CN103529704B - The netted array artificial-muscle design of class skeletal muscle straight line and the neural control method of class - Google Patents

Abstract

The present invention is based on engineering bionics principle, on the basis that microcosmic angle analyses in depth animals skeletal muscle physiology formation and expulsion mechanism, simulation skeletal muscle micromechanism, set up a kind of novel array artificial-muscle tactic pattern, and according to its design feature, in conjunction with skeletal muscle micromechanism and nervous system to the control type of drive of skeletal muscle, the driving process of key design artificial-muscle and control method, make this artificial-muscle more press close to biological skeletal muscle in structure, forms of motion, control method, the driving process etc.

Description

The netted array artificial-muscle design of class skeletal muscle straight line and the neural control method of class

Technical field

The present invention relates to robot Driving technique and bionics techniques, particularly relate to a kind skeletal muscle straight line netted array artificial-muscle construction design method and the neural control method of class thereof.

Background technology

Along with legged type robot is applied more and more widely under the occasions such as battlefield, anti-terrorism, rescue, space exploration, hazardous environment operation, the jump of robot, run, leaping over obstacles, the steadily exercise performance and anti-foreign impacts and complicated landform adaptive faculty such as to advance cause the attention of people gradually.Thus, more and more higher requirement is proposed to the instantaneous moment of joint of robot, the performance such as hunting frequency and power density, and the type of drive of " electric rotating machine+gear train " that current joint of robot generally adopts is because of the restriction by the power-acceleration characteristic of motor itself and the dynamics of gear train, be difficult to meet military and civilian complicated applications environment to the requirement of high-performance joint motions, redundancy large in the urgent need to instantaneous acceleration be strong, the next-generation drive of fast response time and aspect the has superior function such as power density is high.

The joint of animal that skeletal muscle drives is not only simple for structure compact, there is the performance that traditional articulated driving equipment is difficult to reach simultaneously in energy storage, surge capability, speed, agility, energy density, prompt explosion power etc., the human body knee joint being no more than 1Kg as weight can provide 10 times to the driving force of human body weight, the running speed of cheetah is up to speed per hour 130 kilometers, and the biting force of esturaine crocodile is close to 1900Kg.Animals skeletal muscle has logical hierarchical architecture, and the size of initiatively shrinking deflection and the power output produced is made up of its interior microscopic that structure collaborate completes.This hierarchical structure makes animals skeletal muscle have high redundancy, even if skeletal muscle small portion micromechanism forms because the reasons such as damage, pathology can not work, also little to entire effect.The features such as instantaneous acceleration is large, fast response time, the coordination ability are strong, agility is high, self-regulation structure parameter that the hierarchical architecture of skeletal muscle makes skeletal muscle also have.Skeletal muscle has a lot of excellent properties as vertebrate motion muscle, use bionics principle, simulate artificial-muscle and the driving governor thereof of the structure of biological skeletal muscle and expulsion mechanism design, excellent properties and the concision and compact structure of biological skeletal muscle can be obtained.

Today, most artificial-muscle for human simulation object with muscle group power output, displacement and control method, and few carry out analyzing to skeletal muscle micromechanism and micromanagement expulsion mechanism thereof and studies.Simulated animal bone machine hierarchical architecture carries out artificial-muscle Model Design, and artificial-muscle can be made to have good redundancy and better serviceability.For these problems that modern driver exists, the present invention simulates skeletal muscle micromechanism, a kind of line array column artificial-muscle tactic pattern is proposed, make it more press close to biological skeletal muscle in structure, forms of motion, control method, driving process etc., have that instantaneous acceleration is large, redundancy good, the coordination ability is strong, fast response time and power density high.

Summary of the invention

There is the problem that the performances such as redundancy, instantaneous moment, hunting frequency and power density can not meet modern machines people needs for driver in prior art, the invention provides the netted array artificial-muscle design of class skeletal muscle straight line and the neural control method of class.

In order to realize above-mentioned task, the present invention adopts following technical solution:

The method for designing of the netted array artificial-muscle of one kind skeletal muscle straight line, class skeletal muscle straight line netted array artificial-muscle is comprised multiple class half muscle segment driver by the method, multiple class half muscle segment driver forms class muscle fibril driver through series connection, with a class muscle fibril driver for symcenter, other class muscle fibril driver laid out in parallel are formed hexagonal honeycomb meshed symmetric array architecture, all class muscle fibril drivers are aggregated into one respectively by class sarolemma connector by the same layer class half muscle segment driver on each class muscle fibril driver, and polymerization all-in-one-piece class muscle fibril driver two ends are provided with class aponeurosis (aponeuroses) connector formation artificial-muscle, the columns m of such skeletal muscle straight line netted array artificial-muscle and the determination of number of plies n are by following technical step:

(1) maximum target output displacement L is provided according to artificial-muscle application and applied environment _asmaxand the maximum target power output F needed during artificial-muscle motion _asmax;

(2) calculating artificial-muscle parallel-connection counts n, n is layer by layer natural number, if the maximum output displacement of class half muscle segment driver is x _smax, then have

$\frac{L_{\mathrm{as}\max }}{x_{s\max }}\&le;n<\frac{L_{\mathrm{as}\max }}{x_{s\max }}+1---\left(1\right)$

(3) calculating artificial-muscle series connection layer minimum columns k, k is natural number, if the minimum power output f of class half muscle segment driver _smin, then have

$\frac{F_{\mathrm{as}\max }}{f_{s\min }}\&le;k<\frac{f_{\mathrm{as}\max }}{f_{s\min }}+1---\left(2\right)$

(4) calculate netted hexagonal number of total coils c (c is natural number) in artificial-muscle series connection layer, c need meet following formula

$1+6\&CenterDot;\underset{i=2}{\overset{c-1}{\mathrm{\&Sigma;}}}(i-2)\&le;k\&le;1+6\&CenterDot;\underset{i=2}{\overset{c}{\mathrm{\&Sigma;}}}(i-2)---\left(3\right)$

(5) finally determine that artificial-muscle array architecture is columns m is natural number, the number of plies $n=\frac{L_{\mathrm{as}\max }}{x_{s\max }}.$

The present invention also has following other technologies feature:

All class muscle fibril drivers are aggregated into one respectively by class sarolemma connector by the same layer class half muscle segment driver on described each class muscle fibril driver, all kinds of sarolemma connector successively equidistant laid out in parallel.

Polymerization all-in-one-piece class muscle fibril driver two ends are provided with class aponeurosis (aponeuroses) connector, and class aponeurosis (aponeuroses) connector is hexagonal structure.

The neural control method of class of the netted array artificial-muscle of class skeletal muscle straight line of the present invention, comprises the steps:

The first step, according in the control time Δ t that artificial-muscle top level control device resolves, artificial-muscle target output displacement L _asand target power output F during its motion _as;

Second step, advanced row artificial-muscle row control, and calculate artificial-muscle parallel-connection key-course number of plies n, n is natural number, if the maximum output displacement of class half muscle segment driver is x _smax, then have

$\frac{L_{\mathrm{as}}}{x_{s\max }}\&le;n<\frac{L_{\mathrm{as}}}{x_{s\max }}+1---\left(4\right)$

3rd step, select key-course according to power output minimum principle, when moving according to artificial-muscle, the kinetics equation of class half muscle segment driver, judges to produce electromagnetic force f in its motion process _{s (j)}(electromagnetic force f _{s (i) (j)}for all classes half muscle segment driver be in shunt layer j produces total electromagnetic force) monotonicity and work shift x _sselect artificial-muscle key-course, if the electromagnetic force f that class half muscle segment driver produces _{s (j)}for Parallel Control counts the increasing function of j layer by layer, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle starting point layer is key-course; Otherwise, if electromagnetic force f _{s (j)}during subtraction function for variable j, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle stop layer is key-course;

4th step, in control time Δ t, in key-course, the output displacement of each class half muscle segment driver is artificial muscle target output displacement L _as, the class half muscle segment driver in non-controlling layer is in self-locking state, no-output displacement;

5th step, carries out the capable control of artificial-muscle, the work shift x of real-time measure and control layer _s, determine the maximum power output f under this work shift _xmax(x _s);

$f_{\mathrm{xmsx}}\left(x_s\right)=f(x_s,I_{\max })=\frac{d{W}_m(x_s,I_{\mathrm{mas}})}{d{x}_s}=\frac{\&PartialD;}{\&PartialD;x_s}\left[{\&Integral;}_V{\&Integral;}_0^H(B\&CenterDot;\mathrm{dH})\mathrm{dV}\right]---\left(17\right)$

Wherein, I _maxfor the peak coil current of class half muscle segment driver, W _m(x _s, I _max) be magnetic field energy, x _sfor work shift, H is magnetic field intensity, and B is magnetic induction density, and V is the integrating range of volume integral;

6th step, according to target power output F _asdetermine that artificial-muscle series connection layer activates number n _yfollowing formula need be met

f _xmax(x _s)·(n _y-1)≤F _as≤f _xmax(x _s)·n _y(18)

7th step, the reticulate texture according to artificial-muscle belongs to centrosymmetric image, adopts to activate to activate according to centrosymmetric mode, is artificial muscle target power output F by the summation of all activated artificial-muscle series connection layer power output _as, often pair of artificial-muscle series connection layer class power output Central Symmetry is equal;

8th step, passes to controller by each class half muscle segment driver resolved, and by the R-C power in Shared control and Position Hybrid Control, the power and the displacement that complete artificial-muscle in the instruction cycle export task.

The present invention is based on engineering bionics principle, on the basis that microcosmic angle analyses in depth animals skeletal muscle physiology formation and expulsion mechanism, simulation skeletal muscle micromechanism, set up a kind of novel array artificial-muscle tactic pattern, and according to its design feature, in conjunction with skeletal muscle micromechanism and nervous system to the control type of drive of skeletal muscle, the driving process of key design artificial-muscle and control method, make this artificial-muscle more press close to biological skeletal muscle in structure, forms of motion, control method, the driving process etc.

Artificial-muscle of the present invention is consisted of class half muscle segment driver connection in series-parallel, and is not moved by gear train Direct Drive Robot, the features such as thus have execution efficiency high, the life-span is long, and redundancy is good, fast response time, and instantaneous acceleration is large.

Accompanying drawing explanation

Fig. 1 is artificial-muscle connector and reticulate texture numbering thereof;

Fig. 2 is artificial-muscle wiring layout, and wherein 2-1 is class muscle fibril driver, and 2-2 is class aponeurosis (aponeuroses) connector, and 2-3 is that class half muscle segment drives gas, and 2-4 is class sarolemma connector;

Fig. 3 is artificial-muscle construction design method process flow diagram;

Fig. 4 is class sarolemma connector construction figure, and wherein 4-1 is connecting thread hole, and 4-2 is cylindrical hole;

Fig. 5 is class aponeurosis (aponeuroses) connector construction figure, and wherein 5-1 is connecting thread hole, and 5-2 is cylindrical hole, and 5-3 is hinge hole, 5-4 connecting link;

Fig. 6 is artificial-muscle antenna array control schematic diagram;

Fig. 7 detects position from judging pick-up unit hardware structure diagram;

Fig. 8 is the neural control flow chart of artificial muscle's meat;

Fig. 9 is that artificial-muscle drives quadruped robot wiring layout;

Figure 10 is quadruped robot hip joint dynamics cycle simulation curve;

Figure 11 is quadruped robot hip joint artificial-muscle dynamics cycle simulation curve;

Figure 12 is artificial-muscle power output and output displacement conversion schematic diagram

Figure 13 is quadruped robot knee joint endoprosthesis angular motion mechanics cycle simulation curve;

Figure 14 is quadruped robot knee joint artificial-muscle dynamics cycle simulation curve.

The present invention is described in further detail for the embodiment provided below in conjunction with accompanying drawing and inventor.

Embodiment

1. class skeletal muscle straight line netted array artificial-muscle construction design method and redundancy analysis

For avoiding the repeatability of skeletal muscle traditional batch pattern on function and structure, the skeletal muscle half muscle segment connection in series-parallel tier array pattern simplified is proposed, this structural design bionical artificial-muscle connection in series-parallel array structure: class half muscle segment driver class muscle fibril in series driver, become the netted parallel connection of honeycomb fashion to form artificial-muscle by class muscle fibril driver by class aponeurosis (aponeuroses) connector again, its two ends are connected with robot links by class sarolemma connector.The linear electromagnetic class half muscle segment driver that artificial-muscle selects exercise performance good with controlling respective performances is minimum movement unit, its performance is see Li Jing, Qin Xiansheng, Zhang Xuefeng etc. linear electromagnetic drives the design studies of connection in series-parallel array artificial-muscle, China Mechanical Engineering, 23 (08): 883-887,2012.On the basis of artificial-muscle kinetic model, simulate biological nervous system to the control mechanism of skeletal muscle, the neural control method of derivation artificial muscle meat.

Artificial-muscle designs based on bionics principle, first will carry out Simplified analysis to biological skeletal muscle.Skeletal muscle can be divided into 5 layers from macroscopic view to microcosmic, is respectively skeletal muscle, and------muscle fibre---muscle fibril---half muscle segment, because this tactic pattern has repeatability on function and structure, therefore needs during Bionic Design to carry out simplify processes muscle bundle.The structure of skeletal muscles simplified needs to simplify myoarchitecture level, retains the groundwork structure of skeletal muscle again.Owing to all being isolated by different sarolemmas between muscle fibril, between muscle fibre, between muscle bundle and being connected, sarolemma belongs to connective tissue, can not by nervous system ACTIVE CONTROL, and its structure and fuction is similar, therefore carries out compression merging treatment to this part.After simplification, skeletal muscle can be regarded as and consists of sarolemma effect parallel connection muscle fibril, and muscle fibril is in series by sarcomere, and namely skeletal muscle can be regarded as and is made up of by acting on connection in series-parallel half muscle segment.

Artificial-muscle array formatted analog Skeletal Muscle fibrillar structure is designed to netted, as shown in Figure 1.This similar honeycomb, has high strength and high material space ratio.The class muscle fibril driver that definition is in symcenter is numbering 1, be in initial first lap, what be centered around class muscle fibril driver 1 has other 6 class muscle fibril drivers, their positions are the second circle, by clock-wise order be followed successively by numbering 2 to 7(wherein numbering 2 be in upper-right position).By that analogy, c circle (c is netted hexagonal number of total coils in artificial-muscle series connection layer) has 6 (c-1) individual class muscle fibril driver, is followed successively by numbering by clock-wise order the class muscle fibril drive number being in upper-right position is wherein i>=2.

See Fig. 2, the micromechanism that artificial-muscle simulation simplifies skeletal muscle designs: class half muscle segment driver class muscle fibril in series driver, and class muscle fibril driver is by class sarolemma connector, artificial-muscle is formed see Fig. 3 parallel connection, described class sarolemma connector is hexagonal structure, class sarolemma connector center is provided with connecting thread hole 4-1, class sarolemma connector six drift angles are respectively arranged with cylindrical hole 4-2.At polymerization two ends, all-in-one-piece class muscle fibril driver two ends by class aponeurosis (aponeuroses) connector, be connected with robot links see Fig. 4, described class aponeurosis (aponeuroses) connector is hexagonal structure, two connecting link 5-4 have been arranged in parallel at class sarolemma connector center, connecting link is provided with hinge hole 5-3, class sarolemma connector six drift angles are respectively arranged with cylindrical hole 5-2 and threaded connection hole 5-1, and wherein threaded connection hole 5-1 is positioned at and connects in the middle of a cylindrical hole 5-2.

Wherein, the selection of class half muscle segment driver directly affects the exercise performance of artificial-muscle, and straight line voice coil loudspeaker voice coil electromagnetic principle design class half muscle segment driver (meet each other document Li Jing, Qin Xiansheng, Zhang Xuefeng etc. linear electromagnetic drives the design studies of connection in series-parallel array artificial-muscle, China Mechanical Engineering, 23 (08): 883-887,2012) at aspect of performances such as response speed, efficiency and power densities, but there is very large advantage, meet the particular/special requirement of robot to artificial-muscle in instantaneous acceleration etc.As shown in Figure 3, specific design process is as follows for artificial-muscle construction design method process flow diagram:

The first step, first estimates maximum target output displacement L according to artificial-muscle application and applied environment _asmaxand the maximum target power output F needed during artificial-muscle motion _asmax;

Second step, it is natural number that calculating artificial-muscle parallel-connection counts n(n layer by layer), if the maximum output displacement of class half muscle segment driver is x _smax, then have

$\frac{L_{\mathrm{as}\max }}{x_{\mathrm{samx}}}\&le;n<\frac{L_{\mathrm{as}\max }}{x_{s\max }}+1---\left(1\right)$

3rd step, calculating the minimum columns k(k of artificial-muscle series connection layer is natural number), if the minimum power output f of class half muscle segment driver _smin, then have

$\frac{F_{\mathrm{as}\max }}{f_{s\min }}\&le;k<\frac{F_{\mathrm{as}\max }}{f_{s\min }}+1---\left(2\right)$

4th step, calculating netted hexagonal number of total coils c(c in artificial-muscle series connection layer is natural number) following formula need be met

$1+6\&CenterDot;\underset{i=2}{\overset{c-1}{\mathrm{\&Sigma;}}}(i-2)\&le;k\&le;1+6\&CenterDot;\underset{i=2}{\overset{c}{\mathrm{\&Sigma;}}}(i-2)---\left(3\right)$

5th step, finally determines that artificial-muscle array architecture is columns (m is natural number), the number of plies

This biomimetic features design, makes artificial-muscle have good redundancy, some class half muscle segment drive corruption, can not cause too much influence to artificial-muscle motion.As the class half muscle segment driver that is positioned at jth floor No. 2 position because the reasons such as self-locking coil open circuit are locked, can not move, then this layer of class half muscle segment drives it to be all in self-locking state, this layer does not participate in displacement exporting change, other layer of change in displacement is not impacted, the quantitative change of artificial-muscle overall shrinkage is short, still can use within the specific limits.What the class half muscle segment driver moving coil open circuit as being positioned at jth floor No. 2 position caused cannot produce power output, for avoiding because export asymmetric, a moment of torsion can be produced, reduce the efficiency of artificial-muscle, then get its class half muscle segment driver with No. 5 positions, floor Central Symmetry position when this floor moves, all the time relative sliding state is in, therefore during this layer of change in displacement, No. 2 and No. 5 position no-output power, do not affect the work of other class half muscle segment drivers, the total power output of artificial-muscle diminishes, and still can work under certain condition, ensure that good redundancy.

2. the neural control method design of class

Artificial-muscle control effects is mainly manifested on power output and output displacement, and power output and output displacement are determined jointly by all class half muscle segment drivers.Artificial-muscle control object is: each class half muscle segment driver status, electromagnetism power output and output displacement.By capable for artificial-muscle called after artificial-muscle series connection layer (i.e. class muscle fibril driver), artificial-muscle row called after artificial-muscle parallel-connection layer, to connect with jth layer if class half muscle segment driver is in the i-th shunt layer, then uses A (j, i) represent, as shown in Figure 5.The length variations of artificial-muscle is equivalent to the summation (i.e. the length variations of single class muscle fibril driver) of artificial-muscle parallel-connection layer change, the control that artificial-muscle parallel-connection layer carries out is called that row control, its power output is equivalent to the sum total of artificial-muscle series connection layer power output, is called that row controls to the control that artificial-muscle series connection layer carries out.

Artificial-muscle two ends are connected with robot links by class aponeurosis (aponeuroses) connector, and wherein movable less one end is called artificial-muscle starting point, and the artificial-muscle parallel-connection layer near starting point is called artificial-muscle starting point layer; Otherwise movable larger one end is called artificial-muscle stop, and the artificial-muscle parallel-connection layer near starting point is called artificial-muscle stop layer.See in list of references [3] to the design of class half muscle segment driver: if class half muscle segment driver is in self-locking state, under its contracting brake mechanism spring action, mover and stator geo-stationary, do not have power consumption; Be in non-self-lock-ing state, band-type brake coil electricity, mover and stator can relative motions, and contracting brake mechanism consumes stationary electric.Class half muscle segment driver status is comprehensively determined by queues condition, and its ranks state of a control is integrated as shown in Figure 6.

The first rank of advanced units state of a control of artificial-muscle is analyzed: class in shunt layer half muscle segment Drive Status is divided into self-locking state and non-self-lock-ing state.The class half muscle segment driver band brake apparatus being in self-locking state starts, and does not produce output displacement; The class half muscle segment driver band brake apparatus being in non-self-lock-ing state does not start, and produces initiatively power output.On the basis of row state analysis, the analysis of row state of a control is carried out to the class half muscle segment driver being in non-self-lock-ing state: class half muscle segment Drive Status in series connection layer is divided into state of activation and unactivated state.Be in the class half muscle segment driver active slip of state of activation, initiatively power output can be produced; Be in the passive slip of class half muscle segment driver of unactivated state, do not produce initiatively power output.

Simulation skeletal muscle excitation-contraction coupling mechanism designs row control: electrical excitation signal inputs from the joint of nerve-skeletal muscle, by transverse tubule system to myocyte depths unidirectional delivery, controls half muscle segment and shrinks successively.Therefore, finding at artificial-muscle place an artificial-muscle parallel-connection layer for controlling the first floor, from this layer, controlling other shunt layers and moving successively.

The neural control flow chart of artificial muscle's meat is as shown in Figure 8:

The first step, according in the control time Δ t that artificial-muscle top level control device resolves, artificial-muscle target output displacement L _asand target power output F during its motion _as;

Second step, advanced row artificial-muscle row control, and calculating artificial-muscle parallel-connection key-course number of plies n(n is natural number), if the maximum output displacement of class half muscle segment driver is x _smax, then have

$\frac{L_{\mathrm{as}}}{x_{s\max }}\&le;n<\frac{L_{\mathrm{as}}}{x_{s\max }}+1---\left(4\right)$

3rd step, select key-course according to power output minimum principle, when moving according to artificial-muscle, the kinetics equation of class half muscle segment driver, judges to produce electromagnetic force f in its motion process _{s (j)}monotonicity and work shift x _sselect artificial-muscle key-course.If the electromagnetic force f that class half muscle segment driver produces _{s (j)}for Parallel Control counts the increasing function of j layer by layer, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle starting point layer is key-course; Otherwise, if electromagnetic force f _{s (j)}during subtraction function for variable j, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle stop layer is key-course;

4th step, in control time Δ t, in key-course, the output displacement of each class half muscle segment driver is artificial muscle target output displacement L _as, the class half muscle segment driver in non-controlling layer is in self-locking state, no-output displacement;

5th step, carries out the capable control of artificial-muscle subsequently, the work shift x of real-time measure and control layer _s, determine the maximum power output f under this work shift _xmax(x _s).The electromagnetic force computation process that class half muscle segment driver exports is as follows:

Class half muscle segment driver is work shift x _swith the function of working coil current i, assuming that coil current is that under the condition of definite value, magnetic field of permanent magnet is constant, solving of magnetic field force can carry out model solution according to Magnetostatic Field Problems.Maxwell equation is the mathematical model for all macroscopical electromagnet phenomenons, is the basis of Theory of Electromagnetic Field, is also the starting point of the numerical analysis of all kinds Electromagnetic Field and calculating.Be can be expressed as by Ampere circuit law (i.e. circuital law):

Wherein, H is magnetic field intensity, and J is conduction current density vector, for displacement current density.And under static magnetic field condition, therefore have ▽ × H=J, by Maxwell equation (i.e. the principle of continuity of magnetic flux) be:

▽·B＝0(6)

Again because the constitutive relation of electromagnetic field field amount can be derived:

$H=\frac{B}{{\mathrm{\&mu;}}_r{\mathrm{\&mu;}}_0}---\left(7\right)$

Wherein μ _rfor solving the relative permeability of material, μ ₀for permeability of vacuum, B is magnetic induction density.Introduce vector potential A, then have:

B＝▽×A(8)

Can derive:

$J=\&dtri;\&times;(\frac{1}{{\mathrm{\&mu;}}_r{\mathrm{\&mu;}}_0}\&dtri;\&times;A)---\left(9\right)$

According to law of conservation of energy, input electric energy W _eequal magnetic field energy W _mwith the energy W being input to mechanical ports _emsum, then have:

W _e＝W _m+W _em(10)

Utilize the principle of virtual work to calculate electromagnetic force, magnetic field energy is calculated by the potential function in magnetic field to obtain, and FEM (finite element) calculation can be expressed as:

$W_m=\frac{1}{2}{\&Integral;}_{V}B\&CenterDot;\mathrm{HdV}---\left(11\right)$

Adopt the principle of virtual work to calculate magnetostatic power, assuming that under the effect of electromagnetic force, occur relative displacement ds between mover and stator, then coupled magnetic field energy theorem is write as the form of differential:

dW _e=dW _m+dW _em=dW _m+F _emds(12)

DW _e, dW _mwith dW _emrepresent the clean electric energy inputting coupled field in time dt, magnetic field absorbs gross energy and the gross energy being converted into mechanical energy, F _emfor the electromagnetic force that mechanical ports exports, s is the displacement that mechanical ports exports.

For solving of static magnetic field system problem, if system is made up of n loop, in i-th loop, magnetic linkage is changed to d ψ _i, size of current is definite value I _i, then dW _ethe energy that the induction electromotive force produced for resisting magnetic linkage change by power supply produces can be expressed as:

$d{W}_e=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{I}_i\frac{d{\mathrm{\&psi;}}_i}{\mathrm{dt}}\mathrm{dt}=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{I}_{i}d{\mathrm{\&psi;}}_i---\left(13\right)$

Consider that in i-th loop, size of current is definite value I _i(i.e. dI _i=0), then the increment of system stored energy is:

$W_m=\frac{1}{2}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{I}_{i}d{\mathrm{\&psi;}}_i+\frac{1}{2}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&psi;}}_{i}d{I}_i=\frac{1}{2}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{I}_{i}d{\mathrm{\&psi;}}_i=\frac{1}{2}d{W}_e---\left(14\right)$

Can be expressed as according to coupled magnetic field energy differential formulas:

$F_{\mathrm{em}}\mathrm{ds}=d{W}_e-d{W}_m=\frac{1}{2}d{W}_e=d{W}_m---\left(15\right)$

Coil current size is that I(is thought of as steady state value here), W _m(x _s, i) be magnetic field energy.The electromagnetic field of class half muscle segment driver adopts infinite element boundary condition in the hope of determining solution, i.e. unlimited distance B _∞=0.According to the principle of virtual work, between mover and stator, real displacement does not occur, therefore permanent magnet is at virtual displacement x _spower suffered by direction can be expressed as:

$f_s=f(x_s,I)=\frac{d{W}_m(x_s,I)}{d{x}_s}=\frac{\&PartialD;}{\&PartialD;x_s}\left[{\&Integral;}_V{\&Integral;}_0^H(B\&CenterDot;\mathrm{dH})\mathrm{dV}\right]---\left(16\right)$

Determine implementation displacement x _sunder maximum power output f _xmax(x _s) can be expressed as peak coil current I _maxthe electromagnetic force produced under (peak coil current is the rated operational current of class half muscle segment driver here, is determined by processing conditions, and under present processing conditions restriction, getting its rated operational current is 1A) condition.Therefore maximum power output f _xmax(x _s) can be expressed as

$f_{x\max }\left(x_s\right)=f(x_s,I_{\mathrm{mas}})=\frac{d{W}_m(x_s,I_{\max })}{d{x}_s}=\frac{\&PartialD;}{\&PartialD;x_s}\left[{\&Integral;}_v{\&Integral;}_s^H(B\&CenterDot;\mathrm{dH})\mathrm{dV}\right]---\left(17\right)$

6th step, according to target power output F _asdetermine that artificial-muscle series connection layer activates number n _yfollowing formula need be met

f _xmax(x _s)·(n _y-1)≤F _as≤f _xmax(x _s)·n _y(18)

7th step, the reticulate texture of artificial-muscle belongs to centrosymmetric image, in order to ensure its job stability in operational process, activating and activating according to centrosymmetric mode.The summation of all activated artificial-muscle series connection layer power output is artificial muscle target power output F _as, often pair of artificial-muscle series connection layer class power output Central Symmetry is equal.

8th step, passes to controller by each class half muscle segment driver resolved, and by R-C power displacement Shared control, the power and the displacement that complete artificial-muscle in the instruction cycle export task.

3. application example

3.1 key-course application examples

Utilize artificial-muscle kinematical equation to select key-course application example, with a simple example, the system of selection that artificial-muscle controls the first floor is carried out.If class muscle fibril driver starting point is fixed, stop target power output size is definite value f _y0, it is a that target exports absolute acceleration _ms0, target absolute velocity v _ms0, artificial-muscle targeted shrinkage length is Δ l _ms0if only have a class half muscle segment driver j active slip to produce power output, other class half muscle segment driver is all in self-locking state.

The relative acceleration of class half muscle segment driver can be expressed as:

${\stackrel{\&CenterDot;\&CenterDot;}{x}}_{s\left(k\right)}=\left\{\begin{array}{c}{a}_{\mathrm{ms}0},k=j\\ 0,k\&NotEqual;j\end{array}\right.---\left(19\right)$

The relative velocity of class half muscle segment driver can be expressed as:

${\stackrel{\&CenterDot;}{x}}_{s\left(k\right)}=\left\{\begin{array}{c}{v}_{\mathrm{ms}0},k=j\\ 0,k\&NotEqual;j\end{array}\right.---\left(20\right)$

The electromagnetic force size f of class half muscle segment driver needs generation is derived by principle of dynamics _{s (i)}for:

$f_{s\left(i\right)}=f_{y0}-\underset{q=j}{\overset{n}{\mathrm{\&Sigma;}}}{f}_{\mathrm{ests}\left(q\right)}+(n-j)\&CenterDot;M\&CenterDot;{\stackrel{\&CenterDot;\&CenterDot;}{x}}_{s\left(j\right)}+m_a{\stackrel{\&CenterDot;\&CenterDot;}{x}}_{s\left(j\right)}---\left(21\right)$

Class half muscle segment driver is when reality uses, except the external force suffered by two ends and the M.E. that produces between mover and stator (are determined by the selection of artificial-muscle, the pressure etc. that such as electromagnetic force, air pressure and hydraulic pressure produce) impact outside, also can be subject to the joint effect of the power such as gravity, centripetal force, coriolis force, friction force self produced.Self is produced except M.E., is referred to as other power f effectively _{extx (q)}.Wherein, M is class half muscle segment driver gross mass, m _afor subpart quality.

Complete goal task, consumption mechanical energy is the smaller the better, due to mechanical energy W _mecbe expressed as:

W _mec＝f _s( _j)·Δl _ms0(22)

Due to targeted shrinkage length Δ l _ms0for definite value, then the electromagnetic force of class half muscle segment driver j active slip generation is the smaller the better.For judging the punctured position of class half muscle segment driver the best, diverse location electromagnetic force creates a difference and discusses.If completing goal task class half muscle segment driver j required generation electromagnetic force is f _{s (j)}, and class half muscle segment driver j-1 required generation electromagnetic force is f _{s (j-1)}, contrast f _{s (j)}with f _{s (j-1)}then have:

$f_{s\left(j\right)}-f_{s(j-1)}=f_{\mathrm{ests}(j-1)}-M\&CenterDot;{\stackrel{\&CenterDot;\&CenterDot;}{x}}_{s(j-1)}---\left(23\right)$

When completing goal task, when time, the required electromagnetic force produced of class half muscle segment driver is greater than class half muscle segment driver j-1 required generation electromagnetic force, i.e. f _{s (j)}for the increasing function of variable j, the electromagnetic force that thus class half muscle segment driver 1 active slip produces is required minimum electromagnetic force of finishing the work; When time, the required electromagnetic force produced of class half muscle segment driver j is less than class half muscle segment driver j-1 required generation electromagnetic force, i.e. f _{s (j)}for the subtraction function of variable j, the electromagnetic force that thus class half muscle segment driver n active slip produces is required minimum electromagnetic force of finishing the work.

the application example of artificial-muscle on quadruped robot

Replace electric rotating machine with linear electromagnetic array artificial-muscle, for driving quadruped robot to move with diagonal gait, and leg exercise track is compared and analyzed, to verify the serviceability of class muscle driver.Quadruped robot adopts aluminum alloy materials to carry out simulation analysis, and Fig. 9 is shown in by quadruped robot ADAMS model.First carry out the analysis of kinematics policy to the quadruped robot adopting electric rotating machine to drive, the hip joint dynamics cycle simulation curve obtaining quadruped robot RAT simulation calculation is shown in Figure 10.Wherein, the joint angle that Figure 10-(a) is hip joint change curve in time, the joint angle speed change curve in time that Figure 10-(b) is hip joint, the joint angle acceleration change curve in time that Figure 10-(c) is hip joint, the output torque that Figure 10-(d) is hip joint change curve in time.Quadruped robot RAT knee joint dynamics cycle simulation curve is shown in Figure 11.Figure 11-(a) is kneed joint angle change curve in time, Figure 11-(b) is kneed joint angle speed change curve in time, Figure 11-(c) is kneed joint angle acceleration change curve in time, and Figure 11-(d) is kneed output torque change curve in time.

If the right front foot of robot uses artificial-muscle to drive, other tripodias still adopt electric rotating machine to drive, by the following method can pre-estimation artificial-muscle power output scope and displacement range.Figure 12 is shown in by artificial-muscle power output and output displacement conversion schematic diagram.Angle between robot rod member normal is joint angle [234], and the intersection point between normal is articulation point.Definition articulation point is respectively l to the distance of artificial-muscle and rod member attachment point _g1with l _g2, articulation point is h to the distance of artificial-muscle _das, articulation point and artificial-muscle and the angle between rod member attachment point place straight line and rod member normal are respectively γ ₁with γ ₂, the angle of artificial-muscle two attachment point and articulation center is β.

When quadruped robot size and artificial-muscle attachment point are determined, l _g1, l _g2, γ ₁with γ ₂be all fixed value, and Jch={ θ, ω, α, T} are input information, ask artificial-muscle information Mch={l _das, v _das, a _das, f _das.

Angle β can be expressed as:

β＝θ-γ ₁-γ ₂(24)

Angular velocity can be expressed as:

$\stackrel{\&CenterDot;}{\mathrm{\&beta;}}=\frac{\mathrm{d\&beta;}}{\mathrm{dt}}=\frac{d(\mathrm{\&theta;}-{\mathrm{\&gamma;}}_1-{\mathrm{\&gamma;}}_2)}{\mathrm{dt}}=\frac{\mathrm{d\&theta;}}{\mathrm{dt}}=\mathrm{\&omega;}---\left(25\right)$

Angular acceleration can be expressed as:

$\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&beta;}}=\frac{d^2\mathrm{\&beta;}}{d{t}^2}=\frac{d^2(\mathrm{\&theta;}-{\mathrm{\&gamma;}}_1-{\mathrm{\&gamma;}}_2)}{d{t}^2}=\frac{d^2\mathrm{\&theta;}}{d{t}^2}=\mathrm{\&alpha;}---\left(26\right)$

Articulation point is to the distance h of artificial-muscle _dascan be expressed as:

$h_{\mathrm{das}}=\frac{l_{g1}\&CenterDot;l_{g2}\&CenterDot;\sin \mathrm{\&beta;}}{l_{\mathrm{das}}}---\left(27\right)$

Artificial-muscle kinematic parameter modular converter calculates conversion, obtains artificial-muscle output displacement, output speed, output acceleration and power output, and these Parameter transfer are carried out artificial-muscle motion control to bottom peripheral nervous system.

Aggregative formula 24,25,26,27 and trigonometric function knowledge, artificial-muscle output information Mch={l can be derived _das, v _das, a _das, f _dasexpression formula:

$\left\{\begin{array}{c}{l}_{\mathrm{das}}=\sqrt{{l_{g1}}^2+{l_{g2}}^2-2.l_{g1}.l_{g2}.\cos \mathrm{\&beta;}}\\ v_{\mathrm{das}}=\frac{l_{g1}\&CenterDot;l_{g2}\&CenterDot;\sin \mathrm{\&beta;}}{\sqrt{{l_{g1}}^2+{l_{g2}}^2-2\&CenterDot;l_{g1}\&CenterDot;l_{g2}\&CenterDot;\cos \mathrm{\&beta;}}}\\ a_{\mathrm{das}}=\frac{l_{g1}\&CenterDot;l_{g2}\&CenterDot;\cos \mathrm{\&beta;}\&CenterDot;({l_{g1}}^2+{l_{g2}}^2-l_{g1}\&CenterDot;l_{g2}\&CenterDot;\cos \mathrm{\&beta;})-{l^2}_{g1}\&CenterDot;{l^2}_{g2}}{{({l_{g1}}^2+{l_{g2}}^2-2\&CenterDot;l_{g1}\&CenterDot;l_{g2}\&CenterDot;\cos \mathrm{\&beta;})}^{\frac{3}{2}}}\\ f_{\mathrm{das}}=\frac{T\&CenterDot;\sqrt{{l_{g1}}^2+{l_{g2}}^2-2\&CenterDot;l_{g1}\&CenterDot;l_{g2}\&CenterDot;\cos \mathrm{\&beta;}}}{l_{g1}\&CenterDot;l_{g2}\&CenterDot;\sin \mathrm{\&beta;}}\end{array}\right.---\left(28\right)$

Be 12 ~ 15kg to pre-estimating robot quality, according to robot electric rotating machine curve of output, in conjunction with artificial-muscle installation site, estimate thigh power output scope at 70 ~ 100N, range of movement is between 115 ~ 200mm, shank power output scope is at 130 ~ 160N, and range of movement is between 100 ~ 140mm.According to estimated data, the maximum output displacement in conjunction with class half muscle segment driver is x _smax=16mm, minimum power output f _smin=2.8N, can calculate class half muscle segment drive array number in the artificial-muscle at thigh place of robot is 37x6, and in the artificial-muscle at shank place of robot, class half muscle segment drive array number is 61x3.Therefore can calculate quadruped robot dynamics simulation parameter in table 1.

Table 1 quadruped robot simulation parameter

Quadruped robot RAT hip joint artificial-muscle dynamics cycle simulation curve is shown in Figure 13.Figure 13-(a) is artificial muscle length change curve in time, Figure 13-(b) is artificial contraction of muscle speed change curve in time, Figure 13-(c) is artificial contraction of muscle acceleration change curve in time, and Figure 13-(d) is power output change curve in time.The kneed artificial-muscle dynamics cycle simulation curve of quadruped robot RAT is shown in Figure 14.Figure 14-(a) is kneed artificial-muscle length change curve in time, Figure 14-(b) is kneed artificial-muscle contraction speed change curve in time, Figure 14-(c) is kneed artificial-muscle contraction acceleration change curve in time, and Figure 14-(d) is kneed power output change curve in time.In one-period, greatly, artificial-muscle length variations is larger in joint angle change recovery phase, hip joint power output is within 80N, and knee joint power output is within 150N, and the change of driving phase knee joint is less, export more energetically except contacting to earth to have instantaneously, power output in other times is also little.Simulation result shows, the application of linear electromagnetic array artificial-muscle in quadruped robot is feasible, because artificial-muscle is not moved by gear train Direct Drive Robot, thus have execution efficiency high, the life-span is long, and redundancy is good, fast response time, the features such as instantaneous acceleration is large.

Above-mentioned embodiment, for explaining the present invention, does not limit the invention, and in the protection domain of spirit of the present invention and claim, any amendment make the present invention and change, all fall into protection scope of the present invention.

Claims (4)

1. the method for designing of the netted array artificial-muscle of a kind skeletal muscle straight line, it is characterized in that: class skeletal muscle straight line netted array artificial-muscle is comprised multiple class half muscle segment driver by the method, multiple class half muscle segment driver forms class muscle fibril driver through series connection, with a class muscle fibril driver for symcenter, other class muscle fibril driver laid out in parallel are formed hexagonal honeycomb meshed symmetric array architecture, all class muscle fibril drivers are aggregated into one respectively by class sarolemma connector by the same layer class half muscle segment driver on each class muscle fibril driver, and polymerization all-in-one-piece class muscle fibril driver two ends are provided with class aponeurosis (aponeuroses) connector formation artificial-muscle, the columns m of such skeletal muscle straight line netted array artificial-muscle and the determination of number of plies n are by following technical step:

(1) maximum target output displacement L is provided according to artificial-muscle application and applied environment _asmaxand the maximum target power output F needed during artificial-muscle motion _asmax;

(2) calculating artificial-muscle parallel-connection counts n, n is layer by layer natural number, if the maximum output displacement of class half muscle segment driver is x _smax, then have

(3) calculating artificial-muscle series connection layer minimum columns k, k is natural number, if the minimum power output f of class half muscle segment driver _smin, then have

(4) calculating netted hexagonal number of total coils c, c in artificial-muscle series connection layer is natural number, and c need meet following formula

(5) finally determine that artificial-muscle array architecture is columns m is natural number, the number of plies

2. the method for designing of the netted array artificial-muscle of class skeletal muscle straight line as claimed in claim 1, it is characterized in that: all class muscle fibril drivers are aggregated into one respectively by class sarolemma connector by the same layer class half muscle segment driver on described each class muscle fibril driver, all kinds of sarolemma connector successively equidistant laid out in parallel.

3. the method for designing of the netted array artificial-muscle of class skeletal muscle straight line as claimed in claim 1, is characterized in that: polymerization all-in-one-piece class muscle fibril driver two ends are provided with class aponeurosis (aponeuroses) connector, and class aponeurosis (aponeuroses) connector is hexagonal structure.

4. the neural control method of class of the netted array artificial-muscle of class skeletal muscle straight line obtained according to claim 1 design, is characterized in that: this control method comprises the steps:

The first step, according in the control time Δ t that artificial-muscle top level control device resolves, artificial-muscle target output displacement L _asand target power output F during its motion _as;

Second step, advanced row artificial-muscle row control, and calculate artificial-muscle parallel-connection key-course number of plies n, n is natural number, if the maximum output displacement of class half muscle segment driver is x _smax, then have

3rd step, select key-course according to power output minimum principle, when moving according to artificial-muscle, the kinetics equation of class half muscle segment driver, judges to produce electromagnetic force f in its motion process _{s (j)}, i.e. electromagnetic force f _{s (i) (j)}for all classes half muscle segment driver be in shunt layer j produces monotonicity and the work shift x thereof of total electromagnetic force _sselect artificial-muscle key-course, if the electromagnetic force f that class half muscle segment driver produces _{s (j)}for Parallel Control counts the increasing function of j layer by layer, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle starting point layer is key-course; Otherwise, if electromagnetic force f _{s (j)}during subtraction function for variable j, then select work shift x _starget output displacement L can be met _as, and the shunt layer nearest apart from artificial-muscle stop layer is key-course;

4th step, in control time Δ t, in key-course, the output displacement of each class half muscle segment driver is artificial muscle target output displacement L _as, the class half muscle segment driver in non-controlling layer is in self-locking state, no-output displacement;

5th step, carries out the capable control of artificial-muscle, the work shift x of real-time measure and control layer _s, determine the maximum power output f under this work shift _xmax(x _s);

Wherein, I _maxfor the peak coil current of class half muscle segment driver, W _m(x _s, I _max) be magnetic field energy, x _sfor work shift, H is magnetic field intensity, and B is magnetic induction density, and V is the integrating range of volume integral;

6th step, according to target power output F _asdetermine that artificial-muscle series connection layer activates number n _yfollowing formula need be met:

f _xmax(x _s)·(n _y-1)≤F _as≤f _xmax(x _s)·n _y(18)

7th step, the reticulate texture according to artificial-muscle belongs to centrosymmetric image, adopts to activate to activate according to centrosymmetric mode, is artificial muscle target power output F by the summation of all activated artificial-muscle series connection layer power output _as, often pair of artificial-muscle series connection layer class power output Central Symmetry is equal;

8th step, passes to controller by each class half muscle segment driver resolved, and by R-C power displacement Shared control, the power and the displacement that complete artificial-muscle in the instruction cycle export task.