CN108118743B - Hydraulic excavator self-adaptive neural network moment control device and method - Google Patents
Hydraulic excavator self-adaptive neural network moment control device and method Download PDFInfo
- Publication number
- CN108118743B CN108118743B CN201810039705.2A CN201810039705A CN108118743B CN 108118743 B CN108118743 B CN 108118743B CN 201810039705 A CN201810039705 A CN 201810039705A CN 108118743 B CN108118743 B CN 108118743B
- Authority
- CN
- China
- Prior art keywords
- neural network
- valve
- hydraulic motor
- hydraulic
- pressure
- 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
Classifications
-
- E—FIXED CONSTRUCTIONS
- E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
- E02F—DREDGING; SOIL-SHIFTING
- E02F9/00—Component parts of dredgers or soil-shifting machines, not restricted to one of the kinds covered by groups E02F3/00 - E02F7/00
- E02F9/20—Drives; Control devices
- E02F9/22—Hydraulic or pneumatic drives
- E02F9/2203—Arrangements for controlling the attitude of actuators, e.g. speed, floating function
Abstract
The invention provides a self-adaptive neural network moment control device of a hydraulic excavator, which comprises a hydraulic pump and a ratioThe hydraulic motor comprises an example overflow valve, an oil tank, a proportional throttle valve, a one-way valve, a six-way opening center proportional reversing valve, a first proportional overflow valve, a second proportional overflow valve, a first one-way valve, a second one-way valve and a hydraulic motor; the control device is used for controlling the hydraulic motor; the invention also provides a torque control method of the self-adaptive neural network of the hydraulic excavator, wherein the analog output unit sends a control signal u of the six-way open center proportional reversing valve to the six-way open center proportional reversing valve, so as to control the left cavity pressure P of the hydraulic motor of the right two cavities of the hydraulic Ma Dazuo 1 And hydraulic motor right cavity pressure P 2 The hydraulic excavator torque control system can avoid the condition that the flow control gain is constant to zero when the torque control is carried out on the hydraulic excavator, so that the whole system is controllable.
Description
Technical Field
The invention relates to the technical field of torque control of hydraulic excavators, in particular to a device and a method for controlling torque of a self-adaptive neural network of a hydraulic excavator.
Background
As an important engineering machine, hydraulic excavators are widely used in engineering construction, mining industry, and forestry. In order to improve the working efficiency and working accuracy of the hydraulic excavator, scholars at home and abroad have conducted extensive researches on the automated operation of the hydraulic excavator. Automatic work excavators are capable of performing various typical work tasks without human intervention, such as: truck loading and unloading, deep trench digging, land leveling, sidewall cutting and sidewall extrusion. Among them, in order to realize automatic operations of side wall cutting and side wall pressing, it is necessary to control the turning moment of the hydraulic excavator. However, when torque control is performed on a hydraulic excavator, there are two main problems:
(1) For a traditional hydraulic excavator rotary hydraulic system, the main control valve is a six-way open center proportional reversing valve. The control voltage of the proportional reversing valve simultaneously adjusts the outlet pressure of the hydraulic pump and the flow rate flowing into the hydraulic motor. This way of coupling control presents a problem: when the control voltage is not zero and the outlet pressure of the hydraulic pump is lower than the pressure of the left cavity and the right cavity of the hydraulic motor, the hydraulic pump cannot supply oil to the hydraulic motor. For the conventional hydraulic excavator rotary hydraulic system, the pressure dynamic equation of the left and right cavities of the hydraulic motor can be expressed as follows:
wherein: q 1 And q 2 G is the flow rate of the two right cavities flowing into the hydraulic pressure Ma Dazuo 1 And g 2 Flow control gain for six-way open center proportional reversing valve, u c Control voltage k of six-way open center proportional reversing valve q1 And k q2 Is positive coefficient (determined by experiment), p s For outlet pressure of hydraulic pump, p 1 And p 2 Is the pressure of the left cavity and the right cavity of the hydraulic motor, p t Is the oil pressure of the oil tank. As can be seen from equation (1), when u c >0&(p s ＜p 1 ) Or u c <0&(p s ＜p 2 ) At this time, the flow control gain g of the system 1 Or g 2 Always zero. Therefore, effective control of the torque of the hydraulic motor is difficult to achieve with conventional hydraulic excavator swing hydraulic systems.
(2) At present, the academic circles at home and abroad aim at main flow control methods of hydraulic system force and moment, such as: adaptive control, robust control and passive control, and the parameters of a mathematical model of the system need to be known in advance. However, it is very difficult for a hydraulic excavator to obtain parameters of a mathematical model of the system, mainly because: 1) Due to the influence of complex nonlinear characteristics such as nonlinear flow gain, hysteresis, dead zone, saturation, dynamic flow coupling and the like, accurate characteristic parameters of a hydraulic element and a hydraulic system are difficult to obtain; 2) Changes in operating temperature and aging of hydraulic components can cause changes in system parameters.
Accordingly, improvements in the art are needed.
Disclosure of Invention
The invention aims to provide an efficient self-adaptive neural network torque control device and method for a hydraulic excavator.
In order to solve the technical problems, the invention provides a self-adaptive neural network moment control device of a hydraulic excavator, which is used for controlling a hydraulic motor and is characterized in that: the hydraulic pump comprises a hydraulic pump, a proportional overflow valve, an oil tank, a proportional throttle valve, a one-way valve, a six-way open center proportional reversing valve, a proportional overflow valve I, a proportional overflow valve II, a one-way valve I, a one-way valve II and a hydraulic motor;
the six-way open center proportional reversing valve is provided with a valve front oil inlet a, a valve front oil inlet B, a middle oil inlet, an A outlet, a B outlet and a middle oil outlet;
the inlet of the hydraulic pump is communicated with the oil tank, the outlet of the hydraulic pump is divided into three paths, and one path is communicated with the oil tank after passing through the proportional overflow valve; one path of the oil passes through the one-way valve and then is communicated with a valve front oil inlet a of the six-way open center proportional reversing valve; one path is communicated with a middle path oil inlet of a six-way open center proportional reversing valve after passing through a proportional throttle valve;
the middle oil outlet of the six-way open center proportional reversing valve is respectively communicated with the front oil inlet b of the valve and the oil tank;
the outlet of the six-way open center proportional reversing valve A is respectively communicated with a left cavity oil inlet of the hydraulic motor, an inlet of the proportional overflow valve I and a spring cavity oil inlet of the one-way valve I, and an outlet of the proportional overflow valve I and a non-spring cavity oil inlet of the one-way valve I are respectively communicated with an oil tank;
and the outlet of the second proportional overflow valve and the non-spring cavity oil port of the second check valve are respectively communicated with an oil tank.
As an improvement on the self-adaptive neural network moment control device of the hydraulic excavator, the invention comprises the following steps: the torque control device also comprises a low-pass filter, an analog quantity acquisition unit, a torque calculator, a reference torque signal generator, a comparator, an echo state neural network, a comprehensive interference upper bound updater, a neural network parameter updater, a control signal generator and an analog quantity output unit;
a pump pressure sensor is arranged on a pipeline between the hydraulic pump and the proportional overflow valve;
a first hydraulic motor pressure sensor is arranged on a pipeline between an A outlet of the six-way open center proportional reversing valve and a left cavity oil inlet of the hydraulic motor;
a second hydraulic motor pressure sensor is arranged on a pipeline between the outlet B of the six-way open center proportional reversing valve and the right cavity oil inlet of the hydraulic motor;
the pump pressure sensor, the hydraulic motor pressure sensor I and the hydraulic motor pressure sensor II are respectively connected with a low-pass filter, and after filtering, the low-pass filter sends corresponding signals to the analog quantity acquisition unit; the analog quantity acquisition unit acquires and obtains a hydraulic pump outlet pressure signal P s Left chamber pressure P of hydraulic motor 1 And hydraulic motor right cavity pressure P 2 ；
The analog quantity acquisition unit sends a hydraulic pump outlet pressure signal P s To a control signal generator, the analog acquisition unit acquires the hydraulic pressure Ma Dazuo cavity pressure P 1 And hydraulic motor right cavity pressure P 2 Respectively sending the signals to a moment calculator and an echo state neural network;
moment calculator according to pressure P of hydraulic Ma Dazuo cavity 1 And hydraulic motor right cavity pressure P 2 Calculating to obtain an output torque signal tau, and sending the output torque signal tau to a comparator;
the reference torque signal generator generates a reference torque signal tau d And a differential signal of the reference torque signal versus timeAnd will reference moment signal tau d Sending to a comparator a differential signal of the reference torque signal over time +.>Transmitting the echo state data to an echo state neural network;
the comparator is used for comparing the output torque signal tau with the reference torque signal tau d A moment tracking error signal e is obtained, and the moment tracking error signal e is sent to an echo state neural network, a comprehensive interference upper bound updater and a neural network parameter updater;
the comprehensive interference upper bound updater generates a moment tracking error signal e according to the moment tracking error signal eEstimate of upper bound of integrated interferenceAnd the estimated value of the upper bound of the integrated interference is +.>Sending to a control signal generator;
the acoustic echo state neural network is based on the left cavity pressure P of the hydraulic motor 1 Hydraulic motor right chamber pressure P 2 Differential signal of reference moment signal versus timeAnd the estimated value of the output weight of the neural network +.>Calculating to obtain a neural network internal state variable phi, and sending the neural network internal state variable phi to a neural network parameter updater and a control signal generator;
the neural network parameter updater calculates to obtain the estimated value of the output weight of the neural network according to the internal state variable phi of the neural network and the moment tracking error signal eAnd outputting the estimated value of the weight of the neural network +.>Transmitting to an echo state neural network and a control signal generator;
the control signal generator generates a hydraulic pump outlet pressure signal P based on the hydraulic pump outlet pressure signal s And a set relief pressure P preset in the control signal generator set The proportional throttle valve control signal u is obtained by operation Ps The method comprises the steps of carrying out a first treatment on the surface of the The control signal generator outputs an estimated value of the weight according to the neural networkEstimation of the internal state variable phi and the upper bound of the integrated interference of a neural networkValue->And a torque tracking error signal e, calculating to obtain a six-way open center proportional reversing valve control signal u, and calculating the proportional throttle valve control signal u Ps And a six-way open center proportional reversing valve control signal u is sent to an analog output unit;
the analog quantity output unit is respectively connected with the proportional throttle valve and the six-way opening center proportional reversing valve; the analog output unit outputs a proportional throttle control signal u Ps Is sent to the proportional throttle valve to control the opening size of the proportional throttle valve, thereby controlling the outlet pressure P of the hydraulic pump s ；
The analog quantity output unit sends a control signal u of the six-way open center proportional reversing valve to the six-way open center proportional reversing valve, so that the left cavity pressure P of the hydraulic motor with the right two cavities of the hydraulic pressure Ma Dazuo is controlled 1 And hydraulic motor right cavity pressure P 2 And controlling the output torque of the hydraulic motor.
The invention also provides a torque control method of the self-adaptive neural network of the hydraulic excavator, which comprises the following steps: six-way open center proportional reversing valve control signal u controls flow Q flowing into right two cavities of hydraulic Ma Dazuo 1 And Q 2 Flow rate Q flowing into right two chambers of hydraulic pressure Ma Dazuo 1 And Q 2 Control ofAnd->Controlling hydraulic motor left chamber pressure P 1 And hydraulic motor right cavity pressure P 2 Thereby controlling the output torque of the hydraulic motor.
As an improvement on the self-adaptive neural network torque control method of the hydraulic excavator, the specific process of controlling the output torque of the hydraulic motor is as follows:
flow rate Q into right two chambers of hydraulic pressure Ma Dazuo 1 And Q 2 Can be expressed as:
wherein: g 1 And G 2 The flow control gains of the valve front oil inlet a to the outlet A and the valve front oil inlet B to the outlet B of the six-way open center proportional directional valve 7 are respectively shown as u, K and the six-way open center proportional directional valve control signal q1 And K q2 For flow gain, P s For the outlet pressure of the hydraulic pump, P t The pressure of the oil in the oil tank;
the dynamic equation of the pressure of the left cavity and the right cavity of the hydraulic motor is expressed as follows:
and->Respectively represent the pressure P of the hydraulic Ma Dazuo cavity 1 And hydraulic motor right cavity pressure P 2 Differential signal over time, beta e Is the elastic bulk modulus of oil, V 0 The volume of the left cavity and the right cavity of the hydraulic motor is C tm C is the internal leakage coefficient em1 And C em2 The leakage coefficients of the left cavity and the right cavity of the hydraulic motor are obtained;
and->The hydraulic motor left cavity pressure P is obtained through integral operation 1 And hydraulic motor right cavity pressure P 2 The method comprises the steps of carrying out a first treatment on the surface of the Thereby controlling the output torque of the hydraulic motor.
As a further improvement on the torque control method of the self-adaptive neural network of the hydraulic excavator, the invention comprises the following steps:
in the comprehensive interference upper bound updater, an estimated value of the comprehensive interference upper boundThe online real-time update is performed as follows:
wherein Γ is γ ，ω，γ max Is a positive real number, and the output is a real number,estimated value for upper bound of integrated interference +.>A differential signal over time;
obtaining an estimated value of the upper bound of the integrated interference by integral operation>
As a further improvement on the torque control method of the self-adaptive neural network of the hydraulic excavator, the invention comprises the following steps:
in the echo state neural network, the internal state variable phi of the neural network is updated on line and in real time as follows:
wherein c, l is a positive real number, W in ，W intl ，W back Respectively input weight, internal weight and output weight;differential signals of internal state variables phi of the neural network with respect to time;
and obtaining the internal state variable phi of the neural network through integral operation.
As a further improvement on the torque control method of the self-adaptive neural network of the hydraulic excavator, the invention comprises the following steps:
in the neural network parameter updater, the neural network outputs an estimated value of the weightThe online real-time update is performed as follows:
n is a positive integer, Γ C Is an N x N positive definite matrix,and is also provided with
Wherein W is imax And W is imin I=1, N.
Estimation of neural network output weight by integral operationValue->
As a further improvement on the torque control method of the self-adaptive neural network of the hydraulic excavator, the invention comprises the following steps:
the control signal generator outputs an estimated value of the weight according to the neural networkEstimated value of internal state variable phi of neural network and comprehensive interference upper bound +.>And a moment tracking error signal e, generating a signal u by adaptive control a Proportional feedback control signal u s1 Robust feedback signal u s2 A control signal u of the six-way open center proportional reversing valve is formed;
adaptive control signal u a Proportional feedback control signal u s1 Robust feedback signal u s2 Can be calculated as follows:
u s1 ＝k 1 e
k 1 and ω is a positive real number;
the control signal u of the six-way open center proportional reversing valve can be obtained according to the following formula:
u＝-u a -u s1 -u s2 。
the self-adaptive neural network moment control device and method for the hydraulic excavator have the technical advantages that:
the invention can avoid the condition that the flow control gain is constant to zero when the moment control is carried out on the hydraulic excavator, thereby enabling the whole system to be controllable; the invention also provides a torque control method of the self-adaptive neural network of the hydraulic excavator, which utilizes the self-adaptability and the global approximation capability of the neural network to fit an unknown system function in a torque control system and improves the fitting effect by updating the weight of the neural network on line; in addition, the method carries out online estimation on the upper bound of the unknown comprehensive interference and reduces the influence of the unknown comprehensive interference on the system performance in a robust control mode. The method is simple and easy to operate, accurate parameters of the system are not required to be known, so that complicated off-line parameter identification is not required, the system has strong parameter self-adaption and robustness, the torque control precision is high, and the method is also suitable for hydraulic pressure and torque control equipment with six-way open center type proportional reversing valves of other main control valves except a hydraulic excavator.
Drawings
The following describes the embodiments of the present invention in further detail with reference to the accompanying drawings.
FIG. 1 is a schematic diagram of a hydraulic excavator adaptive neural network torque control device;
fig. 2 shows the results of torque control experiments.
Detailed Description
The invention will be further described with reference to specific examples, but the scope of the invention is not limited thereto.
Embodiment 1, a hydraulic excavator adaptive neural network torque control device is used for controlling a hydraulic motor 12, and as shown in fig. 1-2, the hydraulic excavator adaptive neural network torque control device comprises a hydraulic pump 1, a proportional relief valve 2, an oil tank 3, a proportional throttle valve 4, a pump pressure sensor 5, a check valve 6, a six-way open center proportional reversing valve 7, a proportional relief valve 8, a proportional relief valve 9, a check valve 10, a check valve 11, a hydraulic motor 12, a hydraulic motor pressure sensor 13 and a hydraulic motor pressure sensor 14.
The six-way open center proportional reversing valve 7 is provided with a valve front oil inlet a, a valve front oil inlet B, a middle-way oil inlet, an A outlet, a B outlet and a middle-way oil outlet.
The connection relation of the hydraulic excavator rotary hydraulic system is as follows:
the inlet of the hydraulic pump 1 is communicated with the oil tank 3, the outlet of the hydraulic pump 1 is divided into three paths, one path is communicated with the oil tank 3 again after passing through the proportional overflow valve 2, and a pump pressure sensor 5 is arranged on a pipeline between the hydraulic pump 1 and the proportional overflow valve 2; one path is communicated with a valve front oil inlet a of the six-way open center proportional reversing valve 7 after passing through the one-way valve 6 (a non-spring cavity oil port and a spring cavity oil port of the one-way valve 6 are respectively connected with an outlet of the hydraulic pump 1 and the valve front oil inlet a of the six-way open center proportional reversing valve 7); one path is communicated with a middle oil inlet of the six-way open center proportional reversing valve 7 after passing through the proportional throttle valve 4 (an inlet and an outlet of the proportional throttle valve 4 are respectively connected with an outlet of the hydraulic pump 1 and the middle oil inlet of the six-way open center proportional reversing valve 7); the middle oil outlet of the six-way open center proportional reversing valve 7 is respectively communicated with the front oil inlet b of the six-way open center proportional reversing valve 7 and the oil tank 3.
The outlet A of the six-way open center proportional reversing valve 7 is respectively communicated with a left cavity oil inlet of the hydraulic motor 12, an inlet of the proportional overflow valve I8 and a spring cavity oil inlet of the one-way valve I10, and an outlet of the proportional overflow valve I8 and a non-spring cavity oil inlet of the one-way valve I10 are respectively communicated with the oil tank 3; a first hydraulic motor pressure sensor 13 is arranged on a pipeline between an A outlet of the six-way open center proportional reversing valve 7 and a left cavity oil inlet of the hydraulic motor 12;
the outlet B of the six-way open center proportional reversing valve 7 is respectively communicated with a right cavity oil inlet of the hydraulic motor 12, an inlet of the proportional overflow valve II 9 and a spring cavity oil inlet of the one-way valve II 11, and an outlet of the proportional overflow valve II 9 and a non-spring cavity oil inlet of the one-way valve II 11 are respectively communicated with the oil tank 3; a second hydraulic motor pressure sensor 14 is arranged on a pipeline between the outlet B of the six-way open center proportional reversing valve 7 and the right cavity oil inlet of the hydraulic motor 12;
the hydraulic pump 1 supplies oil to the hydraulic motor 12 through the six-way open center proportional directional valve 7.
The pump pressure sensor 5 collects the outlet pressure of the hydraulic pump 1, and the first hydraulic motor pressure sensor 13 and the second hydraulic motor pressure sensor 14 collect the pressures of the left cavity and the right cavity of the hydraulic motor 12.
The moment control device and the self-adaptive neural network moment control method of the hydraulic excavator are as follows:
the torque control device comprises a low-pass filter 15, an analog quantity acquisition unit 16, a torque calculator 17, a reference torque signal generator 18, a comparator 19, an echo state neural network 20, an integrated interference upper bound updater 21, a neural network parameter updater 22, a control signal generator 23 and an analog quantity output unit 24;
the pump pressure sensor 5, the first hydraulic motor pressure sensor 13 and the second hydraulic motor pressure sensor 14 are respectively connected with a low-pass filter 15, after filtering, signals are sent to an analog quantity acquisition unit 16, and the analog quantity acquisition unit 16 acquires a hydraulic pump outlet pressure signal P s Left chamber pressure P of hydraulic motor 1 And hydraulic motor right cavity pressure P 2 ；
The analog quantity acquisition unit 16 transmits a hydraulic pump outlet pressure signal P s To the control signal generator 23, the analog acquisition unit 16 applies the hydraulic motor left cavity pressure P 1 And hydraulic motor right cavity pressure P 2 Are sent to the torque calculator 17 and the echo state neural network 20 respectively,
the moment calculator 17 calculates the left chamber pressure P of the hydraulic motor 1 And hydraulic motor right cavity pressure P 2 Calculating to obtain an output torque signal tau, and sending the output torque signal tau to the comparator 19;
the output torque signal τ can be calculated as follows:
wherein D is m Is the displacement of the hydraulic motor 12. In this example, D m ＝45.4cm 3 /r。
The reference torque signal generator 18 generates a reference torque signal τ d And a differential signal of the reference torque signal versus timeAnd will reference moment signal tau d Sends the differential signal of the reference torque signal to time to the comparator 19>To the echo state neural network 20; tau in this example d And->The following formula can be used for calculation:
the comparator 19 compares the output torque signal τ with the reference torque signal τ d A moment tracking error signal e (e=τ - τ) is obtained d ) And transmits the moment tracking error signal e to the control signal generator 23, the integrated interference upper bound updater 21 and the neural network parameter updater 22;
the upper bound updater 21 generates an estimated value of the upper bound on the basis of the moment tracking error signal eAnd the estimated value of the upper bound of the integrated interference is +.>To the control signal generator 23;
wherein the method comprises the steps ofPush-to-follow online real-time update
Wherein the method comprises the steps ofIs->Differential signal with respect to time Γ γ 、ω、γ max Is a positive real number.
In this example, Γ γ ＝0.001，ω＝0.01，γ max ＝2。
The acoustic echo state neural network 20 is based on the hydraulic motor left chamber pressure P 1 Right chamber pressure P of hydraulic motor 2 Differential signal of reference moment signal versus timeAnd the estimated value of the output weight of the neural network +.>The operation obtains the internal state variable phi of the neural network, and sends the internal state variable phi of the neural network to the neural network parameter updater 22 and the control signal generator 23;
wherein Φ is updated online in real time as follows:
wherein the method comprises the steps ofFor differential signals of phi versus time, c, l are positive real numbers, W in 、W intl 、W back Respectively, an input weight, an internal weight and an output weight.
In this example, c=1, l=0.5, w in 400 x 3 random matrix with matrix elements between-1 and 1, W intl A sparse matrix of 400×400, W back Is a 400 x 1 random matrix with matrix elements between-1 and 1.
The neural network parameter updater 22 calculates an estimated value of the neural network output weight according to the internal state variable phi of the neural network and the moment tracking error signal eAnd outputting the estimated value of the weight of the neural network +.>To the echo state neural network 20 and the control signal generator 23;
wherein the method comprises the steps of(N is a positive integer) online real-time update as follows:
wherein the method comprises the steps ofIs->Differential signal with respect to time Γ C Is an N x N positive definite matrix,
and is also provided with
Wherein W is imax And W is imin Is a positive real number.
In this example, n=400, Γ C =0.0005I (I is 400×400 identity matrix), W imax ＝1，W imin ＝-1，i＝1,...,N。
The control signal generator 23 generates a hydraulic pump outlet pressure signal P based on the hydraulic pump outlet pressure signal s And the set relief pressure P of the proportional relief valve 2 set (the control signal generator 23 presets the set relief pressure P of the proportional relief valve 2) set ) Calculating to obtain the proportional throttle valveControl signal u Ps : control signal u of proportional throttle valve Ps Slowly increasing from zero, the opening of the proportional throttle valve 4 slowly decreases, the middle return oil resistance of the hydraulic pump increases, and the outlet pressure P of the hydraulic pump is improved s Continuing to increase the proportional throttle control signal u Ps So that the hydraulic pump outlet pressure signal P s And the set relief pressure P of the proportional relief valve 2 set The same applies. In this example, the proportional throttle control signal u is adjusted Ps At 3.4V, P can be made s Reach P set 26.2MPa.
The control signal generator 23 outputs an estimated value of the weight according to the neural networkEstimated value of internal state variable phi of neural network and comprehensive interference upper bound +.>And a moment tracking error signal e, generating a signal composed of an adaptive control signalProportional feedback control signal u s1 (u s1 ＝k 1 e) Robust feedback signalControl signal u (u= -u) of six-way open center proportional reversing valve a -u s1 -u s2 ) And the proportional throttle control signal u Ps And a six-way open center proportional reversing valve control signal u is sent to the analog output unit 24; wherein k is 1 Is a positive real number, in this example, k 1 ＝0.13。
The analog output unit 24 is respectively connected with the proportional throttle valve 4 and the six-way open center proportional reversing valve 7. Analog output unit 24 outputs proportional throttle control signal u Ps Is sent to the proportional throttle valve 4, the opening size of the proportional throttle valve 4 is controlled, and the change of the opening of the proportional throttle valve 4 can change the return oil resistance of the hydraulic pump, thereby controlling the outlet pressure P of the hydraulic pump s 。
The analog output unit 24 sends a six-way open center proportional directional valve control signal u to the six-way open center proportional directional valve 7 to control the hydraulic motor left chamber pressure P of the left and right chambers of the hydraulic motor 12 1 And hydraulic motor right cavity pressure P 2 The output torque of the hydraulic motor 12 is controlled. The control mode is shown in formulas (7) and (8); six-way open center proportional reversing valve control signal u controls flow Q flowing into right two cavities of hydraulic Ma Dazuo 1 And Q 2 ，Q 1 And Q 2 Control ofAnd->Thereby obtaining P 1 And P 2 。
Flow rate Q flowing into left and right chambers of hydraulic motor 12 1 And Q 2 Can be expressed as:
wherein: g 1 And G 2 The flow control gains of the valve front oil inlet a to the outlet A and the valve front oil inlet B to the outlet B of the six-way open center proportional directional valve 7 are respectively shown as u, K and the six-way open center proportional directional valve control signal q1 And K q2 Positive coefficient (determined experimentally), K in this example q1 =2 and K q2 ＝2，P s For the outlet pressure of the hydraulic pump, P t Is the pressure of the oil in the oil tank 3.
When torque control is performed on the hydraulic motor 12, the angular velocity of the hydraulic motor 12 is zero. The pressure dynamics equation for the left and right chambers of the hydraulic motor 12 can be expressed as:
and->Respectively represent P 1 And P 2 Differential signal over time. Obtain->And->The P can be obtained through integral operation 1 And P 2 . In addition, in the formulae (4) to (6), all +.>And->Re-integration to get +.>And phi.
Wherein beta is e Is the elastic bulk modulus of oil, V 0 For the volume of the left and right chambers (corresponding to one chamber) of the hydraulic motor 12, C tm C is the internal leakage coefficient em1 And C em2 Is the leakage coefficient of the left and right cavities of the hydraulic motor 12.
As can be seen from comparing the formulas (1) and (7), the hydraulic excavator rotary hydraulic system, the moment control device and the hydraulic pump outlet pressure control method provided by the invention are adopted, and the flow control gain G of the system is obtained 1 And G 2 Always greater than zero (only when P 1 ＝P s Or P 2 ＝P s Flow control gain G 1 And G 2 Is zero) which enables torque control of the hydraulic motor 12.
To verify the effectiveness of the apparatus and method of the present invention, we conducted experiments, the results of which are shown in FIG. 2. As can be seen from FIG. 2 (d), the hydraulic pump outlet pressure P s Always than the hydraulic pressure Ma Dazuo cavity pressure P 1 And hydraulic motor right cavity pressure P 2 High (P) s ＞P 1 And P is s ＞P 2 ) Thereby enabling the flow control gain G of the system 1 And G 2 Always greater than zero. As can be seen from fig. 2 (a) and (b), the output torque signal τ of the hydraulic motor 12 can track the reference torque signal τ well d And the torque tracking error signal e is within + -4 Nm.
Finally, it should also be noted that the above list is merely a few specific embodiments of the present invention. Obviously, the invention is not limited to the above embodiments, but many variations are possible. All modifications directly derived or suggested to one skilled in the art from the present disclosure should be considered as being within the scope of the present invention.
Claims (7)
1. The hydraulic excavator self-adaptive neural network moment control device is used for controlling a hydraulic motor (12), and is characterized in that: the hydraulic pump comprises a hydraulic pump (1), a proportional overflow valve (2), an oil tank (3), a proportional throttle valve (4), a one-way valve (6), a six-way open center proportional reversing valve (7), a proportional overflow valve I (8), a proportional overflow valve II (9), a one-way valve I (10), a one-way valve II (11) and a hydraulic motor (12);
the six-way open center proportional reversing valve (7) is provided with a valve front oil inlet a, a valve front oil inlet B, a middle oil inlet, an outlet A, an outlet B and a middle oil outlet;
the inlet of the hydraulic pump (1) is communicated with the oil tank (3), the outlet of the hydraulic pump (1) is divided into three paths, and one path is communicated with the oil tank (3) after passing through the proportional overflow valve (2); one path is communicated with a valve front oil inlet a of a six-way open center proportional reversing valve (7) after passing through a one-way valve (6); one path is communicated with a middle path oil inlet of a six-way open center proportional reversing valve (7) after passing through a proportional throttle valve (4);
the middle oil outlet of the six-way open center proportional reversing valve (7) is respectively communicated with the front oil inlet b of the valve and the oil tank (3);
the outlet A of the six-way open center proportional reversing valve (7) is respectively communicated with a left cavity oil inlet of the hydraulic motor (12), an inlet of the proportional overflow valve I (8) and a spring cavity oil inlet of the one-way valve I (10), and an outlet of the proportional overflow valve I (8) and a non-spring cavity oil inlet of the one-way valve I (10) are respectively communicated with the oil tank (3);
the outlet B of the six-way open center proportional reversing valve (7) is respectively communicated with a right cavity oil inlet of the hydraulic motor (12), an inlet of the proportional overflow valve II (9) and a spring cavity oil inlet of the check valve II (11), and an outlet of the proportional overflow valve II (9) and a non-spring cavity oil inlet of the check valve II (11) are respectively communicated with the oil tank (3);
the torque control device also comprises a low-pass filter (15), an analog quantity acquisition unit (16), a torque calculator (17), a reference torque signal generator (18), a comparator (19), an echo state neural network (20), a comprehensive interference upper bound updater (21), a neural network parameter updater (22), a control signal generator (23) and an analog quantity output unit (24);
a pump pressure sensor (5) is arranged on a pipeline between the hydraulic pump (1) and the proportional overflow valve (2);
a first hydraulic motor pressure sensor (13) is arranged on a pipeline between an A outlet of the six-way open center proportional reversing valve (7) and a left cavity oil inlet of the hydraulic motor (12);
a second hydraulic motor pressure sensor (14) is arranged on a pipeline between the outlet B of the six-way open center proportional reversing valve (7) and the right cavity oil inlet of the hydraulic motor (12);
the pump pressure sensor (5), the hydraulic motor pressure sensor I (13) and the hydraulic motor pressure sensor II (14) are respectively connected with the low-pass filter (15), and after filtering, the low-pass filter (15) sends corresponding signals to the analog acquisition unit (16); the analog quantity acquisition unit (16) acquires and obtains a hydraulic pump outlet pressure signal P s Left chamber pressure P of hydraulic motor 1 And hydraulic motor right cavity pressure P 2 ；
The analog quantity acquisition unit (16) transmits a hydraulic pump outlet pressure signal P s To a control signal generator (23), the analog quantity acquisition unit (16) acquires the left cavity pressure P of the hydraulic motor 1 And hydraulic motor right cavity pressure P 2 Respectively sending to a moment calculator (17) and an echo state neural network (20);
moment calculator (17) calculates pressure P of the cavity according to the hydraulic pressure Ma Dazuo 1 And hydraulic motor right cavity pressure P 2 Calculating to obtain an output torque signal tau, and sending the output torque signal tau to a comparator (19);
the reference torque signal generator (18) generates a reference torque signal tau d And a differential signal of the reference torque signal versus timeAnd will reference moment signal tau d Sends to a comparator (19) a differential signal of the reference torque signal with respect to time +.>Transmitting to an echo state neural network (20);
the comparator (19) is based on the output torque signal tau and the reference torque signal tau d Obtaining a moment tracking error signal e, and sending the moment tracking error signal e to an echo state neural network (20), an integrated interference upper bound updater (21) and a neural network parameter updater (22);
the upper bound updater (21) generates an estimated value of the upper bound of the integrated disturbance based on the torque tracking error signal eAnd the estimated value of the upper bound of the integrated interference is +.>Send to the control signal generator (23);
the echo state neural network (20) is based on the hydraulic Ma Dazuo cavity pressure P 1 Right chamber pressure P of hydraulic motor 2 Differential signal of reference moment signal versus timeAnd the estimated value of the output weight of the neural network +.>Calculating to obtain a neural network internal state variable phi, and sending the neural network internal state variable phi to a neural network parameter updater (22) and a control signal generator (23);
the neural network parameter updater (22) calculates to obtain the estimated value of the output weight of the neural network according to the internal state variable phi of the neural network and the moment tracking error signal eAnd outputting the estimated value of the weight of the neural network +.>To the echo state neural network (20) and to the control signal generator (23);
the control signal generator (23) is based on the hydraulic pump outlet pressure signal P s And a set relief pressure P preset in a control signal generator (23) set The proportional throttle valve control signal u is obtained by operation Ps The method comprises the steps of carrying out a first treatment on the surface of the A control signal generator (23) outputs an estimated value of the weight according to the neural networkEstimated value of internal state variable phi of neural network and comprehensive interference upper bound +.>And a torque tracking error signal e, calculating to obtain a six-way open center proportional reversing valve control signal u, and calculating the proportional throttle valve control signal u Ps And a six-way open center proportional reversing valve control signal u is sent to an analog output unit (24);
the analog quantity output unit (24) is respectively connected with the proportional throttle valve (4) and the six-way open center proportional reversing valve (7); an analog output unit (24) outputs a proportional throttle control signal u Ps Is sent to the proportional nodeA flow valve (4) for controlling the opening size of the proportional throttle valve (4) to thereby control the hydraulic pump outlet pressure P s ；
The analog output unit (24) sends a six-way open center proportional reversing valve control signal u to the six-way open center proportional reversing valve (7) so as to control the left cavity pressure P of the hydraulic motor (12) in the left cavity and the right cavity of the hydraulic motor 1 And hydraulic motor right cavity pressure P 2 The output torque of the hydraulic motor (12) is controlled.
2. A torque control method of a hydraulic excavator adaptive neural network using the torque control device according to claim 1, characterized in that: six-way open center proportional reversing valve control signal u controls flow Q flowing into right two cavities of hydraulic Ma Dazuo 1 And Q 2 Flow rate Q flowing into right two chambers of hydraulic pressure Ma Dazuo 1 And Q 2 Control ofAnd->Controlling hydraulic motor left chamber pressure P 1 And hydraulic motor right cavity pressure P 2 Thereby controlling the output torque of the hydraulic motor (12), and (2)>And->Respectively represent the pressure P of the hydraulic Ma Dazuo cavity 1 And hydraulic motor right cavity pressure P 2 Differential signal over time.
3. The hydraulic excavator adaptive neural network torque control method according to claim 2, characterized in that the specific process of controlling the output torque of the hydraulic motor (12) is:
flow rate Q into right two chambers of hydraulic pressure Ma Dazuo 1 And Q 2 Can be expressed as:
wherein: g 1 And G 2 The flow control gains of the front oil inlet a to the outlet A and the front oil inlet B to the outlet B of the six-way open center proportional reversing valve (7) are respectively shown as u, which is a control signal of the six-way open center proportional reversing valve, K q1 And K q2 Is positive coefficient, P s For the outlet pressure of the hydraulic pump, P t Is the pressure of the oil in the oil tank (3);
the dynamic equation of the pressure of the left and right cavities of the hydraulic motor (12) is expressed as:
β e is the elastic bulk modulus of oil, V 0 Is the volume of the left cavity and the right cavity of the hydraulic motor (12), C tm C is the internal leakage coefficient em1 And C em2 The leakage coefficients of the left cavity and the right cavity of the hydraulic motor (12);
and->The hydraulic motor left cavity pressure P is obtained through integral operation 1 And hydraulic motor right cavity pressure P 2 The method comprises the steps of carrying out a first treatment on the surface of the Thereby controlling the output torque of the hydraulic motor (12).
4. The hydraulic shovel adaptive neural network torque control method according to claim 3, wherein:
in the integrated interference upper bound updater (21), an estimated value of an integrated interference upper boundThe online real-time update is performed as follows:
wherein Γ is γ ，ω，γ max Is a positive real number, and the output is a real number,estimated value for upper bound of integrated interference +.>A differential signal over time;
obtaining an estimated value of the upper bound of the integrated interference by integral operation>
5. The hydraulic shovel adaptive neural network torque control method according to claim 4, wherein:
in an echo state neural network (20), a neural network internal state variable Φ is updated online in real time as follows:
wherein c, l is a positive real number, W in ，W intl ，W back Respectively input weight, internal weight and output weight;differential signals of internal state variables phi of the neural network with respect to time;
and obtaining the internal state variable phi of the neural network through integral operation.
6. The hydraulic shovel adaptive neural network torque control method according to claim 5, wherein:
in the neural network parameter updater (22), the neural network outputs an estimated value of the weightThe online real-time update is performed as follows:
n is a positive integer, Γ C Is an N x N positive definite matrix,and is also provided with
Wherein W is imax And W is imin I=1, N;
obtaining an estimated value of the output weight of the neural network by integral operation>
7. The hydraulic shovel adaptive neural network torque control method according to claim 6, wherein:
a control signal generator (23) outputs an estimated value of the weight according to the neural networkEstimated value of internal state variable phi of neural network and comprehensive interference upper bound +.>And a moment tracking error signal e, generating a signal u by adaptive control a Proportional feedback control signal u s1 Robust feedback signal u s2 A control signal u of the six-way open center proportional reversing valve is formed;
adaptive control signal u a Proportional feedback control signal u s1 Robust feedback signal u s2 Can be calculated as follows:
u s1 ＝k 1 e
k 1 and ω is a positive real number;
the control signal u of the six-way open center proportional reversing valve can be obtained according to the following formula:
u＝-u a -u s1 -u s2 。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810039705.2A CN108118743B (en) | 2018-01-16 | 2018-01-16 | Hydraulic excavator self-adaptive neural network moment control device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810039705.2A CN108118743B (en) | 2018-01-16 | 2018-01-16 | Hydraulic excavator self-adaptive neural network moment control device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108118743A CN108118743A (en) | 2018-06-05 |
| CN108118743B true CN108118743B (en) | 2023-07-18 |
Family
ID=62232819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810039705.2A Active CN108118743B (en) | 2018-01-16 | 2018-01-16 | Hydraulic excavator self-adaptive neural network moment control device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108118743B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004053337A1 (en) * | 2002-12-10 | 2004-06-24 | Shin Caterpillar Mitsubishi Ltd. | Automatic booster for working machines |
| CN102493517A (en) * | 2011-12-07 | 2012-06-13 | 哈尔滨工业大学 | Slewing system for hybrid hydraulic excavator and driving and braking method for slewing system |
| CN102912823A (en) * | 2012-11-09 | 2013-02-06 | 浙江大学 | Rotary energy saving system of excavator |
| WO2014063490A1 (en) * | 2012-10-26 | 2014-05-01 | 中联重科股份有限公司 | Hydraulic system for controlling boom to rotate and control method therefor and concrete pumping equipment |
| JP2015178863A (en) * | 2014-03-19 | 2015-10-08 | カヤバ工業株式会社 | Control system of hybrid construction machine |
| CN208137007U (en) * | 2018-01-16 | 2018-11-23 | 浙江大学 | Hydraulic crawler excavator adaptive neural network moment control device |
-
2018
- 2018-01-16 CN CN201810039705.2A patent/CN108118743B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004053337A1 (en) * | 2002-12-10 | 2004-06-24 | Shin Caterpillar Mitsubishi Ltd. | Automatic booster for working machines |
| CN102493517A (en) * | 2011-12-07 | 2012-06-13 | 哈尔滨工业大学 | Slewing system for hybrid hydraulic excavator and driving and braking method for slewing system |
| WO2014063490A1 (en) * | 2012-10-26 | 2014-05-01 | 中联重科股份有限公司 | Hydraulic system for controlling boom to rotate and control method therefor and concrete pumping equipment |
| CN102912823A (en) * | 2012-11-09 | 2013-02-06 | 浙江大学 | Rotary energy saving system of excavator |
| JP2015178863A (en) * | 2014-03-19 | 2015-10-08 | カヤバ工業株式会社 | Control system of hybrid construction machine |
| CN208137007U (en) * | 2018-01-16 | 2018-11-23 | 浙江大学 | Hydraulic crawler excavator adaptive neural network moment control device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108118743A (en) | 2018-06-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2109718B1 (en) | System and method for controlling actuator position | |
| US8296024B2 (en) | Calibrating transmission motor torque | |
| Borghi et al. | Energy savings in the hydraulic circuit of agricultural tractors | |
| Focchi et al. | Water/air performance analysis of a fluidic muscle | |
| US20140129035A1 (en) | Excess Flow Control Valve Calibration Method | |
| DE102016109100A1 (en) | System for estimating an adjustment of a pump | |
| US9920780B2 (en) | Slewing drive apparatus for construction machine | |
| US9328821B2 (en) | Hydrostatic drive system | |
| Aschemann et al. | Decentralised flatness-based control of a hydrostatic drive train subject to actuator uncertainty and disturbances | |
| US10066555B2 (en) | Hydraulic system and method for controlling same | |
| US9404516B1 (en) | System for estimating a sensor output | |
| CN108118743B (en) | Hydraulic excavator self-adaptive neural network moment control device and method | |
| Wu | Modeling and experimental evaluation of a load-sensing and pressure compensated hydraulic system | |
| CN112196856A (en) | Hydraulic system of electric transmission equipment and control method thereof | |
| WO2016027463A1 (en) | Hydraulic pump drive system | |
| US9534616B2 (en) | System for estimating a sensor output | |
| Sun et al. | Sliding-mode control of a hydrostatic drive train with uncertain actuator dynamics | |
| Sun et al. | Robust inverse dynamics control for a hydrostatic transmission with actuator uncertainties | |
| WO2016180689A1 (en) | Device and method for controlling work machine | |
| Campanini et al. | Optimized control for an independent metering valve with integrated diagnostic features | |
| CN208137007U (en) | Hydraulic crawler excavator adaptive neural network moment control device | |
| Aschemann et al. | Model-based nonlinear trajectory control of a drive chain with hydrostatic transmission | |
| JP2009282819A (en) | Flow control method | |
| JP2010026576A5 (en) | ||
| Kassen et al. | Combine harvester header height control via robust feedback linearization |
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 |