CN113983033B - Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator - Google Patents
Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator Download PDFInfo
- Publication number
- CN113983033B CN113983033B CN202111240104.6A CN202111240104A CN113983033B CN 113983033 B CN113983033 B CN 113983033B CN 202111240104 A CN202111240104 A CN 202111240104A CN 113983033 B CN113983033 B CN 113983033B
- Authority
- CN
- China
- Prior art keywords
- hydraulic cylinder
- bucket rod
- working condition
- layer
- displacement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F15—FLUID-PRESSURE ACTUATORS; HYDRAULICS OR PNEUMATICS IN GENERAL
- F15B—SYSTEMS ACTING BY MEANS OF FLUIDS IN GENERAL; FLUID-PRESSURE ACTUATORS, e.g. SERVOMOTORS; DETAILS OF FLUID-PRESSURE SYSTEMS, NOT OTHERWISE PROVIDED FOR
- F15B21/00—Common features of fluid actuator systems; Fluid-pressure actuator systems or details thereof, not covered by any other group of this subclass
- F15B21/02—Servomotor systems with programme control derived from a store or timing device; Control devices therefor
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F15—FLUID-PRESSURE ACTUATORS; HYDRAULICS OR PNEUMATICS IN GENERAL
- F15B—SYSTEMS ACTING BY MEANS OF FLUIDS IN GENERAL; FLUID-PRESSURE ACTUATORS, e.g. SERVOMOTORS; DETAILS OF FLUID-PRESSURE SYSTEMS, NOT OTHERWISE PROVIDED FOR
- F15B19/00—Testing; Calibrating; Fault detection or monitoring; Simulation or modelling of fluid-pressure systems or apparatus not otherwise provided for
- F15B19/007—Simulation or modelling
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F15—FLUID-PRESSURE ACTUATORS; HYDRAULICS OR PNEUMATICS IN GENERAL
- F15B—SYSTEMS ACTING BY MEANS OF FLUIDS IN GENERAL; FLUID-PRESSURE ACTUATORS, e.g. SERVOMOTORS; DETAILS OF FLUID-PRESSURE SYSTEMS, NOT OTHERWISE PROVIDED FOR
- F15B21/00—Common features of fluid actuator systems; Fluid-pressure actuator systems or details thereof, not covered by any other group of this subclass
- F15B21/001—Servomotor systems with fluidic control
Abstract
The invention provides a control method and a control device for identifying the working condition of a hydraulic cylinder of a bucket rod of an excavator, wherein the method comprises the following specific steps: s1, respectively acquiring the pressures of an outlet of the electric proportional variable pump, a hydraulic cylinder of the bucket rod and an oil return tank through a pressure sensor, and acquiring the displacement of the hydraulic cylinder through a displacement sensor; s2, comparing the actual displacement signal of the hydraulic cylinder acquired by the displacement sensor in the S1 with the expected displacement signal, and judging the working condition type of the hydraulic cylinder of the bucket rod; and S3, sending a corresponding control signal according to the working condition of the hydraulic cylinder of the arm obtained in the step S2. The device comprises an electric proportional variable pump, a three-phase asynchronous motor, a safety valve, a pressure sensor, a back pressure valve, a three-position three-way electromagnetic directional valve, a displacement sensor, a bucket rod hydraulic cylinder and a controller. In the process of switching among different working conditions, the invention improves the accuracy of the working condition identification result, avoids the phenomenon that the hydraulic cylinder is frequently switched among different working conditions, improves the condition of pressure fluctuation and ensures that the hydraulic cylinder runs more stably.
Description
Technical Field
The invention belongs to the technical field of intelligent identification and control of working conditions of a hydraulic system, particularly relates to an intelligent control method for identifying the working conditions of a hydraulic cylinder of a bucket rod of an excavator, and particularly relates to a control device which is suitable for stably switching between different complex working condition states.
Background
At present, for the control of a bucket rod hydraulic system of an excavator with independent valve ports, different working conditions of a bucket rod hydraulic cylinder are distinguished by a controller through a set threshold according to collected pressure, speed and displacement signals, and then the controller outputs control signals corresponding to the working conditions, so that the bucket rod moves according to a preset planned track. The method for identifying the working condition of the hydraulic cylinder by adopting the set threshold has the advantages that: the working condition state of the hydraulic cylinder can be quickly identified; the control algorithm is simple, and the understanding and application of an excavator operator and a maintenance worker are facilitated; when the working condition of the hydraulic cylinder is judged according to different physical quantities, the threshold value is convenient to adjust, and the algorithm of the controller is correspondingly modified, so that the working condition judgment is respectively carried out according to the collected different kinds of signals.
However, the state of the arm is determined according to signals such as pressure and speed, and when the acquired signals fluctuate near the threshold, the controller cannot accurately identify the working condition, which may cause frequent switching of the system between different states, valve core jitter, and unstable operation of the hydraulic cylinder. Therefore, a controller capable of accurately identifying the working conditions of different hydraulic cylinders needs to be developed, the controller is particularly suitable for accurately identifying the working conditions of the hydraulic cylinders according to collected fluctuation signals of a bucket rod hydraulic system of the excavator, and continuously and stably outputting pump and valve control signals of corresponding working conditions, so that the hydraulic cylinders stably run, and the controller has the characteristics of simple algorithm structure, short time for judging the working conditions, convenience for modifying parameters of excavators of different models, strong universality and the like.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a control method and a control device for identifying the working condition of a bucket rod hydraulic cylinder of an excavator, wherein the working condition of the bucket rod hydraulic cylinder is mainly identified by utilizing a neural network algorithm, so that the pressure and speed fluctuation generated when the bucket rod hydraulic cylinder of the excavator is switched among different states is reduced, and the continuous and stable operation of the bucket rod hydraulic cylinder is realized; and the working condition identification accuracy of the bucket rod hydraulic cylinder is improved, and the problem of frequent switching of different working conditions caused by fluctuation of the acquisition signal near a threshold value is solved.
The invention provides a control method for identifying the working condition of a hydraulic cylinder of a bucket rod of an excavator, which comprises the following specific implementation steps of:
s1, respectively acquiring pressures of an electric proportional variable pump outlet, a rod cavity of a bucket rod hydraulic cylinder, a rodless cavity of the bucket rod hydraulic cylinder and an oil return box through a pressure sensor, and acquiring displacement of the bucket rod hydraulic cylinder through a displacement sensor;
s2, building a neural network identification module, taking the difference value between the actual displacement signal and the expected displacement of the hydraulic cylinder of the bucket rod acquired by the displacement sensor in the step S1 as the input of the neural network identification module, and identifying and judging the working condition type of the hydraulic cylinder of the bucket rod:
s21, building a neural network structure for judging the working condition of the bucket rod hydraulic cylinder:
s211, carrying out data processing on the difference value between the actual displacement signal of the hydraulic cylinder of the bucket rod and the expected displacement acquired in the step S1, and dividing a training set and a testing set according to a ratio of 7: 3;
s212, setting hyper-parameters of the neural network structure: setting the number of neurons of an input layer to be 1 according to the difference value of an actual displacement signal and an expected displacement signal of a hydraulic cylinder of the bucket rod; according to three working conditions of the hydraulic cylinder of the bucket rod, the number of neurons of an output layer is set to be 3; respectively obtaining the number of the hidden layers of the two layers as 10 and 12 according to a formula for determining the number of the neurons;
s22, training the neural network structure which is built in the step S21 and used for judging the working condition of the hydraulic cylinder of the bucket rod by using a BP algorithm of state forward update and error backward propagation:
s221, setting a transfer function and a training function between each layer, training the neural network structure by using the training data set obtained in step S211, and assuming that the input displacement difference is x, obtaining the input and the output of the kth layer are as follows:
hk＝gk(netk)
in the formula, net is input at the k layer, omega is a weight matrix from the k-1 layer to the k layer, b is a bias matrix from the k-1 layer to the k layer, g is a transfer function of the k layer, h is output at the k layer, and T is a transposition symbol;
s222, defining a loss function E of the expected bucket rod hydraulic cylinder output working condition type and the actual bucket rod hydraulic cylinder output working condition type in the training data set, wherein the specific expression is as follows:
wherein y is the actual condition type, y1Outputting the actual working condition identified by the neural network, wherein the lambada 2 is a square symbol;
s223, calculating the partial derivatives of the loss function pair established in the step S222 for identifying the output layer of the hydraulic cylinder of the bucket rod and the weight and the offset of each layer, and updating the partial derivatives through error back propagation, wherein the weight and the offset of the mth layer are respectively assumed to be
And
the specific formula is as follows:
where eta is the learning rate, E is the loss function,
to calculate the bias sign;
s23, testing the neural network which is trained in the step S22 and used for judging the working condition of the hydraulic cylinder of the bucket rod by using the test set obtained in the step S211, storing an effective BP neural network model, and converting the BP neural network model into a simulink module recognition form;
s3, building a signal judgment and generation module, and sending out a corresponding control signal according to the working condition of the arm hydraulic cylinder obtained in the step S2:
s31, if the bucket rod hydraulic cylinder is in an extension working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the rodless cavity end of the bucket rod hydraulic cylinder by adopting a displacement and speed composite control method;
s32, if the bucket rod hydraulic cylinder is in a retraction working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the rodless cavity end of the bucket rod hydraulic cylinder by adopting a displacement and speed composite control method;
and S33, if the bucket rod hydraulic cylinder is in a positioning working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the rodless cavity end of the bucket rod hydraulic cylinder according to the principle of controlling the asymmetric cylinder by the asymmetric valve and the closed loop feedback control principle of hydraulic cylinder displacement.
Preferably, in step S212, the specific expression of the formula for determining the number of neurons is:
in the formula, S is the number of neurons of the hidden layer, P is the number of neurons of the input layer, O is the number of neurons of the output layer, and a is a constant value of 1-10.
Preferably, in step S22, the transfer functions of the input layer to the first hidden layer, the first hidden layer to the second hidden layer, and the second hidden layer to the output layer are a tangent-two sigmoid transfer function tansig, a linear transfer function purelin, and a flexible maximum transfer function softmax, respectively; the training function is a learning rate variable function, slingdx.
Preferably, in step S3, the operating conditions include an extending operating condition, a retracting operating condition and a positioning operating condition.
In another aspect of the present invention, there is provided a control device for identifying the working condition of a hydraulic cylinder of a bucket rod of an excavator, the device comprising an electric proportional variable pump, a three-phase asynchronous motor, a safety valve, a pressure sensor, a back pressure valve, a three-position three-way electromagnetic directional valve, a displacement sensor, a hydraulic cylinder of a bucket rod, and a neural network working condition identification controller, wherein an input end of the electric proportional variable pump is connected to the three-phase asynchronous motor, an output end of the electric proportional variable pump is connected to a first input end of a first three-position three-way electromagnetic directional valve, the safety valve is connected to a first input end of a second three-position three-way electromagnetic directional valve, a second input end of the back pressure valve and a second input end of the second three-position three-way electromagnetic directional valve are sequentially connected to a second input end of the first three-position three-way electromagnetic directional valve, an output end of the first three-position three-way electromagnetic directional valve is connected to a first end of a rod cavity of the hydraulic cylinder of the bucket rod, the displacement sensor is connected with a second end of a rod cavity of the bucket rod hydraulic cylinder, the output end of the second three-position three-way electromagnetic directional valve is connected with a rodless cavity of the bucket rod hydraulic cylinder, and the electric proportional variable pump, the back pressure valve and the three-position three-way electromagnetic directional valve are respectively connected with the neural network working condition recognition controller through a first pressure sensor, a second pressure sensor, a third pressure sensor and a fourth pressure sensor.
Preferably, the neural network working condition recognition controller comprises an extension working condition generation module, a retraction working condition generation module, a judgment output module, a neural network module and a positioning working condition generation module.
Compared with the prior art, the invention has the following advantages:
1. when the bucket rod of the excavator works, the working condition of the bucket rod hydraulic cylinder is identified by using the neural network algorithm, so that the accuracy of the working condition identification result is improved.
2. When the fluctuation phenomenon of the acquired signals is serious, the invention can avoid the problem of frequent switching of the hydraulic system among different working conditions, so that the fluctuation amplitude of the pressure and the speed of the hydraulic cylinder is reduced, the operation is more stable, and the energy loss is reduced. The invention ensures that the running track of the bucket rod is smoother and the displacement precision is improved.
3. The neural network working condition recognition algorithm in the control method can be suitable for different actuators of excavators of different models, and has strong adaptability and good robustness.
Drawings
FIG. 1 is a flow chart of a control method and a control device for identifying the working condition of a hydraulic cylinder of a bucket rod of an excavator according to the invention;
FIG. 2 is a diagram of a hydraulic device of a control device in the control method and the control device for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator according to the invention;
FIG. 3 is a schematic diagram of a controller in the control method and the control device for identifying the working condition of the hydraulic cylinder of the bucket arm of the excavator, according to the present invention;
FIG. 4 is a schematic structural diagram of a BP neural network in the control method and the control device for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator, according to the present invention;
FIGS. 5a and 5b are graphs comparing the displacement difference of the hydraulic cylinder of the control method and the control device for identifying the working condition of the hydraulic cylinder of the arm of the excavator according to the present invention;
FIGS. 6a and 6b are graphs comparing hydraulic cylinder speeds for a control method and a control device for identifying the operating conditions of a hydraulic cylinder of a stick of an excavator according to the present invention;
fig. 7a and 7b are four pressure comparison graphs of the control method and the control device for identifying the working condition of the hydraulic cylinder of the arm of the excavator according to the present invention.
The main reference numbers:
the hydraulic control system comprises an electric proportional variable pump 1, a three-phase asynchronous motor 2, a safety valve 3, a first pressure sensor 4, a back pressure valve 5, a second pressure sensor 6, a first three-position three-way electromagnetic directional valve 7, a third pressure sensor 8, a displacement sensor 9, a bucket rod hydraulic cylinder 10, a fourth pressure sensor 11, a second three-position three-way electromagnetic directional valve 12, an extending working condition generating module 13, a retracting working condition generating module 14, a judgment output module 15, a neural network module 16 and a positioning working condition generating module 17.
Detailed Description
The invention will be described in detail with reference to the accompanying drawings for describing the technical content, the achieved purpose and the efficacy of the invention.
The invention relates to a control method suitable for identifying and judging working conditions of a hydraulic cylinder of a bucket rod of an excavator and automatically selecting corresponding control signals according to different working conditions, which is used for controlling the hydraulic cylinder 10 of the bucket rod and can meet the requirement of continuous and stable operation of the bucket rod. The control method comprises a hydraulic module and a control module, wherein the hydraulic module comprises a three-phase asynchronous motor 2, an electric proportional variable pump 1, a bucket rod hydraulic cylinder 10, a safety valve 3, a back pressure valve 5, a displacement sensor 9, a three-position three-way electromagnetic directional valve and a pressure sensor. The control module consists of five control modules, namely an extension working condition generation module 13, a retraction working condition generation module 14, a judgment output module 15, a neural network module 16 and a positioning working condition generation module 17, and respectively completes the working processes of extension control, positioning control, retraction control, working condition identification and result output of the bucket rod hydraulic cylinder 10. The specific control method is realized in such a way, as shown in fig. 1.
S1, acquiring pressure, displacement and speed signals of two cavities of the hydraulic bucket rod cylinder 10 through sensors and signals of the electric proportional variable pump 1, inputting a target signal and real-time state signals of the hydraulic bucket rod cylinder 10 acquired by the first pressure sensor 4, the second pressure sensor 6, the third pressure sensor 8, the fourth pressure sensor 11 and the displacement sensor 9 into a neural network working condition recognition controller for a neural network recognition module and a signal judgment generation module, and calculating a difference value between the target signal and a feedback signal.
And S2, the neural network module 16 judges the working condition type according to the difference between the actual displacement signal and the expected displacement signal of the hydraulic bucket rod cylinder 10 acquired by the displacement sensor. And the signal generation module is used for simultaneously calculating control signals corresponding to three working conditions of extension, retraction and positioning according to the displacement and pressure signals acquired by the sensor so as to be selected by the signal generation module.
And S3, calculating three different results according to the displacement, speed and pressure signals by the signal judgment and generation module, so as to output the specific working condition of the bucket rod hydraulic cylinder 10, and output a control signal corresponding to one of the three working conditions of corresponding extension, retraction and positioning to the corresponding electric proportional variable pump 1, the first three-position three-way electromagnetic directional valve 7 and the second three-position three-way electromagnetic directional valve 12, thereby realizing the control of the bucket rod hydraulic cylinder 10.
Specifically, in the neural network module 16 of step S2, the transfer functions of the input layer to the first hidden layer, the first hidden layer to the second hidden layer, and the second hidden layer to the output layer are a tangent-two S-type transfer function tansig, a linear transfer function purelin, and a flexible maximum transfer function softmax, respectively; the training function is a learning rate variable function, namely, a thingdx, and can achieve good working condition identification and classification capability by combining the smooth and differentiable transfer function.
In the signal decision generation module of step S3, the operating conditions include an extend operating condition, a retract operating condition, and a position operating condition.
In a preferred embodiment of the present invention, as shown in fig. 2, the control device for identifying the working condition of the hydraulic cylinder of the arm of the excavator comprises an electric proportional variable pump 1, a three-phase asynchronous motor 2, a safety valve 3, a pressure sensor, a back pressure valve 5, a three-position three-way electromagnetic directional valve, a displacement sensor 9, a hydraulic cylinder of the arm 10 and a neural network working condition identification controller; the bucket rod hydraulic cylinder 10 is an asymmetric single-rod double-acting hydraulic cylinder.
The input end of an electric proportional variable pump 1 is connected with a three-phase asynchronous motor 2, the output end of the electric proportional variable pump 1 is connected with the first input end of a first three-position three-way electromagnetic directional valve 7, a safety valve 3 is connected with the first input end of a second three-position three-way electromagnetic directional valve 12, the second input ends of a back pressure valve 5 and a second three-position three-way electromagnetic directional valve 12 are sequentially connected with the second input end of the first three-position three-way electromagnetic directional valve 7, the output end of the first three-position three-way electromagnetic directional valve 7 is connected with the first end of a rod cavity of a bucket rod hydraulic cylinder 10, a displacement sensor is connected with the second end of the rod cavity of the bucket rod hydraulic cylinder 10, the output end of the second three-position three-way electromagnetic directional valve 12 is connected with a rodless cavity of the bucket rod hydraulic cylinder 10, the input ends of the electric proportional variable pump 1, the back pressure valve 5 and the three-position three-way electromagnetic directional valve are respectively connected with a first pressure sensor 4 and a second pressure sensor 6, And the third pressure sensor 8 and the fourth pressure sensor 11 are connected with the neural network working condition recognition controller.
Specifically, as shown in fig. 3, the neural network operating condition recognition controller composed of a neural network recognition module and a signal determination generation module includes an extension operating condition generation module 13, a retraction operating condition generation module 14, a determination output module 15, a neural network module 16 and a positioning operating condition generation module 17, the extension operating condition generation module 13, the retraction operating condition generation module 14 and the positioning operating condition generation module 17 perform calculation simultaneously according to a signal input to the neural network operating condition recognition controller, and three different control signals corresponding to the arm hydraulic cylinder 10 at the same time are obtained. The neural network working condition identification controller judges a specific working condition of three conditions of extension, retraction and positioning of the hydraulic cylinder 10 of the bucket rod of the excavator according to the difference values of displacement, speed and pressure, and selects a calculation result corresponding to the judgment output module 15 according to a working condition identification result to serve as a control signal output controller of the pump and the valve.
The neural network module 16 identifies and judges the current working condition of the hydraulic bucket rod cylinder 10 according to the difference value between the input target signal and the feedback signal; the judgment output module 15 converts the output signal of the neural network module 16 into a decimal number by using a Function module in the simulink, outputs the decimal number, and sends the decimal number to the first three-position three-way electromagnetic directional valve 7, the second three-position three-way electromagnetic directional valve 12 and the electric proportional variable pump 1, so that the accuracy of the controller in identifying the working condition of the hydraulic cylinder 10 of the dipper is improved, the phenomenon that a hydraulic device is frequently switched among different working conditions is avoided, the pressure fluctuation amplitude in the system is reduced, and the stable operation of the dipper is realized; the valve port of the rodless cavity in the extension working condition generation module 13 is opened to the maximum, the side valve of the cavity with the rod controls the pressure, and the electric proportional variable pump 1 adjusts the moving speed of the bucket rod hydraulic cylinder 10; the positioning working condition generating module 17 obtains control signals of the first three-position three-way electromagnetic directional valve 7 and the second three-position three-way electromagnetic directional valve 12 according to displacement closed-loop control, and adjusts the pressure of the hydraulic system by using the electric proportional variable pump 1; the valve opening on the side of the rod cavity in the retraction condition generation module 14 is opened to the maximum, the valve on the side of the rodless cavity regulates the pressure, and the electric proportional variable pump 1 controls the moving speed of the hydraulic bucket rod cylinder 10.
Further, the neural network module 16 performs operation condition recognition on the arm cylinder 10 by using a BP neural network. The actual rule of the BP neural network is an algorithm of forward update of the state and backward propagation of the error, as shown in fig. 4.
The BP neural network forward updating process is as follows:
input layer to first hidden layer:
h1＝g1(net1)
where x is the displacement difference signal of the input layer, net1As a first hidden layer input, ω1As a weight matrix of the input layer to the first hidden layer, b1For the bias matrix of the input layer to the first hidden layer, g1Is the first layer transfer function, h1For the first layer output, T is the transposed symbol.
First hidden layer to second hidden layer:
h2＝g2(net2)
in the formula, net2For the second hidden layer input, ω2A weight matrix from the first hidden layer to the second hidden layer, b2A bias matrix from the first hidden layer to the second hidden layer, g2Is the second layer transfer function, h1For the first layer output, h2For the second layer output, T is the transposed symbol.
Second hidden layer to output layer:
h3＝g3(net3)
in the formula, net3For the output layer input, ω3As a weight matrix from the second hidden layer to the output layer, b3For a bias matrix of the second hidden layer to the output layer, g3Is the third layer transfer function, h2For the second layer of transfusionOut, h3For the third level output, T is the transposed symbol.
The BP neural network error back propagation process is as follows:
updating parameters of an output layer:
in the formula, ω3As a weight matrix from the second hidden layer to the output layer, b3Is a bias matrix from the second hidden layer to the output layer, E is a loss function, η is a learning rate, k is an iteration number,
to calculate the sign of the partial derivatives.
Second hidden layer parameter update:
in the formula, ω2A weight matrix from the first hidden layer to the second hidden layer, b2Is a bias matrix from the first hidden layer to the second hidden layer, E is a loss function, η is a learning rate, k is an iteration number,
to calculate the sign of the partial derivatives.
First hidden layer parameter update:
in the formula, ω1As a weight matrix of the input layer to the first hidden layer, b1Is the bias matrix of the input layer to the first hidden layer, E is the loss function, η is the learning rate, k is the number of iterations,
to calculate the sign of the partial derivatives.
The control method and the control device for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator are further described by combining the embodiment as follows:
s1, respectively acquiring the pressures of the outlet of the electric proportional variable pump 1, the rod cavity of the bucket rod hydraulic cylinder 10, the rodless cavity of the bucket rod hydraulic cylinder 10 and the oil return tank through pressure sensors, and acquiring the displacement of the bucket rod hydraulic cylinder 10 through the displacement sensor 9.
S2, building a neural network recognition module, taking the difference between the actual displacement signal of the arm hydraulic cylinder 10 and the expected displacement acquired by the displacement sensor 9 in step S1 as the input of the neural network recognition module, and recognizing and judging the working condition type of the arm hydraulic cylinder 10, as shown in fig. 4:
s21, building a neural network structure for judging the working condition of the bucket rod hydraulic cylinder 10:
and S211, carrying out data processing on the difference value between the actual displacement signal of the hydraulic bucket rod cylinder 10 and the expected displacement acquired in the step S1, and dividing a training set and a testing set according to a ratio of 7: 3.
S212, setting hyper-parameters of the neural network structure: setting the number of neurons of an input layer to be 1 according to the difference value between the actual displacement signal and the expected displacement signal of the bucket rod hydraulic cylinder 10; according to the three working conditions of the bucket rod hydraulic cylinder 10, the number of neurons of an output layer is set to be 3; respectively obtaining the number of hidden layers of two layers as 10 and 12 according to a formula for determining the number of the neurons, wherein the specific expression for determining the formula for the number of the neurons is as follows:
in the formula, S is the number of neurons of the hidden layer, P is the number of neurons of the input layer, O is the number of neurons of the output layer, and a is a constant value of 1-10.
S22, training the neural network structure which is built in the step S21 and used for judging the working condition of the arm hydraulic cylinder 10 by using a BP algorithm of state forward update and error backward propagation:
s221, setting a transfer function and a training function between each layer, training the neural network structure by using the training data set obtained in step S211, assuming that the input displacement difference is x, the input and the output obtained at the kth layer are as follows:
hk＝gk(netk)
in the formula, net is the input of the k-th layer, ω is the weight matrix from the k-1-th layer to the k-th layer, b is the bias matrix from the k-1-th layer to the k-th layer, g is the transfer function of the k-th layer, h is the output of the k-th layer, and T is the transposition symbol.
S222, defining a loss function E of the expected output working condition type of the hydraulic cylinder 10 of the arm in the training data set and the actual output working condition type of the hydraulic cylinder 10 of the arm, wherein the specific expression is as follows:
wherein y is the real condition type, y1The output is the actual working condition recognized by the neural network, and the lambada 2 is a square symbol.
S223, calculating the partial derivatives of the loss function pair established in the step S222 for identifying the output layer of the arm hydraulic cylinder 10 and the weight and the offset of each layer, and updating the partial derivatives through error back propagation, wherein the weight and the offset of the mth layer are respectively assumed to be the weight and the offset of the mth layer during the nth iteration
And
the specific formula is as follows:
where eta is the learning rate, E is the loss function,
to calculate the sign of the partial derivatives.
And S23, testing the neural network which is trained in the step S22 and used for judging the working condition of the arm hydraulic cylinder 10 by using the test set obtained in the step S211, storing an effective BP neural network model, and converting the BP neural network model into a simulink module recognition form.
S3, building a signal judgment and generation module, and sending out a corresponding control signal according to the working condition of the arm hydraulic cylinder obtained in the step S2:
s31, if the arm hydraulic cylinder 10 is in an extending working condition, control signals of an electric proportional variable pump 1, a rod cavity of the arm hydraulic cylinder 10, a rodless cavity of the arm hydraulic cylinder 10, a first three-position three-way electromagnetic directional valve 7 and a second three-position three-way electromagnetic directional valve 12 are generated by adopting a displacement and speed composite control method, so that the arm hydraulic cylinder 10 extends stably, wherein the speed and the displacement of the arm hydraulic cylinder 10 are controlled by the electric proportional variable pump 1, 40mA current is input into the second three-position three-way electromagnetic directional valve 12, a valve port is opened to the maximum, and the first three-position three-way electromagnetic directional valve 7 regulates back pressure.
The control signal of the electric proportional variable pump 1 is generated by the joint action of displacement and speed control information, and the specific expression is as follows:
Ux,v＝Ux+Uv
Ux＝Kpx[(xr-xrel)+∫(xr-xrel)/TIx]
Uv＝vrA1/nVpmax
in the formula, A1Is the piston area, v, of the arm cylinder 10rThe speed of the arm cylinder 10, n is the rotational speed of the electric proportional variable pump 1, VpmaxIs the maximum displacement, x, of the electric proportional variable displacement pump 1rTo expect displacement of arm cylinder 10, xrelFor actual displacement of arm cylinder 10, Ux,vSignal of the electric proportional variable pump 1, Kpx、TIxProportional coefficient and integral coefficient of PI control.
The control signal of the first three-position three-way electromagnetic directional valve 7 is generated according to the pressure control requirement, and the specific expression is as follows:
U7＝Kp[(pr-p2)+∫(pr-p2)/Ti]
in the formula of U7Is the electrical signal, p, of the first three-position three-way electromagnetic directional valve 7rTo expect the pressure of arm cylinder 10, p2For the rod chamber side pressure, K, of the bucket rod hydraulic cylinder 10p、TiProportional coefficient and integral coefficient of PI control respectively.
And S32, if the bucket rod hydraulic cylinder 10 is in a retraction working condition, generating control signals of the electric proportional variable pump 1, a rod cavity of the bucket rod hydraulic cylinder 10, a rodless cavity of the bucket rod hydraulic cylinder 10, the first three-position three-way electromagnetic directional valve 7 and the second three-position three-way electromagnetic directional valve 12 by adopting a displacement and speed composite control method, and enabling the bucket rod hydraulic cylinder 10 to retract stably. The speed and displacement of the bucket rod hydraulic cylinder 10 are controlled by the electric proportional variable pump 1, 40mA current is input into the first three-position three-way electromagnetic directional valve 7, the valve port is opened to the maximum, and the back pressure of the rodless cavity side is adjusted by the second three-position three-way electromagnetic directional valve 12.
The specific expression of the control signal of the electric proportional variable pump 1 is as follows:
Ux,v＝Ux+Uv
Ux＝Kpx[(xr-xrel)+∫(xr-xrel)/Tix]
Uv＝vrA2/nVpmax
in the formula, A2The working area v of oil on the side of the rod cavityrThe speed of the arm cylinder 10, n is the rotational speed of the electric proportional variable pump 1, VpmaxIs the maximum displacement, x, of the electric proportional variable displacement pump 1rTo expect displacement of arm cylinder 10, xrelFor actual displacement of arm cylinder 10, Ux,vIs the signal of the electrical proportional variable pump 1.
The specific expression of the control signal of the second three-position three-way electromagnetic directional valve 12 is as follows:
U12＝Kp[(pr-pl)+∫(pr-pl)/Ti]
in the formula, p1For actual pressure on the rodless chamber side, U12Is the electrical signal, p, of the second three-position three-way solenoid directional valve 12rTo expect the pressure of the arm cylinder 10, Kp、TiProportional coefficient and integral coefficient of PI control.
S33, if the bucket rod hydraulic cylinder 10 is in a positioning working condition, according to the principle that the asymmetric cylinder is controlled by the asymmetric valve and the hydraulic cylinder displacement closed loop feedback control principle, control signals of the electric proportional variable pump 1, the first three-position three-way electromagnetic directional valve 7 and the second three-position three-way electromagnetic directional valve 12 are generated, when the bucket rod hydraulic cylinder 10 stops moving, the displacement precision requirement is met, and according to the principle that the asymmetric cylinder is controlled by the asymmetric valve, the control signals of the first three-position three-way electromagnetic directional valve 7 and the second three-position three-way electromagnetic directional valve 12 are input according to the ratio of the oil-liquid action areas of two cavities of the bucket rod hydraulic cylinder 10.
According to the displacement feedback PID control, a specific expression of the control signal of the second three-position three-way electromagnetic directional valve 12 is obtained as follows:
in the formula of U12Is the electrical signal, x, of the second three-position three-way solenoid directional valve 12dTo expect displacement of arm cylinder 10, xrelFor actual displacement of arm cylinder 10, Kp、TiProportional coefficient and integral coefficient of PI control.
According to the ratio of the oil action areas of the rod cavity of the bucket rod hydraulic cylinder 10 and the rodless cavity of the bucket rod hydraulic cylinder 10, a specific expression of a control signal of the first three-position three-way electromagnetic directional valve 7 is obtained as follows:
U7＝αU12
α＝A2/A1
in the formula of U7Is the electrical signal of the first three-position three-way electromagnetic directional valve 7, U12 is the electrical signal of the second three-position three-way electromagnetic directional valve 12, A2For the oil action area of the rod chamber of the bucket rod hydraulic cylinder 10, A1Is the oil action area of the rodless cavity of the bucket rod hydraulic cylinder 10, and alpha is the ratio of the oil action areas of the rod cavity of the bucket rod hydraulic cylinder 10 and the rodless cavity of the bucket rod hydraulic cylinder 10.
In the control method, the neural network module 16 is set with the hyper-parameters in the step S212, then training and testing are performed to obtain the recognition accuracy rates of the extension working condition, the retraction working condition and the positioning working condition of 100%, 99.82% and 99.65%, and finally the trained BP neural network is converted into the simulink module recognition and hydraulic device for joint simulation through a command 'genesis (net-1)'.
According to the simulation results shown in fig. 5 and fig. 6 and fig. 7, it can be clearly seen (the neural network module 16 is added in fig. 5a, fig. 6a and fig. 7a, and the neural network module 16 is not added in fig. 5b, fig. 6b and fig. 7 b), by using the neural network working condition recognition controller, the speed and the pressure of the arm hydraulic cylinder 10 are both significantly improved, the working condition is switched to be smoother and smoother, the speed fluctuation is small, and the pressure value is also more stable in 4 to 6 seconds.
The curves in fig. 5 a-5 b are the difference between the target displacement and the displacement of the simulation result; the curves in fig. 6 a-6 b are velocity simulation results for the hydraulic cylinder; the four curves in fig. 7 a-7 b are respectively electricitySimulation result Ps of outlet pressure of proportional variable pump 1 and side pressure P of rodless cavity of bucket rod hydraulic cylinder 101Bucket rod hydraulic cylinder 10 rod cavity side pressure P2Return tank side pressure P0(ii) a In the comparison of the displacement difference in fig. 4, the speed in fig. 5 and the pressure simulation result in fig. 6, the left side is the simulation result of the condition recognition controller of the conventional threshold method, and the right side is the simulation result of the condition recognition controller of the neural network.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements made to the technical solution of the present invention by those skilled in the art without departing from the spirit of the present invention shall fall within the protection scope defined by the claims of the present invention.
Claims (6)
1. A control method for identifying the working condition of a hydraulic cylinder of a bucket rod of an excavator is characterized by comprising the following steps of:
s1, respectively acquiring pressures of an outlet of an electric proportional variable pump, a rod cavity of a bucket rod hydraulic cylinder, a rodless cavity of the bucket rod hydraulic cylinder and an oil return box through a pressure sensor, acquiring displacement of the bucket rod hydraulic cylinder through a displacement sensor, and acquiring the speed of the bucket rod hydraulic cylinder through a speed sensor;
s2, building a neural network identification module, taking the difference value between the actual displacement signal and the expected displacement of the hydraulic cylinder of the bucket rod acquired by the displacement sensor in the step S1 as the input of the neural network identification module, and identifying and judging the working condition type of the hydraulic cylinder of the bucket rod:
s21, building a neural network structure for judging the working condition of the bucket rod hydraulic cylinder:
s211, carrying out data processing on the difference value between the actual displacement signal of the hydraulic cylinder of the bucket rod collected in the step S1 and the expected displacement, and dividing a training set and a testing set according to a ratio of 7: 3;
s212, setting hyper-parameters of the neural network structure: setting the number of neurons of an input layer to be 1 according to the difference value of an actual displacement signal and an expected displacement signal of a hydraulic cylinder of the bucket rod; according to three working conditions of the hydraulic cylinder of the bucket rod, the number of neurons of an output layer is set to be 3; respectively obtaining the number of the hidden layers of the two layers as 10 and 12 according to a formula for determining the number of the neurons;
s22, training the neural network structure which is built in the step S21 and used for judging the working condition of the hydraulic cylinder of the bucket rod by using a BP algorithm of state forward update and error backward propagation:
s221, setting a transfer function and a training function between each layer, training the neural network structure by using the training data set obtained in step S211, assuming that the input displacement difference is x, the input and the output obtained at the kth layer are as follows:
hk＝gk(netk)
in the formula, net is input at the k layer, omega is a weight matrix from the k-1 layer to the k layer, b is a bias matrix from the k-1 layer to the k layer, g is a transfer function of the k layer, h is output at the k layer, and T is a transposition symbol;
s222, defining a loss function E of the expected bucket rod hydraulic cylinder output working condition type and the actual bucket rod hydraulic cylinder output working condition type in the training data set, wherein the specific expression is as follows:
wherein y is the real condition type, y1Outputting the actual working condition identified by the neural network, wherein the lambada 2 is a square symbol;
s223, calculating the partial derivatives of the loss function pair established in the step S222 for identifying the output layer of the hydraulic cylinder of the bucket rod and the weight and the offset of each layer, and updating the partial derivatives through error back propagation, wherein the weight and the offset of the mth layer are respectively assumed to be
And
the specific formula is as follows:
where eta is the learning rate, E is the loss function,
to calculate the bias sign;
s23, testing the neural network which is trained in the step S22 and used for judging the working condition of the hydraulic cylinder of the bucket rod by using the test set obtained in the step S211, storing an effective BP neural network model, and converting the BP neural network model into a simulink module recognition form;
s3, building a signal judgment and generation module, and sending out a corresponding control signal according to the working condition of the arm hydraulic cylinder obtained in the step S2:
s31, if the bucket rod hydraulic cylinder is in an extension working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the rodless cavity end of the bucket rod hydraulic cylinder by adopting a displacement and speed composite control method;
s32, if the bucket rod hydraulic cylinder is in a retraction working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the rodless cavity end of the bucket rod hydraulic cylinder by adopting a displacement and speed composite control method;
and S33, if the bucket rod hydraulic cylinder is in a positioning working condition, generating control signals of the electric proportional variable pump, a first three-position three-way electromagnetic directional valve at the rod cavity end of the bucket rod hydraulic cylinder and a second three-position three-way electromagnetic directional valve at the non-rod cavity end of the bucket rod hydraulic cylinder according to the principle of controlling the asymmetric cylinder by the asymmetric valve and the closed loop feedback control principle of hydraulic cylinder displacement.
2. The control method for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator as claimed in claim 1, wherein in step S212, the specific expression of the formula for determining the number of the neurons is as follows:
in the formula, S is the number of neurons of the hidden layer, P is the number of neurons of the input layer, O is the number of neurons of the output layer, and a is a constant value of 1-10.
3. The control method for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator according to claim 1, wherein in the step S22, the transfer functions from the input layer to the first hidden layer, from the first hidden layer to the second hidden layer and from the second hidden layer to the output layer are a tangent S-type transfer function tansig, a linear transfer function purelin and a flexible maximum transfer function softmax, respectively; the training function is a learning rate variable function, slingdx.
4. The control method for identifying the operating condition of the excavator arm hydraulic cylinder of claim 1, wherein in step S3, the operating condition categories include an extension operating condition, a retraction operating condition and a positioning operating condition.
5. A control device for a control method for identifying the working condition of a stick cylinder of an excavator according to any one of claims 1 to 4, comprising an electric proportional variable pump, a three-phase asynchronous motor, a safety valve, a pressure sensor, a back pressure valve, a three-position three-way solenoid directional valve, a displacement sensor, a stick cylinder and a neural network working condition identifying controller, wherein an input terminal of the electric proportional variable pump is connected to the three-phase asynchronous motor, an output terminal of the electric proportional variable pump is connected to a first input terminal of a first three-position three-way solenoid directional valve, the safety valve is connected to a first input terminal of a second three-position three-way solenoid directional valve, the back pressure valve and a second input terminal of the second three-position three-way solenoid directional valve are sequentially connected to a second input terminal of the first three-position three-way solenoid directional valve, and an output terminal of the first three-position three-way solenoid directional valve is connected to a first terminal of a stick chamber of the stick cylinder And the displacement sensor is connected with the second end of the rod cavity of the bucket rod hydraulic cylinder, the output end of the second three-position three-way electromagnetic directional valve is connected with the rodless cavity of the bucket rod hydraulic cylinder, and the electric proportional variable pump, the back pressure valve and the three-position three-way electromagnetic directional valve are respectively connected with the neural network working condition recognition controller through a first pressure sensor, a second pressure sensor, a third pressure sensor and a fourth pressure sensor.
6. The control device of the control method for identifying the working condition of the hydraulic cylinder of the bucket rod of the excavator as claimed in claim 5, wherein the neural network working condition identification controller comprises an extension working condition generation module, a retraction working condition generation module, a judgment output module, a neural network module and a positioning working condition generation module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111240104.6A CN113983033B (en) | 2021-10-25 | 2021-10-25 | Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111240104.6A CN113983033B (en) | 2021-10-25 | 2021-10-25 | Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113983033A CN113983033A (en) | 2022-01-28 |
| CN113983033B true CN113983033B (en) | 2022-07-01 |
Family
ID=79740931
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111240104.6A Active CN113983033B (en) | 2021-10-25 | 2021-10-25 | Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113983033B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5854993A (en) * | 1996-12-10 | 1998-12-29 | Caterpillar Inc. | Component machine testing using neural network processed vibration data analysis |
| CN101135601A (en) * | 2007-10-18 | 2008-03-05 | 北京英华达电力电子工程科技有限公司 | Rotating machinery vibrating failure diagnosis device and method |
| CN107015476A (en) * | 2017-03-28 | 2017-08-04 | 哈尔滨理工大学 | A kind of position and the cooperative control method of force signal to electrohydraulic servo system |
| CN109778941A (en) * | 2019-03-25 | 2019-05-21 | 江苏徐工工程机械研究院有限公司 | A kind of semi-autonomous digging system and method based on intensified learning |
| CN111255756A (en) * | 2020-03-10 | 2020-06-09 | 常熟理工学院 | Variable speed pump control system in hydraulic system |
| DE102019113765A1 (en) * | 2019-05-23 | 2020-11-26 | Jungheinrich Ag | Method for controlling a hydraulic system of a mobile work machine and mobile work machine |
| CN113431925A (en) * | 2021-07-12 | 2021-09-24 | 南京工程学院 | Electro-hydraulic proportional valve and position control system, control method and fault prediction method thereof |
-
2021
- 2021-10-25 CN CN202111240104.6A patent/CN113983033B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5854993A (en) * | 1996-12-10 | 1998-12-29 | Caterpillar Inc. | Component machine testing using neural network processed vibration data analysis |
| CN101135601A (en) * | 2007-10-18 | 2008-03-05 | 北京英华达电力电子工程科技有限公司 | Rotating machinery vibrating failure diagnosis device and method |
| CN107015476A (en) * | 2017-03-28 | 2017-08-04 | 哈尔滨理工大学 | A kind of position and the cooperative control method of force signal to electrohydraulic servo system |
| CN109778941A (en) * | 2019-03-25 | 2019-05-21 | 江苏徐工工程机械研究院有限公司 | A kind of semi-autonomous digging system and method based on intensified learning |
| DE102019113765A1 (en) * | 2019-05-23 | 2020-11-26 | Jungheinrich Ag | Method for controlling a hydraulic system of a mobile work machine and mobile work machine |
| CN111255756A (en) * | 2020-03-10 | 2020-06-09 | 常熟理工学院 | Variable speed pump control system in hydraulic system |
| CN113431925A (en) * | 2021-07-12 | 2021-09-24 | 南京工程学院 | Electro-hydraulic proportional valve and position control system, control method and fault prediction method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113983033A (en) | 2022-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107882103B (en) | A kind of excavator 3 d pose is shown and Remote Automatic Control System | |
| Feng et al. | A new adaptive sliding mode controller based on the RBF neural network for an electro-hydraulic servo system | |
| Shi et al. | Electro-hydraulic velocity and position control based on independent metering valve control in mobile construction equipment | |
| CN110409546B (en) | Electric control system of excavator and positive flow system excavator | |
| CN102041824B (en) | Automatic control system and control method of working device of excavator | |
| CN113006188B (en) | Excavator staged power matching method based on LSTM neural network | |
| CN109696836B (en) | Intelligent control method for airplane steering engine electrohydraulic servo system | |
| Ge et al. | Research on the performance of hydraulic excavator with pump and valve combined separate meter in and meter out circuits | |
| CN103176409B (en) | Method for fast and accurately realizing concrete pump truck cantilever crane movement locus | |
| CN109884894B (en) | Neural network integral sliding mode control method for electro-hydraulic power-assisted steering system | |
| CN211113856U (en) | Electric control system of excavator and positive flow system excavator | |
| CN109869177B (en) | Electro-hydraulic servo valve control hydraulic frame moving system and displacement control method thereof | |
| CN112659498B (en) | Real-time optimal control method for injection molding machine deep neural network | |
| CN112681443A (en) | Excavating robot joint track control method and control system | |
| CN113983033B (en) | Control method and control device for identifying working condition of hydraulic cylinder of bucket rod of excavator | |
| CN112925208A (en) | Disturbance compensation method for data-driven electro-hydraulic servo system of well drilling machine | |
| CN110455290A (en) | A kind of optimal trajectory planning method of intelligent hydraulic excavator | |
| CN112096696A (en) | Self-adaptive inversion control method for pump-controlled asymmetric hydraulic position system | |
| Lee et al. | Development of the independent metering valve control system and analysis of its performance for an excavator | |
| CN109884887A (en) | A kind of Continuous rotary electrohydraulic servomotor control method based on self-correcting wavelet neural network | |
| Qin et al. | An adaptive robust impedance control considering energy-saving of hydraulic excavator boom and stick systems | |
| CN113139514A (en) | Hydraulic system fault diagnosis method based on fuzzy neural network | |
| CN113431925B (en) | Fault prediction method of electro-hydraulic proportional valve | |
| Coskun et al. | Intelligent PID control of an industrial electro-hydraulic system | |
| Ketonen et al. | Simulation study of a digital hydraulic independent metering valve system on an excavator |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |