rules of the cutting transmission system under different coal rock impacts, and generate an adaptive adjustment amount ΔT according to the change of the working condition e ; the BP neural network optimizes the parameters of the NLSEF and the ESO in the second-order nonlinear active-disturbance-rejection torque controller; The cutting motor rotating speed adjustment module takes the torque sum as a motor controller input to obtain a motor controller output quantity, and adjusts the rotating speed of the cutting motor in real time according to the output quantity. In the formula, y (k) is the system output; e (k) is the difference between the observation output of an extended state observer and the system output; beta1, beta2 and beta3 are constants; alpha1, alpha2 and alpha3 are nonlinear parameters, all of which are positive values, 2. The self-adaptive control system of a shearer cutting transmission system under complex conditions according to claim 1, characterized in that, The tracking differentiator is configured to perform the following actions for the input signal to schedule a transition: where T is the sampling step; r is the velocity factor; d = rh0; h0is the filter factor, ω 11 is the input signal after transition, ω 22 is the differential of ω 11 ; fstis the fastest control synthesis function; The expansion state observer is configured to perform the following actions: In the formula, beta4 and beta5 are proportional factor and differential factor respectively. The nonlinear error feedback control law module is configured to perform the following actions: The BP neural network comprises an input layer, a hidden layer and an output layer; the input layer comprises four neural nodes to track the differential output signal e1, the differential e2 of the differential output signal, the total output y of the control system and 1 as the neuron inputs; the hidden layer is determined as six in combination with the cutting transmission system; and the output layer nodes are six, which correspond to the second-order nonlinear self-disturbance resisting torque controller parameters beta1, beta2, beta3, beta4 and beta5 respectively.

3. The self-adaptive control system of a shearer cutting transmission system under complex conditions according to claim 2, characterized in that, The dynamic adjustment factor includes a change rate factor, an alpha, a disturbance factor beta, and an error factor gamma; wherein γ = |e|, e is a control error, the change rate factor alpha, the disturbance factor beta, and the error factor gamma are weighted and fused to obtain a comprehensive adjustment factor lambda, lambda = s1alpha + s2beta + s3gamma, wherein s1, s2, and s3 are weighting coefficients.

4. The self-adaptive control method for the cutting transmission system of a coal mining machine under complex conditions according to claim 3, characterized in that, The second-order nonlinear active-disturbance-rejection torque module compensates the output torque as follows: e2 = (1 + λ) [β4fal(e1(t), a4, δ) + β5fal(e2(t), a5, δ) - z3 / b0].

5. The self-adaptive control method for the cutting transmission system of a coal mining machine under complex conditions according to claim 4, characterized in that, The adaptive convolutional neural network fuzzy module comprises a data input module which converts the output signal of the tracking differentiator into a format suitable for CNN processing and performs normalization and standardization processing to form a CNN time sequence slice.

6. The self-adaptive control system of a shearer cutting transmission system under complex conditions according to claim 5, characterized in that, The adaptive convolutional neural network fuzzy module adopts an adaptive neural-fuzzy control system ANFIS combined with a multi-layer convolutional neural network structure, specifically, a multi-layer convolutional neural network CNN is introduced before an input layer of the adaptive neural-fuzzy control system ANFIS, the adaptive convolutional neural network fuzzy module is used to improve the working condition adaptive capability of the second-order nonlinear active-disturbance-rejection torque controller, and expert control rules of a cutting transmission system under different coal rock impacts are obtained, and an adaptive adjustment amount ΔT is generated according to working condition changes e ; The multi-layer convolutional neural network CNN comprises a plurality of convolutional layers, activation function layers and pooling layers, extracts high-level features of the input data after the output signal of the tracking differentiator is processed by the multi-layer, and finally outputs a feature vector through a full connection layer; and a feature fusion module fuses the features extracted by the multi-layer convolutional neural network CNN and other input features of the adaptive neuro-fuzzy control system ANFIS, and inputs the features to the adaptive neuro-fuzzy control system ANFIS for adaptive fuzzy reasoning and decision-making. The adaptive neuro-fuzzy control system ANFIS comprises an input layer, a fuzzification layer, a fuzzy rule layer, a fuzzy reasoning layer and a defuzzification layer.

7. The self-adaptive control system of a shearer cutting transmission system under complex conditions according to claim 6, characterized in that, The adaptive neuro-fuzzy control system ANFIS performs fuzzy reasoning. The input layer accepts the fused feature vector, and the fused feature vector input by the ANFIS is defined as x i = [e1(k), e2(k)] T ; The fuzzy layer assigns a linguistic variable value to each node by fuzzy inference rules, and calculates the membership function of each input component belonging to the fuzzy set of each linguistic variable value The membership function is expressed by a Gaussian function, and its formula is where i = 1, 2; j = 1, 2,..., m j , m j is the fuzzy partition number of x i ; c ij and δ ij are the center and width of the membership function, respectively. The fuzzy rule layer is configured to define fuzzy rules, i.e. wherein i1∈ {1,2,...,m1}, i1∈ {1,2,...,m1},..., i1∈ {1,2,...,m1}, j = 1,2,...,m, The final output torque of the self-adaptive torque compensation controller is: The fuzzy inference layer combines the fuzzy rules, calculates the applicability of each rule, and performs normalization calculation, i.e. The de-obfuscation layer converts the obfuscated inference result into a clear control output In the formula, w ij is the connection weight of the fourth layer node and the output layer node.

8. The self-adaptive control method for the cutting transmission system of a coal mining machine under complex conditions according to claim 7, characterized in that, When the adaptive neuro-fuzzy control system ANFIS performs adaptive fuzzy reasoning and decision, a performance index function is defined where t i is the target output; the gradient descent method is used to calculate Then, a first-order gradient optimization algorithm is used to adjust w ij , c ij , δ ij , the first-order gradient to be solved is: The adaptive fuzzy inference parameter learning algorithm is:

9. The self-adaptive control method for the cutting transmission system of a coal mining machine under complex conditions according to claim 8, characterized in that, An adaptive convolutional neural network fuzzy module is configured to perform the following actions: T tra = -1 x (T me - T ste ), wherein T me is the electromagnetic torque output by the motor itself, T ste is a torque opposite to the direction of the electromagnetic torque output by the motor, and T tra is a steady-state target value of the electromagnetic torque of the motor; the adaptive convolutional neural network is continuously trained in combination with an input feature vector of a controller to obtain an adjustment amount ΔT e of the compensation electromagnetic torque of the control system. The superposition module superimposes the PI control output torque Te1. T e2 = (1 + λ) [β4fal(e1(t), a4, δ) + β5fal(e2(t), a5, δ) + ΔT e ]- z3 / b0 The cutting motor rotating speed adjustment module outputs the motor control total output compensation torque of the coal winning machine self-adaptive control system as: The method comprises the following steps: T eM = Te1+ (1 + λ) [β4fal(e1(t), a4, δ) + β5fal(e2(t), a5, δ) + ΔT e ]- z3 / b0.

10. A control method based on a self-adaptive control system of a shearer cutting transmission system under complex working conditions, characterized in that, S1, determining whether the cutting resistance of the drum of the coal winning machine changes, if yes, determining that the working condition is full coal, and executing step S2, otherwise, determining whether the working condition is coal and rock, if yes, executing step S2, otherwise, determining that the working condition is fault, and executing step S2. ​ S2, determine the motor speed and impedance, input the signal to the adaptive torque compensation controller, the controller compensates the electromagnetic torque in real time according to the signal, outputs the electromagnetic torque to adjust the cutting motor speed, the traction speed and the roller speed; S3, judge whether the cutting impedance changes, if yes, execute step S2, otherwise end.

Images