阅读说明：本技术 基于单神经元与改进粒子群算法优化双电机高速高精度同步控制方法 (Based on single neuron and improve particle swarm algorithm optimization bi-motor high-speed, high precision synchronisation control means ) 是由 王�华 吴杰 洪荣晶 于 2019-09-02 设计创作，主要内容包括：一种基于单神经元与改进粒子群算法优化双电机高速高精度同步控制方法,其特征是包括以下步骤：步骤1：根据永磁同步电机的矢量控制原理设计粒子群优化两个电机的PI控制器中K<Sub>P</Sub>、K<Sub>I</Sub>两个参数；步骤2：对粒子群算法进行改进提高其优化PI控制器的速度与精度；步骤3：通过对评价指标的设计,来保证单电机平稳启动,提高双电机启动时的高精度同步控制；步骤4：将优化后的参数赋值给电机PI控制器；步骤5：通过单神经元耦合器的设计,完成针对任意型号的两个电机的实时耦合计算,实现实时耦合控制。本发明能够明显改善在不同型号电机的双电机高转速下的高精度同步控制,并保证了在初始启动阶段、平稳运行阶段和负载扰动阶段均具有较高的转速同步精度。(One kind is based on single neuron and improves particle swarm algorithm optimization bi-motor high-speed, high precision synchronisation control means, it is characterized in that the following steps are included: step 1: designing K in the PI controller of two motors of particle group optimizing according to the principle of vector control of permanent magnet synchronous motor P 、K I Two parameters；Step 2: particle swarm algorithm being improved and improves its speed and precision that optimize PI controller；Step 3: by the design to evaluation index, to guarantee single motor smooth starting, improving high-precise synchronization control when bi-motor starting；Step 4: giving the parameter assignment after optimization to motor PI controller；Step 5: by the design of single neuron coupler, completing to calculate for the real-time coupling of two motors of disposable type, realize coupling control in real time.The present invention can be obviously improved the control of the high-precise synchronization under the high revolving speed of bi-motor of different model motor, and ensure that and all have higher revolving speed synchronization accuracy in initial start up phase, even running stage and load disturbance stage.)

1. one kind is based on single neuron and improves particle swarm algorithm optimization bi-motor high-speed, high precision synchronisation control means, feature Be the following steps are included:

Step 1: K in the PI controller of two motors of particle group optimizing is designed according to the principle of vector control of permanent magnet synchronous motor_P、 K_ITwo parameters；

Step 2: particle swarm algorithm being improved and improves its speed and precision that optimize PI controller；

Step 3: by the design to evaluation index, to guarantee single motor smooth starting, improving high-precision when bi-motor starting Synchronously control；

Step 4: giving the parameter assignment after optimization to motor PI controller；

Step 5: according to cross-coupling control method, coupler is established using single neural Meta-algorithm, it is collected using two motors Real-time revolving speed is completed to calculate for the real-time coupling of two motors of disposable type, realizes coupling control in real time.

2. according to claim 1 synchronous with particle swarm algorithm optimization bi-motor high-speed, high precision is improved based on single neuron Control method, it is characterized in that the step 3 specifically comprises the following steps:

Specifically, initializing particle by chaotic maps, chaotic maps method is as follows:

In formula, τ is mapping function, z⁽ⁱ⁾For chaos sequence；

Further, design performance evaluation index guarantees single motor smooth starting, improves high-precise synchronization when bi-motor starting Control.The improved time of design and the integral (IITAE) of absolute error product are as follows:

Wherein, p-overshoot controls precision；The output of u (t)-controller；Y (t)-speed setting value；y_sp(t)-actual speed value； E (t)-systematic error；ω₁、ω₂、ω₃- weight, and ω₃>>ω₁；

Kinematic nonlinearity inertia weight is devised, is shown below:

In formula, ω (t)-inertia weight；ω_int- initial weight (generally takes 0.9)；ω_finalWeight (one when-greatest iteration As take 0.4)；T-current iteration number；t_max- maximum number of iterations；

Further, in order to execute mutation operation to the particle still accumulated, to change particle direction of advance, reach and beat The purpose of shot；Variation rule is as follows:

In formula, X_is(t+1) for after t+1 iteration i-th of particle in the position of s dimension space；η is to obey [0,1] Gaussian Profile Stochastic variable；C is the value chosen according to the actual situation.

3. according to claim 2 synchronous with particle swarm algorithm optimization bi-motor high-speed, high precision is improved based on single neuron Control method, it is characterized in that the step 3 may additionally include following steps:

Comprehensive method as claimed in claim 2 has obtained a kind of Modified particle swarm optimization algorithm, is mentioned using dynamic inertia weight The precision and convergence rate of high algorithm, and enhance by particle mutation the ability of searching optimum of algorithm, algorithm is being protected Optimum results are faster obtained while demonstrate,proving rotational speed optimization precision, the speed of particle and location updating method are as follows after improving:

v_is(t+1)=ω (t) v_is(t)+c₁ r_is(p_is(t)-x_1s(t))+c₂ r_2s(p_gs(t)-x_gs(t))

x_is(t+1)=x_is(t)+v_is(t+1)

In formula, i=[1, m], s=[1, S], Studying factors c₁And c₂It is nonnegative constant, inertia weight ω is nonnegative number, r₁And r₂For Mutually independent pseudo random number obeys being uniformly distributed on [0,1].

4. according to claim 5 synchronous with particle swarm algorithm optimization bi-motor high-speed, high precision is improved based on single neuron Control method, it is characterized in that the step 5 may additionally include following steps:

According to the single neuron coupler of cross-coupling control method, Hebb learning algorithm is supervised to the defeated of single neuron using having It is connected each other between the deviation for entering, exporting and exporting and weight, controller output are as follows:

Wherein normalize weight w_i' (k) are as follows:

In formula, η 1, η 2, η 3 are learning rate；w_iIt (k) is weight；x_iIt (k) is condition input signals；U (k) is output；Δe(k) For control deviation increment；E (k) is the deviation of value of feedback and setting value.

Technical field

The present invention relates to motor high-speed, high precision synchronously control fields, more particularly to one kind is based on improvement population (IPSO) With the bi-motor high-speed, high precision synchronisation control means of single neuron.

Background technique

In industrial automation production field, the driving method of single motor is in fields such as heavy load, long range and multivariable Controls It closes, has been increasingly difficult to meet the needs of modern industrial technology development, the occasion that it requires that bi-motor works at the same time increasingly increases It is more.Especially cut to pieces in tooth lathe in numerical control, the superiority and inferiority of the revolving speed net synchronization capability between cutter and workpiece by the system that directly affects can By property and Gear Processing effect.

It is a kind of based on Principles of Gear Connection that tooth technology is cut in numerical control, which to pieces, is carried out efficiently using dedicated serrated knife tool of cutting to pieces, high-precision tooth Take turns processing method.For this processing method by traditional gear hobbing, gear shaping is complex as a movement, inside, in external gear process, It cuts serrated knife tool to pieces and is both equivalent to hobcutter, and be equivalent to pinion cutter, have an inclined setting angle between cutter and workpiece, pass through The rotating ratio of workpiece and cutter is controlled to realize high speed, high finishing tooth processing.Due to cutting this special technique of tooth to pieces, in cutter shaft While rotation, workpiece spindle also can flexibly be changed in synchronous rotary by changing the speed ratio of cutter shaft and work spindle The type and shape of converted products, therefore it is very high to the synchronism requirement of cutter shaft and workpiece spindle to cut tooth processing to pieces.In process In, due to needing to shorten process time and using DRY CUTTING, so revolving speed required for cutter shaft and work spindle is high, this is just Cutting the synchronous control technique of the dual-servo-motor axis of tooth machine to pieces to numerical control, more stringent requirements are proposed.

Current existing bi-motor synchronisation control means is not directed under the high revolving speed of bi-motor required for cutting tooth operating condition to pieces High synchronism required precision studied.And existing bi-motor simultaneous techniques is tended not in view of two motor models are different In the case of the influence to bi-motor synchronization accuracy of difference and single motor control performance that compensates of synchronous error.Therefore the present invention Net synchronization capability at high speeds is deposited for existing bi-motor synchronisation control means to be declined and starting and having load The problems such as synchronous effect when disturbance is insufficient devises a kind of high based on improvement population and single neural Meta-algorithm optimization bi-motor Accurate synchronization control method has biggish Practical Project value.

Summary of the invention

It is high based on single neuron and improvement particle swarm algorithm optimization bi-motor high speed that the purpose of the present invention is to provide one kind Accurate synchronization control method.In view of the deficiency of the prior art, single motor controller is optimized using improvement particle swarm algorithm And single neuron optimizes new motor rotating speed genlock, controls synchronous rotational speed, can be obviously improved not With the high-precise synchronization control under the high revolving speed of bi-motor of model motor, and ensure that in initial start up phase, even running rank Section and load disturbance stage all have higher revolving speed synchronization accuracy.

The invention adopts the following technical scheme:

One kind is based on single neuron and improves particle swarm algorithm optimization bi-motor high-speed, high precision synchronisation control means, including Following steps:

Step 1: the PI controller of two motors of particle group optimizing is designed according to the principle of vector control of permanent magnet synchronous motor Middle K_P、K_ITwo parameters.

Step 2: particle swarm algorithm being improved and improves its speed and precision that optimize PI controller.

Step 3: by the design to evaluation index, to guarantee single motor smooth starting, improving height when bi-motor starting Accurate synchronization control.

Step 4: giving the parameter assignment after optimization to motor PI controller.

Step 5: according to cross-coupling control method, coupler being established using single neural Meta-algorithm, is acquired using two motors The real-time revolving speed arrived is completed to calculate for the real-time coupling of two motors of disposable type, realizes coupling control in real time.

The step 3 specifically comprises the following steps: specifically, initializing particle, chaotic maps method by chaotic maps It is as follows:

In formula, τ is mapping function, z⁽ⁱ⁾For chaos sequence.

Further, design performance evaluation index guarantees single motor smooth starting, improves high-precision when bi-motor starting Synchronously control.The improved time of design and the integral (IITAE) of absolute error product are as follows:

Wherein, p-overshoot controls precision；The output of u (t)-controller；Y (t)-speed setting value；y_sp(t)-practical turn Speed value；E (t)-systematic error；ω₁、ω₂、ω₃- weight, and ω₃> > ω₁。

Kinematic nonlinearity inertia weight is devised, is shown below:

In formula, ω (t)-inertia weight；ω_int- initial weight (generally takes 0.9)；ω_finalPower when-greatest iteration Weight (generally taking 0.4)；T-current iteration number；t_max- maximum number of iterations.

Further, it is reached to execute mutation operation to the particle still accumulated to change particle direction of advance To the purpose for breaing up particle.Variation rule is as follows:

In formula, X_is(t+1) for after t+1 iteration i-th of particle in the position of s dimension space；η is to obey [0,1] Gauss The stochastic variable of distribution；C is the value chosen according to the actual situation.

The step 3 may additionally include following steps: comprehensive method as claimed in claim 2 has obtained a kind of improvement particle Colony optimization algorithm improves the precision and convergence rate of algorithm using dynamic inertia weight, and enhances calculation by particle mutation The ability of searching optimum of method enables algorithm faster to obtain optimum results while guaranteeing rotational speed optimization precision, after improving The speed and location updating method of particle are as follows:

v_is(t+1)=ω (t) v_is(t)+c₁ r_is(p_is(t)-x_1s(t))+c₂ r_2s(p_gs(t)-x_gs(t))

x_is(t+1)=x_is(t)+v_is(t+1)

In formula, i=[1, m], s=[1, S], Studying factors c₁And c₂It is nonnegative constant, inertia weight ω is nonnegative number, r₁ And r₂For mutually independent pseudo random number, being uniformly distributed on [0,1] is obeyed.

The step 5 may additionally include following steps: according to the single neuron coupler of cross-coupling control method, use There is supervision Hebb learning algorithm to connect each other between the deviation and weight of the input of single neuron, output and output, controls Device output processed are as follows:

Wherein normalize weight w_i' (k) are as follows:

In formula, η 1, η 2, η 3 are learning rate；w_iIt (k) is weight；x_iIt (k) is condition input signals；U (k) is output；Δe It (k) is control deviation increment；E (k) is the deviation of value of feedback and setting value.

Compared with prior art, the invention has the advantages that；

One kind of the invention optimizes bi-motor high-speed, high precision synchronously control side based on improvement population and single neural Meta-algorithm Method, the first high synchronization accuracy under targetedly raising single motor control effect, the high speed of Lai Tigao bi-motor.It utilizes Improve particle swarm algorithm the PI controller parameter of two motors is optimized, can guarantee iteration initial stage be always maintained at compared with Big weight makes the state for keeping algorithm iteration early period focusing on global optimizing, and the iteration later period keeps lesser weight, stresses part Optimizing.And mutation operation is introduced, it avoids algorithm from falling into local optimum with this, accelerates convergence rate and low optimization accuracy, improve The whole control effect of single motor.

Further, by designing new evaluation index, the corresponding overshoot for punishing single motor improves single motor starting effect Fruit improves the starting synchronization accuracy of bi-motor.

It is calculated later by cross-coupling control method using single neuron in raising bi-motor Integral synchronous precision aspect Method establishes coupler, using the requirement revolving speed and collected real-time revolving speed of two motors, completes two electricity for being directed to disposable type The real-time coupling of machine calculates, and realizes coupling control in real time.

In conclusion the present invention, which compensates for the net synchronization capability that existing bi-motor synchronisation control means is deposited at high speeds, to be had Decline and the problems such as synchronous effect when starting and having load disturbance is insufficient.It realizes high for disposable type bi-motor Fast high-precise synchronization control.

Detailed description of the invention

Fig. 1 is control method structure chart of the present invention；

Fig. 2 is that the present invention improves particle swarm algorithm flow chart；

Fig. 3 is kinematic nonlinearity inertia weight change curve of the present invention；

Fig. 4 is single neuron coupled structure figure of the present invention；

Fig. 5 is Modified particle swarm optimization of the present invention (IPSO) and conventional particle group optimizes (PSO) fitness value variation diagram；

Fig. 6 a-Fig. 6 d is control method of the present invention and ordinary particle group algorithm optimization bi-motor synchronously control side The experimental result comparison diagram of method, in which:

Fig. 6 a is commonsense method motor rotating speed figure；

Fig. 6 b is motor rotating speed figure of the present invention；

Fig. 6 c is commonsense method synchronous error figure；

Fig. 6 d is synchronous error figure of the present invention.

Specific embodiment

The present invention is further elaborated with reference to the accompanying drawing.

Such as Fig. 1 to Fig. 6 d.

Referring to Fig. 1, the present invention is a kind of to optimize bi-motor high-speed, high precision based on single neuron and improvement particle swarm algorithm Synchronisation control means includes the following steps:

Step 1, the PI controller of two motors of particle group optimizing is designed according to the principle of vector control of permanent magnet synchronous motor. Select the revolving speed evaluation index value of permanent magnet synchronous motor as input value, by the K of the motor speed ring PI controller_P、K_IIt is worth conduct Output valve, wherein K_P、K_IIt is ratio P, integral two parameters of I of speed ring PI controller.

Step 2, particle swarm algorithm is improved and improves its speed and precision that optimize PI controller, please refer to Fig. 2.

Step 2.1, particle swarm algorithm parameter is initialized, it is assumed that in a S dimension search space, initialize a population rule Mould is the population of N, initializes the position X of group within the allowable range_i=(x_i1,x_i2,…,x_iS) and speed V_i=(V_i1, V_i2,…,V_iS), wherein i represents i-th of particle in population, x by i=1,2 ..., m_iDIt is particle i in the position of S dimension, v_iS For particle i S dimension speed；Position value interval is [- x_max,x_max], as parameter K_P、K_IValue interval；Speed takes Value section is [- v_max,v_max], as parameter K_P、K_IThe value interval of pace of change；Interval range can generally be known according to existing Know and experience provides, set v here_max=k x_max(0.1≤k≤1.0)。

Step 2.2, to arbitrary i, s, in [- x_max,x_max] in by chaotic maps generate x_iS；To arbitrary i, s, [- v_max,v_max] in by chaotic maps generate v_iS；Chaotic maps function is as follows:

In formula, τ is mapping function, z⁽ⁱ⁾For chaos sequence.

Step 2.3, calculate the fitness value of each particle, engineering in practice frequently be time and absolute error product Integral (ITAE):

In formula, y (t)-speed setting value；y_sp(t)-actual speed value；ITAE is fitness value.

Step 2.4, to each particle by its fitness value and its desired positions P lived through_isFitness value compared Compared with, if preferably, as current desired positions.Wherein individual extrema representation (individual) the particle i of particle is immediately The adaptive optimal control angle value P undergone in search process_bestWith optimal location P_is=(P_i1,P_i2,…,P_is)；The current desired positions of particle It is determined by following formula:

In formula, f (x) is the objective function minimized, P_iFor current desired positions.

Step 2.5, the desired positions P each particle lived through its fitness value and the overall situation_gsFitness value carry out Compare, if preferably, as current global desired positions.Global desired positions indicate entire population in all previous iteration Adaptive optimal control angle value g in the process_bestWith optimal location P_gs=(P_g1,P_g2,…,P_gs)；It should be noted that individual extreme value is each Particle has one, the entire population of global extremum only one.

Step 2.6, the speed of particle and position are updated respectively by following formula, update method is as follows:

v_is(t+1)=ω v_is(t)+c₁ r_is(p_is(t)-x_1s(t))+c₂ r_2s(p_gs(t)-x_gs(t))

x_is(t+1)=x_is(t)+v_is(t+1)

In formula, i=[1, m], s=[1, S], Studying factors c₁And c₂It is nonnegative constant, inertia weight ω is nonnegative number, r₁ And r₂For mutually independent pseudo random number, being uniformly distributed on [0,1] is obeyed.

Since inertia weight ω controls influence of the previous speed to latter speed, when ω is larger, previous speed is affected, Ability of searching optimum is stronger；When ω is smaller, latter speed is affected, and local search ability is stronger.Therefore it is non-to devise dynamic Linearly decreasing weight is shown below:

In formula, ω (t)-inertia weight；ω_int- initial weight (generally takes 0.9)；ω_finalPower when-greatest iteration Weight (generally taking 0.4)；T-current iteration number；t_max- maximum number of iterations.

Its with the number of iterations change curve as shown in figure 3, kinematic nonlinearity inertia weight ω change curve be lordosis after Recessed decreasing function.Since the global optimizing performance that conference improves system was arranged in the value of ω, local optimal searching performance is reduced, it is on the contrary .So it is recessed after the curve lordosis of the kinematic nonlinearity inertia weight of design, guarantee to be always maintained at iteration initial stage larger Weight, make the state for keeping algorithm iteration early period focusing on global optimizing, the later period keeps lesser weight, stresses local optimal searching. It avoids algorithm from falling into local optimum with this, accelerates convergence rate, improve algorithm performance.

Further, by the observation to algorithm operational process, discovery in the circulating cycle, still has the position of some particles to exceed The boundary of search space, and due to the limitation of algorithmic rule, these particles are all placed to the side in space after going beyond the scope Edge, more and more build-up of particles together, cause the similitude of particle higher and higher, are unfavorable for finding globally optimal solution.It borrows Variation thought in genetic algorithm of reflecting, mutation operator is introduced into algorithm, executes mutation operation to the particle accumulated, thus Change particle direction of advance, achievees the purpose that break up particle.Variation rule is as follows:

In formula, X_is(t+1) for after t+1 iteration i-th of particle in the position of s dimension space；η is to obey [0,1] Gauss The stochastic variable of distribution；C is the value chosen according to the actual situation.

Operation in this way, not only reduces the similarity of particle, while having expanded particle itself search space, makes grain Son jumps out current location, scans in bigger space, improves a possibility that finding the more figure of merit.

A kind of Modified particle swarm optimization algorithm has to sum up been obtained, the precision and receipts of algorithm are improved using dynamic inertia weight Speed is held back, and enhances the ability of searching optimum of algorithm by particle mutation, algorithm is enabled to guarantee rotational speed optimization precision Optimum results are faster obtained simultaneously.The speed of particle and location updating method are as follows after it is improved:

v_is(t+1)=ω (t) v_is(t)+c₁ r_is(p_is(t)-x_1s(t))+c₂ r_2s(p_gs(t)-x_gs(t))

x_is(t+1)=x_is(t)+v_is(t+1)

Step 2.6, if meeting termination condition, solution is exported；Otherwise step 2.3 is returned to.

Step 3, design performance evaluation index guarantees single motor smooth starting, and high-precision when improving bi-motor starting is same Step control.Due to Performance Evaluating Indexes in practice frequently be time Yu absolute error product integral (ITAE), this can lead It causes to produce speed overshoot to guarantee the rapidity of single motor starting, so that the starting for reducing bi-motor synchronizes essence Degree.And ITAE is the transit time and transient error quantified in system response process, without in compartment system dynamic process Overshoot error.So using the system of ITAE, there is larger overshoot in the phase before responding, cannot stablize rising and reach setting speed, Influence net synchronization capability when electric motor starting.The integral (IITAE) of improved time Yu absolute error product are devised thus.

In formula, p-overshoot controls precision；The output of u (t)-controller；Y (t)-speed setting value；y_sp(t)-practical turn Speed value；E (t)-systematic error；ω₁、ω₂、ω₃- weight, and ω₃>>ω₁。

Limit excessive overshoot by the introducing of p, IITAE enabled to guarantee that speed ring is stablized, it is accurate and without compared with Track rotary speed instruction to big overshoot.

Step 4, it gives the parameter assignment after optimization to motor PI controller, realizes optimal control.

Step 5, according to cross-coupling control method, coupler is established using single neural Meta-algorithm, utilizes adopting for two motors The real-time revolving speed collected is completed to calculate for the real-time coupling of two motors of disposable type, realizes coupling control.

Traditional bi-motor cross-coupling synchronously control is using same ratio value as two electricity of speed Coupling Control Unit direct compensation The speed error of machine, but this ratio value needs artificial experience to adjust, synchronism is not high.And artificial neural network is not only to this Kind nonlinear system can be approached arbitrarily, and can realize control to nearly all uncertain and nonlinear system.But Conventional neural networks correction time and Weight Training are long and composed structure is complicated, and single neuron is similar with neural network, can be with Any Nonlinear Function Approximation, structure is relatively easy, high reliablity, is easy to realize in the controls, can be controlled in real time System.Single neuron coupled structure please refers to Fig. 4.

Hebb learning algorithm is supervised to phase between the deviation and weight of the input of single neuron, output and output using having Mutually connection, controller output are as follows:

Wherein normalize weight w_i' (k) are as follows:

In formula, η 1, η 2, η 3 are learning rate；w_iIt (k) is weight；x_iIt (k) is condition input signals；U (k) is output；Δe It (k) is control deviation increment；E (k) is the deviation of value of feedback and setting value.

Lasting adjustment by single neural Meta-algorithm to weight improves the adaptive of single neuron speed coupler and oneself Learning ability.Single Neuron Controller may be implemented under the high revolving speed of two motors of different model under cross coupling structure The on-line tuning of synchronously control ensure that the high synchronization accuracy under the high revolving speed of two motor of different model.

If Modified particle swarm optimization (IPSO) fitness value variation diagram of Fig. 5 is to learn, under identical fitness index, Although two algorithms are all gradually reduced the fitness of motor, traditional PS O, which optimizes 57 times, reaches convergence, and IPSO is at the 53rd time Fitness value is less than traditional PS O after iteration, shows the better convergence efficiency of IPSO.It sees on the whole, PSO is in the 4th After iteration, fitness value just no longer has a greater change, and has fallen into local optimum, and IPSO is because of kinematic nonlinearity inertia weight Recessed design after curve is first convex, so that its state for being in global optimizing before 53 iteration, avoids and be trapped in local optimum The case where solution, and stress local optimal searching in 53 times to 88 times iteration, obtain the PI parameter more more outstanding than PSO.

Such as Fig. 6 a-Fig. 6 d, showing given two motor speed is 3000rpm, and two motor speeds are applied after stablizing to motor 1 Add the fluctuation of speed figure of 15Nm load disturbance.Motor after IPSO-IITAE optimizes reduces super in the No Load Start stage Tune amount improves the steady-state performance of motor.And by single neuron coupler, when motor 1 is by load disturbance, motor 2 is done Corresponding rotation speed change improves performance of noiseproof out.In the No Load Start stage, starting is synchronized while overshoot reduces 3.8% Precision is compared and increases 2.6 times；In stable operation stage, due to the collective effect of IPSO-IITAE and single neuron, stable state is missed Difference is reduced to 5rpm by 20rpm；When system is by load disturbance, control precision is obviously improved, and is greatly strengthened The performance of noiseproof of system.From above data it is found that system is given by IPSO-IITAE adjusting rotating-speed tracking, while passing through list Neuron coupler makes its Fast synchronization to the compensation of two motor speed rings.Therefore system is regardless of when there is non-loaded and No Load Start Revolving speed synchronous error is all significantly reduced, and system shows superior robustness, synchronism and rotating-speed tracking.

Although the present invention gives specific case study on implementation, it is not intended to limit the scope of the present invention, is not departing from Under the premise of present inventive concept, the various changes and improvements that engineers and technicians make technical solution of the present invention in this field, Protection scope of the present invention should all be fallen into.The claimed technology contents of the present invention are all described in the claims.